Stability and Stabilization of Biocatalysts

STABILITY AND STABILIZATION OF BIOCATALYSTS STABILITY AND STABILIZATION OF BIOCATALYSTS Progress in Biotechnology Vo...

Author: F.J. Plou | J.L. Iborra | P.J. Halling | A. Ballesteros

71 downloads 4070 Views 37MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

STABILITY AND STABILIZATION OF BIOCATALYSTS

STABILITY AND STABILIZATION OF BIOCATALYSTS

Progress in Biotechnology Volume 1 New Approaches to Research on Cereal Carbohydrates (Hill and Munck, Editors) Volume 2 Biology of Anaerobic Bacteria (Dubourguier et al., Editors) Volume 3 Modifications and Applications of Industrial Polysaccharides (Yalpani, Editor) Volume 4 Interbiotech '87. Enzyme Technologies (Bla~ej and Zemek, Editors) Volume 5 In Vitro Immunization in Hybridoma Technology (Borrebaeck, Editor) Volume 6 Interbiotech '89. Mathematical Modelling in Biotechnology (Bla~ej and Ottov&, Editors) Volume 7 Xylans and Xylanases (Visser et al., Editors) Volume 8 Biocatalysis in Non-Conventional Media (Tramper et al., Editors) Volume 9 ECB6: Proceedings of the 6th European Congress on Biotechnology (Alberghina et al., Editors) Volume 10 Carbohydrate Bioengineering (Petersen et al., Editors) Volume 11 Immobilized Cells: Basics and Applications (Wijffels et al., Editors) Volume 12 Enzymes for Carbohydrate Engineering (Kwan-Hwa Park et al., Editors) Volume 13 High Pressure Bioscience and Biotechnology (Hayashi and Balny, Editors) Volume 14 Pectins and Pectinases (Visser and Voragen, Editors) Volume 15 Stability and Stabilization of Biocatalysts (Ballesteros et al., Editors)

Progress in Biotechnology 15

STABILITY AND STABILIZATION OF BIOCATALYSTS Proceedings of an International Symposium organized under auspices of the Worldng Party on Applied Biocatalysis of the European Federation of Biotechnology, the University of Cordoba, Spain, and the Spanish Society of Biotechnology Cordoba, Spain, April 19-22, 1998

Edited by A. B a l l e s t e r o s

Department of Biocatalysis, CSIC Institute of Catalysis, Madrid, Spain

F. J. P l o u

Department of Biocatalysis, CSIC Institute of Catalysis, Madrid, Spain

J. L. I b o r r a

Department of Biochemistry and Molecular Biology "B", Faculty of Chemistry, University of Murcia, Murcia, Spain P. J. H a i l i n g

Department of Pure & Applied Chemistry, University of Strathclyde, Glasgow, UK

(,

ELSEVIER Amsterdam

- Lausanne - New York - Oxford - Shannon - Singapore - Tokyo 1998

ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 EO. Box 211, 1000 AE Amsterdam, The Netherlands

© 1998 Elsevier Science B.V. All rights reserved. This work and the individual contributions contained in it are protected under copyright by Elsevier Science B.V., and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier Science Rights & Permissions Department, PO Box 800, Oxford OX5 1DX, UK; phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-mail: [email protected]. You may also contact Rights & Permissions directly through Elsevier's home page (http://www.elsevier.nl), selecting first 'Customer Support', then 'General Information', then 'Permissions Query Form'. In the, USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (978) 7508400, fax: (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W 1P 0LP, UK; phone: (+44) 171 436 5931; fax: (+44) 171 436 3986. Other countries may have a local reprographic rights agency for payments. Derivative Works Subscribers may reproduce tables of contents for internal circulation within their institutions. Permission of the publisher is required for resale or distribution of such material outside the institution. Permission of the publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Contact the publisher at the address indicated. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the publisher. Address permissions requests to: Elsevier Science Rights & Permissions Department, at the mail, fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. First edition 1998 Transferred to digital printing 2006 Library of Congress Cataloging in Publication Data A catalog record from the Library of Congress has been applied for. ISBN: 0-444-82970-9 ~ ) The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed and bound by CPI Antony Rowe, Eastbourne

Preface The aims of the Working Party on Applied bioeatalysis of the European Federation of Biotechnology are: i) to increase the understanding of bioeatalysis, in particular its commercial applications; ii) to take initiatives in areas of growing scientific and industrial interest and importance in the field of applied bioeatalysis; iii) to identify key topics which may be rate-limiting the development of scientific and technological capabilities in applied biocatalysis and to take steps to stimulate these areas. In the light of these aims, the Working Party has organised over a dozen symposia since 1985. In 1992 in Maastricht (The Netherlands) the WP held a symposium on Stability and Stabilization of Enzymes (ref. 1). We felt that six years later would be a good time to organize a second symposium. This would bring up to date the work already going on, and identify possible breakthroughs in the research. This time the scope of the conference was broadened by including whole cell biocatalysts in addition to enzymes. The international symposium on "Stability and Stabilization of Biocatalysts" was held in Cordoba, Spain from April 19-22, 1998. It was possible to bring together 210 participants, representing every continent, to learn from 150 oral and poster communications. Delegates had the opportunity to hear lectures on stability and stabilization grouped within the topics: inactivation mechanisms and reactors; stabilization by chemical modification; noncovalent processes in solution; protein engineering and thermophile enzymes; immobilized enzymes; non-conventional media; and whole cells. The papers included in these Proceedings are grouped within the above mentioned topics. We hope that the symposium, and this book, which contains most of the papers presented in Cordoba, will make a useful contribution to this key area of applied biocatalysis. We thank Professors J.M. Marinas and D. Luna, Department of Organic Chemistry, University of Cordoba, Spain, for their valuable contribution to the organization of the congress.

The Editors, Madrid, Murcia and Strathclyde, May 1998.

REFERENCE I. W.J.J. van der Tweel, A. Harder and R.M. Buitelaar (Eds.),Stabilityand Stabilization of Enzymes, Elsevier,Amsterdam, 1993.

This Page Intentionally Left Blank

vii

Acknowledgements The Organizing Committee gratefully acknowledges the support of the following sponsors: Diputaci6n de C6rdoba Ayuntamiento de C6rdoba Universidad de C6rdoba (Programa propio) Junta de Andalucia Comisirn Interministerial de Ciencia y Tecnologia (CICYT) Consejo Superior de Investigaciones Cientificas (CSIC) Instituto de Cat~ilisis y Petroleoquimica, CSIC Caja Murcia AVE-RENFE IBERIA Airlines Caja Sur

a This Page Intentionally Left Blank

ix

Contents Preface Acknowledgements

vii

Inactivation mechanisms and Reactors Diagnosing the inactivating process of enzymes O. Misset, A. van Dijk Stability and stabilization of glucose-fructose oxidoreductase from Zymomonas mobilis against irreversible inactivation during substrate turnover in biochemical reactors B. Nidetzky, M. Ftirlinger, D. Haltrich, K.D. Kulbe

19

Reactor performance under thermal inactivation and temperature optimization with chitin-immobilized lactase A. Illanes, L. Wilson, C. Altamirano, A. Aillaphn

27

Fast in-situ-characterization ofbiocatalysts long-term stability M. Boy, A. Dominik, H. Voss

35

Chemical and thermal stability of ferulic acid (feruloyl) esterases from Aspergillus C.B. Faulds, F.O. Aliwan, R.P. de Vries, R.W. Pickersgill, J. Visser, G. Williamson

41

Stability properties of calf intestinal alkaline phosphatase U. McKeon, B. O'Connor, C. 0'Ffig~tin

47

Stabilization of [3-glucanase and its effect on the substrate hydrolysis pattern I. Markovi6, B. Markovi6-Dev~i6, S. Gamulin, N. Pavlovi6

53

Some factors affecting the behavior of anhydrous ~x-chymotrypsin at high temperature D. Pirozzi, G. Greco Jr.

59

Horseradish peroxidase stability in the course of phenols oxidation A.M. Egorov, Yu.L. Kapeluich, T.A. Pastuhova, M.Yu. Rubtsova, D.N. Cherepanov

65

The effect of impurities of crude olive residue oil on the operational stability of the Candida rugosa lipase immobilized in polyurethane foams A.C. Correia, S. Ferreira-Dias

71

Approaches for improved identification of mechanisms of enzyme inactivation M. Polakovi6, P. Vr~tbel, V. B~tle~

77

Stability and stabilization of a-l,4-D-glucan phosphorylases R. Grie[31er, B. Mtiller-Fembeck, S. D'Auria, F. La Cara, B. Nidetzky

83

Oxidation by hydrogen peroxide of D-amino acid oxidase from Rhodotorula gracilis V. Obreg6n, I. de la Mata, F. Ram6n, C. Acebal, M.P. Castill6n

89

Cosolvent effect on the synthesis of ampicillin and cephalexin with penicillin acylase C. Aguirre, J. Baeza, A. Illanes

95

Conversion and stability studies on enzyme-membrane reactor with lipase immobilized by different methods T. Venyige, E. Cshnyi, Cs. Sisak

101

Study of the deactivation process of the glucose oxidase-catalase enzymatic system by means of simulation L.E. Romero, D. Cantero

107

Stabilization by chemical modification

113

Stabilization ofhydrolases by chemical modification with fatty acids or polyethylene glycol F.J. Plou, M.V. Calvo, M. Ferrer, A. Ballesteros

115

Modification of the carbohydrate moiety in ribonuclease B and its influence on the protein stability probed by limited proteolysis U. Arnold, R. Ulbrich-Hofinann

121

Comparative studies on selective modification of e-amino groups in lipases and phospholipase A2 B.C. Koops, A.J. Slotboom, H.M. Verheij

127

Analysis of the thermal deactivation kinetics of a-chymotrypsin modified by chemoenzymatic glycosilation M.A. Longo, D. Combes

135

Increasing the operational stability of flavoproteins by covalent cofactor binding W.J.H. van Berkel, R.H.H. van den Heuvel, M.W. Fraaije, C. Laane

141

Properties of PEG-modified microbial proteases. Activity and stability studies T.M. Fatum, A. Agerlin Olsen, C.C. Fuglsang, D. Otzen

147

The effect of crosslinking on thermal inactivation of cellulases J. Bilen, U. Bakir

151

Studies of the stability of acid phosphatase (A. niger) by cross-linking with glutaraldehyde and soil humates N. Ortega, L. Berzal, M.D. Busto, M. Perez-Mateos

157

Non-covalent processes in solution

163

Folding and association versus misfolding and aggregation of proteins R. Jaenicke

165

Activity of monoclonal antibodies in prevention of in vitro aggregation of their antigens B. Solomon, T. Katzav-Gozanski, R. Koppel, E. Hanan-Aharon

183

Stability of monodeamidated forms of ribonuclease A F. Catanzano, G. Graziano, S. Capasso, G. Barone

189

Pressure effects on protein oligomeric dissociation C. Balny

197

Effect of high hydrostatic pressure on enzyme stability V. AthOs, D. Combes

205

Thermodynamic stability of a monomeric derivative of bovine seminal ribonuclease G. Barone, F. Catanzano, G. Graziano, V. Cafaro, G. D'Alessio, A. Di Donato

211

Thermodynamic stability of ribonuclease P2 sulfolobus solfataricus and some mutants G. Graziano, F. Catanzano, P. Fusi, P. Tortora, G. Barone

217

Stabilization of ct-chymotrypsin by DMSO E. Flaschel, L. Ebmeier

223

Thermal stabilization by its ligands ofNADP÷-isocitrate dehydrogenase from the thermophilic cyanobacterium Phormidium laminosum M.A. Pardo, M.J. Llama, J.L. Serra

229

The characterization of protein stability from hydrogen exchange: maximum entropy analysis of co-solvent effects on exchange rates B. Souhail, I.M. Plaza del Pino, J.M. Sanchez-Ruiz

235

Preferential hydration changes upon protein unfolding in water-cosolvent mixtures H.O. Hammou, I.M. Plaza del Pino, J.M. Sanchez-Ruiz

241

Charactbrizing protein-cosolvent interactions coupled to protein refolding by kinetic calorimetry M.C. Boulaich, A. Parody-Morreale, J.M. Sanchez-Ruiz

247

Characterizing cosolvent effects on protein stability: cold-denaturation of ubiquitin in the presence of guanidine B. Ibarra Molero, J.M. Sanchez-Ruiz

251

Acetylcholinesterase-ethanol interactions: inactivation, substrate and ligand exclusion F. Ortrga, D. Garcia, J.-L. Marty

257

xii Effect of a mixed stabilizer-denaturant system on the stability of enzyme activity at high temperatures: lysozyme as a model Y. Sangeeta Devi, U.B. Nair, R. Bhat

263

A simple folding method for high level production of the hydrophobic disulfide bonded hepatitis B X protein by inclusion body route and its structural analysis I. Marczinovits, J. Molnhr, M.Z. Kele, P.T. Szab6, T. Janfiky

269

Protein engineering and Thermophile enzymes

275

Rigidity of thermophilic enzymes A. Fontana, V. de Filippis, P. Polverino de Laureto, E. Scaramella, M. Zambonin

277

Improvement of thermal stability of a diagnostic enzyme, Streptomyces cholesterol oxidase, by random and site-directed mutageneses and a structural interpretation Y. Murooka, Y. Nishiya, M. Toyama, M. Aoike, M. Yamashita, N. Hirayama

295

Protein engineering for thermostabilisation of proteins: some theoretical rules and application to a 13-glucanase E. Querol, J. Pons, J. Cedano, M. Vallmitjana, F. Garcia, C. Bonet, J. P6rez-Pons, A. Planas, A. Mozo-Villarias j//

303

Mechanisms of stabilization of the [3-glycosidase from the hyperthermophilic archaeon

Sulfolobus solfataricus M. Moracci, M. Ciaramella, L.-H. Pearl, B. Cobucci Ponzano, M. Rossi

311

A strategy for engineering thermostability: the case of cyclodextrin glycosyltransferase J.C.M. Uitdehaag, B.W. Dijkstra

317

Thermophilic esterases and the amino acid "traffic rule" in the hormone sensitive lipase subfamily G. Manco, F. Febbraio, M. Rossi

325

Cloning and stabilization of NAD-dependent formate dehydrogenase from Candida boidinii by site-directed mutagenesis H. Slusarczyk, M. Pohl, M.-R. Kula

331

Usefulness of bacterial thermostable enzymes in clinical chemical analysis K. Tomita, K. Nomura, T. Miura, H. Shibata

337

Effect of temperature and Ca2÷ on the degree of multiple attack exhibited by mesophilic and thermophilic a-amylases A. Kramhott, B. Svensson

343

Degradation and denaturation of stable enzymes D. Thompson, R. Fernhndez-Lafuente, C. Mateo, D.A. Cowan, J.M. Guisfin, R. Daniel

349

xiii

Non-conventional media

353

Engineering stability of enzymes in systems with organic solvents V.V. Mozhaev

355

Inactivation of enzymes at the aqueous-organic interface P.J. Hailing, A.C. Ross, G. Bell

365

Exploiting hysteresis for high activity enzymes in organic media J.Partridge, P.J. Hailing, B.D. Moore

373

Limited proteolysis of proteins by thermolysin in trifluoroethanol P. Polverino de Laureto, E. Scaramella, M. Zambonin, V. De Filippis, A. Fontana

381

Enzyme inactivation by inert gas bubbling M. Caussette, A. Gaunand, H. Planche, B. Lindet

393

Effects of water-miscible solvents on the stability and specificity of cyclodextringlucosyltransferases A.D. B lackwood, C. Bucke

399

Stabilization of immobilized enzymes against organic solvents: complete hydrophilization of enzymes environments by solid-phase chemistry with polyfunctional macromolecules R. Femhndez-Lafuente, C.M. Rosell, L. Caanan-Haden, L. Rodes, J.M. Guishn

405

Effect of sorbitol on immobilized cx-chymotrypsin thermostability in low-water system T. de Diego, P. Lozano, M.J. lqiguez, J.L. Iborra

411

Stability of immobilized enzyme-polyelectrolyte complex against irreversible inactivation by organic solvents V. Levitzky, P. Lozano, A. Gladilin, J.L. Iborra

417

Enhancement of invertase activity in organic media for oligosaccharide synthesis S. Bielecki, R.I. Somiari

423

Effects of crown ethers on the activity of enzymes in peptide formation in organic media D.J. van Unen, J.F.J. Engbersen, D.N. Reinhoudt

429

Recovery of the activity of an immobilized lipase after its use in fat transesterification S. Ferreira-Dias, C.S. Duarte, V. Falaschi, S.R. Marques, J.H. Gusm~o, M.M.R. da Fonseca

435

Effects of lipid-bome compounds on the activity and stability of lipases in microaqueous systems for lipase-catalyzed interesterification X. Xu, C.-E. Hoy, J. Adler-Nissen

441

xiv Stabilisation of lipases for activity in ammoniolysis F. van Rantwijk, A.C. Kock-van Dalen, R.A. Sheldon

447

Water sorption isotherm as a tool to explore hydration of the microenvironment of biocatalysts J.M. Shnchez-Montero~ R.M. de la Casa, J.V. Sinisterra

453

Thermal stability of free and immobilised Pseudomonas cepacia lipase in aqueous and organic media G. Pencreac'h, J.C. Baratti

459

The stabilization of enzyme in organic solvent at low temperature H.J. Lee, J.R. Kim, Y.J. Yoo

465

A comparative study of thermal inactivation of enzymes in supercritical carbon dioxide A. Giel3auf, T. Gamse, E. Klingsbichel, H. Schwab, R. Marr

471

Studies of the stability of aminoacylase in some organic solvents L. Boross, J. Koshry, 1~. Stefanovits-Bhnyai, C. Sisak, B. Szajhni

477

Cutinase activity and enantioselectivity in supercritical fluids N. Fontes, M.C. Almeida, S. Garcia, C. Peres, J. Grave, M.R. Aires-Barros, C.M. Soares, J.M.S. Cabral, C.D. Maycock, S. Barreiros

483

Effect of pressure on enzyme activity in compressed gases M.C. Almeida, N. Fontes, E. Nogueiro, S. Garcia, C. Peres, A. Silva, M. Carvalho, S. Barreiros

487

Immobilized enzymes

493

Activity and structural stability of adsorbed enzymes W. Norde, T. Zoungrana

495

Covalent immobilization of glucose oxidase on A1PO4 as inorganic support. F.M. Bautista, M.C. Bravo, J.M. Campelo, A. Garcia, D. Luna, J.M. Marinas, A.A. Romero

505

The effect of site-specific immobilization on the thermal stability of thermolysin-like neutral proteases J. Mansfeld, G. Vriend, B. Van den Burg, G. Venema, V.G.H. Eijsink, R. UlbrichHofmann

513

Immobilization ofhydantoin cleaving enzymes from Arthrobacter aurescens DSM 3747 - Effect of the coupling method on the stability of the L-N-carbamoylase M. Pietzsch, H. Oberreuter, B. Petrovska, K. Ragnitz, C. Syldatk

517

Thermoinactivation of l~-xylosidase immobilized on nylon M.J. Duefias, P. Estrada

523

XV

Increase of thermal stability of tannin acyl hydrolase by covalent immobilization through its carbohydrate side chain: application in tea cream solubilisation P. Nicolas, J.L. Sauvageat, S. Reymond, E. Raetz

529

Stability in the presence of organic solvents of dextransucrase from Leuconostoc mesenteroides NRRL B-512F immobilized in calcium-alginate beads M. Alcalde, F.J. Plou, M.T. Martin, M. Remaud, P. Monsan and A. Ballesteros

535

Immobilisation and characteristics of glucose oxidase immobilised on Convective Interaction Media (CIM) disks A. Podgomik, M. Vodopivec, H. Podgomik, M. Barut, A. Strancar

541

Immobilization of spinach leaf hydroperoxide lyase L.M. Simon, Sz.J. Mfirczy, M. Kotorman, Sz.A. N6meth, B. Szajhni

547

A new immobilization system for Candida rugosa lipase: characterization and applications K. Carbone, M. Casarci

553

Soybean lipoxygenases: purification and stability of the free and immobilized enzymes A. Chikere, B. Galunsky, V. Kasche

559

Stabilization of lipase B from Candida antarctica by immobilization on different supports M. Arroyo, J.M. S~nchez-Montero, J.V. Sinisterra

565

Covalent immobilization of crude and partially-purified lipases onto inorganic supports: stability and hyperactivation A.R. Alchntara, I. Borreguero, M.T. L6pez-Belmonte, J.V. Sinisterra

571

Immobilization of alkaline phosphatases on various supports B. Surinenaite, V. Bendikiene, B. Juodka

577

Stabilization of serine proteases by immobilization V. Bendikiene, B. Juodka

583

Whole cells

589

Immobilized Cells: Plasmid stability and plasmid transfer J.N. Barbotin, D. Mater, M. Craynest, J.E. Nava Saucedo, N. Truffaut, D. Thomas

591

Stabilization of immobilized cells systems using a modified metal surface, fructose polymer levan and high cell concentration M. Bekers, E. Ventina, A. Karsakevich, I. Vina

603

Plasmid stability in recombinant Saccharomyces cerevisiae expressing the EXG1 gene in free immobilized cultures A. Guillhn, T. Lfi Chau, E. Roca, M.J. Nfifiez, J.M. Lema

611

xvi Bioreductions by Pyrococcusfuriosus at elevated temperatures E. van den Ban, H. Willemen, H. Wassink, H. Haaker, C. Laane

619

Stability of free and immobilized Mycobacterium sp. cells in aqueous and organic media P. Fernandes, J.M.S. Cabral, H.M. Pinheiro

625

Operational stability of immobilized C. reinhardtii cells: an approach to its potential as biocatalyst for N-consuming processes I. Garbayo, C. V ilchez

631

Induction and stability of cholesterol oxidase from cells of a Rhodococcus J. Kreit, P. Germain

639

Stability of functionally active fusion proteins during their biosynthesis and isolation from expressing bacterial cells L.M. Vinokurov, A.V. Yakhnin, T.V. Ivashkina, V.N. Ksenzenko, Yu.B. Alakhov

645

Miscellaneous Production ofbio-esters by immobilized lipases K. B61afi-Bak6, L. Gubicza, E. Cs~nyi, M.A. Galhn, J.A. Moreno

653

An aspergillusflavus strain promoting oleic acid esterification in isooctane V. Loscos, G. Caries, B. Perpifia, M. Torres, N. Sala, R. Canela

657

Covalent enzyme immobilization in different porous polymer membranes H.-G. Hicke, M. Becker, G. Malsch, M. Ulbricht

661

Functionalised cross-linked polyvinyl alcohol as new matrix for lipase immobilization E. Cernia, G. Milana, G. Ortaggi, C. Palocci, S. Soro

667

Enzymatic biphasic membrane reactors for the synthesis of chiral products H.A. Sousa, J.P.S.G. Crespo

673

Coimmobilization of enzymes and cells on chitosan and derivatives A.B. Martin, M. Picciolato, A. Heras

679

Parameters affecting the activity ofMucor miehei esterase 30 000 in a solvent freesystem M. Karra-Chaabouni, S. Pulvin, D. Thomas

685

Recombinant antigens by fusion of antigenic epitopes to a GST partner J. Molnhr, Marczinovits, M. Kiss, S. Husz, G. T6th, L. Dorgai, M. Khlmhn

691

Modification of metal substrates and its application to the study of redox proteins T. Pineda, J.M. Sevilla, A.J. Roman, R. Maduefio, M. Blazquez

697

xvii

Influence of some inducers on activity of ligninolytic enzymes from corncob cultures of Phanerochaete chrysosporium in semi-solid-state conditions S. Rodriguez Couto, M.A. Longo, C. Cameselle, A. Sanromhn

703

NMR study of hydration of liquid phase during lipase catalysed esterification in non aqueous media C. Sarazin, C. Roblot, B. Decagny, F. Ergan, J.N. Barbotin, J.P. Srguin

709

ot-chymotrypsin-catalysed synthesis of N-acetyl-L-tyrosine esters in organic media K. L~tszl6, L.M. Simon

713

Purification and characterization of penicillin V acylase from Streptomyces lavendulae R. Torres, I. de la Mata, M.P. Castill6n, M. Arroyo, J. Torres, C. Acebal

719

Studies on the regioselective acylation of sugars catalyzed by lipase in tert-butanol V. Sereti, H. Stamatis, F.N. Kolisis

725

Activity of cardosins A and B in the presence of organic solvents A.C. Sarmento, M. Barros, E. Pires

731

Effect of fermentation conditions in the enzyme activity and stereoselectivity of crude lipase from Candida rugosa R.M. de la Casa, A. SLnchez, J.V. Sinisterra, J.M. S~nchez-Montero

735

Biotransformations catalyzed by Candida rugosa lipase partially purified by precipitation and by organic solvents treatment S. Chamorro, A.R. Alchntara, J.M. S~.nchez-Montero, J.V. Sinisterra

741

Effect of pH and temperature on immobilised Thiobacillusferrooxidans cells over nickel alloy fibre J.M. G6mez, D. Cantero

747

Index of authors

753


Inactivation mechanisms and Reactors


Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Halling (Editorg) 9 1998 Elsevier Science B.V. All rights reserved.

Diagalosing the inactivating process o f e n z y m e s Onno Misset and Alard van Dijk Research and Development, Food Specialties Division, Gist-brocades P.O. Box 1, NL 2600 MA, Delft, The Netherlands

1.

INTRODUCTION

Enzymes are widely used in a large variety of industrial applications (for an excellent text book see [1 ]). The commercial products appear in various formulations such as liquids, powders, granulates, tablets etceteras containing either single enzymes or mixtures of enzyme activities. An important aspect of enzymes is their stability, both during production and storage as well as their ultimate application. Elevated temperatures, extremes of pH, highand low salt concentrations and high water activity are known to be responsible for a low or reduced stability. Therefore, enzymes are produced and applied under conditions such as to retain maximal activity. Several methods are available to improve the stability properties of enzymes [2]. The use of stabilising additives is well-known and widespread and also protein engineering has proven to be successful in developing intrinsically more stable mutant enzymes. Chemical modification and immobilisation are, although also recognised as stabilising techniques, much less applied Despite the enormous amount of literature on stabilising additives, relatively little is known about the precise reasons why particular stabilisers work as good as they do. This is at least in part caused by the lack of knowledge about the actual enzyme inactivating processes. Also the success of protein engineering is significantly increased if the inactivating process is known at the molecular level. Of course some common inactivation mechanisms will play a role, but each enzyme has its own 'individual' sensitivities. Knowledge of these sensitivities can direct the stabilising process and contribute to the development of cost-effective enzyme products. In this article we will focus on how to identify the inactivating process. Hereto we propose a series of coherent experiments that is designed to quickly and unambiguously identify said inactivating event. This proposal emerged from extensive discussions between enzymologists, protein- and analytical chemists and formulation technologists from our company. We combined knowledge on enzyme stability (section 2) and the various analytical methods and their potential to be applied to enzyme products (section 3) in a multidisciplinary mode in order to arrive at proposal "diagnosing the inactivating processes of enzymes" (section 4).

2.

STABILITY OF ENZYMES

The stability of industrial enzymes is put to the test during all the various stages of its life as described earlier [2]. The first stage is the production process at the enzyme manufacturer's and comprises usually fermentation, downstream processing, formulation and storage. The subsequent stage involves transport and usage by the customer in a specific application. The enzyme molecules have to be able to survive all the different processing steps, each with its own physical and chemical conditions, in order to end up in the enzyme product, preferably at very high yields and then to deliver its promised performance in the customer's application. In order to do so, enzyme manufacturers are selecting and developing enzymes with already a high intrinsic stability and low sensitivity toward all kinds of inactivating processes. However, since enzymes still are protein molecules, they do possess a vulnerable three-dimensional structure with reactive amino acid side chains (acids, bases, nucleophiles etceteras).

~ non-covalent t

" 1 .., covalent

denaturation aggregation precipitation

oxidation deamidation Maillard proteolysis racemization hydrolysis

Figure 1. Scheme for the irreversible inactivation of an enzyme. N is the native enzyme; D represents an ensemble of reversibly, partially denatured forms.

The irreversible inactivation of an enzyme can be represented by the scheme in Figure 1 ([3] and reproduced from [2]). The enzyme molecule normally exists in its native, active state N which is in equilibrium with the partially denatured, enzymaticaUy inactive state D. At rising temperatures, the enzyme tends to unfold in a co-operative process. The temperature at which [N] = [D] is called the melting temperature Tin. The Tm of enzymes may vary from 30 to >100 ~ but, in aqueous solution is usually somewhere between 60 and 80 ~

Two different inactivating processes can be discerned: non-covalent and covalent. At higher temperatures, the unfolding process can proceed to such an extend that refolding is no longer possible because of denaturation accompanied by aggregation and precipitation. As a result, these non-covalent processes cause irreversible inactivation. The covalent processes involve chemical modification of the enzyme molecule because sensitive amino acids react with for instance oxidants, acids or bases, reducing sugars etceteras. Also included here is 'simple' proteolysis that may occur in case either the enzyme product is 'contaminated' with proteases (e.g. also produced by the production organism) or in case these are present in the application of the customer. The following physical and chemical factors are considered bad for enzyme stability because they either promote the non-covalent unfolding and/or aggregation/precipitation processes or enhance the rate of the covalent processes: 9 high temperature 9 extremes ofpH 9 extremes of ionic strength 9 high water activity

Solid- versus liquid enzyme preparations: solvent effects. The solution state of an enzyme has important effects on its thermodynamic stability: in solution the enzyme will have different stability characteristics than in the dry state. Drying is known to often cause structural changes, but fortunately most of them are reversible [4]. Enzymes are hygroscopic. In the dry form they absorb water. The stability of the enzyme is very much dependent on the moisture content and, at low contents, T m (or better Zg for solids) decreases with increasing content (Tg: glass transition temperature). In addition to an improved thermodynamic stability, the dry enzyme is also better resistant against microbiological degradation or chemical inactivation (except gas mediated inactivations like oxidation). From a stability point of view the dry state is therefore preferred for enzyme storage. Nevertheless, liquid enzyme formulations are still used because: 9 Cost aspect: evaporating water is very expensive! 9 The enzyme is inactivated by the drying process or, 9 The application requires a liquid formulation (costumer demand) The choice of solvent system of the liquid formulation is very important. Although water is the common solvent for enzymes, co-solvents are often added to improve the stability of the native protein. They are often added in high concentrations [5] and can either be other solvents (e.g. glycerol) or solids (e.g. sucrose) and all aim at reducing the water activity. Inactivation and changes in the protein molecule Irreversible inactivation of enzymes is always associated with a change in one or more of its basic properties as summarised in Table 1. In order to be able to diagnose the inactivating process, it is therefore essential to avail of analytical techniques that allow to monitor these changes preferably in a sensitive and unambiguous way. In the next paragraphs (section 2.2), the effect of the different covalent modifications will be discussed followed by a discussion of the various analytical techniques (section 3).

Table 1 - Basic properties of enzyme molecules. PROPERTY

DETERMINED BY

Structure primary secondary tertiary quaternary 'Molecular weight

Amino acid sequence Occurrence of alpha-helices and beta-strands/sheets 3-Dimensional folding Subunit composition Determined by the chemical composition of the protein molecule (i.e. amino acid composition plus cofactors, metal ions etceteras). .........

Charge overall

distribution

- Hydrophol~icity

2.1

Resultant of positive and negatively charged amino acid residues. Strongly determined by pH. Zero at pI (isoelectric point), overall negative when pH>pI and viceversa Location of the charged and/or polar side chains of amino acids on the surface of the protein molecule. Together with the hydrophobicity important for functional and physical properties of enzymes. Occurrence of amino acids with hydrophobic side chains at the surface of the enzyme molecule. Together with charge distribution important for functional and physical properties of enzymes. . . . . . . .

Covalent modifications

Many covalent changes in protein molecules have been described in the literature as being causative of inactivation: oxidation, deamidation, MaiUard reactions, racemization, proteolysis and polypeptide chain breakage (Figure 1). The likelihood that these processes occur depends on whether the reactive amino acid side chains are accessible to the reacting species or not. This means that in the enzyme in the N-conformation only surface-exposed side chains will react. In the partially denatured structures D, many more residues may become exposed and the irreversible inactivation may proceed at higher rates. Table 2 summadses the reactivity of the various amino acids and the effect of the covalent modifications on either the charge/polarity of the enzyme molecule or the molecular weight (M.W.) - see also Table 1. It is clear from Table 2, that most chemical modification cause changes in the polarity and charge of the enzyme molecule as well as its molecular weight. Yet, the extend and sign of the changes varies with the type of chemical modification.

Table 2. Reactivity of amino acids and effect On enzyme molecule A M I N O ACID

REACTIVITY

P R O P E R T Y OF ENZYME

none

none

like normal organic acids (formation of esters, reduction to alcohols potent nucleophile, acylation, alkylation, arylation, Maillard reactions

charge and M.W.

MOLECULE AFFECTED

aliphatic acidic

Ala, Val, Leu, Ile, Pro, Gly Glu, Asp

basic

Lys

charge and M.W.

charge and M.W.

His acylation, alkylation, electrophilic substitution oxidation Arg

polar

sulphurcontaining aromatic

Ser, Thr Asn, Gln Met, Cys Trp, Tyr, Phe

charge and M.W. can form heterocyclic condensation product with 1,2 and 1,3 dicarbonly compounds like alcohols deamidation alkylation oxidation electrophilic substitution oxidation, alkylation

polarity, M.W. charge polarity/charge and M.W. polarity/charge and M.W.

Effect on activity Covalent modification of an amino acid within the active site of the enzyme will certainly lead to an altered (lower) activity of the enzyme because the interactions between enzyme and substrate/product are affected. When the enzyme is not completely inactive, we can speak of a cripple enzyme. Whether this is the case depends on the position and the role of the amino acid in the enzyme molecule. Covalent modification of amino acid residues outside the active site usually has not an effect on the activity of the enzyme, but, the modifications may reduce the (thermo)-stability of the enzyme. This can be explained in terms of a shift in the equilibrium between the (modified) forms of N and D

2.1.1 Oxidation Oxidation of amino acids can occur by oxygen (02 - whether or not activated by metal ions), hydrogen peroxide or other stronger oxidants such as peroxy acids (peracetic acid). These reactions proceed via electrophilic addition or, more general, via radical addition. The latter makes the reaction rather pH-independent. The most sensitive amino acids are the sulphur containing cystein and methionine. Less sensitive are the aromatic amino acids tryptophane, tyrosine, histidine and phenylalanine. Table 3 shows that the oxidation products of these amino acids possess a different chemical structure which will disturb the enzyme structure and hence its (catalytic) properties.

Table 3. Oxidation of amino acids AMINO ACID OXIDATIONPRODUCT Cystein -CH2-SH

Methionine -CH2-CH2-S-CH3

TryPtophane

Cystine -CH2-S-S-CH2Cysteic acid ..... -CHz-SO~H. . . . Sulfoxide -CH:-CH2-SO-CH3 Sulfon -CH2-CHz-OSO-CH~ Oxindole-alanine N-formylkynurenine

OXIDANT

.

.

Oxygen H202 Performic acid . . . . . . H202 Chloramine T Performic acid

Histidine

H202 Iodine H202, Periodate, Iodine

Tyrosine

H~O2, Periodate,Iodine

The oxidised form of the enzyme can be regarded as a cripple form of the enzyme molecule. The change in molecular weight can be very small: the oxidation of one methionine toward its sulfoxide only increases its molecular weight by 18 dalton and leaves the charge almost unaffected. Oxidation of one cystein toward the cysteinic acid introduces one additional negative charge and increases the molecular weight by 54 dalton. 2.1.2 Deamidation

This reaction involves the hydrolysis of the uncharged amino acids glutamine (Gin) and asparigine (Asn) into the negatively charged glutarnicacid (Glu) and asparticacid (Asp). The molecular weight is affectedjust by the substitutionof a NH2-group by an oxygen-atom (l dalton). The reaction increases with increasing pH and temperature [6]. Since deamidation is a hydrolytic reaction, it may be expected to proceed much slower in powders although this will be dependent upon the hydration. Since the reactiondoes not require specialchemicals, it may occur during all stages of the life cycle of the enzyme. The effect of introducing the negative charge(s) in the enzyme molecule can be instability,inactivityor both.

2.1.3 Maillard reaction Browning of protein samples upon heating with sugars was first observed by Maillard in 1912 [7]. This reaction, also referred to as glycation of protein, is not only very important in the food industry, but also plays a role in ageing of proteins like the eye lens crystallins and in certain diseases such as diabetes mellitus. The chemistry of the Maillard reaction is very complex and not fully understood. Usually, the consecutive reactions are divided on the basis of product formation: early and advanced glycation products. The first step in the glycation process is the reaction between the aldehyde group of a reducing sugar (such as glucose) and the free amino group of a protein. The latter is either the alpha-amino group of the N-terminus of the protein or the gamma-amino group of the lysine side chain. The result is a still labile Schiff base which undergoes a rearrangement to form the more stable Amadori product. The Schiff base and the Amadori product are called the early glycation products. The steps following the ketoamine formation are not completely elucidated. A large number of reactions follow, resulting in complex carbohydrate-protein complexes. These structures can possess chromophoric (browning) and fluorophoric properties and are able to cross link protein chains [8]. These compounds are called advanced glycation end products. Also, chain breakage has been reported to occur as a result of the Maillard reaction. Other studies revealed that not only lysines undergo a reaction with sugars, but also tryptophanes and tyrosines can undergo modification with advanced products of the Maillard reaction (e.g. dialdehydes) [9] Conditions or compounds which favour the Maillard reactions are high(er) temperature, organic acids (such as sorbic acid). The reaction is supposed to involve the unprotonated aminogroup of lysine residues. This means that, with a pK~ of approximately 10, the reaction increases with increasing pH. As a result of the chemical modification of the enzyme molecule, both its molecular weight and charge distribution (and colour) will severely be affected, accompanied by both inactivation and a further decreased stability. 2.1.4 Racemization This process converts the (natural) L-amino acids to the D-form and can occur with each amino acid of an enzyme except glycines. It has been reported that only Asn and Asp are the most susceptible to racemization [10]. Furthermore, threonine and isoleucine contain an asymmetric carbon atom in their side chain which can, in principle, undergo racemization. The effect of racemization can be (partial) unfolding accompanied by loss of activity, although some authors consider that the contribution of racemization to thermoinactivation is negligible [11 ]. The process takes place preferentially at the extremes of pH and at high(er) temperatures.

2.1.5 Proteolysis Proteolysis is the process in which the enzyme molecule (i.e the substrate) is degraded by a protease. The sites of proteolytic attack are determined mainly by the specificity of the protease, but also involves flexible parts of the substrate which are located at the surface. Factors which promote proteolytic activity are the temperature and partially denaturing conditions for the substrate. The effect of other conditions such as the pH and the presence of metal ions are also depending on the type of protease. In general, the higher the concentration of both protease and its substrate, the faster the rate of proteolysis and, therefore, the rate of

10 inactivation. Proteolytic degradation usually leads to inactivation because the enzyme will be cut into small fragments.

2.1.6 Polypeptide chain breakage Apart from chain breakage by proteolysis, non-enzymatic breakage can occur at the extremes of pH extremes and elevated temperatures. Especially Asn-Gly bonds appear to be very sensitive. Effects similar to those after proteolysis.

3.

ANALYTICAL METHODS - WHAT ARE THEY WORTH?

There are various methods available which can be used to study the biochemical properties (see Table 4 for an overview). These can be categorised into a few distinct groups that will be shortly commented.

3.1 Amino acid analysis Two different methods are commonly applied in order to supply information about either the N-terminal amino acid sequence of the native enzyme molecule or (tryptic or CNBr-cleaved) fragments thereof or the quantitative amino acid composition. Comparison of the results with those for the non-inactivated enzyme may indicate if and where chemical modification occurred. Both methods are neither very accurate nor reproducible. Therefore, at the least they can only be indicative rather then very conclusive. For instance, quantitative amino acid analysis has a standard deviation of at least 10% which means that it is impossible to conclude for instance that 1 out of 5 methionines is chemically modified or similar reductions took place for other amino acids. 3.2 Electrophoresis Polyacrylamide gel electrophoresis (PAGE) is a very easy and therefore commonly performed experirnent. It can be carried out under several different conditions. In the presence of the surfactant sodium dodecyl sulphate (SDS-PAGE) the enzyme molecule becomes completely unfolded and coated with the negatively charged surfactant. Because of the constant charge/size ratio, the enzymes only migrate in the gel according to their (subunit) molecular weight. The changes in molecular weight that can be detected are in the order of >500 (compare with subunit molecular weight ranging from 15,000 - 50,000 for most industrial enzymes). Migration of enzymes in their native state (Native PAGE and Capillary Zone Eleetrophoresis) depends on both the charge and molecular weight. A change of even one charge (e.g. caused by deamidation, oxidation, Maillard reactions ...), gives already a significant change in nmning behaviour (as we observed also with protein engineering by which charged amino acid residues were mutated - not shown here). With isoelectric focusing (IEF) the enzymes migrate to the pH-value in the gel where their net charge is zero. Therefore, changes in charge earl also be detected very sensitively with IEF, provided that the enzymes remain soluble under these conditions and that the pI's are within the pH-range of the technique (2500 Dalton) or isoelectric point (theoretical calculations show that one charge mutation changes the pI of an enzymes by 0.10-0.15). A complication with industrial enzymes is that the enzyme of interest may be a minor component in the enzyme mixture. This means that a 2D-gel may contain several spots and it can be difficult if not impossible to identify the interesting spots. In some cases this problem can be solved by using enzymograms for detection of specific activities (e.g. gelatin test for proteases). However, inactivated species or enzyme fragments will not show up in such staining methods, making the method of limited use. The solution is to have polyclonal antibodies against the protein of interest that make it possible to identify the relevant protein (or protein fragments). Such an enzym-specific 'staining' is crucial in the problem analysis and requires: 9 Purified enzyme of interest 9 Preparation ofpolyclonal antibodies against the purified enzyme of interest 9 A reference 2D-pattern of the intact (purified) enzyme of interest The development of these relatively simple tools is essential for a quick and thorough analysis of the causes of enzyme instability. Purification of the enzyme of interest is essential in the other stages as well.

Conclusion r r

In stage II the enzyme product will be analysed by 2D-gel electrophoresis (IEF and SDS-PAGE) using regular or specific (enzymogram or antibody) staining. Antibody staining requires purification of the enzyme and antibody production

16 In case stage II indicates no covalent modifications of the enzyme of interest (within the resolution of the techniques described), the analysis enters stage III where smaller covalent changes or changes in the conformational properties of the enzyme can be detected. In the opposite case where a clear covalent modification can be demonstrated, the analysis enters stage IV in order to further unravel the precise nature of the chemical modification. Stage III- Is there a minor change in molecular weight or a conformational change in the protein with no mass changes? A small change in molecular weight arising from limited proteolysis (a few amino acids N- or C-terminal) or an electrically 'neutral' chemical modification (oxidation of methionine to the sulfoxide form) still can have occurred but undetectable by the method applied in stage II. As a result of its high sensitivity and high resolution (dalton range) mass spectroscopy is the method of choice to look at this range of minor modification. Alternatively, the enzyme instability is caused by a loss of the 3-dimensional conformational integrity of the enzyme molecule. This can be detected by methods such as nuclear magnetic resonance (NMR), Fourier transform infrared spectroscopy (FT-IR) and preferred by us - circular dichroism (CD). These analytical tools can only be applied to purified enzymes samples, i.e. single components in a defined buffer system. This requires: 9 Purification of the enzyme of interest 9 A reference spectrum the (unmodified) enzyme of interest In some cases a useful reference spectrum may be recorded of the crude sample but in the majority of cases the matrix will interfere.

Conclusion r r r r

Analysis in stage III requires purification of the enzyme of interest from both the inactivated as well as the reference product followed by, Recording of a mass spectrum to identify small changes in molecular weight and to identify critical residues (of peptides) Recording CD (optionally NMR or FTIR) spectrum and comparison with reference spectrum of unmodified reference enzyme Determination of Tin

Stage IV. Determination of the cause of the activity change and the identification of critical residues involved. It has been concluded in stage II that there is a covalent modification leading to enzyme instability and/or inactivation. In stage IV we will now focus on the question: what type of covalent modification (proteolysis, deamidation, oxidation, (de)glycosylation, Maillard) and optionally - what is (are) the sensitive site(s) in the amino acid sequence. The most straightforward analysis is mass spectrometry. This requires at least a partially purified enzyme often in combination with a peptide map of the enzyme [12].

Proteolysis Multiple bands on SDS-PAGE indicate proteolysis; purification of the fragments followed by mass spectroscopy, directly gives information about the sensitive sites in the protein, provided that the amino acid sequence is known (for the majority of industrial enzymes this is the case). Proteolysis can be caused by a contaminating protease or, in case

17 the enzyme of interest is a protease: autodigestion. These two options can be distinguished by diluting the sample and follow the rate of inactivation as a function of the dilution factor. In case of autodigestion the rate will be independent of enzyme concentration (first order process), in the other case it will be concentration dependent (second order process).

Modifications, other than proteolysis, leaving a soluble enzyme. In case of other types of covalent modifications, assuming the modified protein is still soluble, the modified protein will probably still behave similarly to the intact protein in chromatographic procedures, and can be purified accordingly. Covalent modifications all give rise to specific changes in MW; therefor, mass analysis of the purified protein (with or without peptide mapping) will directly give information about the type of modification. Deamidation, for instance, gives rise to a change in MW of only 1 dalton whereas oxidation and Maillard reactions will give rise to larger changes in MW, the size of which characterises the specific event (the MW change for the oxidation if two thiols to a disulphide differs from that for the oxidation of a thiol to a sulfoxide). Identification of the site of modification requires knowledge of the amino acid sequence of the protein; it can be determined using MS-MS analysis (two coupled mass spectrometers, separated by a collision chamber) of the intact protein or of MS analysis of the peptide fragments after enzymatic or chemical cleavage and purification. The inactivation process may be an inherent protein property or be catalysed by a matrix component; the two options can be distinguished via a dilution series, as described above. Modifications, other than proteolysis, resulting in insoluble protein. In case the modified protein becomes insoluble (e.g. due to aggregation), it will first have to be solubilised using chaotropic agents (urea, guanidine chloride) or detergents after which the same track can be followed as described above for the soluble modified protein. The purification procedure, however, will probably differ from the one for the native protein. Conclusion r Analysis in stage IV requires purification of the enzyme of interest from both the inactivated as well as the reference product followed by: r Mass spectroscopy: conclusion about the identity of the major covalent changes r Fragmentation of enzyme (peptide mapping) r Determination of sensitive site after fragmentation r Determination whether the inactivation process is an inherent enzyme property

5.

ACKNOWLEDGEMENT

We like to acknowledge our colleagues Ruud Barendse, Koos Batist, WiUem Bijleveld, Raoul Cassaigne, Wim Dusoswa, Jan-Metske van der Laan, Gabrie Meesters, Erry Oosterom, Marcel Pi~t, Dick Schipper, Mirjam Scholte, Paul Schuurhuizen, Marcel van Tilborg and Bert Vroemen from Gist-brocades for providing experimental data, reviewing the literature, stimulating discussions and financial support.

18 References

[ 1] Godfrey, T and West, S.I. (eds), Industrial Enzymology, Stockton Press, New York, 1996 [2] Misset, O. in Stability and stabilisation of enzymes, Studies in Organic Chemistry, 47, (1993) 111-131 (Tweel, W.J.J. van den, Harder, A. and Buitelaar, R.M. eds), Elsevier Science Publishers, Amsterdam [3] Lumry, R. and Eyring, H., J. Phys. Chem., 58, (1954), 110. [4] Griebenow, K. and Klibanov, A.M., Proc. Nat. Acad. Sc., 92, (1995), 10969-10976 [5] Timasheff, S.N. in Methods in Molecular Biology, 40, 253-269 (Shirley, B.A. eds), Humana Press Inc., Totowa (1995) [6] Aher, T.J. and Klibanov, A.M., Science, 228 (1985), 1280 [7] Maillard, L.C., C.R. Seances Acad. Sci. 154, (1912), 66-68 [8] Yen, G.C. et al. Food Sci. Technol. 29, (1989), 273-288 [9] Ashoor, S.H. et al. J. Food. Sci. 49, (1984), 1206-1207 [ 10] Zhao, M. et al. Bioorganic Chemistry, 17, (1989), 36-40 [ 11 ] Zale, S.E. and Klibanov, A.M. Biochemistry, 27 (1986), 5432-5444 [12] Williams, K.R. and Stone, K.L., Mol. Biotechnol. 8, (1997), 155-167

Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

19

Stability and stabilization of glucose-fructose oxidoreductase f r o m Zymomonas

mobilis a g a i n s t i r r e v e r s i b l e

inactivation

during

substrate

turnover

in

b i o c h e m i c a l reactors B. Nidetzl~, M. F'tirlinger, D. Haltrich and K. D. Kulbe Division of Biochemical Engineering, Institute of Food Technology, Universitiit fur Bodenkultur Wien (BOKU), Muthgasse 18, A- 1190 Wien, Austria 1. I N T R O D U C T I O N Gluconic acid and sorbitol are chemical commodities with many applications in food and chemical industries. Both compounds are industrially produced from glucose whereby gluconic acid is derived by chemical or microbial oxidation and sorbitol is obtained by chemical reduction. The ethanologenic bacterium Zymomonas mobilis produces an enzyme, glucosefructose oxidoreductase (GFOR), that can synthesize both products in a simultaneous manner, thus converting a substrate solution equimolar in glucose and fructose into products [ 1,2]. The simultaneous utilization of glucose and fructose for production of, respectively, gluconic acid and sorbitol is thought to add significant value to the educt mixture, even though separation of the products is required after exhaustive substrate conversion. GFOR contains tightly bound cofactor, E-NADP(H), and catalyses two individual half reactions [3]: (i)

D-fructose + E-NADPH D-sorbitol + E-NADP

(ii)

D-glucose + E-NADP D-glucono-1,5-1actone + E-NADPH

(ii)'

D-glucono-1,5-1actone + H20 D-gluconic acid

D-Glucono-1,5-1actone is unstable in aqueous solution at around neutral pH and hydrolyses spontaneously. The net reaction is a transfer of reduction equivalents from glucose to fructose mediated by enzyme-bound NADP(H), and the cofactor cycles between its oxidized and reduced form. GFOR seems to be specific for the acceptor molecule, fructose, but not for the donor aldose. As consequence, several aldose sugars other than glucose such as xylose, galactose and even lactose [4] can be oxidized into the corresponding aldonolactone, subsequently hydrolysing into aldonic acid. Compared with conventional coenzyme-dependent enzyme systems, one major advantage of GFOR is the activity of the enzyme without exogenous NADP(H), which is an expensive and quite unstable molecule. Relative to free NADP(H) that displays a half-life of no more than about 24 to 72 h [5], the GFOR-bound

20 NADP(H) seems to be significantly stabilized to chemical decomposition. Furthermore, binding to GFOR increases the apparent molecular mass of NADP(H) to about 41 kDa, compared with 0.7 kDa for free NADP(H). Use of macromolecular coenzyme greatly assists continuous process development. Retention of small molecules such as NADPffl) in immobilized enzyme reactors or ultrafiltration membrane reactors is not easily accomplished [6], and coenzyme losses will add to coenzyme-specific product costs. Especially for low-price high-volume products such as gluconic acid the contribution of coenzyme-specific product costs to total product costs could be critical. For conversion of glucose and fructose two different forms of GFOR could be useful biocatalysts: permeabilized cells of Z. mobilis and cell-free (partially) purified GFOR. Permeabilization of the bacterial cells was shown to yield a stable and apparently fully active biocatalyst [7]. In comparison with permeabilized Zymomonas the advantage of cell-free GFOR could be the utilization of higher enzyme activities per unit reactor volume and thus a significant increase in productivity. However, compared with stability of GFOR encapsulated by the cellular matrix of Z. mobilis, the operational stability of the isolated enzyme was shown to be low, even when a reaction temperature of only 25 ~ was employed [8]. The inactivation reduced the total turnover number for GFOR drastically and hampered efficient use of the isolated enzyme in biochemical reactors [8]. In contrast, a continuous reactor loaded with permeabilized cells was operated at 39~ for several days without apparent decrease in enzyme activity [7]. Interestingly, inactivation of free GFOR seemed to be coupled to substrate turnover because the enzyme was stable under otherwise identical incubation conditions when no catalytic action occurred [8]. Autoinactivation during catalysis (as with glucose oxidase 191) or product-induced inactivation of GFOR are interpretations that seemed consistent with the experimental observations. Our recent studies aimed at elucidating the mechanism of GFOR inactivation in some detail [ 10] and, based on this mechanistic understanding, improving the operational stability of the enzyme during substrate conversion [11]. Some of the results are reviewed here.

2. INACTIVATION OF GFOR DURING SUBSTRATE TURNOVER Judging from the full stability of GFOR at 25 ~ in the absence of turnover for at least 24 h, rapid inactivation during substrate conversion was a surprising result (Figure 1, A). The final extent to which enzyme activity was lost during the reaction was dependent on initial substrate concentration. A correlation between catalytic activity and inactivation was proposed [8] because remaining GFOR activity was stable when conversion of substrate had been completed in batch reaction. At high concentration of glucose and fructose, however, the fast depletion of enzyme activity with time did not allow complete substrate conversion (Figure 1,

21 A), unless fresh enzyme was added to restart the reaction. The inactivation of GFOR was irreversible and did not depend on one particular degree of purification of the enzyme. Hence, a negative effect of contaminating proteins or enzyme activities seemed unlikely.

~-

lOO

"~ ",1=:1

8o

c~

60

r~

40

N

20

DTr~

0

~

o 0

20

40

60

80

100

Incubation time (h) Figure 1. Stability of GFOR during substratc turnover at 25 ~ using 1 M glucose and fructose as substrates (according to ref. [8]). (A) Enzyme activity without addition of DTr. (B) Enzyme activity when 10 mM DTT was added during the reaction. 2 M KOH was used for pH control (pH 6.5 in 0.1 M potassium phosphate buffer). We tried to derive a simple mathematical model that would give a satisfactory description of enzyme inactivation as function of substrate conversion. An apparent kinetic correlation between rate of reaction and rate of enzyme inactivation was found, that is d[S]/dt = k- d[E]/dt, where [S] and [El are substrate and enzyme concentration, t is the reaction time and k is an empirical parameter that lumps the effects that lead to inactivation of GFOR [8]. Although the model is capable of describing well the time-dependent product formation including enzyme inactivation [8,11], the resulting values for k seem consistent only for a limited range of experimental conditions (Table 1). One major drawback of the model is that it cannot account for the marked substrate concentration-dependence of the rate of GFOR inactivation [8], and

further analyses have shown thatindce~ high concentrationsof fi'uctoseratherthan glucose are

responsiblefor stabilizationof G F O R activity(Table I).

Itwas found empiricallythattwo factorsother than high fructoseconcentration crucially

determine the operational stabilityof G F O R during substrateconversion in batch or continuous

reactors [I 1]: thiolprotectionwith dithiothreitol(DTr) in concentrationof 5 to 10 m M being the most efficientreagent and use of weak bases such as carbonate, "Irisor Imidazole rather than hydroxide for p H control, i.e.,neutralizationof gluconic acid (Table 1). Note that D T r was

22 effective only when present during the conversion. Supplementation of DTT to partially inactivated GFOR failed to restore enzyme activity anew, but slowed the loss of remaining activity during further substrate turnover (Figure 1, B). The operational stability of GFOR was good and thus efficient use of enzyme activity for product synthesis could be made when GFOR was retained by a 10-kDa ultrafiltration membrane in a continuously operated tank reactor with constant discharge and 10 mM DTI" and 2 M Tris (as base) were employed (of. Table 1). Under these conditions a stable conversion could be maintained for about 500 h. The corresponding loss of enzyme activity was no more than about 20%. Table 1. Rate-dependent inactivation of GFOR during substrate turnover in a continuous ultrafiltration membrane reactor under different conditions at 25 ~ (according to ref. [11]). i

(2 M base)

Thiol stabilisation Inactivation with DTI' constant (k x 102)

Half-life of GFOR activity (h)

0.1 / 0.1

KOH

10mM

0.180

= 100

0.5 / 0.1

KOH

10 mM

0.100

> 200

1.0 / 1.0

"Iris

10 mM

0.003

> 500

1.0 / 1.0

"Iris

none

0.012

- 200

Substrate

pH control

(M Fru / M Glc)

i

3. M U L T I S T E P MECHANISM OF GFOR INACTIVATION 3.1. Destabilizing effect of the lactone product

The stability of GFOR was determined in the presence of fructose, glucose, sorbitol or sodium gluconate (0.5 M each). Compared with the control, i.e., GFOR in buffer, none of these compounds seemed to negatively affect long-term stability of the enzyme at 25 ~

As

consequence of this result, both the NADPH- and the NADP-form of GFOR are considered to be stable enzyme forms. The effect of glucono-1,5-1actone on GFOR stability was difficultly quantitated because of the rapid hydrolytic decomposition of this compound into gluconic acid in aqueous solution at about neutral pH. Using the alternate product xylono-1,4-1actone (100 mM) that is less prone to hydrolysis than glucono-1,5-1actone (note that xylose is substrate for the oxidizing half reaction of GFOR), time-dependent irreversible losses of enzyme activity were observed [10]. Since pH changes (due to formation of xylonic acid) were very small in this reaction, destabilization of GFOR seems to be caused only by interactions of the enzyme with the lactone product. It is important to note, however, that excess of exogenously added

23 gluconolactonase, i.e., the enzyme that catalyzes the hydrolysis of glucono-1,5-1actone, could not stabilize GFOR activity during the conversion, hence in the presence of lactone. 3.2. Time-course of inactivation The time-dependence of GFOR inactivation provides some insight to the underlying mechanism. It was found that the loss of GFOR activity with reaction time was preceded by a characteristic lag time (Figure 2). The lag time was dependent on fructose concentration and increased with increasing fructose. This result suggested a series mechanism of inactivation in which the native enzyme, N, reacts to give one or several intermediate enzyme forms, [N*], with destabilized conformation relative to N, but fully retained enzyme activity. [N*] inactivate(s) irreversibly to give I, thus a plausible model is: N [N*] -> I

(1)

According to model (1), binding of the lactone product to GFOR would trigger the N -> [N*] transition. It is interesting to note that in addition to high fructose concentrations thiol protection using 1 mM DTI" could significantly increase the lag time prior to the onset of inactivation (Figure 2).

~ v,,,~

100

o

80 6o

~

40

"~ 20 ~ ~,,,4

o

0 0

10 Reaction time (11)

20

Figure 2. Inactivation of GFOR during turnover is preceded by a lag period (according to reL [ 10]). Pure GFOR (0.5 I.tM) was incubated at 30~ in 50 mM phosphate-buffer, pH 6.4, in the absence or presence of substrate: El, no substrate; II 1.5 M substrate, 9 0.5 M substrate, O 0.5 M substrate and 1 mM Dq~F. Control of pH was with 2 M Tris, and the substrate concentration was kept constant using pH-controlled feed of glucose and fructose. 3.3. Involvement of reactive thiols

The effect of D'I~ for stabilizing enzyme activity suggests that reactive thiols may be involved in the inactivation. To ascertain a putative role of thiols for stability of GFOR, chemical modification of the enzyme was carded out using N-ethyl-maleimide fNEM) which is

24 a reagent quite specific for protein sulfhydryl groups. The modification reaction was carded out in the absence and presence of substrates. In each case the time-dependence of enzyme inactivation in the presence of NEM obeyed an apparent fast-order reaction for which a plot of the natural logarithm of residual enzyme activity vs. incubation time is linear. As shown in Figure 3 that compares the inactivation constants for different reaction conditions, GFOR was most rapidly inactivated by NEM when glucose and fructose were present at the same time, hence when catalysis occurred and lactone product was formed. Compared with the results obtained for GFOR in buffer, glucose had a weak destabilizing effect whereas fructose could even afford slight protection to the enzyme. The pH change because of gluconic acid formation was negligible for the reaction time studied, and GFOR was stable in the absence of NEM.

m

~

.o 6

..-~4 O

~,~

2 0 Buffer

Glc

Fruc

Glc/Fruc

Figure 3. Inactivation of GFOR as result of chemical modification of thiols using NEM. GFOR (0.3 gM) was incubated with 20 mM NEM in 50 mM potassium phosphate buffer, pH 7.3, at 25~

in the absence or presence of substrates (1 M each). The inactivation constant is the

negative slope of the plot In (residual enzyme activity) vs. incubation time. The results in Figure 3 implicate exposure and deactivation of reactive thiols in the inactivation mechanism of GFOR, especially during substrate turnover. The exposure of sulfhydryl groups seems to be the result of conformational changes of GFOR, thus probably reflecting the N -> [N*] transition induced by the lactone [10]. Because the reactive form of the sulfhydryl group is the thiolate anion that is favored at high pH, use of strong bases for pH control in line with local pH inhomogeneities due to imperfect mixing of the reactor volume is expected to increase the sensitivity of GFOR to thiol deactivation with subsequent loss of enzyme activity, which was observed (Table 1 and ref. [ 11]).

25

3.4. Aggregation determines the loss of enzyme activity A line of experimental evidence suggested that thiol deactivation is not directly responsible for loss of enzyme activity. First, GFOR previously inactivated by chemical modification using methylmethane-thiosulfonate (MMTS) which is a reagent that forms mixed disulfides with protein thiols in a readily reversed manner could not be re-activated by treatment with excess of DTT. Second, colorimetric titration of reactive thiols in N and [N*] with Ellman's reagent did not lead to immediate loss of enzyme activity [10]. Third, using light scattering as probe of the extent of protein aggregation in samples taken from a typical batch reaction, evidence for a time-dependent increase in formation of high-order protein associates was found. Protein aggregation was irreversible and was correlated kinetically with enzyme inactivation [ 10]. Model (1) should therefore be extended to, N [N*] -> [N*]deact-> I

(2)

where [N*]deact is an intermediate with deactivated thiol residues but retained enzyme activity. [N*]deact is expected to be prone to formation of high-order protein associates and will thus aggregate irreversibly. Recent analysis has shown that thiol inactivation in GFOR proceeds sequentially and, by measuring surface hydrophobicity as function of reaction time, provided an explanation for the high tendency to aggregation of intermediate [N*]deaet [10]. 4. S T A B I L I Z A T I O N OF G F O R Improved performance of isolated GFOR during conversion in a continuously operated ultraf'fltration membrane reactor has been reported recently, and stabilization of enzyme activity was based on the empirical finding that use of DTF together with a weak base for pH control can confer acceptable operational stability to the free enzyme [11 ]. Results of mechanistic studies provide now a clearer explanation to these observations and allow to devise other and probably more efficient strategies to maintain GFOR in an active, non aggregated state I I01. Because fructose and lactone product compete for binding to the NADP-form of GFOR, it seems reasonable to assume that high concentrations of fructose displace gluconolactone from the binary GFOR-gluconolactone complex and thus stabilize N relative to N* during turnover. The difference in the equilibrium dissociation constants [12] of GFOR.fructose (> 1 M) and GFOR.gluconolactone (0.35 mM) indicates that fructose can only poorly compete with gluconolactone for binding to GFOR, and thus a very large excess of fructose would have to be present at all times. Thiol protection on the other hand will stabilize [N*] rather than N. Because the deactivation of exposed thiols did not immediately lead to inactivation, prevention of aggregation seems to be most important in order to avoid GFOR inactivation. Use of bovine serum albumine (BSA; 5 g/L) as anti-aggregation reagent could indeed stabilize well GFOR

26 activity during conversion [6]. The effect of BSA on reducing the aggregation of GFOR could unfortunately not be determined by light scattering because the presence of a large excess of other protein interfered with the measurements. Finally, we discuss whether the results described here can explain why GFOR is stable when encapsulated by the cell matrix of Z. mobilis [7]. The destabilizing interaction of the enzyme with glucono- 1,5-1actone can probably

not be avoided in the eeUular reaction system as such, even though a periplasmic gluconolactonase is known to be present in Z. mobilis. The entrappment of GFOR could however reduce the conformational flexibility of the enzyme in such a way that structural transitions observed with free GFOR are efficiently prevented (of. model (2)). The results with soluble enzyme suggest that avoidance of aggregation of [N*]deaet would be sufficient to prevent the enzyme inactivation.

REFERENCES o

U. H. Chun and P. L. Rogers, Appl. Microbiol. Biotechnol., 29 (1988) 19.

2.

M. Zachariou and R. K. Scopes, J. Bacteriol., 167 (1986) 863.

3.

M. Hardman and R. K. Scopes, Eur. J. Biochem., 173 (1988) 203.

4.

M. Satory, M. Fiidinger, D. Haltrich, K. D. Kulbe, F. Pittner and B. Nidetzky,

o

Biotechnol. Lett., 19 (1997) 1205. Chenault and G. M. Whitesides, Appl. Biochem. Biotechnol. 14 (1987) 147.

6.

K. D. Kulbe and H. Chmiel, Ann. N. Y. Acad. Sci., 542 (1988) A.'!.'!..

7.

B. Rehr, C. Wilhelm and H. Sahm, Appl. Microbiol. Biotechnol. 35 (1991.) 144.

8.

D. Gollhofer, B. Nidetzky, M. Ftirlinger and K. D. Kulbe, Enzyme Microb. Technol.

o

17 (1995) 235. C. Bourdillon, C. Hervagault and D. Thomas, Biotechnol. Bioeng., 27 (1985) 1619.

10.

M. Fiirlinger, D. Haltrich, K. D. Kulbe and B. Nidetzky, Eur. J. Biochem., 251

11.

(1998) 955. B. Nidetzky, M. Ftirlinger, D. Gollhofer, D. Haltrich and K. D. Kulbe, Biotechnol. Bioeng. 53 (1997) 623.

12.

M. Hardman, M. Tsao and R. K. Scopes, Eur. J. Biochem., 205 (1992) 715.


27

Reactor p e r f o r m a n c e under thermal inactivation and temperature optimization with chitin-immobilized lactase A. IUanes, L. Wilson, C. Altamirano and A. Aillapfin Escuela de Ingenieria Bioquimica, Universidad Cat61ica de Valparaiso. PO Box 4059, Valparaiso, Chile

ABSTRACT Kinetics of thermal inactivation of chitin-immobilized lactase (CIL) was studied in the absence and presence of catalytic modulators. Results were in good agreement with two-phase series mechanism of inactivation. The inhibitory product galactose was a positive modulator of enzyme thermal stability, while the opposite held for the substrate lactose. Differential equations were developed and solved to describe packed-bed reactor operation with CIL, considering lactose and galactose modulation. The model was experimentally validated in a laboratory packed-bed reactor. A more stable CIL has been developed and their kinetic (Vm~, Kin, KI) and inactivation parameters (kd have been determined as explicit functions of temperature. A strategy for temperature optimization of CIL reactor operation is presented in which the experimental information gathered has been included.

1. INTRODUCTION Enzyme thermal inactivation, although a well reported phenomenon, has seldom been studied in the presence of catalytic modulators to develop sound models to describe enzyme reactor performance [1-2]. Thermal inactivation in the presence of catalytic modulators has been described already and characterized in terms of modulation factors, defined in equation 2 [3]. Substances that interact with the enzyme during catalysis are potential modulators of its thermal stability and, therefore, modulation factors are relevant parameters for evaluation of reactor performance. Chitin immobilized lactase (CIL) was used as a model enzyme; in this case, both the substrate lactose and the inhibitory product galactose are potential modulators. In fact, galactose was proven to be a positive modulator of CIL stability while lactose had the opposite effect [4]. Kinetic parameters (kcat, Km and KI) and inactivation parameters (kij) are all temperaturedependent. Therefore, temperature explicit functions for all of them will allow to develop a model for reactor temperature optimization.

28 2. THEORETICAL BACKGROUND

2.1. Enzyme inactivation Enzyme thermal inactivation can be conveniently modeled based on a series type mechanism [5]. Reaction kinetics and enzyme inactivation for CIL according to a two-stage series mechanism can be represented by the following scheme: E3P

E3

E3S

E2P

E2

E2S

Tk~(1-n,p)Tk,

kcat'

9 E2+P+Q

Tk,(1-n,s) ~t

*- E I + P + Q

Kp K E: lactase; S: lactose; P: galactose (P); Q: glucose. Considering a f'mal inactive stage, the following expression is obtained:

":[I+A. k,

eo

k,

k2 - k I

(1)

k 2 -k I

Based on the modulation hypothesis, different first-order transition rate constants will exist for each enzyme species (free enzyme E, and secondary complexes ES and EP, equation (1) being applicable to any of them, with: k u = k i (1 - nij) (2) where subscript i refers to the inactivation stage, subscript J denotes the modulator and nu are modulation factors. In addition, the rate equation for lactose hydrolysis with CIL is [4]:

o,x, ....... l kcat "e v(X)

/ X + / m / +1 0 - x)

Km K;-1

K

so

From a material balance of all enzyme species, considering the proposed two-stage series mechanism, and the rate equation for lactose hydrolysis, enzyme inactivation under reactive conditions will be described by: de

= e - k 1 9[

dt

(1 - A). exp (- k I 9t)-[1 - a(X)-N I(X)] exp(- k 1 9t)+ (k I 9A (it 2 - k l ) )- [exp (- k , . t ) - exp (- k 2 9t)] +

A?k2"[exp(-k!'t),exP(,k2"t)]?[!-a(X)'Nz(X)]

]

+ (k 2 - k , ) e x p (-k, . t ) + k I 9A. [exp ( - k , . t ) - e x p ( - k 2 .t)]] where:

(4)

29 NI(X ) : n l s +nip .K m 9X / [ K p ( I - X ) ] "

N2(X ) =n2s +n2p . g m . X / [ K p ( I - X ) ]

2.2. Reactor performance under thermal inactivation Batch reactor performance will be described by the simultaneous resolution of equation (5), which represents the material balance over the reactor [6], and equation (4)" dX _ kc, , 9e(t).a(X) (5) dt so Packed-bed reactor performance, considering pseudo-steady-state and plug flow regime, will be described by the simultaneous resolution of equations (6), which represents the material balance over the reactor [6], and equation (4): dX e ( t , z ) . c r ( X ) - e - S

dz

F-s o

(6) 2.3. Temperature optimization Temperature has two opposite effects, increasing both reaction and inactivation rates. Kinetic and inactivation parameters are functions of temperature; therefore, reactor temperature can be optimized provided explicit functions for all of them are validated and a sound objective function is defined. Arrhenius-type functions can be proposed for all rate constants: kc, t =kclt,0 9e x p [ ' E T ]

;

Vmx=Vmx,0-exP[RET]

kij = ku'~ exp R . T ]

Affinity parameters (Km and Kp) can be considered as equilibrium constants and, therefore, thermodynamic correlations can be proposed: K

m

= K~0.exp R'T

;

K p = Kp, 0 - e x p AH~

LR.T

Such expressions can be introduced in equations (4), (5) and (6) to derive temperature explicit equations that will optimize reactor temperature. Specific productivity (the amount of product produced per unit mass of enzyme and unit time) may be used as a preliminary objective function. A more elaborated cost function will be the ultimate to consider.

3. METHODOLOGY 3.1. Materials Enzyme was a partially purified lactase (I3-D galactoside galactohydrolase, E.C.3.2.1.23) from Kluyveromyces marxianus var marxianus NRLL u produced and immobilized on activated chitin as already described [4]. A more stable CIL was used for temperature optimization studies, obtained by optimizing immobilization in terms of enzyme and glutaraldehyde to chitin ratios and particle size and including an acid pretreatment step. USP lactose was used for reactor runs. All other reagents were analytical grade fro m Merck (Darmstadt, Germany) or Sigma (Saint Louis, Mo, USA). One international unit (IU) of CIL

30 activity was defined as the amount of enzyme that hydrolyses one l~aol of lactose per minute at pH 6.6 and 40 ~ C in a 200 g/L solution. 3.2. Kinetics of enzyme inactivation Experiments in the absence and presence of modulators were done in batch reactors with temperature control. Air space was kept to a minimum to avoid enzyme inactivation. For experiments in the presence of lactose and galactose, medium replacement was done every eight hours. Lactose and galactose concentrations were 10"Kmand 10"Kl, to ensure equilibrium displacement to the corresponding enzyme complexes. In the former case, conditions were established to keep the enzyme saturated with substrate and enzyme to substrate ratio was low enough to keep conversion below 1%. Inactivation parameters were detennined by non-linear regression (Statistica, Stat Soft Inc, Tulsa, OK, USA) of experimental data to equation (1); modulation factors were determined from equation (2). 3.3. Reactor modelling and packed-bed reactor operation Model for reactor operation was obtained by solving equations [4] and [6] (or [5]), using modified Euler method. Computer resolution was done in Visual Basic 3.0 for Windows. Model was experimentally tested in a continuous packed-bed reactor with CIL. Reactor was a jacketed Pyrex-glass column, 2 cm in internal diameter and 18 cm in height, with catalyst bed height of 16 cm. Pseudo-steady-state was obtained after four to five residence times. 3.4. Temperature optimization Kinetic parameters at different temperatures were determined by linear regression of initial rate data in Lineweaver-Burk plots. Concentration range was 30 to 585 mM for lactose and for 30 to 180 mM for galactose. Inactivation parameters at different temperatures, in the absence of modulators, were determined as previously described. Temperature ranged from 15 to 40 ~ Temperature correlations for affinity parameters and rate constants were validated in Arrhenius-type plots.

4. RESULTS AND DISCUSSION 4.1. Kinetics of enzyme inactivation Kinetics of CIL inactivation at 28 ~ in the absence of modulators and in the presence of saturating concentrations of lactose and galactose is presented in Figure 1. Results were well represented by a two-stage series type mechanism of inactivation for all enzyme species, being a much better fit than one-stage first-order kinetics. From equation (1), shown by solid lines in Figure 2, inactivation parameters for the free enzyme E and the complexes ES and EP were determined by non-linear regression, as shown in Table 1. G-alactose was a positive modulator, modulation factors being positive in both stages. Lactase was a negative modulator in both stages of CIL inactivation. There are no reports on the particular effect of substrate and products on yeast lactase inactivation at operation temperatures. However, a two-stage mechanism of inactivation was proposed for the inactivation of yeast lactase in the presence of milk [7]. Experiments, however, were conducted at variable and non-saturating concentrations of sugars, obscuring conclusions. Negative modulation by substrate has been reported for glucose isomerase [8] and attributed to quaternary structure dissociation promoted by the substrate, which might be the case for lactose in K.marxianus lactase. Product modulation has

31 seldom been reported: in the case of penicillin acylase, strong protection was exerted by 6aminopenicillanic acid and negative modulation by phenylacetic acid [3]. Product protection can be explained by increasing hydrophobicity in the active site and by favouring a rigid configuration [9]. 100-

!. h

X X g

9

9

j.J

2O

0 0

|

!

5

i0

,

L

i

|

1.5

2O

30

35

40

415

TIE (Imam)

Figure 1. Kinetics of thermal inactivation of CIL at 28 ~ (X): phosphate buffer; (O): 545 mM lactose in buffer; (A): 870 mM galactose in buffer. Table 1 Inactivation parameters for CIL at 28 ~ according to a two-stage series-type mechanism Condition, Enzyme Specie Buffer E + galactose EP + lactose ES i i,

A

nlj

g~

k]. 102 (h'!) 7.74

k2"102 (hl ) 7.75

0.77

6.79

2.56

0.77

0.123

0.669

0.992

8.27

10.97

0.77

-0.068

-0.416

0.921

n2j

0.974

l, .,..

4.2. Reactor modelling and packed-bed reactor operation Packed-bed reactor operation was designed for an initial substrate conversion of 0.9. Reactor performance is presented in Figure 2. As seen, the model describes reactor performance very well and only in the first part conversion was somewhat higher than predicted. At prolonged times the model departs somewhat, but this is out of the region in which the inactivation mechanism was validated and is of no practical significance. Model not considering modulation factors was a very poor representation of reactor performance, which validates modulation factors as important parameters for proper reactor design. Enzyme decay profiles through catalyst bed height (data not shown) were consistent with the model, inactivation rate being lower near reactor outlet where protecting galactose concentration was higher.

32

0.$ o.B O,T

9

O.II

O.a i OA 0.3 0.2

9

0,1 0

.

"

l

0

10

20

! ....

30

40

Time (houri)

Figure 2. Continuous packed-bed reactor performance with CIL at 28 ~ a) model considering modulation; b) model not considering modulation; (e) experimental data. 4.3. Temperature optimization

Kinetics of CIL inactivation at different temperatures in the absence of modulators is presented in Figure 3. Results were in all cases well represented by a two-stage series-type mechanism of inactivation. As seen from figures 2 and 1, this CIL was more stable than the used formerly. Values for kinetic and inactivation parameters in the absence of modulators, at different temperatures, are presented in Table 2. Table 2 Kinetic and inactivation parameters of CIL at different temperatures Temperature (~ Km(m Vmax * 116.1 18.3 15 164.6 21.9 20 22.5 186.6 23.9 25 211.0 24.8 27.5 241.8 27.5 ....30 ..... 3il.7 31.7 .... 32.5 371.0 32.6 35 383.0 37.9 37.5 415.0 38.3 * ttmoles glucose/min.g catalyst ,

....

.

.

.

.

.

Kp (raM) 37.0 56.1 69.5 80.1

k]. 102 (h "l) k2.102 (h"z)

109.7 135.1

7.85 8.05 15.20 21.19

0.29 1.61 2.02 2.99

200.8

26.76

4.73

A

.

.

0.75 0.72 0.71 0.31 .

.

.

.

.

0.30

....

Thermodynamic correlations for temperature dependence were validated for Km and Kp. Arrhenius-type correlations were validated for kcat (Vmax)and ki, the following expressions being obtained: K m = 1.92" 10 .6 .exp R - T J

K p = 3.76-

.exp R . T J

33

Vmx

kI

--

fJ

F- 10.423

9.59.109

= 2.50-1013-

exp

I"-R--qr21"16]

k2

= 5.73- 10u

.I-" 38.20]

Correlation coefficients were in all cases around 0.99. A was not a defmite function of temperature, being rather constant at temperatures over and below 30 ~ However different values were obtained for both ranges: A = 0.73 T < 30 A = 0.31 T > 30 Batch reactor performance, not considering modulation effects (all n i j = 0), was simulated to determine optimum temperature, by solving equations (4) and (5) with temperature explicit functions for kinetic and inactivation parameters. Optimum temperature was determined considering specific productivity as objective function. Results are presented in Figure 4, temperature optimum being 26 ~ Kinetics on CIL inactivation in the presence of saturating concentrations of lactose and galactose is now underway to determine temperature explicit functions for modulation factors.

Figure 3. Kinetics of CIL inactivation in the absence of modulators:m: 25 ~ 27.5 ~ A: 30 ~ 4,: 32.5 ~ O: 37 ~

Figure 4. Temperature optimization for batch reactor with CIL

5. CONCLUSIONS .

CIL thermal inactivation was well described by a model based on a two-stage series mechanism, lactose and galactose being modulators. 9 A model for continuous packed-bed reactor performance with CIL, based on modulated thermal inactivation, was experimentally validated. 9 Explicit temperature functions were validated for all kinetic parameters and inactivation parameters in the absence of modulators

34 Batch reactor operation with CIL, not considering modulation effects, was simulated to determine optimum temperature. Results are underway to determine temperature explicit functions for modulation factors, to be included in the optimization model

REFERENCES 1. A. IUanes, M.E. Zufdga, Contreras, S. and A. Guerrero, Bioproc. Eng., 7 (1992) 199. 2. J.Y. Houng, H.Y. Yu and K.C. Chen. Biotechnol. Bioeng., 41 (1993) 451. 3. A. Illanes, C. Altamirano, M.E. Zufiiga, Bioteclmol. Bioeng., 50 (1996) 609. 4. A. Illanes, C. Altamirano, A. Aillap~n, G. Tomasello and M.E. Zuffiga, Enzyme Microb. Teclmol. (1998) (accepted for publication). 5. A. Sadana and J.P.Henley, Enzyme Microb. Technol. 7 (1985) 50. 6. E. CJalindo, O.T. Ramirez (eds.), Advances in Bioprocess Engineering, pp 467-472, Kluwer Academic Publ., Dordrecht, 1994. 7. R. Mahoney and T. Wilder. J. Dairy Res., 55 (1988) 423. 8. W. van den Tweel, A. Harder and R. Buitelaar (eds.), Stability and Stabilization of Enzymes, pp 111-131, Elsevier, Amsterdam, 1993. 9. I. Villaume, D.Thomas and M. Legoy. Enzyme Microb. Technol. 12 (1990) 506.

Nomenclature A molar activity ratio between initial and intermediate stage e enzyme activity e0 initial enzyme activity F actose feed flow-rate k~t catalytic rate constant k~ first-order transition rate constants Km Michaelis constant for lactose Kp inhibition constant by galactose R universal gas constant R 2 determination coefficient s lactose concentration so initial (inlet) lactose concentration S cross section of reactor t time T temperature v reaction rate Vma~maximum reaction rate ( = l~t. e) X substrate conversion ( = [s0-s]/s0) z axial dimension in reactor s liquid volume fraction in catalyst bed

Acknowledgements This work was supported by Grant 1971029 from FONDECYT, Chile


35

Fast in-situ-characterisation o f b i o c a t a l y s t s l o n g - t e r m stability M. Boy, A. Dominik, H. Voss SFB Biocatalysis, Institute of Biotechnology, Graz University of Technology, Petersgasse 12, A-8010 Graz, Austria

1. INTRODUCTION A major bottleneck in developing industrial biocatalysts is the lack of a method to determine the catalyst process stability quickly and efficiently [1]. In particular, the determination of these properties is difficult, often inaccurate and expensive because highly active, selective and stable biocatalysts should be achieved [2,3,4]. To solve this problem, a new method for fast in-situ-characterisation of activity and longterm stability of enzyme carrier catalysts has been developed. To accelerate the process of ageing the temperature is increased continuously during the experiment. On the basis of the proposed dynamic model it is possible to determine information pertaining to enzyme kinetics, selectivity, interaction between carrier, enzyme and substrate as well as information regarding the stability of the enzyme carrier complex, including reversible and irreversible activation and inactivation parameters. In contrast to common, isothermal techniques these parameters can be estimated for the whole temperature range in one short experiment. Working under industrial process conditions allows one to determine the whole relevant information necessary to characterise the (carrier- bound) biocatalyst. This knowledge allows the design and optimisation ofbiocatalytic industrial processes within a short period of time [5,6]. 2. T H E O R Y

To demonstrate the method of fast in-situ-characterisation of activity and long-term stability of enzyme carrier catalysts a single step deactivation with reversible deactivation of the native and active enzyme F_~ is presented. The equilibrium constant K gives the ratio of reversible deactivated enzyme ED. and F_~. The sum of Es and ED, deactivates irreversibly to ED with the deactivation rate coefficient k d.

~E~* In mathematical terms the mechanism can be described by a mass balance and a differential equation:

[Eo,]+[E,]

(1)

36

.[e]

dt

At the beginning of the process the whole enzyme is in the native form. The deactivation rate coefficient kd and the equilibrium constant K are temperature dependent according to the Arrhenius law: Eo

ka = ka.| -e R.r

(3)

[E~

(4)

[E I Ed.r,,v

K=K|

R.r (5) In addition to the reversible and irreversible enzyme deactivation, the biochemical reaction is also a function of temperature. In general, the biochemical reaction rate can be expressed in separable terms of temperature T, substrate concentrations ci and enzyme concentration Cen~me(eq. (6)). A frequently used expression for the kinetics of enzymatic reactions is the Michaelis-Menten-equation. It is possible to define the maximum reaction rate rm~x as a product of rate coefficient for the enzyme-catalysed reaction k~ and the native enzyme concentration [Es], as demonstrated in equation (7). In the case of substrate saturation equation (7) simplifies to (8). Adequate substrate concentrations and flow rates were chosen during the experiments to fulfil this assumption.

r= f ( r , ci,ce~m)= f, (r). f2 (ci). f3 (Cenzym )

IS] "K. [S--------] + __kr

r-r

.[EN ]

(6) (7)

The definition of the specific enzymatic activity "a" [mol/g.s] in equation (9) and (8) together with (10) for the temperature dependence of the rate coefficient k~ finally leads to equation (11): (9) a--~rmax

[E]o

k r = h| "e R.T = A

(10)

[E~] -e-L [EN] a = A. [El0' = A.~-e "'r 9[El0

(11)

The set of equations (1)-(5) and (11) allows the calculation of the time course of the specific enzymatic activity dependent on temperature. 3. MATERIALS AND METHDOS 3.1. Enzymes and chemicals The goal of the present study is to demonstrate the application of the newly developed method for characterising free and carrier-bound enzymes during the development of carrierbound biocatalysts. A lipase of Mucor miehei was examined both as a free enzyme and immobilised on different carriers. Novozyme 388 (Novo Nordisk) was used as a free enzyme. Additionally, a carrier bound enzyme of this source is commercial available from Novo Nordisk (Lipozyme IM). Support material is a highly porous anion exchanger resin of the

37 phenolic type (particle size 0,2-0,6mm). As a parallel enzyme, we immobilised Novozyme 388 adsorptive on a hydrophobic organopolysiloxane-carrier (DELOXAN ~, propylsiloxane modified ethylenediaminopolysiloxane, particle size 0,2-0,4 mm, Degussa AG). The carrier was saturated with 170 ~gprotein/gdry carrier with a yield of 80 %. A highly active carrier-bound enzyme with 30000 IU/g~y c~meraccording to the tributyrine standard assay [7] is achieved. As a model reaction, the hydrolysis of triglycerides was chosen. Tributyrine (Fluka) was emulsified in water according to the literature[7]. 3.2. Experimental conditions

Experiments with the free enzyme were carried out in jacket-cooled, flat-flange standard glass reaction vessels with 2000 ml total volume (Schott). A fed-batch process was used. The initial reaction volume was 200 ml. The feed rate depends on the enzymatic activity. Substrate saturation should be achieved during the whole experiment. The feed rate to fulfil this assumption, the used enzyme amount and the applied temperature program are presented in Table 1. The feed rate was attained by a gear pump (MC-Z, Ismatec). The substrate solution was stored under stirring on a magnetic stirrer in a ice bath. The speed of a 3-blade-impeller stirrer with 70 mm diameter was 300 rpm. Temperature control was reached by a computercontrolled external heating bath. pH-value was kept constant at pH 8 using an autotitrator (Schott Titroline). We used 0,1 mol/1 potassium hydroxide solution as a titrant. For immobilised enzymes continuous processes in a jacket-cooled, fiat-flange standard glass reaction vessel with 500 ml total volume (Schott) and 250 ml working volume were carried out. 300 rpm stirrer speed of the used 3-blade-impeller stirrer with 70 mm diameter was found to be enough to neglect external mass transfer resistance. The flow rate depends both on enzymatic activity and the amount of enzyme amount used. Flow rates, enzyme amount and temperature programs are presented in Table 1. The feed was attained by a gear pump (MC-Z, Ismatec). The working volume was held constant by pumping out the excess solution by a vacuum filtration aspirator pump through a glass frit. Due to higher activity, 1 mol/1 potassium hydroxide solution was used as titrant to keep the pH-value constant at pH 8. The time course of titrant necessary to neutralise the fatty acids produced is automatically recorded, corrected and converted into activity. Table 1 Experimenta! conditions Carrier Reactor Feed rate Enzyme Enzyme Enzyme material type [ml/h] amount preparation source fed-batch 200 14 pl Free enzyme Novozyme 388 528 0,015g ~ = Immobilised Novozyme organopoly- continuous siloxane stirredtank enzyme 388 Immobilised LipozymeIM enzyme

anion

reactor continuous

resin

reactor

exchanger stirredtank

214

0,499gd~~

Temperature Program 10-70 ~ in 7 h const. 6,4 ~ for 2 h, 6,4-66,4 in 7h 10-70 ~ in 7 h

38 4. PARAMETER ESTIMATION x

measured

,,

c a l c u l a t e d specific activity

........

native e n z y m e c o n c e n t r a t i o n

..... 25000

,

'

temperature ,

,

9

. . . . . . . . . . . . . . . . . . . . . . . 9 .-

E" .v

specific activity

20000.

.

,,

i ~

, ~

,

,

-

s

"-.

0,03

x

10000,

o,o2

O

5000

O

0

(D

0,05 ~ 0,04~

~

15000,

~r

0,06 %; 70

l !

s

~ > .n

|

0

~

~

~

.

.

~ .

0,00

10

O 50 .o .

-~-

40

c:

30~

8 8

0,01 E .

6 0 ,..., == =

"

|

20E

10F-

N

c "'

Time [h]

Figure 1. Cleavage of tributyrine by Novozyme 388 immobilised on organopolysiloxane. Figure 1 shows the measured and simulated specific activity, the temperature profile and the computed concentration of native enzyme of the lipase Novozyme 388 immobilised on organopolysiloxane. In order to determine the temperature dependent coefficients of equations (3), (5) and (11), the measured specific activity is evaluated in sections. In the first time interval ( 2 - 4 . 5 h) it is possible to neglect reversible and irreversible deactivation of the enzyme because of the low temperature (6-26~ and short operation time. This assumption leads to equation (12). If the enzyme is saturated by a high substrate concentration or by a high flow rate, equation (11) finally simplifies to (13): [EN]/[E]o=I.

(12)

E.

a = A| .e R.r (13) By presenting the specific activity during this time interval in an Arrhenius plot, the activation coefficients A~ and E=could be determined. With increasing temperature the fraction of reversibly deactivated enzyme increases. Irreversible deactivation plays a minor role because of moderate temperature and operation time. A mass balance of the enzyme leads to equation (14). Combining equation (4) and (14) with (11 ) results in equation (15) to determine the equilibrium constant K:

(,4)

[E]o = _ E....~a

K

R.r -1 (15) a During the time interval between 5 and 6.2 h the equilibrium constant K could be calculated from the measured specific activity "a" and from the previously determined activation coefficients A~ and E,. An Arrhenius plot of K leads to the reversible deactivation coefficients K~ and Er At the end of the temperature program, the fraction of irreversibly denaturated enzyme increases. When combining equations (1), (2), (4) and (11), one finally obtains a formula allowing an evaluation of the coefficients of irreversible deactivation. =

A|

39

ka

=

/

K

I+K

.

Ear~ , _ R

.

d 1 T___.~I dt a

da dt

(16)

The equilibrium constant K is temperature dependent according to equation (5). During the time interval between 7 and 8 h k d, could be calculated from the measured specific activity "a" and the previously determined activation and reversible deactivation coefficients. The graphical presentation of kd in an Arrhenius plot leads to the determination of the irreversible deactivation coefficients k~| and Ea, irr.

5. RESULTS AND DISCUSSION The presented method was used to determine the influence of immobilisation on the examined lipase. For all experiments carried out to determine the influence of immobilisation, a first-order deactivation mechanism with a reversible deactivation was found to be the better fitting compared to a first-order deactivation mechanism without reversible deactivation. It has to be stressed in general, that it is not the claim of the presented method to proof a mechanism, but to find a model with a set of parameters which fits the experimental data well and can be used to calculate enzyme characteristics and process strategies. The coefficients of the different immobilised and free enzymes are presented in Table 2. Half-life time (Fig. 2) and "Turn-over-Number" (TON) (Fig. 3) were calculated. TON expresses how much substrate can be converted until a defined residual activity is reached.

TON =

~a(t)dt (17) 0 For example, the residual activity can be given by a minimum conversion with a fixed amount of enzyme used in a reactor. Table 2 Activation and deactivation coefficients of free and carrier-bound M u c o r m i h e i lipase. Activation coefficients Deactivation coefficients Reversible Irreversible A| E, K,o E~ k,t| E~i, [mol/g.s] [kJ/mol] [.] [kJ/mol] [I/s] [kJ/mol] Novozyme388 175,5 32,53 3,6561019 120,52 5,2311033 231,79 Novozyme 388/ 13,45 26,74 7,5671021 134,44 6,733.103s 262,66 organopolysiloxane Lipozyme IM 16~5104 60,19 1,3631019 114~50 1,6171049 328,10 Figure 2 clearly shows the big advantage of the presented method for fast in-situcharacterisation of activity and long-term stability. The half-life time ofLipozyme IM is about 140 days. This period is necessary to measure the half-life time in an isothermal experiment. To determine the half-life times of Lipozyme IM in the whole temperature range from 10 to 70 ~ in 10~ intervals using the common isothermal techniques would take about 3700 years (of course, even during 140 days other phenomena beside thermal deactivation (e.g. microbial growth) may be responsible for shorter half-life times). To get the same information using the new method one 12 h-experiment is enough. Figure 3 shows the TON of free and immobilised M u c o r m i h e i lipase. 300 U/g were defined as minimum specific activity. Lipozyme IM reaches this lower level only in the

40 temperature range between 28 and 54~ It has a maximum in the temperature course of the TON. This temperature is an optimum operation point if the costs of the enzyme plays an important role in the overall process. 101~

Lipozyme IM [ ....... Novozyme 388

108 ~" 106 11)

E

....,

"', ,",,"-',-',"~='~"~~~...,..~."

o

E

388

1016

101'

,.:..

-~- 101z Z O 101~ t" 108 "~ Os

104 102

~_. 100

Lipozyme IM ....... Novozyme 388 ~:.~:. :... immob. Novozyme 388 ...,.

~":':.-..

0 1 E =.1 04

"r" 10 .2

lo

2o

3o

40

so

60

A

a0

Temperature [*C] Figure 2. Half-life time of free and carrierbound Mucor mihei lipase.

Z 102

0

I--

:.. 9.

9

0

,

,

,

.

,

.

,

.

,

.

|

,

!

,,.,

,

"'|'.."

10 20 30 40 50 60 70 80 90

Temperature [*C] Figure 3. Turn-over-Number of free and cartier-bound Mucor mihei lipase.

6. CONCLUSIONS The new developed method for fast in-situ-characterisation of activity and long-term stability of enzyme carder catalysts enables researchers, producers and users of enzymes and carrier-bound enzymes to characterise biocatalysts quickly and, therefore, accelerates the process of choosing and developing biocatalysts and carriers. Commercial available biocatalysts could be very easy characterised for a special problem. A quick process optimisation regarding process conditions, reaction media and finally the costs of the catalyst thus becomes possible. REFERENCES

1. Cabral, J.M.S., Best, D., Boron, L., Tramper, J.: Applied Biocatalysis, Harwood publ., Chur 1994. 2. Mozhaev, V.V.: Mechanism-based strategies for protein thermostabilization. TIBTECH Vol. 11, 1993. 3. Gupta, M.N. (ed.): Thermostability of enzymes. Springer, Berlin 1993. 4. Vrabel, P., Polakovic, M., Stefuca, V., Bales, V.: Analysis of mechanism and kinetics of thermal inactivation of enzymes: Evaluation of multitemperature data applied to inactivation of yeast invertase. Enzyme and Microbial Technology 20, 348-354, 1997. 5. VoB, H.: DP 195 46 192.4 1995. 6. Boy, M., Dominik, A., Vog, H.: Fast determination of biocatalysts process stability. Process Biochemistry (submitted). 7. Weber, K. H., Stecher, H., Faber, K.: Some Properties of Commercially Available Crude Lipase Preparations. In: Preparitive Biotransformations, Editor: S.M. Roberts, Wiley, New York, 1995, S. 5:2.1-5:2.10.

Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Halling (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

Chemical

and t h e r m a l

stability

41

of ferulic

acid (feruloyl)

esterases

from

AspergiUus. C.B. Faulds ~, F.O. Aiiwan ~, R.P. de Vries b, Williamson ~.

ILW. Pickersgill ~, J. Visser b and G.

aBiochemistry Department, Institute of Food Research, Norwich Research Park, Colney, Norwich NR4 7UA, UK. bMGIM, Wageningen Agricultural University, Dreijenlaan 2, 6702 HA, Wageningen, The Netherlands CFood Macromolecular Science Department, Institute of Food Research, Early Gate, Reading RG6 6BZ, UK. ABSTRACT Ferulic acid (feruloyl) esterases are a novel subclass of the carboxylic acid esterases, and are able to hydrolyze esterified phenolic acids from plant cell walls. The active sites of feruloyl esterases show high homology with the serine active sites of lipases, and a model of a feruloyl esterase (FAE-III) from Aspergillus niger was derived based on the structure of four lipases. Stability of this feruloyl esterase was compared to a feruloyl esterase from A. tubingensis using chemical and thermal denaturation. These enzymes show a 92.5% identity in their amino acid sequences. The A. tubingensis feruloyl esterase is more sensitive to degradation than the A. niger FAE-III. These enzymes have potential in the food and agro-chemical industries.

1. INTRODUCTION Ferulic acid (feruloyl) esterases (FAE) are a subclass of the carboxylic acid esterases (EC 3.1.1.1) and are able to hydrolyze esterified phenolic acid from plant cell-walls. These types of linkages are found in a wide variety of cell-walls [1 ], and the enzymes are secreted by the microorganism to facilitate total breakdown of the cell-wall polymers. Feruloyl esterases have been purified from bacterial and fungal sources [e.g. 2-8], and are induced during growth on lignocellulosic material. They differ from other cell-wall degrading esterases in their specificity for cinnamoyl esters (as shown by Km values between 10-6 and 2 x 10.5 M for some substrates). Five genes encoding enzymes with feruloyl esterase activity have recently been sequenced [5, 810], showing very little conserved identity, apart from the amino acids adjacent to the active site serine. Modification of nucleophilic serine residues has indicated a catalytic role for serine in these esterases [4], and the GXSXG motif common in lipases together with aspartate and histidine residues which make up the catalytic triad appear to be conserved. From its primary sequence, a model of the tertiary structure of FAE-III from Aspergillus niger was based on

42 comparisons with lipases, in open and closed lid conformations, from Rhizopus delemar, Rhizomucor miehei, Humicola lanuginosa and Penicillium camembertii ([10], Aliwan, Faulds, Pickersgill & Williamson, unpublished results). Aspergillus species have been shown to produce a number of feruloyl esterases, but the gene encoding one (faeA) has been sequenced to date from two species, A. niger and A. tubingensis [ 10]. This gene shows a high degree of nucleotide sequence identity (87.1%) between the two strains and contains one conserved intron. Translation of the DNA sequence of both genes resulted in amino acid sequences of 281 and 280 for A. niger and A. tubingensis, respectively, each containing a conserved signal peptide of 21 amino acids. The esterases from A. niger (FAE-III) and A. tubingensis contain a single putative N-glycosylation site, and an alignment of their amino acid sequence showed a 92.5% identity, with a calculated molecular mass of 28314 and 28261, respectively [ 10]. In this report, we compare the A. niger and A. tubingensis gene products over-expressed in a protease-deficient strain of A. niger and subsequently purified, and discuss possible reasons for one form to be more resistant to proteolysis. 2. MATERIALS AND METHODS Fungal strains and culturing. A. niger NW154 containing a prtF mutation (pyrA6 prtF28) and A. tubingensis NW241 (pyrA2 fwnA1) have been described previously [10]. Functional constructs containing the faeA genes were transformed into either the A. niger NW154 prtF mutant or A. tubingensis NW241 [10]. Purified esterase was obtained from A. niger NW154 transformants. Cultures were grown on 1% (w/v) oat spelt xylan containing minimal media for up to 6 days at 25~ Supernatants were collected by filtering through miracloth. Purification and characterization of feruloyl esterases. Recombinant esterase was purified from culture supernatants after 3 days of growth on oat spelt xylan as described previously [3]. Molecular masses were determined by SDS-PAGE (Novex 10% Tricine gels) calibrated with known standards and by electrospray mass spectroscopy on a Micromass Platform positive spray mass spectrophotometer. Western analysis of culture supernatant samples was performed using polyclonal antibodies raised in mice against purified FAE-III. Enzyme assays:. Feruloyl esterase activity in culture supernatants and throughout the purification was determined using methyl ferulate (50 lxM) as a substrate at pH 6.0 and 37~ [3]. One unit of enzyme activity is the amount of enzyme which released 1 lxmol of product per minute under these conditions. Chemical and thermal inactivation of FAE-III were Chemical and thermal inactivation. determined as previously described [ 11]. Determination of N-terminal sequences. Amino acid sequences were determined from electroblots of an SDS-PAGE separation of purified A. tubingensis FAE on an Applied Biosystems Procise Sequencer. Modeling of FAE-III. The A. niger FAE-III model was generated using the program MODELLER [ 12], with four aligned lipases giving the input protein scaffold and the sequence alignment giving the relationship between this scaffold and the A. niger FAE-III sequence. The

43 molecules were displayed on a Silicon Graphics Indigo 2 with the program O [13] and the protein cartoon was generated using MOLSCRIPT [ 14]. 3. RESULTS AND DISCUSSION Over-expressing transformants of both A. niger and A. tubingensis, containing either thefaeA gene from A. niger or the faeA gene from A. tubingensis, were grown for 6 days, culture supernatants concentrated 5-fold and subjected to Western analysis (Figure 1). On SDSPAGE, multiple bands were visible (results not shown), ranging from the mature protein at 36 kDa (approx. 30% of total protein for the A. tubingensis FAE and 95% for the A. niger FAEIII) to smaller bands down to 22 kDa. Upon Western blot analysis, it was apparent that only the A. tubingensis enzyme gave a multiple banding pattern, irrespective of the host strain of Aspergillus. The recombinant esterase from A. niger (FAE-III) showed partial degradation when expressed in A. tubingensis, only upon Western analysis (Figure 1, lanes 1 and 3). Thus, it appears that the A. niger feruloyl esterase is less sensitive towards degradation by natural Aspergilli proteases.

Figure 1. Western blot degradation patterns for feruloyl esterases from Aspergillus. Lane 1, transformant NW154::pIM3207.7 (A. niger faeA in A. niger); lane 2, NW154::pIM3208.5 (A. tubingensis faeA in A. nige0; lane 3, NW241::pIM3207.7 (A. niger faeA in A. tubingensis); lane 4, NW24 l::pIM3208.5 (A. tubingensis faeA in A. tubingensis); lane 5, purified FAE-III from A. niger [3]. Cultures grown on oat spelt xylan for 6 days. With courtesy of Appl. Environ. Microbiol. [ 10]. The recombinant esterase from A. tubingensis was purified by ammonium sulphate precipitation, hydrophobic interaction chromatography and anion-exchange chromatography. SDS-PAGE of each stage of the purification still resulted in the presence of multiple bands for the denatured enzyme. Mass spectroscopy of the enzyme aider anion-exchange revealed a single mass of 29724, compared to the A. niger FAE-III mass of 29756. Thus, the feruloyl esterase from A. tubingensis was apparently cleaved at certain points on the protein, but this did not appear to alter the native molecular mass. From Figure 1, there appears to be a number of these cleavage sites. N-terminal sequencing of the first five residues of the mature A. tubingensis protein (36 kDa) and the second largest band on SDS-PAGE (26 kDa) gave identical results, presumably the lower band is produced by C-terminal truncation of the mature protein. A well separated band at 22 kDa was also N-terminally sequenced (5 residues). This band contained two sequences of equal signal: one corresponding to the N-terminal sequence (A S T Q G); the other a sequence X

44 E V H, which corresponds to a sequence 95 residues in from the N-terminus (Figure 2). In the protein sequence for the A. tubingensis esterase, X corresponds to cysteine. Figure 2 shows an alignment of the two Aspergillus feruloyl esterase sequences. There are 15 amino acid changes between the two sequences, and one extra residue at position 185. The degraded band sequence at 22 kDa encompasses one of these amino acid changes at the point of cleavage: A. niger (D *C E V H) and A. tubingensis (S *C E V H). Thus, it appears that an aspartic acid residue instead of a serine at position 93 may contribute to the proteolytic stability of FAE-III. ANFAE ATFAE

ASTQC ASTQC

ANFAE ATFAE

ISEDL YNRLV ISEDL YSRLV

EMATI EMATI

SQAAYADLCN SQAAY ADLCN

IYNAQ IYNSQ

45 45

TDING WILRD DTSKE TDING WILRD DSSKE

IITVF RGTGS DTNLQ LDTNY TLTPF DTLPQ IITVF RGTGS DTNLQ LDTNY TLTPF DTLPQ

90 90

ANFAE ATFAE

CNDCE VHGGY YIGWI CNSCE VHGGY YIGWI

S V Q D Q V E S L V KQQAS Q Y P D Y A L T V T G H S L G SVQDQ VESLV QQQVS QFPDY / ~ ' r V T GHSLG

135 135

ANFAE ATFAE

ASMAALTAAQ ASLAA LTAAQ

ANFAE ATFAE

P E T T Q Y F R V T HSND_G IPNLP P A D E G Y A H G G V E Y W S P D T T Q Y F R V T HAND_G IPNLP P A D E G Y A H G V V E Y W S

ANFAE ATFAE

VCTGD EVQCC VCTGD EVQCC

LSATY DNVRL YTFGE LSATY DNIRL YTFGE

IPSTI IPSTI

PRSGN QAFAS PRS N Q A F A S

E A Q G G Q G V N D A H _ T T Y FGMTS E A Q G G Q G V N N AH_TTY FGMTS

GACTW GHCTW

IKGEK IKGEK

YMNDA YMNDA

FQVSS FQASS

180 179

VDPYS AQNTF VDPYS AQNTF

225 224 260 259

Figure 2. Alignment of the deduced amino acid sequence for mature feruloyl esterases from A. niger and A. tubingensis. The sequence identity is 92.5%. The region homologous to the putative serme lipase catalytic triad is underlined, and the points of change are indicated by a * (from [101)

A model of A. niger F AE-III (Figure 3), based on sequence identity with four fungal lipases, shows that the location of this amino acid change lies on an exposed loop of the ct/13 barrel structure, and so the residue is accessible to proteolytic attack while in the native form. Activity of the purified esterases from A. tubingensis was compared to FAE-III from A. niger against methyl ferulate. Activities were very similar:- FAE-III (4.38 l.tmol/min/mg protein) and A. tubingensis FAE (4.46 U/mg protein). This indicates that the differences in amino acids (16 residues) are not important for activity on methyl ferulate, and that proteolytic attack on A. tubingensis FAE does not interfere with the catalytic mechanism. Initial results on the stability of the two esterases to thermal inactivation suggests that they behaved in a very similar manner at 60~ losing 50% of their activity in 60-80 minutes. Thermal inactivation of FAE-III from A. niger has been found to be reversible, the first order rate constant of unfolding (v) was 0.76'10 .3 per second, and AG* was 101.9 kJ/mol [11].

45

Figure 3. A schematic ribbon diagram of the 3-D fold of A. niger FAE-HI monomer model. The model was based on the alignment of FAE-III with four lipases: Humicola lanuginosa [ 15], Penicillium camembertii [16], Rhizomucor miehei [17] and Rhizopus niveus [18], which showed significant homology (Z-score larger than 4.0). The N-terminus is located at the bottom right of the figure and the C-terminus is located near the active site. The location of the loop containing the D'-~S amino acid change is indicated. In summary, two feruloyl esterases from Aspergillus show 92.5% amino acid sequence identity, similar physical characteristics, and specific activity against methyl ferulate, but the enzyme from A. niger (FAE-III) is more resistant to degradation than the esterase from A. tubingensis.

This work was supported by a Biotechnology and Biological Sciences Research Council (BBSRC) Realizing Our Potential Award (ROPA), the Industrial Research Center (Tripoli, Libya) for a Ph.D. scholarship to FOA, and the BBSRC, UK. We wouM like to thank Dr. Fred Mellon (Institute of Food Research) for carrying out the electrospray mass spectroscopy on the two enzymes. REFERENCES

[1] [2] [3] [4]

R.D. C.B. C.B. P.A.

Hartley and E.C. Jones, Phytochemistry 16 (1977) 1531. Faulds and G. Williamson, J. Gen. Microbiol. 137 (1991) 2339. Faulds and G. Williamson, Microbiol. 140 (1994) 779. Kroon, C.B. Faulds and G. Williamson, Biotechnol. Appl. Biochem. 23 (1996) 255.

46

[5] [6]

[7] [8] [9] [10]

[11] [12] [13] [14] [15] [16] [17] [18]

L.M.A. Ferreira, T.M. Wood, G. Williamson, C. Faulds, G.P. Hazlewood, G.W. Black and H.J. Gilbert, Biochem. J. 294 (1993) 349. W.S. Borneman, L.G. Ljungdahl, R.D. Hartley and D.E. Akin, Appl. Environ. Microbiol. 58 (1992) 3762. A. Castanares, S.I. McCrae and T.M. Wood, Enzyme Microb. Technol. 14 (1992) 875. B.P. Dalrymple, Y. Swadling, D.H. Cybinski and G-P. Xue, FEMS Micro. Letts. 143 (1996) 115. B.P. Dalrymple and Y Swadling, Microbiol. 143 (1997) 1203. R.P. de Vries, B. Michelsen, C.H. Poulsen, P.A. Kroon, R.H.H. van den Heuvel, C.B. Faulds, G. Williamson, J.P.T.W. van den Hombergh and J. Visser, Appl. Environ. Microbiol. 63 (1997)4638. G. Williamson and J. Vallejo, Int. J. Biol. Macromol. 21 (1997) 163. A. Sali and T.L. Blundell, J. Mol. Biol. 234 (1993) 779. T.A. Jones, J-Y. Zou, S.W. Cowan and M. Kjeldgaard, Acta Cryst A47 (1991) 110. P. Kraulis, J. Appl. Cryst. 24 (1991) 946. U. Derewenda, L. Swenson, Y. Wei and Z.S. Derewenda, J. Lipid Res 35 (1994) 524. S. Yamaguchi, K. Takeuchi, T. Mase, K. Oikawa, T. McMullen, U. Derewenda, R.N. McElhaney, C.M. Kay and Z.S. Derewenda, Prot. Engin. 9 (1996) 789. Z.S. Derewenda, U. Derewenda and G.G. Dodson, J. Mol. Biol. 227 (1992) 818. M. Kohno, J. Funatsu, B. Mikami, W. Kugimiya, T. Matsuo and Y. Morita, J. Biochem. 120 (1996) 505.


47

Stability Properties o f C a l f Intestinal Alkaline P h o s p h a t a s e Ultan McKEON, Brendan O'CONNOR & Ciarhn 0'F/kG.~qq" School of Biological Sciences, Dublin City University, Dublin 9, Ireland.

Inactivation of alkaline phosphatase (APase) in the range 48-65~ fits a first order decay; an Arrhenius plot gives estimates of low-temperature half-lives. APase tolerates urea but not GdC1. Micromolar amounts EDTA lead to inactivation, as do millimolar amounts 1,10phenanthroline. APase is sensitive to SDS but retains significant activity in presence of the solvents DMSO, DMF or THF (up to 60% (v/v) in buffer).

1. INTRODUCTION Alkaline phosphatase (Orthophosphate monoester phosphohydrolase; EC 3.1.3.1; APase) is used to prevent self-annealing of DNA by cleavage of 3'-terminal phosphate groups (1). Its ability to utilize luminescent substrates has led to its use as a reporter enzyme in nonradioactive DNA probes (2) and in sensitive enzyme immunoassays (3). Its levels in serum can be diagnostic in clinical chemistry (4) and it is included in commercial control sera. Long-term persistence of enzyme activity can be a critical parameter in these various types of diagnostic or molecular biology products. It would be useful to have some indication of the baseline stability characteristics of APase and of its ability to withstand various kinds of inactivating influences. The tetrameric, zinc-containing calf intestinal APase (5) is widely used. Here we explore the intrinsic stability properties and limitations of this enzyme and we report on its tolerances of heat, denaturants, chelating agents, detergent and water-miscible organic solvents.

2. EXPERIMENTAL Boehringer Mannheim calf intestinal APase (product # 108 138) was diluted 1/1500 in 2amino-2-methyl-1-propanol/metal ion buffer (6) for use and assayed by the method of IFCC [intemational Federation of Clinical Chemistry] (6). 4-nitrophenyl phosphate and other assay reagents were from Sigma, as were urea, guanidine chloride [GdC1], EDTA, 1,10phenanthroline and SDS. Aldrich supplied 1,7- and 4,7-phenanthrolines.

"Author for correspondence.Tel +35317045288; Fax +35317045412; Email [email protected]

48 Dimethylsulphoxide [DMSO] was from BDH while acetone, dimethylformamide [DMF] and tetrahydrofuran [THF] were from Labsean (Dublin, Ireland). For the temperature profile, aliquots of APase were placed (after preheating to 35-40~ for 10 min at temperatures in the range 35-70~ then removed onto ice to prevent further denaturation. Residual activity was later determined under optimal conditions. APase was heated continuously at 48~ 50~ 55~ 60~ and 650C to obtain thermoinactivation curves. Residual activities of samples, removed at intervals and stored on ice, were assayed. An exponential decay curve of remaining activity against time was plotted to estimate the first order rate constant, k, at each temperature. Urea and GdC1 stock solutions (5M) were prepared according to (7), but using Tris, pH 8.3, instead of MOPS buffer. Incubation of the enzyme in the denaturant took place at 30~ for 2 h, after which time enzyme activity was determined. The effects of EDTA (pH 8.3), 1,7-phenanthroline, 1,10-phenanthroline and 4,7phenanthroline on activity were determined. After incubation at 370C for 4h, samples were extensively dialysed (40C) against 10mM Tris-HC1, pH 8.3. Activities were then measured. Enzyme was treated with various v/v concentrations of THF, DMF and DMSO and assayed after 90 min at room temperature. APase was treated with various w/v concentrations of SDS. After 2h at 30~ catalytic activity was measured.

3. RESULTS & DISCUSSION

Thermostability Studies. The temperature profile experiment gave 57~ as halfinactivation temperature, Tso. All thermoinactivation curves in the range 48~176 fitted a first order decay curve and gave linear semi-log replots. Values for the first-order rate constant, k, and half lives (t~t2= 0.693/k) are shown in Table 1. Table 1 Values of first-order rate constant, k, for APase thermoinactivation Temperature

48~

50~

55~

60~

65~

k, min"

0.021+ 0.001

0.028+ 0.001

0.063+ 0.002

0.143+ 0.008

0.30+ 0.04

An Arrhenius plot of these data yielded a value of 142 kJ.mol-! for the inactivating event under the conditions used. Extrapolation of the Arrhenius plot permits estimation of k (and of likely half lives) at temperatures of interest - see Table 2. Note that the value for 0~ takes no account of possible freezing effects. The values obtained under our particular experimental conditions may not be generally applicable. Franks (8) has discussed strengths and weaknesses of accelerated stability testing.

49 Table 2 APase half-lives estimated from Arrhenius plot. Temperature

0~

4~

20~

25~

37~

Half-life, h

6360

2568

88.8

33.6

3.6

Half-life, days

265

107

3.7

1.4

0.15

The DNA dephosphorylation protocol of Maniatis et al. (1) used calf intestinal APase for 1 h at 37~ followed by inactivation at 68~ for 15 min in SDS. The present results indicate that the enzyme remains active throughout the 37~ incubation and will quickly undergo inactivation at 68~ In the measurement of PCR products, Yang et al. (9) incubated APase at 37 ~ for 30 min in an enzyme-substrate reaction. Nakagami et al. (10) used APase in an enzyme-conjugated DNA probe system; they carried out hybridization of probe with target DNA at 42~ and found that a 30 min hybridization period gave the greatest sensitivity, with decreased sensitivities at longer times. These protocols appear to make good use of the intrinsic thermal stability characteristics of APase. de la Fourniere and colleagues (11) studied thermal and pH stabilities of bovine intestinal APase by Fourier transform infrared spectroscopy [FTIR]. At temperatures up to 70~ (or at pH values down to 5.4), activity loss was not accompanied by significant FTIR changes; i.e. there were no notable conformational alterations. At 80~ (or at pH 3.4), however, APase began to unfold after it had completely lost activity. FTIR results were confirmed by H/2H exchange and circular dichroism. Denaturant Stability. Fig. 1 shows the effects of increasing concentrations of urea and GdCI on APase activity. Urea up to 3M exerts a slight activation effect and approximately 80% activity is retained even at 5M urea. GdC1 has a similar activating effect up to 0.5M but APase activity declines steeply at GdC1 concentrations above 0.5M and is almost absent at 2M GdC1. GdC1 is a more potent denaturant than urea (7). At low concentrations, both denaturants slightly activate APase, perhaps by altering its conformation slightly to favour catalysis. 0.1M GdC1 caused tertiary structural changes in the monomeric horseradish peroxidase without affecting activity (12). Nakagami et al. (10) noted that APase retained 40% activity after 3 h in 6M urea. Effect of EDTA and Phenanthrolines. Micromolar EDTA concentrations have a potent adverse effect on APase (Fig. 2), as one would expect for a Zn-containing enzyme. This is most likely due to chelation of the catalytically-essential Zn 2§ ion and is confirmed by Fig. 3, which shows the effects of phenanthrolines on APase activity. The chelating 1,10phenanthroline adversely affects APase (but at millimolar concentrations) while the nonchelating 4,7- and 1,7-phenanthrolines (13) have little or no effect. Note that exhaustive dialysis of EDTA-containing samples took place prior to assay to prevent EDTA interference in the assay mix. Femley (5) has stated that the kinetics of EDTA inhibition are complicated and depend on both time and substrate. Satoh (14) has used the immobilized apoenzyme [Zn-free APase] for microanalysis of Zn(II) ions. This system could be used at least 120 times; treatment with a chelating agent regenerated the reactor following each measurement. Maniatis et al. (1) used EDTA together with high temperature to terminate the dephosphorylation reaction.

50 140 120 100 .,..4 9

p. 80

< 0

60

.,.,.1 l,.,,,d

40

o

GdCI

o

Urea

20 I

I--

1

2

I

I

I

3

4

5

C oncentration(M

)

Fig. 1. Effects of denaturant on APase activity.

100

p.

80

< 0

r .!.=,

60

40

: 20

"--.... 0

10

20

30

EDTA(~tM)

Fig. 2. Effects ofEDTA on APasr activity.

40

50

60

70

51 120 100

80

~-

e A

0

9 0,8 13 < | 0,6 .>_ ..,., (u 0,4

O

9

O

9

9

O

~.,.. E z: 3,5

II

O

[]

n

[]

9

m

9

9

t

t

I

I

2

3

4

5

CLL ~U.

0 lml

o

[]

0

I

i

t

i

i

6

7

8

9

10

=9 ~ 2,5 -~ 2o9 2

2

0,2

t i

i

i

2

3

4

i

i

i

i

i

5

6

7

8

9

1

10

Time (days)

Time (days)

1,2 ......

4

1

A

A

:E o,8

9

"6 < 0,6 (D .__. ~9 0,4 .=.

~,3,5

._~ mo2,5 9- -

U.

A

O)

o I

9

A

~E

A

A

I

I

I

I

I

I

2

3

4

5

6

7

1

8

1

~me(days)

9

A

A

I

I

I

I

I

i

i

2

3

4

5

6

7

8

Time (days)

12

3,5 A ~.

1 .m

3

2,5

~ 9 0,8 9

A

>,O

o:: 0,2

> -.~

m

--r 0,4 r n~

~

02

@ 8

o

-->'m P ' 6 1,5 E

0,6

o o

o o 9

9

9

v.=,.

- = E o,s 1

I

I

I

2

3

4

I

I

,

,

1

5

6

7

8

9

Time(days)

,

10 11

9

9

o 9

9

9

8

g

10

1

0

0

1

2

3

4

5

6

7

~rne (days)

Fig. 3- Batch operational stability tests to the polyurethane immobilized C. rugosa lipase during the hydrolysis of crude olive residue oil. A- Relative activity-data presented with reference to the initial hydrolysis rate of batch number 1 (1 batch = 23 hours). B- Application of l't order inactivation model. (Exp.1) 25% FFA crude oil, water/TG = 19.0; (Exp.2) 25% FFA crude oil, water/TG=4; (Exp.3) 25% FFA oil previously treated by activated earths, water/TG = 19.0; (Exp.4) similar to exp.3, but the oil was also treated by alumina; (Exp.5) 11% FFA crude oil, water/TG = 19.0;(Exp. 6) similar to exp.5, but the oil was treated by activated charcoal (see text for details).

76 When the 11% FFA crude oil was used (Fig.3- exp.5 and 6), the stability was lower than the observed with 25% FFA oil (Fig3-exp.l-4). This may be ascribed to the inhibitory effect of higher amounts of chlorophyls and secondary oxidative products in the first oil (Table 1). In fact, when pigments and copper were removed by adsorption to charcoal (exp.6), a slight increase in stability was observed as compared to experiment 5. Inactivation caused by FFA was not to be expected since the partition coefficient for oleie acid between the support and the organic solvent was close to zero 13. The removal of ehlorophyls from the more acidic oil with activated earths and alumina, also promoted an increase in lipase stability (Fig.3-exp.3 and 4). In the present study, the experimental data do not fit a first order exponential decay model (Fig.3B). The activity profiles suggest a non linear inactivation kinetics with sucessive inactivation stages 2~ for every experimental conditions tested. Further studies have to be carded out to understand the inactivation process caused by the impurities in crude oils.

Aknowledgements The authors are grateful to Prof. F/ttima Peres, Escola Sup. Agrhria de Castelo Branco, Portugal, for the Fe and Cu determinations. This study was supported by grant n ~ JNICT

PBIC/C/AGR/2316/95, Portugal. REFERENCES

I Brady, C.; Metcalfe, L.; Slaboszewski, D.; Frank, D (1988) Jr. Am. Oil Chem. Soc. 65 917921. 2 Dtinhaupt, A., Lang, S., Wagner, F. (1992), Biotechnol. Lett., 14 (10) 953-958. 3 Wisdom, R.A., Dunnill, P., Lilly, M.D. (1985), Enzyme Microb. Technol., 7 (11) 567-572. 4 Linfield, W. M. (1988), In: Proceedings World Conference on Biotechnology for the Fats and Oils Industry, (T. H. Applewhite, ed.), Am. Oil Chem. Soc., Champaign, USA, 131-133. 5 Posorske, L.H., LeFebvre, G.K., Miller, C.A., Hansen, T.T., Glenvig, B.L. (1988), J. Am.Oil Chem. Soc., 65, (6) 922-926. 6 Wang, Y., Gordon, M.H. (1991), J. Agric. Food Chem., 39, 1693-1695. 7 Legier, V., del Guist, C., Comeau, L. (1994), Rev. Fr. Corps Gras, 41 (3/4) 45-52. 8 Kuncov/t, G., Malrterovfi, Y., Loveckfi, P. (1994), Biotechnol. Techn., 8 (8) 535-540. 9 F. Dias, S., Vilas-Boas, L., Cabral, J.M.S., Fonseca, M.M.R. (1991), Biocatalysis, 5:21-34. l0 Ferreira-Dias, S. and da Fonseca, M.M.R. (1995) Bioprocess Eng., 13 (6) 311-315. 11 Hoq, M.M.; Yamane, T.; Shimizu, S.; Funada, T.; Ishida, S.(1984), J.Am. Oil Chem Soc., 61,776-781. 12 Kang, S. T.; Rhee, J. S.(1989), Biotechnol. Bioeng. 33: 1469-1476. 13 Ferreira-Dias, S., da Fonseca, M.M.R. (1995), Bioprocess Eng., 12 (5) 327-337. 14Ferreira-Dias, S., da Fonseea, M.M.R. (1995), Biocatalysis and Biotransformations, 13 (2) 99-110. 15 Lowry, R.R.; Tinsley, I.J. (1976) ,/. Am. Oil Chem. Soc. 53: 470-472. 16 Ferreira-Dias, S.; Fonseca, M.M.R. (1993) Biotechnol. Techn. 7: 447-452. 17 Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J. (1951), J.BioLChem., 193, 265 18 Peres, M.F.P. (1995),Sobre a oxidar de azeite virgem catalisada por metais, Tese Mestrado Cirncia e Tecnologia de Alimentos, UTL, Lisboa.. 19 Fassett, D.W. (1963)In: Industrial Hygiene and Toxicology, Interscience Publ., vol.2, pp: 2032-2033. 20 Henley, J.P., Sadana, A. (1986), Biotechnol. Bioeng.,28:1277-1285


77

A p p r o a c h e s for i m p r o v e d i d e n t i f i c a t i o n o f m e c h a n i s m s o f e n z y m e i n a c t i v a t i o n M. Polakovi~,, P. Vr~ibel* and V. B~le] Department of Chemical and Biochemical Engineering, Faculty of Chemical Technology, Slovak University of Technology, Radlinsk6ho 9, 812 37 Bratislava, Slovak Republic

This publication reviews the results of the authors' studies dealing with the identification of mechanisms of enzyme inactivation. It presents both a critical assessment of traditional approaches of evaluation of enzyme inactivation kinetics from uniresponse experiments as well as the procedures of better design and evaluation of experiments which improve the analysis of inactivation mechanism and its corresponding kinetics. These methods are based on the simultaneous evaluation of inactivation experiments carried out at several temperatures or pHs. The inactivation mechanism is thus evaluated from the data set containing larger amount of information compared to a single inactivation curve when the influence of the factor of variation is unequivocally defined, for example, by the Arrhenius equation in the case of temperature or protonation and deprotonation equilibrium constants in the case of pH. It is demonstrated furthermore that the complexity of the mechanism of the thermal inactivation of free enzyme can be enhanced by utilising the combination of selective immobilization and multitemperature evaluation when the immobilization prevents some reactions in the ordinary inactivation pathway to occur. All methods are illustrated on case studies using yeast invertase.

1. DRAWBACKS OF CONVENTIONAL EVALUATION [1] Enzyme inactivation is generally explained as a chemical process involving several phenomena like aggregation, dissociation into subunits, or denaturation (conformational changes), which occur simultaneously during the inactivation of a specific enzyme [2,3]. However, the principal information for elucidating the kinetics of enzyme inactivation is obtained through the measurement of enzyme activity, which is determined as the rate of enzyme reaction at specified conditions. Consequently, enzyme activity gives the best quantitative information on the phenomenon of activity loss, but it is difficult to relate this to the changes occurring in the enzyme structure. The total enzyme activity, A, is generally equal to the sum of the activities of active enzyme forms, A i . T h e activity, a, is usually given as a relative value of the initial activity, Ao,

' Present address: KluyverLaboratory for Biotechnology, Department of Biochemical Engineering, Delft University of Technology, Julianalaan 67, 2628 Delft, The Netherlands

78

E Ai

E ?'ici (1)

A

2A, o 2r,c,o i

i

where Yi are the molar activities and c i are the molar concentrations of enzyme forms. Equation (1) reflects the methodological problems encountered in the analysis of the kinetics of enzyme inactivation. As the inactivation is represented by a set of reactions among different enzyme forms, for the evaluation of kinetics it would be highly useful to know the content of individual enzyme forms. Further complications during the analysis of inactivation kinetics are that the molar activities of enzyme forms are generally unknown. Thus, it is difficult to decide from the inactivation curve alone whether a reaction is reversible or irreversible, whether parallel reactions are taking place on the same enzyme form or not, or whether intermediates are active or inactive. It has been demonstrated how simple kinetic equations (mainly first-order kinetics) can disguise a more complex kinetic behaviour [4-6]. Different inactivation mechanisms of equal complexity can however correspond to the same integral activity-time relationship and the number of equivalent mechanisms increases exponentially with the number of parameters in the kinetic equation [1]. Thus, neither isothermal inactivation data exhibiting non-first order kinetics do not provide sufficiently consistent information upon which a particular mechanism of inactivation could be declared. A careful statistical analysis of extensive set of inactivation data found in literature showed that any pattern of enzyme inactivation behaviour (biphasic, grace-period or activation-period) can be described by a relatively simple exponential function [1]. Such functions represent lumped kinetic equations and any simple mechanisms which could be identified from them are dismissed if the validity of the mechanisms is tested at several temperatures [ 1].

2. MULTITEMPERATURE EVALUATION [7] If the kinetic parameters are evaluated using a simultaneous fit of the inactivation data obtained at several temperatures, this provides an opportunity to improve the analysis of the mechanism and kinetics of inactivation. The improvement of the analysis consists in: 1. the inactivation mechanism should be validated at each temperature; 2. the kinetic equation may contain parameters which should not change with temperature; 3. the parameters dependent on temperature should provide a consistent temperature relationship, for example, the rate constants could be expected to obey the Arrhenius equation. Figure 1 presents the inactivation of yeast invertase at four temperatures from 40 to 60~ charaeterised by the so-called biphasic behaviour. In order to measure the benefits gained by the evaluation of inactivation kinetics from multitemperature data, isothermal data have been evaluated as well. The most distinctive difference between the quality of isothermal and multitemperature evaluation was found at the isozyme mechanism and is presented in Fig. 1. This mechanism was unsuitable for describing the inactivation of yeast invertase at all temperatures simultaneously whereas the corresponding isothermal model fitted the individual inactivation experiments very well.

79

1.0 ""

- - ~d.

-O

0.8

r - 0.6 >,~ :L-'_ .> o

C

6

0

(3

m r e~

G)

(3

4

20

10

0

100

200

Time

300

400

500

(min)

Figure 3. Effect of sorbitol on the synthesis of cephalexin with B. megaterium PA. (1): control; (0):20% (w/v) sorbitol

4. CONCLUSIONS Better results were obtained with PA from B. megaterium both for the synthesis of ampicillin and cephalexin. 7-ADCA was a better nucleophile than 6-APA, higher yields being obtained in the production of cephalexin than ampicillin with both PA. Sorbitol increased yield and rate of synthesis of both antibiotics. However, the reported yields are still low to compete with existing chemical synthesis, despite the extremely mild conditions required for the enzymatic synthesis. Preliminary results indicate that pH, enzyme to substrate ratio and substrate to substrate ratio are important process variables, so that ample room exists for process optimization PA from B. megaterium is worthwhile to be studied in other organic solvent systems that produce no enzyme inactivation.

I00 REFERENCES 1. 2. 3. 4.

J. Shewale and V. Sudhakaran, Enzyme Microb. Technol., 20 (1997) 402. D. Nam, C. Kim and D. Ryu, Biotechnol. Bioeng.27 (1985) 953. V. Kasche, U. Haufler and L. Riechmann, Methods in Enzymol. 136 (1987) 280. A. IUanes, R. Tortes, O. Cartagena, A. Ruiz and M. Vdsquez, Process Biochem. 29 (1993) 263. 5. K. Balashingham, D. Warburton, P. Dunnill and M.D. Lilly, Biotechnol. Bioeng. 276 (1972) 250. 6. C. Hyun and J. Kim, Biotechnol. Bioeng. 42 (1993) 800.

Acknowledgements This work was supported by chilean FONDECYT, Grant N~ The authors express their gratitude to SINQUISA S.A., Lima, Peru for their kind supply of 6-APA and penicillin GK.


101

Conversion and stability studies on enzyme-membrane reactor with lipase immobilized by different methods T. Venyige, E. Cs~myi and Cs. Sisak Research Institute of Chemical and Process Engineering, Pannon University of Agricultural Sciences, Egyetem u. 2,. H-8200 Veszpr6m, Hungary

1. INTRODUCTION Nowadays the enzyme-membrane reactors (EMR) have proved as a suitable system to carry out bioprocesses in which the components taking part in the reaction are in two immiscible liquid phases and the biocatalyst can be immobilized on or inside the membrane. They are favourably applied e.g. for enzymatic hydrolyses [1-3] and esterification [4,5] but there are several examples for other reactions, e.g. peroxidation of fatty acids [3]. The importance of membrane containing reactors is increasing further with the spreading of integrated systems. To perform simultaneously the biocatalytic reaction and the product removal it is obvious to utilize the opportunities of the membrane separation. Among the so-called extractive bioconversions, the first results about the formation of bioflavours was published ten years ago. The procedure was based on the finding that natural ethyl esters can be formed by in situ enzymatic esterification of fatty acids and ethanol produced from glucose via fermentation by yeast cells [6,7]. On the basis of the recent investigations [5], it has been cleared up that EMR is also applicable to ethyl oleate production if the ethanol content of substrate is of 10-30 wt%. *In the present studies, the performance of an enzyme-membrane reactor applied to ethyl oleate production is discussed. The main question to be answered is, how the reactor operates - concerning the ester formation and stability - under the circumstances of the really integrated extractive biocatalytic reactor system: i.e. at low ethanol content and in presence of living yeast cells. 2. MATERIALS AND METHODS 2.1 Materials Mucor michel lipase (LYPOZYME, 10000 LU/g, NOVO Nordisk Denmark) was used as catalyst. Aqueous ethanol (REANAL Hungary) solution of 60-160 g dm 3 initial concentration or ethanol containing fermentation broth produced by Saccharomyces bayanus

The financial supports of European Commission through COPERNICUS Grant No ERBCIPACT 923011 and German-Hungarian Intergovernmental Co-operation through Grant No. D/46-96 are gratefully acknowledged. " Acknowledgement:

102 IST 154 cells, and oleic acid (REANAL Hungary) were applied as substrates. Chemicals (from REANAL and Sigma Chemical Company) were of reagent grade.

2.2. Immobilization methods

Ultrafiltration/adsorption method: Lipase solution (100 cm 3, l g) was ultrafiltered (flow rate 4 cm 3 min -~) through the membrane and enzyme was immobilized by adsorption on the inner wall of the fibres. The process was followed by determination of protein contents of both the initial solution and permeate. The ratio between the amounts of dissolved and immobilized lipase was 3:2. The strength of the enzyme layer was checked by flowing distilled water through the membrane tubes: no protein wash-out was found [2]. Coating method: The undiluted enzyme solution (1 cm 3) was spread on the inner wall of the dry fibres applying alternated spreading and drying steps. After last drying, the fibres were coated with hexanoic solution of silicone rubber (DOW CORNING 3140 RTV, ingredients: methyl-trimetoxysiloxane, trimethylated silica, polymethylsiloxane). The 20 cm 3 solution contained 3 g rubber. In order to harden the silicone coating, the module was rinsed with inert gas for 24 hours. Since the polymerisation of silicone rubber takes place in presence of water, the gas stream was moistened. Emulsion method: The undiluted enzyme solution was mixed with the silicone solution (see above) and shaken in a test tube for 5 minutes. The emulsion was spread on the inner wall of the dry fibres as it was described above, with alternated spreading and drying steps. The drying procedure was the same as in case of the coating method. 2.3. Reactor configuration Experiments were carried out in a hollow fiber membrane bioreactor with 0.0188-0.0376 m 2 surface area. The hydrophylic membrane tubes ( asymmetric cellulose acetate, cut off: 40 000 Da) had 2 mm diameter and 200 mm length. During the reaction, 300 cm 3 oleic acid was recirculated on the enzyme side of the membrane and 100 cm 3 aqueous ethanol solution on the other side. Water produced could permeate through the membrane into the aqueous phase and ester remained in the organic phase. Concerning the studies in the extractive bioconversion system, the ethanol fermentation process was performed in a fluidized-bed immobilized-cell fermenter and the esterification in EMR described above. The scheme of the integrated system is illustrated in Fig. 1.

Figure 1. The scheme of the integrated fermenter/enzyme reactor system

103

Saccharomyces bayanus cells were immobilized by Ca-alginate gel entrapment. The fluidized-bed reactor was run in semicontinuous regime: periodically glucose solution of 40 wt % was added to the reaction mixture and the same quantity of broth was taken off. The effective volume of the reactor was 300 cm 3, its diameter 36 mm, the volume of the bed of immobilized-cell particles was 50 c m 3. The fermentation broth was recirculated in the primary side of the membrane reactor. 2.4. Analytical methods The samples were analyzed by HPLC (Merck-Hitachi L-6000A pump, Merck refractive index detector RI-71, temperature 70~ Nucleosil 100-10 C 18, 250x4 mm column with metanol:water (60:40) as eluent (flow rate 0,8 cm3/min) for acid and ester, PL Hi-PLex 300x7.7 mm column 5 mmol dm 3 H2SO 4 solution for ethanol (flow rate 0,7 cm3/min) were used.

3. RESULTS AND DISCUSSION

3.1. Esterification in enzyme-membrane reactors with iipase immmobilized by different methods Fig. 2 shows the bioconversion run in enzyme-membrane reactor with lipase immobilized by the ultrafiltration/adsorption method. The membrane surface was 0.0376 m 2. It can be seen on the figure that the rate of ester formation at the repeated adding of ethanol (dotted line) approximates the initial one, i.e. no considerable inactivation of enzyme can be supposed during the period.

1,6 1,4

:~~t........~

-

1,2 ~

Et(OH)

1

Ester

~ 0,8

~ 0,6 ~ 0,4

o,2 0 0

1

I

t

2

3

Time (days)

4

Figure 2. Time course of esterification in enzyme-membrane reactor with lipase immobilized by ultrafiltration/adsorption method. Time course of esterification applying the coating method is illustrated in Fig. 3. A model medium with non metabolizing Saccharomyces bayanus cells was used as aqueous phase (initial ethanol cone.: 163 mg/g, glucose cone.: 24 mg/g). Other parameters were the same as above. Comparing the initial slopes of the ester concentration vs. time curves in Figs. 2 and 3, it is unambiguous that the productivity of the enzyme-membrane reactor is higher in case of the coating method.

104 The experimental conditions applied to study the productivity of EMR with lipase immobilized by emulsion method were almost the same as that given previously, with exceptions of the enzyme fixation method and the volume of the organic phase (200 cm3). On the basis of the results (see Fig. 4), it is proved that some improvement in the productivity can be achieved using the emulsion method in comparison to the two other ones.

1,4

1

g

m

+__...._..-4

0,6 O

lira

0,4

Et(OH)

-

f

0,2

--a- Ester

0 .

I

I

I

t

I

0

0,2

0,4

0,6

0,8

1

1,2

Time (days) Figure 3. Time course of esterification in enzyme-membrane reactor with lipase immobilized by coating method.

1,6t

1,4~1,2-

" Et(OH) -.m-Ester

/~

1 0,8 I

0,6

elm

-a-t -a-t - - " " A

i

O

;~ 0,4

t

0,2 0 "~. 0

I

t

t

1

2

3

Time (days) Figure 4. Time course of esterification in enzyme-membrane reactor with lipase immobilized by emulsion method. 3.2. Extractive bioconversion in integrated system The organic phase of 200 c m 3 volume was recycled in enzyme-membrane reactor with 0.0188 m 2 membrane surface area. The time course of the run (Fig. 5) shows that quasi stationary conditions exist concerning ethanol concentration along a six day range. The ester level was increasing with almost constant slope, what implies probably that the equilibrium concentration would be much higher.

105

~0~ ~0,6

:. Et(OH) Ester e Glucose

O

~0,4 'r

t~--e i

0,2 __ O=9 0

, e------.e-....e.., e -

t~.

2

--e-.e" I

4 T ~ e (days)

,

I

ii

6

8

Figure 5. Time course of extractive bioconversion in fluidized-bed immobilized cell reactor/enzyme membrane reactor system.

3.3. Working stability of the enzyme-membrane reactor The setting up of the EMR was based on our experiences of its application to enzymatic hydrolysis of vegetable oils and fats. The membrane tubes were fixed by epoxy glue and it has been proved to bc fairly resistant to the chemicals applied [5]. Unfortunately, it was found during the esterification runs that the ethyl oleate/oleic acid mixture causes the corrosion of the glue. Due this effect, the aqueous and organic phases were mixed after 100-200 hours and the experiments had to be stopped. After testing the possibilities to solve this problem, a more advanced reactor structure with flat membrane has been elaborated for the further investigations. 4. C O N C L U S I O N S

Based on the analysis of these studies,conclusions can be drawn as follows: Comparing the differentimmobilization methods of lipasc (Table I) significantadvantage can bc found for the coating and emulsion procedures. Although the same method was used in the last two cases, the difference can probably be explained by the considerable different medium composition and conditions. A relativelyhigh ester concentration can be achieved in the secondary side of the enzyrnc membrane reactor even in case of low (30-40) ethanol concentrations in the primary side. If the operational stabilityof E M R will bc improved by the application of the reactor of new type, presumably flavour esteryields can be enhanced further. The presence of living cells and other components of the broth did not reduced dramatically the ester formation rate. Consequently, the further studies on extractive bioconversion reactor system consisting of the immobilized cell fermenter and E M R seem to bc promising for natural flavour esterproduction.

106 Table 1. Values of initial ester formation rate related to membrane surface area

Medium in the primary side

Immobilization method

Ethyl oleate formation rate (g day"m"2).........

Aqueous ethanol

UF/adsorption

239

Aqueous ethanol/ glucose/biomass

Coating

638

Aqueous ethanol/ glucose/biomass

Emulsion

854

Fermentation broth with living cells

Emulsion

414

REFERENCES

1. 2. 3. 4. 5.

D.R. Wu, S.M. Cramer and G. Belfort, Biotechnol. Bioeng. 41 (1993) 979. K. B61afi-Bak6, A. Dombi, L. Szab6 and E. Nagy, Biotechnol. Techniques. 8 (1994) 671. S.T. Bouwer, F.P. Cuperus and J.T.P. Derksen, Enzyme Microb. Tehnol. 21 (1997) 291. M. Habulin and Z. Knez, J. Membr. Sci. 61 (1991) 315. E. Cshnyi, K. B61afi-Bak6, T. Venyige and Cs.Sisak, Proc 1 st E u r . Congr. on Chem Eng. Florence. Vol 4. (1997) 2627. 6. M.R. Aires-Barros, J.M.S. Cabral and J.M Novais, Biotechnol. Bioeng. 29 (1987) 1097. 7. A.C. Oliveira, and J.M.S. Cabral, J. Chem Tech. Biotechnol. 52 (1991) 219.


107

Study of the deactivation process of the glucose oxidase-catalase enzymatic system by m e a n s of simulation L. E. Romero* and D. Cantero Biological and Enzymatic Reactors Research Group. Department of Chemical Engineering, Food Technology and Environmental Technologies. Faculty of Sciences. University of Cadiz. 11510. Puerto Real. Cadiz. Spain.

This paper reports the simultaneous deactivation of co-immobilized glucose oxidase and catalase in calcium alginate beads. For this purpose, the study is conducted in a continuous stirred tank reactor containing the reaction medium. After a comprehensive review of the existing literature, a three-parametered irreversible deactivation general model is put forward for each enzyme. 1. INTRODUCTION Nowadays, glucose oxidase (GOD) and catalase (CAT) are considered to be one of the enzymatic systems of most interest and use for both sensors and industrial processes. Glucose oxidase and catalase are nature related because they both participate in the enzymatic pool of those microorganisms able to oxidize glucose to gluconic acid [1]. Glucose oxidase has the systematic name of 13-D-glucose: oxygen-l-oxidoreductase (EC 1.1.3.4.) and catalase is H202: 1-1202 oxidoreductase (EC 1.11.1.6.). Both enzymes (GOD) and (CAT) are very stable against pH and temperature; consequently, their pH range is very wide (pH 3-8) [2,3]. If temperature of operation is considered, both enzymes show catalytic activity from 10~ to 60~ [2, 3, 4]; however, they are highly sensitive against hydrogen peroxide [5]. GLUCOSE~

/ [

3N

+

+

DEACTIVATIO~ HYDROGEN- l

Figure 1. Hydrogen peroxide deactivating effects.

]

108 1.1. Deactivation mechanism of glucose oxidase. Existing literature provides evidence for a "ping-pong" model to be used in the irreversible deactivation of the GOD enzyme [5]. Such model assumes that hydrogen peroxide works by deactivating the three states this enzyme is likely to adopt: oxidized state, reduced state, and complex GOD-oxygen state. Further to enzyme activity and reaction time, a representative equation would be:

a=exp(-klP 0

(1)

where: a: relative activity of the enzyme (between 0 and 1) k1: deactivation coefficient. P: concentration of hydrogen peroxide. t: time of reaction. 1.2. Deactivation mechanisms of catalase.

Two different models have been employed for the deactivation mechanisms of catalase, i.e. Tse and Gough's [5], and Vasudevan and Weiland's [6]. Tse and Gough's model is similar to that for GOD. This means that the enzyme is deactivated in the oxidized state, in the reduced state, as well as in the complex enzyme-peroxide state. It provides the following equation: a = exp

(-kot - kiP t)

(2)

In this equation, a new kinetic constant appears, this is k~ Such constant is not affected by hydrogen peroxide concentration. Finally, the model by Vasudevan and Weiland [6] assumes that there exist three states of the enzyme. These states are: enzyme, enzyme-peroxide, and enzyme-peroxide-peroxide. A mathematical treatment of this mechanism provides the following equation:

a = exp

-

klpn p2 t

(3)

P+k 2 +~ k3 Where kl, k2 and k~ are deactivation constants. It can be observed that, whereas the mathematical expression is more complicated than the previous one, the mathematical function is very similar. 2. MATERIALS AND METHODS 2.1. Reactants Experiments were developed by the use of glucose oxidase from SIGMA (G-1262, crude extract). The enzyme had 10.600 A.U./g solid for conditions of oxygen saturation (pH 5.1 and 35~ Catalase employed was from SIGMA (C-3515) with 7080 A.U./mg solid.

109 2.2. Reaction medium The experiments were developed using D-(+)-glucose from PANREAC dissolved in a pH 5.1 buffer (acetic acid/calcium acetate). 2.3. Calcium aiginate gel (CAG) beads C AG beads were prepared by means of the reaction of alginic acid and calcium chloride. A general technique of ionotropic gelling was used for this purpose [7]. Calcium alginate gel beads behaved as carriers for both glucose oxidase and catalase. 2.4. Reactor and starting up The reactor used was a 300 mL stirred tank reactor with a working volume of 250 mL. All the experiments were developed in continuous operation, working at different residence times and with several substrate initial concentrations. The temperature of operation was 35~ and pH was 5.1 -as these are regarded as the optimum operating conditions for glucose oxidase. 2.5. Software The simulation software was developed in C-language using LabWindows (2.2.1.) from National Instruments, and it was finally compiled by Microsoft (6.0). The developed software acquires process variables data and parameters related with the calculus algorithm (finite differences method), providing the simulation results. These results can be simultaneously filed and displayed [8]

' I-'i o

REACTION RATES l

~ J CAT P"I

MASS BALANCES EFFECTIVENESS FACTORS SUSTRATE

PRODUCT

1

! ~176

l l|l

ii

i

I

Figure 2. KINETIC M O D E L FOR THE. SYSTEM

3. RESULTS AND DISCUSSION. As a result of in-depth experimental studies concerned with the kinetic behaviour of GOD and CAT immobilized on CAG beads, a kinetic model is finally proposed. This model is made up of several equations, such as (i) equations for the enzymes (reaction rates and deactivation); (ii) mass balances for substrate and product both inside the particles and inside the reactor; (iii)

110 conversion and effectiveness factors. Unlike traditional models applied to enzymatic processes [9], the model put forward here shows deactivation equations. After a comprehensive review of the existing literature, a three-parameters irreversible deactivation general model is put forward for each enzyme:

aGO D = e x p -

kI(GoD)P2

t , aCA T exp " = -

t

k3(GOD ) + P

k3(cAr) + P

Where: "' ac,oD" and "'acAr " are the relative activities of GOD and CAT; "k" 1 to 3 are the deactivation coefficients for each enzyme; "P" is the hydrogen peroxide concentration; and "t" is the time of reaction. The adjusted values of these coefficients are: kl(c, oD) = 5- 10 "5 (mM s) "I kI(CAT) = 3- 10"2 (mM s) "1 k2(GOD) =

2. 10 "3 s"1

k2(CAT)'-5- I0 3 s~

ka(c, oD) =

1.0 mM

k3tCAT)= 1.2 mM

With the aim of obtaining the adjusted values of deactivation coefficients, these two deactivation equations were included iy: a simulation program together with all those equations necessary to predict the behaviour of tl '~. system. Once developed, it was possible to validate the simulation program thanks to a vast experimentation carried out. Results obtained facilitate the knowledge of the enzymatic oxidation process from glucose to gluconic acid and hydrogen peroxide, starting from its initial stage along the whole process until the complete deactivation of the system.

A

E 15 r

Z

o .,..,

@

9

I--

~ 10 I-, Z

o z o o

5

0

6

12

18-

24

30

36

42

TIME (h)

Figure 3. Experimental substrate concentration ( t ) , experimental product concentration (11); theoretical substrate concentration ( ) and theoretical product concentration ( . . . . ) versus time.

111 In Figure 3 it is possible to check the experimental and simulated concentrations of glucose (susbstrate) and hydrogen preroxide (product) in the reaction medium. As it can be observed, the system works at peak performance along the first ten hours of the process. Then, the system gradually begins to deactivate, substrate concentration returns to its inital value, and product concentration finally falls down to zero due to the wash-out effect. The developed simulation software allows for the calculation of the deactivation profiles of both enzymes GOD and CAT (Figures 4 and 5). Thus, it can be observed that relative activities of the enzymes can be predicted with the same equation, although their behaviour is completely different for each case. Figure 4 shows simulated profiles of relative activity versus dimensionless radius for different times. It should be observed that at the surface of particles r = 0 and at the center of particles r = 1. The enzyme glucose oxidase follows a "progressive conversion" model in which activity decays with the same degree in all radial positions. This behaviour is explained by an irreversible enzyme deactivation which increases along with the process time. Thus, the time the enzyme is located in an environment with unfavourable conditions seems to be the most relevant factor in the deactivation kinetics.

,

2h 0.8

Q I

r=l

)0.6

(3

Oh

r=O

5h 0.4

0.2 .I 0

25h

1Oh ,

,

0.2

0.4

'

! 0.6

0.8

DIMENSIONLESS RADIUS (r)

1

Figure 4. Simulated profiles of relative activity GOD versus dimensionless radius.

On the other hand, catalase deactivation (Figure 5) can be represented by a "decreasing nucleus" model from the surface to the center of the particles. Figures 4 and 5 also contribute to conclude that CAT deactivates earlier than GOD. This way, the relative activity of catalase is zero after five hours of process and, at this point, GOD still maintains a 60% of its activity (relative activity 0.6). As a conclusion, a sequential deactivation of both co-immobilized enzymes can be observed: CAT decays in the first place, and GOD in the second place, each one with a different deactivation model.

112

0.8

0

0

3h

I-" 0.6

4h

0.4 0.2

0

0.2

0.4

0.6

0.8

1

DIMENSIONLESS RADIUS (r) Figure 5. Simulated profiles of relative activity CAT versus dimensionless radius.

REFERENCES 1. R6hr, M., Kubicek, C.P. and Kominek, J. Biotechnology 0Ed. Rehm, H.J. and Reed, G.) Verlag Chemic Gmbh, Weinheim (1983). 2. Kozhukharova, A., Kirova, N., Popova, I. Batsalova, K. and Kunchev, K.. Biotechnol. Bioeng,, 32 (1988) 245. 3. Jiang, B. and Zhang, I.. Eur. Polym. J., 29 (1993) 1251. 4. Fortier, G. and Bdanger, D. Biotechnol. Bioeng., 37 (1991) 854. 5. Tse, H.S.P. and Gough, D.A. Biotechnol. Bioeng., 29 (1987) 705. 6. Vasudevan, P.T. and Weiland, R.H. Biotechnol. Bioeng,, 36 (1990) 783. 7. Smidsr~d, O. and Skjg~k-Bra~k,G. TmTECH, 8 (1990) 71. 8. Romero, L.E.. PhD. Thesis. University of Cadiz, Spain (1996). 9. B6dalo, A., G6mez, J.L., G6mez, E., Bastida, J. and Martinez, E. Biotechnol. Prop=, 9 (1993) 166.

Stabilization by chemical modification



115

Stabilization of hydrolases by chemical modification with fatty acids or polyethylene glycol F.J. Plou, M.V. Calvo, M. F e r r e r a n d A. B a l l e s t e r o s

Departamento de Biocat~isis, Instituto de Cat~lisis, CSIC, Campus Universidad Aut6noma, 28049 Madrid, Spain 1. I N T R O D U C T I O N One of the m o s t i m p o r t a n t aims of enzyme engineering is to e n h a n c e the stability of enzymes. In particular, the t e m p e r a t u r e stability of e n z y m e s u s e d in l a u n d r y d e t e r g e n t s (lipases, proteases, amylases) is a n i m p o r t a n t factor in productivity. Protein stabilization h a s b e e n achieved by several m e t h o d s (Shami et al., 1989), including immobilization, site-directed m u t a g e n e s i s , chemical modification, crosslinking a n d selection of e n z y m e s from thermophilic organisms. Chemical modification of an enzyme m a y lead to a n alteration of its surface and can t h u s be u s e d to i m p a r t d e s i r e d properties to the biocatalyst. The hydrophile-lipophile b a l a n c e (HLB) of a n enzyme s e e m s to be i m p o r t a n t in some of its properties (recognition of s u b s t r a t e s , binding, stability, etc.). In this context, the in vivo p o s t - t r a n s l a t i o n a l modification of proteins by the covalent a t t a c h m e n t of fatty acids a n d lipids is a w i d e s p r e a d p h e n o m e n o n in n a t u r e b u t as yet not completely u n d e r s t o o d (Schultz et al., 1988). Several m e t h o d s for the acylation of proteins in vitro have b e e n also described. Polyethylene glycol (PEG) is a synthetic p o l y m e r with a m p h i p a t h i c properties. The hydrophilic n a t u r e of P E G m a k e s it possible to modify e n z y m e s in a q u e o u s solution, a n d its h y d r o p h o b i c c h a r a c t e r e n a b l e s the modified e n z y m e s to function in h y d r o p h o b i c e n v i r o n m e n t s . PEG-modified e n z y m e s have b e e n reported to be soluble a n d active in a p o l a r organic solvents. In c o n t r a s t with the v a s t n u m b e r of studies on synthetic or t h e r a p e u t i c a l applications of PEG-enzymes, the effect of this modification on enzyme stability h a s b e e n analyzed by only a few r e s e a r c h e r s {Pasta et al., 1988; Calvo et al., 1995}. Subtilisin is a serine e n d o p e p t i d a s e (MW 27500) t h a t is secreted in large a m o u n t s from a wide variety of Bacillus species. The large d a t a b a s e for subtilisin m a k e s it a n attractive model enzyme to elucidate s t r u c t u r e - f u n c t i o n relationships. The lipase p r o d u c e d by Candida rugosa (formerly C andida cylindracea ) is being extensively u s e d in research, owing to its high activity in hydrolytic as well as in synthetic reactions (Otero et al., 1990). This y e a s t p r o d u c e s two m a i n isoforms (lipases A a n d B) with similar a m i n o acid content, N-terminal s e q u e n c e a n d molecular weight, w h i c h differ in n e u t r a l s u g a r content, hydrophobicity, stability We thank Comunidad de Madrid and Fundaci6n Caja de Madrid for financial support. This research was financed by the Spanish CICYT (project BIO96-1574-CE).

116 and substrate specificity (R~a et al., 1993). Our group h a s developed a novel procedure for the purification of lipases A and B based on hydrophobic interaction chromatography (Rua and Ballesteros, 1994). In the present work, and imitating the hydrophobization that takes place in n a t u r e for m a n y proteins, we have lowered the HLB of subtilisin Carlsberg - a protein containing 9 L y s + the amino terminal (Ala)- by chemical acylation using acyl chlorides of different chain length. In addition, we have modified lipases A and B from C. rugosa by covalent coupling of their amino groups with PEG activated by two different m e t h o d s (see Fig. 1). The effect of chemical modification on the activity and stability of the corresponding enzymes was further analyzed.

CH'(OCH'CH2'"O c' O--NH,

CH3 (OCH2CH2)nO~_ .... N

CH3 (OCH2CH2)n O ' ~ ~ N

OH3 (OOH2OH2)nO ~

O

CHa(OCH2CH2)nO~O

(~NH

~

2

+ HCI (I)

0

~~'--NH~)

CHz(OCI'I2CH2)nO~;--NH-~

+ HO--

([B

O-~

o CH3 (CH2)n CII--Cl

o ~--- NH2 --- CH3 (CH2)nCII--NH~[~

+ HCI (IH)

Figure 1. Scheme of the chemical modification of enzymes with: (I) methoxypolyethylene glycol activated with triazine (TZ-PEG); (II) methoxy-polyethylene glycol activated with succirdmidyl carbonate (SC-PEG); (III) fatty-acid chlorides.

2. MATERIALS AND METHODS Subtilisin Carlsberg (protease type VIII from B. licheniformis}, palmitoyl and octanoyl chlorides, sodium cholate, p-nitrophenyl butyrate (pNPB), tributyrin, Succ-Ala-Ala-Pro-Phe-pNA, methoxy-polyethylene glycol (MW 5000) activated with cyanuric chloride (TZ-PEG) a n d C. rugosa lipase were from Sigma. Methoxypolyethylene glycol (MW 2000 and 5000) activated with succinimidyl carbonate (SC-PEG) was from Calbiochem. 2,4,6-Trinitro-benzenesulfonic acid and PMSF were from Fluka. Phenyl-Sepharose and S e p h a d e x G - 5 0 were supplied by Pharmacia. Sodium dodecyl sulfate (SDS) was from Serva.

2. I. Lipase p u r i f i c a t i o n Lipase from C. rugosa was purified using an FPLC system (Pharmacia) into isoenzymes A and B following the protocol of Rua and Ballesteros (1994) with Phenyl-Sepharose as matrix.

117

2.2. E n z y m e a s s a y s The a m i d a s e activity of subtilisin was d e t e r m i n e d spectrophotometrically at 410 n m in 0.1 M phosphate buffer (pH 7.0), containing 0.1 M Succ-Ala-Ala-Pro-Phe-pNA a s s u b s t r a t e . The hydrolytic activity u s i n g p-nitrophenyl b u t y r a t e a s s u b s t r a t e was followed spectrophotometricaUy at 346 nm. The a s s a y mixture contained 0.32 mM s u b s t r a t e in 50 mM p h o s p h a t e buffer (pH 7.2), including 4% acetone. Lipase activity u s i n g tributyrin was a s s a y e d titrimetrically in a pH-stat {Radiometer) u s i n g 5 0 r a M NaOH. The reaction mixture contained 68 mM tributyrin, 0.1 M NaC1, 0.1 M CaC12, i mM Tris-HC1 (pH 7.2), including 3% acetonitrile. One u n i t of activity (U) is the a m o u n t of enzyme t h a t hydrolyses 1 pmol of s u b s t r a t e / m i n u n d e r the above conditions.

2.3. C h e m i c a l m o d i f i c a t i o n The acylation of subtilisin w a s performed in a p H - s t a t according to a method previously d e s c r i b e d (Plou a n d Ballesteros, 1994). Palmitoyl chloride and octanoyl chloride were employed as acylating agents; the former was u s e d in p r e s e n c e of sodium cholate to p r o d u c e a n h o m o g e n e o u s dispersion. Chemical modification of lipases with activated PEG was carried out using a pH-stat. Lipase (25 mg) w a s dissolved in 6 mL of 50 mM p h o s p h a t e buffer {pH 8.0). Then, PEG w a s a d d e d a n d the mixture m a i n t a i n e d at pH 8.0 and 25~ with 50 mM NaOH. The m o l a r excess of activated PEG referred to the amino g r o u p s in the protein was 10:1. Once the b a s e c o n s u m p t i o n curve leveled off, the u n b o u n d PEG was removed by ultrafiltration - u s i n g an Amicon PM 30 m e m b r a n e - w a s h i n g with 1 mM p h o s p h a t e buffer (pH 7.0). Finally, the resulting PEG-lipase was freeze-dried.

2.4. D e t e r m i n a t i o n of t h e d e g r e e of s u b s t i t u t i o n The c o n t e n t of free lysines before a n d after acylation with fatty acids was determined u s i n g 2,4,6-trinitrobenzenesulfonic acid according to Fields (1971}. In order to avoid the effect of the p e p t i d e s g e n e r a t e d by autolysis of subtilisin, leading to an o v e r e s t i m a t i o n of a m i n o groups, s a m p l e s were inhibited irreversibly with PMSF and c h r o m a t o g r a p h i e d over S e p h a d e x G-50. The c o n t e n t of a m i n o g r o u p s in the protein before and after the chemical modification with PEG w a s d e t e r m i n e d by the fluorescamine m e t h o d , using a Perkin Elmer LS50B spectrofluorometer, as described by Stocks e t al. (1986). F l u o r e s c a m i n e was b o u n d to the free a m i n o groups, and the intensity of the fluorescence e m i s s i o n was m e a s u r e d at 475 n m after excitation at 390 nm.

2.5. A s s a y o f s t a b i l i t y A diluted solution (1-2 ~tM) of enzyme in the corresponding buffer {in absence or p r e s e n c e of SOS) w a s i n c u b a t e d at the desired t e m p e r a t u r e . The r e m a i n i n g activity with Succ-Ala-Ala-Pro-Phe-pNA for subtilisin or with pNPB for C. r u g o s a lipase was m e a s u r e d at intervals. The c o r r e s p o n d i n g curves were fitted to a single exponential (subtilisin, lipase B} or a double exponential decay (lipase A) using the SIMFIT p a c k a g e {Bardsley, 1995}. Buffers employed were 0.1 M p h o s p h a t e (pH 7.0) for subtilisin, a n d 50 mM p h o s p h a t e (pH 7.2) or 0.1 M b i c a r b o n a t e (pH 9.0) for lipases.

118

3. RESULTS AND DISCUSION 3.1. Acylation of subtilisia with fatty acids The kinetic c o n s t a n t s for the hydrolysis of n a t u r a l a n d synthetic s u b s t r a t e s catalyzed by native a n d modified subtilisins were evaluated. Slight differences in kcat a n d Km were o b s e r v e d in the hydrolysis of synthetic s u b s t r a t e s . However, u s i n g n a t u r a l s u b s t r a t e s - s u c h as c a s e i n - the chemical modification p r o d u c e d a d e c r e a s e in the catalytic efficiency, possibly d u e to steric h i n d r a n c e ; the effect w a s more appreciable with palmitoyl-subtflisin c o m p a r e d with octanoyl-subtilisin (Plou a n d Ballesteros, 1994) The stability of acyl-subtilisins in a q u e o u s m e d i u m (pH 7.0) at 45~ a n d 65~ w a s studied. Low enzymatic c o n c e n t r a t i o n s were u s e d in order to minimize autolysis a n d / o r aggregation processes. As s h o w n in Table 1, the h y d r o p h o b i z a t i o n of subtilisin with acyl chlorides (C8 or C16) e n h a n c e s significantly its stability. The stabilization effect (ratio of the t h e r m o i n a c t i v a t i o n c o n s t a n t s of native a n d modified enzymes) is a b o u t 1S-fold at 45~ a n d a b o u t 2.S-fold at 65~ Table 1 Half-lives (t112) of native a n d fatty-acid modified subtilisins at pH.7 Fatty acid

Degree of

tl/2 (h)

acylationl (%)

45~

65~

Native

--

2.4

1.4

Octanoic acid

57

30.4

2.9

Palmitic acid

60

37.2

4.4

1 Subtilisin Carlsberg c o n t a i n s 10 a m i n o residues.

3 . 2 . M o d i f i c a t i o n o f l i p a s e s w i t h PEG Chemical modification of the a m i n o g r o u p s of lipases A a n d B with PEGtriazine (TZ-PEG) a n d PEG-succinimidyl c a r b o n a t e (SC-PEG) w a s carried out. S u b s t i t u t i o n degrees r a n g i n g from 5 0 - 7 0 % were achieved (lipases A a n d B c o n t a i n 25 a n d 21 lysine r e s i d u e s per mol respectively). It is i m p o r t a n t to note t h a t w h e n PEGsooo is activated with triazine, two c h a i n s of PEG are b o u n d to the biocatalyst per single s u b s t i t u t i o n {Fig. 1). The stability of TZ-PEGsooo-lipases A a n d B w a s studied at different v a l u e s of pH. At pH 7.2, the modification does not p r o d u c e a significant stabilization {results not shown). However, at pH 9.0, the TZ-PEGsooo-lipases A a n d B are c o n s i d e r a b l y more stable t h a n the native enzymes. Fig. 2 s h o w s the inactivation curves for the native a n d modified l i p a s e s A a n d B, respectively. The stabilizing effect, in t e r m s of the inactivation c o n s t a n t s , is 9-fold for lipase A a n d 20-fold for lipase B. Similar stabilization effects were f o u n d u p o n modification with SC-PEG. The stability of native a n d chemically-modified p r e p a r a t i o n s of lipases A a n d B i n c u b a t e d with v a r i o u s c o n c e n t r a t i o n s of SDS (0.04-0.5%) w a s also examined. For lipase A, the modification with PEG gives rise to p r e p a r a t i o n s with higher stability

119 (about 4-fold); however, the m e t h o d of activation of PEG w a s less i m p o r t a n t t h a n the degree of s u b s t i t u t i o n in this effect. 100 ~"

,--

80

~

,-,9

o

60

4O =1 .,.-i

m

4 Z L ~

20 0 1

2

3

4

5

6

T i m e (h) Figure 2. Kinetics of e n z y m e inactivation at pH 9.0 o b s e r v e d with lipase A (O), TZPEOsooo-lipase A ([:]), lipase B (O) a n d a n d TZ-PEGsooo-hpase B (tt). The i n a c t i v a t i o n of l i p a s e B at 0 . 2 % SDS w a s so fast t h a t it w a s not possible to d e t e r m i n e a n a c c u r a t e v a l u e for kin. Fig. 3 s h o w s the inactivation c u r v e s for lipase B a n d PEG-lipase B at 0 . 2 % (w/v) SDS, r e p r e s e n t e d in a s e m i l o g a r i t h m i c scale. Fig. 4 s h o w s the i n a c t i v a t i o n c u r v e s for lipase B a n d PEG-lipase B at 0 . 0 4 % (w/v) SDS. At both c o n c e n t r a t i o n s of SDS, the l i p a s e s modified with PEGsuccinimidyl c a r b o n a t e are m o r e stable t h a n t h o s e with PEG-triazine, especially at 0.04% (w/v) SDS. In all c a s e s , the modification with PEG stabilizes significantly lipase B with r e g a r d s to the d e l e t e r i o u s effect of SDS. The b i p h a s i c c h a r a c t e r of the inactivation c u r v e s of s a m p l e s u s i n g SC-PEG m i g h t reflect the p r e s e n c e of p o p u l a t i o n s of PEG-lipase with different d e g r e e s of s u b s t i t u t i o n . 100

> ~

10

a)

x 0

i 25

I

1

1

50

75

I00

Time

125

(min)

Figure 3. S e m i l o g a r i t h m i c plot of the inactivation of lipase B (O} a n d the s a m p l e s modified TZ-PEGsooo (A), SC-PEGsooo (O) a n d SC-PEG2ooo (0) i n c u b a t e d in a q u e o u s solution at 25~ a n d pH 7.2, in p r e s e n c e of 0.2 % (w/v) SDS.

120 1 O0

L.

~

~

.~

I

~

~

i

80 .+a .P,I

.~9 .l.a {j

60 4O

"'~ ~)

ID

20

...

0

1

2

I

3

4

Time

5

6

7

8

(h)

Figure 4. Stability of lipase B (O) and the samples modified TZ-PEGsooo (A), SCPEGsooo (O) and SC-PEG2ooo (0) incubated in aqueous solution at 25~ and pH 7.2, in presence of 0.04 %(w/v) SDS. In conclusion, chemical modification of enzymes with fatty acids or amphipatic polymers such as PEG may be exploited as a means of obtaining preparations with enhanced stability u n d e r harsh conditions: high temperature, alkaline pH and presence of anionic surfactants. The increased resistance of acyl-subtilisin and PEG-lipases to these d e n a t u r a n t s is of great interest from the point of view of structural implications a n d / o r the biotechnological potential of hydrolytic enzymes.

REFERENCES

Bardsley, G.W. (1995) SIMFIT: "A package for simulation, curve-fitting and statistical analysis in the life sciences", Manchester University, U.K. Cairo, M.V., Plou, F.J., Pastor, E. and BaUesteros, A. Biotechnol. Lett. 17 (1995) 171-176. Calvo, M.V., Plou, F.J. and BaUesteros, A. Biocatal. Biotransf. 13 (1996) 271-285. Fields, R. Biochem. J., 124 (1971) 581-590. Otero, C., Pastor, E., Fernandez, V.M. and Ballesteros, A. Appl. Biochem. Biotechnol. 23 (1990) 237-247. Pasta, P., Riva, S. and Carrea, G. FEBS Lett. 236 (1988) 329-332. Plou, F.J. and BaUesteros, A. FEBS Lett. 339 (1994) 200-204. Rfla, M.L., Diaz-Maurifio, T., Fem~_ndez, V.M., Otero, C. and Ballesteros, A. Biochim. Biophys. Acta 1156 (1993) 181-189. Rfia, M.L. and BaUesteros, A. Biotechnol. Tech. 8 (1994) 21-26. Schultz, A.M., Henderson, L.E. and Oroszlan, S. Annu. Rev. Cell Biol. 4 (1988) 611-647. Shami, E.Y., Rothstein, A. and Ramjeesingh, M. Trends Biotechnol. 7 (1989) 186-190. Stocks, S.J., Jones, A.J.M., Ramey, C. W. and Brooks, D.E. Anal. Biochem. 154 (1986) 232-234.

Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborraand P.J. Halling (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

121

M o d i f i c a t i o n o f t h e c a r b o h y d r a t e m o i e t y in r i b o n u c l e a s e B a n d its influence on the protein stability probed by limited proteolysis

Ulrich Arnold and Renate Ulbrich-Hofmann Martin-Luther University Halle-Wittenberg, Department of Biochemistry and Biotechnology, Institute of Biotechnology, Kurt-Mothes-Strat~e 3, D-06120 Halle, Germany

The non-glycosylated ribonuclease A (RNase A), its glycosylated form, ribonuclease B (RNase B), and two partially deglycosylated species of RNase B were compared with respect to their kinetic thermal stabilities determined by limited proteolysis with thermolysin. The modifications of the carbohydrate moiety in ribonuclease B were obtained by treatment with endoglycosidase F and c~-mannosidase, respectively, yielding GlcNAc-RNase and GlcNAc2Mart3-RNase (Man3-RNase). The results reveal that the modified RNase species possess nearly the same stability as RNase B independent of the size of the carbohydrate moiety. In contrast, the rate of proteolysis decreased with increasing volume of the carbohydrate chain due to steric shielding of the primary cleavage sites in the ribonuclease molecule. 1. INTRODUCTION Glycosylation as the mostly occurring natural modification of proteins is suggested to function as traffic marker, signal modifier, or regulator of biological activities (Rademacher et al., 1988; Lis & Sharon, 1993). Moreover, it is known to effect the stability of the protein structure toward denaturants (Jaenicke, 1991) and proteolytic degradation (Rudd et al., 1995). Glycosylation might increase the life time of the protein molecule by prevention of deamidation of Asn side chains which results in a decreased stability of the protein (Catanzano et al., 1997). Ribonuclease A (RNase A) and ribonuclease B (RNase B) possessing identical primary and tertiary protein structures (Berman et al., 1981; Williams et al., 1987) but differing by a carbohydrate chain (GlcNAc2Mans_9) attached to Asn34 in RNase B are an ideal model system for studying effects due to glycosylation. Interestingly, the glycosylation site is situated in the structural region that was identified as the first unfolding region in RNase A under thermal denaturation (Arnold et al., 1996). The location of the unfolding region is not changed due to the attachment of the carbohydrate moiety in RNase B

122 (Arnold et al., 1998) whereas the kinetic as well as thermodynamic thermal stabilities of RNase B are higher than those of RNase A (Arnold & Ulbrich-Hofmann, 1997). In the present paper, the influence of the size of the carbohydrate moiety on the kinetic thermal stability and the proteolytic resistance of the RNase molecule is studied. Therefore, partially deglycosylated RNase B species were prepared and their unfolding and proteolyis rate constants were determined by limited proteolysis with thermolysin.

o Ash34 RNase A

0 GlcNAc

'~ Man

GlcNAcRNase

Man3RNase

RNase B

Figure 1. Schematic presentation of the modification of the carbohydrate moiety in the RNase molecule. 2. MATERIALS A N D METHODS

RNase A and RNase B from Sigma, MO, USA, were purified on a MONO S FPLCcolumn (Pharmacia Biotech, Sweden) using 50mM Tris-HC1 buffer, pH8.0. c~-mannosidase and thermolysin from Sigma, MO, USA, and endoglycosidase F from Boehringer, Germany, were used without further purification. All other chemicals were the purest ones commercially available. Preparation of GlcNAc-RNase and Man3-RNase. After adjusting the pH to 4.5 with 1 M sodium acetate buffer, 1 mg of purified RNase B were treated with endoglycosidase F (0.5 U) or c,-mannosidase (1.0 U) at 35~ for 22 h. The reaction products were purified on a MONO S FPLC-column and identified by MALDI mass spectrometry (Reflex TM Bruker-Franzen, Germany). Thermal denaturation and proteolysis. In a typical experiment, 15 ~tL of thermolysin (0.1-0.8 mg/mL) in 50 mM Tris-HC1 buffer, 10 mM CaC12, pH 8.0, and 15 I~L of RNase (1.0 mg/mL) were added to 120 ~tL of preincubated 50 mM Tris-HC1 buffer, pH 8.0. After distinct time intervals, samples of 18 I~L were removed and mixed with 6 btL of 50 mM EDTA.

123 Electrophoresis and densitometric evaluation. Electrophoresis was carried out on a Midget electrophoresis unit (Hoefer) according to Sch~igger & v o n Jagow (1987) but using 10, 14, and 18% acrylamide for the sampling, spacer, and separating gels. The SDS-PAGE gels were stained with silver nitrate according to Blum et al. (1987) or stained with Coomassie Brillant Blue G250 and scanned at 595 nm using a densitometer CD 60 (Desaga). Determination of the unfolding rate constants. The rate constants of proteolysis (kproteolysis) w e r e calculated for several thermolysin concentrations from the decrease of the peak areas of intact RNases in the scanned SDS-PAGE gels as a function of time of proteolysis according to Arnold and Ulbrich-Hofmann (1997). From the respective pairs of kproteolysis and the thermolysin concentration [P]0, the unfolding rate constants (ku) were estimated in adoption to the approach by Imoto et al. (1986) with

ku [P]0 kproteolysis=

and B + [P]0

lim kproteolysis= ku [Pl0-*~

where B is a constant. 3. RESULTS The carbohydrate moiety of RNase B was shortened by its treatment with endoglycosidae F or ~-mannosidase, which are known to cleave the oligosaccharide chain between the di-N-acetylchitobiose moiety and at Man~l--~X bonds, respectively. The partial deglycosylation and purification procedures of the ribonucleases led to homogenous products in the SDS-PAGE (Fig. 2) which were identified as GlcNAc-RNase and Man3-RNase by MALDI mass spectrometry. Surprisingly, the deglycosylation of RNase B by c~-mannosidase did not lead to Manl-RNase but reaction stopped at the Man3 level. This result is in contrast to the results by Rudd et al. (1994) where, however, different reaction conditions were applied.

Figure 2. SDS-PAGE of RNase B (lane 1), Man3-RNase (lane 2), GlcNAc-RNase (lane 3), and RNase A (lane 4). Lane 5 shows the reference proteins soybean trypsin inhibitor (21 kDa), cytochrom c (12.4 kDa), and bovine pancreatic trypsin inhibitor (6.5 kDa).

124 For comparing the kinetic thermal stabilities of the different glycosylated RNase species, the method described in Arnold and Ulbrich-Hofmann (1997) was applied. Both the naturally occurring RNases and the partially deglycosylated RNases were treated with thermolysin of different concentrations under thermal denaturing conditions (50-60~ In all cases, increasing thermolysin concentrations as well as increasing temperatures led to increased proteolysis rates for the RNases. Even if the proteolytic degradation of the glycosylated RNase species was markedly decelerated by the more or less voluminuous carbohydrate moieties, the rate constants of proteolysis revealed a saturation behavior with increasing thermolysin concentrations. Therefore, the unfolding rate constants could be estimated by extrapolation of the proteolysis rates to infinite protease concentrations (Table 1). Table 1.

Unfolding rate constants (ku) of modified and non-modified RNases (xl0 -4 s-l). T (~ 50 55 60

RNase A

GlcNAc-RNase

Man3-Rnase

RNase B

8.25 + 0.95 39.6 + 2.8 159 + 21

2.39 + 0.17 18.6 + 1.3 92.7 + 10

2.74 + 0.19 21.4 +1.8 77.6 + 5.8

2.66 + 0.17 17.5 + 5.2 76.5 + 10.9

Figure 3. EYRING-plot of the unfolding rate constants of the modified and non-modified RNases ( 9 RNase A, 9GlcNAcRNase, ~ Man3-RNase, and 9RNase B).

Within the experimental error, all the non-RNase A species reveal the same unfolding rate constants, which clearly differ from those of RNase A at all temperatures studied. From the EYRING-plot of the ku-values (Fig. 3) the activation enthalpy AH# and the activation entropy AS# were estimated to be 263 + 5 kcal moN

125 and 508 + 9 cal mo1-1 K -1 for RNase A and 298 + 7 kcal mo1-1 and 608 + 23 cal mo1-1 K -~ for RNase B. Nearly identical values to those of RNase B were obtained for both the modified species. On the other hand, the proteolytic susceptibility of the different glycosylated RNase species was compared under conditions where the concentration of thermolysin and not the rate of unfolding of the RNases limits the rate of proteolytic degradation, i.e. at high temperatures and low protease concentration (Fig. 4). The results demonstrate that the rate of proteolysis is strongly influenced by the modification of the ribonuclease molecule. 100 -

Figure4. Time course of the proteolytic degradation of ribonucleases by thermolysin at a 250:1 ratio ( w / w ) at 65~ The kproteolysis -values are 8.46 x 10 -3 s -1, 7.23 x 10 -3 s-1, 5.13 x 10 .3 s -1, and 4.33 x 10 -3 s-1 for R N a s e A ( - ) , GlcNAcRNase ( 9), Man3-RNase ( a ), and RNase B ( 9).

50

t~- "

10

._.

\

r

~ ..Q 0 (1,}

2

0.5

l_

0.1

0

i

i

i

250

500

750

1000

time (s)

4. D I S C U S S I O N Limited proteolysis proved to be a powerful tool both for the detection of delicate changes of the protein structure due to modifications and the determination of the kinetic stability. The comparison of the four RNase species revealed that the proteolytic degradation which starts at the Asn34-Leu35 and the Thr45-Phe46 peptide bonds (Arnold et al., 1996,1998) is hindered by the glycosylation at Asn34. It is decelerated the more the size of the carbohydrate moiety increases. However, it was possible to estimate the thermal unfolding constants by extrapolation of the kproteolysis-values to infinite protease concentrations. Here, the steric shielding effects were negligible and it was found that all the non-RNase A species possess nearly the same unfolding rate constants and consequently the same kinetic thermal stability. This means that the stability difference between RNase A and RNase B has to be attributed to the attachment of the first carbohydrate unit to Asn34. According to the computer modelling studies by Woods et al. (1994), a hydrogen bond between the side chain of Lys37 and GlcNAcl might be responsible for the stabilization effect of the glycosylation in RNase B.

126 ACKNOWLEDGEMENTS

We thank Dr. A. Schierhom, Martin-Luther University, Halle, for performing MALDI mass spectrometry measurements and the Deutsche Forschungsgemeinschaft, Bonn, Germany, for supporting this work. REFERENCES

Arnold, U., Rticknagel, K. P., Schierhorn, A. & Ulbrich-Hofmann, R. (1996) Thermal unfolding and proteolytic susceptibility of ribonuclease A, Eur. J. Biochem. 237, 862869. Arnold, U. & Ulbrich-Hofmann, R. (1997) Kinetic and thermodynamic thermal stabilities of ribonuclease A and ribonuclease B, Biochemistry 36, 2166-2172. Arnold, U., Schierhorn, A. & Ulbrich-Hofmann, R. (1998) Influence of the carbohydrate moiety on the proteolytic cleavage sites in ribonuclease B, J. Prot. Chem. /7, 397-405. Berman, E., Wakers, D. E. & Alerhand, A. (1981) Structure and dynamic behavior of the oligosaccharide side chain of bovine pancreatic ribonuclease B, ]. Biol. Chem. 256, 3853-3857. Blum, H., Beier, H. & Gross, J. H. (1987) Improved silver staining of plant proteins, RNA and DNA in polyacrylamide gels, Electrophoresis 8, 93-99. Catanzano, F., Graziano, G., Capasso, S. & Barone, G. (1997) Thermodynamic analysis of the effect of selective monodeamidation at asparagine67 in ribonuclease A, Prot. Sci. 6, 1682-1693. Imoto, T., Yamada, H. & Ueda, T. (1986) Unfolding rates of globular proteins determined by kinetics of proteolysis, J. Mol. Biol. 190, 647-649. Jaenicke, R. (1991) Protein stability and molecular adaptation to extreme conditions, Eur. J. Biochem. 202, 715-718. Lis, H. & Sharon, N. (1993) Protein glycosylation. Structural and functional aspects, Eur. ]. Biochem. 218,1-27. Rademacher, T. W., Parekh, R. B. & Dwek, R. A. (1988) Glycobiology, Ann. Rev. Biochem. 57, 785-838. Rudd, P. M., Joao, H. C., Coghill, E., Fiten, P., Saunders, M. R., Opdenakker, G. & Dwek, R. A. (1994) Glycoforms modify the dynamic stability and functional activity of an enzyme, Biochemistry 33, 17-22. Sch~igger, H. & von Jagow, G. (1987) Tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis for the separation of proteins in the range from I to 100 kDa, Anal. Biochem. 166, 368-379. Williams, R. L., Greene, S. M. & McPherson, A. (1987) The crystal structure of ribonuclease B at 2.5-A resolution, ]. Biol. Chem. 262,16020-16031.


127

Comparative studies on selective modification of E-amino groups in lipases and phospholipase A2 Bart C. Koops, Arend J. Slotboom and I-Iubertus M. Verheij Department of Enzymology and Protein Engineering, C.B.L.E., Utrecht University, PO Box 80054, 3508 TB Utrecht, The Netherlands Several lipases and phospholipase A2 (PLA2) were chemically modified with ethyl caprylimidate, the capryl ester of p-hydroxyphenyldimethylsulfonium methyl sulphate (CsDSP-ester), the capryl carbonate of p-hydroxyphenyldimethylsulfonium methyl sulphate (C8DSP-carbonate) and polyethylene glycol monomethyl ether activated with p-nitrophenyl chloroformate (pNP-MPEG). PLA2 could be easily modified to high degrees, whereas modification of lipases was less efficient. The poor modification with the C8-DSP-ester, C8DSP-carbonate, and pNP-MPEG can be explained by hydrolysis of the reagents by the lipases and by impurities present in the crude lipase preparations. Purification of the lipases decreases the hydrolysis rate of the reagent, but the enzymatic hydrolysis of the reagent remains considerable. For purified Candida rugosa lipase the hydrolysis of pNP-MPEG is slow enough to give a relatively high degree of modification when reacted with this reagent. The degree of modification for Candida antarctica lipase B and Staphylococcus hyicus lipases could be enhanced by increasing the pH of the modification reaction and addition of the reversible inhibitor, 2,4-dichlorobenzeneboronic acid. 1. I N T R O D U C T I O N During the past decade, the use of enzymes as biocatalysts has been established as an important tool in organic synthesis. Due to their ability to discriminate between enantiomers they can be utilised as biocatalysts for the kinetic resolution of racemates to produce chiral building blocks for pharmaceuticals and fine chemicals. Furthermore, hydrolytic enzymes can catalyse the reverse reaction in organic solvents. By far the most frequently used biocatalysts in organic synthesis are lipases (triacylglycerol hydrolases, EC 3.1.1.3) because they are relatively cheap, available from many sources and have a broad substrate selectivity. However, the use of lipases in organic synthesis on industrial scale has two major disadvantages. First, enzymes are insoluble in non-aqueous solvents whereas most of the reactants are soluble in these media. For such heterogeneous reactions the activity of the lipase is low and therefore a large amount of the biocatalyst is needed. Second, their stability under conditions used in industrial applications is rather low and the biocatalyst can only be used for a short period of time. These disadvantages make the use of lipases as biocatalyst in industrial processes approximately ten times too expensive. Obviously improving the performance of enzymes in organic media should concern their activity, solubility and stability in these media. An often described strategy is changing the surface properties of the enzymes by covalent coupling of hydrophobic or amphiphilic groups to the e-amino group of lysine residues. Most papers on this subject describe either reductive alkylation with aldehydes (1), amidination with imidates (2), acylation with esters of phydroxyphenyldimethylsulfonium methyl sulphate (DSP-esters) (3) or the introduction of a polyethylene glycol moiety (pegilation) with activated polyethylene glycol monomethyl ethers (MPEG) of different molecular weights (4, 5). All these publications describe an increased activity and thermal or solvent stability in organic media. However, only for pegilated enzymes

128 an increase in the solubility in organic solvents has been described. MPEG was previously introduced by activation with (toxic) cyanuric chloride. As an alternative, milder reagents for pegilation are used, such as MPEG activated with p-nitrophenyl chloroformate (pNP-MPEG) (6), or succinimidyl succinate (suc-MPEG) (7) There are numerous articles on the topic of chemical modification of enzymes, but only a limited number of lipases have been subjected to modification. In this article we describe the modification of several lipases with the four modifying agents depicted in Figure 1. All these reagents react with the E-amino group of lysines and with the m-amino group. Amidination of proteins with imidoesters results in retention of the positive charge at the lysines. Modification of proteins with the DSP-ester, DSP-carbonate, and pNP-MPEG results in a neutral amide linkage for the first or carbamate linkage for the latter two reagents.

J H

100

+ Cl-

C7H15

~8o

O--C2Hs /

,CH3

Xc.~

CH3SO4-

C7Hls-O--CO--O-CsH4-(CH3 CHaSO4CH3

4

-

C

.o_

C7H1s --CO-O--CsH4-S./+

3

100

CH3-(OCH2CH2)110--O--CO--O-CsH4-NO2

Figure 1. Structures of the modification reagents. 1. Ethyl caprylimidate. 2. CSDSP-ester 3. Cs-DSP-carbonate 4. pNPMPEG5000. C6H4 is para substituted phenyl.

80

~

..'g =_60

60 =o ,..e ~.

"~ 40

40 o~

~

"O

e

20 0

20

0

,,,J,,,I,,,i,,,i 20 40 60 molar excess

80

,,,

0 100

Figure 2. Degree of amidination and activity of PLA2 as a function of the molar excess of ethyl caprylimidate. Degree of amidination determined with ninhydrin (o) or with amino acid analysis (e). Residual activity after modification (m)

2. MATERIALS AND METHODS 2.1. Enzymes Candida antarctica lipase B, SP 525 (CALB) and Rhizomucor miehei lipase, Novozym 388 (RML) were kind gifts of Novo Nordisk (Bagsvaerd, Denmark). Cutinase from Fusarium solani pisi and Pseudomonas glumae lipase (PGL) were generously donated by Prof. Dr. M.R. Egmond (Unilever R.L., Vlaardingen, The Netherlands). Candida rugosa lipase, Lipase MY (Meito Songyo, Japan) (CRL) and Pseudomonas fluorescence lipase, Amano P (PFL) were generous gifts of Dr. G.J.M. van Scharrenburg (Solvay-Duphar, Weesp, The Netherlands). Bacillus subtilis lipase (BSL) was a kind gift from Dr K. Schanck-Brodrtick from Universit6 Catholique De Louvain (Louvain-la-neuve Belgium). Geotrichum candidum lipase (GCL) was obtained from Biocatalysts Ltd (England). Porcine pancreatic lipase (PPL) was obtained from Calbiochem (Los Angeles, USA). Porcine pancreatic phospholipase (PLA2) was obtained as described previously (8). Staphylococcus hyicus lipase (SHL) was isolated as described before (9). 2.2. Chemicals Ethyl chloroformate was from Acros Chimica. 4-(Methylmercapto)phenol was obtained from ICN Biomedicals. Dimethyl sulphate, octyl chloroformate and p-nitrophenyl

129 chloroformate were purchased from Aldrich. 2.4-Dichlorobenzeneboronic acid was obtained from Johnson Matthey. Sodium deoxycholate (DOC) was from Merck. Po' ",~thylene glycol monomethyl ether (MW 5000) was purchased from Fluka. Sigma lipase sabstrate (stabilised olive oil emulsion) was obtained from Sigma. All other chemicals were of analytical grade. Ethyl caprylimidate, C8-DSP-ester, and pNP-MPEG5000 were synthesised in analogy to the procedures described by Suydam et al. (10), Kouge et al. (11) and Veronese et al. (6), respectively. The C8-DSP-carbonate was synthesised analogously to the Cs-DSP-ester. Thin-layer chromatography was done on TLC plates from Merck (silicagel DC Fertig). 1HNMR measurements were done on a 300 MHz Brucker or Varian machine. Mass spectrometry data was obtained on a JMS SX 102/102 FAB Spectrometer. Small-scale gel filtration PD-10 columns (Sephadex G-25) were purchased from Pharmacia.

2.3. Preparation and partial purification of the crude lipases All lipases except BSL and RML were dissolved in distilled water, centrifuged to remove insoluble matter, dialysed against distilled water for three days at 40C and lyophilised prior to use (crude lipase preparations). Under these conditions no loss of enzymatic activity occurred. BSL was dialysed against 10 mM glycine buffer pH 10 to prevent precipitation. Because RML contains cellulases it was passed over a PD-10 column for desalting. Partial purification of CALB, RML and CRL was done on a DEAE-cellulose column equilibrated with 25 mM Tris.HCl pH 7.0, 7.5 and 7.0, respectively. BSL was purified on a CM-cellulose column equilibrated with 10 mM sodium borate pH 9.0. All purifications were carried out at 4~ Under these conditions CALB did not bind to the column. RML and CRL were eluted with a linear salt gradient (0 - 0.5 M NaCI). BSL was stepwise eluted with 0.1 M NaCI, 0.5 M NaC1 and finally with 10 mM NaC1 in 10 mM sodium borate at pH 12. All fractions with lipase activity were combined, dialysed against distilled water or 10 mM glycine buffer pH 10 and lyophilised. PPL was purified according to the procedure described by Verger et al. (12). After purification the CALB, RML, CRL and BSL preparations contained approximately 30-50% lipase (w/w) and PPL contained more than 90% lipase (w/w) based on SDS-PAGE.

2.4. Modification of PLA2 and lipases PLA2 was amidinated by the procedure adapted from Wofsy and Singer (13). Phospholipase A2 solutions (0.2%) in 0.1 M borate buffer at pH 8.5 were reacted with ethyl caprylimidate at different molar ratios of the reagent over the protein amino groups (Figure 2) at room temperature. The pH of the reaction mixture was re-adjusted to 8.5 with 2 M NaOH. After 4 hours the remaining lipolytic activity was measured and the reaction mixture was passed twice over a PD-10 column to remove excess reagent and hydrolysis products. The degree of modification was determined with ninhydrin. Alternatively, the reaction mixtures were dialysed 16 hours against 50% DMF, 24 hours against distilled water and prepared for amino acid analysis. The modified enzyme preparation was stored at -20~ The lipases were modified similarly, except that after modification the excess reagent was not removed. RML was amidinated at 0~ to prevent proteolytic degradation. Phospholipase A2 and lipases were acylated with the C8-DSP-ester and Cs-DSP-carbonate according to the procedure of Kawasaki et al. (14) with minor changes. To 900 lxl of the protein solution (0.56 mg/ml in 50 mM pyrophosphate buffer, pH 8.0), 100 ~tl of the DSP-ester (or DSP-carbonate), dissolved in the same buffer, was added at 40C. Based on the concentration of the amino groups, different molar excesses of the reagent were added (Table 2). After 24 hours for the DSP-ester modification, and 3 hours for the DSP-carbonate modification the degree of acylation was determined with ninhydrin. The proteins were pegilated as described by Veronese et al. (6). Protein solutions (1 mg/ml) in 50 mM pyrophosphate buffer at pH 8.0 or 50 mM borate buffer pH 8.8 were reacted with activated pNP-MPEG5000 at different molar ratios of the reagent over the protein amino groups (Table 2). After 24 hours reaction at 4~ the degree of modification was measured with ninhydrin. In case that a reversible inhibitor was used during the modification, 20 mM of 2,4dichlorobenzeneboronic acid was added to the lipase solution prior to addition of the reagent.

130

2.5. Determination of the degree of modification. The concentration of amino groups before and after modification was determined in triplicate with ninhydrin (15-17). Therefore, 50 gl of protein solution (1-2 mg/ml) was mixed in a 1.5 ml eppendorf vial with 100 gl distilled water, 50 gl 4 M acetate buffer pH 5.1 and 200 I.tl ninhydrin reagent. The mixture was vortexed and heated at 100*C for 15 minutes, cooled at 100C for 10 minutes and diluted with 1 ml 50% ethanol. After vortexing, the eppendorfs were centrifuged at 14000 rpm and the absorbance at 570 nm was measured against a blank. The extinction coefficient of ninhydrin was determined with a glycine solution of known concentration. The degree of modification was calculated as (1 - [NH2]after / [NH2]before). 100. Alternatively, the amidinated proteins were dialysed in 50 % DMF overnight followed by dialysis against distilled water for 24 hours in stead of the PD-10 column. The dialysed samples were hydrolysed in constant boiling HCI and analysed by amino acid analysis. In this case the degree of modification (%) was calculated as (1 - [lys]after/[lys]before).100.

2.6. Enzymatic activity measurements pH-Stat measurements: (Phospho)lipase activity was assayed at 40"C and pH 8.0 with a pHStat titration set (Radiometer, Copenhagen, Denmark) consisting of a TTTld pH meter, a SBR2c recorder and an ABU12 autoburette containing a 20 mM NaOH solution. Four assay conditions were used. (i) Tributyrin / Triton: the reaction mixture consisted of 120 mM Triton X100, 60 mM tributyrin, 100 mM NaCI, 5 mM CaC12 and 5 mM Tris.HC1. (ii) Tributyrin / DOC: the reaction mixture consisted of 110 mM tributyrin, 2 mM DOC, 150 mM NaC1 and 5 mM Tris.HCl. (iii) Sigma lipase substrate (stabilised olive oil emulsion 50% v/v): the reaction mixtul:e consisted of Sigma lipase substrate, 6 times diluted in 50 mM NaC1, 10 mM CaC12 and 5 mM Tris.HC1. (iv) Egg-yolk lecithin: the reaction mixture consisted of 20 ml 6 mM DOC plus 10 ml of the following mixture: one egg-yolk in 130 ml 20 mM CaC12. One (phospho)lipase unit is defined as the amount of enzyme which liberates 1 grnol of fatty acid per minute. Spectrophotometric assay: Lipase activity was assayed at room temperature with 250 gM acyl p-nitrophenyl esters (pNP-ester) in 100 mM Triton X100, 50 mM NaC1, 5 mM CaC12 and 5 mM Tris.HC1 at pH 8. The pNP-esters were dissolved in acetonitrile prior to mixing with the Triton solution (1% acetonitrile in the assay). For cutinase and CALB p-nitrophenyl butyrate was used as substrate, for SHL, CRL and RML p-nitrophenyl caprylate was used. To 1 ml substrate solution, 10 gl of lipase solution was added. The release of p-nitrophenoxide ion following lipase hydrolysis of the substrate is recorded at 400 nm using a Shimadzu UV 1205 spectrophotometer connected with a Shimadzu U-125MU recorder. The apparent molar extinction coefficient of the p-nitrophenoxide ion in the presence of 100 mM Triton X 100 was found to be 10.000 M-1.era-1. One lipase unit was defined as the amount of lipase which liberates 1 ~nol ofp-nitrophenoxide ion per minute. The rate of hydrolysis of the DSP-ester, DSP-carbonate and pNP-MPEG by lipases was measured in a similar way. The reaction mixture contained the reagent (150-300 ~IVI) in 50 mM pyrophosphate buffer (pH 8.0). Upon addition of lipase the release of DSP or p-nitrophenol was followed by measuring the increase in absorbance at 269 or 400 nm, respectively. The release of DSP or p-nitrophenol due to chemical modification is negligable since the amount of protein in the activity test is very low.

2.7. Inhibition of lipases by 2,4-dichlorobenzeneboronic acid. Lipase activity was measured with p-nitrophenyl esters as described above. In the assay 10 gl of the inhibitor (I) in methanol was added before adding the enzyme. The inhibitor concentration was varied between 100 nM and 1 raM. The reciprocal hydrolysis rates were plotted against the inhibitor concentration and the data were fitted to equation (1). The intercept with the x-axes represents the negative value of Kiapp.

• =

1 +

1 )

KM

1

"is] Vmax + (Vmax "IS]"

1 ).[I]

(1)

131 3. R E S U L T S

3.1. Amidination of PLA2 and lipases with ethyl caprylimidate PLA2 was reacted with ethyl caprylimidate in a molar excess (mol / mol amino group) ranging from 2 to 100. By taking samples at different time intervals and measuring the concentration of amino groups, we determined that after 4 hours the reaction reached a maximum (data not shown). The results of amidination of PLA2 are depicted in Figure 2. Obviously, increasing the molar excess of imidate results in an increase in amidinated amino groups with a maximum at 80%. The difference in the degree of amidination found with the ninhydrin method and amino acid analysis is probably due to selective loss of modified PLA2 during the purification of the reaction mixture on the PD-10 columns in the first method. It was found that PLA2 with a high degree of modification sticks better to the PD-10 column than PLA2 with a low degree of modification. The rapid decrease of the PLA2 activity with increasing amidination is not surprising, since modification of the o~-amino group inactivates PLA2 (18, 19). For the modification of crude lipase preparations it was assumed that the average reactivity of amino groups in different proteins would be similar. Therefore, the lipases were reacted with a 2, 25 and 50 fold molar excess of ethyl caprylimidate , to obtain lipases of which approximately 25, 50 and 75% of their amino groups were amidinated. To test this assumption we subjected the lipases modified with a 50 fold molar excess to amino acid analysis. It was found that 25-32% of the amino groups were amidinated (Table 1). While PLA2 lost 92% of its activity upon modification, most lipases remained active. 3.2. Modification of PLA2 and lipases with Cs-DSP-ester,

and pNP-MPEGsooo.

Cs-DSP-carbonate

PLA2 was modified with Cs-DSP-ester, C8-DSP-carbonate and pNP-MPEGs000 at 4~ During 48 hours samples were taken to determine the degree of modification. For the modification reaction of the C8-DSP-ester and pNP-MPEGs000, no increase in the degree of modification was found after 24 hours. For the modification with Cs-DSP-carbonate the reaction reached a maximum after 3 hours. Obviously, an increase in the molar ratio of reagent over protein amino groups resulted in an increase in the degree of modification (Table 2). As was found for the amidination of PLA2, the activity decreased with increasing acylation in a similar way (data not shown). Table 1. Degree of amidination and lipase activity before and after modification with a 50 fold molar excess of ethyl caprylimidate. activity Enzyme amidination e native enzyme .... amidinated enzyme (%) (O/m~) (U/ml~) PLA2 79 1200d 96d CRL 29 39 b 6b GCL _f 27 c 1c CALB 25 181 b 200 b PPL 28 2470 b 552 b PFL _f 63 a 40 a PGL 32 458 a 314 a RML _f 98 c 82 e BSL 27 8b 5b cutinase 32 532 a 236 a a Tributyrin/Triton assay, b Tributyrin/DOC assay, c Sigma lipase substrate, d Egg-yolk assay. e The degree of amidination was measured with amino acid analysis, f The concentration of lysine after modification could not be determined accurately

132 Table 2. Degree of modification of PLA2 and (purified) lipases with Cs-DSP-ester, Cs-DSPcarbonate and pNP-MPEG. The number of lysine residues for each enzyme molecule are: PLA2: 9, CRL: +_21, PPL: 22, CALB: 9, RML: 7, SHL: 32 and cutinase: 6. Enzyme molar excess Degree of modification (%) Cs-DSP-ester Cs-DSP-carbonate pNP-MPEG50o0 PLA2 2 16 17 n.d. PLA2 5 30 47 44 PLA2 10 65 n.d. n.d. PLA2 20 79 n.d. 79 CRL* 5 0 0 n.d. CRL* 20 0 0 n.d. CRL 20 4 n.d. 48 PPL* 5 6 9 n.d. PPL* 20 8 11 n.d. PPL 20 35 n.d. n.d. CALB* 5 0 2 n.d. CAI.~* 20 10 12 n.d. RML 20 6 n.d. n.d. SHL 5 0 0 4 SHL 20 0 0 15 cutinase 5 n.d. n.d. 1 cutinase 20 n.d. n.d. 10 * Crude lipase preparations, n.d. = not determined. Subsequently, lipases were modified with the three reagents under the same conditions as PLA2. Compared to PLA2, the same molar ratio of reagent over amino groups resulted in lower degrees of modification (Table 2). A possible reason for this lower degree of lipase modification could be that the lipases are able to hydrolyse the modifying reagents or that impurities in the crude lipase preparation destroy the reagent. Therefore, we measured the hydrolytic activity of the lipases towards these reagents before and after partial purification. We found that after purification the rate of hydrolysis was lower, but that the partially purified lipases were still able to hydrolyse the reagents (Table 3). Modification of partially purified CRL and PPL with the Cs-DSP-ester resulted in a low (4%) and substantial (27%) increase in the degree of modification. The hydrolysis rates of some lipases presented in Table 3 seem very low but under the conditions used for modification these activities are too high. CRL for example, with an activity of 0.2 U/mg towards the Cs-DSP-ester, degrades 50% of a 20 fold molar excess of this reagent in 10 minutes. Pegilation of CRL, which has a low rate of hydrolysis towards pNP-MPEGs000, resulted in the highest degree of modification (48%). We conclude that the chemical modification of most lipases is severely hampered by the capability of the lipases to hydrolyse the reagent before modification can take place. Table 3. Hydrolysis rates of partially purified lipases towards the different reagents and Kiapp values of. 2,4-dichlorobenzeneboronic acid for the lipases. Lipase hydrolysis rate Kiapp .... Cs-DSP-'ester Cs-DSP-carbonate 'pNP-MPEG5000 (U/mg) (U/mg) (U/mg) lxM cutinase 203 10 SHL 112 31 CALB 0.5 0.09 CRL 0.2 0.04 RML 0.3 0.02 The substrate concentrations were 270 lttM for the DSP-ester, and 210 l.tM for pNP-MPEG.

1" 210 0.01 2.5 3 5.5 0.006 180 0.02 1430 160 ~tM for the DSP-carbonate

133 In order to get higher degrees of modification, it is necessary to decrease the activity of the lipases and/or increase the rate of the modification reaction. Increasing the rate of the modification reaction can be done by increasing the pH since the e-amino group of lysine residues are less protonated (and thus more reactive) at a higher pH. However, the pH-range is limited because these reagents are unstable due to alkaline hydrolysis. To decrease the lipase activity, we (reversibly) inhibited several lipases by 2,4-dichlorobenzeneboronic acid and determined their Kiapp values. As can be seen in Table 3 both SHL and CALB have high affinities for the inhibitor. Modification of SHL with pNP-MPEG5000 in the presence of 2,4dichlorobenzeneboronic acid increased the degree of modification only from 15 to 22%. When SHL was modified in the presence of the inhibitor at pH 8.8 the degree of modification substantially increased to 42%. Modification of CALB with the same reagent at pH 8.8 in the presence of the inhibitor resulted in 48% modification. 4. D I S C U S S I O N For the modification of the crude lipase preparations with imidoesters an easy determination of the degree of amidination with ninhydrin is not possible. The reason is that the excess of the imidoester interferes with the ninhydrin reaction, and thus has to be removed. However, upon removing the excess imidoester on a PD-10 column, not only the reagent but other impurities (amino groups) present in the crude lipase preparation are removed as well. Since the degree of amidination is calculated from the difference in the amount of amino groups before and after modification, it is clear that with this strategy large errors are obtained. Our assumption that the results obtained for amidination of PLA2 could be extrapolated to lipases proved to be not entirely true, since lower degrees of amidination were found for the lipases. Modification of lipases with an extremely high molar excess (300 fold) does not lead to a higher degree of modification, most probably because the insoluble hydrolysis products of the imidoester give rise to practical problems. Modification of lipases with the DSP-ester, DSP-carbonate, and pNP-MPEG also gives lower degrees of modification than with PLA2. The reason for this difference is clear, since the lipases and impurities in the lipase preparation are able to hydrolyse the reagents. Even though partial purification of the lipases results in a lower rate of hydrolysis of the reagent, most lipases hydrolyse a substantial part of the reagents before the modification reaction can take place. The relatively high degree of pegilation of CRL in the absence of inhibitor, can be reached because of a low hydrolysis rate. Only the use of a reversible inhibitor in combination with an increase in the pH improved the degree of modification for SHL and CALB. For cutinase, RML and CRL, the affinity for the reversible inhibitor was too low to be used in this way. In literature, modification of lipases with a DSP-ester and pNP-MPEG has only been described for Phycomyces nites lipase (3, 20) and CRL (21-24), respectively. A recently described modification reagent polyethylene glycol succinimidate, which is the most reactive pegilation reagent under neutral conditions (7), has only been described for CRL. We suggest that modification of other lipases with this reagent has its limitations, since this activated ester may be hydrolysed by many lipases. Based on our results, we conclude that setting up a general strategy for lipase modification should be done with (at least partially) purified lipases. Moreover, either a reagent should be used that cannot be hydrolysed by the lipases, or the reaction should be carried out in the presence of a potent competitive inhibitor.


This research was funded by the Dutch Ministry of Economics through IOP-Katalyse. We would like to thank Ruud Dijkman for the synthesis of p-hydroxyphenyldimethylsulfonium methyl sulphate and its esters, Dr. J.J. Kettenes- van den Bosch and C. Versluis for recording the NMR and Mass spectra, and H. Ravestein and P. Overkamp for the amino acid analyses. REFERENCES

Basil, M., Ampon, K., Wan Yunus, W. M. Z., Razak, C. N. A., and Salleh, A. B. J.

o

Mol. Catal. B: Enzym. 3, (1997) 171-176. .

.

.

5. 0

.

8. .

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

Basri, M., Ampon, K., Yunus, W. M. Z., Razak, C. N. A., and Salleh, A. B. J. Am. Oil Chem. Soc. 69, (1992) 579-583. Murakami, M., Kawasaki, Y., Kawanari, M., and Okai, H. J. Am. Oil Chem. Soc. 70, (1993) 571-574. Inada, Y., and Matsushima, A. Biocatalysis 3, (1990) 317-328. Matsushima, A., Kodera, Y., Hiroto, M., Nishimura, H., and Inada, Y. J. Mol. Catal. B: Enzym. 2, (1996) 1-17. Veronese, F. M., Largajolli, R., Boccfa, E., Benassi, C. A., and Schiavon, O. Appl. Biochem. Biotechnol. 11, (1985) 141-152. Bremen, U., and Gais, H. J. Tetrahedron-Asymmetry 7, (1996) 3063-3066. Nieuwenhuizen, W., Kunze, H., and de Haas, G. H. Methods Enzymol. 32B, (1974) 147-154. Simons, J.-W. F. A., Boots, J.-W. P., Slotboom, A. J., and Verheij, H. M. J. Mol. Catal. B: Enzym. 3, (1997) 13-23. Suydam, F. H., Greth, W. E., and Langerman, N. R. J. Org. Chem. 34, (1969) 292296. Kouge, K., Koizumi, T., Okai, H., and Kato, T. Bull. Chem. Soc. Jpn. 60, (1987) 2409-2418. Verger, R., de Haas, G. H., Sarda, L., and Desnuelle, P. Biochim. Biophys. Acta 188, (1969) 272-282. Wofsy, L., and Singer, S. J. Biochemistry 2, (1963) 104-116. Kawasaki, Y., Murakami, M., Dosako, S., Azuse, I., Nakamura, T., and Okai, H. Biosci. Biotechnol. Biochem. 56, (1992)441-444. Moore, S., and Stein, W. H. J. Biol. Chem. 176, (1948) 367-388. Moore, S., and Stein, W. H. J. Biol. Chem. 211, (1954) 907-913. Hirs, C. H. W. Methods Enzymol. 11, (1967) 325-329. Abita, J.-P., and Lazdunski, M. Eur. J. Biochem. 30, (1972) 37-47. Slotboom, A. J., and de Haas, G. H. Biochemistry 14, (1975) 5394-5399. Takahashi, Y., Tanaka, K., Murakami, M., Kawanari, M., Kawasaki, Y., Tatsumi, K., and Okai, H. Biosci. Biotechnol. Biochem. 59, (1995) 809-812. Basil, M., Salleh, A. B., Ampon, K., Yunus, W. M. Z., and Razak, C. N. A. Biocatalysis 4, (1991) 313-317. Basil, M., Ampon, K., Wan Yunus, W. M. Z., Razak, C. N. A., and Salleh, A. B. J. Chem. Technol. Biotechnol. 64, (1995) 10-16. Hem~iz, M. J., S~nchez-Montero, J. M., and Sinisterra, J. V. Biotechnol. Tech. 10, (1997) 917-922. Hem~iiz, M. J., S~chez-Montero, J. M., and Sinisterra, J. V. Biotechnol. Bioeng. 55, (1997) 252-260.

Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.U Iborra and P.J. Hailing (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

r

135

Analysis of the thermal deactivation kinetics of modified by chemoenzymatic glycosylation Maria Asunci6n LONGO ~ and Didier COMBES b

"Department of Chemical Engineering, University of Vigo, Lagoas-Marcosende, 36200 Vigo, Spain. bDrpartement de Grnie Biochimique et Alimentaire, UA-CNRS 544, Institut National des Sciences Appliqures, Complexe Scientifique de Rangueil, 31077 Toulouse Crdex, France.

I. INTRODUCTION Enzyme stability is a crucial factor to determine whether application of biocatalysis will be commercially successful. Catalytic proteins loose part of their activity when they are subjected to the action of heat, extreme pH or proteases. During the last decades, much research has focused on the improvement of enzymes behaviour in the conditions in which they were to be used, and especially on the increase of their thermal stability. The production of heat-resistant enzymes would allow carrying out enzymatic reactions at higher temperatures, and therefore, increasing conversion rates and substrates solubility and reducing the risk of microbial growth and the viscosity of the reaction medium [1]. Several strategies have been proposed: use of soluble additives [2], immobilisation [3], protein engineering [4] and chemical modification [1, 5]. The covalent binding of appropriate modifiers to the enzyme's surface residues appears as the most popular approach, and very interesting results have been reported. Protein surface hydrophilic/hydrophobic balance seems to play a major role on enzyme thermal stability [6]. Non polar aminoacids are organised on protein surface as hydrophobic clusters, whose distribution may determine properties such as solubility, activity, specificity and stability towards pH, temperature or proteolysis [ 1]. Thus, the modification of protein surface characteristics by chemical binding of modifiers appears as a good strategy to improve biocatalyst performance. Among the investigated modifiers, polysaccharides are outstanding for the improvements in enzyme properties achieved. [7, 8]. Several strategies may be used to assess enzyme thermal stability [9]. Among them, determination of half-life time at a certain temperature is often chosen, since it provides information about enzyme residual activity, an important parameter regarding industrial applications of biocatalysts. Nevertheless, when the effects of enzyme surface modification on thermal stability are investigated, a detailed study of enzyme deactivation kinetics could be necessary. Some enzymes show exponential decay of activity with time, while others present different non-exponential kinetics [10]. A number of models have been proposed to explain the mechanism of enzyme thermal deactivation [ 10-12].

136 In the present work, a-chymotrypsin has been modified by attachment of glycosidic chains of various lengths. This modification has been carried out by a procedure combining chemical and enzymatic reactions [13], and it induced a global increase in the surface hydrophilic character of the protein. The effects of enzyme surface glycosylation on thermal stability in aqueous environment have been investigated, considering both the variation in halflife times, and the study of the deactivation kinetics. So, both half-life times and kinetic and thermodynamic deactivation parameters could be compared, in order to verify whether the results were coincident or not. In some cases, the deactivation parameters pointed out certain amelioration in enzyme thermostability during one step of the deactivation mechanism, even if the observation of half-life times had indicated a decrease in thermostability.

2. EXPERIMENTAL

2.1. Material Bovine pancreas a-chymotrypsin (EC 3.4.21.1), CNBr and N-acetyl-L-tyrosine ethyl ester (ATEE) were obtained from Sigma. Radioactive labelled sucrose was from HEN Research Products. Levansucrase from Bacillus subtilis (EC 2.4.1.10) was kindly provided by Eridania Beghin-Say. All other reagents were analytical grade. 2.2. Assay of a-chymotrypsin hydrolytic activity The substrate utilised for assay of a-chymotrypsin hydrolytic activity was N-acetyl-Ltyrosine ethyl ester (ATEE), as previously reported [14]. 2.3. Chemoenzymatic glycosylation of a-chymotrypsin Glycosylation of the enzyme was carried out by a two step chemoenzymatie technique previously developed [ 13]. This method involves chemical binding of sucrose molecules to the enzyme surface in a first step, followed by a second step in which a glycosyltransferase transfers fructose residues from a donor to the enzyme-bound glycosidic moieties, lengthening them in a progressive way. The levansucrase from B. subtilis has been utilised as glycosidic chain lengthening enzyme. a-chymotrypsin in control solutions and samples was separated from residual reagents by anion exchange chromatography, using a Mono-Q HR 5/5 column connected to an FPLC (Fast Protein Liquid Chromatography) system, both from Pharmacia LKB. The column was eluted at 1 ml/min with a gradient of 0.02 M piperazine buffer pH 9.5 as eluent A and the same buffer containing 1 M NaCI as eluent B. Protein fractions were collected, dialysed and freeze-dried. Glycosylation degrees were determined by radioactive labelling [ 13]. 2.4. Determination of surface hydrophilic character of the enzymes Native and modified enzymes hydrophilicity was evaluated from the reversed phase chromatography profiles obtained in an HPLC Hewlett Packard 1090 equipped with a Nucleosil C18 5 ttm column (220x2.1 nun, SFCC). Following sample injection, the column was eluted at 0.5 ml/min with a gradient of H20/0.1% TFA as eluent A and ethanol/0.1% TFA as eluent B. After a 5 min step of 100% A, a linear gradient from 40% to 100% B in 15 rain

137 was applied. Eluted proteins were evaluated from the on-line measurement of the optical density at 280 nm. 2.5.

Thermal stability of the enzymes

2.5.1. Enzyme thermal deactivation curves Samples and controls were diluted to the desired concentration (5 g/l) in 10.4 M HCI, in order to minimise the risk of proteolysis. Tight-closed tubes containing 0.5-ml aliquots were placed in a water bath at 50~ taken out at regular time intervals and cooled down in ice. The residual activities were measured and referred to the initial values. 2.5.2. Determination of enzyme deactivation kinetic parameters The experimental enzyme deactivation data were fitted to modelled theoretical curves using a non-linear regression procedure based on the Marquardt-Levenberg method of iterative convergence included into the Sigmaplot 1.02 (1994) software. All the relevant deactivation parameters were calculated.

3. RESULTS AND DISCUSSION The well-known bioeatalyst a-chymotrypsin has been modified, using a chemoenzymatic glycosylation method, in order to produce several enzyme forms glycosylated at various degrees. The extent of sugar binding achieved [13] ranged from 12 to 85 moles of grafted monosaccharide per mol of enzyme (0.08 to 0.55 g g'~). The hydrophilic/hydrophobic balance of the produceA enzymes was assessed from their reversed phase chromatography profiles. Attachment of glycosides brought about a decrease in elution times, which could be related to a progressive increase in the global hydrophilic character of the biocatalyst with glyeosylation. Then, thermal stability of some selected modified enzymes was thoroughly investigated, and compared to that of the native form. The evaluation of half-life times (Table 1) showed a very slight increase in thermostability for the less glycosylated forms (half-life times 5-10% higher than the control), and a decrease for the others, a-chymotrypsin bound to 19 moles of monosaccharide per mol of enzyme (0.123 g.g'~) showed a half-life time 28% lower than the non-modified form. When higher glycosylation degrees were obtained, half-life times decreased remarkably (up to 80% in some cases). Nevertheless, a more detailed study of the thermal denaturation patterns for glycosylated a-chymotrypsin provided further information. The deactivation curves indicated a variable fractional rate of deactivation, and therefore, pointed out that more than one type of bond must be broken before a total loss of activity occurs (Figure 1). The results were fitted to a series-type enzyme deactivation model, which involves two first-order steps with one active precursor (El) and a final enzyme state (E2) with possible nonzero activity [ 15]. E

kl .~ E am

l~ .~Ea2

138

This model assumes that E, Et and E2 are enzymatic homogeneous states, with several specific activities, k~ and k2 are the first order deactivation rate constants (min"t) for each step, and oh and a2 are the ratio of specific activities EriE and E2/E respectively. Residual enzyme activity (a) could be expressed as a function of time by the following equation:

a~k~ k2-k;l a,k~ exP("k~t)-(lo~k:~)(czl--m)exp(-k2t)+

a = 1+ l ~ - k ,

~2

Parameters at, kt and k2 for native and modified a-chymotrypsin have been calculated from the experimental data, using a non linear regression method based on a MarquardtLevenberg iterative convergence algorithm. The ratio a2 was considered equal to 0, which means that the final form of the enzyme is totally deactivated [ lO, 15].

- 1,0 0,8

A

1,0

0,6

.~ 0,8

0,4 ~~

"~t

"~176

.-~ 0,6

9

"~176

0,4 i

I

I

0

5

10

I

I

I

15 20 25 Time (min)

'i

I

30

0

5

I

'I

'

I

I

10 15 20 25 Time (rain)

0,2

~7 I

0,1

30

Figure 1. Thermal denaturation of glycosylated a-chymotrypsin A. Increase in half-life times. B. Decrease in half-life times. Moles ofmonosaccharide per mol of enzyme: Q 0 ; A 12; V 19; ~7 45

The thermodynamic parameter standard free energy (AG,~ of the thermal deactivation process has also been calculated for each step of the mechanism, from the equation: -AG,~ RT In [keTCh]

139 where ks is Boltzmann's constant (J K'l), h is Planck's constant (J min), K is the gas constant (J mol ~ K'~),T is the temperature (K) and ki is the deactivation rate constant (min'~). The values of the kinetic and thermodynamic parameters of deactivation are shown in Table 1. Table 1 Half-life times, kinetic and thermodynamic parameters of deactivation for glycosylated achymotrypsin (5 g/l solution in 104 M HCI, 50~ .....

,

Monosaccharide per

Half-life time

enzyme (mol mol"~)

(min)

0

16.2

12

otl

kl

k2

AG~

(min"l)

(min'b

0d morb

0.318

0.0814

3.04x10 "f~

246.3

17.2

0.420

0.115

9.0x10 "l~

242.4

19

11.6

0.412

0.164

2.08x10 "~

251.6

45

1.5

0.277

0.838

0,171

185.9

In some cases, the evolution of thermal deactivation after glycosylation, as indicated by half-life times did not totally agree with the values of the kinetic and thermodynamic parameters of deactivation obtained. For instance, binding of or to 12 moles of monosaccharide per mol of enzyme brought about a slight increase in half-life time, but the deactivation rate constants k~ and k2 increased and the standard free energy decreased after glycosylation. The study of half-life times indicated a stabilisation of the protein after modification, while the thorough study of the deactivation kinetics pointed out, to a certain extent, the opposite effect. On the other hand, the enzyme bound to 19 moles of monosaccharide per tool of protein, whose half-life time decreased in 28% compared to the control, undergone a global stabilisation effect, according to the observed increase in AG~ (= 5 ld mol'~). Furthermore, an increase in the value of r (0.318 for the control and 0.412 for the glycosylatod enzyme) and a decrease in the deactivation rate constant k~, were also detected. The former would imply a higher specific activity of the intermediate form El, and the latter a stabilisation effect during the second step of the deactivation process, which would account for the contradictory trends indicated by half-life time and AG~ results. The other glycosylated form studied (45 moles of monosaccharide per tool of enzyme) showed similar tendencies in half-life times and kinetic and thermodynamic parameters of deactivation. When ot-chymotrypsin was bound to small amounts of glycosides, its thermostability remained nearly constant or underwent certain amelioration, while for higher glycosylation degrees, it decreased remarkably. The present work seems to support the occurrence of a close relationship between the surface hydrophilic character and the thermal stability of enzymes. The mechanisms of thermal stabilisation of enzymes in aqueous environment have been thoroughly studied, and it seems

140 clear that the interactions between protein surface hydrophobic clusters and surrounding water play a key role in the heat-induced denaturation of biocatalysts [16]. Some researchers have proposed the defivatization of enzyme surface with hydrophilic molecules as a way to protect the non-polar clusters from contact to water, and therefore to increase thermal stability [1718]. The modifiers should trap the surrounding water molecules and create a protecting shield to the hydrophobic clusters. Nevertheless, it has to be considered that these hydrophilic modifiers could also attract water, and facilitate its contact with enzyme surface instead of protecting it against damaging interactions. This hypothesis would support the results presented here, in which the significant increase in hydrophilic character of the enzyme surface by glycosylation brought about a decrease in half-life times. Nevertheless, phenomena provoked by enzyme surface glycosylation appear to be quite complex. It is very difficult to know the protein sites to which glycoside molecules have been bound, and therefore the interactions induced. A thorough study of the enzyme deactivation kinetics will provide a more detailed information about the consequences of the modifications, and could reveal secondary stabilisation or destabilisation effects. In the present work, a stabilisation of the enzyme during the second step of the thermal deactivation process has been detected for some degrees of modification, in spite of the lowering in half-life times. This effect could be due to the binding of some molecules to sites in which advantageous interactions would occur.

REFERENCES 1. Mozhaev, V.V., Berezin, I.V. and Martinek, K. CRC Crit. Rev. Biochem., 173 (1988) 147. 2. Schmid, R.D. Adv. Biochem. Eng., 12 (1979) 41. 3. Monsan, P. and Combes, D. Methods Enzymol., 137 (1988) 584. 4. Gupta, M.N. Biotechnol. Appl. Biochem., 14 (1991) 1. 5. Roig, M. G. and Kennedy, J. F. CRC Cfit. Rev. Biotechnol., 12 (1992) 391. 6. Dill, K. A. Biochemistry, 29 (1990) 7133. 7. Lenders, J.P. and Crichton, R.R. Biotechnol. Bioeng., 26 (1984) 1343. 8. Marshall, J.J. TIBS, 3 (1978) 79. 9. Rupley, J. A., Gratton, E. and Careri, G. TIBS, 8 (1983) 18. 10. Henley, J.P. and Sadana, A. Enzyme Microb. Technol., 14 (1985) 42. 11. Lumry, R. and Eyring, H. J. Phys. Chem., 58 (1954) 110. 12. Gianfreda, L., Marrucci, G., Gfizzuti, N. and Greco, G. Biotech. Bioeng., 27 (1985) 877. 13. Longo, M.A. and Combes. D. J. Mol. Catal. B: Enzymatic, 2 (1997) 281. 14. Lozano, P., Combes, D. and lborra, J.L.J. Biotechnol., 35 (1994) 9. 15. Sadana, A. and Henley, J.P. Biotech. Bioeng., 30 (1987) 717. 16. Mozhaev, V. V. and Martinek, K. Enzyme Microb. Technol, 6 (1984) 50. 17. Mozhaev, V. V., Siksnis, V. A., Melik-Nubarov, N. S., Galkantaite, N. Z., Denis, G. J., Butkus, E. P., Zaslavsky, B. Yu., Mestechkina, M. and Martinek., K. Eur. J. Biochem., 173 (1988) 147. 18. Vankovfi, H., Pospisilov~, M., Tich~ M. and Tukovfi, J. Biotechnol. Techniques, 8 (1994) 375

StabilityandStabilizationof Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

141

Increasing the operational stability of flavoproteins by covalent cofactor binding Willem J. H. van Berkel, Robert H. H. van den Heuvel, Marco W. Fraaije and Colja Laane* D e p a r t m e n t of Biomolecular Sciences, Laboratory of Biochemistry, Wageningen Agricultural University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands

The active site residue Asp170 of the covalent flavoprotein vanillylalcohol oxidase was selectively changed by site-directed mutagenesis. Glu170 and Ser170 replacements resulted in flavinylated vanillyl-alcohol oxidase variants. However, substitution of Asp170 by Asn prevented covalent flavinylation without affecting flavin binding. The creation of flavindissociable mutant enzymes provides challenging possibilities to address the role of covalent flavinylation in the operational stability of vanillyl-alcohol oxidase.

1.

INTRODUCTION

Vanillyl-alcohol oxidase (VAO, EC 1.1.3.13) from Penicillium simplicissimum is a covalent flavoprotein acting on a wide range of lignin-derived phenolic c o m p o u n d s (1). VAO-mediated reactions of biotechnological relevance include the production of vanillin from creosol, the conversion of eugenol to coniferyl alcohol (2), and the enantioselective hydroxylation of short-chain 4-alkylphenols (3). The crystal structure of VAO, the first structure of a histidyl-FAD containing flavoenzyme (4), and sequence alignments suggest that VAO belongs to a new family of structurally related flavoenzymes (5,6). Many members of this family contain a histidyl-flavin, a feature shared by less than 5 % of all flavoproteins.

* This research was performed within the framework of the Innovation Oriented Research Program (IOP) Catalysis of the Dutch Ministry of Economy Affairs (project no. IKA96005).

142 The rationale for covalent flavinylation is unclear as is the mechanism by which the covalent bond is formed (7). As the efficiency of commercial flavoenzymes is often hampered by the dissociation of the flavin cofactor under stressful operational conditions, tailor-made covalent flavoenzymes could be of great value for biotechnological applications. Covalent bond formation may i) prevent inactivation of the cofactor, ii) improve resistance against proteolysis, and iii) improve protein stability. We have started to address the mechanism of covalent flavinylation by the characterization of VAO variants. The knowledge gained from these studies will be valuable for the future design of biotechnologically relevant flavoenzymes, made highly stable by the covalent linkage of the cofactor. Here we report on the preliminary characterization of Asp170 variants.

2.

2.1

MATE~ALS AND METHODS

Site-directed mutagenesis The vaoA gene has been cloned into pEMBL19(-) to give pIM3972 (5).

The plasmid pBC11, which is identical to pIM3972 except for a silent mutation at position 882, was used as vector for site-directed mutagenesis. The PstI-SalI fragment from pBC11 was cloned into pUCBM20 yielding a suitable construct for PCR-based mutagenesis. The oligonucleotide 5'-CTTGATGTACCGXXXCTTGGTGGCGGT-3' (where XXX denotes the replacement of GAT for GAG (D170E), TCT (D170S) and AAC (D170N), respectively) was used as primer for the construction of Asp170 variants. Mutated PstI-SalI fragments were excised from pUCBM20 and ligated into P s t I / S a l I digested pBC11. Succesful insertions were confirmed by plasmid sequencing.

2.2

Enzyme purification Enzyme purification was performed essentially as described previously (5). Highly pure Asp170 variants were obtained by introducing Superdex 200 PG 50/1000 gel filtration as final chromatography step. 2.3

Activity measurements Vanillyl-alcohol oxidase (VAO) activity was determined at 25 ~ in airsaturated 50 mM potassium phosphate, 1 mM vanillyl-alcohol, pH 7.5, by measuring the production of vanillin at 340 nm (1). 2.4

Gel electrophoresis SDS-PAGE was carried out as described (8). After electrophoresis the gel was incubated for 5 min in 4% acetic acid. Upon illumination with UV light, the fluorescence of protein-bound 8tx-(N3-histidyl)-FAD can be observed.

143

2.5

Analytical gel filtration

Analytical gel filtration was performed on a Superdex 200 HR 10/30 column using a Pharmacia .~kta system (9). 2.6

Guanidinium hydrochloride (GdmHC1) induced unfolding

Samples of VAO variants (2 ~M) were incubated at 25 ~ in 50 mM potassium phosphate buffer pH 7.5, containing 4 M GdmHC1. Fluorescence spectral analysis was done as described in (10).

3.

RESULTS AND DISCUSSION

3.1

General properties

The Asp170 variants were purified from E. coli TG2 in a three-column procedure, involving hydrophobic chromatography, hydroxyapatite chromatography and gel filtration. Analytical gel filtration established that all Asp170 variants were octameric and their yield after purification was comparable to that of the native recombinant enzyme (5). Illumination of unstained SDS-PAGE gels revealed that D170E and D170S were highly fluorescent, whereas no flavin fluorescence was detected with D170N. In line with this, treatment with 5% trichloroacetic acid yielded yellow protein precipitates for D170E and D170S and a yellow supernatant for D170N. This clearly demonstrates that the Asn170 replacement results in non-covalent flavin binding.

3.2

Catalytic properties

The Asp170 variants were less active with vanillyl-alcohol than the native enzyme. As can be seen from Table 1, almost no activity was observed with D170N. The highest activity was observed with D170E, t~ut the catalytic efficiency of this mutant was considerably decreased in comparison with the native enzyme (Table 1). These results suggest that Asp170 plays a crucial role in substrate conversion, which is in line with the proposed reaction mechanism (4).

144 Table 1 Kinetic parameters of VAO variants.

Enzyme

kc a t

Km

kcat / Km

s-1

~M

M-ls -1

VAO

3.3

160

20630

D170E

1.3

340

3820

D170S

0.01

nd

nd

D170N

< 0.005

nd

nd

nd, not determined

3.3

Flavin binding

D170N was eluted over a Biogel P-6DG column in 50 mM potassium phosphate, 2M urea, 1M KBr, pH 7.5. Under these conditions, and distinct from many other flavoproteins (11), no loss of flavin was observed. This shows that the covalent attachment of the flavin in vanillyl-alcohol oxidase is not essential for tight binding of the cofactor. 3.4

GdmHCl-unfolding

VAO activity is lost in 2 M GdmHC1 (12). When VAO was incubated in the presence of 4 M GdmHC1, a rapid increase in tryptophan fluorescence was observed (Fig. 1A). Moreover, the fluorescence emission maximum shifted from 340 to 355 nm, indicative of protein unfolding. Upon unfolding, no change in fluorescence intensity of p r o t e i n - b o u n d 80~-(N3-histidyl)-FAD was observed (inset Fig. 1A). This is consistent with the low fluorescence quantum yield of 8(x-(N3-histidyl)-FAD at neutral pH (13). When D170N was incubated in the presence of 4 M GdmHC1, a rapid shift in the tryptophan fluorescence emission maximum was followed by a slow increase in tryptophan fluoresence intensity (Fig. 1B). Unfolding of D170N was accompanied by a strong increase of flavin fluorescence (inset Fig. 1B). In analogy to the tryptophan fluorescence, flavin fluorescence increased slowly with time (inset Fig. 1B), suggesting that these secondary changes reflect flavin dissociation.

145

0.5

u.I

o

Z U.l 0 or) uJ IZ:

uJ o 0.4 z u.l

o.8

o

0.6 0.2

0

.-J u.

0.4

ill n"

0.2

-\ ,,

..i

~

n-

~,,,

~ L% %

320

~

450

,. -, _ _~

I

0

0

400

550

600

650

WAVELENGTH tnml

_

i

F ......

360

J

500

] 440

i

I

480

520

560

W A V E L E N G T H (nm)

1

o rr

S

0.5

0.8

~'~

0.6

,"

..',< ", \

/ ~

z 0.4

ix< \ /

,,,0.3

',,,',k

/,-., \

~

\

t

\

0.4 50

0.2 o 320

",

:i ~....._

50

WAVELENGTH (nm)

1

360

400

440

480

520

560

W A V E L E N G T H (nm)

Figure 1. Fluorescence emission spectra of VAO variants. Excitation w a s at 295 n m ( t r y p t o p h a n fluorescence) or 450 n m (flavin fluorescence, inset). (A) native VAO (---) and after incubation for 30 s in 4 M GdmHC1 (--). (B) D170 N (---) and after incubation for 30 s (- - -) or 30 min ( - - ) in 4 M GdmHC1.

146 4.

CONCLUSION

This is the first report on the properties of VAO variants. The results show that the generation of stable flavin-dissociable mutant enzymes is feasible and furthermore suggest that Asp170 is involved in activating the flavin ring for covalent attachment. Future protein-engineering will be directed towards the construction of highly active flavin-dissociable VAO variants. This will allow to study the operational stability of such enzymes in relation to their covalently-bound counterparts. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

12.

13.

M.W. Fraaije, C. Veeger and W.J.H. van Berkel, Eur. J. Biochem., 234 (1995) 271. W.J.H. van Berkel, M.W. Fraaije and E. de Jong, Process For Producing 4-Hydroxycinnamyl Alcohols, European Patent Application 0710289B1 (1997). F.P. Drijfhout, M.W. Fraaje, H. Jongejan, W.J.H. van Berkel and M.C.R. Franssen, Biotechnol. Bioeng., (1998) in press. A. Mattevi, M.W. Fraaije, A. Mozarelli, L. Olivi, A. Coda and W.J.H. van Berkel, Structure, 5 (1997) 907. J.A.E. Benen, P. S~_nchez-Torres, M.J.M. Wagemaker, M.W. Fraaije, W.J.H. van Berkel and J. Visser, J. Biol. Chem., 273 (1998) 7865. M.W. Fraaije, W.J.H. van Berkel, J.A.E. Benen, J. Visser and A. Mattevi, Trends Biochem. Sci., (1998) in press. M. Mewies, W.S. McIntire and N.S. Scrutton, Protein Sci., 7 (1998) 7. M.W. Fraaije, M. Pikkemaat and W.J.H. van Berkel, Appl. Environ. Microbiol., 63 (1997) 435. M.W. Fraaije, A. Mattevi and W.J.H. van Berkel, FEBS Lett., 402 (1997) 33. W.J.H. van Berkel, A.G. Regelink, J.J. Beintema and A. de Kok, Eur. J. Biochem., 202 (1991) 1049. F. Miiller and W.J.H. van Berkel, Chemistry and Biochemistry of Flavoenzymes Vol. I (F. Mfiller, ed.) CRC Press, Boca Raton, Florida (1991) 261. W.J.H. van Berkel, M.W. Fraaije, E. de Jong and J.A.M. de Bont, Flavins and Flavoproteins XI (K. Yagi, ed.) W. de Gruyter, Berlin, New York (1993) 799. E. de Jong, W.J.H. van Berkel, R.P. van der Zwan and J.A.M. de Bont, Eur. J. Biochem., 208 (1992) 651.


147

Properties o f P E G - M o d i f i e d Microbial Proteases. Activity and Stability Studies

T. M. Fattma=, A. Agerlin Olsen', C. C. Fuglsang' and D. Otzenb

"Protein Discovery, Novo Nordisk A/S, Novo All6, 2880 Bagsvaerd, Denmark bDepartment of Biochemistry, University of Lurid, P.O Box 124, S-22100 Lund, Sweden

Attachment of PEG to enzymes has been reported to prevent (auto)proteolysis,' increase solubility and activity in organic solvents and to shield the enzyme from recognition by the immune system. 2 We now report results of activity and stability studies of PEG modified subtilisins. 1. Derivatization Two microbial subtilisins (Sub 1 and Sub2) were modified by attachment of succinimidyl carbonate activated mPEG 5000 in 80-100 times molar excess. The activated mPEG was coupled to the e-amino groups of the lysine side-chains to form stable urethane bonds. 3 0 CH3"{O

0 O--

0 +

NH2-Enzyme

~

CH3..,,O

O

O

n

NH-Enzyme +

HOO

2. Activity The catalytic activity of the PEG conjugates Table 1 towards peptide and protein substrates was Activity against Suc-AAPF-pNA tested and compared to the parent enzymes (figure 1). The Micha~lis Menten kinetic con(s") (M) (M"s "l) stants of the enzymes and derivatives against Subl 7.18E+05 1.160 6.19E+05 Suc-Ala-Ala-Pro-Phe-pNA are summarised in Subl-PEG 5.90E+05 1.052 5.61E+05 Table 1. The PEG-enzymes all have well preSub2 5.97E+05 0.369 1.62E+06 served activity, with l~t and K~ only slightly below the parent enzymes. The specific activ- Sub2?PEG 4.23E+05 0.338 1.25E+06 ity against a protein substrate (dimethyl casein) was typically 10-20% lower than the parent enzyme. This is probably due to steric hindrance near the active site. .

.

.

.

.

.

148

9 Sub1

* Subl-PEG 9 Sub2 9 Sub2-PEG

Activity

0

i--

0.0

I

I

I

i

I

0.5

1.0

1.5

2.0

2.5

IS]

Figure 1 Activity against Suc-AAPF-pNA 3. C o n f o r m a t i o n

To determine whether PEG induces any structural changes in the proteases, circular dichroism spectra of Sub 1 and Subl-PEG were obtained (figure 2). Both near UV and far UV spectra of the two proteins are identical, indicating that neither the tertiary nor the secondary structure are changed to any discernible extent by PEG conjugation. Far-UV CD-spectra

Near-UV CD-spectra 0.25

I

I

I

I

i

I

I

/

E 0.00

'7,

O "7,

O

E -0.25.

~

I

,=

u 25

T

o

E

--

0

to

---Subl-PEG

8O

,

100-

->-9

80 9

l

:::3

9 I

,

'

,

,

,

>,

20

33

loo

30

..

-_>r

80

< ~

60

= r n,'

20

o~

2o

Time (min)

(e)

~

.m

=

I

lo

0

I

lO

Time (min)

(d)

< 60 .~_ 4o ID ~: 20 N o

I

n

40

9

9 9

0

,

0

9

9

9

i

....

9

10

20

Time (min)

100-

>

< _

CU

60 40

U~ ~D

rr'

i

2O

o

I

0

2

|

i i

,

9 |

4

6

Time

(min)

.

.

8

10

Figure 3: The effect of DMS crosslinking on thermal inactivation of cellulases. % residual total cellulase activities are given at 65 ~ (a), 67.5 ~ (b), 70 ~ (c), 72.5 ~ (d) and 75 ~ (e). 0.2 % DMS was used. The data points were the averages of two separate experiments. ( | : crosslinked, e: native.)

156 DMS crosslinking seemed to increase thermostability at 65 and 70 ~ upon longer incubation periods, at 75 ~ crosslinking was not beneficial and better thermal stabilities were observed at 67.5 and 72.5 ~ Similar results were obtained for endoglucanase and cellobiohydrolase. Although DMS crosslinking seemed to change thermal inactivation, the changes were not very significant. REFERENCES

1. C.P. Kubicek, R. Messner, F. Gnaser, L.R. Mach, and M.E. Kubicek-Pranz, Enzyme Microb. Technol., 15 (1993) 90. 2. C.P. Kubicek, Adv. Biochem. Eng., 45 (1992) 1. 3. A. Singh and K. Hayashi, Advanc. Appl. Microb., 40 (1995) l. 4. O.C. Fagain andR. O'Kermedy, Biotech. Adv., 9 (1991) 351. 5. L.Gianfreda and M.R. Scarfi, Enzyme Stabilization:State of Art, Kluwer Academic Publishers, Netherlands, 1991. 6. J.H. Tae, Method. Enzymol., 91 (1983) 580. 7. G.E. Means and R.E. Feeney, Chemical Modification of Proteins, Holden Day Inc., California, 1971. 8. N. Nelson, J Biol. Chem., 153 (1944) 375. 9. Y.S. Rajput and M.N. Gupta, Enzyme Microb. Technol., 10 (1987) 143. 10. E.F. Hartree, Anal. Biochem., 48 (1972) 422.


157

Studies on the stability of acid phosphatase (A. niger) by crosslinking with glutaraldehyde and soil humates N. Ortega, L. Berzal, M.D. Busto and M. Perez-Mateos Department of Biotechnology and Food Science, University of Burgos, Plaza Misael Bafiuelos s/n, Burgos 09001, Spain

1. INTRODUCTION Enzymes in the aqueous phase of soil are generally short-lived because they are easily inactivated by adsorption, denaturation or degradation [1]. Some enzymes, however, can persist for a long time because they are protected by humic colloids [2]. For example, urease [3], oxidases [4], proteases [5] or hydrolases [5] have been extracted from different soils as stable enzyme-humic complexes. In fact, we can regard humic matter as a highly charged polyelectrolyte having a random configuration and which is dominated chemically by a highly recalcitrant phenolic 'backbone' [7]. On the other hand, chemical cross-linking of enzymes can be easily achieved with bifunctional reagents (e.g. glutaraldehyde, which contain two identical reactive groups reacting with nucleophilic side chains of amino acids [8]). In this sense, Busto et al proposed to use glutaraldehyde to introduce functional groups into soil humic polymers which could then react covalently with water-soluble enzymes [9]. One part of the reagent is used to introduce an appropriate functionality into the reactive functional group from the polymer's backbone, and the other part is used to bind the enzyme. In the present work, the stability towards thermal and proteolytic deactivation of acid phosphatase native and immobilised by cross-linking with glutaraldehyde and soil humates is compared.

2. METHODS 2.1. Soil

Samples of an Umbric Dystrochrept soil were collected from the surface (0-10 cm), airdried, sieved ( "'

80

(4)

N

with J(, M', M as unfolded, intermediate and structured monomer, N as native dimer and k1, k2 as I st- and 2nd-order rate constants. How the intermediates along the pathway can be monitored depends on the specific structure-function relationship for a given quaternary structure. In most cases, biological function requires the native state of association so that the final ratedetermining step can be monitored by measuring activity. Preceding steps may be accessible to spectral analysis, H-D exchange, cross-linking, HPLC and a wealth of other methods [18,19]. Figure 2 i l l u s t r a t e s the reactivation of cytosolic (s-MDH) and mitochondrial malate dehydrogenase (m-MDH), showing that closely related proteins may obey different folding-association mechanisms: in the case of s-MDH, dimerization is diffusion controlled, whereas m-MDH follows sequential uni-bimolecular kinetics [3]. 100 i'.-.,.b. A

n,-

0 0

.

-1 1~176B;.,-,oH

=f- - .._~f. - ' - ~ "--..

loo i -

. =

~

I < "':' i Oo "[/=I

I , , , 0 , I 9 ,-0.5 1 1.5 2-2/, 0 0.5 1 1.5 2 6 1 0 0--0.5 1 1.5 2 Time (h) Time (h) Time (h) Figure 2. Reactivation kinetics of porcine s-MDH and m-MDH. A. s-MDH: Denaturation in 6 M GdmCl renaturation in phosphate buffer pH 8 at D~ 3 40-300 nM shows no concentration dependence. B. m-MDH: Denaturation at . or in 5 M GdmCl: renaturation as in A: CMnU= 2 (I). 4 ( D ) , 10 (Z~), 35 (0), 88 ( = ) and 143 nM (A). Curves c a l c u l a ~ ] according to a uni-bimolecular mechanism. C. m-MDH. Reactivation at CMDH,= 65 nM after denaturation at pH 2 (o), in 6 M GdmCl (m). 8 M urea ( A ) . anG in the presence of 10 mM NAD+ (>95% saturation) ( 0 ) [1,3].

171

4. OFF PATHWAYREACTIONS

There are three stages where side reactions on the folding pathway may compete with proper folding and association the hydrophobic collapse, the merging of domains and the docking of subunits. Examples are- ( i ) transient interactions in the collapse of cytochrome c or B-lactoglobulin, ( i i ) domain mismatch of tryptophan synthase B2 and octopine dehydrogenase, and (i i i) inclusion body formation. At all three levels, 'recognition' is involved in the sense that specific substructures or surfaces have to be pre-formed in order to proceed on the folding path toward maximum packing density and minimum hydrophobic surface area. Collapse and domain merging involve intramolecular rearrangements. Due to the high local concentration of the reacting groups or surfaces within one and the same polypeptide chain, they are not significantly affected by neighboring molecules, i . e . , they obey I st - order kinetics with the slowest isomerization reaction determining the overall rate. In the case of domain proteins, the r e l a t i v e s t a b i l i t i e s of the domains and the contributions of the domain interactions to the overall s t a b i l i t y are crucial. The significance of the linker peptide connecting two well-defined domains has been studied by "grafting" experiments, e.g., by mutually exchanging the linker peptides of B- and v-crystal l in (Figure 1A) [8]. In both transplants, domain contacts dominate over subunit contacts. The recombinant separate domains do not interact with each other, stressing the local-concentration argument. Evidently, separate domains or other substructures may very well recognize each other and form complexes' RNaseS [20] and the folding and association of nicked lactate dehydrogenase [11] may serve as examples. The fact that in the l a t t e r case the yield does not exceed 15 % illustrates the importance of side reactions. In the case of the related monomeric octopine dehydrogenase, kinetic partitioning upon reactivation yields 70% native, and 30 % non-native protein [21] ~ U - ~- I N ~ IN

N

(70%)

(5)

(30%)

It is assumed that I N is "irreversibly denatured" as a consequence of wrong domain interactions. Repetitive denaturation/renaturation yields 70% N per cycle so that after a 2nd and 3rd cycle only ca. 50 and 35% of the original a c t i v i t y are recovered.

172

Obviously, i t is the kinetic partitioning between folding/association on one side, and misfolding/misassembly, on the other that determines the wellknown decrease in the yield of reactivation at high protein concentration. In going from single-chain domain proteins to protein assemblies, i t is wellestablished that this side reaction becomes the dominant effect in the overall mechanism of folding and association (Figure 3A) [22]. The underlying kinetic mechanism for an oligomer made up of n subunits is:

(m+n)~--~

kI k2 (m+n) M'- > n M ~ Mn,native (6)

v({ m

M'm

Here, ~6~, M', and M stand for the unfolded, collapsed and structured monomer, k1, k2 and k>2 for r a t e - l i m i t i n g I st-, 2nd and higher than 2nd-order rate constants, and ~m and M m for the products of aggregation as side reaction of folding/association. Evidently, the latter equation corresponds to Equ. (4) which describes the limiting case at high dilution at which premature c o l l i s i ons of misfolded molecules can be ignored, as shown in Figure 3B, i t is sufficient to quantitatively describe the observed inverse concentration dependence of reactivation and aggregation. Considering the underlying molecular mechanism, three questions need to be answered. ( i ) what is the committed step in aggregate formation, i . e . , at which stage along the sequential reaction are aggregates formed, ( i i ) when is the structured monomer committed to end up as the native protein and (i i i) what is known about the structure of aggregates and their constituent polypeptide chains. With respect to the f i r s t two problems, commitment to aggregation was shown to be a fast reaction, whereas the kinetics of the 'commitment to renaturation' followed precisely the slow kinetics of the overall reactivation. This means that there are fast precursor reactions on the folding path (collapsed states) all of which allow s t i l l aggregation, whereas. after a certain intermediate has been formed, slow shuffling leads 'one way' to the native state [23]. Regarding the third question, i . e . , the structure of aggregates, electron microscopy and circular dichroism indicate that wrong subunit interactions give rise to irregular networks with a broad distribution of highly structured particles at least 10 times the size of the native proteins. They resemble the native protein in its spectral properties, as far

173

as turbidity allows this conclusion (Figure 3A). For structural characteris tlcs of inclusion bodies and protein deposits, cf. [22,24]. A

B

.. I00~ ~ "~ 75

C

.

I

100 .,.,.

so

0

001

01

1.0

ClD H (pM)

8--_1

10

_--

01

5O

~

0001 001

~

0.1

O

10

~

DO lpM)

10

113(3

0

1.0

L- Arginine (M]

20

Figure 3. Kinetic partitioning between folding and aggregation. A. % reactivation (o), renaturation (Q) and light scattering (Z~) for porcine muscle LDH; denaturation at pH 2 renaturation at pH 7 (10~ [3,17] B. Reactivation calculated according to the competition of lSZ-order folding and diffusioncontrolled aggregation [16]. C. Effect of arginine on the reactivation and reoxidation of tissue plasminogen activator [25].

5. KINETICPARTITIONING. PRACTICALAND CELLULARASPECTS 5.1. Aggregation and inclusion body formation As has been mentioned, aggregation in vitro and inclusion body formation in the cell correspond to each other 9overexpression leads to high local concentrations of folding, yielding precipitates instead of native protein. One approach to cope with this problem is the use of weaker promotors, this way reducing the concentration. Since the aggregation reaction is of >2nd order, this may lower the local level of aggregation-competent folding intermediates below a c r i t i c a l concentration, thus favoring correct folding and association. However, there are two reasons why this approach is of no practical use- f i r s t , the overall yield of the recombinant protein per gram cell mass is drastically decreased, and second, its purification requires a f u l l scale separation of the guest molecule from the bulk of the host proteins, whereas inclusion bodies with their characteristic low heterogeneity allow highly simplified purification procedures for their fractionation. Thus, in many cases, down-stream processing starting from inclusion bodies has been the method of choice. In order to work at sufficiently low protein concentration during reconstitution, discontinuous 'pulse dilution' may be applied a certain amount of the protein is subjected to dilution and reactivated at the

174

sub-critical protein concentration: approaching the final value of reconstitution, the next portion of the concentrated solution of the denatured protein is added, and so on until the whole batch is transferred. Apart from keeping the concentration of the denatured protein low, this approach has the advantage that the increasing level of the renatured protein exerts a stabilizing effect on folding intermediates comparable to the one used routinely by adding, e.g.. serumalbumin. In this context, additives such as arginine may strongly increase the yield by shuffling aggregates back on the productive folding path (Figure 3C). L i t t l e is known about specific groups involved in the aggregation reaction. Early systematic experiments suggested hydrophobic interactions as well as covalent disulfide linkages to be of major importance. Recent in v i t r o and in vivo studies confirm this result [22,25,26]; test cases have been rhodanese and bovine growth hormone: for the f i r s t , detergents were shown to compete with aggregation, whereas BGH mutants with enhanced hydrophobic surface area showed enhanced aggregation [22,24-28]. The partitioning between folding and aggregation has been most intensively studied in the case of the tailspike protein (Tsp) from Salmonella bacteriophage P22 and the numerous mutants of this protein which either increase or suppress aggregation ( Figure 4) [30]. The wildtype trimer is highly stable and shows a close similarity in its in vivo and in v i t r ~ folding behavior. Upon release from the ribosome or upon dilution from denaturant solutions, A

B

C

~ams

Figure 4. Structure and folding mechanism of Tsp from bacteriophage P22. A. Polypeptide backbone of the truncated tailspike trimer, one subunit traced in bold. The molecule is folded into a right-handed parallel #-helix. B shows a section through the #-helix part, illustrating the side-by-side association of the subunits. C. Scheme for the in v i t r o and in vivo assembly of Tsp [22.29].

175

the polypeptides fold into an assembly-competent conformation bulk of the #sheet secondary structure and their aromatic amino-acid side chains are close to the native state, but even as 'protrimers' they are s t i l l highly unstable and prone to aggregation. The anomalous s t a b i l i t y is only acquired in a slow rearrangement reaction when the intertwined parallel #-helices merge to form the native trimer. Correspondingly. a significant part of the Tsp folding reaction occurs after subunit association. During maturation in vivo as well as refolding in v i t r o , the fraction of chains capable of maturing to native Tsp decreases with increasing temperature, the rest aggregates. Temperaturesensitive folding ( t s f ) mutations reduce the folding yield at elevated temperatures, whereas second-site suppressor mutations (su) improve folding under such conditions. Both types of mutations act by altering the s t a b i l i t y of tailspike folding intermediates [22,29].

5.2. Reconstitution in the presence of accessory proteins Inclusion bodies are the product of intracellular aggregation. They differ from aggregates formed in the test tube by their high packing density and large size which may sometimes span the entire diameter of the cell [23]. As a consequence, they can be easily harvested and washed by fractionated centrifugation, favoring their use in down-stream processing of recombinant proteins. The isolation of the desired protein follows exactly the same routine of in v i t r o denaturation-renaturation described before. For detailed guidelines, cf. [17,26]. Recently, the repertoire of methods has been extended by attempts to mimic in vivo conditions with respect to folding catalysts and chaperone proteins. Three examples may serve to illustrate the results. The f i r s t refers to the GroE system, the major molecular chaperone from E. coli and, at the same time, the role model of chaperone action. The molecule represents a K+- dependent weak ATPase, consisting of GroEL, a double-disc with 7-fold symmetry and 14 identical 60 kDa subunits, forming a hollow cylinder, which is closed by one or two heptameric GroES-lids serving as ATPase inhibitors. ATP. ADP and the substrate-protein are involved in drastic conformational changes that alter the interior of the cage from strongly hydrophobic to hydrophilic. As long as stretches of hydrophobic residues of the folding polypeptide chain are exposed, this shows high a f f i n i t y to the chaperone. During the off-phases, 'annealing' continues until the substrate has reached its native state. This does not bind to the chaperone. Thus. considering the kinetic partitioning between folding and aggregation, the chaperone does nothing but keeping aggre-

176

gation-competent folding intermediates below a c r i t i c a l concentration [30]. This simple explanation is confirmed by in v i t r o experiments using c i t r a t e synthase (CS) as substrate [31]. After unfolding and subsequent d i l u t i o n of the denaturant, only a small portion of the enzyme regains activity, while the major part aggregates. Renaturation in the presence of GroEL leads to the formation of a stable binary complex with non-native CS. thus i n h i b i t i n g reactivation completely. Only after GroES and ATP are added, the enzyme is released and reactivated. Stoichiometric amounts of GroEL are s u f f i c i e n t to block aggregation. The slow increase in light scattering after the addition of GroES and ATP shows that the released protein is s t i l l in its non-native state, prone to unspecific aggregation (Figure 5). The previous experiments clearly show that the GroE system does not rescue aggregated protein, nor is i t a folding catalyst accelerating the rate of folding. On the contrary, in the case of CS, reactivation is slowed down, probably due to the binding and release processes taking place during i t e r a t i v e annealing. For a detailed discussion of chaperones and their various modes of action, see [22,30,32.33]. The second example refers to the reactivation of a denatured and reduced immunotoxin (B3(Fo)-PE38KDEL) composed of the VH region of a carcinomaspecific antibody, connected by a f l e x i b l e linker to the corresponding VL chain, which is in turn fused to truncated Pseudomonas exotoxin. The chimeric protein contains three disulfide bonds, one in each antibody domain, and one in the toxin part. Upon renaturation, aggregation of non-native polypeptide A 9

c75

9- - -

.9

o

o:

B 75~

~

=

....

._c

._o

.,.

.--

75

~,

so

so_~ ~ so

25

25_

0

' ~' 0 100 200 300 400 Citrate s y n t h a s e ( n M )

"

~

C 60

._

; ~

o

~ ,

~ 20

~

i 0

50

~ 100 150 Time ( m l n )

"]

~

0

0

2 Time

i /, 6 (mJn)

Figure 5. Reconstitution of porcine citrate synthase after preceding denaturation in 6M GdmCl. A. Concentration dependence of reactivation (o) and aggregation (e). B. Reactivation in the presence of GroEL without GroES and ATP ( B ) , with GroES and ATP (e), and with the addition of GroES and ATP after 30 (~7) and 60 min ( w ) , respectively. C. Aggregation of the enzyme during reconstitution in the absence of GroE (o), in the presence of GroEL ( n ) , and in the presence of GroEL/GroES and ATP; addition at zero time of reconstitution ( w ) . after 30 sec (e), and after 4 min ( v ) [31].

177

chains and the formation of incorrect disulfide linkages lead to >90 % inacrive molecules. In order to improve the yield, E. Coli GroE and DnaK as well as protein disulfide isomerase (for SH/SS exchange) were applied. Both GroE and DnaK are found to influence the reaction: GroEL alone inhibits reactivation, whereas the complete GroEL/ES system significantly increases the yield of active protein. DnaK exhibits the same effect. PDI synergistically stimulates the reactivation, duplicating the yield compared to the non-enzymatic disulfide bond formation [34]. The third example deals with the renaturation, purification and down-stream processing of recombinant antibody fragments expressed in E.Coli. In this case, temperature, protein concentration, redox buffers, folding catalysts and destabilizing additives were varied in order to optimize the renaturation process [35,36]. It turned out that, in the case of the oxidized Fab fragment, peptidyl prolyl isomerases accelerate the refolding reaction and, at the same time, increase the yield of the native final product, thus proving that catalysis affects the kinetic partitioning by enhancing a rate-limiting step on the folding path. 5.3. Immobilization of proteins Folding/unfolding of immobilized proteins not only allow to avoid the off-pathway aggregation reaction but also opens the way to study the vectorial character of protein self-organization by comparing the results of folding starting from the N- or the C-terminal end of the polypeptide chain. Using ~glucosidase with polyionic N- or C-terminal t a i l s to reversibly immobilize the protein to an ion-exchange resin as a model, i t was shown that the enzyme (with an Arg 6 t a i l ) shows long-term s t a b i l i t y over weeks. Its reactivation yield, after preceding guanidine denaturation, is increased at least 5-fold, and the upper l i m i t of protein concentration without aggregation is shifted from-10 ~g/ml to -5 mg/ml [37]. The fact that both refolding from the C- and N-terminal end yield active enzyme is clear evidence that the vectorial character of protein biosynthesis, i . e . , the cotranslational folding from the Nto the C-terminal end of the polypeptide chain, does not determine the threedimensional structure of proteins. 5.4. Cel I ul ar aspects As taken from the classical examples of protein folding studies, three steps may be r a t e - l i m i t i n g in the self-organization of proteins, disulfide shuffling, proline cis-trans isomerization and association. On the other hand.

178

chains and the formation of incorrect disulfide linkages lead to >90 % inactive molecules. In order to improve the yield. E. Coli GroE and DnaK as well as protein disulfide isomerase (for SH/SS exchange) were applied. Both GrcE and DnaK are found to influence the reaction: GroEL alone inhibits reactivation, whereas the complete GroEL/ES system significantly increases the yield of active protein. DnaK exhibits the same effect. PDI synergistically stimulates the reactivation, duplicating the yield compared to the non-enzymatic disulfide bond formation [34]. The third example deals with the renaturation, purification and down-stream processing of recombinant antibody fragments expressed in E.Coli. In this case, temperature, protein concentration, redox buffers, folding catalysts and destabilizing additives were varied in order to optimize the renaturation process [35.36]. It turned out that, in the case of the oxidized Fab fragment. peptidyl prolyl isomerases accelerate the refolding reaction and, at the same time, increase the yield of the native final product, thus proving that catalysis affects the kinetic partitioning by enhancing a rate-limiting step on the folding path. 5.3. Immobilization of proteins Folding/unfolding of immobilized proteins not only allow to avoid the off-pathway aggregation reaction but also opens the way to study the vectorial character of protein self-organization by comparing the results of folding starting from the N- or the C-terminal end of the polypeptide chain. Using ~glucosidase with polyionic N- or C-terminal tails to reversibly immobilize the protein to an ion-exchange resin as a model, i t was shown that the enzyme (with an Arg 6 t a i l ) shows long-term s t a b i l i t y over weeks. Its reactivation yield, after preceding guanidine denaturation, is increased at least 5-fold, and the upper l i m i t of protein concentration without aggregation is shifted from-10 pg/ml to -5 mg/ml [37]. The fact that both refolding from the C- and N-terminal end yield active enzyme is clear evidence that the vectorial character of protein biosynthesis, i . e . . the cotranslational folding from the Nto the C-terminal end of the polypeptide chain, does not determine the threedimensional structure of proteins. 5.4. Cel I ul ar aspects As taken from the classical examples of protein folding studies, three steps may be rate-limiting in the self-organization of proteins, disulfide shuffling, proline cis-trans isomerization and association, On the other hand.

179

there are two enzymes, localized in the appropriate cellular compartments to catalyze the f i r s t two steps, and there are 'machineries' of chaperones and 'cohortes' of helper proteins to assist the third reaction [16.30.32,33]. The development in all three areas during the last few years has been breathtaking. In spite of that, both folding catalysis and chaperone action are far from being understood. Only a few aspects w i l l be outlined in the following paragraphs. For details reference is made to a number of recent monographs and reviews [5,26,38,39]. The biological significance of protein disulfide isomerase (PDI), peptidyl prolyl cis-trans isomerase (PPI) and chaperones is now well-established, and structural data for representatives of all three types of accessory proteins have been reported with s u f f i c i e n t l y high resolution so that mechanistic information is available. DsbA, the PDI-homolog in Eocoli, shows a close relationship to thioredoxin. Its second domain may be involved in the communication with a more complex machinery, including a membrane-spanning unit (DsbB) involved in the regeneration of DsbA. The anomalous redox and s t a b i l i t y properties of DsbA (with the reduced form of the enzyme showing higher s t a b i l i t y than the oxidized one) have been elegantly resolved by mutant studies [40,41]. Apart from its PDI activity, the enzyme acts as a subunit in a number of other enzyme complexes the common denominator seems to be that in al I cases recognition of nascent polypeptide chains is involved. PPIs seem to be ubiquitous enzymes catalyzing the rotation around the X-pro peptide bond and inhibiting signal transduction processes. The three-dimensional structures of two different representatives of the PPI family have been elucidated, supporting a "catalysis-by-distortion" mechanism. The substrate specificity is low. Obviously, in the cell, PPIs serve various functionstheir function as trigger factor and their location in the ER, as well as their effect on the formation of collagen in fibroblasts support the assumption that they are involved in in vivo folding. Considering the complex regulatory processes in higher cells, definitive proof, e.g., in connection with the maturation of collagen, s t i l l needs further investigation [42-44]. With respect to chaperones, even for the best-known systems, DnaK/DnaJ/GrpE and GroEL/GroES, significant information is s t i l l lacking. Therefore, generalizations have to be taken with care. This holds (i) for the stoichiometry and role of co-chaperones and the components in the cycle of ATP hydrolysis, (i i) for the apparent lack of substrate speci f i c i t y . (i i i ) for the conformational state of the bound polypeptide, (iv) for the interaction with other cellular

180

components toward 'chaperone machines', (v) for the question whether chaperones catalyze protein folding, acting as helper proteins and "unfoldases", etc. Obviously, "resurrection" of misfolded and misassembled proteins ( i f i t occurs) is the exception rather than the rule, for steric reasons, because of the limited space in the "cage", and for energetic reasons, because aggregates are commonly trapped in a deep energy well. Regarding the state of association of the released protein substrate, direct proof for the monomeric state, i .e. against oligomerization on the chaperone, came from reconstitution experiments with glutamin synthetase, where the formation of the native dodecamer has been shown to occur in solution, after release of the subunits from the binary GroEL-complex [45]. What drives the release, and how the polypeptide chain in the chaperone-substrate complex looks like, is s t i l l unresolved [30]. In summary, the picture that emerges from studies on pro- and eukaryotes clearly shows that off-pathway reactions in the cell are blocked by a whole arsenal of components from the moment of translation. The reason why, in spite of this protection, protein biosynthesis and subsequent structure formation commonly does not yield 100% is obvious: interactions involved in chaperone action must not be too strong, because otherwise the nascent polypeptide chain would be trapped. Thus, during the off-reaction, molecules may escape from the folding path, ending up in degradation. There are a number of aspects of in vivo folding that have not been considered: genome organization, nucleic acid secondary structure, codon usage, cotranslational modification, etc. The reason is that at present there is no clear evidence proving any of these to play a significant role in the folding/association and misfolding/ misassembly reactions of proteins in the cellular environment. ACKNOWLEDGMENTS Work performed in the author's laboratory has been generously supported by Grants of the Deutsche Forschungsgemeinschaft, the Fonds der Chemischen Industrie, the Alexander von Humboldt Stiftung, the Max Planck Gesellschaft and the European Community.

REFERENCES 1. 2. 3. 4.

R. Jaenicke (ed.), Protein Folding, North Holland, Amsterdam, 1980. C.M. Dobson and O.B. Ptitsyn (eds.). Curr. Opin. Struct. Biol. 7 (1997) I. R. Jaenicke, Progr. Biophys. Mol. Biol. 49 (1987) 117. R. Jaenicke, Biol. Chem. 379 (1998) 237.

181

5. R.B. Freedman, in Protein Folding (T.E. Creighton, ed.), W.H. Freedman, New York, 1992. 6. R. Rudolph, in Protein Engineering: Principles and Practice (J.L. Cleland and C.S. Craik. eds.), 283, Wiley-Liss Inc., New York, 1996. 7. R. Jaenicke, Eur. J. Biochem. 202 (1991) 715. 8. R. Jaenicke, Progr. Biophys. Mol. Biol. (1998) in press. 9. R. Jaenicke, FASEB J. I0 (1996) 84. I0. D.B. Wetlaufer, Proc. Natl. Acad. Sci. USA 70 (1973) 697. i i . D.V. Laurents and R.L. Baldwin, Biophys. J. (1998) in press. 12. R.Lo Baldwin, J. Biomol. NMR 5 (1995) 103. 13. M.P. Schlunegger, M.J. Bennett and D. Eisenberg, Adv. Prot. Chem. 50 (1997) 61. 14. R. Jaenicke, Biochemistry 30 (1991) 3147. 15. T. Kiefhaber, R. Rudolph, H.-H. Kohler and J. Buchner, Bio/Technology 9 (1991) 825. 16. R. Jaenicke, Curr. Topics Cell. Reg. 34 (1996) 209. 17. R. Rudolph, G. BOhm, H. Lilie-and R. Jaenicke, in Protein Structure: A Practical Approach, 2nd Ed. (T.E. Creighton, ed.), 57, IRL, Oxford, 1997. 18. W.A. Eaton, V. Munoz, PoA. Thompson, C.-K. Chan and J. Hofrichter, Curt. Opin. Struct. Biol. 7 (1997) I0. 19. F.M. Richards and P.J. Vithayathil, J. Biol. Chem. 234 (1959) 1459. 20. U. Opitz, R. Rudolph, R. Jaenicke, L. Ericsson and H. Neurath, Biochemistry 26 (1987) 1399. 21. W. Teschner, R. Rudolph and J.-R. Garel, Biochemistry 26 (1987) 2791. 22. R. Jaenicke and R. Seckler, Adv. Protein Chem. 50 (1997) I. 23. M.E. Goldberg, R. Rudolph and R. Jaenicke, Biochemistry 30 (1991) 2790. 24. G. Georgiou, ACS Symp. Ser. 516 (1993) 53. 25. R. Rudolph, in Modern Methods in Protein and Nucleic Acid Research (H. Tschesche, ed.) 149, de Gruyter, Berlin, New York, 1990. 26. A. Mitraki and J. King, Bio/Technology 7 (1989)690. 27. S. Tandon and P.M. Horowitz, j . Biol. Chem. 261 (1986) 15615. 28. D.N. Brems, S.M. Plaisted, H.A. Havel and C.-S.C. Tomich, Proc. Natl. Acad. Sci. USA 85 (1988) 3367. 29. A. Mitraki and J. King, FEBS Lett. 307 (1992) 20. 30. M. Beissinger and J. Buchner, Biol. Chem. 379 (1998) in press. 31. J. Buchner, M. Schmidt, M. Fuchs, R. Jaenicke, R. Rudolph, F.X. Schmid and T. Kiefhaber, Biochemistry 30 (1991) 1586.

182

32. A.L. Fink and Y. Goto (eds.) Molecular Chaperones in the Life Cycle of Proteins, M. Dekker, New York, 1998. 33. B. Bukau (ed.) Molecular Biology of Chaperones, Harwood, Amsterdam. 1998. 34. J. Buchner, U. Brinkmann and I. Pastan. Bio/Technology I0 (1992) 682. 35. H. Lille, K. Lang, R. Rudolph and J. Buchner, Protein Sci. 2 (1993) 631. 36. H. Lilie, S. McLaughlin, R.B. Freedman and J. Buchner. J. Biol. Chem,269 (1994) 14290. 37. G.Stempfer, B. HOll-Neugebauer, E. Kopetzki and R. Rudolph, Nature Biotechnol. 14 (1996) 329, 481. 38. C.B. Anfinsen, J.T. Edsall, F.M. Richards and D.S. Eisenberg (eds.), Accessory Folding Proteins, Adv. Protein Chem. 44, Academic Press, San Di ego, 1993. 39. S. Betts, C. Haase-Pettingell and J. King, Adv. Protein Chem. 50 (1997) 243. 40. J.C.A. Bardwell, Mol. Microbiol. 14 (1994) 199. 41. M. Huber-Wunderlich and R. Glockshuber, Folding and Design, (1998) in press. 42. T. Zarnt, T. Tradler, G. Stoller, C. Scholz, F.X. Schmid and G. Fischer, J. Mol. Biol. 271 (1997) 827. 43. F.X. Schmid, L. Mayr, M. MOcke and E.R. SchOnbrunner, Adv. Protein Chem. 44 (1993) 25. 44. G. Fischer, T. Tradler and T. Zarnt, FEBS Lett. (1998) in press. 45. M.T. Fisher, J. Biol. Chem. 268 (1993) 13777.

Stability and Stabilizationof Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

183

Activity o f m o n o c l o n a l antibodies in prevention o f in vitro aggregation o f their antigens Beka Solomon, Tamar Katzav-Gozanski, Rela Koppel and Eilat Hanan-Aharon Department of Molecular Microbiology & Biotechnology, Tel-Aviv University, Ramat Aviv 69978, Tel-Aviv, Israel

The availability of monoclonal antibodies (mAbs) which bind to a specific antigen at distinct and well-defined sites led to a better understanding of the effect of highly specific antigen-antibody interactions on antigen behavior. We found that appropriate mAbs interact with their antigens at sites where unfolding is started, leading to stabilization, refolding and suppression of further aggregation. These effects were found to be related to the localization of the antigen site of the antibody. In the following we study the interaction of carboxypeptidase A (CPA) and Alzheimer's 13-amyloid peptide (~AP) with some of their selected monoclonal antibodies. From a large panel of mAbs we selected two mAbs which exhibit significant chaperone-like activity in the refolding and prevention of CPA aggregation induced by heat and excess of Zn ions. CPA exposed to aggregation conditions maintains its solubility and catalytic properties in the presence of mAbs CP10 and CP 9. In vitro aggregation of ~AP (1-40) was inhibited by the presence of one of its mAbs, called AMY-33, but not by mAb 6F/3D, reinforcing the dependence of prevention of aggregation on the localization of aggregating epitopes. Identification of sequences or regions that may play a role in the protein folding pathway via appropriate mAbs may lead to understanding and prevention of protein aggregation.

1. INTRODUCTION

Protein aggregation is a major problem that extends into the mechanisms of human diseases and the fundamental aspects of protein folding, expression and function. Data from several laboratories(1-3) suggest that aggregation is non-specific in the sense that addition of other proteins can influencethe extent of aggregation of a certainprotein. The availabilityof monoclonal antibodies has facilitatedunderstanding how highly specific antigen-antibody interactions affect antigen stability. The complementary conformation between the interactingregions of the antibody and its antigen may confer high specificityand stabilityto the immunocomplex formed (4). Like the ubiquitous chaperones (5-7) mAbs raised against specific,native antigens may assistin antigen refolding (8-10) by recognizing incompletely folded epitopes and inducing their native conformation. By appropriate selection,mAbs have been found to bind to predefined locations on protein or peptide antigens without inhibitingtheirbiological activity(11). The identificationof such classes of sequences that participatein folding-unfolding and/or solubilization-aggregation

184 processes suggests the existence of effective solutions for the prevention of aggregation (1214). Interesting results were obtained from the study of the interaction of carboxypeptidase A (CPA) and/or Alzheimer's 13-amyloid peptide (13AP) with some of their selected monoelonal antibodies. CPA, a well-characterized zinc-exopeptidase that exhibits both peptidase and esterase activity, occupies a prominent position in the bibliography of metalloenzymes (15). A large number of mAbs have been prepared towards native enzyme and their properties have been widely investigated (11,16). Some of these antibodies bind to the enzyme with a relatively high binding constant at a location remote from the enzyme's active site and assist in refolding of heat-denatured enzymes (10). From a large panel of mAbs we selected two mAbs which which were found to exhibit significant chaperone-like activity in the prevention of CPA aggregation induced by heat and excess of Zn ions (17). In the other antibody-antigen complex we found that in vitro aggregation of 13AP (140) was inhibited by the presence of one of its mAbs, called AMY-33, but not by mAb 6F/3D, reinforcing the dependence of prevention of aggregation on the localization of aggregating epitopes. ELISA measurements and determination of residual enzymic activity as a probe of monitoring the native structure are used in our study to follow the inhibitory effect of different mAbs on prevention of aggregation of CPA and/or 13AP.

2. MATERIALS AND METHODS

2.1. Carboxypeptidase A aggregation CPA was obtained as an aqueous crystalline suspension (Sigma Chemical Co.). The crystals were washed with double-distilled water, centrifuged, and dissolved in 0.05 M Tris/HC1/O.5 M NaC1 buffer, pH 7.5. Protein concentration was determined by UV absorbance at 278 nm and by the Bradford method using BSA as a standard (18). The enzymic activities of CPA and its immunocomplexes were determined spectrophotometrically at 254 nm, using hippuryl-DL-13-phenyl-lactic acid as esterase substrate in 0.5 M PBS/0.5 M NaCl buffer, pH 7.5 (19). The preparation and characterization of the mAbs CPl0 and CP 9 chosen for the present study were described previously (16, 20). The optimal denaturation temperature of 50~ was chosen for CPA, because at this temperature the enzyme is easily denatured and the relevant mAbs have been shown to retain their binding affinity (21). The effect of immunocomplexation between CPA and its mAbs on enzyme aggregation was monitored by determination of the residual enzymic activity of CPA, as follows. CPA (0.1-1 ~tg) in PBS in the presence or absence of excess of ZnC12 was measured after exposure to aggregation conditions by incubation at 50~ with or without mAbs CPt0 and CP 9 (100 ~tl in PBS), at antibody/CPA molar ratios l:l. The enzymic activities of the immunocomplexes formed were measured as described above. Data are expressed as the percentage of activity reported to the enzymic activity of CPA before denaturation (being considered 100%).

185

2.2. 13-Amyloid Peptide aggregation Synthetic [3AP (1-40) was obtained from Sigma. For in vitro induced aggregation, the reaction mixture tubes containing 200 ~tl of an aqueous solution of [~AP (2.5 x 10-5 mM) were incubated for 3 h at 37~ Aggregated [3-amyloid samples were removed by centrifugation for 15 min at 15,000 x g. To determine the residual soluble ~AP, the supernatants were then incubated for another 60 min with an excess of mAb AMY-33 and/or 6F/3D to produce immunocomplexed [3AP. The two commercial mAbs used in this study, raised against peptides 8-17 and 1-28 respectively of [3AP, were anti-human [3-amyloid mAb 6F/3D (Accurate Chemicals, Westbury, NJ, USA) and mAb AMY-33 (Zymed, San Francisco, CA, USA). Rabbit polyclonal antibodies raised against synthetic ~AP-(1-40) were obtained from Boehringer Mannheim, Germany. In another set of experiments mAbs, at equimolar antibody/antigen concentrations, were added to the reaction mixtures before the first incubation period of 3h at 37~ The amount of 13AP left in solution under the various conditions was measured by ELISA as follows. Rabbit anti-[3-amyloid (1-40) antibody (100 ng/well) was attached to epoxy-coated Microtiter plates for 16 h at 4~ Atter the plates were washed with phosphate-buffered saline (PBS) containing 0.005% Tween 20, the residual epoxy groups were blocked by adding 1% low-fat milk. Before use, the plate was thrice washed with PBS/Tween 20 and then dried. The soluble immunocomplex of anti-13-amyloid/13AP obtained as described above was added to the plates for an additional 1 h at 37~ and bound mAb was measured by horseradish peroxidase-labeled goat-anti-mouse antibody. Degradation of the O-phenylenediamine substrate by HRP was monitored at A495 according to manufacturer's instructions by using an ELISA reader. The amount of mAb bound was assumed to be proportional to the amount of soluble amyloid peptide that remained in the reaction tube after incubation at the various aggregation conditions specified. The data represent the mean of three replicates. The standard deviations of the intra-assay and inter-assays were

DMSO

10

o 08 o ~ ,

>

concentration,

Co.so 1% (v/v) [] o v 30 L~ .to O 40 0 20

0.6 o 20

04

3O

0.240

00

0

9

I

15

~

I

,

I

30 45 Time, min

,

I

60

~

I

75

Figure 1" Deactivation of ot-chymotrypsin at 50 ~ and pH 6.0 When the temperature was lowered to 39 and 30 ~ just the opposite effect was observed: with increasing DMSO concentration o~-chymotrypsin was stabilized against thermal deactivation - as to be seen particularly well in Fig. 2. o~-Chymotrypsin solutions stored at temperatures of less than 44 ~ exhibited a remarkable increase of activity with time. In the absence of DMSO this activity increase during the initial deactivation phase was only observed for a maximum temperature of 39 ~ (data not shown). The degree of activation increased by lowering the temperature to 30 ~ By adding 10 % (v/v) DMSO this phenomenon was also observed at 44 ~ Higher DMSO concentrations caused the activity to stay constant at its maximum value for a longer time. However, the activity maximum flattened with increasing DMSO concentration. An increase of pH from 6.0 to 7.5 and 8.5 had a drastic effect on the stability of otchymotrypsin in the absence of DMSO - as shown in Fig. 3 for experiments at pH 7.5. Thus, the activity at pH 7.5 had dropped by 6 1 % after six days, whereas the activity at pH 6 had only dropped by 19 % after the same period of time.

226

1.4 3O

1.2 1.0 == 0.8 D

D

Parameter: couso !'1

(9

L.

0.4

0 V

m

IZ

O

1% (vlv) 0

Z~ 10

-~ 0.2 , O

0

concentration,

DMSO

0.6

I

,

20 30 40 I

0

.

I

1

9

2

I

9

3

I

9

4

l

,

I

5

9

I

6

.

7

8

Time, days Figure 2" Deactivation of o~-chymotrypsin at 30 ~ and pH 6.0 1.4 30 20

1.2

40

Z~

10

0.8 t~ :3

"o o,m

f/I CD L

"~

0.6

Parameter: DMSO

0.4 0.2 0.0

con centration,

coNso 1% (v/v) rl o ~7 30 Zl 10 O 40 O 20 ' 0

9

I

I

i

l

2

I

i

.

3

I

i

4

I

5

i

I

6

9

i

7

.

8

Time, days

Figure 3" Deactivation of o~-chymotrypsin at 30 ~ and pH 7.5 The series-type of model did explain at least partly the observations with respect to the stability of ~-chymotrypsin. During thermal treatment a stable intermediate enzyme species (El) was formed the specific activity of which was higher than that of the initial species (E). The rate of formation, and particularly the rate of disappearance of this intermediate species, were influenced by the presence of DMSO. The rate of disappearance decreased at temperatures lower than 44 ~ in the presence of DMSO. At higher temperatures DMSO accelerated the rate of deactivation. Apparently, DMSO was able to stabilize o~-chymotrypsin by stabilizing the intermediate

227 species against further attack. The formation of the intermediate species was accelerated by increasing the temperature and the pH. At 50 ~ the reaction was so fast that the intermediate species could no longer be observed - as shown in Fig. 1. It is remarkable that there was only a small influence of pH on the stability of ~chymotrypsin in solutions containing 20, 30, and 40 % (v/v) DMSO at 30 ~ - as shown in Fig. 2 and 3. Table 1 summarizes the half-life of the initial enzyme species. These have been calculated in the case of experiments at 50 ~ as

In(2)

tl/2= k---~

(4)

In(2) tl/2 _ /" ,kl:~

andas

Table 1" Half-life of o~-chymotrypsin T

pH

,

for those at 30- 44 ~

(tl/2 / h) in the presence of DMSO

DMSO-concentration

/ % (v/v)

-

0

10

20

30

40

6.0 7.5 8.5 6.0 6.0 6.0

38.1 6.37 6.60 4.58 2.68 1.13

91.1 28.1 26.3 13.1 4.62 1.12 ,

181 114 110 35.3 6.42 0.88

242 185 268 79.1 n.d. 0.72

342 544 318 n.d. n.d. 0.35

~

30 30 30 39 44 50

(5)

.

,

The influence of DMSO on the main rate constant of deactivation (k2) is summarized in Fig. 4 for temperatures at which DMSO showed a stabilizing action. The activity ratio (~) was found to be normally about 1.3. Thus, the intermediate enzyme species seemed to be 30 % more active than the initial one. 10

-3 -

Parameter: pHi "7 r ...,.

E

--r'l--

- and

te mperature

8.5;

--Z~--7.5;

10-4

/ ~

30

--0--

6.0;

30

--~7--6.0;

30 39

e-

C

o

1 0. 5

1 0 .6

J

0

i

I

10

i

I

20

,

,

I

30

,

I ,,,

40

DMSO concentration, % (v/v)

Figure 4: Influence of DMSO on the main rate constant of deactivation

228

Finally, the activity of ~-chymotrypsin has been determined in the presence of DMSO and three different substrates in order to test, if a selective inhibition of proteolysis may eventually stabilize the enzyme against autohydrolysis. The data are gathered in Fig. 5. 120

lOO =~

~

80

~

60

~

4o

, O~ 0

rl

Parameter: s ubstrate - - / k - - SUPHEPA ~ - - 0 - AZOCASEIN - - E l - - BTEE

20 0

0

10 20 30 DMSO concentration, % (v/v)

40

Figure 5: Influence of DMSO on different reactions of ~-chymotrypsin Obviously, there was an inhibitory effect of DMSO on all of the reactions tested at least at high DMSO content. However, the effect seemed to be less pronounced and less specific for proteolysis - as tested by the solubilization of azocasein - in order to be able to interpret the stabilization of ~-chymotrypsin by DMSO as owing to the inhibition of autodigestion. Thus, the results have shown that soluble ~-chymotrypsin may be stabilized by DMSO. This was true even for alkaline conditions, a domain of particular interest for synthetic reactions. Even a continuous use of ~-chymotrypsin in ultrafiltrationmembrane reactors (Fiaschei eta/. 1982) may be envisaged. In addition, it would be of quite some interest to know, if this stabilizing principle may be found for other enzymes as well. REFERENCES Flaschel E., Wandrey C., Kula M.-R. (1982), Adv. Biochem. Eng./Biotech. 26, 73-142 Kisee H., Fujimoto K., Noritomi H. (1988), J. Biotechnol. 8, 279-290 Moresoli C, Flaschel E., Renken A. (1992a), Biocatalysis 5,213-231 Moresoli C., Zaza P., Flaschel E., Renken A. (1992b), Biocatalysis 5, 203-211 Pliura D.H., Jones J.B. (1980), Can. J. Chem. 58, 2633-2640


229

T h e r m a l stabilization by its ligands of NADP+-isocitrate d e h y d r o g e n a s e f r o m the thermophilic c y a n o b a c t e r i u m Phormidium laminosum* Miguel A. Pardo, Maria J. Llama and Juan L. Serra§ Enzyme and Cell Technology Group, Department of Biochemistry and Molecular Biology, University of the Basque Country, Faculty of Sciences, P.O. Box 644, E-48080 Bilbao, Spain

The thermostability of the NADP§ isocitrate dehydrogenase [Isocitrate: NADP § oxidoreductase (decarboxylating), EC 1.1.1.42] purified to homogeneity from the thermophilic non-N2-f'lxing cyanobacterium Phormidium laminosum was studied. The catalytic activity of the enzyme increased up to 55~ and the calculated activation energy (Ea) for catalysis was 9.5 +__0.1 kcal mol ~. In the absence of ligands the enzyme was fully stable until 50~ but more than 75% of its activity was lost when it was incubated for 15 min at 60~ and it was completely inactive at 650C. The thermal inactivation obeyed a first-order reaction and the calculated Ea for enzyme denaturation was 91.5 _ 0.1 kcal mol1. The presence of either D,Lisocitrate or NADP § did not stabilize the enzyme, but the combined presence of D,L-isocitrate and Mn 2§ increased significantly the enzyme thermostability and, at least, 50% of the activity remained after 15 min at 73~ The calculated Eo for thermal inactivation of the enzyme-D,Lisocitrate-Mn 2§ ternary complex was 109 +_0.1 kcal mol-~. These results together with the fact that the Km values of the enzyme for D,L-isocitrate and NADP § depended on the nature of the divalent metal cation used as a cofactor suggest that the D,L-Isocitrate-metal 2+ complex is the true substrate of the enzyme.

1. INTRODUCTION In cyanobacteria the NADP§ isocitrate dehydmgenase (NADP-IDH) is the enzyme responsible to provide the cell with 2-oxoglutarate which is the carbon skeleton acceptor of the ammonium resulting from nitrate assimilation. Since the tricarboxylic acids cycle is incomplete in these photosynthetic prokaryotes (due to the lack of both 2-oxoglutarate dehydrogenase and succinyl-CoA synthetase activities [ 1]) IDH activity plays a key role in the linking of nitrogen and carbon metabolisms [2]. Nitrogen starvation caused in the thermophilic filamentous non-N2-flxing cyanobacterium P. laminosum important changes in morphology and biochemical composition of cells accompanied with a noticeable increase on the cellular

* Supported by grants (DGICYT PB95-0509 and DGES PB96-1471) from the Spanish Ministry of Education and Culture. M.A.P. is the recipient of a scholarship from the Basque Government. To whom correspondence should be addressed

230 level of several enzymes involved directly in nitrate assimilation as well as in NADP-IDH activity. Although the cyanobacterial NADP-IDH has been characterized in Anabaena sp. PCC 7119 [3] and Anabaena cylindrica [4] and purified to homogeneity from nitrogen-sufficient cells of Anacystis nidulans [5], Synechocystis sp. PCC 6803 [6] and Anabaena sp. PCC 7120 [7] no data on the effect of temperature on the stability of the enzyme have been reported so far. In this paper we report the thermostability of the enzyme from the thermophilic cyanobacterium P. laminosum in the presence and absence of its ligands. The enzyme was purified to homogeneity from nitrogen-starved cells which showed 8 to 10-times higher specific activity than nitrate-grown cells. A purification step consisting in the incubation for 15 min at 65~ of the crude extract supplemented with D,L-isocitrate and Mn ~* appeared as a key step in the purification strategy.

2. MATERIALS AND METHODS 2.1. Materials Mono Q HR 5/5 column and Phenyl Sepharose CL-4B were from Pharmacia Biotech (Uppsala, Sweden). DEAE-ceUulose (DE-52) was from Whatman (Maidstone, U.K.). Reactive Red-120 Agarose and D,L-isocitrate were from Sigma Chemical Co. (St. Louis, U.S.A.). NADP was from Boehfinger (Mannheim, Germany). Inorganic salts and other chemicals were reagents of analytical grade supplied by E. Merck (Darmstadt, Germany).

Organismand cell culture Phormidium laminosum (Agardh) Gom. strain OH-I-p.Cll was originally obtained from Prof. R. W. Castenholz (University of Oregon, Eugene, OR, U.S.A.). Cells were grown photoautotrophicaUy [8] and maintained routinely in axenic monoalgal cultures at 45~ in liquid medium D supplemented with 0.5 g 1t NaHCO3 [9] which contained 10 mM nitrate as the sole nitrogen source. Cultures were stirred by an air stream and illuminated by white fluorescent lamps with a photon flux density at the surface of the vessels of about 100 I.tmol rn2 st. To obtain nitrogen-starved cultures, cells were grown as indicated above in glass containers with 60 1 of medium D containing only 1 mM KNO3 as nitrogen source [10]. Cells grew until nitrate exhaustion and then the culture was maintained in the nitrogen-free medium for further 60-65 h under the same light and temperature conditions used for growth. 2.2.

2.3. Enzyme assay NADP-IDH activity was routinely measured at 45~ by following the reduction of NADP + at 340 nm in 50 mM Tris-HC1 buffer, pH 7.0, containing 1.7 mM D,L-isocitrate, 0.2 mM NADP § and 2 mM Mn2§ 4 mM Mg 2+) in a final volume of 1 ml. The reaction was started by the addition of enzyme. One unit (U) of activity catalysed the appearance of 1 lamol NADPH2 per min under standard assay conditions. The affinity (Kin) of the enzyme for its ligands was determined in both the Mn 2§ and Mg2§ reactions. Kinetic parameters were calculated fitting data by non-linear regression to the Michaelis-Menten equation. Protein was determined according to Peterson [11] using crystalline bovine serum albumin as the standard.

231

2.4. Thermostability studies Thermal stability was determined by assaying at 45~ the residual activity of aliquots of the pure enzyme incubated for 15 min at different temperatures in the presence or absence of ligands. The activation energy (Ea) for the decarboxylating activity and for thermal inactivation of enzyme were calculated from the slope of Arrhenius plots of logarithm of catalytic or inactivation first-order rate constants versus the reciprocal of the absolute temperature. The slope corresponded to -Ed2.303 R, where R is 1.987 cal K t mol t.

2.5. Enzyme purification The enzyme was purified from nitrogen-starved cells of P. laminosum to electrophoretical homogeneity by an unpublished procedure which includes heat treatment and several conventional low-pressure chromatographic steps performed at 4~ followed by a final ionexchange step using FPLC at room temperature. Cells were harvested by centrifugation, washed and resuspended in 20 mM Tris-HCl buffer (pH 7.0) containing 0.5 mM EDTA and 8 mM 2-mercaptoethanol. Cells were disrupted by sonication (Soniprep 150, MSE, Mabor Royal, U.K.) in an ice-bath and the preparation was centrifuged at 12.000 x g for 60 min at 4~ Solid D,L-isocitrate (to 1.7 raM) and MnSO4 (to 4 mM) were added to the cell-free extract which was then incubated at 65~ for 15 min. The preparation was cooled and centrifuged at 11.000 x g for 60 min at 4~ The supematant was decanted and chromatographied through DEAE-cellulose (DE-52), Reactive Red-120 Agarose and Phenyl Sepharose CL-4B. Finally, a homogeneous IDH preparation was obtained after ion-exchange chromatography in a Mono Q HR 5/5 column.

2.6. Polyacrylamide gel electrophoresis (PAGE) analysis The progress of enzyme purification, the determination of molecular mass of the native enzyme and the formation of high-size aggregates of enzyme during its heat-denaturation in the presence of ligands were analysed by nondenaturing PAGE. Electrophoresis was run in a PhastSystem equipment using 8-25% (w/v) acrylamide gradient PhastGels (Amersham Pharmacia Biotech, Uppsala, Sweden) or 10% (w/v) homogeneous acrylamide gels in a BioRad (Hercules, U.S.A.) equipment. Protein bands were stained with 0.2% (w/v) Coomassie brilliant blue R-350.

3. RESULTS AND DISCUSION

3.1. Purification of NADP-IDH The purification protocol used in this work allowed the purification of enzyme from 600 to 650-fold to a specific activity of 500 U/mg protein with a final activity yield of 25%. This specific activity value is much higher than those reported for the enzyme purified from nitrategrown cells of Synechocystis (15.7 U mgt protein) [6] and Anabaena (25.0 U mg ~ protein) [7], and probably it represents the highest specific activity for a NADP-IDH reported so far for the enzyme isolated from photosynthetic organisms. The initial heat treatment of the crude extract at 65~ in the presence of D,L-isocitrate and Mn 2§ was crucial since it removed most of the chlorophyllous pigments present in the crude extract and increased by near 3-fold the NADP-IDH specific activity.

232

3.2. Affinity of NADP-IDH for its ligands The enzyme required for its full activity the obligatory presence of either Mg 2+ and Mn 2* as an essential activator, although other divalent cations were also partially effective. The K= values for D,L-isocitrate and NADP + calculated in the Mn 2. and Mg2+-dependent reactions were significantly different (data not shown). Moreover, the calculated K= for Mg 2§ (159 +_0.5 I.tM) was near 7-fold higher than that for Mn 2§ (24 • 3.0 I.tM) despite the fact that Mg 2§ is considered to be the NADP-IDH physiological cofactor [6]. 3.3. Effect of temperature on the activity and stability of NADP-IDH The effect of temperature on NADP-IDH activity was examined by monitoring the appearance of NADPH2 in the complete reaction mixture incubated at temperatures ranging from 25 to 65~ The activity increased linearly with temperature up to 55*(3. The Eo for catalysis calculated from an Arrhenius plot was 9.5 _ 0.1 kcal mol ~. The thermostability of the enzyme was studied in detail in the absence or presence of its ligands, all of them supplied at the same concentrations than those used in the standard assay for activity (Fig. 1). The enzyme showed similar stability against denaturation when incubated alone or in the presence of either NADP or D,L-isocitrate. The presence of Mg ~+ or Mn 2+ alone increased the enzyme thermostability, but in a lower extent than when each one of these cations was added in combination with D,L-isocitrate.

80-

Ala exchange appears to be a strategy followed by nature in engineering protein stability, as well as rigidity. The same amino acid replacement was proposed and utilized by Matthews et al. (1987) for enhancing the stability of T4 lysozyme, considering that the Gly~Xaa exchange is expected to indirectly stabilize the folded protein by decreasing the chain entropy of the unfolded state. This strategy of protein stabilization has been successfully employed by Margarit et al. (1992) for enhancing the thermostability of the mesophilic neutral protease from B. subtilis. Single and double Gly~Ala mutants of this

287 protease, which is homologous to thermolysin, were more stable than the wild-type protein. Moreover, it was shown that mutational effects were cumulative (Wells, 1990). The most rigid amino acid residue in proteins clearly is Pro, as a result of the steric constrains of the pyrolidine ring of this residue. Accepting the view that there is a correlation between rigidity and stability, one would predict that thermophilic enzymes should possess enhanced content of Pro residues. Indeed, Suzuki et al. (1987) has proposed the "proline rule" for thermophilic enzymes, in that they contain an enhanced number of Pro residues in respect to their mesophilic counterparts. The introduction of Pro residues into a protein was therefore utilized as a strategy for stabilizing proteins. Mutants of oligo-l,6-glucosidase from B. c e r e u s with multiple Pro exchanges (up to nine residues) were thus produced and shown to be significantly and cumulatively stabilized against thermal denaturation (Watanabe et al., 1994). The same Xaa~Pro replacement for enhancing protein stability was also proposed by Matthews et al. (1987) on the basis of entropic considerations of protein stability (see above). Since the Pro residue is a helix-breaker (O'Neil & De Grado, 1990), the Xaa~Pro replacement should occur at the level of loops in the protein molecule. If the same exchange is placed in the middle of a helical segment, the protein becomes less thermostable (Nakamura et al., 1997). The concept of protein stabilization by rigidification has been recently applied by De Filippis et al. (1998) introducing by a semi-synthetic procedure cx-amino-isobutyric acid (Aib, or cx-methylalanine) at selected positions in a folded 62-residue fragment of thermolysin (C-terminal subdomain 255-316). The geminal methyl groups of Aib severely reduce the conformational space of this amino acid residue and thus restrict the peptide backbone torsion angles to those characteristic of the helical configuration. Consequently, the Xaa---~Aib exchange is expected to rigidify and stabilize a protein in analogy to the Gly---~Ala or Xaa---~Pro exchange (Matthews et al., 1987). Indeed, a double Ala~Aib replacement in thermolysin fragment 255-316 enhanced the melting temperature from 63.5 to 71.5~ (ATm= 8~ (De Filippis et al., 1998). 7. DISCUSSION The rigidity of the stable enzymes that can be isolated from thermophilic and extremophilic microorganisms is documented by a variety of experimental findings. The major effect of this protein rigidity is in reducing the catalytic efficiency of thermophilic enzymes in respect to that of their mesophilic counterparts and thus thermophilic enzymes are usually poor catalysts at room temperature. It is proposed that a rigid protein structure does not allow correct proximity and positioning of the key residues required for an efficient catalysis, which requires instead a somewhat flexible protein architecture to bind and properly accomodate the substrate at the active site, to perform the catalytic evem and to release the product of the enzymatic reaction (Alber et al., 1983; Fersht, 1985). Considering that extremophiles appeared earlier during biological evolution (Stetter et al., 1990; Stetter, 1993), it can be proposed that the more flexible (and allosteric) mesophilic enzymes derived from thermophilic ones and, since at

288 ambient temperature catalysis is reduced, evolution created more efficient biocatalysts by improving their plasticity (for an alternative view see Danson et al., 1996). Having proposed that thermophilic enzymes are more rigid molecules, one should expect that psychrophilic enzymes should be even more flexible than the mesophilic ones. Indeed, the thermolabile enzymes isolated from psychrophilic microorganisms have been shown to possess improved catalytic activity in respect to that of their mesophilic or thermophilic counterparts. The molecular basis for this effect has been analyzed utilizing the structural models of some psychrophilic enzymes produced by homology modelling (Feller et al., 1996; Narinx et al., 1997) or by direct determination of their X-ray structure (Aghajari et al., 1996; Feller et al., 1996). In general, it has been observed that psychrophilic enzymes possess enhanced overall flexibility in respect to their mesophilic or thermophilic enzymes, as judged from physicochemical data (Feller & Gerday, 1997). It has been proposed that the structural factors favouring the conformational flexibility of psychrophilic enzymes likely reside in a reduced number of salt bridges, shorter chain loops, less proline residues in loops, low hydrophobicity of protein core(s) and weaker binding of calcium ion(s) (Feller et al., 1996; Narinx et al., 1997). In general, it can be stated that the results of structural, functional and stability studies on psychrophilic enzymes parallel those conducted on thermophilic and extremophilic enzymes, since the results of both studies indicate the same factors and interactions as modulators of protein stability/rigidity/flexibility (see Feller & Gerday, 1997; Gerday et al., 1997, for a review). As an example, a higher or lower content of Pro residues in loops appears to correlate with enhanced or reduced thermostability in thermophilic and psychrophilic enzymes, respectively (Feller & Gerday, 1997). In conclusion protein flexibility appears to be a main reason for the enhanced activity of psychrophilic enzymes, while protein rigidity explains why thermophilic enzymes are poor catalysts at room temperature. Nevertheless, enzyme catalysis is a complex process involving a proper stereochemistry of the active site, specific binding of the substrate, participation of functional (acid or basic) groups characterized by specific pK values, a microenvironment at the active site of defmed dielectric constant, etc. Since proteins must move in order to function, an efficient active site requires some mobility, but it is clear that the molecular mechanism of enzyme catalysis is the result of a subtle balance between a variety of energetic, structural and dynamic effects and, consequently, perhaps it is not appropriate to overemphasise the role of protein mobility in catalysis. We are inclined to consider therefore protein rigidity of thermophilic enzymes an often made observation, but we are aware that exceptions can exist. For example, a laboratory evolution technique has been devised by using an engineered E. coli containing the LeuB gene coding for the isopropylmalate dehydrogenase from the extreme thermophile Thermus thermophilus (Topt 80~ Upon prolonged incubation at moderate/ambient temperature, E. coli mutants able to grow rapidly were obtained, producing isopropylmalate dehydrogenase mutants which were shown to retain the unusual thermostability of the wild-type enzyme from Th. thermophilus, while displaying high catalytic activity even at ambient temperature (Suzuki et al., 1997). Since it is possible

289 therefore to produce by evolution stable proteins still maintaining high catalytic activity at mesophilic temperatures, it is clear that the inverse correlation between thermostability and functional efficiency of a protein is a common finding and not a rule. It can be concluded therefore that there are several ways to develop a functional protein, as well as to improve protein stability. The major conclusion deriving from the vast m o u n t of expefimemal data accumulated so far from protein engineering studies (Matthews, 1987, 1993), as well as from studies on thermophilic enzymes (Jaenicke et al., 1996; Fomana, 1990, 1991), is that guidelines only and not generally applicable rules can be advanced for explaining at a molecular level and for modulating protein structure/stability/function, since protein mutations can be strongly context dependem (Russell & Taylor, 1995; Arnold, 1998). Acknowledgements. This study was conducted with financial support from the Italian National Research Council (CNR) and from the Ministery of University and Scientific Research (MURST). REFERENCES Adams, M.W.W. (1993) Enzymes and proteins from organisms that grow near and above 100~ Annu. Rev. Microbiol. 47, 627-658. Adams, M.W.W., Perler, F.B. & Kelly, R.M. (1995) Extremoenzymes: Expanding the limits of biocatalysis. Biotechnology 13, 662-668. Aghajafi, N., Feller, G., Gerday, C. & Haser, R. (1996) Crystallization and preliminary X-ray studies of (z-amylase from the antarctic psychrophile Alteromonas haloplanctis A23. Protein Sci. 5, 2128-2129. Alber, T. (1989) Mutational effects on protein stability. Annu. Rev. Biochem. 58, 765-798. Alber, T., Gilbert, W.A., Ringe, D. & Petsko, G.A. (1983) The role of mobility in the substrate binding and catalytic machinery of enzymes. In Mobility and function in proteins and nucleic acids, Ciba Symposium 93, Pitman, London, pp. 4-24. Argos, P., Rossmann, M., Grau, U., Zuber, H., Frank, G. & Tratschin, J. (1979) Thermal stability and protein structure. Biochemistry 25, 5698-5703. Arnold, F.H. (1998) Enzyme engineering reaches the boiling point. Proc. Natl. Acad. Sci. USA 95, 203 5-2036. Bae; K.H., Jang, J.S., Park, K.S., Lee, S.H. & Byun, S.M. (1995) Improvement of thermal stability of subtilisin J by changing primary autolysis site. Biochem. Biophys. Res. Commun. 207, 20-24. Banks, R.D., Blake, C.C.F., Evans, P.R., Haser, R., Rice, D.W., Merret, M. & Phillips, A.W. (1979) Sequence, structure and stability of yeast phosphoglycerate kinase. Nature 279, 773777. Barnes, L.D. & Stellwagen, E. (1973) Enolase from the thermophile Thermus X-1. Biochemistry 12, 1559-1565. Beaucamp, N., Hofman, A., Kellerer, B. & Jaenicke, R. (1997) Dissection of the gene of the bifunctional PGK-TIM fusion protein from the hyperthermophilic bacterium Thermotoga maritima: Designe and characterization of the separate triose phosphate isomerase. Protein Sci. 6, 2159-2165.

290 Braxton, S. & Wells, J.A. (1992) Incorporation of a stabilizing Ca2§ loop into subtilisin BPN. Biochemistry 31, 7796-7801. Chakrabartty, A., Schellman, J.A. & Baldwin, R.L. (1991) Large differences in the helix propensities of alanine and glycine. Nature 351, 586-588. Daniel, R.M., Cowan, D.A., Morgan, H.W. & Curran, P. (1982) A correlation between protein thermostability and resistance to proteolysis. Biochem. d. 29, 181-193. Danson, M.J., Hough, D.W., Russell, R.J.M., Taylor, G. & Pearl, L. (1996) Enzyme thermostability and thermoactivity. Protein Eng. 9, 629-630. De Filippis, V., De Antoni, F., Frigo, M., Polverino de Laureto, P. & Fontana, A. (1998) Enhanced protein thermostability by Ala--->Aib replacement. Biochemistry 37, 1686-1696. Delepierre, M., Dobson, C.M., Selvarajan, S., Wedin, R.E. & Poulsen, F.M. (1983) Correlation of hydrogen exchange behaviour and thermal stability of lysozyme. s MoL BioL 168, 687692. Dill, K.A. (1990) Dominant forces in protein folding. Biochemistry 29, 7134-7155. Efiink, M.R. & Ghiron, C.A. (1976) Exposure of tryptophanyl residues and tryptophan dynamics. Biochemistry 16, 5546-5551. Effink, M.R. & Ghiron, C.A. (1981) Fluorescence quenching studies with proteins. Anal. Biochem. 114, 199-227. F/lg/tin, C.0. (1995) Understanding and increasing protein stability. Biochim. Biophys. Acta 1252, 1-14. Feller, G. & Gerday, C. (1997) Psychrophilic enzymes: Molecular basis of cold adaptation. Cell. MoL Life Sci. 53, 830-841. Feller, G., Narinx, E., Arpigny, J.L., Aittaleb, M., Baise, E. & Gerday, C. (1996) Enzymes from psychrophilic organisms. FEMS MicrobioL Rev. 18, 189-202. Fersht, A. (1985) Enzyme structure and mechanism, 2nd. ed., Freeman, New York. Fersht, A.R. & Serrano, L. (1993) Principles of protein stability derived from protein engineering experiments. Curr. Opin. Struct. BioL 3, 75-83. Fontana, A. (1990) How nature engineers protein (thermo)stability. In Life in extreme conditions (di Prisco, G., ed.), Springer Verlag, Heidelberg, pp. 89-113. Fontana, A. (1991) Analysis and modulation of protein stability. Curr. Opinion Biotechnol. 2, 551-560. Fontana, A. (1997) Rigidity of thermophilic enzymes. Proc. 8th Eur. Congress on Biotechnology, Budapest, 17-21 August 1997, Commun. TU-138. Fontana, A., Fassina, G., Vita, C., Dalzoppo, D., Zamai, M. & Zambonin, M. (1986) Correlation between sites of limited proteolysis and segmental mobility in thermolysin. Biochemistry 25, 1847-1851. Fontana, A., Polverino de Laureto, P. & De Filippis, V. (1993) Molecular aspects of proteolysis. In Stability and stabilization of enzymes (van den Tweel, W., Harder, A. & Buitelaar, M., eds.), Elsevier Sci. Publ., Amsterdam, pp. 101-110. Fujita, S., Oshima, T. & Imahori, K. (1976) Purification and properties of D-glyceraldehyde-3phosphate dehydrogenase from the extreme thermophile Thermus thermophilus HBS. Eur. ,I. Biochem. 64, 57-68. Gerday, C., Aittaleb, M., Arpigny, J.L., Baise, E., Chessa, J.P., Garsoux, G., Petrescu, I. & Feller, G. (1997) Psychrophilic enzymes: A thermodynamic challenge. Biochim. Biophys. Acta 1342, 119-131.

291 Goldberg, A.L. & Dice, J.F. (1974) Intracellular protein degradation in mammalian and bacterial cells. Annu. Rev. Biochem. 43, 83 5-864. Griffiths, M.W. & Sundaram, T.K. (1973) Isocitrate lyase from a thermophilic Bacillus: Effect of salts on enzyme activity. J. BacterioL 116, 1160-1169. Guagliardi, A., Manco, G., Rossi, M. & Bartolucci, S. (1989) Stability and activity of a thermostable malic enzyme in denaturants and water-miscible organic solvents. Eur. J. Biochem. 183, 25-30. Gupta, M. (1993) Thermostability of enzymes, Springer Verlag, Berlin. Gurd, F.R.N. & Rothgeb, M. (1979) Motions in proteins, Adv. Protein Chem. 33, 73-165. Hochachka, P.W. & Somero, G.N. (1984) Biochemical adaptation, Princeton Univ. Press, Princeton, NJ. Holmquist, B. & Vallee, B.L. (1976) Esterase activity of zinc proteases. Biochemistry 15, 101107. Huber, R. (1979) Conformational flexibility and its functional significance in some proteins. Trends Biochem. Sci. 4, 271-274. Jaenicke, R. (1981) Enzymes under extremes of physical conditions. Annu. Rev. Biophys. Bioeng. 10, 1-67. Jaenicke, R. (1991) Protein stability and molecular adaptation to extreme conditions. Eur. J. Biochem. 202, 715-728. Jaenicke, R. (1996) Stability and folding of ultrastable proteins: Eye lens crystallins and enzymes from thermophiles. FASEB J. 10, 84-92. Jaenicke, R., Schurig, H., Beaucamp, N. & Ostendorp, R. (1996) Structure and stability of hyperstable proteins: Glycolytic enzymes from the hyperthermostable bacterium Thermotoga maritima. Adv. Protein Chem. 48, 181-269. Kidokoro, S., Miki, Y., Endo, K., Wada, A., Nagao, H., Miyake, T., Aoyama, A., Yoneya, T., Kai, K. & Ooe, S. (1995) Remarkable activity enhancement of thermolysin mutants. FEBS Lett. 367, 73-76. Kraut, J.(1988) How do enzymes work? Science 242, 533-540. Lazaridis, T., Lee, I. & Karplus, M. (1997) Dynamics and unfolding pathways of a hyperthermophilic and a mesophilic rubredoxin. Protein Sci. 6, 2589-2605. Margarit, I., Campagnoli, S., Frigerio, F., Grandi, G., De Filippis, V. & Fontana, A. (1992) Cumulative stabilizing effects of glycine to alanine substitutions in Bacillus subtilis neutral protease. Protein Eng. 5, 543-550. Matsumura, M., Signor, G. & Mat-thews, B.W. (1989) Substantial increase of protein stability by multiple disulfide bonds. Nature 142, 291-293. Matsumura, M., Yasumura, S. & Aiba, S. (1986) Cumulative effect of intragenic amino acid replacements on the thermostability of a protein. Nature 323, 356-358. Matthews, B.W. (1987) Genetic and structural analysis of the protein stability problem. Biochemistry 26, 6885-6888. Matthews, B.W. (1991) Mutational analysis of protein stability. Curr. Opin. Struct. Biol. 1, 1721. Matthews, B.W. (1993) Structural and genetic analysis of protein stability. Annu. Rev. Biochem. 62, 139-160. Matthews, B.W., Nicholson, H. & Becktel, W. (1987) Enhanced protein thermostability from site directed mutations that decrease the entropy of unfolding. Proc. Natl. Acad. Sci. USA 84, 6663-6667.

292 McLendon, C. & Radany, E. (1978) Is protein turnover thermodynamically controlled? J. Biol. Chem. 253, 6335-6337. Menrndez-Arias, L. & Argos, P. (1989) Engineering thermal stability: Sequence statistics point to residue substitutions in or-helices. J. Mol. Biol. 206, 397-406. Nakamura, S., Tanaka, T., Yada, R.Y. & Nakai, S. (1997) Improving the thermostability of Bacillus stearothermophilus neutral protease by introducing proline into the active site helix. Protein Eng. 10, 1263-1269. Narinx, E., Baise, E. & Gerday, C. (1997) Subtilisin from psychrophilic antarctic bacteria: Characterization and site-directed mutagenesis of residues possibly involved in the adaptation to cold. Protein Eng. 10, 1271-1279. Nayal, M. & Di Cera, E. (1994) Predicting calcium-binding sites in proteins. Proc. Natl. Acad. Sci. USA 91, 817-821. Ogasahara, K., Tsunasawa, S., Soda, Y., Yutani, K. & Sugino, Y. (1985) Effect of single amino acid substitutions on the protease susceptibility of tryptophan synthase alpha subunit. Eur. J. Biochem. 150, 17-21. O'Neil, K.T. & De Grado, W.F. (1990) A thermodynamic scale for the helix-forming tendencies of the commonly occurring amino acids. Science 250, 646-651. Perutz, L. & Raidt, H. (1975) Stereochemical basis of heat stability in bacterial ferredoxins and in haemoglobin A2. Nature 255, 256-259. Querol, E., Perez-Pons, J.A. & Mozo-Villarias, A. (1996) Analysis of protein conformational characteristics related to thermostability. Protein Eng. 9, 265-271. Rees, D.C. & Adams, M.W.W. (1995) Hyperthermophiles: Taking the heat and loving it. Structure 3, 251-254. Rehaber, V. & Jaenicke, R. (1992) Stability and reconstitution of D-glyceraldehyde-3-phosphate dehydrogenase from the hyperthermophilic bacterium Thermotoga maritima. J. Biol. Chem. 267, 10999-11006. Roche, R.S. & Voordouw, G. (1978) The structural and funcional roles of metal ions in thermolysin. CRC Crit. Rev. Biochem. 5, 1-23. Rosr, J.R., Salto, R. & Craik, C.S. (1993) Regulation of autoproteolysis of the HIV-1 and HIV2 proteases with engineered amino acid substitution. J. Biol. Chem. 268, 11939-11945. Russell, R.J.M. & Taylor, G.L. (1995) Engineering thermostability: Lessons from thermophilic proteins. Curr. Opin. Biotechnol. 6, 370-374. Schellman, C. (1986) Proteolysis as a probe for motility in mutant bacteriophage T4 lysozyme. Biophys. J. 49, 4932. Shoichet, B.K., Baase, W.A., Kuroki, R. & Matthews, B.W. (1995) A relationship between protein stability and protein function. Proc. Natl. Acad. Sci. USA 92, 452-456. Signor, G., Vita, C., Fontana, A., Frigerio, F., Bolognesi, M., Toma, S., Gianna, R., De Gregoriis, E. & Grandi, G. (1990) Structural features of neutral protease from Bacillus subtilis deduced from model-building and limited proteolysis experiments. Eur. J. Biochem. 189, 221-227. Somero, G.N. (1995)Proteins and temperature. Annu. Rev. Physiol. 57, 43-68. Sonnleitner, B. & Fiechter, A. (1983) Advantages of using thermophiles in biotechnological processes: Expectations and reality. Trends Biotechnol. 1, 74-80. Stellwagen, E. & Barnes, L.D. (1976) Structural and functional properties of thermophilic enolase. In Enzymes and proteins from thermophilic microorganisms (Zuber, H., ed.), Birk~tuser Verlag, Basel, Experientia Suppl. 26, 223-228.

293 Stellwagen, E., Cronlund, M.M. & Bames, L.D. (1973) A thermostable enolase from the extreme thermophile ]hermus aquaticus YT-1. Biochemistry 12, 1552-1559. Stetter, K.O. (1993) The lessons of archaebacteria. In ,Early life on hearth (Bengston, S., ed.), Nobel Symposium 84, New York, Columbia University Press, pp. 101-109. Stetter, K.O., Fiala, G., Huber, R. & Segerer, A. (1990) Hyperthermophilic microorganisms. FEMS MicrobioL Rev. 75, 117-124. Sundaram, T.K., Wright, I.P. & Wilkinson, A.E. (1980) Malate dehydrogenase from thermophilic and mesophilic bacteria: Molecular size, subunit structure, amino acid composition, immunochemical homology and catalytic activity. Biochemistry 19, 2017-2022. Suzuki, S. & Imai, T. (1982) Abnormally high tolerance against proteolysis of an exo-oligo-l,6glucosidase from thermophile Bacillus thermoglucosidicus RP 1006, compared with its mesophilic counterpart from Bacillus cereus ATCC 7064. Biochim. Biophys. Acta 705, 124126. Suzuki, S., Yamagishi, A. & Oshima, T. (1997) Adaptation of a thermophilic isopropylmalate dehydrogenase to mesophilic temperature. Proc. 8th Eur. Congress on Biotechnology, Budapest, 17-21 August 1997, Commun. TU- 137. Suzuki, Y., Oishi, K., Nakano, H. & Nagayama, T. (1987) A strong correlation between the increase in number of proline residues and the rise in thermostability of five Bacillus oligo1,6-glucosidases. AppL MicrobioL BiotechnoL 26, 546-551. Tainer, J.A., Roberts, V.A. & Getzoff, E.D. (1992) Protein metal-binding sites. Curr. Opinion BiotechnoL 3, 378-387. Toma, S., Campagnoli, S., Margarit, I., Gianna, R., Grandi, G., Bolognesi, M., De Filippis, V. & Fontana, A. (1991) Grafting of a calcium-binding loop of thermolysin to Bacillus subtilis neutral protease. Biochemistry 30, 97-106. Tsuboi, M., Ohta, S. & Nakanishi, M. (1978) Structural fluctuation of protein and thermophily. In Biochemistry of thermophily (Friedman, M., ed.), Academic Press, New York, pp. 251266. Van den Burg, B., Eijsink, V.G.H., Vriend, G., Veltman, O.R. & Venema, G. (1998a) Rendering one autolysis site in Bacillus subtilis neutral protease resistant to cleavage reveals a new fission. BiotechnoL AppL Biochem. 27, 125-132. Van den Burg, B., Vriend, G., Veltman, O.R., Venema, G. & Eijsink, V.G.H. (1998b) Engineering an enzyrne to resist boiling. Proc. Natl. Acad. Sci. USA 95, 2056-2060. Varley, P.G. & Pain, R.H. (1991) Relation between stability, dynamics and enzyme activity in 3phosphoglycerate kinases from yeast and Thermus thermophJlus. J. MoL BJoL 220, 531-538. Veronese, F.M., Boccfi, E., Schiavon, O., Grandi, C. & Fontana, A. (1984) General stability of thermophilic enzymes: Studies on 6-phosphogluconate dehydrogenase from B. stearothermophJlus and yeast, d. AppL Biochem. 6, 39-47. Vieille, C. & Zeikus, J.G. (1996) Thermoenzymes: Identifying molecular determinants of protein structural and functional stability. Trends BJotechnol. 14, 183-191. Vieille, C., Burdette, D.S. & Zeikus, J.G. (1996) Thennoenzymes. Biotechnol. Annu. Rev. 2, 183. Vihinen, M. (I987) Relationship of protein flexibility to thermostability. Protein Eng. 1, 477480. Vihinen, M., Torkkilla, E. & Riikonen, P. (1994) Accuracy of protein flexibility predictions. Proteins Struct. Funct. Gen. 19, 141-149.

294 Vogt, G. & Argos, P. (1997) Protein thermal stability: Hydrogen bonds or internal packing7 Folding & Design 2, 540-546. Vogt, G., Woell, S. & Argos, P. (1997) Protein thermal stability, hydrogen bonds and ion pairs. J. MoL BioL 269, 631-643. Watanabe, K., Masuda, T., Ohashi, H., Mihaya, H. & Suzuki, Y. (1994) Multiple proline substitutions cumulatively thermostabilize Bacillus cereus ATCC-7064 oligo-l,6glucosidase. Eur. s Biochem. 226, 277-283. Wells, J.A. (1990) Additivity of mutational effects in proteins. Biochemistry 29, 8509-8517. Wraba, A., Schweiger, A., Schultes, V., Jaenicke, R. & Z~tvodszky, P. (1990) Extremely thermostable D-glyceraldehyde-3-phosphate dehydrogenase from the eubacterium Thermotoga maritima. Biochemistry 29, 7584-7592. Wiathrich, K. & Wagner, G. (1979) Amide proton exchange and stability of proteins. J. MoL BioL 130, 31-3 7. W0thrich, K., Roder, H. & Wagner, G. (1980) Internal mobility and unfolding of globular proteins. In Protein folding (Jaenicke, R., ed.), Elsevier/North Holland Biomedical Press, Amsterdam-New York, pp. 549-564. Yan, B.X. & Sun, Y.Q. (1997) Glycine residues provide flexibility for enzyme active sites. J. BioL Chem. 2 72, 3190-3194. Zamost, B.L., Nielsen, H.K. & Starnes, R.L. (1991) Thermostable enzymes for industrial applications, d. lnd. MicrobioL 8, 71-82. Zhang, T., Baase, W.A., Shoichet, B.K., Wilson, K.P. & Matthews, B.W. (1995) Enhancement of protein stability by the combination of point mutations in T4 lysozyme is additive. Protein Eng. 8, 1017-1022. Zuber, H. (1981) Structure and function of thermophilic enzymes. In Structural and functional aspects of enzyme catalysis (Eggerer, H. & Huber, R., eds.), Springer Verlag, Heidelberg, pp. 114-127.


295

I m p r o v e m e n t o f thermal stability o f a diagnostic e n z y m e , Streptomyces cholesterol oxidase, b y r a n d o m and site-directed m u t a g e n e s e s and a structural interpretation Y. Murooka a, Y. Nishiya b, M. Toyamaa , M. Aoike a, M. Yamashita ~, and N. Hirayama c aDepartment of Biotechnology, Graduate School of Engineering, Osaka University, Yamadaoka, Suita, Osaka 565-0871, bTsuruga Institute ofBiotechnology, Toyobo Co. Ltd., Toyo-cho, Tsuruga, Fukui 914, r of Biotechnology, Faculty of Engineering, Tokai University, Nishino, Numazu, Shizuoka 410-03, Japan

1. INTRODUCTION Cholesterol oxidase (313-hydroxysteroid oxidase, EC 1.1.3.6) is a FAD-dependent bifunctional enzyme that catalyzes both the oxidation of cholesterol (5-cholesten-3[3-ol) to the intermediate 5-cholesten-3-one with the reduction of molecular oxygen to hydrogen peroxide, and the isomerization of steroid with a trans A:B ring junction to reduce 4cholesten-3-one. This enzyme is industrially important, and is commonly used for the enzymatic transformation of cholesterol. It is also useful for the clinical determination of serum cholesterol by coupling with a related enzyme for the assessment of arteriosclerosis and other lipid disorders and of the risk of thrombosis. Another use of the enzyme is in the microanalysis of steroids in food specimens. We previously cloned (1) and sequenced (2) the gene for cholesterol oxidase (choA) from Streptomyces sp. SA-COO. The secretory overproduction of Streptomyces cholesterol oxidase (ChoAs) in a Streptomyces host-vector system was demonstrated (3). Recently, the expression of the choA gene from Streptomyces in Escherichia coli by genetic modification was also reported (4). The crystal structure of cholesterol oxidase (ChoAB) from Brevibacterium sterolicum was determined and refined at 1.8A resolution (5). Li et al. reported the structure of a complex of cholesterol oxidase with the steroid substrate dehydroisoandrosterone, refined at 1.8A resolution (6). The steroid is buried within the protein in an internal cavity which, in the free enzyme crystal structure, is occupied by a lattice of water molecules. The hydroxyl group of the steroid substrate is hydrogen-bonded to both the flavin ring system of the FAD cofactor and a bound water molecule. The overall topology of this domain is very similar to that of other FAD-binding proteins. The amino acid sequence of ChoAs is homologous with that of ChoAB (59.2% identity). In particular, residues composing the steroid- and the FAD-binding sites are highly conserved between ChoAs and ChoAB. The three-dimensional structure of ChoAs modeled on the basis of its homology to ChoAB provides a reasonable point for analyzing the structure-activity relationships of thermostable mutants of ChoAs created by random mutagenesis. Here we report on successful enhancement of the thermal stability of ChoA by means of in vitro random mutagenesis and by site-directed mutagenesis techniques. The mutational effects are discussed using a structural model of ChoAs based on the tertiary structure of

296 ChoAs. We also describe other effects of the amino acid substitutions, which provide information for further improvements in the functionality of enzyme. 2. MATERIALS AND M E T H O D S The host strain used was E. coli JMI09. Plasmids pCOll7 coding for C h o A (4) and pBluescript-KS were used for additional plasmid construction. Plasmid pCOll0 (7), prepared for random mutagenesis of the choA gene, contains a unique SphI site on the choA gene. Recombinant strainswere grown in Terrificbroth or on L agar. The antibiotic used was ampicillin(50 ttg/ml). The transformation of E. coli, isolation of the plasmid, cleavage of the D N A with restriction enzymes, polymerization with Klenow fragment and ligation with T4 D N A ligase were carried out according to the common method. The nucleotides were sequenced on an A L F D N A sequencer. Random mutagenesis of the choA gene was carried out using P C R techniques. P C R was done under standard conditions with Tth D N A polymerase (Toyobo Co., Osaka, Japan), except that dTTP or dCTP was one fifth the volume of other deoxyribonucleotidc phosphates. The amplified fragments and pCOl10 were digested with SphI and ApaI, respectively. The insert fragments and vector were separated by agarosc gel electrophoresis, purified and ligated. These constructs were used to transform E. coli JM109 and transformants were plated onto indicatorplates. Cholesterol oxidase-producing colonies selected were grown on L agar at 30~ and heat treated by incubation at 50~ for 16 h. After the heat treatment, positive clones having enzyme activitywere selected by the filter-paper method (1) based on a colorimetric assay of cholesterol with cholesterol oxidase and peroxidase (8). These isolates were thermostablc enzyme-producing clones, because E. coli JM109 (pCOl l0) producing the wild-type C h o A showed no activity after the heat treatment. Site-directed mutagenesis was performed with a Transformer kit (Clontech Laboratories, Palo Alto, CA) as described in the manufacturer's instructions. Successful mutations were verified by DNA sequencing. The mutagenesis oligonucleotides used were synthesized with a Model 381A DNA synthesizer (Applied Biosystems, Foster City, CA). Each recombinant strain was grown to the stationary phase at 30~ in Terrific broth with agitation at 180 r.p.m. Cells were harvested by centrifugation and crude extract was prepared by vortex mixing the cells with glass beads. Ammonium sulfate was added to the cell-free extract and the precipitate collected by centrifugation, dissolved in 50 mM phosphate buffer (pH 7.0) and dialysed against the same buffer. The enzyme was purified from the dialysate by DEAE-Sepharose column chromatography followed by gel filtration to homogeneity. Cholesterol oxidase activity was measured by the method of Allain et al. (8). Enzyme solutions (0.5-1.0 U/ml) were prepared by dilution with 20 mM potassium phosphate (pH 7.0) containing 0.2% bovine serum albumin. The appearance of quinoneimine dye formed by coupling with 4-aminoantipyrine, phenol and peroxidase was measured at 500 nm by spectrophotometry. One unit of activity was defined as the formation of 1 M o f hydrogen peroxide (0.5 ttM of quinoneimine dye) per minute at 37~ and pH 7.0.

297 3. RESULTS 3.1. Construction of ChoA structural model Sequence alignment was carried out using the Needleman-Wunsch algorithm (9). The ChoAs sequence showed a 59.2% homology with ChoAB. A complex between ChoAB and dehydroisoandrosterone, an inhibitor of cholesterol oxidase, determined by X-ray crystallography (6), provided a basis for three-dimensional structure modeling of ChoA (Figure 1). The ChoA model was constructed using the QUANTA software package (QUANTA 4.0; Molecular Simulations, Burlington, MA). The ChoAB coordinates were obtained from the Brookhaven Protein Databank (10). The initial model was refined by energy minimization using the steepest descent method followed by the conjugate gradient method (11). The minimization calculations were conducted using the CHARMm module of QUANTA. All calculations were performed on an Indy workstation (Silicon Graphics, Palo Alto, CA). 3.2. Random mutagenesis to isolate thermostable mutant cholesterol oxidases The thermostable cholesterol oxidase was isolated from recombinant cells carrying a randomly mutated gene. The SphI-ApoI fragment of pCO110 encoding a part of ChoA (from position 97 to position 355 in the precursor sequence composed of 546 amino acid residues) was used as the target DNA. Since the structural information of ChoAB, indicated that this region does not contain the major parts of the FAD-binding domain and the active site cavity (5), the enzyme activity of ChoA would be conserved by random mutagenesis of this region. Approximately 90% of recombinant cells carrying a randomly mutated gene showed cholesterol oxidase activity. After incubation at 50~ for 16 h, the cholesterol oxidase activities of four clones among 5000 colonies were detected on the indicator plates, although no activity was detected in the wild-type-producing strain, E. coli JM109 (pCOll0) under the same condition. Recombinant plasmids were extracted from the four clones and the DNA sequences of the inserted fragments were determined. A single base change, resulting in the replacement of one amino acid residue, was found in each choA gene. All of the amino acid substitutions were located within a limited region (SerlO3 to Thr, Va1121 to Ala, Arg135 to His and Vail45 to Glu). The four ChoA mutants were designated S103T, V121A, R135I-I and V 145E, respectively. 3.3. Characterizations of wild-type and mutant cholesterol oxidases The four recombinant strains producing the mutant ChoA were cultured and these cholesterol oxidases were purified to homogeneity. All the purified enzymes and the wildtype enzyme migrated as a single protein band in SDS-PAGE analysis; the molecular mass of each enzyme was determined to be-55 kDa. The specific activities ofS103T, V121A, R135H and V145E were found to be 86.6, 72.0, 68.9 and 63.5 U/mg, respectively, while that of the wild type was 67.7 U/mg (with 1.0 mM cholesterol as the substrate, at 37~ and pH 7.0). The mutant enzymes thus had specific activities similar to or slightly higher than that of the wild-type enzyme. The thermal stabilities of these mutant enzymes were compared with that of the wild-type enzyme. All the mutant enzymes were more thermally stable than the wild type at 50, 55 and 60~ (Figure 2). The half-lives of the wild-type, S103T, V121A, R135H and V145E were calculated to be 7.8, 11.3, 12.2, l l.0 and 24.1 min, respectively. The increased thermal stability of V145E was significant.

298

100 V145E

A

v

>,10 >

V121A

..=,

IS103T R135H

ro~

c r

1

,Wild type

E

~D tr

"10 Figure 1. Representation of the overall folding of Stm/~omyces cholesterol oxidase that is constructed by hon~logy modeling. The FAD molecule (red balls) and de~droisoatxtm~ t m e (gray balls) are indicated. Typical mutation sites are also indicated.

"

I~O

40

60

Incubation time at 60"C (min) Figure 2. Effect of temperature on the stability of wild-type and mutant cholesterol oxidases.

The effects of pH on the stability and activity of the mutant and wild-type cholesterol oxidases were also compared. The stabilities of their pH were similar to that of the wildtype enzyme. However, the pH profile of VI45E differed from that of the wild-type and other mutant enzymes. The optimum pH of Vl45E was shifted to alkali and encompassed a broad range between acid and alkali. This significant alteration in pH in V145E may be caused by introduction of the negatively charged amino acid, glutamate. The Km values for cholesterol were calculated for the purified wild-type and mutant enzymes. The Km value for $103T, R135H and V145E, 13, 30 and 11 ~M, respectively, were similar to that for the wild-type enzyme, which was calculated to be 13 ~M. It was difficult to determine a K= value for V121A because the substrate-velocity curve was different from the Michaelis-Menten type. 3.4. Site-directed mutagenesis of V145 and other amino acid residues Since ChoAa was more thermostable than the ChoAs, we searched the differences of amino acid residues between ChoAs and ChoAa. We predicted some amino acid residues that might have effect on the stability of the enzyme and thus substituted V145, Q286, L119, P357 and $379 by other amino acid residues (Figure I). These mutant enzymes were also purified to homogeneity and characterized. The VI45D and VI45E mutants shiRed the optimum temperatures to 67~ but not other V145Q, Q286R, LII9A, LII9F, P357N and $379T mutant enzymes (Figure 3). The thermal stabilities of these mutant enzymes were compared with that of the wild-type enzyme (Figure 4). VI45E and VI45D were markedly thermostable. The V145 of ChDAs is corresponding to QI08 of ChDAB (Figure

299 120

~2o

100

100 A

,~v* 8 0 -

o~~

(D ._> 40

~(J6 0 \

Wild type

.i.

n- 2 0 ~ 0

>

\

~

.-> 60

---u-,~- -

V145Q Q286R ,

30

0)

V145D V145E

-" a- -

,

40

50

.

l

l

60

-> 9 40

"9- 0 - - -

~ 20-

--&-.--e--=- -

u

70

O

80

Wild type L119A L119F P357N S379T l

30

!

40

Temperature (*(3)

50

!

60

i

70

Temperature (~

Figure 3. Optimum temperatures of wild-type and mutant cholesterol oxidases.

1201

120

~oo.

--,..

l oo-:~~.~--,

80, ..~ =

"~

'~

Go.

40

\\

~

2 0 '

oJ. 30

Wild type

~ -- ~"" "

-- i-

V145D V145E V145Q

~

,~

40

50

__

60

-o "~

40

a~ rr

\ \ =,,,

~

80,

"~

\\

- Q286R

~

60

Temperature (*(3)

---On Wild type

----D--- L119A 2 0 ~ -- A--. Ll19F ----e--- P357N --L. S379T

o-t

70

30

40

50

60

70

Temperature (~

Figure 4. Thermal stabilities of wild-type and mutant cholesterol oxidases. The enzyme in 10raM ILP buffer (pH7.5) was incubated at various temperatures for 15 min.

!

80

300 5). The VI45Q mutation resulted in slightly increasing the optimum temperature and slightly stable than the wild-type. Q286R, L119F and $379T mutants were more thermally unstable than the wild-type. 3.5. Construction of multiple mutant enzymes A combination of mutations sometimes affecting protein stability shows additivity. Thus, we tried to construct multiple mutant enzymes by site-directed mutagenesis. The VI45E mutation was chosen as the base because of its higher level of thermal stability. As close residues easily influence one another, the S103T, V121A and R135H mutations were added in that order. Two-, three- and four-point mutant enzymes were designated M2 (S 103T and V145E), M3 (S103T, V121A and V145E) and M4 (SI03T, VI21A, RI35H and V145E), respectively. Each multiple mutant cholesterol oxidase was purified to homogeneity. These mutant enzymes were the same M.W. as that of the wild-type as determined by SDS-PAGE analysis. The specific activities of M2, M3 and M4 were found to be 39.9, 47.8 and 26.7 U/mg, respectively. The stabilities of M2 and M3 were elevated compared with that of V145E (Figure 6). The half-lives of M2 and M3 were 32.4 and 52.2 min, respectively, showing that the mutational effects of amino acid positions 103, 121 and 145 are additive. The stability of M4, however, decreased in comparison with those of V145E, M2 and M3.

100

S103T, V121A, V145 E (M3)

~.

'S103T, V145E (M2) VI45E (MI)

~> 10

:~ fo r U~ r

S103T,V121A, R135H,V145E (M4)

E ID al

-

9

20

9

40

-

9

60

Incubation time at C~'C (rain) Figtae 5. View of the residues arot~ V145 o f ~ p t o m ) r ~ C~A(A) arid

Figure 6. Effect of temperature on the stability of multiple mutant cholesterol oxidases.

O 108 ofBrev/bacterqwn C~A. Moreover, M4 exhibited maximum activity at 50~ whereas the optimum temperature of other enzymes was 60~ Thus, the R135H mutation has a negative influence on the other mutations. The optimum pH of the multiple mutant enzymes changed to encompass a broad range. The Km values of M3 and M4 for cholesterol could not be estimated because the substrate-velocity curves were similar to that of mutant VI21A. The Km value of M2 was calculated to be 21 M . The multiple mutant enzymes thus inherited the features of the V145E and/or V121A mutants.

301 4. DISCUSSION

We succeeded in creating thermostable cholesterol oxidases by random and site-directed mutageneses. The three-dimensional structure of ChoA modelled on the basis of its homology to ChoAB helps to analyze the structure-stability relationships of the mutants. The structural model also helps us to deepen our understanding of the reaction mechanism of cholesterol oxidase. Replacement of the V145 by E or D could introduce not only a hydrogen bond between E145 or D145 and D134, but also a salt bridge between E145 or D145 and R147 (Figure 5). These newly formed attractive interactions could contribute to the stabilization of the conformation and increase the thermal stabilities of V145E and V145D. The interactions of E145 with D134 and R147 would certainly influence the deprotonation and protonation of D134 and R147. The drastic change in the optimum pH ofV145E is considered to be due to these interactions. The two residues corresponding to positions 135 and 145 of ChoA lie on each of two antiparallel 13-strands and are located near each other. Replacement of these residues by H and E, respectively, in M4 will cause a very short non-bonded contact between an atom at the ~osition of H and C• of E. The interaction would destructively influence the molecular conformation there and reduce the thermal stability. M2, containing two point mutations, was finally selected as the commercial type of ChoA. Although three point mutations, M3 was the most thermostable mutant, the substrate-velocity curve was different from that of Michaelis-Menten type on account of the V121A mutation. Therefore, M3 was unsuitable for application of enzymatic assay. Arising from the effect of the V145E mutation, M2 exhibits a broad range of optimum pH from acid to alkali. This makes the variant suitable for clinical application, because various buffers and pHs are encountered depending on the assay reagent components. Development of M2 for commercial production is now in progress. Recently, we have succeeded in creating new enzymes that catalyze either the oxidation of cholesterol or the isomerization of 5-cholesten-3-one, as well as in enzyme improvement with respect to either or both reaction rates by the substitution of some amino acids. These findings will provide several ideas for the design of more powerful enzymes.

REFERENCES

1. 2. 3. 4. 5. 6. 7. 8.

Y. Murooka, T. Ishizaki, O. Nimi, and N. Maekawa, Appl. Environ. Microbiol., 52 (1986) 1382-1385. T. Ishizaki, N. Hirayama, H. Shinkawa, O. Nimi, and Y. Murooka, J. Bacteriol., 171 (1989) 596-601. I.'Molnar, K.-P. Choi, N. Hayashi, and Y. Murooka, J. Ferment. Bioeng., 72 (1991) 368-372. N. Nomura, K.-P. Choi, M. Yamashita, H. Yamamoto, and Y. Murooka, J. Ferment. Bioeng., 79 (1995) 410-416. A. Vrielink, L.F. Lloyd, and D.M. Blow, J. Mol. Biol., 219 (1991) 533-554. J. Li, A. Vrielink, P. Brick, and D.M. Blow, Biochemistry, 32 (1993) 11507-115151 Y. Nishiya, N. Harada, S.-I. Teshima, M. Yamashita, I. Fujii, N. Hirayama, and Y. Murooka, Protein Eng., 10 (1997) 231-235. C.C. Allain, L.S. Poon, C.G.S. Chan, W. Richmond, and P.C. Fu, Clin. Chem., 20 (1974) 470-475.

302 9. S.B. Needleman and C.D. Wunseh, J. Mol. Biol., 48 (1970) 443-453. 10. F.C. Bemstein, T.F. Koetzle, G.J.B. Williams, E.F. Meyer, M.D. Brice, J.R. Rodgers, O. Kennard, T. Shimanouchi, and Y. Tasumi, J. Mol. Biol., 112 (1977) 535-542. 11. I. Fujii, N. Hirayama, M. Yamashita, and Y. Murooka, Abstract Book of Annual Meeting of Nippon Seibutsu Kougaku (the Society for Fermentation and Bioengineering, Japan), (1994) 106.


303

Protein engineering for thermostabilisation of proteins: some theoretical rules and application to a [3-glucanase E. Querol a, J. Pons a, J. Cedano a, M. Vallmitjana a, F. Garcfaa, C. Bonet b, J. P6rez-Pons a A. Planas c & A. Mozo-Villarfas b alnstitut de Biologia Fonamental and Departament de Bioqufmica i Biologia Molecular. Universitat Autbnoma de Barcelona. Bellaterra. 08193 Barcelona. Spain. bDepartament de Ci~ncies M~diques B~siques, Facultat de Medicina, Universitat de Lleida. E25198 Lleida, Spain. CLaboratori de Bioqufmica. Institut Qufmic de Sarri~. Universitat Ramon Llull. 08017 Barcelona. Spain.

Protein thermostability has been investigated by two approaches. (A) Computational. To study the relationship between thermostability and conformational characteristics of proteins, 195 single amino acid residue replacements have been analysed for several protein conformational characteristics. From the analyses, some general rules arise which suggest where amino acid substitutions can be made to enhance protein thermostability. (B) Experimental. Glucohydrolases are biotechnologically important enzymes. We are analysing by site directed mutagenesis the structure/function relationship of two bacterial glucohydrolases, a 1,3-1,4-13-glucanase and a 13-glucosidase. We have determined the key residues for catalysis and substrate binding, and redesigned the stability and specificity of the glucanase. A glucanase thermorresistant mutant (N57A) has been obtained. 1. INTRODUCTION An important goal of protein engineering for biocatalysis is the enhancement of protein thermostability. The use of thermostable enzymes in industrial applications offers the benefits of increased rates of reaction, higher substrate solubility, decreased media viscosity, longer enzyme shelf lives than at normal storage temperatures and lowered risk of microbial contamination when reactions are carried out at high temperatures.

However, although a

number of successful examples of stabilisation of proteins exist, the mechanism of their thermostabilisation is far from being understood (see [ 1] for a review). Up to the present moment the successful examples of stabilisation of proteins have been achieved by methods based not far from intuition or trial and error. Two general approaches have been followed to

304 analyse the stability of proteins. First, the comparison between homologous proteins from thermophiles and mesophiles has provided some insight into the reasons why related proteins performing the same functions could have very different stability. Second, the analysis by protein engineering of the contribution of the different interactions that take place in a protein has resulted in some general rules about possible ways to increase the stability of a protein [2 11]. The first approach has the limitation that it is often difficult to identify the important determinants involved in a specific case since, in general their sequences have diverged significantly. It is important to examine proteins that are at the beginning of divergence, when their structures are still almost identical and the effect of single differences can be easily analysed. Following the second approach, protein engineering, several strategies have been proposed, for example: (a) reducing the difference in entropy between folded and unfolded proteins, which in practice means reducing the number of possible conformations in the unfolded state, (b) stabilising the dipoles of m-helices, (c) increasing the number of hydrophobic interactions and packing ratio in the interior core and, (d) reducing the area of water-accessible hydrophobic surface. Nevertheless, simple patterns that characterise stabilising interactions and quantitatively predict the effects of amino acid substitutions have eluded identification. An additional problem is the fact that in interpreting the stabilisation in terms of specified local interactions, one has to consider that the free energy of stabilisation of globular proteins in solution represents a marginal difference of large numbers. Thus, important contributions coming from small rearrangements of side chains or structural elements may escape detection. The difference in AAG between mesophilic and thermophilic enzymes is about 5-7 kcal.mo1-1, which can be accomplished by a few hydrogen bonds or by two salt bridges inside the protein globule.

This marginal energy for stabilisation of proteins is clearly a result of natural

selection. There appears to be a selective pressure for the marginal stability, probably to facilitate such processes as polypeptide folding and the flexibility required by the "native" conformation in the protein function. On the other hand, there is no selective pressure to make proteins stable at temperatures above those which they encounter in vivo. In order to gain insight into a variety of structural and physico-chemical characteristics leading to thermostabilisation several properties were analysed, suggesting some general rules for stabilization. In addition, they have been applied to the redesign of glucohydrolases. Glucohydrolases are biotechnologically important enzymes. We are analysing, by protein engineering, the structure/function relationship and determining the key residues for catalysis and substrate binding, and redesigning the stability and specificity of two enzymes, 1,3-1,4-13D-glucan 4-glucanohydrolase (EC 3.2.1.73) and a ~l-glucosidase (EC 3.2.1.21). These enzymes are utilized in the food industry, in brewing, feedstuffs, for the enzymatic synthesis of carbohydrates or glycoconjugates favoring their transglycosylating activity, etc. Among other

305 site-directed mutants with improved catalytic efficiency, etc., we have designed 13-glucanase thermorresistant mutants. We have previously determined the key catalytic residues by protein engineering and the 3D structure by crystallography [ 12 - 15]. 2. MATERIALS AND METHODS 2.1. Data from thermostable and mesostable isoproteins The wild type and thermostable mutant proteins used in this database are: o~-amylase; o~lantitrypsin; calbindin; carboxyl esterase; chloramphenicol acetyltransferase; cytochrome C; ferredoxin; [~-galactosidase; glucose dehydrogenase; glyceraldehyde phosphate dehydrogenase; 3-isopropylmalate dehydrogenase; kanamycin nucleotidyltransferase; lactate dehydrogenase; T4, hen egg-white, bird and human lysozymes; luciferase; malate dehydrogenase; neutral protease; oligo-l,6-glucosidase; protein HU; ~ cro and arc repressors; ribonucleases A, T1 and barnase; RNA polymerase subunit; subtilisins; superoxide dismutase; triosephosphate dehydrogenase; tryptophan synthase; tyrosinase; xylanase; xylose isomerase. The bibliography used to obtain this data is reported in a previous article [ 10]. Most of the thermostable forms have been obtained by site-directed mutagenesis, few of them emerge from the analysis of native isoproteins, in which the mesostable and thermostable forms differ only in one amino acid residue. From 195 single amino acid residue replacements -164 different positions on protein sequences- reported elsewhere [ 10] several protein conformational characteristics have been analysed. These are: type of residue replacement; conservative versus non conservative substitution; replacement being in a homologous stretch of amino-acid residues; change in hydrogen bonds, van der Waals and secondary structure propensities; solvent accessible versus inaccessible replacement; type of secondary structure involved in the substitution; the physico-chemical characteristic to which the thermostability enhancement can be attributable and, finally, the relationship of the replacement site to the folding intermediates of the protein, when known. Details of the computer analysis are described elsewhere [ 10]. 2.2. Experimental Bacterial strains, chemicals, PCR site directed mutagenesis, protein purification, enzyme kinetic assays, equilibrium urea denaturation, enzyme inactivation and thermotolerance (ts0) measurements, have been performed as previously reported [ 13, 15 - 17]. 3. RESULTS AND DISCUSSION. 3.1. Theoretical results. From the 195 substitutions (164 different positions on protein sequences), 121 represent a

306 different type of residue replacement. Since the whole space of directional changes has 380 substitutions (20 x 19), a quite large set of the potential substitution space has been explored by Nature or by protein engineers. As could be expected (see Table I), most replacements are conservative according to the Dayhoff matrix of replacements [ 18]. Table I. Some propensities related to thermostable replacements Secondary structure Potentiality for Conservative (Dayhoff) propensity hydrogen bonding Conserved Non conserved Maintainor Decrease Maintain or Decrease increase increase 146/195 49/195 162/195 33/195 121/195 74/195 It was also found that most replacements lead to maintain or increase the secondary structure propensity, analysed according to a classical scale of secondary structure propensities [19]. Most of the replacements lead to a residue with the same or enhanced potential for hydrogen bonding lying on its side chain, independently of their sequential and structural context. This is in agreement with a recently reported analysis [ 11]. The type of secondary structures involved in the replacements are shown in Table II. As can be seen, most of them lie in regular secondary structure positions either in a-helices or in [5-strand. Table II. Secondary structure involved Total a-helix 6 13-strand turn loop coil

3 2 2 2

On the N-cap 16

.....

On the C-cap 6

6 nd

3 nd

nd nd

nd nd

nd = No data available

If the above results are explainable in terms of conservation of protein structure and function, the results related to the rate of external versus internal replacements are more difficult to explain. Of 164 mutations, 121 were done in solvent accessible regions and 43 were on buried regions. This could be related to the fact that mutations within the core of a protein are likely to affect more than one energy term. Therefore evolution could choose those mutations being conformationally less expensive, which correspond to residues at the solvent exposed area. The putative amphipathicity of the regular secondary structures in which the mutants lie has also been checked. From the 109 point mutations, 23 lie on different amphipathic a-helix or ~-strand secondary structures (19 on helices and 4 on strand). However, this result may also

307 reflect the biased selection done by researchers. We have checked whether the internal replacements would primarily correspond to internal residues from amphipathic structures. We have found the following: 15 external, 16 internal amphipathic-helix residues, and 4 external, 2 internal amphipathic-strand residues, results that do not suggest any special tendency. Another interesting result comes from the homology/similarity analyses of the amino acid sequence stretches in which the mutated residue lies (we do not refer to the conservation of the specific replaced residue, which has been analysed above). This has been done by the Clustal algorithm [20] on those proteins having a number of sequences to allow this analysis (in fact it was the whole set of proteins used in this work except lysozymes and repressors); 113 regions in total as seen in Table III: Table III. Replacement on conserved stretches of mult!ali~ned homologous sequences On regular secondary structure On loops, turns or coils Highly conserved region 46 36 Medium conserved region 6 10 Non conserved 7 8 ,

,

,

Most of the replacements match conserved stretches of residues, specially those lying on a regular secondary structure. It is a well known fact that regular secondary structure stretches are highly conserved in protein sequences, however to our surprise most substitutions lying on loops, turns and coil regions, correspond to conserved regions too. Thermostability is certainly related to protein structure and conserved external regions are important for structure indeed. It is usually assumed that replacements leading to enhanced protein thermostability correspond to rigid regions. We have checked it for those cases in which these data (B-factors) are either reported by the authors or can be obtained from the Protein Data Bank. We found 37 replacements in flexible regions, 38 in rigid regions (113 were found in regions on unknown rigidity), suggesting that a simple increase of rigidity does not necessarily lead to thermostablity enhancement. Finally, a survey of the physicochemical basis of the thermostability enhancement, when indicated by the authors, show that they are diverse. This list is biased by the fact that sitedirected mutants are chosen by researchers to accomplish, as far as possible, some physicochemical expectations. Table IV summarises these results. In a subset of replacements leading either to a thermostability enhancement of 5~ or higher, or to a doubling of the thermoresistence, we have found an outstanding improvement of secondary structure propensity with respect to the whole set of 195 substitutions: Most of these substitutions lie on conserved and solvent-accessible regions.

308 Table IV. Physical and chemical contributions to AG to which thermostability enhancement be attributed "Better hydrogen bonding Better hydrophobic internal packing Enhanced secondary structure propensity Helix dipole stabilization Argo's replacements Removal of residues sensitive to oxidation or deamidation Burying hydrophobic accessible area Improved electrostatic interactions Strengthening intersubunit association Decrease chain strain Salt bridge optimisation Better Van der Waals Better affinity for calcium Improved AH upon substitution Unspecified improved AG upon substitution Unknown or not described

can 18 16 12 10 10 10 7 6 6 5 4 3 2 1

25 60

From the above results we suggest the following general rules to enhance protein thermostability: - Look for conserved stretches of residues upon multiple alignment. - Make substitutions conservative according to the Dayhoff matrix. Maintain or enhance the secondary structure propensity upon substitution.

-

- Replace preferentially solvent accessible residues. - Replace residues on N- and C-cap of m-helices (PDB or predicted) by introducing, if possible, a negative charge at N-cap or positive at C-cap of the helix. Replace residues on the N-cap and C-cap of l~-strands introducing, if possible, a negative -

charge at N-cap or positive at C-cap of the strand. Substitute with residues which improve or increase the number of potential hydrogen bonding

-

or van der Waals contacts. Whenever possible consider thermal damage on specific amino acid residues: deamidation of

-

Asn and Gin (the sequential motive Gly-Asn is a well known deamidation target [21]); oxidation of methionines; splicing the peptidic bond besides aspartic residues and substitutions of free thiol cysteins [21 - 25]. And also consider some suggestions elsewhere reported, as for example: - Argos' replacements (Lys ~ Arg, Asp ~ Glu, Gly ~ Ala) [2]. -

Substitute with prolines in loops [26].

- Introduce an additional disulfide bridge [27-29]. -

Introduce an additional metal binding site [30].

- Introduce new glycosylation sites [31].

309 Obviously, the best strategy for increasing the stability of a specific protein by sitedirected mutagenesis would require a thorough knowledge of its three-dimensional structure and of the molecular mechanisms responsible for its heat inactivation. There is still a limited number of proteins that fulfill these requirements. Since the above rules only predict wide regions of the protein as targets for site directed mutagenesis, in order to design individual replacements we have checked three different computational methods which define protein pseudopotentials. We find that each performs well only for one protein.

3.2. Site directed mutagenesis of a bacterial [3-glucanase The selected stretch of residues to be mutated constitutes a major loop covering the active site cleft. It is a solvent exposed region, highly conserved on most of the Bacillus glucanases (Figure 1). In addition to enhance thermostability, we were interested on redesigning and improving the catalytic and specificity properties of the enzyme (for more information see [ 15]). The general strategy was an Alanine scanning mutagenesis of the whole region followed by saturation mutagenesis of selected residues.

B. l i c h e n i f o r m i s B. m a c e r a n s B. p o l y m y x a B. subt ilis B. a m y l o l i q u e f a c i e n s

51 64 DGYSNGNMFNCTWR DGYSNGGVFNCTWR DGYSNGQMFNCTWR DGYSNGNMFNCTWR DGYSNGDMFNCTWR

Figure 1. Alignment of the loop amino acid sequences from different Bacillus glucanases. The designed single point mutations were: D51A, G52A, Y53A, $54A, N55A, G56A, N57A, M58A, F59A, N60A, T62A, W63A, R64A, M58D, M58S, M58V, M58W, M58G, M58F, M58N, M58P, M58L, M58I, M58Q, M58R, M58E, M58Y, M8T. A second strategy was the introduction of an additional disulfide bridge embracing the N- and C-terminal tails of the protein structure, by means of the double mutant $34C+K242C. Thermotolerance was determined at 65~ as the incubation time required to irreversibly inactivate the enzyme to 50% of its initial activity (ts0). One mutant, N57A showed higher thermotolerance (70%) than the wt enzyme. It is well known than Asn residues are often involved in protein deamidation. This effect is more pronounced when Asn is next to a Gly residue in a highly solvent exposed region [21 ], which is the case in this glucanase. Therefore, we suggest than the thermotolerance is due to removing the Asn residue. Finally, the strategy of introducing an additional disulfide bridge by means of the double mutant $34C+K242C failed. We got, at lower expression levels than for wt and other mutants, a protein with the same thermoresistence than the wild type. Finally, it deserves to be mentioned that, in addition to enhance thermostability, we were interested on

310 redesigning and improving the catalytic and specificity properties of the enzyme (for additional information see [15]). ACKNOWLEDGMENTS

This research was supported by grants BIO97-0511-CO2 and IN94-0347 from the CICYT (Ministerio de Educaci6n y Ciencia, Spain) and by the Centre de Refer~ncia de R+D de Biotecnologfa de la Generalitat de Catalunya. REFERENCES.

1. Gupta, M. N. (ed.) (1993) Thermostability of enzymes. Springer-Verlag, Berlin. 2. Argos, P., Rossmann, M.G., Grau, U. M., Zuber, H., Frank, G. and Tratschin, J. D. (1979) Biochemistry 18,5698. 3. Imanaka, T., Shibazaki, M. and Takagi, M. (1986) Nature, 324, 695. 4. Querol, E. and Parrilla, A. (1987) Enzyme Microb.Technol., 9, 238. 5. Menendez-Arias, L. and Argos, P. (1989) J. Mol. Biol., 206, 397. 6. Mrabet, N. T., Van der Broeck, A., Van den Brande, I., Stanssens, P., Laroche, Y., 20. Higgins, D.G. and Sharp, P. M. (1989) Comp. Appl. Bioscien., 5, 151. 7. Fersht, A. R. and Serrano, L. (1993) Curr. Opin. Struct. Biol., 3, 75. 8. Nosoh, Y. and Sekiguchi, T. (1993) In Gupta, M. N. (ed.) Thermostability of enzymes. Springer-Verlag, Berlin, p. 182. 9. Serrano, L., Day, A. G. and Fersht, A. R. (1993) J. Mol. Biol., 233,305. 10. Querol E., P6rez-Pons J.A., Mozo-Villarfas A. (1996) Prot. Engineer., 9,265. 11. Vogt G., Woell S., Argos P. (1997) J. Mol. Biol., 269, 631. 12. Planas A., Juncosa M., Lloberas J., Querol E., FEBS Letters (1992), 308, 141. 13. Juncosa M., Pons J., Dot T., Querol E, Planas A. (1994), J. Biol. Chem. 269, 14530. 14. Hahn M., Pons J., Planas A., Querol E., Heinemann U. (1995), FEBS Letters 374,221. 15. Pons J., Querol E., Planas A. (1997) J. Biol. Chem. 272,13006. 16. Juncosa M, Pons J., Planas A., Querol E. (1994), BioTechniques 16, 820. 17. Pons J., Querol E., Planas A. (1995), Prot. Engineer. 8, 939. 18. Dayhoff, M. O. (ed.), (1978) Atlas of Protein Sequence and Structure. Natl. B iomed. Res. Found., Washington DC. Vol 5. Suppl. 3. 19. Privilege, P., Fasman, G. (1989) In Fasman, G. (ed.) Prediction of protein structure and the principles of protein conformation. Plenum Press, New York, 391. 21. Volkin D.B., Mach H., Middaugh C.R. (1995) Meth. Mol. Biol., 40, 35. 22. Ahern, T. J. and Klibanov, A.M. (1985) Science, 228, 1280. 23. Zale, S. E. and Klibanov, A.M. (1986) Biochemistry, 25, 5432. 24. Volkin, D. B. and Klibanov, A. M. (1987) J. Biol. Chem., 262, 2945. 25. Tomazic, S. J. and Klibanov, A. M. (1988) J. Biol. Chem., 263, 3086. 26. Matthews, B. W., Nicholson, H. and Becktel, W. J. (1987) Proc. Natl. Acad. Sci. USA, 84, 6663. 27. Perry, I. J. and Wetzel, R. (1984) Science, 226, 555. 28. Pantoliano, M. W., Ladner, R. C.,Bryan, P. N., Rollence, M. L., Wood, J. and Poulos, T. L. (1987) Biochemistry, 26, 2077. 29. Matsumura, M., Signor, G. and Matthews, B. W. (1989) Nature, 342, 291. 30. Toma, S., Campagnoli, S., Magarit, I., Gianna, R., Grandi, G., Bolognesi, M., DeFilippis, V. and Fontana A. (1991) Biochemistry, 30, 97. 31. Olsen, O. and Thomsen, K. K. (1991) J. Gen. Microbiol., 137, 579.

Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 9 1998Elsevier Science B.V. All rights reserved.

Mechanisms of hyperthermophilic

stabilization of the ~-glycosidase a r c h a e o n S ulfolobus solfataricus

311

from

the

M. Moracci a, M. Ciaramella a, L.-H. Pearl b, B. Cobucci Ponzano a and M. Rossia, c aInstitute of Protein Biochemistry and Enzymology, CNR, Via Marconi 10, 80125, Naples, Italy" bStructural Biochemistry Section, Department of Biochemistry and Molecular Biology, University College London, Gower Street, London, WCIE 6BT, United Kingdom CDepartment of Organic Chemistry and Biochemistry, University of Naples "Federico II", Via Mezzocannone 16, 80100, Naples, Italy

1. I N T R O D U C T I O N The interest arose in these years for enzymes and proteins from hyperthermophilic microrganisms comes both from the biotechnological advantages offered by stabilised biocatalysts or because the molecular basis of protein stability is a central problem in protein biochemistry far from being completely solved. In this regard, enzymes that are active and stable at temperatures up to 100~ have been chosen as model system for studying the adaptation of molecules to high temperatures. Thermostability in proteins is the result of the balance of two opposite factors: flexibility and rigidity (Jaenicke et al. 1996). Even few weak interactions, as five hydrogen bonds or salt bridges, can contribute significantly to stability. The comparison of the amino acid sequences of thermophilic and mesophilic

This work was supported by CNR Target Project Biotechnology and Bioinstrumentation, by CNR Progetto Finalizzato Ingegneria Genetica, by EC project: "Biotechnology of extremophiles" contract n. BIO2-CT93-0274and by EC FAIR project: "Enzymatic Lactose Valorization" contract n. 1048/96.

312 molecules provided indications on some preferences in terms of the amino acid composition of proteins from hyperthermophiles. However, these results were descriptive and could not explain in detail the molecular mechanisms of stabilization. More recently, the increasing number of 3D-structures of enzymes and proteins from hyperthermophiles (Day et al. 1992; Russell et al. 1994; Hennig et al. 1995; Kornd6fer et al. 1995; Yip et al 1995) allowed to shed some light on the determinants of structural stability. One of the thermophilic enzymes more extensively studied is the ~-glycosidase (EC 3.2.1.x) from the extreme thermoacidophilic archaeon Sulfolobus solfataricus (Ssl3-gly). This enzyme was purified to homogeneity, its gene was cloned and expressed in yeast, and the enzyme was fully characterised. It is a retaining glycosyl hydrolase with wide substrate specificity and remarkable exo-glucosidase activity against oligosaccharides, which are hydrolysed from the non-reducing end (for a review on this enzyme see Moracci et al. 1994). The amino acid sequence places the enzyme in glycosyl hydrolase family-1 along with archaeal, bacterial and eukaryal enzymes. More recently, the residues involved in catalysis were identified and the 3D-structure resolved at 2.6 ~ (Moracci et al. 1996; Aguilar et al. 1997). Like other enzymes from hyperthermophilic Archaea, Ss~-gly is extremely stable to heat, with an half-life of 48 hours at 85~ and displays optimal activity at temperatures higher than 85~ These features, together with the availability of the 3D-structure, make of SsI3-gly an attractive model system for the study of the structure-function relationship in proteins at high temperatures.

2. THE 3D-STRUCTURE OF SS~-GLY Ss[~-Gly has been crystallized in its native, tetrameric form using multiple isomorphous replacement, and refined at 2.6 ~ resolution. The enzyme shows the classic ([~o08 fold that has also been observed in the two mesophilic members of the glycosyl hydrolase family-1 crystallized so far: the cyanogenic [3-glucosidase from Trifolium repens (Barret et al. 1995)and the phospho-[3-glucosidase from

Lactococcus lactis (Weismann et al. 1995). The Ss[3-gly enzyme crystallizes as a tetramer with a dimer in the asymmetric unit; each monomer contacts two other monomers in the tetramer through extensive interactions involving the last 25 residues at the C-terminal. The intermolecular contacts involving the C-terminal strand appear to contribute significantly to the overall stability of the enzyme.

313

2.1 Factors producing thermostability Several theories, mainly based on the comparison of the amino acid sequences of thermophilic and mesophilic molecules, have been proposed in order to explain the molecular origins of the enhanced stability of enzymes from hyperthermophiles. When this analysis is applied to Ss[~-gly it is clear that most of these rules do not hold, since the enzyme shows a lower content of alanine and isoleucine and an higher content of tryptophan and asparagine than the mean of the mesophiles (Aguilar et al. 1997). In contrast, two structural features might contribute to thermal stabilization in Ss~-gly. This enzyme maintains a surprising high number of buried hydrophilic cavities than is generally observed, with one buried solvent molecule on 11.4 amino acid residues versus a 1:27 ratio in proteins in general. This feature has only been observed so far in Ss~-gly. The second one is the high number of ionic groups involved in ion-pairs. The Ss~-gly tetramer contains 524 charged groups including the (z-amino and the carboxyl groups; 58% of these are involved in ion-pair interactions and around 60% of them occur as part of multiple ion-pair networks with at least three charge centres. The networks tend to occur in non contigous positions covering the surface and spanning different domains and subunits. In this way, the networks could act as an electrostatic cross-link between folded structural elements. The abundance of ion-pairs arranged in large networks has been found in other enzymes from hyperthermophiles (Hennig et al. 1995; Yip et al 1995). Interestingly, arginines occur frequently in ion-pair networks for their ability to form multidentate interactions: this might explain the higher number of arginines found in enzymes from hyperthermophiles. These findings suggest that large networks of ion-pairs are important for hyperthermostability; however, although the stabilizing effect of salt bridges in proteins was proposed more that 20 years ago (Perutz and Raidt, 1975), how electrostatic interactions contribute to protein stability is still object of debate (Goldman 1995). Recent papers showed that ion-pairs do not affect the thermostability of several hyperthermophilic enzymes (Russell et al. 1994). However, Jaenicke's group has recently demonstrated the role played by a ionic network in the stabilization of the Dglyceraldehyde-3-phosphate dehydrogenase from Thermotoga maritima (Pappenberger et al. 1997). Hence, the importance of ionic networks in each hyperthermophilic enzymes have to be checked experimentally. In Ss[~-gly, site-directed mutagenesis was applied in order to test the stabilizing effect of the largest network which occurs at the tetrameric interface between the carboxy termini. This ionic network involves 16 ion-pairs interactions in which

314 the penultimate Arg488 of each monomer is hydrogen bonded to the side chain of Glu345 in the same monomer and salt bridged to the (x-carboxyl group of the Cterminal His489 of a non-crystallographic symmetry related monomer (Figure 1).

Figure 1. Cartoon of the Ss~gly tetrameric

interface.

For

clarity, only two, out of four, Arg488-His489

interactions,

are highlighted.

3. COMPARISON OF WILD-TYPE AND MUTANT SS[~-GLY The mutation Arg488Ala was introduced by site-directed mutagenesis. No significant differences in behavior during purification could be detected; the steady state kinetic constants and the profile of thermal activation of the m u t a n t were similar to those of the wild type, suggesting that the mutation caused only minor alterations of the native structure.

3.1 Inactivation kinetics

The kinetic of thermal inactivation of the wild type and Arg488Ala m u t a n t enzymes was followed by incubating the samples at 75~

and at 85~

increasing time spans, and measuring the residual activity at 65~ The wild type Ss[~-gly, incubated at 75~

for

(Figure 2A).

after an activation in the first 2.5 hrs,

maintained 80% of the initial activity after 5 h. The half-lives were 24 h and 2 hrs at 75~

and 85~

respectively. In contrast, the mutant Arg488Ala showed a

significant reduction of the thermal resistance with half-lives of 3.5 hrs and 10 min at 75~ and 85~

respectively. At 85~ no activation could be found for both

315 the enzymes. Interestingly, when both the wild type and the mutant enzymes were incubated in 8M urea at 20~ no inactivation was observed for more than 8 hrs (data not shown). The thermal inactivation at 75~ with sodium chloride 4M is shown in Figure 2B: the denaturation profile of the wilde type Ss~-gly is comparable with that

obtained at 85~ without salt, with an half-life of 2.5 hrs. In contrast, the half-life of the Arg488Ala mutant, at these conditions, was 50 min. A

B

120

120

~oOoo

100 i ~

~ I,,,4

100 c

0

80

OQO

El

,p,l

0

El

t O I

9--.

40

0

~" 2o

80

~ ..=

60

~

40

% 0000

o

O -

o

I

l l

50

I I m

100

i..

150

i .

200

J

250

.

n.

300

o

o

o o

0

350

o

O0

~ 2o

m

0

9~ u

0

El El

a

I

I

I

I

I

50

100

150

200

250

300

350

time (rnin)

time (rain)

Figure 2. Thermal resistance of wild type (circle) and mutant (square) Ss~-gly. (A) 75~ (open symbols) and 85~ (closed symbols). (B) 75~ and 4M NaC1. These results show the importance of the Arg488 residue in the thermal stabilization and suggest that the inactivation of the mutant is caused by perturbation of the ionic network at the tetrameric interface. In fact, sodium chloride resulted more effective than urea on Ss~-gly, indicating the prevalence of ionic interactions in the stabilization of this enzyme. Furthermore, the faster salt-inactivation observed for wild type Ss~-gly, if compared with the mutant, can be explained assuming that the wild type, with an intact ion-pair network, is more

sensitive

than the Arg488Ala mutant

to the

increased

counterion

concentration.

4. CONCLUSIONS

The strong denaturating effect of the ionic strength has already been shown on hyperthermophilic enzymes indicating the importance of ion pairs in thermal

316 stabilization (Tomschy et al. 1994). Here, heat and salt-induced inactivation experiments on the wild type and mutant enzyme, give support to the observation derived from the 3D-structure, that a large ion pair network at the tetrameric interface stabilizes Ss[3-gly. Studies are in progress to identify the contribution to stability of additional residues of the network, and to analyze the structural effect of such mutations by kinetic and equilibrium denaturation.

REFERENCES

Aguilar C., Sanderson I., Moracci M., Ciaramella M., Nucci R., Rossi M. and Pearl L.-H., J. Mol. Biol., 271 (1997) 789 Barret, T., Suaresh, C.G., ToUey, S.P., Dodson, E.J., and Hughes, M.A., Structure, 3 (1995) 951 Day M.W., Hsu B.T., Joshua-Tor L., Park J.B., Zhou Z.H., Adams M.W.W. and Rees D.C., Protein Sci., 1 (1992) 1494 Goldman, A., Structure, 3 (1995) 1277 Hennig, M., Dairmont, B, Sterner, R., Kirschner, K., and Jansonius, J.N., Structure, 3 (1995) 1295 Jaenicke R. Schuring H., Beaucamp N. and Ostendorp R. Adv. Protein Chem., 48, (1996) 181 Kornd6fer, I., Steipe, B., Huber, R., Tomschy, A., and Jaenicke R., J. Mol. Biol., 246 (1995) 511 Moracci, M., CiarameUa, M., Nucci, R., Pearl, L.H., Sanderson, I., Trincone, A., and Rossi, M., Biocatalysis, 11 (1994) 89 Moracci M., Capalbo L., Ciaramella M. and Rossi M. Protein Eng., 9 (1996) 1191 Pappenberger, G., Schurig, H., and Jaenicke R., J. Mol. Biol., 274 (1997) 676 Perutz, M.F., and Raidt, H., Nature, 255 (1975) 256 Russell, R.J.M., Hough, D.W., Danson, M.J., and Tailor, G.L., Structure, 2, (1994) 1157 Tomschy, A., B6hm, G., and Jaenicke R., Protein Eng., 7 (1994) 1471 Yip, K.S.P., Stillman, T.J., Britton, K.L., Artymiuk, P.J., Baker, P.J., Sedelnikova, S.E., Engel, P.C., Pasquo, A., Chiaraluce, R., Consalvi, V., Scandurra, R., and Rice, D.W., Structure, 3 (1995) 1147 Weismann, C., Beste, G., Hengstenberg, W., and Schultz, G.E., Structure, 3 (1995) 961


317

A strategy for engineering thermostability: the case of cyclodextrin glycosyltransferase Joost C.M. Uitdehaag and Bauke W. Dijkstra* BIOSON Research Institute and Laboratory of Biophysical Chemistry, Groningen Biomolecular Sciences and Biotechnology Institute (GBB), University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands. Nowadays, protein engineering strategies to make thermostable enzymes are considered to be dependent on the mechanism of thermo-inactivation. For the enzyme cyclodextrin glycosyltransferase (CGTase), this mechanism is believed to be irreversible unfolding. To reduce the enzyme's inactivation rate, a sensible engineering strategy would be to stabilize flexible loops near the active site. These loops can be detected using the atomic B factors in any of the five CGTase X-ray structures. A comparison of thermolabile and thermostable CGTases shows that Nature has indeed chosen a strategy of stabilizing flexible loops near the active site. On this basis, some site-directed mutagenesis experiments are suggested.

1. I N T R O D U C T I O N

Cyclodextrin glycosyltransferases (CGTases) are enzymes that have the unique ability to produce cyclodextrins (o~(1->4) linked circular oligoglucosides) from starch. CGTases are evolutionary related to o~-amylases, and both enzymes are commonly used during the industrial liquefaction of starch. This is done by jet-cooking at high temperatures to keep the starch soluble, which requires use of thermostable enzymes, like the B. licheniformis o~-amylase I or the Thermoanaerobacter CGTase 2. in contrast, thermolabile CGTases have been isolated with very interesting specificities, like the production of a single size of cyclodextrin 2. To take advantage of such a specificity in an industrial context, thermostability will have to be engineered into these thermolabile CGTases through site-directed mutagenesis. * corresponding author

318

This is a challenging task, which requires foremost detailed insight into the enzyme's individual structure and function 3. Fortunately, for CGTases, five different three-dimensional structures have been solved 48, with sequence homologies > 60%, of which two are thermostable CGTases 6,7. These structures were extensively compared by Knegtel et al. 7 in combination with biochemical data, which identified many sites contributing to thermostability. Below we will provide reasons for focusing on two sites in particular, and suggest mutations to enhance thermostability in CGTases. From this, the consequences for other (~-amylase like proteins will be discussed.

2. SITES OF THERMOLABILITY

The proper approach to stabilize an enzyme depends on whether the thermoinactivation process is reversible or irreversible 9. In the case of reversible inactivation, there is an equilibrium between functional (folded) and unfunctional (unfolded) enzyme. Consequently, engineering thermostability means shifting this equilibrium towards the functional state by either stabilizing the folded state or destabilizing the unfolded state. This has been the focus of numerous strategies in the past 3'9'1~ However, in the case of (x-amylases, and thus probably also for the related CGTases, thermo-inactivation proceeds by irreversible unfolding 11, which means that once the enzyme has unfolded, it can never become functional again. Here thermostability is determined by the rate of unfolding, which depends on the free energy difference between the folded state and the transition state for unfolding 9. To decrease the unfolding rate this energy d i f f e r e n c e has to be increased. Therefore, a protein engineering approach should aim at making structural differences between folded and transition states as high in energy as possible. In contrast, it is ineffective to modify sites in the protein that have a similar structure in folded and transition state. The structures of folded and transition states are probably very alike. This is indicated by the fact that unfolding is often a highly cooperative process 3, meaning that its transition state comes early in the unfolding route. Therefore, structural differences between folded and transition state are likely to be confined to the sites where the unfolding begins. These are the loops that are structurally not very well stabilized (flexible). Moreover, unfolding of the loops closest to the active site would

319

have a direct effect on the catalytic process. Therefore, the flexible loops near the active site should be our engineering target in CGTase. Identifying flexible loops can be done using the atomic temperature factors (B factors) that are incorporated in any X-ray structure determination of sufficient high resolution. Such a B factor is directly related to the fluctuations around the average position of an atom 12, and correlates strongly with the calculated entropy of a loop13. Analyses of B factors, however, in studies of stability, are sparse14.

3. COMPARING

FLEXIBILITY IN CGTASES

To identify the flexible loops near the active site in CGTases, we first had to define the 'active site'. On the basis of a structure of a CGTase-maltononaose complex 15, we concluded that residues 70-270 and 310-350 (from a total of 686, using Bacillus circulans strain 251 CGTase numbering) were in or near the active site. Subsequently, for the various CGTases, equivalent C(~-positions were found from the pairwise superpositions of their X-ray structures. The B factors of those Ce~ atoms were taken to represent the local flexibility. If a particular amino acid position was deleted in a structure, it was assigned a B factor of 0.01 A 2, to account for deletion as a tactic to reduce flexibility. Subsequently, all B factors were divided by the average B factor of their structure, to enable comparisons. The CGTases were divided into two groups, thermostable (thermophile)" B. stearothermophilus 6 (Bstea) and Thermoanaerobacterium 7 (Tabium ) and thermolabile (mesophile): B. circulans strain 2515 (Bc251), B. circulans strain 84 (Bc8) and B. sp. 10118 (consisting of two independent molecules). Within each group, the normalized B factors were averaged. The thermolabile B factors were plotted versus the thermostable B factors in figure 1. For most residues, there is no consistent difference in B factor between thermostable and thermolabile CGTases: these dots lie close to the diagonal in figure 1. Some dots lie above the diagonal, indicating residues that are more flexible in thermostable CGTases, but these are all non-active site residues (white circles). In contrast, all positions near the active site that have consistently high B factors in thermolabile CGTases (grey circles on the right in figure 1) lie below the diagonal, and are thus stabilized in the thermostable CGTases. This shows that Nature has indeed targeted the flexible loops around the active site, which comprise residues 89-92 and 335-336. These residues were also mentioned by Knegtel et aL7 as part of the loops 88-95 and 335-339, which can now be singled

320

out as probable prime determinants of thermostability. Interestingly, another study of CGTases 16 located thermostability in the first and second N-terminal domains, where also the two loops 88-95 and 335-339 are located.

Figure 1. The average normalized B factors (see text) of the C(~ atoms in three thermolabile (x-axis) and two thermostable (y-axis) CGTase structures. Each circle represents one equivalent amino acid position in all five superimposed CGTase structures. A grey circle indicates a residue near the active site, a white circle a residue distant from the active site, residues mentioned in the text are numbered. A position with a high x-coordinate or a high y-coordinate signifies a consistent large flexibility in thermolabile or thermostable CGTases respectively. The diagonal is drawn for reference; residues deviating from the diagonal were targeted by evolution towards thermostability. The labelled residues were checked manually for involvement in crystal contacts in all CGTase structures. Omitting these crystal contacts from the averaging procedure and redrawing the figure still showed the special positions of the labelled residues.

321

4. MAKING

MUTATIONS

The loops 88-95 and 335-339 were discussed by Knegtei et aL 7 Loop 88-95 forms substrate binding site +315, and it was concluded that it was stabilized in Tabium CGTase compared to Bc251 CGTase by hydrophobic interactions attained through insertion of an extra residue at position 92. In addition an Asn 88 Pro substitution might reduce entropy of the unfolded loop in the transition state, in accordance with the 'proline rule '1~ In Bstea CGTase, the 88-95 loop is one residue shorter than in Bc251 CGTase, and residue Asp 89 augments the internal hydrogen bonding network, which is very loose in Bc251 CGTase. To stabilize the 88-95 loop by protein engineering, one can envisage to transfer the entire sequence from a thermostable to thermolabile CGTase, including surrounding residues that interact with it. Alternatively, point mutations could be rationally devised to stabilize the loop. For example in Bc251 CGTase, a mutation Y89D could improve internal hydrogen bonding, deletion of residues 89-91 is possible, or a S90P mutation could reduce flexibility.

Figure 2: Structure of the loop comprising residues 330-344 in diverse CGTases: Bc251 (white), Bstea (grey) and Tabium (black). Indicated are individual residues contributing to loop stability. The picture was made with MOLSCRIPT 17.

322

The other flexible loop near the active site, with residues 335-339, is close to the important catalytic residue Asp 328, and therefore its instability could directly interfere with catalysis. Knegtel et aL7 found that the loop 335-339 is one residue shorter in the thermophile Tabium CGTase and that it is more stabilized by hydrogen bonds and hydrophobic contacts, compared to the mesophile Bc251 CGTase (see figure 2). Interestingly, again a proline is present (Pro 341), which reduces the flexibility in the Tabium loop, whereas none of the thermolabile CGTases have Pro in the loop 335-339. In the thermophile Bstea CGTase, residues 335-339 are also stabilized by a proline, Pro 339, in addition to a Lys 376Asp 335 salt bridge (figure 2). Again, a mutational strategy to stabilize this loop can involve mimicking nature, by for instance introducing in Bc251 CGTase the R339P and K341P mutations, or by deleting residue Asn 335. However, due to sequence differences in the 335-339 loop, point mutations might have unexpected effects, and modelling and crystallographic techniques will have to be used to relate the biochemical behaviour to the structure. Fortunately, the speed of modern day protein crystallography is such, that it can easily be used as an analytical tool to study mutations.

5.

CONCLUSIONS

The wealth of structural information available about CGTases has enabled us to understand the structural basis of thermostability in this enzyme class 7. If thermoinactivation proceeds via irreversible unfolding, as is the case for CGTases, the flexible loops around the active site should be the first targets for a stability engineering strategy. This notion is supported by comparing B factors in five mesophile and thermophile CGTase structures, where Nature appears to have targeted loops 88-95 and 335-339. Stabilization of these loops has occurred through a promiscuous use of salt-bridges, proline substitutions, residue deletions and hydrophobic interactions, all factors which have been implicated in thermostability 9,1~ This shows that not the type of interaction is essential, but its strength and position, which depends on an enzyme's individual three-dimensional structure. Therefore, the results for CGTases cannot straightforwardly be applied to related enzymes with limited sequence homology, like the e~-amylases. However, if denaturation by irreversible unfolding is the general mechanism for all enzymes from the (z-amylase family 11'18, flexible loops near the active site might be a

323

common target to enhance thermostability. It should be noted that two known stabilizing mutations in B. licheniformis (~-amylase, H133Y and A209V 18 are not positioned in regions of high flexibility. The structure of these mutations is however not known, and their influence on flexibility is therefore uncertain. Currently we are working on CGTase mutants to test our theories, and rationalize the phenomenon of thermostability in this enzyme class. ACKNOWLEDGEMENTS

The authors thank Dr. Finn Drablos for providing a CGTase and e~-amylase sequence alignment. Conny Uitdehaag and Prof. Lubbert Dijkhuizen are thanked for their comments on the manuscript. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

Machius, M., Wiegand, G. & Huber, R. J. MoL BioL 246, 545-559 (1995). Pedersen, S., Dijkhuizen, L., Dijkstra, B.W., Jensen, B.F. & Jorgensen, S.T. Chemtech 19-25 (1995). Jaenicke, R. Eur. J. Biochem. :202, 715-728 (1991). Klein, C. & Schulz, G.E.J. MoL BioL 217, 737-750 (1991). Lawson, C.L., van Montfort, R., Strokopytov, B., Rozeboom, H.J., Kalk, K.H., de Vries, G., Penninga, D., Dijkhuizen, L. & Dijkstra, B.W.J. Mol. BioL 236, 590600 (1994). Kubota, M., Matsuura, Y., Sakai, S. & Katsube, Y. Denpun Kagaku 38, 141-146 (1991). Knegtel, R.M.A., Wind, R.D., Rozeboom, H.J., Kalk, K.H., Buitelaar, R.M., Dijkhuizen, L. & Dijkstra, B.W.J. MoL BioL :256, 611-622 (1996). Harata, K., Haga, K., Nakamura, A., Aoyagi, M. & Yamane, K. Acta Cryst. i:)52, 1136-1145 (1996). Shaw, A. & Bott, R. Curr. Opin. Struct. BioL 6, 546-550 (1996). Watanabe, K., Masuda, T., Ohashi, H., Mihara, H. & Suzuki, Y. Eur. J. Biochem. 226, 277-283 (1994). Tomazic, S.J. & Klibanov, A.M.J. BioL Chem. 263, 3086-3091 (1988). Drenth, J. Principles of protein X-ray crystallography (Springer-Verlag, New York, 1994). Meirovitch, H. & Hendrickson, T.F. Proteins Struct. Funct. Genet. 29, 127-140 (1997). Fontana, A. How nature engineers protein (thermo)stability. in Life Under Extreme Conditions (eds. di Prisco, G.) 89-113 (Springer-Verlag, Berlin, 1991). Strokopytov, B., Knegtel, R.M.A., Penninga, D., Rozeboom, H.J., Kalk, K.H., Dijkhuizen, L. & Dijkstra, B.W. Biochemistry 35, 4241-4249 (1996). Kaneko, T., Song, K.-B., Hamamoto, T., Kudo, T. & Horikoshi, K. J. Gen. Microbiol. 135, 3447-3457 (1989). Kraulis, P.J.J. AppL Crystallogr. 24, 946-950 (1991). Declerck, N., Machius, M., Chambert, R., Wiegand, G., Huber, R. & Gaillardin, C. Prot. Eng. 10, 541-549 (1997).



325

Thermophilic esterases and the amino acid "traffic rule" in the h o r m o n e sensitive lipase subfamily Giuseppe Manco*, Ferdinando Febbraio* and Mos~ Rossi *~ *Institute of Protein Biochemistry and Enzymology, CNR, via Marconi 10, 80125, Naples, Italy *University of Naples "Federico II", via Mezzocannone 16, 80100, Naples, Italy

We recently reported on the identification, over-expression and partial characterisation of a new esterase from the thermoacidophilic eubacterium Bacillus acidocaldarius homologous to the hormone sensitive lipase (HSL)-like sub-group of the esterase/lipase family. A comparative analysis of the amino acid compositions in this family quite unexpectedly revealed some trends in amino acid changes that are presumably related to both protein thermal stability and/or thermophilicity. Here we further extended this analysis to an acetylesterase (AES) of E.coli and to a new homologous gene from the hyperthermophilic archaea Archeoglobus fulgidus. Moreover an analysis of data with the statistical approach of Argos et al.(1979) confirmed the aforementioned trends in amino acid changes going from psychrophilic to hyperthermophilic members of the HSL group and revealed some new features.

1. INTRODUCTION Some studies (1, 2) have suggested that esterases, lipases and cholinesterases belong to a large family of phylogenetically related proteins with representatives in the domains of Eukarya and Bacteria and include proteins lacking enzymatic activity. Three subfamilies have been identified: the C group includes cholinesterases from Vertebrate and Invertebrate, lipases from Fungi, a number of esterases and some non-enzymatic proteins; the L group includes lipases from Vertebrate and Bacteria, lipoprotein lipases, lecithin-cholesterol acyltransferases and related non-enzymatic vitellogenins from flies; the third group, H, was also named HSL by Hemil~i et al. (1) since these authors reported the cloning and sequencing of a gene from B. acidocaldarius which encoded a protein of unknown function (ORF3) homologous to the HSL from human and rat. In this group were included also: the N-acetylphosphinothricintripeptide deacetylase from Streptomyces hygroscopicus; the acetylhydrolase from Streptomyces viridochromogenes; the lipase 2 from Moraxella TA144 (MOL); an esterase from Acinetobacter calcoaceticus (ACE); the esterase AES encoded by the ORF203 of E. coli; Vsh5, a protein from the slime mold Dictiostelium discoideum; and the recently sequenced human liver arylacetamide deacetylase (HDAC). Unfortunately, it is unclear at the moment whether this sequence similarity implies comparable biochemical properties for these enzymes. New enzymes from thermophilic bacteria could add new insights into the evolutive

326 relationships in the esterase/lipase family from one hand and from the other they could allow fundamental studies in the field of protein stability for this class of proteins. Moreover they are giving rise to growing interest in the biotransformations field. To date, carboxylesterases have been purified and partially charaeterised only from Sulfolobus acidocaldarius (3), Bacillus stearothermophilus (4-6) and B. acidocaldarius (7). We recently reported on the purification of a new esterase from B. acidocaldarius and demonstrated its identity with the aforementioned ORF3 (8). The gene was over-expressed in E. coli purified and partially characterised to get more insight into the structure-function relationship of this interesting representative of the HSL group (9). Here we report the alignment of an hyphotetical protein from A. fulgidus (Accession number: AEOOO782 from GenBank TM database, Release 103), with several members of the HSL group and an analysis of the respective amino acid compositions with the aim to identify amino acid changes that are related to both protein thermal stability and/or thermophilicity.

2. MATERIAL AND METHODS

2.1. Computer analysis The procedure for the multisequenees alignment was as previously described (8). In short the alignment between Candida rugosa lipase (CRL), Torpedo californica acetylcolinesterase (TAC) and Geotrichum candidum lipase (GCL) based on 3-D superpositions was taken from ref. (10). EST2, AFES, and MOL sequences, automatically aligned by the MAXHOM (11) method of the PHI) server (http:www.embl-heidelberg.de) were manually aligned to the above three sequences on the basis of secondary structure correspondences. Secondary structure predictions were obtained using the PredictProtein PHDsec program (12) from the PHD server. For HSL the secondary structure-driven alignment to CRL, TAC and GCL was taken from ref. (13) although the protein sequence was from mouse, which displays a 27.3 % identity with EST2 in a 289 amino acid overlap using the FastA program of the GCGSequence analysis package (14). Gaps were kept to minimum and, where possible, were not allowed inside the structural elements. 2.2. Amino acid composition analyses The amino acid composition analyses were all obtained from the respective published sequences by using the AAeomp program (SBDS). 2.3. Statistical analysis The statistical analysis of aligned sequences was perfomed according to Argos et al. (15) by generating an "ad hoe" program writed in Q-basic language on a Hewlett Packard personal computer. 3. RESULTS AND DISCUSSION In a previous report (8) we predicted for EST2, a carboxylesterase from the thermoacidophilic eubacterium B. acidocaldarius encoded from the ORF3 gene (1), the or/13 hydrolase fold of the esterase/lipase family. Moreover we made a secondary structure-driven multisequence alignment of EST2 with other members of the HSL subfamily.

b3

CRL

(2 8)

- FAEPP

.... VGNLRFKD.

TAC

(3 5)

- FAEPP

.... VGNMFRR

MOL

(8 5 )

-LAIDPK.

GCL EST2

HSLM AFES

CRL

(2 9) (I)

(253)

(4)

- FADPPVG.

.D L R F K H P Q

. .P V P Y S G S L D G Q K F T L Y G P L C M Q Q N P E G T Y E E N L P K A A L D L V M O S

VSRLLSL..

MVWI YGGAFVFGSSASYPGNGYVKESVEM

EST2

LVY.____~GGGWVVGDLETH.

MOL

.QHERVERVEDRT

. .D V Y N G K Y L A Y T E

MLFFHGGGFCIGDIDTH.

. .D P V C R V L A K D G .

.G Q P V V F V S

.E V V L V S L S Y R V G A F G F L . .W A V V S V D Y R M A P

VVHIHGGGFVAQTSKSH.

CRL

TYKGKPLFRAGIMOSGAMVPSDA.

. . VDGIYGNEI

. . .S R D L F R R A I L Q S G S P N C P W A .

.S V S V A E G R R R R A V E L G R N L N C N L N

C S I E s H . . .D A L C R R I A R L S

LVYYHGGGFVI

.V P I I S I D Y S L A P

86 GCL TAC

EST2

MOL HSLM AFES

CRL GCL TAC

EST2

MOR

HSLM AFES

TYNGKKLFHSAI

LQSGGPLPYFDSTSVGPESAYA

. . .V R V P D G I M A A .

YPVTDYEAEYP YPVTTLQ

.... EDFI KHQILIYPVVNF.

QLCPSYIVVAEL.

DILRDEG.

NLPPALIITAEY.

DPLRDEG.

Figure 1:

.............

EGYLLTGGMMLWFRDOYLNSL

.............

EGLWI

FPELSGRKP

LDQKIMSWFSEQYFSR

[38

" u I,., FLS KQLSGLPVLGTFHSNDIVFODYLLGSG

DPMLDDS.

KLYAEALNKAGVKVEIENFE

VMFARRLRDLGQPVTL~E

EVFGQMLRRAGVEASIVRYR

.L R I D P S

. NTPGF.

.S D E E L I H C L R E K K P Q E L I D V E W N V L P F D S I F R F S

.S L L E F G

LAYAELLQKEGVQVQTYTVL

" us., FGESAGSMSVMCHIL

...............

...............

...............

...............

.............

...... DLIHGFAQFYS__~L. SPG .......... HQGLGNQTTYI

...... GVLHGFINYYPV.

LKA ..........

...... DLPHGFLSLAALCRET.RQATEFCVORI

WNDGDN

.I FGESAGAMSVAHQLVAYGGDN

..

FGESAGGASVGMHILSPG.

KI FVGGDSAGGNLAAAVS

IMARDSG

. . LAYSSLRLSYLPRPDGVNITDDMYALVEGKY

. 9. F G L L P Q F L G F G P R P D G N I .......

QKTT.

I PDAAYELYSGRY

FVPVIDGEFFPTSLESMLNGNF

EELTHPWFSPVLYP

HSGLPQSHPLI

SVMHG

..............

u2 ,,, SLIYNNAFIAFA.

DLS

..............

DNT

FA ..............

DLE

S PTAESVRPTESMRRspLLAPDSM

EEDKFNPLASVI

............. SA. YRRYFI SFAN. YT. AEEEALSRRIMHYWATFAKTG-

...... GAPHGFINLMSV.

V

13r.4 ,~

GCDASASDNETLACLRSKSSDVLHSAQNSYDLKDL

...... ASSWLNS

..........

~3,~

. KLACLRGVS.SDTLFDATN.

EGLLLDHNDAEVFNSAYTQ

S 1 GVMGLVQRDTSLFLRDLRLG

VAPTP

...... RSRAL

.L G A S P S R I V L S G D S A G G C L A A L v t F K N S L A D

EHKFPAAVYDCYDATKWVAENAEE

.............

YPA

DEAA

D .F H L D P A R I A V G G D S A G G N L A A V T . S I L A K E R G

EYPAPTALKDCLAAYAWLAEHSQS

DEGT I LAPVAINATTTPnvNRWTYLATQLHNLPFLGTFHGSDLLFQYYAGPWS DEGSFFL... LYGAPGFSkgTYLYFFNHRASneWMGVIHGYEIEFVFGLPFVKELN.

GLPPAYIATAQYYD.PLRDVG. SLPPVHIVACAL_.

..........

.F G G D P K T V T I

KPDSP

KSTTV

........

EAPFPRALEECFFAYCWAVKHCDDLLGSTGERICLAGDSAGGNLCITVSLRAAAYG

0:6, ~ NA.. G C . G S A S D .

.FGGDPKVM

EHKFPAAVEDAYDALQWIAERAA

. .S W E L Y G

[37 " u I~., u 2~,, A N I IP _V _I~ N N D E G T F F G T s r E Y F K O L F n g T K Y S AKVPYITGNQE. KKTQILLGVNK.

FDLL~.

VRVYQQ.

135 FGGDPTKVTI

SQEAPGNVGLLDQRMALQWVHDNIQF

............

l~1 7

. . .G P A Y I A F _ _ Q _ ~ Y P S T G Y D P A H P P A S I E E N A

. . .L P R P L A Q L P L .

ALHG.

............

.N S T V V S V D Y R L A P

IKGRNGDIR

...................

TAEGNTNAGLHDQRKGLEWVSDNIAN

............

HSLM

AFES

.G G D A I

............

RAVVFSVDYRLAP

.H E F C H ~ C A Q T G .

.E p Y L K S W A O E L G .

INYRTGPYGFL

........

SWQDKTIANADGGDMTVRCYQSTQNSERKST.

%., .I KAEGSANAGLKDQRLGMQWVADNIAA.

84 GKPI.IHVSVNYRVSSWGFLAGDE

MDLPGRTLKVRMYRPEGVEPP

...... LARLISYDLREGQDSKVLNSLAKSEGPRLDVRPRPHQAP

9P P E A F E M P L T S D P R L T V T I S P P L A H T G P A P V

a,,, 83 MLWIFGGGFEVGGTSTFPPAQMITKSIAM.

EMWNPNREMSEDCLYLNIWVPSPRP

...............................

-MPIDPVYYQLAEYFDSLpfSSAREYREAINRIYEERNRQLS.

...... ANLPV

...... DAKLPV

GPLyqGSVSMNEDCLYLNVFRPAGTKP

....................

9. P A P D Y K H L S A Q Q F R S Q Q S L F P P V K K E P V A E V R E F D

.L R Q K F G T D A V S L ~ P S V W Q Q N A D A S G S T E N A V

MVWIYGGGFYSGSSTL.

...............

.... PEPKKPWSGVWNASTYPNNCQQYVDEQFPGFSGS

GCL

TAC

82 ..... KVFEAVLPLSEDCLTINVVRPPGTKAG

..... PFTGSYQGLKANDFSSACMQLDPGNAISLLDKVVGLGKILPDNLR.

MPLDPVIQOVLDOLNPd!4__.

-SLANMAS...

c%,,2

9. L K

TD-

H-

ATKALVRIAEKLRDALA

INEFACLVQNLLTSEGDKPNLRA RLI LTPPAAPLN

ARDAINQIAALLVFD

Multiple sequences alignment of AFES with some HSL subfamily enzymes and with CRL, TAC and GCL. The SwissProt codes are: CRL, p20261; GCL, p04058; TAC, G406508; MOL, p24484; HSLM, p54310; The sequence of AFES, was taken from GenBank TM database, Release 103 Accession number: AEOOO782, whereas that of EST2 was taken from ref. (1). [3-strands and or-helices of CRL (16) are indicated as single or double lines, respectively. The stretches of hypothetical [3-strands or a-helices in EST2 (8) and HSL (13), estimated using the PredictProtein PHDsec program, are indicated in the same way. The serine, aspartic and histidine residues, most likely to be at the active site in the HSL-group, are indicated by an asterisk. The residue couples in lowercase letters indicated insertions that are not shown. Finally, the residues which are conserved among all sequeces or in the HSL group are indicated in bold.

328 Searching the GenBank TM database we have identified a new hypothetical protein from the hyperthermophilic sulphate-reducing archaeon A. fulgidus reported to share 42% identity with the A. calcoaceticus esterase, a member of the HSL group. In Figure 1 we show a multisequence alignment which includes the new sequence. The alignment revealed the presence of highly conserved residues already identified by Cygler et al. (10) in the esterase/lipase family, as well as residues which are common to the HSL group. It is worth mentioning that the region corresponding to the lid in CRL (16) was partially deleted in the AFES sequence. This finding, suggests that the enzyme is not a lipase. The location of the elements belonging to the catalytic triad Ser...Glu/Asp...His proved particularly interesting. Table 1 Amino acid compositions comparison ASE AMINO MOL ACE (37~ ~ ACIDS (5~ ~' (30~ e'

Ala Cys Asp Glu Phe Gly His Ile Lys Leu Met Asn Pro Gin Arg Ser Thr Val Trp Tyr

# mole% 50 11.55 8 1.85 26 6.00 19 4.39 9 2.08 26 6.00 17 3.93 19 4.39 21 4.85 52 12.01 6 1.39 21 4.85 25 5.77 25 5.77 13 3.00 26 6.00 25 5.77 25 5.77 6 1.39 14 3.23

HSL* (37~ a'

mole% # mole% # mole% # 10.32 28 8.70 32 10.03 80 9 2.82 14 1.81 (2.19) 4 1.24 6.58 31 4.00 21 6.52 21 5.33 48 6.19 (5.47) 16 4.97 17 4.70 33 4.26 15 4.66 15 6.58 58 7.48 16 4.97 21 6 1.88 18 2.32 10 3.11 3.76 22 2.84 19 5.90 12 8 2.51 16 2.06 16 4.97 13.81 45 13.98 37 11.60 107 3.13 19 2.45 7 2.17 10 8 2.51 19 2.45 (2.19) 10 3.11 6.27 55 7.10 (6.72) 23 7.14 20 5.64 24 3.10 19 5.90 18 5.64 56 7.23 13 4.04 18 4.08 72 9.29 17 5.28 13 5.96 41 5.29 (5.31) 17 5.28 19 4.08 40 5.16 13 4.04 13 5 1.57 6 0.77 3 0.93 5.33 16 2.06 10 3.11 17

EST2 (65~ a'

AFES (83oc)e'

# mole% # mole% 37 11.94 28 9.00 1 0.32 3 0.96 21 6.77 22 7.07 20 6.45 24 7.72 12 3.87 16 5.14 22 7.10 17 5.47 8 2.58 6 1.93 11 3.55 23 7.40 12 3.87 10 3.22 34 10.97 25 8.04 6 1.94 5 1.61 6 1.94 11 3.54 28 9.03 16 5.14 12 3.87 11 3.54 16 5.16 24 7.72 13 4.19 20 6.43 8 2.58 5 1.61 23 7.42 25 8.04 4 1.29 3 0.96 16 5.16 17 5.47

Amino acid compositions were all obtained from the respective published sequences. * In the case of HSL the results do not change significantly (data in parenthesis) if the phosphorylation domain [1] is excluded from computation. a, This is the temperature midpoint of source microorganism. The location of the active serine (Ser 160) is clearly indicated by the conserved pattern "GlyXISX2Gly" and its topological position in a 13-eSer-a motif aider strand 135 (Fig. 1) (17). In the C-terminal part of the protein, His 285, which was well conserved in all members of the HSL subfamily (HGF/A motif) as well as in CRL, TAC and GCL, is most likely to be the

329 histidine of the active site. The acidic element of the triad could be Asp 255 according to the alignment with the EST2 sequence (8). Site-directed mutagenesis experiments recently made on the correspondig Asp and His residues in human HSL (13) and in EST2 (G. Manco, unpublished results) are in agreement with this assignment Thus the HSL subfamily includes hydrolases from Bacteria, Archaea and Eukarya, as well as from psychrophilic, mesophilic, thermophilic and hyperthermophilic sources. Comparative studies within this group would be of valuable interest in the field of protein stability determinants. We previously observed (9) some trends of amino acid changes in the HSL subfamily going from the psychrophilic to the thermophilic enzymes. Now we have extended this analysis to the AES and AFES enzymes. In Table 1 we show the amino acid compositions of several members of the HSL group. As a result some trends in amino acid changes were still evident (Glu and Cys) some other were not confirmed (Pro and Asn) and finally a tendency to change Thr was more evident. Note that the Cys content in the AFES was higher than in EST2. A possible explanation for this behaviour is that being A. fulgidus an anaerobic archaeon the enzyme is much less sensitive to Cys oxidation. With the aim to analyse more carefully these results we decided to exploit the method of Argos et al.(15) which should be able to test statistically the significance of the amino acids drifts towards the "warmer" sequences. A program able to analyse automatically a multisequence alignment was generated and used to analyse the sequences indicated in Table 1. The multisequence alignment (not shown) was based on the same criteria as those used for the alignment of Figure.1. The results obtained are shown in Table 2 and diagrammatically in Figure 2. The outpout of the program is a matrix of the ratios Eij between the net weighted exchange per degree Celsius to the standard error. Only values of Eij_>_l.7 c were taken into account and ranked from the higher to the lower. Table 2 Representative amino acid changes Exchanges with Cold ~ Hot Ei] _>1.70 L Q L S T T Q Q E

=~ =~ =~ =~ =~ =~ ~ ~ =~

V P A A L R K R A

5'82 2.97 2.28 2.23 2.19 2.16 2.16 2.05 2.04

Cold

L E L C Q G N S

~

Hot

~ ~ ~ ~ =~ =~ ~ =~

Y R E L N R I P

Exchanges with Eij >_1.70

1.89 1.88 1.88 1.80 1.77 1.72 1.72 1.70

First of all it is clear that the observed changes are not coincident with those reported from Argos et a/.(15) except for the Ser--~Ala change in the cold--hot direction. However we confirm the existence of "psichrophilic" and "thermophilic" residues with Gly, Ser, and Gin belonging to the former and Ala, Arg, Leu, Glu and Asn belonging to the latter. In contrast with the Argos data on the "cold" side we also found Leu, Thr, Glu, Asn and Cys whereas on the "hot" side we have Val, Pro, Ile, Leu, Lys, and Tyr. The difference in results could be attributed to the higher temperature range spanned by the sequences we used and/or to the

330

1 4.-.

N ~._.. Q . . . ~

P 4--

S -.~

Y

Figure 2. The observed exchanges in the direction of thermophilic preferences. The marked arrows indicate that a single mutation event in the codon was necessary for the amino acid change. higher difference in the evolutive distances between the tested species. Alternatively differences could be family-specific. Much more enzyme families need to be analysed to answer this point. As shown in Figure 2 several of found exchanges could be easily obtained by means of single base change per codon. Other exchanges, especially those involving Leu require additive mutational events. A possible explanation could be the higher Leu content of all sequences respect AFES. REFERENCES

1. Hemil~i,H., Koivula, T.T., and Palva, I. (1994) Biochim. Biophys. Acta 1210, 249-253. 2. Krejei, E., Duval, N., Chatonnet, A., Vincens, P., and Massouli~, J. (1991) Proc. Natl. Acad. Sci. USA 88, 6647-6651. 3. Sobek, H. and Gorish, H. (1989) Biochem. J. 250, 453-458. 4. Matsunaga, A., Koyama, N., and Nosoh, Y. (1974) Arch. Biochem. Biophys. 160, 504513. 5. Owusu, R. K., and Cowan, D. A. (1991) Enzyme Microb. Technol. 13, 158-163 6. Wood, A. N., Femandez-Lafuente, R., and Cowan, D. A. (1995) Enzyme Microb. Technol. 17, 816-825. 7. Manco, G., Di Gennaro, S., De Rosa, M., and Rossi M. (1994) Eur. J. Biochem. 221,965972. 8. Maneo, G. Adinolfi, E., Pisani, F.M., Carratore, V., and Rossi, M. (1997) Protein and Peptide Letters 4, 375-382. 9. Manco, G. Adinolfi, E., Pisani, F.M.,Ottolina,G., Carrea, G. and Rossi, M. (1998) Biochem. d. in press. 10. Cygler, M., Schrag, J. D., Sussman, J. L., Harel, M., Silman, I., Gentry, M. K., and Doctor, B. P. (1993) Protein Science 2, 366-382. 11. Sander C. and Schneider, R. (1991) Proteins Struct. Funct. Genet. 9: 56-58. 12. Rost, B., and Sander, C. (1993) I Mol. Biol. 232, 584-539. 13. Contreras, J.A., Karlsson M., Osterlund, T., Laurell, H., Svensson, A., and Holm, C. (1996) J. Biol. Chem., 271, 31426-31430. 14. Devereux, J., Haeberli, P., and Smithies, O. (1984) Nucleic Acids Res. 12, 387-395. 15. Argos, P, Rossmann, M.G., Grau, U.M., Zuber, H., Frank, G., and Tratschin, J.D. (1979) Biochemistry, 18, 5698-5703. 16. Grochulski, P., Li, Y., Sehrag, J.D., BouthiUier, F., Smith, P., Harrison, D., Rubin, B., and Cygler, M. (1993)I Biol. Chem. 268,12843-12847. 17. Derewenda, Z.S. and Derewenda, U. (1991) Biochem. Cell. Biol. 69, 842-851.


331

Cloning and stabilization of NAD-dependent formate dehydrogenase from C a n d i d a boidinii by site-directed mutagenesis H. Slusarczyk, M. Pohl and M.-R. Kula Institut ~ r Enzymtechnologie, Heinrich-Heine Universitat Duesseldorf, Forschungszentrum Juelich, D-52426 Juelich, Germany

The gene of the NAD-dependent formate dehydrogenase from the yeast Candida boidinii was cloned by polymerase chain reaction and expressed in E. coil The recombinant enzyme was stabilized by site-directed mutagenesis. Two cysteine residues probably located at the surface of the protein were exchanged against various aliphatic amino acids and the mutant enzymes were tested for stability against oxidation by air in the presence of catalytic amounts of Cu(II). All mutants were significantly more stable against oxidative stress than the wildtype enzyme. Additionally, the mutant enzymes were characterized with respect to their temperature optima, kinetic constants and activation energy of the enzymatic reaction. The results of these studies indicated that the catalytic properties of the mutant enzymes have not been altered by the mutagenesis compared to the wildtype enzyme.

1. INTRODUCTION NAD-dependent formate dehydrogenase (EC 1.2.1.2) is found in many methylotrophic bacteria and yeasts [ 1]. The enzyme catalyzes the oxidation of formate to carbon dioxide with the concomitant reduction of NAD to NADH. Because of the favourable thermodynamic equilibrium of the reaction the enzyme is used for in situ regeneration of NADH during asymmetric synthesis of chiral compounds [2,3].The enzyme is efficiently formed as an intracellular enzyme by the yeast Candida boidinii, if methanol is the only energy and carbon source available [4].In order to perform enzymatic reductions on large scale a production and downstream process has been established [5]. Though FDH from C. boidinii (Cb-FDH) can be prepared in large quantities and was used in several enzymatic redox reactions [2,3] the enzyme is less stable against oxidative stress [6] resulting in higher costs for the biocatalyst in synthesis reactions. Therefore, the enzyme from the yeast was cloned and expressed in E. coli. This paper deals with the stabilization of the recombinant enzyme by site-directed mutagenesis and the enzymatic characterization of the resulting mutants. Their properties are compared to the native enzyme from C. boidinii.

332 2. MATERIALS AND METHODS

2.1 Cloning of the fdh-gene The cloning procedure is described elsewhere [7] 2.2 Site-directed mutagenesis In vitro site-directed mutagenesis was performed according to the method of rio et al. [8] using pUC-FDH as a template. 2.3 Expression in E coli Recombinant bacteria were cultivated on LB-medium containing 100 ~tg AmpiciUin/ml. Expression was induced by addition of 1 mM IPTG (isopropyl-13-D-thiogalactoside) aider A550 reached 0.5. The cells were harvested by centrifugation 3 h ~ e r induction. Cell desintegration was carried out by glass beads in a mixer mill and the cell free crude extract were analyzed for activity and protein content according to the method of Bradford [9], using BSA as a standard. 2.4 Measurement of FDH acitivity FDH activity was assayed spectrophotometricaUy at 340 nm using a Shimadzu photometer UV160. The assay mixture was composed of 162 mM sodium formate and 1.62 mM NAD in 100 mM potassium phosphate buffer, pH 7.5. The mixture was thermostated to 30~ unless otherwise stated and the reaction was started by addition of limiting amounts of enzyme. The increase in absorbance at 340 nm was followed and activities calculated by using e = 62201 mol1 cmq for NADH at 340 rim. One unit is defined as the amount of enzyme which catalyzes the reduction of 1 ~mol NAD per min at pH 7.5 and 30~ 2.5 Kinetic constants Kinetic constants were derived from duplicate measurements of initial velocity under conditions varying one substrate and saturating concentrations of the second substrate. The corresponding data were fitted to a model for double substrate kinetics according to MichaelisMenten, using the software 'Scientist for Windows'(Microsofl| 2.6 Temperature dependencies of enzyme activity Temperature dependencies of the activity of the native, the recombinant wildtype enzyme (rec-FDH) and the mutant enzymes have been tested by increasing the incubation temperature under standard assay conditions. Activation energies were calculated from Arrhenius plots using the linear segments of the lnvmax versus 1/T plots. 2.7 Thermal inactivation Thermal inactivation experiments were carried out by incubating the native enzyme in 100 mM potassium phosphate, pH 7.5, over a temperature range between 30 an 70~ After 20 rain aliquots were removed and the residual activity and protein content were determined. 2.8 Stability in presence of Cu(IO All enzymes were tested for stability against oxidative stress by incubation in the presence of 10 ~M CuCI2 in 100 mM potassium phosphate, pH 7.5, at 25~ Controls were incubated without CuC12. At certain time intervalls, aliquots were removed and residual activity was determined.

333 3. RESULTS AND DISCUSSION

The fdh gene from Candida boidinii was amplified from a genomic DNA template by polymerase chain reaction and cloned into pUC 18. The Cb-FDH open reading frame (ORF) codes for a 364-aa protein (40 370 Da) which shows highest homology to FDH from C. methylica [10]. Only a single aa exchange could be detected: Leu at position 87 in Cb-FDH corresponds to Ser in Cm-FDH. The wt-fdh gene was cloned into the expression vector pBTac2 and transformed into E. coli JM105. FDH was overproduced after induction with IPTG to about 20-25 % ofE. coli soluble cell protein. The thermal stability of Cb-FDH was tested by stepwise increasing the incubation temperature up to 70~ At several time intervals aliquots were removed and subsequently the residual activity and the soluble protein content were determined. Fifty percent of activity was lost after incubation at 57~ for 20 min. Similar values have been obtained for FDHs from the yeasts Kloeckera sp.(55~ and Pichia pastoris (54~ [ 11]. Thermal inactivation of the CbFDH above 55~ is paralleled by a loss of soluble protein in the enzyme preparation. The loss of activity at higher temperatures is caused by an aggregation process, which is not readily reversible lowering the temperature again. Kelly and Zydney [12] examined the role of oxidation of sulfhydryl groups in bovine serum albumin (BSA) on fouling during microfiltration. They found that aggregation of BSA during the filtration process could be minimized by adding sulfhydryl groups-modifying agents like iodacetamide or dithiothreitol to a prefiltered protein solution. The results presented in their study demonstrated that B SA aggregation in these filtration systems occurred via an intermolecular thiol-disulfide interchange reaction, possibly in combination with thiol oxidation reactions. Based on this observations two cysteine residues, Cys23 and Cys262 in the recombinant wtFDH (rec-FDH) were exchanged by various aliphatic amino acids by site-directed mutagenesis resulting in four different mutants. Using the cfistal structure of FDH from the bacterium Pseudomonas sp. 10] [ 13] as a model, the substituted cysteine residues are likely to be located at the surface of the protein and therefore highly accessible for oxidation by molecular oxygen. The stability of the mutant enzymes in comparison with the wt-enzyme against oxidative stress was estimated by incubation of partially purified enzyme preparations in the presence of catalytic amounts of divalent copper ions. Traces of divalent ions like Cu(II) or Fe(II) were shown to promote oxidation processes at thiol groups[ 14,15].

334

Figure 1. Residual activities ofwt-FDHs and various mutant enzymes incubated in 100 mM potassium phosphate, pH 7.5, for 15 h at 25~ in absence and presence of 10 ~tM CuCI2, respectively.

All mutant enzymes showed a significantly higher stability in presence of CuCI2 compared to Cb-FDH as well as rec-FDH (Figure 1). Even after 15 hours of incubation in presence of Cu(II) ions a residual activity of about 70% could be determined in all mutant enzyme samples whereas the wt-activity decreased to about 8%. The results suggest that the inactivation of the wt-enzymes (Cb-FDH and rec-FDtO is caused by oxidation of free sulfhydryl residues. Further studies characterizing the inactivation process are under investigation. In order to compare the catalytic and enzymatic properties of the mutant enzymes and the wt-enzymes, all enzymes were characterized with respect to their temperature optima, kinetic constants and activation energy, respectively. Figure 2 demonstrates that both the wt-enzymes and the mutant enzymes show broad temperature optima between 45 - 55~ Obviously, the temperature optima of the mutant enzymes are slightly reduced to 50~ compared to the wtenzymes. This effect is probably due to a decrease in conformational stability of the mutant enzymes caused by the amino acid substitutions.

335

100 --A--C23S -- x7-- C262V - o - C23SC262V - +- C23SC262A /

80 o~" 60 "~

,\ i ~

~~

40 20

10

'

20'

'

3'0

'

' ;o 40' temperature [~

'

60'

'

70'

Figure 2. Effects of temperature on the activity of Cb-FDH, rec-FDH and various mutant enzymes. The linear segments of the Arrhenius plots are used for determinating the activation energies. In all enzyme preparations FDH-activity is proprotional to the incubation temperature in the range of 20 - 40~ and the calculated activation energies are similar or increased compared to the wt-enzyme (Table 1). Table 1 Kinetic constants and activation energies of Cb-FDH, rec-FDH and various mutant enzymes. Km Formate [mM]

Km HAD [pM]

Ki NADH [BM]

activation energy [kJ/mol]

Cb-FDH

5.6 +/- 0.4

45 +/- 2.9

17 +/- 1.2

55

recFDH

5.9 +/- 0.4

37 +/- 3.0

21 +/- 1.8

54

FDH-C23S

5.5 +/- 0.4

44 +/- 3.5

16 +/- 1.3

62

FDH-C262V

4.3 +/-0.5

41 +/-5.2

8.8 +/-1.2

55

FDH-C23S/C262V

4.5 +/- 0.6

44 +/- 5.6

11 +/- 1.5

59

FDH-C23S/C262A

5.0 +/-0.3

45 +/-3.0

13 +/- 0.9

68

As demonstrated in Table 1, the kinetic constants of the mutant enzymes compared to the wtenzymes are not significantly altered.

336 The present study represents the succesful stabilization of NAD-dependent formate dehydrogenase from C. boidinii by site-directed mutagenesis of two oxidation-sensible cysteine residues. These amino acid exchanges did not influence the catalytic properties of the enzyme. The potential of the mutant enzymes as stable biocatalysts for the in situ regeneration of NADH is under investigation.

ACKNOWLEDGEMENT This work was supported by BMBF AZ 03D0031 C7.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

V.O. Popov and V.S. Lamzin, Biochem. J. 130 (1994) 625. A.S. Bommarius, K. Drauz, W. Hummel, M.-R. Kula and C. Wandrey, Biocatalysis 10 (1994) 37. G. Krix, A.S. Bommarius, K. Drauz, M. Kottenhahn, M. Schwarm, M.-R. Kula, J. Biotechnol. 53 (1997) 29. H. Sahm, Adv. Biochem. Eng. 6 (1977) 77. D. Weuster-Botz, H. Paschold, B. Striegel, H. Gieren, M.-R. Kula, and C. Wandrey, Chem. Eng. Technol. 17 (1994) 131. H. Sch0tte, J. Flossdorf, H. Sahm and M.-R. Kula, Eur. J. Biochem. 62 (1976) 151. H. Slusarczyk, PhD Thesis, D0sseldorf University, Germany, 1997. S.N.Ho, H.D. Hunt, R.M. Horton, J.K. Pullen and L.R. Pease, Gene 77 (1989) 51. M.M. Bradford, Anal. Biochem. 72 (1976) 248. S.J. Allen and J. Holbrook, Gene 162 (1995) 99. J.J. AUais, A. Louktibi and J. Baratti, Agile. Biol. Chem. 47 (1983) 2547. S.T. Kelly and A.L. Zydney, Biotechnol. Bioeng. 44 (1994) 972. V.S. Lamzin, Z. Dauter, V.O. Popov, E.H. Harutyunyan and K.S. Wilson, J. Mol. Biol. 236 (1994) 759. J.M. Kolthoff and B.R. Willeford jr., J. Am. Chem. Soc. 80 (1958) 5673. L. Michaelis, J. Biol. Chem. 84 (1929) 777.


U s e f u l n e s s of b a c t e r i a l t h e r m o s t a b l e analysis

337

e n z y m e s in clinical chemical

K. Tomita a, K. Nomura ~, T. Miura" and H. Shibata b aCollege of Engineering, Kanto Gakuin University, 4834 Mutsuura, Kanazawa-ku, Yokohama 236-8501, Japan bHead Office, Iatron Laboratories, Inc., 1-11-4 Higashikanda, Chiyoda-ku, Tokyo 101-0031, Japan

Since enzymes have many excellent characteristics, e.g., high substrate and reaction specificities, they have become popular in clinical chemical analysis. However, many of the usual enzymes tend to lose their activities on storage, especially in a solution state, even at rather low temperature, which limits their uses remarkably. On the other hand, enzymes from thermophiles are known to be superior to usual enzymes in thermal and storage stabilities. The author et al. [1] have successfully prepared many thermostable enzymes from a thermophilic bacterium, Bacillus stearothermophilus (BS). Moreover, we have obtained several enzymes from a mesophilic bacterium, Zymomonas mobilis (ZM), and found they also have good thermal stability[2]. In this paper, we show their usefulness as tools in clinical chemical analysis, with focusing the application of two kinds of them, i.e., glucokinase (GlcK, EC 2.7.1.2) and alanine dehydrogenase (AlaDH, EC 1.4.1.1), to the determination of some components in a serum. 1. MATERIAL AND METHODS 1.1. Material Glck was prepared from BS cell or ZM cell. AlaDH was prepared from Escherichia coll C 600 cell carrying the plasmid-encoding BS AlaDH gene [3]. Glucose 6-phosphate dehydrogenase (G6PDH, EC 1.1.1.49, from Leuconostoc mesenteroides) was obtained from Boehringer Mannheim GmbH, Mannheim, Germany. The other chemicals were commercial products of analytical grade. 1.2. Methoda The enzymatic activities of GlcK and AlaDH were measured at 30~ on the bases of the change in absorbance at 340 nm due to the formation of NADPH (GlcK coupled with G6PDH) or NADH (AlaDH) according to the usual method.

338

Four kinds of reagents for the determination of the components in biological fluids, mainly sera, were compounded from adequate reactants mentioned in the Sections of 2.3.-2.6. A sample was added to the reagent, incubated at 30~ (glucose determination) or 37~ (the others), and the formation of NADPH or NADH was spectrophotometricaUy measured at 340 nm. 2. RESULTS AND DISCUSSION 2.1. Properties of GlcK The main properties of BS GlcK [4] and ZM GlcK [2] were as follows. The molecular weights were determined to be 68,000 (BS) and 65,000 (ZM), and 32,000 (BS) and 33,000 (ZM) for the subunits. The optimal pHs for the enzymatic activities were 8.5 (BS) and 7.0-8.0 (ZM). BS GlcK was stable in a wide pH range over 8, and ZM GlcK was stable in a pH range from 6.0 to 8.0. Noteworthily, both GlcKs were fully active even after treatment for 60 rain at 60~ (BS) and 50-55~ in the buffer solution, whereas hexokinase (HK, EC 2.7.1.1) from yeast rapidly lost its activity at temperature above 35~ (Figure 1).

100 75 50 9~

25

~

0 0

25

50

75

100

Temperature (~

Figure 1. Thermal stabilities of both GlcKs dissolved in the buffer solution. Comparison with yeast HK. After incubation for 15 rain at different temperatures, the remaining enzymatic actim'ties were measured. 0 : BS GlcK; A: ZM GlcK; m: yeast HK.

Both GlcKs showed no activities when fructose, mannose and glucosamine were used as phosphoryl acceptors. On the other hand, both the mammarian GlcK and the yeast HK were fully active to these monosaccharides. The Km values were

339 determined to be 0.1 mM (BS) and 0.2 mM (ZM) for glucose and 0.05 mM (BS) and 0.5 mM (ZM) for ATP, being much lower than those of mammarian GlcK and the yeast HK. 2.2. Properties of AlaDH The molecular weight was determined to be 230,000, and 38,000 for the subunit. The optimal pH for the enzymatic activity was 10.4. AlaDH was stable over a wide pH range above 7, moreover, over a wide temperature range. AlaDH was fully active even after treatment for 60 rain at 70~ in the buffer solution. AlaDH showed no activity when L-leucine, L-isoleucine and L-valine were used as hydrogen donors. The Km values were determined to be 10.0 mM for L-alanine and 0.26 mM for NAD § These properties were similar to other AlaDH from mesophilic Bacillus sp., except for pH and thermal stabilities.

2.3. Determination of glucose In clinical chemical analysis, the determination of glucose is a very important factor for the diagnosis of diabetes mellitus. Although the method using HK as the key enzyme (HK method) is recommended as the reference method for glucose determination by several organizations, e.g., American Association for Clinical Chemistry[5], the properties of HK are much inferior to those of GlcK as described above. The application of BS GlcK to the glucose determination (GlcK method) [6] was

1t,511tl

2500 N 211110 o 15110

5OO 0

10

20

30

40

Time(day) Figure 2. Storage stability of the reagent solution for the determination of glucose. Comparison of GlcK method and HK method. After storage for several ten days at room temperature, each reagent solution was used for the determination of glucose in a serum sample. O: GlcK method; m: HK method.

340 attempted according to the principle 9Glucose is phosphorylated by GlcK in the presence of ATP to yield glucose 6-phosphate (G6P); G6P is converted to 6phosphogluconate (6PG) by G6PDH in the presence of NADP, which results in the formation of NADPH; finally, glucose is determined as the change in absorbance at 340 nm. The assay had a high degree of precision, i.e., coefficient of variation (CV) was 0.7-1.3% (within-run reproducibility) and 1.1-2.3% (between-run reproducibility ), and was linear up to about 8000 mg 1"1. A good correlation was observed between GleK method and HK method for both plasma and urine samples. The outstanding superiority of GlcK method to HK method was the storage stability of the reagent solution. The solution based on GlcK method was found to be stable at least for about one month at room temperature (Figure 2) [7] and 21 months at 10~ [8] , whereas that based on HK method was stable for only a few days at room temperature (Figure 2) [7]. ZM GlcK was also applied to glucose determination [2], and gave as excellent results as BS GlcK. As for the stability of the reagent solution, ZM GlcK gave a little inferior results to BS GlcK, but much superior to HK. 2.4. Determination of creatine kin~ae activity The determination of creatine kinase (CK, EC 2.7.3.2 ) activity is much in use for the diagnosis of myocardial infarction or progressive muscular dystrophy. The current method based on using HK (HK method) involves the same disadvantage in the stability of the reagent solution as that shown above. The new method [9] was also developed by using BS GlcK based on the principle: CK acts on creatine phosphate in the presence of ADP to yield ATP; followed by two reactions coupled with GlcK and G6PD, respectively, ATP causes a formation of NADH. The assay had a high degree of precision, i.e., CV was 1.12.2% (within-run reproducibility), and was linear up to about 2000 U 1-'. It was noteworthy that the reagent solution was found to be stable at least for one week at room temperature and for 13 months at 10 ~ [8]. The conventional HK method gave only one day stability at room temperature. 2.5. Determination of magnesium ion The determination of magnesium ion is a matter of significance for the diagnosis of renal diseases and gastrointestinal disorders. The colorimetric method, in which Xylidyl Blue is utilized as a detector of magnesium ion, is currently used, but this method has a defect in specificity. The atomic absorption spectrophotometry is looked upon as the most reliable method, but requires an expensive instrument. The reagent for magnesium ion determination [10] was successftflly composed with using GlcK on the basis of the paper of Tabata et a]. [11]. The principle of the assay was based on the fact that magnesium ion-ATP complex acts as a substrate of GlcK, as shown in Figure 3. The concentration of magnesium ion is then determined as an increasing rate in absorbance at 340 nm due to NADPH turnover. This simple and rapid spectrophotometric method was found to have a good correlation with the atomic

341 absorption spectrophotometry and to be very little affected by several cations in a serum such as K § Na § Zn s+, Ca s§ NH4+ and so on. The present method was recognized as to be more specific to magnesium ion than the Xylidyl Blue method. The assay had a high degree of precision, i.e., CV was 0.85-1.7% (within-run reproducibility) and 1.25-2.15% (between-run reproducibility), and was linear up to about 100 mg 1-'. This linearity range was considerably wider than that of the Xylidyl Blue method. The present reagent had an excellent stability at least for about one month at 10~ in liquid form. It should be emphasized that this method is the world's first practical enzymatic method for measuring metal ion. GlcK Glucose

~

Mg- ATP ATP

~

G6P M g - ADP Mga+ ~

""~ADP

G6PDH G6P + NADP +

6PG + NADPH + H §

Figure 3. The schematic principle of GlcK method for the determination of magnesium ion.

2.6. Determination of,/-glu~myltransferase activity Determination of~/-glutamyltransferase (v .GT, EC 2.3.2.2) activity is routinely employed in the diagnosis of hepatic diseases. The colorimetrlc method, in which ~/-glutamyl 4-nitroanilide (,/-Glu4NA) is utilized as a substrate for ,/-glutamyl donor, is most commonly used in current, but ~/-Glu4NA has a serious problem, i.e., poor solubility in water. The present method [12] was ultraviolet spectrophotometric one, in which the readily soluble substrate ~/-glutamyl dipeptide was used as the ~/-glutamyl donor and thermostable amino acid dehydrogenase as the attxiliary enzyme. In the preliminary experiment, ~/-glutamylalanine (~/-GluAla) was found to be the most specific and sensitive donor substrate. This method is consisted of two enzymatic reactions: First, ~/-GluAla releases L-alanine by the action of~/-GT in the presence of glycylglycine(GlyGly); second, L-alanine is converted to pyruvate by AlaDH in the presence of NAD, accompanied by NADH formation. NADH is continuously monitored at 340 nm. The present method is strictly stoichiometric, since one NADH is produced from one ~/-GluAla by every ~/-GT action. The assay had a high degree of precision, i.e., CV was 0.8-2.8% (within-run reproducibility) and about 2.2% (between-run reproducibility), and was linear up to about 700 U 1-1. The present method well correlated to the ,/-Glu4NA method. Although this method had about one-third

342 of the sensitivity to the T -Glu4NA method, it was adequate for the practical assay. The storage of the reagent solution did not affect the measuring activity at least over about 30 days at 10~ 3. CONCLUSIONS The bacterial thermostable enzymes, GlcK and AlaDH, had high degrees of stability, especially thermal and storage stab'dities, which made them useful in clinical chemical analysis. The other properties were also well suited for the analysis, i.e., low Km values, high substrate specificities and so on. These enzymes were successfully applied to the establishment of new enzymatic methods which gave long term stability of the reagent solutions as well as rapid, accurate and reliable results.

REFERENCES 1. K.Tomita, K.Nagata and K.Okada, Ann. N.Y. Acad. Sci., 672 (1992) 178. 2. K.Tomita and K.Nomura, Ann. N.Y. Acad. Sci., 750 (1995) 338. 3. Y.Sakamoto, S.Nagata, N.Esaki. H.Tanaka and K.Soda, J. Ferment. Bioeng., 69 (1990) 154. 4. S.Kamei, K.Tomita, K.Nagata, H.Okuno, T.Shiraishi, A.Motoyama, A. Ohkubo and M.Yamanaka, J. Clin. Biochem. Nutr., 3 (1987)1. 5. P.Duncan, J.Neese, D.Bayse and C.Stewart, Clin. Chem., 20 (1974) 882. 6. K.Tomita, S.Kamei, K.Nagata, H.Okuno, T.Shiraishi, A.Motoyama, A. Ohkubo and M.Yamanaka, J. Clin. Biochem. Nutr,. 3 (1987) 11. 7. K.Tomita, K.Nagata, H.Kondo, T.Shiraishi, H, Tsubota, H.Suzuki and H.Ochi, Ann. N. Y. Acad. Sci., 613 (1990) 421. 8. K.Tomita, K.Nomura, H.Kondo, K.Nagata and H.Tsubota, J. Pharm.Biomed. Anal., 13 (1995) 477. 9. H.Kondo, T.Shiraishi, M.Kageyama, K.Nagata and K.Tomita, J.Clin.Biochem. Nutr., 3 (1987) 17 10. T. Shiraishi, H. Suzuki, H.Ochi, K. Kawahara, H. Kondo, K. Nagata and K.Tomita, Jap. J. Clin. Chem., (in Japanese) 20 (1991) 37. 11. M.Tabata, T.Kido, M.Totani and T.Murachi, Agric. Biol. Chem., 50 (1986) 1909. 12. H.Kondo, M.Hashimoto, K.Nagata, K.Tomita and H.Tsubota, Clin. Chim. Acta, 207 (1992) 1.


343

Effect of temperature and Ca 2+ on the degree of multiple attack exhibited by mesophilic and thermophilic a - a m y l a s e s B. Kramhofi and B. Svensson Carlsberg Laboratory, Department of Chemistry, Gamle Carlsberg Vej 10, DK-2500 Copenhagen Valby, Denmark. The activity of (z-amylases depends on Ca 2§ and temperature. In the starch hydrolysis these depolymerases exert a distinct number of catalytic events per enzyme-substrate encounter referred to as the degree of multiple attack. The present results indicate that the degree of multiple attack for the thermophilic a-amylase from Bacillus licheniformis is reduced from about 2 to 1 following an increase in the Ca2+-concentration from 0.1 mM to 5 mM. The degree of multiple attack is unchanged following an increase in the temperature from 37 ~ to 80 ~ at both Ca2§ The multiple attack by an (z-amylase from a related thermophilic alkalophilic Bacillus is reduced at the lower temperature and in low Ca 2§ but is unaffected by the temperature at the higher Ca2§ At 37 ~ and at the low Ca 2+concentration a reduced multiple attack was exhibited by the mesophilic a-amylase from porcine pancreas and isozyme 2 from germinated barley seeds. In contrast, barley a-amylase 1 maintained the level of multiple attack found at the high Ca2§ 1. INTRODUCTION The degree of multiple attack exhibited by a-amylases is defined as the average number of catalytic events following cleavage of the first glycosidic bond during the life-time of an individual enzyme-polysaccharide complex [1]. The degree of multiple attack for a given enzyme thus contributes to the understanding of the nature of the productive enzyme-substrate interaction. Using a substrate with a high degree of polymerization, the degree of multiple attack for a number of a-amylases was previously determined to vary from 1.9 for the oramylase from Aspergillus oryzae to 6 for porcine pancreatic a-amylase [ 1]. The aim of the present study was to determine the role of Ca 2§ and of temperature in the degree of multiple attack exhibited by a number of thermophilic and mesophilic a-amylases having extended substrate binding sites. The chosen a-amylases from bacteria and higher plants all have an active site containing 9-10 consecutive binding subsites each accommodating one substrate glucosyl residue [2-5]. In contrast, pancreatic a-amylase, which is known to exert a high degree of multiple attack, has an active site comprising only 5 subsites [6]. Ca2. is required for activity of all a-amylases [7-9]. The different three-dimensional structures have a common Ca2+-site that stabilizes a small domain containing side chains essential for the substrate binding by bridging this domain and the catalytic (13/or)s-barrel domain. This characteristic domain interplay occurs at the subsites in the direction of the nonreducing end from the bond to be cleaved [4,10,11 ]. Various additional Ca2"-sites exist in B.

344

fichemformis and barley a-amylases [4,10].

In the multiple attack mechanism the nonreducing end product of the hydrolysis is envisaged to reorganize in the substrate binding cleft for subsequent catalytic events.

2. METHODS The enzymatic digests were performed with amylose (Sigma) as substrate at a concentration of 0.1% in 2 % DMSO in the appropriate buffer. The degree of polymerization (DP) of the amylose was 440 determined as described by Robyt and French [1]. In the case of the thermophillic Bacillus and the barley cx-amylases, the buffer was 20 mM Na-acetate, pH 5.5 with either 0.1 mM or 5 mM CaCI2. For porcine pancreatic a-amylase 20 mM MES, pH 6.9 was used with CaC12 added as described above. Enzymatic digests were performed in a small scale. Samples (1 ml) from a 10 ml reaction mixture of substrate with enzyme were removed at intervals during the initial phase of substrate degradation. Hydrolysis of substrate was monitored after iodine staining as the decrease in "blue value" measured at 620 nm [ 1,12]. In separate experiments it was found that the initial stage of substrate hydrolysis corresponds to a decrease in blue value of about 20%. This was usually obtained within 20-45 min after addition of the enzymes (data not shown). The Bacillus enzymes were kind girls from Novo Nordisk, Denmark. The barley a-amylase isozymes were prepared in-house as described earlier [5]. Porcine pancreatic o~-amylase was from Sigma. "l~he amounts of enzyme used in the digests were as follows: Bacillus licheniformis a-amylase: 10-30 mU/ml; a-amylase from alkalophilic Bacillus: 30-100 mU/ml; barley a-amylase,isozyme 1 (AMY1): 25-70 mU/ml and isozyme 2 (AMY2): 10-25 mU/ml. One unit is defined as the amount of enzyme that causes an increase in absorbance of 1.0 at 620 nm within 15min at 37 ~ using insoluble Blue Starch (Pharmacia) as a substrate [9,13]. Porcine pancreatic a-amylase was used at a concentration of 5-10 mU/ml, where 1 unit, according to Sigmae is defined as the amount of enzyme liberating 1 mg of maltose from starch at pH 6.9 and 20 ~ The enzyme action was arrested by addition of trichloroacetic acid to a final concentration of 0.1 mM. The. remaining polymeric amylose was subsequently precipitated with 67% ethanol [ 1], and lefi: overnight at 4 ~ The contents of reducing sugar in the total digest (RVt) and of the ethanol precipitate (RVv) were determined by the copper-bicinchoninate method adapted to a microtiter plate assay using glucose as a standard [14]. The degree of multiple attack is defined as (RVt/RVv)-I as described previously [1], and was calculated using analytical data obtained from several samples from each digest with blue values in the range of 80-90% of the initial value. 3. RESULTS AND DISCUSSION

3.1. Thermophilic a-amylases The degrees of multiple attack were determined for two bacterial thermophilic a-amylases at two different temperatures and at two different Ca2+-concentrations (Table 1). Significant degrees of multiple attack, in the range 1-3, are seen for both Bacillus a-amylases. This means that during each encounter of enzyme with the substrate, in average 1-3 glycosidic bonds are hydrolyzed in addition to the first cleaved bond. The a-amylase from B. licheniformis has a slightly reduced degree of multiple attack in 5 mM CaCI2 compared to the value obtained at

345 the low CaCl2-concentration. The specific activity of the a-amylase from B. licheniformis, measured with insoluble Blue Starch as substrate, increases somewhat but not dramatically as a function of Ca 2§ in the present concentration range [3]. The combined observations are difficult to interpret in relation to the three-dimensional structure of the enzyme, where three Ca2§ and one Na+-ion by crystallography are found to stabilize the active conformation [4]. Finally, it is shown that an increase in the temperature of the enzymatic digests from 37 ~ to 80 ~ does not affect the degree of multiple attack, suggesting that the stability of the enzyme-substrate complex is not temperature-dependent (Table 1). Table 1 De~rees of multiple attack exhibited by thermophilic a-amylases 0.1 mM CaCI2

5 mM CaCI2

Enzyme source

37 ~

80 ~

37 ~

80 ~

B. licheniformis

2.2

2.1

1.2

1.5

Alkalophilic Bacillus

1.9

2.6

2.6

2.5

The degree of multiple attack exhibited by the alkalophilic Bacillus ct-amylase tends to be reduced at the low Ca 2+ -concentration, but only at 37 ~ This suggests that at this Ca 2§ concentration the stability of the enzyme-substrate complex is temperature-dependent as well as Ca2+-dependent. At 5 mM CaCI2 the temperature of the enzymatic digest has no effect on the degree of multiple attack, perhaps reflecting a more efficient stabilization of the structural integrity at the higher concentration of CaC12.

3.2. Mesophilic or-amylases The degrees of multiple attack were determined for mesophilic enzymes from a higher plant (barley) and from an animal source (porcine pancreas), respectively (Table 2). All values are obtained from digests performed at 37 ~ Table 2 Degrees of multiple attack in mesophilic a-amylases Enzyme

0.1 mM CaCI2

5 mM CaCI2

Barley AMY1

1.4

1.1

Barley AMY2

0.5

1.2

Porcine pancreatic or-amylase

4.7

5.9

The degree of multiple attack exerted by both barley a-amylase isozymes is lower than for the bacterial and mammalian enzymes. In the case of the AMY1 isozyme, reduction of the

346 Ca2+-concentration of the digest does not significantly change the degree of multiple attack. In contrast, the degree of multiple attack of the AMY2 isozyme is strongly CaZ§ and has the lowest value (0.5) in the present study in 0.1 mM CaCI2 (Table 2). This may be a consequence of the lower affinity for Ca2+ of AMY2 compared to AMYI [8]. The effect of Ca2§ on the degree of multiple attack observed with the barley or-amylase isozymes parallels the Ca2§ of the activity of these enzymes towards insoluble Blue Starch. The activity of AMY1 is almost Ca2+-independent in the used concentration range, whereas the specific activity of AMY2 is strongly reduced at low Ca2§ and has a maximum at 5-15 mM CaClz [ 13]. Both isozymes contain structural Ca2+[ 10], but how the present results can be related to the three-dimensional structures of the enzymes remains to be elucidated. Interestingly, the ligands for the three Ca2§ bound in barley a-amylase are conserved between AMY1 and AMY2 [ 10]. Since the degree of multiple attack represents an average estimate and is defined as the number of bonds hydrolyzed after the first bond cleavage in each enzyme-substrate encounter, the result obtained with AMY2 (Table 2) indicates that at a low Ca2§ AMY2 acts on amylose by multiple attack as well as by the multichain mechanism. The multichain mechanism is the classical random action, where the enzyme hydrolyzes only one bond per productive encounter with the substrate [ 1]. As previously reported porcine pancreatic a-amylase exhibits a high degree of multiple attack.[ 1]. This is confirmed in the present study, where a value close to 6 was found (Table 2). The degree of multiple attack determined for pancreatic a-amylase using the ethanol precipitation technique adapted to small scale digests agrees well with previous results obtained using substrates with DP values ranging from 17 to 1000 [ 1,15,16]. Thus, the degree of multiple attack found for porcine pancreatic a-amylase is several fold greater than that exhibited by the thermophilic bacterial as well as the mesophilic plant a-amylases in the present study. It is tempting to speculate that this property might be related to the fact that porcine pancreatic a-amylase possesses a shorter active site with only 5 subsites [ 11] compared to the 9-10 subsites long binding sites found in the other a-amylases used in this study [2-6]. Also in the case of the porcine pancreatic or-amylase some dependence on the Ca2+-concentration is observed (Table 2). This agrees with the finding that the enzyme molecule contains the structural Ca2§ found in all the other a-amylases [4,10,11 ]. The increasing insight into the role of Ca2+ and other ions on the functional conformation of the different a-amylases from B. licheniformis, barley and porcine pancreas [4,10,11,17] renders it highly relevant to assess the influence of Ca2+ on enzymatic properties including the mechanism of multiple attack for which the structural basis is presently poorly understood. ACKNOWLEDGMENT The kind gift of the Bacillus enzymes by Henrik Bisg~rd-Frantzen (Novo Nordisk, Denmark) is gratefully acknowledged. REFERENCES

1. J.F. Robyt and D. French, Arch. Biochem. Biophys. 122 (1967) 8. 2. H. Outtrup, H. Bisg~,rd-Frantzen, P. Rahbek Ostergaard, M. Dolberg Rasmussen and P. van der Zee, Patent no. PCT WO 95/26397. 3. D. Tull, H. BisgS,rd-Frantzen and B. Svensson, in preparation.

347 4. M. Machius, N. Declerck, R. Huber and G. Wiegand, Structure 6 (1998) 281. 5. E.H. Ajandouz, J. Abe, B. Svensson and G. Marchis-Mouren, Biochim. Biophys. Acta 1159 (1992) 1932. 6. E.H. Ajandouz and G. Marchis-Mouren, Carbohydr. Res. 268 (1995) 267. 7. B.L. Vallee, E.A. Stein, W.N. Summerwell and E.H. Fisher, J. Biol. Chem. 234 (1959) 2901. 8. D.S. Bush, L. Sticher, R. van Huystee, D. Wagner and R.L. Jones, J. Biol. Chem. 264 (1989) 19392. 9. I. Matsui and B. Svensson, J. Biol. Chem. 272 (1997) 22456. 10. A. Kadziola, J.-i. Abe, B. Svensson and R. Haser, J. Mol. Biol. 239 (1994) 104. 11. M. Qian, R. Haser, G. Buisson, E. Du~e and F. Payan, Biochemistry 33 (1994) 6284. 12. J.M. Bailey and W.J. Whelan, J. Biol. Chem. 236 (1961) 969. 13. K.W. Rodenburg, N. Juge, X.-J. Guo, M. Sogaard, J.-C. Chaix and B. Svensson, Eur. J. Biochem. 221 (1994) 277. 14. J.D. Fox and J.F.Robyt, Anal. Biochem. 195 (1991) 93. 15. A.K. Mazur and H. Nakatani, Arch. Biochem. Biophys. 306 (1993) 29. 16. H. Kondo, H. Nakatani, K. Hiromi, R. Matsuno, J. Biochem. 84 (1978) 403. 17. A. Kadziola, M. Sogaard, B. Svensson and R. Haser, J. Mol. Biol. (1998) in press.


Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 1998 Elsevier Science B.V.

349

Degradation and denaturation of stable enzymes Dion Thompson a, Roberto Fem/mdez-Lafuente b, Cesar Mateo c, Don A. Cowan b, Jose M. Guisan r and Roy Daniel a. Thermophile Research Unit, University of Waikato Private Bag 3105, Hamilton, New Zealand. a

b Department of Biochemistry University College London, United Kingdom. c Departamento de biocat/disis. Instituto de Catalisis. CSIC, Campus Universidad Autonoma 28049 Madrid, Spare.

1. INTRODUCTION We are interested in the mechanisms which govern the upper temperature limit for enzyme activity. The loss of enzymic activity at high temperatures can be separated into two broad classes. The first is the loss of tertiary and secondary structures. The second is the loss of primary structure through covalent modifications or cleavage. These two processes have been referred to as denaturation and degradation respectively I. The aim of the present study is to investigate the relative roles of denaturation and degradation in activity loss in an enzyme system which has reasonable stability at temperatures above 100~ with a view to understanding which of these two processes limits high temperature enzyme stability. The enzyme chosen for study was a xylanase from a stain of Thermotoga maritima 2. This enzyme has a half life of 5 to 10 minutes at 100~ in the absence of stabilisers. Stabilisation was provided by sorbitol and multi-point covalent immobilisation giving rise to a half life of 2 minutes at 140~ when immobilised and suspended in molten sorbitol.

2. IRREVERSIBLE ACTIVITY LOSS The loss of recoverable activity of an enzyme due to heat denaturation typically follows first order kinetics, eqn(1), with respect to concentration and time, although close examination may revel that second order kinetics provides a marginally better approximation 3.

350 A graph of the logarithm of residual activity against time has a slope of-k. Occasionally, a semi-log graph revels a biphasic behaviour whereupon a sharp transition from a high to a low rate constant is made at some stage through time 2. However, no such behaviour was observed in this work.

d[A] dt where:

-

(~)

-k[A]

[A] = concentration of active enzyme k = the rate constant. t = time

The first order rate constant of xylanase activity loss was determined for a range of temperatures and with different stabilising agents. Figure 1 presents these rate constants as a function of temperature in the form of an Arrhenius plot. The slope of an Arrhenius plot represents the activation energy of the reaction, which in this case is the transition from active to inactive enzyme. High activation energies are characteristic of unfolding of macromolecules while lower values are generally found for covalent modifications (eg ref 4).

In(k) Halflife 4ilsec

tee x34amse noso,t~ I ~

~ 50%sort~ A-----~.-__..,

freexylanase

2

esec

O-

9 41sec \ ~..~.

irm'~lised li ..... n o ~ "'" .... x ~ "-.~.

e,o-2 N 5 mn 9

-. .....

9

,,

tee~l~

-4 !38 rrin I I

-6 5hm -8 341"~ 24

1135

127 I

119 i

111 I

104 I

I

245

25

255

26

265

27

901

oC

275 10001~

Temm~e

Figure 1: Arrhenius plot of the first order rate constants for irreversible loss of activity.

351 Three aspects of the denaturation profiles stand out from Figure 1. First, loss of activity of the free xylanase in low concentrations of sorbitol, 60 X ,,i.,a o 50 & 40 Ill X ,It& = 30 & 9 -~ 20 w • 9 9 10 iv, Xxx a, I 0 0 80 20 40 60 T i m e (rain)

9

I

100

120

Figure 4: pH influence on lysozyme inactivation kinetics 9 :pH = 4 ;m:pH = 6,2 ;A:pH = 7 ; :....:pH ---8. Experimental conditions" Phosphate buffer 0.01M; Lysozyme concentration" 0.05 g.L-~; Temperature- 70~ Nitrogen flow rate 950 mL.min-1

3.3. Effect of physico-chemicai parameters on enzyme inactivation kinetics. Several authors have demonstrated that physico-chemical parameters such as pH and ionic strentgh act on protein adsorption processes (5,20). Enhancing protein adsorption process might result in an increase of inactivation kinetics at gas-liquid interface. Figure 4 gives the pH effect on lysozyme inactivation at a 50 mL. min ~ nitrogen flow rate. Gas bubbling efficiency strongly depends on pH. A pH close to the enzyme isoelectric point enhances the

397 inactivation. No activity loss is observed for experiments carded out without bubbling in the same conditions of pH, temperature and duration. The same kind of effect has been observed for the other enzymes (data not shown). Table 1 shows the influence of NaC1 concentration on PME half life with and without bubbling. Without bubbling, NaC1 exerts a protective effect on PME. The PME half life increases with NaC1 concentration. With bubbling, PME half life decreases very strongly. At 2M NaCI, the PME half life decreases from 91 min to 9.5 min when bubbling at 100 mL. min-~" Table 1 Effect of NaC1 concentration on PME inactivation NaC1 (M) 0

1

2

3

Half life (min), Q = 0 mL.min~

62

101

91

95

Half life (min), Q= 100 mL.min ~

48

53

9,5

5,6

Experimental conditions: PME concentration 0,008 g.L~; Citrate buffer 0,01M; pH 4,5; Nitrogen flow rate : 0 and 100 mL.min~; Temperature : 54.5~

4. DISCUSSION Inactivation is strongly enhanced by gas bubbling for the three studied enzymes, lysozyme, lipase and pectinmethylesterase. Nitrogen bubbling in an enzyme solution generates irreversibly inactivated enzyme molecules. For the studied enzyme concentrations, the inactivation kinetics of lysozyme and PME in the presence of nitrogen bubbles have been found of first order. For lipase, the change of the reaction order throughout the inactivation process can be explained by the formation of inactivated lipase species. Inactive lipase species induce a competition effect at gas-liquid interfaces between active and inactive species which consequently slows down lipase inactivation kinetics. Efficiency of gas bubbling on enzyme inactivation strongly depends on pH and ionic strentgh. The same tendency was observed for the three enzymes : the closer the pH is to the enzyme isoelectric point (pI) the more efficient the inactivation process is. (pI of lysozyme, lipase and PME are respectively 11, 4.3 and 4.4). The pH and ionic strength act on the enzyme adsorption process and enhance the inactivation kinetics. Our results seem to demonstrate that inactivation of lysozyme, lipase and PME promoted by gas bubbling is mainly governed by an interfacial process.

398 REFERENCES

1. 2. 3. 4. 5. 6. 7.

Lindet B., Caussette M., Delepine S. and Planche H. (1995). Patent FR 9515239. Norde W., Adv. Coll. Int. Sci., 25 (1986) 267-340. Schmidt C.F., Zimmermann R.M., Gaub H.E., Biophys. J., 54 (1990) 577-588. Ball A., Jones R.A.L., Langrnuir, 11 (1995) 3542-3548. Buijs J., Norde W., Langmuir, 12 (1996) 1605-1613. Cassells J.M., HaUingG P.J., Enzyme Microb. Technol., 12 (1990) 755-759. Blanco R.M., Hailing P.J., Bastida A., CuestaA C., Guisan J.M., Biotechnol. Bioeng., 39 (1992) 75-84. 8. Ghatorae A.S., Guerra M.J., Bell G. Hailing P.J., Biotechnol. Bioeng., 44 (1994) 13551361. 9. Ghatorae A.S., Bell G. Hailing P.J., Bioteehnol. Bioeng., 43 (1994) 331-336. 10. Adams D.J., Evans M.T.A., Mitchell J.R., Phillips M.C., Rees P.M., J. Polymer Sci., Part C, 34, (1971) 167-179. 11. Graham D.E., Phillips M.C., J. Coll. Int. Sci., 70, 3 (1979) 403-439. 12. Graham D.E., Phillips M.C., J. Coll. Int. Sci., 76, 1 (1980) 240-249. 13. Ter-Minassian-Saraga L., J. Coll. Int. Sci., 80, 2 (1981) 393-401. 14. Hunter J.R., Carbonell R.G. Kilpatrick P.K., J. Coll. Int. Sci., 143, 1 (1991) 37-53. 15. Mac Ritchie F., Anal. Chim. Acta, 249 (1991) 241-245. 16. Tronin A., Dubrovsky T., Dubroskaya S., Radicchi G. and Nicolini C., Langrnuir 12 (1996) 3272-3275. 17. Inbar I. and Miller, I., Biochim. Biophys. Aeta, 452 (1976) 417-430. 18. Xu S. and Damodaran S., J. Coll. Int.Sci. 159 (1993) 124-133. 19. Rouse, A. H. and Atkins, C., Food Teehnol 6: (1952) 291-294. 20. Wannerberger K., Amerbrant T., J. Coll. Int. Sci., 117 (1996) 316-324.

Stability and Stabilization of Biocatalysts A. BaUesteros, F.J. Plou, J.L. Iborra and P.J. Halling (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

Effects of water-miscible solvents on the stability and specificity of cyclodextringlucosyltransferases Anne D. Blackwood and Christopher Bucke School of Biosciences, University of Westminster, 115 New Cavendish Street, London W lM 8JS, UK 1. INTRODUCTION. Cyclodextringlucosyltransferase (EC 3.2.1.19 [CGTase]) catalyses several reactions with starch as substrate, including the synthesis of three cyclodextrins(CD) (t~,13 and ?) with 6,7 and 8 glucose units joined by a-1,4 linkages. It also catalyses three transferase reactions, the hydrolysis of starch to linear dextrins, the disproportionation of two linear dextrins into linear oligosaccharides of different sizes and the coupling of CDs with linear dextrins. Several species and strains of bacteria produce CGTases all of which appear to be capable of catalysing all the reactions but which differ in their specificities. The CDs are the commercially important products of CGTase action: the separation and purification of the individual cyclodextrins is an expensive part of the production procedure. Consequently there is a need to find methods of improving the selectivity of CGTases to produce improved yields of individual CDs. Cyclodextrin yield and specificity can be improved by the inclusion in reaction mixtures of complexants which form insoluble selective inclusion complexes with CD molecules (1,2). There is also mention in the literature of increases in CD production achieved by the inclusion of low M.W. polar solvents such as ethanol (3) and acetone (4). The mechanism by which this effect occurs is less clear. To extend knowledge of this phenomenon, the effects of a range of organic co-solvents on the activities of CGTases from a Thermoanaerobacter sp. (5) and Bacillus circulans strain 251 (6) have been investigated. Both strains produce predominantly ~-CD in normal process conditions but the enzymes have been found to differ in their responses to the organic co-solvents used. 2. MATERIALS AND METHODS. 2.1. Enzymes and Chemicals. T-CGTase (from Thermoanaerobacter sp.) with a specific activity of 120 NU/g (based on its a-amylase activity measured by the Phadebas method) was a gift of Novo-Nordisk. B-CGTase (from Bacillus circulans strain 251) with a specific activity (for 13-CD production) of 270 U/mg was provided by Professor Lubber Dijkhuizen,

399

400 University of Groningen. All anhydrous organic solvents were stored over 4-A molecular sieves. Phadebas tablets (cross-linked insoluble blue-dyed starch) were from Pharmacia. Paselli SA2, a 2 dextrose equivalent maltodextrin preparation, was from Avebe, the Netherlands. 2.2. Enzyme Stability. T-CGTase was lyophilized from 0.1M Na phosphate buffer pH 6.0. Freeze-dried enzyme (lmg/ml) was suspended in 10 ml of an aqueous-organic solvent solution (aqueous phase was 0.1M Na-phosphate buffer pH 6.0) and incubated at for 24 h at 25~ Samples were then diluted 1000-fold and assayed for hydrolytic activity for 15 min at pH 6.2 and 60~ by measuring the absorbance at 620 nm of the blue dye released from Phadebas tablets. Enzyme activity was compared with a standard czamylase from Bacillus licheniformis. 2.3. Enzyme assays. Synthesis of I3-CD was followed using Paselli SA2 (5% w/v) as substrate by the formation of a colourless inclusion complex with phenolphthalein. One unit of activity produces 1~tmol of 0-CD per min. Solvent effects on CGTase activity were determined by incubating 0.2 U/ml of enzyme with 5%(w/v) Paselli SA2 in 10raM Na-citrate buffer, pH 6.0, at 50~ for 24 h. Samples were taken at regular time intervals, added to 100 B1 of 1M HC1 and boiled for 5 rain. Products were analysed by HPLC using a Dionex PAl anion-exchange column eluted with 150mM NaOH with Na acetate gradient and a flow rate of lml/min. Detection was by pulsed amperometric detector (PAD). The coupling activity of T-CGTase was measured in the presence of various additives (2.5 g/1 maltose and/or 10% v/v solvent) using either cx or ~3-CD as donor. T-CGTase (0.2U/ml) was added to the reaction in 10raM Na-citrate buffer, pH 6.0 and the residual CD after 4 h incubation at 50~ determined by HPLC. 3. RESULTS. The stability of the Thermoanaerobacter sp CGTase on exposttre to 12 polar organic solvents at a range of aqueous-organic solvent concentrations is summarised in Table 1. At low solvent concentrations the enzyme retained over 70% of its hydrolysis activity relative to enzyme incubated in the absence of organic solvent. At higher solvent concentrations T-CGTase was inactivated only in protein-dissolving solvents such as DMSO and in DMF, diethylacetamide (DMA) and pyridine. Table 1 provides also the partitioning coefficients, log P, for each solvent as measured in a standard octanol-water two-phase system Log P has been suggested as a reliable predictor ofbiocatalyst function in organic solvents (7): solvents with high log P values generally give the best retention of enzyme activity, while hydrophilic solvents with log P less than 2 are thought to remove the hydration "shell" of the enzyme, thus eliminating biocatalytic activity. DMSO and DMF both significantly denature TCGTase and have low log P values but other polar solvents of similar hydrophilicity do not denature the enzyme under these conditions. The enzyme maintained max~um activity, with even a slight increase in activity in all solvents at 20% (v/v) solvent concentration. This encouraged the investigation of solvent effects on enzyme specificity.

401 The effects of 8 polar solvents at 10% (v/v) concentration on the total conversion of Paselli SA2 to CDs by both T-CGTase and B-CGTase was examined. The results are summarised in Table 2. In all of the solvents tested there was an increase in the overall conversion of maltodextrin to cyclodextrin products compared with a solvent-free control. Maximum conversion, 66%, was achieved with the addition of 10% (v/v) ethanol. Addition of DMSO, DMA and t-butanol as co-solvents produced an TABLE 1. Log P values of solvents and relative residual hydrolysis activity ofTCGTase after 24 h incubation in solvent-aqueous buffer mixtures at 24~ % residual activity as a function of organic solvent concentration (v/v) Solvent

logP

20

40

50

70

100

Triethylene glycol dimethyl ether Dimethylsulfoxide Dioxane N,N-Dimethylformamide N,N-Dimethylacetamide Acetonitrile Ethanol Acetone Ethylene glycol diethyl ether Tetrahydrofuran Pyridine t-Butanol

- 1.86

107

99

99

93

77

- 1.3 - 1.1 - 1.0 -0.81 -0.33 -0.24 -0.23 0.28

108

87 80 86 70 87 100 97 98

81

75

104 83 98 92 105 118

103 87 93 83 91 90 95 100

3 22 82 89 94 108

7 80 72 74 61 64

0.49 0.71 0.78

102 82 100

85 78 91

77 71 96

73 27 90

66 57 70

104

0

13

increase in [3-CD formation with corresponding decreases in the ratios of both a and yCDs. DMF and acetone also produced increased synthesis of ~-CD with a decrease in ~/-CD production. Addition of acetonitrile, ethanol or tetrahydrofuran resulted in an increase in r formation at the expense of both [3 and y-CDs. The product selectivity of B-CGTase, determined under similar conditions is shown in Table 3. As with T-CGTase, each solvent produced an increase in the overall conversion of Paselli SA2 to CDs but the increases were less pronounced. Maximum conversion of maltodextrins to CDs (51%) was achieved with DMF as co-solvent. In contrast to TCGTase, each case addition of co-solvent resulted in an increase in the formation of [3CD at the expense of both ~ and ~/-CD. In other experiments (data not shown) it was demonstrated that with T-CGTase the inclusion ofjust 2% (v/v) solvent in the reaction mixture caused an increase in either r (acetonitrile) or [~-CD (DMA). Production of the favoured CD increased further with increases in co-solvent concentration up to 20% (v/v). The reason for this increase in CD production in the presence of small concentrations of polar organic solvents is unclear.

402 TABLE 2. Conversion of maltodextrins to cyclodextrins and product specificity ofTCGTase after 24 h incubation in 10% v/v organic solvent-buffer mixtures. Product ratio (%) Solvent

Substrate conversion to CDs (%)

r

13

7

None Dimethylsulfoxide Dimethylformamide Dimethylacetamide Acetonitrile Ethanol Acetone Tetrahvdrofuran

40 50 47 57 54 66 55 57

30 24 32 16 58 51 30 48

50 66 57 68 32 39 64 43

20 10 10 16 10 10 6 9

TABLE 3. Conversion ofmaltodextrins to cyclodextrins and product selectivity of BCGTase after 24 h incubation in 10% (v/v) organic solvent-aqueous buffer mixtures. Product ratio (%) Substrate conversion Solvent

to CDs (%)

~

13

None Dimethylsulfoxide Dimethylformamide Acetonitrile Tetrahydrofuran t-Butanol

40 45 51 45 42 44

15 11 9 7 10 9

65 77 77 76 76 82

20 12 14 16 14

9

No precipitation of product was observed in any of the organic co-solvent-aqueous buffer mixtures. It is plausa'ble that the solvents affect reactions which consume the CDs after synthesis or otherwise limit yields. In the conditions of industrial production of CD, yield is limited by product inhibition (8). In B-CGTase, which has five structural domains (6), [3-CD causes inhibition by interfering with the function of a maltose binding site located in the raw starch-binding E-domain of the enzyme as well as interfering with catalysis at the active site (9). The CD bound to the maltose binding site could interfere with the guidance of linear maltodextrin chains from this binding site to the active site located in the A-domain of the enzyme (6,10). Solvent interacting with the CD could alter the conformation of the CD and hinder or prevent its binding with the maltose binding site and/or the active site. It was found that addition of 0.2 mM I3-CD to T-CGTase did indeed markedly decrease the initial rate of cyclisation activity. The inhibition was non-competitive and

403 followed Hill kinetics. The inclusion of 10% (v/v) DMA had no effect on this inhibition. The reverse (coupling) reaction also lowers the yields of CD obtained: small linear dextrins which form in the reaction mixture act as acceptors for glucose residues from the CD. Inhibition of this reaction by organic co-solvents might explain the increases observed in CD production. The hydrolysis and transglycosylation of both a- and 13CD with and without solvents and with and without maltose as acceptor were followed by HPLC (Table 4). TABLE 4. Percentage of CD remaining after 4 h incubation with T-CGTase in the presence and absence of various additives (2.5 g/1 maltose or 10% v/v organic solvent) %CD remaining Additives CGTase CGTase CGTase CGTase CGTase CGTase

+ + + + +

acetonitrile dimethylacetamide maltose maltose + acetonitrile maltose + dimethylacetamide

a-CD

~-CD

76 92 69 39 74 35

92 82 95 65 65 82

In the absence of solvents or maltose, ct-CD was hydrolysed to a greater extent than 13CD and hydrolysis of both CDs was increased in the presence of maltose. The extent of hydrolysis was affected by the addition of and by the nature of co-solvent. Acetonitrile (10% v/v) reduced the extent of hydrolysis of a-CD but increased the hydrolysis of ~-CD. In contrast, addition of DMA increased hydrolysis of a-CD but decreased 13-CD hydrolysis. In the presence of maltose, acetonitrile protected et-CD against hydrolysis but had no effect on hydrolysis of 13-CD, but DMA had the reverse effect, protecting 13-CD but not ct-CD. 4. DISCUSSION. It seems likely that the effects of organic solvents on CGTase could be a combination of various factors which depend on the bacterial origin of the enzyme and the nature of the co-solvent. It is clear that the solvent effects are more subtle than can be explained simply by the log P of the solvent and it is possible that water molecules may be displaced by the solvents at specific sites either on the surface of the enzyme molecule or in the active site region. The effectiveness of low concentrations of solvents encourages speculation that solvents may interact more effectively at locations in the active site. The determination of the molecular basis of observed differences in enzyme selectivity is one of the most challenging tasks for molecular biologists. The CGTases provide a group of enzymes eminently suitable as model systems for this task. Organic solvents seem likely to be useful as additional tools in this effort.

404 References.

1. Armbruster, F.C. In: Proceedings of the 4 th International Symposium on Cyclodextrins. (Huber, O. & Szetli, J. Eds). Kluwer Academic Publishers, Dordreeht, 1988, 33-39. 2. Schmid, G., Huber, O.S. & Eberle, H. J.. lbid, 71-77. 3. Tomita, K., Tanaka, T., Fujita, Y. & Nakanishi, K. J. Ferment. Bioeng. 70 (1990) 190-192. 4. Lee, Y.D. & Kim~ H.S. Enzyme Microb. Tectmol. 13 (1991) 499-503. 5. Norman, B.E. & Jorgensen, S.T. Denpun Kagaku 39 (1992) 101-108. 6. Lawson, C.L., van Montfort, 1~, Stropotykov, B., Rozeboom, H.J., Kalk, I~H., de Vries, G.E., Penninga, D., Dijldauizen, L & Dijkstra, B.W.J. Mol. Biol. 236 (1994) 590-600. 7. Laane, C., Boeren, S., Vos, K. & Veeger, C. Bioteclmol. Bioeng. 30 (1987) 8187. 8. Bergsma, J., Bruinenberg, P.M., Hokse, H. & Meiberg, J.M.B. In: Proceedings of the 4th International Symposium on Cyclodextrins. (Huber, O. & Szejtli, J. Eds). Kluwer Academic Publishers, Dordrecht, 1988, 41-46. 9. Penninga, D., van der Veen, B.A., Knegtel, IkM.A., van Hijum, S.A.F.T., Kozeboom, H.J., Kalk, I~H., Dijkstra, B.W. & Dijkhuizen, L. J. Biol. Chem. 271 (1996) 32777-32784. 10.Knegtel, I~M.A., Strokoptykov, B., Penninga, D., Faber, O.G., Rozeboom, H.J. Kalk, K.H., Dijldmizen, L. & Dijkstra, B.W.J. Biol. Chem 270 (1995) 2925629264.


405

Stabilization of immobilized enzymes against organic solvents: complete hydrophylization of enzymes environments by solidphase chemistry with poly-functional macromolecules. R. Fernandez-Lafuente a, C.M. Rosell a, L. Caanan-Haden b, L. Rodes b, J.M. Guisan a* a Departamento de Biocatalisis. Instituto de Catalisis. CSIC. 28049 Madrid. Spain. b Centro de Ingenieria Genetica y Biotecnologia. La Habana. Cuba.

1. INTRODUCTION The performance of enzyme biotransformations in non-conventional media is becoming increasingly popular. The use of cosolvents, non-miscible organic solvent and supercritical fluids (1-2) may be an excellent tool to shift thermodynamic equilibrium towards synthesis, to increase the solubility of reagents, to "modulate" enzyme properties, etc. However, the interaction of enzyme molecules with high concentrations of cosolvents or with hydrophobic solvent interfaces may promote their rapid inactivation. This is hindering the industrial applicability of such exciting biotransformations, when working in organic media, hydrophobic solvents may directly interact with the enzyme surfaces, promoting deleterious conformational changes. From this point of view, we may assume that a substantial increase in the "hydrophilicity" of the enzyme surface could prevent many deleterious effects of hydrophobic solvents (direct interaction and conformational changes) (3). However, the achievement of a substantial increase in the hydrophilicity of an enzyme surface via site-directed mutagenesis is not a very simple task (4). This tool for protein engineering requires the design of a large number of site-mutations on enzymes whose 3D structure need to be well known. However, the structure of many industrial enzymes has not been elucidated. In any case, the design of a large number of mutations could promote important alterations in the folding of the protein. In this point, it is worthy to note that the chemical modification of previously folded enzymes should only involve surface groups. Thus, protein engineering via chemical tools could have some advantages and it could be rationally performed on enzymes with unknown structure. However, a large number of chemical modifications on the enzyme surface could also promote dramatic reductions of the enzyme activity. "Hydrophilization" of enzymes via chemical modification of immobilized derivatives with polyfunctional macromolecules. This approach could have a number of advantages:

406

i.- Solid-phase chemical modification of proteins may be much easier than modification of soluble enzymes (5-6). A number of key features strongly facilitate the chemical modification of immobilized enzymes: intermolecular processes are now impossible, the process is very easily controlled, modification can be carried out in a wide range of experimental conditions, etc. ii.- As any other chemical modification, modification with poly-functional macromolecules is always directed towards the enzyme surface and it permit to incorporate any type of chemical moiety "on or very close to" the enzyme surface. iii.- The controlled modification with poly-functional macromolecules may promote dramatic changes in the environment of every enzyme molecule but with a minimal chemical modification of the enzyme surface. iv.- In addition to direct modification of the enzyme, the possible co-immobilization of enzymes and macromolecules on large support surfaces could permit strong modifications on the nano-environment of the enzyme molecules without any chemical modification on the enzyme structure. v.- This approach can be developed on immobilized-stabilized derivatives (e.g. multipoint covalently immobilized derivatives). In this way, the stabilization by "hydrophilization" could be additive to previous stabilization by "rigidification" associated to immobilization. Penicillin G acylase from Escherichia coil (PGA) is a very interesting enzyme with current industrial application for the selective hydrolysis of penicillin G to produce 6amino penicillanic acid (a key intermediate for the synthesis of r~-Iactam antibiotics) (7). Moreover, other interesting prospects of PGA in fine chemistry have been already reported (7, 8) synthesis of antibiotics, resolution of racemic mixtures of chiral intermediates, protection/deprotection reactions. PGA is very sensitive to the deleterious effects of organic solvents and this feature strongly limits its synthetic applicability. The chemical modification with polyfunctional macromolecules as a tool to stabilize a multipoint covalently immobilized PGA is here reported. Two different approaches are discussed: a.- the co-immobilization of the enzyme plus poly-aminated macromolecules and b.- the direct slight chemical modification of the immobilized derivatives with poly-aldehydes. Finally, the combined effect of both stabilizing approaches was also checked.

2. M E T H O D S . Enzyme activity was checked by hydrolyzing 10mM penicillin G in 100mM sodium phosphate buffer pH 8 at 37~ Penicillin G acylase was immobilized in 10% glyoxyl agarose kindly donated by Hispanagar and prepared as previously described (6). Immobilization was performed in the presence of phenylacetic acid as previously described (5,6), offering 101U per ml of support (that means less than 3% of maximum support capacity for protein immobilization). Aldehyde-dextrans (molecular weigh of 6,000 D) were prepared as described elsewhere (9). The polyethylenimine used has a molecular weight of 25,000 D. Modifications with aldehyde-dextrans were performed at 4~ and finished after 24h by adding sodium borohydride (9). Inactivations were performed at different temperatures and pH values. To initialize the experiment, the derivative was washed with 10 volumes of the desired organic solvent/aqueous buffer mixture, then suspended in 9 volumes of this mixture.

407

3.- R E S U L T S .

3.1. Co-immobilization of PGA and polyethylenimine. The co-immobilization required a three-step protocol because of the polyamine/enzyme competition by the support. Polyethylenimine is far much reactive with the support than the protein and prevents the covalent attachment of the enzyme to the support if direct competition is permitted. The experimental protocol was the following: 1.- the enzyme is immobilized for 4.5 hours. 2.- an excess of polyamine is added 3.- the suspension was reduced with sodium borohydride after 30 minutes. Reflecting the absence of chemical modification of the enzyme, PGA activity was fully preserved during the co-immobilization protocol (Table 1). However, the stability in the presence of DMF of PGA co-immobilized with polyethylenimine is almost identical to the standard derivative (Table 1).

TABLE 1.- Effect of the different treatments in the activity and stability of the immobilized PGA in the presence of 70% DMF. Treatment

Residual Activity,

(%)a

Relative Stability b

None

100

1

Polyamine/PGA Co-immobilization

100

1.05

Aldehyde -dextra n

70-75

0.85

Combined approach

77-82

50

Inactivations were performed at 4~ a Residual activity refers to the initial activity offered in the immobilization. Relative stability takes as the unit the stability of standard non-modified PGA derivative under the described conditions.

3.2. Modification of immobilized PGA with polyaldehyde-dextran. The chemical modification of the immobilized PGA caused a small reduction in the enzyme activity (25-30%). However, Table I shows that this chemical modification did not promote any significant stabilization of PGA in the presence of 70% DMF.

3.3. Modification with aldehyde-dextran of co-immobilized polyamine-PGA. The modification of the co-immobilized polyethylenimine-PGA derivative with aldehyde dextran promoted a significant stabilization of the enzyme in the presence of

408 70% DMF (a 50-fold factor) with a moderate decrease in the enzyme activity (activity decreased by around 20%) (Table 1). TABLE 2.- Effect of the combined approach in the PGA stability under different conditions

Organic solvent

Concentration

pH

T, C ~

Stabilization" factor

Dioxane

70%

7

4

300

Dioxane

80%

7

4

320

Tetraglyme

90%

7

4

>1,000

Tetraglyme

90%

8

4

>1,000

8

60

6

7

63

5

a Stabilization factor is the increment of the half-life of the modified derivative compared to the standard derivative under the described conditions. Table 2 shows that the stabilizing effects of this modification may be extrapolated to other organic solvents. In fact, these stabilizing effects increased when the apolarity and size of the organic solvent increased (e.g., maximum stabilization is detected against tetraglyme, a 1,000-fold factor). This modified derivative remained almost fully active after 3 months of incubation in 90% of tetraglyme at pH 8 and 4~ while the unmodified derivative presented a half life under two days. On the other hand, PGA thermostability is only marginally improved by this treatment (Table 2). These results seemed to reinforce the idea that the main effects of the chemical treatment performed has not been the rigidification of the enzyme structure (5,6), but the reduction in the concentration of the organic co-solvent molecules that are able to interact with the enzyme. 4. DISCUSSION

Co-immobilization of PGA with poly-amines or the chemical modification of PGA derivatives with dextrans, when separately performed, did not exert significant stabilizing effects against the effects of organic solvents. It seems that PGA has many surface areas that are highly sensitive to organic solvents. However, a substantial increase in the stability of PGA against organic solvents was achieved when both stabilizing approaches were sequentially performed. Very likely, the enzyme became completely surrounded by three hydrophilic structures: the support, the co-immobilized polyethylenimine and the chemically anchored dextrans. This combined approach for chemical modification of immobilized enzyme derivatives seemed to yield a complete hydrophilic nano-

409

environment around the external surface of the enzyme (Scheme). As a consequence of such a mild chemical modification of the enzyme, PGA preserves a high percentage of catalytic activity in spite of being much more stable against organic solvents. At first glance, this strategy for stabilization of enzymes is based in general features of enzyme structure and enzyme mechanisms of inactivation and it seems to have general applicability to the most of industrial enzymes. Perhaps some enzymes might undergo important stabilization via a single stabilizing approach. However, the performance of the combined approach is very facile and hence we propose it as a highly suitable approach for stabilization of enzymes to work in organic media.

This stabilization of PGA derivatives with regard to the deleterious effect of organic solvents does not seem due to the impossibility of the solvent to interact with the enzyme. The very open structure of the polymers surrounding the enzyme surface and the high recovery of catalytic activity suggest that only an open hydrophilic physical barrier has been created all around the enzyme molecule. Dramatic stabilization could be explained via a more simple partition effect generated by the polymeric structures. The presence of these open and highly hydrophilic polymers fully surrounding the enzyme surfaces could promote a significant reduction of the concentration of solvents in the immediate surrounding of the enzyme (e.g. by a 25 - 50 % of the concentration in the bulk solution). This reduction of apparent solvent concentration could be enough to promote dramatic stabilizing effects. The stability of enzyme derivatives may be

410

decreased by orders of magnitude when simply doubling the concentration of solvent (e.g. from 40 to 80%) (8). The combined stabilizing approach for enzyme stabilization has been performed on multipoint covalently immobilized derivatives. In this way, stabilizing factors (hydrophilization) were additive to stabilizing effects promoted by that immobilization protocol (rigidification). Thus, PGA derivatives were 107 fold more stable (against the presence of 90% of organic cosolvent (tetraglyme)) than one-point covalently immobilized derivatives could be prepared. As far as we know, any stabilization strategy (genetic, chemical and so on) have been previously reported to yield so extremely high stabilizing factors. ( 11 - 16). REFERENCES

1.-. Kasche, V. (1986) Enzyme Microb. Technol., 8, 2-16. 2.- Khmelnitsky, Y.,L., Beiova, A.B., Levashov, A.V. (1991) FEBS Lett, 284, 267-269 3.- Fangxiao, Y., Russell, A.J. (1996) Biotechnoi. Bioeng., 49, 709-716. 4.- Alvaro, G., Russell, A.J. (1991) Methods in enzymology, 202, 620-643. 5..-Femandez-Lafuente, R., Rosell, C.M., Rodriguez, V., Guisan, J.M. (1995). Enzyme Microb. Technol. 17, 517-523. 6.- Fernandez-Lafuente, R., Rosell, C.M., Alvaro, G., Guisan, J.M. (1992) .Enzyme Microb. Technoi 14, 489-495. 7.-Sheawale, J., G., Deshpande, B.S, Sudhakaran, V.K., Ambedkar (1990) Process Biochem. Int., 6, 97-103. 8.-Femandez-Lafuente, R., Rosell, C.M., Guisan, J.M.. (1996) Biotechnol. Appl. Biochem.24, 139-143. 9.-Guisan, J.M., Rodriguez, V., Rosell, C.M., Soler, G., Bastida, A., FemandezLafuente, R., Garcia-Junceda, E. (1997) in Methods in Biotechnology 1. Immobilization of enzymes and Cells (Bickerstaff, G.F. ed) Humana Press. Totowa, New Jersey. 10.-Alvaro, G., Femandez-Lafuente, R., Blanco, R.M., Guisan, J.M. (1990) Appl. Biochem. Biotechnol., 26, 181-195. 11.-Fernandez-Lafuente, R., Rosell, C.M., Piatkowska, B., Guisan, J.M. (1996) Enzyme Microb. Technol., 17, 517-523 12.-Waldman, H., Sebastian, D. (1994) Chem. Rev., 94, 911-937 13.- Favino, T.F., Fronza, G., Fuganti, C., Fuganti, D., Grasselli, P., Mele, A. (1996) J. Org. Chem., 61,8975-8979. 14.-Cardillo, G., Tolomelli, A., Tomasini, C. (1996)J, Org. Chem., 61, 8651-8654. 15.- van Straten, N.C.R., Duynstee, H..I., de Vroom, E., van der Marel, G.A., van Boom, J.H. (1997) Liebigs Ann/Recueil, 1315-1220. 16.- Margolin, A.L. (1993) Tetrahedrum Lett., 34, 1239-1242.

Stability and Stabilization of Biocatalysts A. BaUesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

411

Effect of sorbitol on immobilized a-chymotrypsin thermostability in low-water system. T. de Diego, P. Lozano, M.J. Niguez and J.L. Iborra. Department of Biochemistry and Molecular Biology "B" and Immunology. Faculty of Chemistry. University of Murcia. P.O. Box. 4021, E-30100 Murcia, Spain. The influence of sorbitol (0 to 80 mmol/g of protein) on the thermal stability of immobilized a-chymotrypsin has been studied in a low water system (hexane:ethanol:water, 69:30:1 v/v) from 30 to 60 ~ The results obtained were analyzed by a two-step series-type deactivation model, involving an active intermediate enzyme state, with excellent agreement. In all cases, the immobilized derivative exhibited non-first order deactivation kinetic, where the main protective effect of sorbitol was observed in the first-step of the deactivation profile. For all assayed temperatures, sorbitol played a clear stabilizing role only at the lowest assayed concentration (20 mmol sorbitol/g enzyme), as it was observed by both thermodynamic and kinetic analysis of deactivation profiles. However, only for the lowest assayed temperature, the increase in sorbitol concentration enhanced proportionally the etm3ane stability, obtaining a maximum increase in half-life of 40-times. The overall influence of sorbitol was related with the water activity (Aw) in the microenvironment of the enzyme. 1. INTRODUCTION Understanding parameters that control the activity and stability of enzymes in nonconventional media is essential to optimize bioeatal3r processes with industrial interest [ 1, 2] Synthetic reactions with hydrolytic enzymes, containing only a small amount of water are especially attractive (i.e. enzymatic peptide synthesis) because the equilibrium is fully directed to the synthetic way [3-5]. However, the catalytic activity of enzymes suspended in organic solvents strongly depends on the amount of water available to maintain the active conformation [6-8]. Additionally, in low water systems, the enzyme deactivation processes can be due to the direct interactions of protein with solvent molecules from, the bulk organic liquid at the phase interface, or dissolved in an aqueous phase around the enzyme [3]. Different strategies have been developed to enhance the enzyme stability in low water systems. Immobilization and/or chemical modification of enzymes to preserve the hydrophilic microenvironment of catalyst have been shown as advantageous to improve both the catalytic action and the thermostability of enzymes [6,7]. The addition of water-activity depressing agents (i.e. polyols, salts) induce favourable partitioning of available water and its retention near the enzyme, involving the increase in stability in organic media [2,9]. The aim of this paper is to analyze the thermostability of ~-chymotrypsin immobilized in hexane:ethanol:water (69:30:1 v/v) and the influence of sorbitol, as stabilizing agent, by using a series-ta/jae kinetic model. * Acknowledgements. This work was partially supported by the "Comisi6n Interministerial de Ciencia y Tecnologia", CICYT,N~BIO96-1016-C02-01.Spain.

412 2. MATERIALS AND METHODS 2.l.Materials cx-Chymotrypsin (EC 3.4.21.1) type II from bovine pancreas was purchased from Sigma Chemical Co. N-Acetyl-L-tyrosine ethyl ester (ATEE, Sigma Chemical Co.) was used as standard substrate. Celite 545 (0,01-0,04 mm particle size) was obtained from Merck. All remaining reagents were of analytical grade and used without additional purification. 2.2. Methods The immobilization method was as follows: 15 nag of a-chymotrypsin were dissolved in 7 rnL of 0.1 M phosphate buffer pH 7.8 and mixed with 1 g of oolite. The mixture was shaken for 30 rain at room temperature and then lyophilized. The immobilized derivative showed an esterase activity toward N-acetyl-L-tyrosine ethyl ester of 12.8 U/mg support and a water content of 15.4 mg of H20/g support, measured by the Karl-Fischer method [8]. The esterase activity of cx-chymotrypsin was determined by the pH-stat method using a videotritrator VIT-90 equipped with autoburette and sample station (Radiometer, Copenhagen) [8, 9]. The protocol was as follows: a 3 mL sample of 50 mlVl ATEE in 30% w/v aqueous ethanol solution containing 20 mM CaCI2 was placed into a thermostated (40 ~ reaction vessel. The reaction was started by addition of 50 lxL of suspension of immobilized derivative in organic media previously homogenized. The pH was maintained constant at 7.0 by continuous addition of 50 ~ NaOH as tritrant. One unit of activity was defined as the amount of enzyme that hydrolyzed 1 ~tmol of ATEE/min under standard conditions of assay. The stability in low-water-system was determined as follows: into a screw-capped test tube containing 1GO nag of immobilized derivative, 30 IXL aqueous sorbitol solution ( 0 - 4 M) were added. Then, the system was completed by the addition of 2.97 mL of the organic solvent (hexane:ethanol, 69:30 % v/v), and the mixture was incubated at 30, 40, 50 or 60 ~ At regular intervals of time, homogeneous aliquots of 50 lxL were extracted from the incubation mixture and the residual activity was determined immediately as described above. The results obtained were fitted to model theoretical curves using a non-linear regression program of iterative convergence by the Marquardt-Levenberg algorithm method included in the Sigmaplot 2.0 (1994) software. Water activity was determined using a humidity and temperature digital indicator HUMIDAT-IC II (Novasina, Zurich, Switzerland), with a humidity sensor model BS-3(4)/PP (Novasina). The humidity sensor was checked with control saturated salts solutions (LiCI, Aw= 0.11, Mg(NO3)2,Aw = 0.54; BaCI2, Aw = 0.91) for the overall measuring range. 3. RESULTS AND DISCUSSION The influence of temperature on the immobilized cz-chymotrypsin stability in hexane:ethanol:water system, 69:30:1% v/v has been studied between 30 and 60 ~ with and without the presence of sorbitol during 100 hours. Figure 1 shows the deactivation profiles, depicted by experimental points of immobilized cx-chymotrypsin derivative in some assayed conditions. As it can be seen, the increase in temperature enhanced clearly the loss in activity of the immobilized derivative. However, the substitution of the aqueous fraction of the medium ( 1 % v/v) by 1 M sorbitol solution in water (20 mmol/g of protein) resulted in a clear improvement of enzyme stability. This results should be explained by the "immobilization" of sorbitol in the microenvironment of the enzyme and its ability to preserve the water shell

413 around the protein molecule, which aids to maintain the essential solvophobic interactions of the native structure [2, 9]. Analysis of these data on c~-chymotrypsin deactivation by a first-order model did not represent adequately the enzyme deactivation kinetics, being noticeable their biphasic lOOq character [8 -11 ]. In this order, a series-type model of enzyme deactivation was applied 80 [10]. This model involves two first order steps, with one active intermediate (El) and a final enzyme state (E2) with possible non-zero activity, as follows, ~ 4o ,~ ...e ...... kl k2 E ~ E1 > E2 (1) 20

Of,1

(/,2

E"

0

-

2

-

4

90

95

100

where kl and k2 are first-order deactivation Time ( h ) rate constants (t1"1); and oh and 0~2 are the ratios of specific activities E1/E and E2/E, respectively. Considering as initial conditions Figure 1. Deactivation profiles of Et=o = Eo, and E1 t=o = E2 t=o = 0, the specific immobilized ct-chymotrypsin in hexane: activity of each enzyme form can be ethanol:water, 69:30:1% v/v at different determined by the integration of the rate temperatures [30 (o), 40 (11), 50 ( 0 ) and equation of each step [11]. At it was 60 ~ (A)]. Dotted line shows the experimentally proved, the final state of the deactivation profile of enzyme with 20 immobilized derivative (E2) is fully mmol sorbitol / g protein in hexane: deactivated, and the ct2 parameter of the ethanol: water (69:30:1% v/v). model should be equal to 0 [8].The fractional remaining activity a can be given by a weighted average function of specific activities of all enzyme states yields, as follows: [E + oh E1 + or2 E2] a=

= 1 + [~lkl/(k2-kl)]" { [exp(-klt)] - [exp(-k2t)] }

(2)

E0 The fit of the experimental data of Figure 1 by this equation using a non-linear regression method of iterative convergence yields the theoretical curves depicted in the same Figure. The suitability of the model was indicated by the good agreement between experimental and theoretical data. By using this model, the kinetic of the deactivation process of the immobilized derivative was analyzed at different temperatures (30 - 60~ Additionally, the influence of sorbitol concentration into the aqueous fraction of the medium (hexane:ethanol:water (69:30:1) was also studied. In all cases, kl was significant larger than k2, and k2 was practically unchanged and very close to zero. Thus, an eventual stabilization of the average enzyme activity a, at a level lower than ctl was observed. Figure 2 shows the evolution of deactivation parameters (Ctl and kl) of the immobilized derivative as a function of both the temperature and the sorbitol content (relative to the amount of protein). As it can be seen, both parameters were very

414 sensitive to the increase in temperature, increasing the k~ parameter (from 30 to 60 ~ the k~ parameter was increased in 15-times) and decreasing the oq parameter, respectively. Additionally, the presence of sorbitol not change the rate constant of activity decay (k~), showing that the deactivation process during the first step of the proposed model was dependent of the assayed temperature rather than the protective effect of sorbitol. The positive influence of sorbitol was observed by the enhancement of the o.4 specific activity of the intermediate state (czl). 0.2 However, only for 30 ~ the increase in the cz~ parameter was proportional to the sorbitol 0.0 content, while for the remaining assayed temperatures, the stabilization effect of sorbitol was only observed at the lowest assayed ~ 2 concentration. The high hydrophilicity of 1 sorbitol allows to establish that the major 0 amount of this additive should be placed into the 0 20 40 60 80 microenvironment of the immobilized enzyme. Sorbitol content Thus, the enhancement of the immobilized cz( m m o l / g protein) chymotrypsin stability by sorbitol must be explained by both the maintenance of the Figure 2. Effect of sorbitol content of the enzyme structure and the interaction with water immobilized derivative on the kinetic molecules. Gekko and Timasheff[12] concluded deactivation parameters at different tempethat the low hydrophobicity of polyols ensured ratures [30 (o), 40 (m), 50 ( 0 ) and 60 ~ that the solvatation shell around the exposed (A)]. non-polar groups remain intact at the assayed temperature, leading to protein stabilization. Furthermore, the presence of ethanol, as hydrophilic vehicle, into the organic media must give a partition of sorbitol concentration from the micro- to the macroenvironment of the immobilized enzyme. Thus, by an increase in temperature, ethanol enhanced its ability to reduce the sorbitol concentration near to the enzyme, and this fact should be related with the loss in its stabilizing function. Additionally, as Aw-reducing agent, sorbitol reduces the concentration of "free" water molecules in the microenvironment of the enzyme by increase in the hydrogen bonds interaction. Figure 3 shows the evolution of the Aw parameter of the immobilized enzyme as a function of the sorbitol content. As it can be seen, in this "dry" system, the initial Aw (0.67) is dearly reduced with respect to an aqueous medium, and the increase in sorbitol concentration produced an additional Aw reduction. So, the negative effect of sorbitol at high concentration (Figure 2) could be related with its ability to remove water molecules from the essential hydration layer of the enzyme, which could contribute to the break out of hydrophobic intramolecular interactions, to finally yield inactive protein molecule [2, 8, 9]. In this way, at all assayed temperatures, an optimum sorbitol concentration was observed (20 mmol/g protein) due to the most favourable environment.

415

0.70

2nm) of the tryptophan residues were observed before and after treatment of the enzymes with SC-CO2 at temperatures below 65~ (see also table 2). Table 2 Fluorescence emission maxima of esterase EP10 and lipase from Candida cylindracea (for experimental details read section materials and methods) Temperature [~ Esterase EP 10 Lipase from Candida cvli~acea ,

untreated 35 45 55 65

335 335 335 335 334

,

339 339 339 340 341

These results indicate that no conformational change of the studied proteins occurs caused by treatment with SC-CO2. Above this temperature significant differences between the temperature stabilities of the enzymes can be found. Lipase from Candida cylindracea was incubated at the same temperatures for 1 hour in a dryer (under air-oxygen) because it was unclear by which mechanism enzyme inactivation occurs. Interestingly the enzyme is more stable in air-oxygen than in SC-CO2 (see table 3). These results indicate that enzyme

475 inactivation at temperatures > 85~ is not caused by oxidation of certain amino acid residues (due to the use of technical carbon dioxide for all experiments) because lipase from Candida cylindracea is more stable in air. Table 3 The residual activities of lipase from Candida cylindracea after treatment with SC-CO, (1 hour, 150 bar) an.d under air oxygen our, atmospheric pressure). Temperature [~ SC-COz ..... Air-oxygen 75 85 95 105

111.4 • 2.8 94.3 + 5.6 57.9• 3.0• 1.6

127.3 + 5.8 121.8 • 2.8 131.0• 1.6 94.8 + 17.8

A possible explanation for the lowest temperature stability of esterase EP10 is the low purity of the enzyme preparation. The results shown in Figure 2 fit very well to this hypothesis, because a decomposition process may cause enzyme inactivation (after about 4 hours almost all of the unstable compound is consumed and enzyme inactivation reaction is slowed down). The moderate stability of this enzyme against pressurization/depressurization can not be explained by the fact that a crude enzyme was used (figure 3), because at 35~ a decomposition process should be of less importance. By the experiments we have performed, it is difficult to speculate whether enzyme purity is an important factor for enzyme stability in SC-CO2. Further experiments are necessary to answer this question. Therefore in the near future the stabilities of preparations with different purities of one enzyme will be compared at the department. Kasche et al. [6] observed first that enzymes with S-S bridges (c~chymotrypsin,trypsin) were denatured to a lesser degree than the enzyme without cysteine (penicillin amidase). It is intended to increase the number of enzymes used for enzyme stability studies under comparative conditions to test the hypothesis if the number of disulfid-bridges is really effecting to enzyme stability in SC-CO2. REFERENCES

.

.

5. 6.

T.W. Randolph, H.W. Blanch, J.M. Prausnitz and C.R. Wilke, Biotechnol. Lett., 7 (1985) 325. D.A. Hammond, M. Karel and A.M. Klibanov, Appl.Biochem.Biotechnol., 11 (1985) 393. K. Nakamura, Y.M. Chi, Y.Yamada, and T. Yano, Chem.Eng.Commun., 45, (1986) 207. O. Aaltonen, M. Rantakyla ChemTech. 21 (1991), 240. E.Catoni, E.Cernia and C.Palocci, J.Mol.Cat. A:Chem., 105,(1996) 79. V. Kasche, R. Schlothauer and G. Brunner, Biotechnol.Lett., 10, (1988) 569.

476

.

9. 10. 11. 12. 13. 14. 15.

E Lozano, A. Avellaneda, R. Pascual and J.L.Iborra, Biotechnol.Lett., 18, (1996) 1345. J. Zagrobelny, F.V. Bright, Biotechnol. Prog., 8, (1992) 421. P. Pasta, G. Mazzola, G. Carrea and S. Riva, Biotechnol.Lett.,II,(1989) 643. S. Kamat, J.Barrera, E.J.Beckman, A.J.Russell, Biotechnol.Bioeng.,40 (1992) 158. S. Kamat, G.Critchley, E.J.Beckman, and A.J. Russell, Biotechnol.Bioeng.,46 (1995) 610. S.-H. Yoon, O.Miyawaki, K.-H.Park and K.Nakamura, J.Ferment.Bioeng., 82 (1996) 334. D.C. Steytler, P.S. Moulson, and J. Reynolds, Enzyme Microb.Technol. 13 (1991)221. E.Cemia, C.Palocci, F.Gasparrini and D. Misiti, Chem.Biochem.Eng. 8 (1994) 1. H. Michor, R.Marr, T. Gamse, T.Schilling, E.Klingsbichel and H.Schwab, Biotechnol.Lett. 18 (1996) 79.

Stability and Stabilization of Biocatalysts A. BaUesteros, F.J. Plou, J.L. Iborra and P.J. Hailing (Editors) 9 1998 Elsevier Science B.V. All rights reserved.

477

Studies on the stability o f a m i n o a c y l a s e in some organic solvents L. Boross a J. Kosfiry~ [~. Stefanovits-BhnyaP C.

Sisak b

and B. Szajhni c

aDepartment of Chemistry and Biochemistry, University for Horticulture and Food Industry, Budapest, Hungary bResearch Institute of Chemical Engineering, Pannon University of Agricultural Sciences, Veszprrm, Hungary cCOVENT Co. Inc., Budapest, Hungary

1. INTRODUCTION Aminoacylase (N-acylamino acid amidohydrolase, EC 3.5.1.14) is a hydrolase which catalyses not only hydrolysis of N-acyl-L-amino acids but their reverse synthesis in organic solvents of low water content [1-3]. Nowadays the application of reverse hydrolytic processes in non-conventional media is of increasing interest [4,5]. However, organic solvents influence not only the catalytic properties of enzymes but also their stability. Our recent results about the effects of dimethylformamide and dioxane on the stability of aminoacylase are presented in this paper.

2. MATERIALS AND METHODS 2.1. Materials Aminoacylase isolated from pig kidney with a specific activity of 2,000-3,000 U mg ~ protein was a commercial preparation (A 3010, Grade I) of SIGMA Chemical Co (St. Louis, MO, USA). N-Acetyl-DL-methionine was purchased from REANAL Factory of Laboratory Chemicals (Budapest, Hungary). N,N-dimethylformamide (DMF), 1,4-dioxane and other chemicals were commercial products of A.R. grade. DMF and dioxane were dried by distillation from phophorous pentoxide to remove the water content followed by a treatment with Molecularsieve Union Carbide Type 3 A (Fluka AG). Aminoacylase immobilized on a polyacrylarnide-type bead support (ACRYLEX C100) possessing carboxylic groups activated by water soluble carbodiimide with an activity of 3200 U g-~ dry gel was also a product of REANAL. The binding capacity of the support was 6.2+0.3 meq/g solid and particle size was 100-320 ).tm. 2.2. Assay of enzyme activity The activity of aminoacylase was assayed at 25 ~ by the ninhydrin method adapted by [6] using N-acetyl-DL-methionine as substrate. The changes in absorbance were detected at 570 nm. One unit of activity is defined as the amount of enzyme that catalyses the

478 liberation of 1 ~tmole L-methionine in 1 hour at 25 ~ and at pH 7.0. Samples containing DMF or dioxane were assayed in less than 15 see after mixing the solvent or after appropriate incubation time.

3. RESULTS AND DISCUSSION We have shown earlier [1] that DMF and dioxane inhibit aminoacylase in concentrations higher than 2-5%. Above 25-30% solvent concentration the hydrolytic activity of the enzyme is lost. The effect of these organic solvents on the stability of aminoacylase was studied both in native and immobilized form on a derivatized polyacrylamide-type bead support containing earboxyl groups (Aerylex C-100). The enzyme was mixed with aqueous organic solvent solutions and the hydrolytic activity was measured within 15 see and after longer incubation in 0.1 M potassium phosphate buffer (pH 7.0) containing 0.015 M N-aeetyl-DL-methionine (Fig. 1 and 2). We have found that in the ease of very short (15 see) incubation in aqueous media with low concentrations of either solvent, the activity of aminoaeylase is the same as that of untreated enzyme. However, during longer incubation the enzyme loses its activity and the rate of inactivation is increasing with the enhancement in the concentration of organic solvents. Because of the extensive dilution (125 times) of starting mixtures the organic solvent content of the test solutions is negligible. It was found that inactivation by DMF follows an apparent first order reaction kinetics in lower DMF concentrations but in higher concentrations the time curves of inactivation became complex (Fig. 3). According to the preliminary data of inactivation caused by dioxane it seems that the process is much more complex (Fig. 4). The chemical nature and polarity of DMF and dioxane are different. It is suggested that DMF is able to promote the dissociation of aminoaeylase dimer and the unfolding processes of dimer and monomer are parallel. In the ease of dioxane it seems that the unfolding is faster than dissociation of dimer. Probably dioxane perturbes the hydrophobic core of the protein molecule. The covalent immobilization of aminoaeylase on Aerylex C-100 increases its stability against the organic solvents mentioned above (Fig. 5). Aminoacylase preserves at least 40% of its original activity in mixtures even of 50% content of organic solvents after 20 min incubation. The rate of inactivation is faster in 50% dioxane containing mixture than in 50% DMF. The semilogarithmic representation of residual activities shows that inactivation of immobilized aminoaeylase follows a complex kinetics in both solvent mixtures composed of at least a fast and a slow phase. A detailed study of the problem is in progress in our laboratory. In an earlier study [7] it has been demonstrated that the solvent mixtures inside the gel particles and outside are the same.

ACKNOWLEDGEMENT This work is supported the Hungarian Scientific Research Fund (Grant No. T-015558). The authors wish to express their thanks to Miss S. M. Firisz and A. B6day for skillful technical assistance.

479

Figure 1. Effect of DMF concentration and incubation time on the stability of native aminoacylase in different DMF - 0.1 M phosphate buffer (pH 7.0) mixtures. 5 mg ml ~ enzyme solution containing different DMF in 0.1 M phosphate buffer (pH 7.0) were prepared and after appropriate incubation time samples were transferred (125 times dilution) into activity test solutions containing 0.015 M N-acetyl-DL-methionine in 0.1 M phosphate buffer (pH 7.0). The specific activity (3200 ~tmol h 1 mg ~ protein) of the control i.e. enzyme solution with no DMF was taken as hundred percent.

Figure 2. Effect of dioxane concentration and incubation time on the stability of native aminoacylase in different dioxane - 0.1 M phosphate buffer (pH 7.0) mixtures. 5 mg ml ~ enzyme solution containing different dioxane in 0.1 M phosphate buffer (pH 7.0) were prepared and after appropriate incubation time samples were transferred (125 times dilution) into activity test solutions containing 0.015 M N-acetyl-DL-methionine in 0.1 M phosphate buffer (pH 7.0). The specific activity (3200 gmol h 1 mg -~ protein) of the control i.e. enzyme solution with no dioxane was taken as hundred percent.

480

~'r ~,

.~, &~

3.54 I 3.52 3.5

10%DMF

3.48

v

3.46

~

~

9

20% DMF 30% DMF

3.44

3.42 0

10

20

30

40

50

Incubation time (min)

Figure 3. Plot of the logarithm of the residual enzyme activity as a function of incubation time in different DMF - 0.1 M phosphate buffer (pH 7.0) mixtures. 5 mg ml 1 enzyme solution containing different DMF in 0.1 M phosphate buffer (pH 7.0) were prepared and after appropriate incubation time samples were transferred (125 times dilution) into activity test solutions containing 0.015 M N-acetyl-DL-methionine in 0.1 M phosphate buffer (pH 7.0). The specific activity (3200 ~tmol h ~ mg ~ protein) of the control i.e. enzyme solution with no DMF was taken as hundred percent.

6)

E

3.6 3.4 3.2

0%dioxane A . ~ 5 % dioxane lO%dioxane 15%dioxane ~,~ 20%dioxane

'7 r'A 3 ~.c_. E ~ 2.8

.~, Q" 2.6 ~9 _o~

o

~

2.4 2.2 0

10

20

30


Figure 4. Plot of the logarithm of the residual enzyme activity as a function of incubation time in different dioxane - 0.1 M phosphate buffer (pH 7.0) mixtures. 5 mg ml 1 enzyme solution containing different dioxane in 0.1 M phosphate buffer (pH 7.0) were prepared and after appropriate incubation time samples were transferred (125 times dilution) into activity test solutions containing 0.015 M N-aeetyl-DL-methionine in 0.1 M phosphate buffer (pH 7.0). The specific activity (3200 ktmol h ~ mg ~ protein) of the control i.e. with no dioxane solution was taken as hundred percent.

481

a) 100 90

o~" v

80

.>_ o r m r= "o

60

~9

(I) n,

70 50

---.~

DMF

~

Dioxane

-.---~

DMF

40

30 20

10 I

I

I

20

40

60


b) 3.7

"o ,_

3.5

E

3.3

&

"7,

t- m A

Dioxane

E ~

._z-

2.9

~

2.7 2.5 I 0

I

I

20

40

''

I

60


Figure 5. Effect of incubation time on the hydrolytic activity of immobilized aminoacylase in 50% DMF or dioxane containing 0.1 M phosphate buffer (pH 7.0) mixtures (a) and semilogarithmic representation (b). 10 mg solid ml "! immobilized enzyme was suspended in 50% DMF or dioxane containing 0.1 M phosphate buffer (pH 7.0) and after appropriate incubation time enzyme was separated by centrifugation and was transferred into activity test solutions containing 0.015 M N-acetyl-DL-methionine in 0.1 M phosphate buffer (pH 7.0). The activities were impressed in % of the control i.e. non-incubated solvent-phosphate buffer suspensions containing immobilized enzyme. 4. R E F E R E N C E S

1. L. Boross, J. Kos~ry, t~. Stefanovits-B~ayai, C. Sisak and B. Szajhni, J. Biotechnol., (1998) (accepted for publication). 2. J.L. Iborra, J.M. Obon, A. Manjon and S.M. Canovas, Biocatalysis, 7 (1992), 37. 3. J. Koshry, C. Sisak, B. Szajhni and L. Boross, Biocatalysis, 11 (1994) 329. 4. K. Farber and M.C. Franssen, Trends in Biotechnol., 13 (1993) 63.

482 5. F. Terradas, M. Feston-Henry, P.A. Fitzpatrick and A.M. Klibanov, J. Am. Chem. Soc. 115 (1993) 390. 6. B. Szajhni, Acta. Biochim. Biophys. Acad Sei. Hung., 15 (1980) 223. 7. C. Sisak, J. Koshry, B. Szajhni and L. Boross, Proceedings of Third International Symposium on Biochemical Engineering, Stuttgart (1995) 190.


483

Cutinase activity and enantioselectivity in supercritical fluids Nuno Fontes, M.Concei~.o Almeida, Silvia Garcia, C61ia Peres, Jo~o Grave, M.Raquel Aires-Barros *, Cl/mdio M. Soares, Joaquim M. S. Cabral ~, Christopher D. Maycock and Susana Barreiros* Instituto de Tecnologia Quimica e Biol6gica, Universidade Nova de Lisboa, Quinta do Marquis, Apt. 127, 2780 Oeiras, Portugal Centro de Engenharia Biol6gica e Quimica, Instituto Superior T6cnico, Av. Rovisco Pais, 1096 Lisboa Codex, Portugal

We studied the performance ofFusarium solam pisi cutinase, immobilized on a zeolite, in supercritical carbon dioxide and ethylene. Water activity had a strong effect on the catalytic activity of the enzyme, unlike pressure to 300 bar. The enzyme was very selective towards one of the isomers of 1-phenylethanol, with an enantiomeric excess of virtually 100 % which did not depend on water activity, pressure or solvent. Computer modeling of the structures of transition states for the two enantiomers led to structural differences which explained the preference of the enzyme for the (R)-isomer.

1. INTRODUCTION The properties of supercritical fluids (SCF) which make them interesting solvents for extraction also make them attractive media for conducting enzymatic reactions [ 1]. The rapid growth of the market for enantiopure compounds [2] opens interesting perspectives for the use of SCF as solvents for enantioselective biotransformations. The state of hydration of an enzyme can have a pronounced effect on its activity in nonaqueous media [3,4]. The effect of pressure on enzyme activity is usually moderate in nonaqueous liquid solvents at pressures to a few hundred bar, although often larger in magnitude than in aqueous media [5-7], and can be very significant close to the critical point of a SCF [1,6,8]. Both enzyme hydration and pressure can affect enzyme enantioselectivity [9-11]. We studied the effect of enzyme hydration and pressure on the catalytic activity and enantioselectivity of immobilized cutinase from Fusarium solani pisi suspended in supercritical CO2. The influence of the medium was probed by obtaining a few data

* Author to whom correspondence should be addressed We acknowledge the support of Junta National de lnvestigacr Cientifica e Tecnol6gica (Portugal) through the contract PRAXIS 2/2.1/BIO/34/94 and the grant PBIC/C/2037/95 (C.M.S.), and of Funda~;5.o Calouste Gulbenkian through a grant (S.B.).

484 points in ethylene. The observed enantioselectivity was analyzed on the structural differences between transition states formed by the two enantiomers.

basis of

2. RESULTS AND DISCUSSION Due to the difficulty in distinguishing between the state of hydration of the enzyme molecules and that of the support in the immobilized enzyme preparation, we correlated enzymatic properties to the water activity, aw. To quantify this parameter, we built a scale of aw vs. water concentration in the solvent, in the presence of both substrates. Our results are shown in Figure 1. The fact that aw showed a linear dependence on water concentration indicated that the activity coefficient of water in the solvent mixtures was constant over the limited concentration range covered up to saturation. The solubility of water in CO2 increases significantly as pressure is raised from 80 to 250 bar at constant temperature [ 12], and this is reflected in the relative positions of the lines.

0.8 O

> 0.6 0

e~

0.4 0.2

0

0.4

0.8 1.2 water concentration/g dm-3

1.6

2

Figure 1. Water activity in the (CO2 + vinyl butyrate + 1-phenylethanol) solvent mixtures at 35 ~ as a function of water concentration. LeR line, 80 bar; right line, 250 bar. The salt hydrate pairs used and the respective aw values at 35 ~ [13] were: Na2HPO4.2/0 (0.19), Na2HPO4.7/2 (0.69), Na2HPO4.12/7 (0.90). More details of the experimental conditions given in reference 14.

Using the scales in the figure, we studied the dependence of enzyme activity on aw. Our results are shown in Figure 2. The activity of the enzyme is strongly dependent on the amount of water available to it, reaching a maximum at about aw = 0.5 and then declining, although not to zero. In water saturated ethylene at 80 bar and 15 ~ and at otherwise identical experimental conditions, the activity of cutinase was similar to that in saturated CO2. It is reasonable to conceive that an increase in temperature to 35 ~ in ethylene would

485 give rise to an increase in enzyme activity. This again supports the notion of an adverse effect of CO2 on enzyme function, whether or not the enzymes are immobilized [3,15,16]. The effect of pressure on the catalytic activity of an enzyme depends on the magnitude and the sign of the effective activation volume for the reaction, "AV~ [ 17]. Figure 2 shows a great similarity between the activity of cutinase, converted into pressure independent units by dividing by the molar volume of the solvent, at the two pressures tested, over the whole aw range. The calculation of accurate AV~ values would require more data but the results obtained so far make it reasonable to expect small in magnitude AVis for the immobilized cutinase preparation used. 100

'

80

Q C~ C3

60 Q

O

~ 4o

Q~S~

Q O

Q

"~ 2o 0

i

0

I

0.2

t

I

I

I"

0.4 0.6 water activity

l

"

I

0.8

,

1

Figure 2. Relative Vmax/K~ for the cutinase catalyzed transesterification of vinyl butyrate by 1-phenylethanol in CO2 as a function of water activity, at 3 5 ~ Full symbols, 80 bar; open symbols, 250 bar. More details of the experimental conditions and analysis given in reference 14.

The enzyme was virtually 100 % selective towards the (R)-isomer of phenylethanol, regardless of solvent, aw and pressure. To understand this, we did a computer modeling of the structures of transition states for the two enantiomers. These differed in the position of the methyl group attached to the chiral center, as seen in Figure 3. From this figure, a plausible explanation for the enzyme preference for the (R)-isomer emerges: the burial of the methyl group in the case of the (R)-isomer and the van der Waals interactions of this group with tyrosine 119 which immobilize and stabilize the transition state. A high degree of complementarity between the protein and the complex leads to a more efficient catalytic cycle. We note that the nonaqueous solvent may penetrate the active site to some extent [7,18] and could contribute to the stabilization of the above methyl group of the (S)-isomer. But apparently, these interactions are less effective from the standpoint of catalysis than those of the methyl group with protein residues in the (R)-isomer conformation.

486

Figure 3. Hypothetical transition states for transesterification of the two enantiomers of 1phenylethanol by cutinase. Details of the modeling are given in reference 14. REFERENCES 1. Kamat, S.V., Beckman, E.J., Russell, A.J. (1995) Crit. Rev. Biotechnol. 15, 41. 2. Koskinen, A.M.P. Enzymatic Reactions in Organic Media, Koskinen, A.M.P., Klibanov, A.M., Eds., Blackie Academic & Professional, 1996, p. 1. 3. Borges de Carvalho, I., Corr~a de Sampaio, T., Barreiros, S. (1996) Biotechnol. Bioeng. 1996, 49, 399-404. 4. Bell, G., Janssen, A.E.M., Hailing, P.J. (1997) Enzyme Microb. Technol. 20, 471. 5. Kim, J., Dordick, J.S. (1993) Biotechnol. Bioeng. 42, 772. 6. Fontes, N., Nogueiro, E., Elvas, A.M., Corr~a de Sampaio, T., Barreiros, S. (1998) Biochimica et Biophysica Acta 1383, 165. 7. Michels, P.C., Dordick, J.S., Clark, D.S. (1997) J. Am. Chem. Soc. 119, 9331. 8. Chaudhary, A.K., Kamat, S.V., Beckman, E.J., Nurok, D., Kleyle, R.M., Hadju, P, Russell, A.J. (1996) J. Am. Chem. Soc. 118, 12891. 9. Orrenius, C., Norin, T., Hult, K., Carrea, G. (1995) Tetrahedron: Assymetry 6, 3023. 10. Kamat, S.V., Beckman, E.J., Russell, A.J. (1993) J. Am. Chem. Soc. 115, 8845. 11. Ikushima, Y., Saito, N., Arai, M., Blanch, H.W. (1995) J. Phys. Chem. 99, 8941. 12. Chrastil, J.(1982) J. Phys. Chem. 86, 3016. 13. Hailing, P.J. (1992) Biotechnol. Tech. 6, 271. 14. Fontes, N., Almeida, M.Conceig~o, Peres, C., Garcia, S., Grave, J., Aires-Barros, M.Raquel, Soares, C.M., Cabral, J.M.S., Maycock, C.D., Barreiros, S. (1998) Ind. Eng. Chem. Res., in press. 15. Kamat, S., Critchley, G., Beckman, E.J., Russell, A.J. (1995) Biotechnol. Bioeng. 46, 610. 16. Almeida, MC., Ruivo, R., Maia, C., Freire, L., Corr~a de Sampaio, T., Barreiros, S. (1998) Enzyme Microb. Technol. 22, in press. 17. Van Eldik, R., Hubbard, C.D. Chemistry under Extreme or Non-Classical Conditions,, van Eldik, R., Hubbard, C.D., Eds., Wiley and Spektrum, 1997, p.53. 18. Parker, M.C., Moore, B.D., Blacker, A.J. (1995) Biotechnol. Bioeng. 46, 452.


487

Effect o f pressure on e n z y m e activity in c o m p r e s s e d gases M.Concei~ao Almeida, Nuno Fontes, Eugtnia Nogueiro, Silvia Garcia, Ctlia Peres, Anttnio Silva, Manuel Carvalho and Susana Barreiros* Instituto de Tecnologia Quimica e Bioltgica, Universidade Nova de Lisboa, Quinta do Marquis, Apt. 127, 2780 Oeiras, Portugal We studied the effect of pressure to 300 bar on the catalytic activity of subtilisin Carlsberg suspended in compressed ethane, and of immobilized Candida antarctica lipase B - Novozym 435 - suspended in compressed carbon dioxide, ethane and propane. Subtilisin activity in ethane decreased with increasing pressure and this effect was more pronounced close to the critical point of the solvent, resulting in large in magnitude pressure dependent activation volumes. The catalytic activity of Novozym was hardly affected by pressure in any of the solvents.

1. INTRODUCTION High pressure may be used advantageously in many areas of biotechnology, e.g. to prolong storage time and prevent microbial contamination in foods [ 1,2], to improve protein stability at high temperature [3], to affect enzyme activity [4] and selectivity [5]. The effect of pressure on the catalytic activity of an enzyme depends on the magnitude and the sign of the effective activation volume for the reaction, AV~. AV~ values for enzymatic reactions in aqueous solvents are generally of the order of + 50 cm 3 mol ~ [6]. Comparatively larger in magnitude AVis have been obtained for biocatalytic processes in nonaqueous media, organic [7-9] and supercritical or near-critical [8,10]. In the latter case, small changes in pressure close to the critical point can have a great effect on reactions rates. Here we examined the effect of pressure to 300 bar on the catalytic activity of subtilisin Carlsberg suspended in compressed ethane, and of immobilized Candida antarctica B lipase -Novozym 435 - suspended in compressed carbon dioxide, ethane and propane.


2.1. Subtilisin Carisberg Water may affect the catalytic performance of enzymes in nonaqueous solvents in * Author to whom correspondence should be addressed We acknowledge the support of Junta Nacional de Investigac~o Cientifica e Tecnol6gica (Portugal) through the contract PRAXIS 2/2.1/BIO/34/94 and of Funda~o Calouste Gulbenkian through a grant (s.B.).

488 many ways, e.g through its impact on enzyme dynamics [11,12], the stabilization of transition state charges [ 13], the drive for hydrolysis over synthesis. It is thus very important to quantify enzyme hydration and to refer the observed enzymatic properties to this parameter. In the case of subtilisin, we quantified enzyme hydration by means of a mass balance with the water initially present in both the enzyme and the solvent, and the water present in the solvent at water partitioning equilibrium with the enzyme. The plot of activity as a function of enzyme hydration is a bell-shaped curve for lyophilized subtilisin Carlsberg [11,14]. Comparisons of the catalytic performance of the enzyme in different solvents should be performed at a fixed point of that curve. This procedure should also be applied to a single solvent when its physical properties vary significantly with changes in experimental conditions, as is the case with compressible solvents not far from the critical point. The effects of these changes on the water sorption isotherm for the enzyme must be accounted for, so that variations in its state of hydration may be corrected. Figure 1 shows the effect of pressure on the catalytic activity of subtilisin suspended in compressed ethane. The parameter k~t/Km was calculated by taking the fraction of intact active centers to be 15 % and invariant with pressure [9]. The line represents experiments performed in the presence of higher concentrations of alcohol [8], whereas the symbols represent points obtained with alcohol concentrations about five times lower. The use of low concentrations of substrates in the latter case ensured that the reaction mixture behaved very much like the pure solvent. The direct effect of pressure on the reaction rate constant for an elementary process, k, is determined by the activation volume, AV~, which is the difference in volume between the activated and ground states [15]: [(alnk/aP)T =-AV~/RT], where P is the pressure, T the

4.5 ~

'

O

4

I

~ 3.5 O

_= 2.5

O

i

50

I

100

o

'1

o

I

150 200 Pressure/bar

o

I

250

o

300

Figure 1. Dependence of the logarithm of the catalytic activity of subtilisin for the transesterification of N-acetyl-L-phenylalanine ethyl ester by 1-propanol on pressure, in ethane at 3 5 ~ Enzyme hydration was approximately 15 %. Concentrations of 1-propanol of nearly 0.9 M (line) and 0.2 M (symbols). More details of the experimental conditions and analysis given in reference 8.

489 temperature and R the gas constant, with k based on pressure independent units. The catalytic activity of subtilisin decreases with increasing pressure, yielding positive AV#s. The variable plotted on the Y-axis of the figure was converted into pressure independent units by dividing by the molar volume of the solvent at each pressure of interest. Using the molar volume of the solvent instead of that of the mixture may be an oversimplification close to the critical point, as discussed [8]. This could artificially lower the catalytic efflciencies obtained at lower pressures, and thus lead to smaller in magnitude activation volumes. The line in Figure 1 corresponds to a constant activation volume of 139 cm 3 mol 4 [8]. The symbols in that figure lead to pressure dependent AV%, plotted in Figure 2. Such large in magnitude AV% should reflect a local density enhancement of the solvent around the solute molecules [16].

5OO !

o 400

o 300 E 200

.~ ]oo < o

:

50

',

100

i

I

150 Pressure/bar

I

I

200

250

Figure 2. Dependence of the activation volume for the subtilisin catalyzed transesterification of N-acetyl-L-phenylalanine ethyl ester by 1-propanol on pressure, in ethane at 35 ~ Concentrations of 1-propanol of nearly 0.2 M.

2.2. Novozym 435 At water partitioning equilibrium between enzyme and solvent, there is one value of the water activity, aw, which characterizes all of the phases present in the heterogeneous system. This value is related to the state of hydration of the enzyme. For immobilized enzyme preparations, it is not possible to distinguish between the hydration of the enzyme mlecules and the hydration of the support. However, as long as aw is kept constant and solvents do not differ appreciably in hydrophylicity, enzyme hydration in the various media should be the same for a given aw [17,18]. Aw may be controlled by addition of salt hydrate pairs to the reaction medium [19]. We used this technique indirectly to build a scale of aw vs. water concentration, and could show that the transesterification activity of Novozym 435 was nearly insensitive to aw over the whole aw range in compressed propane, ethane and CO2 [20]. This allowed us to neglect

490 changes in water activity brought about by changes in solvent density when studying the effect of pressure on enzyme kinetics. The results we obtained are shown in Figures 3 and 4. Because the bulk enzymatic preparation was found to suffer from internal diffusional limitations [20], we included in the present study the smaller enzyme particles. As the figures show, pressure to 300 ba~rhas a very small effect on the performance of the enzyme 0.15 m

!1

"7 =9 0.12

A r-3

!

o

Lx

0.09

E E

~ 9 0.06

:~ 0.03 9 ,,,,4 x-.-

50

100


250

;,

300

Figure 3. Dependence of the initial rates of the Novozym catalyzed transesterification of nbutyl acetate by 1-hexanol on pressure, at 35 ~ Symbols for solvents: V, carbon dioxide, A, ethane; r'i, propane. Open symbols, bulk enzymatic preparation; full symbols, 100-300 IxM diameter enzyme particles. More details of the experimental technique given in reference 20.

"7 .~

0.1

~

0.08

i

0.06

~

AA r---3

~ . r--q

.

.

A. r--q

.

A r--q

0.04 /

"=~g'=0.02

1

50

100


250

3OO

Figure 4. Dependence of the initial rates of the Novozym catalyzed transesterification of nhexyl acetate by 1-butanol on pressure, at 35 ~ Symbols for solvents as in Figure 3.

491 in all three solvents. This could be explained in part by the fact that we were never as close to the critial point of any the solvents as in the previous study with subtilisin, but should also reflect a particular characteristic of the enzyme. Figures 3 and 4 again present evidence of a lowered catalytic activity of an enzyme in C02 [21,22]. REFERENCES

1. Pothakamury, U.R., Barbosa-Chnovas, G.V., Swanson, B.G., Meyer, R.S. (1995) Chem. Eng. Prog., March, 45-53. 2. Mozhaev, V.V., Heremans, K., Frank, J., Masson, P., Balny, C. (1994) TIBTECH 12, 493-501. 3. Mozhaev, V.V., Lange, R., Kudryashova, E.V., Balny, C. (1996) Biotechnol. Bioeng. 52, 320-331. 4. Mozhaev, V.V., Heremans, K., Frank, J., Masson, P., Balny, C. (1996) Proteins 24, 8191. 5. Guthmann, O., Schwerdtfeger, R., Ricks, A., Antranikian, G., Kasche, V., Brunner, G. (1996) in High Pressure Chemical Engineering, von Rohr, P., Trepp, C., eds., Process Technology Proceedings 12, Elsevier, p. 127-131. 6. Morild, E. (1981) Adv. Prot. Chem. 34, 93-166. 7. Kim, S., Johnston, K.P. (1987) Ind. Eng. Chem. Res. 26, 1206-1213. 8. Fontes, N., Nogueiro, E., Elvas, A.M., Corr~a de Sampaio, T., Barreiros, S. (1998) Biochimica et Biophysica Acta 1383, 165. 9. Michels, P.C., Dordick, J.S., Clark, D.S. (1997) J. Am. Chem. Soc. 1997, 9331-9335. 10. Kamat, S.V., Beckman, E.J., Russell, A.J. (1995) Crit. Rev. Biotechnol. 15, 41-71. 11. Affieck, R., Xu, Z.F., Suzawa, V., Focht, K., Clark, D.S., Dordick, J.S. (1992) Proc. Natl. Acad. Sci. USA 89, 1100-1104. 12. Burke, P.A., Griffin, R.G., Klibanov, A.M. (1992) J. Biol. Chem. 267, 20057-20064. 13. Xu, Z.-F., Affleck, R., Wangikar, P., Suzawa, V., Dordick, LS., Clark, D.S. (1994) Biotechnol. Bioeng. 43, 515-520. 14. Corr~a de Sampaio, T., Melo, R.B., Moura, T.F., Michel, S., Barreiros, S. (1996) Biotechnol. Bioeng. 50, 257-264. 15. Savage, P.E., Gopalan, S., Mizan, T.I., Martino, C.J., Brock, E.E. (1995) AIChE J. 41, 1723. 16. Kim, S., Johnston, K.P. (1987) Am. Chem. Soc. Symp. Ser. 329, p.42-55. 17. McMinn, J.H., Sowa, M.J., Charnick, S.B., Paulaitis, M.E. (1993) Biopolymers 33, 1213-1224. 18. Parker, M.C., Moore, B.D., Blacker, A.J.(1995) Biotechnol. Bioeng. 46, 452-458. 19. Zacharis, E., Omar, I.C., Partridge, J., Robb, D.A., Hailing, P.J. (1997) Biotechnol. Bioeng. 55,367-374. 20. Almeida, M.C., Ruivo, R., Maia, C., Freire, L., Corr~a de Sampaio, T., Barreiros, S. (1998), Enzyme Microb. Technol. 22, in press. 21. Kamat, S., Critchley, G., Beckman, E.J., Russell, A.J. (1995) Biotechnol. Bioeng. 46, 610-620. 22. Borges de Carvalho, I., Corr~a de Sampaio, T., Barreiros, S. (1996) Biotechnol. Bioeng. 49, 399-404.


Immobilized enzymes



495

Activity and structural stability of adsorbed enzymes W. Norde and T. Zoungrana Laboratory for Physical Chemistry and Colloid Science, Wageningen Agricultural University, P.O. Box 8038, 6700 EK Wageningen, The Netherlands*)

A proteolytic enzyme, {x-chymotrypsin, and a lipolytic enzyme, cutinase, were adsorbed from aqueous solution onto sorbent surfaces of varying hydrophobicity and morphology. With both enzymes the affinity of adsorption is larger for the hydrophobic surface. Water-s01uble, flexible oligomers attached at the sorbent surface cause a reduction of the protein adsorption affinity. Circular Dichroism (CD) spectroscopy as well as Differential Scanning Calorimetry (DSC) indicate severe structural perturbations in the enzymes resulting from adsorption. The degree of structural perturbation, that is the fraction of the enzyme molecules that is perturbed by adsorption, decreases with decreasing hydrophobicity of the sorbent surface, with increasing degree of coverage of that surface by the enzyme, and with the presence of watersoluble oligomers at the sorbent surface. The specific activities of the enzymes are reduced upon adsorption, more or less following the extent of structural perturbation. When the enzyme is adsorbed its specific activity is much less sensitive to temperature variation.

1. INTRODUCTION In various biocatalytic processes, especially in the food and biomedical industries, enzymes immobilized on a (solid) matrix are employed. Immobilization allows a flow process in the bioreactor and it facilitates recovery of the enzyme. Thus, using immobilized enzymes may render the process less costly and/or more sustainable. Among the various methods of enzyme immobilization [1 ] physical adsorption may be the simplest one and it can be performed under relatively mild conditions. To use them successfully adsorbed enzymes must still be biologically active. This requirement should not be taken for granted, since upon adsorption the enzyme changes its environment which may lead to a rearrangement in the 3-dimensional protein structure. This, in turn, could very well affect the enzymatic activity. In this paper we discuss the influence of adsorption on the protein structure and on the enzymatic activity. The protein structure was studied by differential scanning calorimetry (DSC) and by circular dichroism (CD) spectroscopy. These two techniques are complementary" DSC provides information on the overall protein structural stability with respect to thermal unfolding, whereas CD probes the sub-molecular, secondary structure of the protein. *) The support from the Graduate School VLAG, The Netherlands, and from the Human Capital and Mobility program of the European Commissionis gratefully acknowledged.

496 The enzymatic activity was established as the specific activity, SA (= activity per unit mass of enzyme), - temperature profiles. Performing these types of experiments the relation between protein structure and enzymatic activity was evaluated. This was done for the enzymes in solution as well as in the adsorbed state, thereby gaining information on the effect of adsorption on the structure-function relationship of the enzymes.

2. MATERIALS 2.1. Enzymes and Substrates ct-chymotrypsin was supplied by Merck, Germany (lot no. 102307). N-acetyl tyrosine ethylester (ATEE), was purchased from Sigma. Cutinase from Aspergillus niger orizae was received from Unilever Research Laboratory, Vlaardingen, The Netherlands. Para-nitrophenyl butyrate (pNPB) was supplied by Sigma. In Table 1 some characteristics of the two proteins are given. 2.2. Sorbent materials The sorbents were supplied as colloidal particles. Teflon particles (DuPont de Nemours) consist of perfluoroalkoxyfluorocarbon resin [2]. This material does not contain UV-adsorbing double bonds or aromatic groups. The refractive index of this teflon (1.35) is comparable to that of water (1.33) so that it shows only very little scattering in the far UV. Because of the negligible UV absorption and the insignificant light scattering the teflon particles are suitable as sorbents in the investigation of the circular dichroism of adsorbed proteins. PS are negatively charged polystyrene particles prepared as described in references [3] and [4]. PS-(EO)8 are the same PS particles on which oligo- (ethylene oxide) moieties are covalently grafted [5]. Each of these oligomers consists of 8 monomers of ethylene oxide. Because the average density of (EO)8 on the PS surface is ca. one oligomer per 2.5 nm 2 [5] and the (EO)8oligomers are highly soluble in water, the PS-(EO)8 particles are sterically stabilized in an aqueous environment.

Table 1 Physical chemical characteristics of the proteins and the sorbent particles proteins

molar mass (Da)

dimensions (nm 3)

isoelectric point (pH-units)

o~-Chymotrypsin Cutinase

25200 20600

5.1 x 4.0 x 4.0 4.5 x 3.0 x 3.0

8.1 7.8

sorbents

surface morphology

water wettability

surface charge at pH 7.1

Teflon PS PS-(EO)8 Silica Aerosil OX-50 Silica Ludox HS-40

smooth smooth hairy smooth smooth

hydrophobic hydrophobic hydrophobic hydrophilic hydrophilic

negative negative negative negative negative

specific surface area (m2 g-l) 12.5 10.3 12.8 44 208

497 Aerosil OX-50 (Degussa) consists of pyrogenic non-porous silica particles. This silica was used except for the CD-measurements in which Ludox HS-40 (Du Pont de Nemours) was supplied. The silica particles of Ludox HS-40 are ultrafine; the large area/volume ratio allows a small volume fraction of silica in the samples so that interference of the circular dichroism signal from the protein by light scattering from the particles can be greatly reduced. Additional properties of the sorbent materials are included in Table 1. 3. METHODS

3.1. Adsorption Adsorption experiments were carried out at room temperature (ca. 22~ in phosphate buffer pH 7.1 and 0.01 M ionic strength. Varying amounts of protein were added to a constant amount of sorbent material so that the final volume of the mixture was 10 cm 3. The 10 cm 3tubes in which essentially no head-volume was present, were incubated for 16 hours in an endover-end rotator. Then, the samples were centrifuged and the protein concentration in the clear supematant was determined by UV absorbance at 280 nm. The amount of protein adsorbed was calculated from the difference between the protein concentrations before and after adsorption.

3.2 Desorption The protein-covered particles in the precipitate after centrifugation (see sec. 3.1) were redispersed in 10 cm 3 0.01 M phosphate buffer pH 7.1 or in 0.01 M NaCI solution, incubated for 16 hours in an end-over-end rotator and centrifuged again. The supematant was examined by UV for the presence of dilution-induced protein desorption.This procedure was executed at various elevated temperatures to check heat-induced desorption.

3.3 Circular dichroism spectroscopy CD spectra were recorded in the far UV region (185-260 nm) in a Jasco J-600 spectropolarimeter using protein concentrations ranging between 0.3 and 1.0 mg cm -3 in a 0.2 mm cuvette. From the far UV-CD spectra the secondary structure of the protein may be inferred [6]. The CD spectrum is in particular sensitive to the a-helical content of the protein: the ahelix content was estimated from the ellipticity at 222 nm. With the ultrafine silica particles (Ludox HS - 40) the adsorbed amount of protein could not be accurately determined because of incomplete separation of the particles from the solution. Hence, with these systems the influence of adsorption on the protein structure was assessed by comparing spectra of samples containing varying ratios of adsorbed and unadsorbed proteins.

3.4. Differential scanning calorimetry DSC measurements were performed in a Setaram Micro-DSC III. The sample cell contained 1-3 mg of protein and the reference cell the appropriate blank material (buffer solution or sorbent dispersion). The cells were thermally equilibrated at 20~ inside the calorimeter before heating at a rate of 0.5~ per minute up to 100~ whereafter they were cooled down to 20~ and re-heated at the same rate to 100~ again.

498

3.5. Enzymatic activity measurements Enzymatic activities of (x-chymotrypsin and cutinase at pH 7.1 were calculated from the rates of hydrolysis of ATEE and pNPB, respectively. The enzymes were incubated in a solution of ATEE (0.002 M) or pNPB (0.001 M) in 0.01 M NaCl at various temperatures.


4.1. Adsorption isotherms The adsorption data are displayed as adsorption isotherms in Figure 1, where the amount of protein adsorbed per unit sorbent surface area, F, is plotted vs. the concentration of protein in solution after adsorption, Cp. The initial rising part of the isotherms, (dF/dcp)Cp --> 0 reflects the affinity of the protein to adsorb. Thus, with both tx-chymotrypsin and cutinase the affinity for the hydrophobic surfaces is much higher than for the hydrophilic silica surface. Adsorption saturation of 2.0-2.5 mg m -2 is compatible with a complete monolayer of more or less native protein molecules. The adsorptions of (t-chymotrypsin on teflon and on PS go beyond monolayer coverage of nativelike molecules. On these hydrophobic surfaces a second layer of protein molecules is adsorbed (possibly triggered by structural rearrangements in the molecules adsorbed in the first layer) or the protein molecules are perturbed severely in such a way that more mass can be accommodated in a single layer. Cutinase adsorption does not exceed monolayer coverage. On the silica surface adsorption saturation is relatively low and the S-shaped isotherm is characteristic of a co-operative adsorption process, that is the firstly adsorbed molecules facilitate adsorption of the subsequent ones. This would lead to two-dimensional aggregates of cutinase at the silica surface. (b) E 56 f

(a)

~

O)

E 4

o4 RE 2 I:~

E

C

3

1

0

0.2

0.4

0.6 0.8 Cp/ mg cm-3

0

0.1

0.2

I

0.3 ep / mg cm-3

01,

Figure 1. Adsorption isotherms of o~-chymotrypsin (a) and cutinase (b) for different sorbent materials. 9 silica; 9 PS-(EO)8; x PS; o teflon. T = 22~ For both proteins dilution does not lead to desorption from the hydrophobic surfaces, at least if F < 2 mg m -2. This last condition suggests that the protein more loosely attached beyond F = 2 mg m -2 is adsorbed in a second layer. In the case of hydrophilic silica the adsorption is irreversible with respect to dilution for F ~< 0.5 mg m -2. In all systems increasing the temperature does not lead to detectable desorption. The other experiments, i.e. CD spectroscopy, DSC and enzymatic activity measurements were performed under non-desorbing conditions.

499 4.2. Influence of adsorption on the three-dimensional protein structure The CD spectra of the proteins, both in solution and adsorbed on silica and teflon, are shown in Figure 2. 25000 i._ 20000 0

I

(a)

(b)

-

~--\

15000

10000 5000

i

~\

\

__.

"~ " x \ E-IO000 -5000

-15000

185

I 195

i 205

I 215

I 225

235

wavelength /

nm

245

185

195

I 205

I 215

I 225

I 235

I 245

wavelength / nm

Figure 2. Circular dichroism spectra of ot-chymotrypsin (a) and cutinase (b), in solution ( ), adsorbed (1.0 mg m -2) on teflon ( n _ m ) and incubated with silica ( m m n ) . According to the low ellipticity at 222 nm the helical content in dissolved tx-chymotrypsin is relatively small. This is in agreement with literature data [7,8]. After adsorption the spectra of o~-chymotrypsin are largely shifted, especially at the hydrophobic teflon surface. The larger effect on the protein structure at the hydrophobic teflon surface is in agreement with the adsorption isotherm data where, for that sorbent, severe structural changes in the first adsorbed layer were inferred. Moreover, at 222 nm, representative of the tx-helix content, the shifts for the hydrophobic and hydrophilic surfaces are in opposite directions. Teflon induces more o~helix in tx-chymotrypsin. Such a trend has been reported more often on changing the environment of the protein from polar (aqueous) into non-polar [2,9,10]. The spectra for the various o~-chymotrypsin-silica mixtures point to a decreased helical content. The spectra are essentially invariant with the adsorbed/unadsorbed protein ratio, suggesting that after 16 hours of incubation all the protein has been exchanged between the silica surface and the solution and that after release from the silica surface the structure remains in the conformation as it was in the adsorbed state. The CD spectrum for cutinase in solution indicates that this protein contains far more o~helix than o~-chymotrypsin. The effect of the teflon surface on the cutinase structure is relatively mild. Unlike ot-chymotrypsin, cutinase shows a (small) decrease in the amount of helix. The spectral shift for cutinase on the hydrophilic silica points to a helix reduction; it is relatively large which may be related to aggregation of the cutinase molecules on the silica surface, as discussed in see. 4.1. A decreased helical content induced by adsorption on silica has also been reported for other proteins [ 11,12]. A helix reduction in proteins at a hydrophilic surface and opposite structural responses for different proteins upon adsorption at a hydrophobic surface may be explained as follows. When a protein molecule arrives at the sorbent surface, at one side of the molecule the aqueous

500 environment is replaced by the sorbent material. Apolar parts of the protein that are buried in the interior of the dissolved molecule may become exposed to the sorbent surface where they are still shielded from contact with water. As a consequence, intramolecular hydrophobic interaction becomes less important as a factor stabilizing the globular structure of the adsorbed molecule. Because hydrophobic interaction between amino acid side groups in the protein's interior supports the formation of o~-helices, a reduction of that interaction tends to destabilize helices. Decrease of the helical content is expected to occur only if the peptide units released from the helices can form hydrogen bonds with the sorbent surface. This would be possible with oxide surfaces (e.g., glass, silica, metal oxides) or with residual water molecules remaining at the (hydrophilic) sorbent surface. Then, a decrease in the helix content may lead to an increased rotational mobility along the polypeptide chain and, hence, an increased conformational entropy of the protein. However, if the peptide units do not have the possibility to form hydrogen bonds with the sorbent surface, as is the case with hydrophobic sorbents, adsorption may induce extra intramolecular hydrogen bonding, thereby promoting the formation of secondary structures as or-helices and 13-sheets. Thus, whether adsorption on a hydrophobic surface causes an increased or a decreased ordering in the protein structure depends on the delicate balance between energetically favourable hydrogen bonds on the one hand and conformational entropy of the protein on the other. The CD spectra represent an average structure of a large ensemble of protein molecules; they do not give information about the variation in the conformational states of the protein population. That kind of information may be provided by DSC data. All transitions observed in the DSC thermograms are reflected in a single endothermic peak which indicates that the thermal unfolding of the protein, in solution as well as in the adsorbed states, occurs in a single step, characteristic for one-domain proteins. As an example, the DSC thermogram for dissolved ct-chymotrypsin is shown in Figure 3. The thermogram allows the evaluation of the denaturation enthalpy AdH as the transition peak area and the denaturation temperature Td as the temperature at half-peak area.

40

50

60 temperature (~

70

80

Figure 3. Differential scanning calorimetry of o~-chymotrypsin in solution, showing the heat-induced structural transition.

Cooling and re-heating showed that the temperature-induced structural transitions proceed irreversibly. The irreversibility is probably due to aggregation of the unfolded protein molecules. It is generally agreed [e.g. 13,14] that aggregation involves a relatively small (exothermic) enthalpy change, so that AdH essentially represents the enthalpy of the unfolding process. The DSC data, AdH and Td are summarized in Table 2.

501

Table 2 Differential scanning calorimetry between 20 ~ and 100~ of ot-chymotrypsin and cutinase in solution and adsorbed on various surfaces transition adsorbed amount (mg m -2) temperature (~ 49

ot-chymotrypsin in solution

%N

enthalpy of transition (J g-l) 20.4

100

no transition no transition no transition

tx-chymotrypsin adsorbed on teflon

0.70 1.10 1.60

ct-chymotrypsin adsorbed on PS

0.63 0.96 1.34

49 48

1.7 4.4

9 20

o~-chymotrypsin adsorbed on PS-(EO) 8

0.42 0.98 1.30

46 49 50

4.2 10.3 12.4

20 50 61

t~-chymotrypsin adsorbed on silica OX-50

0.13 0.34 0.49

50 47 48

9.0 15.0 16.9

44 74 83

53

30.0

100

cutinase in solution cutinase adsorbed on teflon

0.66 1.00 1.66

cutinase adsorbed on silica OX-50

0.10 0.14 0.24 . . . . . .

no transition

no transition no transition no transition 55 51 52

6.0 9.2 20.5 ,

,

20 31 68 ,,

For both ct-chymotrypsin and cutinase AdH is smaller when the proteins are adsorbed. However, Td is not significantly affected by adsorption. A decreased value for AdH recorded at the same Td indicates that the protein molecules in the adsorbed layer are in at least two different conformational states: a fraction is in a native-like conformation whereas the remainder has adopted a (or more) different conformation(s) that is thermo-stable up to, at least, 100~ This structural heterogeneity is expressed in the last column of Table 2, where %N (calculated from the value for AdH relative to that for the protein in solution) refers to the fraction of molecules that, below Td, are in a native-like conformation. It is evident that %N is smaller at the hydrophobic surfaces. This is in agreement with the general trend, reported in the literature [e.g. 15,16], that the extent of structural rearrangements in adsorbed proteins increases with increasing hydrophobicity of the sorbent surface. The influence of the (EOs)oligomers on the PS surface results in more adsorbed protein molecules retaining a native-like structure. Apparently, the (EOs)-moieties protect the protein molecules to some extent against

502

structural perturbation, probably by preventing intimate interaction between the protein and the PS surface. It is further interesting to note that %N increases with increasing degree of coverage o f the sorbent surface. This too is in line with the generally reported relation between the extent of structural changes and adsorbed amount [11, 17, 18], which is usually interpreted in terms of the ratio of the rates of deposition at the surface and of structural adaptation. The CD spectra for the two proteins adsorbed at teflon reveal that the structurally altered thermo-stable states (at least for T < 100~ still contain a significant amount of secondary structure. This may also be true for the perturbed states of the proteins at the other surfaces. (a)

12 -

0.3

~9

zx 9

6

o -~--Ro--1o

~9

20

-

30

9

o~

9

40

9

EE0.03 9

.~ 0.01 - o t0

L%

50

60

rl

l 20

i

X

l

xo

m

l 40

i

{

-

(c)

0.15 -

I 50

9

I

I 60

0.10 -

o

o

X

X

9

l 2O

I 30

0

I 40

zx

9 I 0

I 5O

0

X

o o

0.5

o

X

01

X

I

20

r

9

0

I 6O

I 70

Ox

l 60

I 70

I:~

X X

o

I 40

I

30

I 50

(c)

1.2 0.8 0.4

,

9

0

(b)

~

X

9

2.0

.o_ 1.6

9

0.05 0

~ >, N

m

9

i 30

2O

_->'1.0 _-2

o X

,,

?

o)

O

.,~ o 0.20

0

o

9 0

E 1.5

El

x

X

30

t:::

9 0

E

X

0

2

~

9

El

o

o

ox

~

A A

E m 2.5 -

9

o

9

9 o

4

o__~o___~o__~_o__o=o_o~.o___L ~

(b)

-'-9 0 . 0 2 -

9

"E

~.

9

E

E ~' N

9

9

S

W W ,~ 0.04

8

A

0.1

~

10

z~

0.2

(a)

-

o

o

o

o

[] o

I

i

I

40 50 temperature / ~

....

~

, I

60

0

__Xt

20

v

ix

30

X

XI

40

X

X

o

I

o

I

50 60 temperature / ~

I

70

Figure 4. Specific enzymatic activities of ot-chymotrypsin (left) and cutinase (fight). a: A in solution, 9 adsorbed on silica; 9 adsorbed on PS-(EO)8; x adsorbed on PS; o adsorbed on teflon b: adsorbed on silica at various surface coverages (as indicated in Table 2). c: adsorbed on teflon at various surface coverages (as indicated in Table 2). For more detailed information is referrred to the text.

503

4.3. Influence of adsorption on the enzymatic activity It is of course relevant to know whether the adsorbed enzymes in the native-like state are as active as the ones in solution and, also, whether the adsorbed perturbed molecules still possess enzymatic activity. The results of the enzymatic activity measurements are given in Figure 4 where the specific activity is plotted vs. temperature. Both enzymes lose activity upon adsorption. The activity of o~-chymotrypsin at the hydrophobic surfaces of teflon and PS has disappeared almost completely. The presence of the (EO)8-oligomers at the PS surface provides some protection against adsorption-induced inactivation. Most enzymatic activity is retained at the hydrophilic silica surface. For ot-chymotrypsin the influence of the sorbent surface on the enzymatic activity closely follows the trend in %N. For cutinase the relation between the activity and the structural rearrangements is more ambiguous. First, on the silica surface, where %N = 67, the activity is remarkably low. This could be due to reduced accessibility of the enzyme's active sites caused by surface aggregation of the cutinase molecules (cf. sec. 4.1). Second, on teflon, where %N = 0, cutinase is still somewhat active. Apparently, the structural perturbation is such that it does not completely destroy the enzymatic activity. This result is not so surprising considering the relatively small shift in the CD spectrum of cutinase due to adsorption on teflon. Finally, the observation that after adsorption the activities of ct-chymotrypsin and cutinase are less sensitive to temperature variation, so that at high temperature, say 500-60 ~ C, the specific activity in the adsorbed state may exceed that in solution could be of great relevance for the application of immobilized enzymes in bioengineering.

5. CONCLUSIONS The conclusions from this study may be summarized as follows: ot-chymotrypsin and cutinase adsorb on hydrophobic and hydrophilic surfaces. The affinity of the proteins to adsorb (the strength of the protein-sorbent interaction) increases with increasing hydrophobicity of the sorbent surface. The proteins undergo conformational change s upon adsorption, that is a part of the protein population in the adsorbed layer is structurally perturbed whereas the rest remains native-like. The structural alteration at the hydrophilic silica surface involves a decrease in the helical content of the proteins, whereas at the hydrophobic teflon surface ct-chymotrypsin and cutinase show opposite trends with respect to their change in secondary structure. The opposite trends can be explained by the delicate balance between energetically favourable hydrogen bond formation and the resulting effect on the conformational entropy of the protein. The fraction of native-like adsorbed molecules increases with decreasing strength of the protein-sorbent interaction and also with increasing sorbent surface coverage. The reduction in the enzyme's specific activity is more or less correlated to the fraction of the native-like structure in the adsorbed layer: for o~-chymotrypsin this correlation is better than for cutinase. It seems that the perturbed structure of adsorbed cutinase still contains enzymatic activity. The presence of water-soluble ethylene oxide oligomers at the sorbent surface causes a decrease in the adsorption affinity of ot-chymotrypsin and, hence, a larger number of native-like molecules in the adsorbed layer, which, in turn, results in a higher specific enzymatic activity. In the adsorbed state the activities of o~chymotrypsin and cutinase are less sensitive to temperature variation.

504 REFERENCES 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

M.H. Pinheiro, J.F. Kennedy and J.M.S. Cabral, in Interfacial Phenomena and Bioproducts, J.L. Brash and P.W. Wojciechowski (eds.), Marcel Dekker, New York, 1984, pp. 311-350. M.C.L. Maste, W. Norde and A.J.W. Visser, J. Colloid Interface Sci., 196 (1997) 224. J.W. Goodwin, J. Hearn, C.C. Ho and R.H. OttewiU, Colloid & Polymer Sci., 252 (1974) 464. J.W. Goodwin, R.H. Ottewill and R. Pelton, Colloid & Polymer Sci., 257 (1979) 61. M.C.L. Maste, A.C.P.M. van Velthoven, W. Norde and J. Lyklema, Colloids Surfaces A: Physicochem. Eng. Aspects, 83 (1994) 255. N.J. Greenfield, Anal. Biochem., 235 (1996) 1. N. Greenfield and G.D. Fasman, Biochemistry, 8 (1969) 4108. P. Manavalan and W.C. Johnson, Anal. Biochem., 167 (1987) 76. H.H.J. de Jongh and B. de Kruijff, Biochem. Biophys. Acta, 1029 (1990) 105. H. Vogel, F. J~ihnig, V. Hoffmann and J. Stiimpel, Biochim. Biophys. Acta, 733 (1983) 201. W. Norde and J.P. Favier, Colloids Surfaces, 64 (1992) 87. A. Kondo and J. Mihara, J. Colloid Interface Sci., 177 (1996) 214. S.P. Manly, K.S. Matthews and J.M. Sturtevant, Biochemistry, 24 (1985) 3842. P.L. Privalov and L.V. Medved', J. Mol. Biol., 159 (1982) 665. U. J6nsson, B. Ivarsson, I. LundstrSm and J. Berghem, J. Colloid Interface Sci., 90 (1982) 148. F. Grinnell and M.K. Feld, J. Biomed. Mater. Res., 15 (1981) 363. V. Ball, A. Bentaleb, J. Hemmerle, J.C. Voegel and P. Schaaf, Langmuir, 12 (1996) 1614. A. Kondo, F. Muramaki and K. Higashitani, Biotechnol. Bioeng. 40 (1992) 889.


Covalent immobilization support

505

of g l u c o s e o x i d a s e o n A1PO4 a s i n o r g a n i c

F. M. Bautista, M. C. Bravo, J. M. Campelo, A. Garcia, D. Luna, J. M. Marinas and A. A. Romero. Organic Chemistry Department, Faculty of Sciences, University of Cordoba, Avda. San Alberto Magno s/n, E-14004 C6rdoba, Spain.*

A procedure was studied to obtain a covalent attachment of glucose oxidase (GOD) to the surface of amorphous A1PO4 used as support. Immobilization of the enzyme is carried out through the a-amino group of lysine residues through an aromatic Schiffs-base. The enzymatic activities of native and immobilized GOD systems were obtained in the oxidation of ~-D(+)-glucose to D-gluconic acid. In different experiments the percentage of immobilized enzyme was in the range 50-90 % while the percentage of catalytically active immobilized enzyme was always near 90 %. Moreover, GOD immobilization increased its efficiency and operational stability when repeatedly used with respect to soluble enzymes.

1. I N T R O D U C T I O N Over recent years, there has been considerable interest in the development of some methods for the covalent attachment of biologically active species to the surface of inorganic materials [1]. This blend has brought to light many new applications in the fields of chemical sensing and biocatalysis [2,3]. Immobilization on the surface of a support material has been proposed to decrease mass transfer limitations associated to several immobilization techniques, such as entrapment or adsorption in gels [4]. Thus, support material should provide a large surface area suitable for enzyme reactions. In this respect, a sol-gel material such as A1PO4, previously reported as metal support [5], could be an adequate support component for the covalent attachment of enzymes. Many functionalized linkers have been used in immobilization procedures throughout the E-NH2 groups in the external lysines of the enzyme as binding group [6,7]. However, the most common way is through glutaraldehyde [8], or some other aliphatic aldehydes [9], to obtain a Schiffs-base, In this study attempts are made to find a suitable procedure to obtain a covalent attachment of Glucose Oxidase (GOD) to an inorganic support such as A1PO4. Immobilization of the enzyme is carried "This work was subsidized by a grant from the DGICYT (PB92/0816), Mimsterio de Educacion y Cultura and from the Consejeria de Educacion y Ciencia (Junta de Andalucia), Espafia.

506 ~/I--oH+ H2N-CH2-~-~-NH2---~ #I-- NH - CH2-

0

0

i 20

55 ~ (6th use) ( ..................

""

~g

,

40

Time (hours)

0

20

40

Time (hours)

0

20

40

Time (hours)

Figure 4. Influence of the reaction temperature on the catalytic activity of native (dotted lines) and immobilized enzyme under standard reaction conditions. The used immobilized enzyme is always the same throughout along the experiment, so that on increasing the reaction temperature there is a new re-use.

successive reactions after separation and washing of solid biocatalyst. Thus, after every assay (between 2 and 4 days continuously working) the solid was easily recovered after centrifugation, washed and then reused with fresh substrate mixture. Results in Fig. 3(b) indicates that, after re-use in six cycles, including filtration after centrifugation in stirred tubes and solid washing with phosphate buffer, working continuously along 23 days, residual activity was 25%. Besides, according to the results in Fig. 4, immobilization also led to an important increase in the thermal stability of GOD. Here we can see how immobilized GOD is able to work after re-use conditions while the soluble enzyme was inactivated as soon as the reaction temperature increased. Thus, at 35 ~ while a re-used immobilized GOD is able to finish the conversion at a good reaction rate, soluble GOD is inactivated after 10 hours. At increased temperatures inactivation of soluble GOD is faster, while the same immobilized enzyme is able to continuously remain active, instead of the re-use conditions indicated. Many references can be obtained in the literature where immobilized GOD is used as biosensor [2,13]. Polymer membranes are in most cases used as GOD carriers [14] and, in the best cases, the operational stability was 2 to 4 weeks [2,13] and the immobilized enzyme 40% [14]. Comparatively, references to inorganic supports are always very reduced. In these cases, silica gel is the most common support where immobilization is carried out by entrapment [3] or by

512 using glutaraldehyde as covalent linker [15]. However, the results here presented lead us to conclude that GOD covalently immobilized on an inorganic support such as A1PO4 could be advantageously used in most described applications. 4. CONCLUSIONS From the results obtained we can conclude that due to their textural and acid-base properties, amorphous A1PO4 may be a very adequate support component for the covalent attaching of organic chains to give new hybridized composites [16]. In this connection, the linker here studied open many possibilities associated to the facility through polymerization for obtaining organic "spacers" of variable length, attached to the A1PO4 surface. In this way, tailormade organic/inorganic composites when appropriately functionalized, may incorporate a variety of organic molecules including biocatalysts and enzymes, like GOD, with a high efficiency and yield. REFERENCES 1. E. Katchalski-Katzir, Trends Biotechnol., 11 (1993) 471. 2. G.G. Guilbault, Analytical Uses of Immobilized Enzymes, Marcel Dekker, New York, 1984. 3. Y. Tatsu, K. Yamashita, M. Yamaguchi, S. Yamamura, H. Yamamoto and S. Yoshikaw, Chem. Lett., (1992) 1615. 4. A. Wojcik, J. Lobarzewski and T. Blaszczynska, J. Chem. Tech. Biotechnol., 90 (1990) 287. 5. F.M. Bautista, J.M. Campelo, A. Garcia, R. Guardefio, D. Luna, J.M. Marinas and M.C. Ordofiez, Stud. Surf. Sci. Catal., 78 (1993) 227. 6. J.V. Sinisterra and A.R. Alcantara, J. Mol. Catal., 84 (1993) 327. 7. M. Shimomura, H. Kikuchi, H. Matsumoto, T. Yamamuchi and S. Miyauchi, Polym. J., 27 (1995) 974. 8. V. Bulmus, H. Ayhan and E. Piskin, Chem. Eng. J., 65 (1997) 71. 9. R.M. Blanco, J.M. Guisan and P.J. HaUing, Biotechnol. Lett., 11 (1989) 1277. 10. F.M. Bautista, J.M. Campelo, A. Garcia, R. Guardefio, D. Luna, and J.M. Marinas, J. Mol. Catal. A. Chem., 104 (1996) 229. 11. J.M. Campelo, A. Garcia, D. Luna, and J.M. Marinas, J. Catal., 111 (1988) 106. 12. F.M. Bautista, C. Bravo, J.M. Campelo, A. Garcia, D. Luna, and J.M. Marinas, Fabrication Procedure of Activated Inorganic Solids for Covalent Immobilization of Lipases and other Enzymes, Spanish Patent No. P9601847 (1996). 13. F. Mizutani, S. Yabuki and S. Ijima, Anal. Chim. Acta, 300 (1995) 59. 14. J. Liu and Y. Chung, Angew. Makromol. Chem., 219 (1994) 101. 15. J. Jen, J. Zen, F. Cheng and G. Yang, Anal. Chim. Acta, 292 (1994) 23. 16. T. Koyano, M. Saito, Y. Miyamoto, K. Kaifu and M. Kato, Biotechnol. Prog., 12 (1996) 141.


513

The effect of site-specific i m m o b i l i z a t i o n o n the t h e r m a l stability of thermolysin-like neutral proteases J. Mansfeld a, G. Vriend b, B. Van den Burg ~, G. Venema c, V. G. H. Eijsinkd, and R. Ulbrich-Hofmann a ,Martin-Luther University Halle-Wittenberg, Department of Biochemistry/Biotechnology, Institute of Biotechnology, Kurt-Mothes-Strat~e 3, D-06120 Halle, Germany bBiocomputing, EMBL, Meyerhofstrat~e 1, D-69117 Heidelberg, Germany cDepartment of Genetics, Center for Biological Sciences, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands dDepartment of Biotechnological Sciences, Agricultural University of Norway, P. O. Box 5051,1432 As, Norway

Starting from a cysteine-free mutant of the thermolysin-like neutral protease from Bacillus stearothermophilus, cysteines were introduced into different positions on the enzyme surface by site-directed mutagenesis. The mutant enzymes were immobilized via the SH-groups to Activated Thiol-Sepharose 4B and their thermostabilities were compared to those of the soluble enzymes. The results showed that the effects of immobilization on stability strongly depend on the site of attachment. Binding of the enzyme via engineered cysteines in the critical unfolding region between residues 56 and 69 led to a considerable increase of thermal stability, whereas the immobilization via a cysteine introduced remote from the unfolding region yielded less stabilization. An extremely strong stabilization was obtained upon binding via T56C where the half-life at 75 ~ was increased by the factor of 24. 1. INTRODUCTION Several members of the bacterial genus Bacillus produce extracellular neutral proteases that resemble thermolysin, the extremely stable protease from Bacillus thermoproteolyticus. The different thermolysin-like _proteases (TLPs) are characterized by large differences in thermal stabilities [1]. TLPs have been shown to be irreversibly inactivated at elevated temperatures as a result of autolysis [2]. Previous extensive mutational studies led to the assumption that the local unfolding processes preceding autolysis, involve surface-located regions, mainly in the N-terminal domain of the enzyme [3]. In the case of the TLP from B. stearothermophilus (TLP-ste), the solvent-exposed region between residues 56 and 69 has been shown to be crucial for thermal stability [3]. These findings prompted us to use TLP-ste for probing our model of protein stabilization [4] which attributes the success of protein stabilization by immobilization to the fixation of a specific structural region of the protein molecule,

514 namely the region where the unfolding l~rocess starts. For this purpose, cysteine residues were introduced by site-directed mutagenesis into the above-mentioned critical unfolding region as well as in other regions of the protein. The various single cysteine-containing TLP-ste mutant enzymes were immobilized site-specifically via their free sulfhydryl groups, and the effects of immobilization on stability were studied. 2. RESULTS A N D DISCUSSION 2.1. Design of mutants

Suitable positions for the introduction of cysteine residues were selected by studying a three-dimensional model of TLP-ste (built by homology, using the crystal structures of thermolysin and the TLP from B. cereus, as described previously [1]; TLP-ste has 86 % sequence identity with thermolysin). Solvent exposition and accessibility to functional groups on the carrier material were prerequisites. Furthermore, mutations expected to result in significant clashes or other negative side effects were excluded. Starting from the cysteine-free mutant of TLP-ste C288L, the mutants G8C, T56C, N60C, $65C, N181C were constructed by site-directed mutagenesis and expressed successfully in B. subtilis DBl17. With the exception of the mutant N60C, expression levels were similar to wild-type TLP-ste. Wild-type and mutant enzymes had similar specific activities towards casein showing that the selected mutations had only marginal effects on the catalytic properties o f the enzyme. 2.2. Immobilization of mutant enzymes on Activated Thiol-Sepharose 4B The mutant enzymes were purified by Bacitracin-silica affinity chromatography [5] in the presence of fl-mercaptoethanolin order to orevent intermolecular disulfide formation. Immediately prior to binding to "Activated Thiol-Sepharose 4B (Amersham-Pharmacia Biotech), the mutant enzymes were subjected to gel filtration on Sephadex G25-superfine to remove salts and tg-mercaptoethanol. The ~es~llet~ot the site-specific immobilization of mutant TLP-ste enzymes are shown in

Table 1: Immobilization of TLP-ste mutants on Activated Thiol-Sepharose 4B i

Mutant

G8C T56C N60C $65C N181C C288L thermolysin

,

,w

,,

Activity of immobilized enzyme

Amount of bound protein

Relative activity

Coupling yield

[U/~]

Imp/~ carrier]

[%]

[%]

100.0 52.3 75.2 73.0 75.2

18.5 16.4 41.5 36.6 41.5

i

2.6 31.1 33.8 34.8 33.8

0.33 0.22 0.33 0.60 0.33

0.6 0.9

0.03 0.09

,,

0.3 0.6

515 In contrast to conventional immobilization strategies using the amino or carboxyl groups of proteins, the immobilization procedure used i n this study yields an uniform orientation of the enzyme molecule on the carrier material and permits the study of local binding effects on the thermal stability of the protease. Under the conditions used, nonspecific binding could be reduced to a minimum, as shown by the low coupling yield of the cysteine-free mutant C288L and thermolysin (Table 1).

2.3. Thermal stability Thermal stabilities of soluble and immobilized mutant enzymes were determined by following the time course of thermal inactivation at pH 7.5. In all cases, inactivation followed first-order kinetics. Introduction of cysteines into the TLP-ste mutant C288L strongly affected the thermal stabilities of the enzymes, varying from marginal effects to significant stabilization or destabilization effects. The haIf-lives of ttie soluble enzymes ranged from 0.9 min for N60C to 19.9 min for N181C at 75 ~ in comparison to 10.2 min for the wild-type enzyme. Intermolecular disulfide formation and, presumably, oxidation of cysteines are assumed to be the reason for the lower thermal stability of some of the mutants. The results of the comparison of thermal stabilities of soluble and immobilized mutant enzymes are given in Table 2. Table 2: Comparison of thermal stabilities of immobilized and soluble mutant enzymes at 75 ~C The enzymes were incubated at 75 oC in 0.05 M Tris buffer pH 7.5/5 mM CaC12. Aliquots removed after different time intervals were assayed for activity towards casein at 37 ~ Mutant G8C T56C N60C $65C N181C

.

Ratio of half-lives of immobilized and soluble enzymes 1.33+0.05 23.70+0.61 5.075:0.19 9.56+0.35 3.20+0.08

The binding of the enzyme via engineered cysteines in the critical unfolding region (T56C, N60C, S65C)led to a considerable increase of thermal stability (Table 2). In contrast, the immobilization via a cysteine introduced remote from the unfolding region (N181C) yielded only small stabilization. Immobilization of the protease via the introduced cysteine at position 8 did not result in significant stabilization of the enzyme (Table 2), whereas the formation of a disulfide b r i d g e between cysteines 8 and 60 stabilized the enzyme remarkably [6]. This confirms that the region around residue 60 is crucial for stability. Anextremely strong stabilization of the enzyme by immobilization was observed with the T56C mutant (by the factor 24 in comparison to 5 - 10 for the other positions tested in the critical region ) . By computer modelling, this effect could be explained by additional favorable interactions of one of the carboxyl groups in the spacer molecule of the carrier material with the Ca2*-binding site 3 of the enzyme Previous studies had shown that this Ca2*-binding site is one of the determinan'ts of thermal stability of TLP-ste [7, 8].

516 The results of these site-specific immobilization studies, using genetically introduced cysteines at predefined locations into TLP-ste, show that the attachment site at the protein surface is significant for the success of stabilization by immobilization.

Acknowledgements This work was supported by a research grant from the Leopoldina, the Deutsche Akademie der Naturforscher, and by the Deutsche Forschungsgemeinschaft, Bonn.

REFERENCES [1] Vriend, G. and Eijsink, V. G. H. (1993) J. Computer-Aid. Mol. Design 7, 367396 [2] VanBiochem.den j.Burg'272,B.~o,_~,E~sink' V. G. H., Stulp, B. K., and Venema, G. (1990) [3] Eijsink, V. G. H., Veltman, O. R., Aukema, W., Vriend, G., and Venema, G. (1995). Nature Struct. Biol. 2, 374 - 379 [4] Ulbrich-Hofmann, R., Golbik, R., and Damerau, W. (1993) Stability and Stabilization of Enzymes (van den Tweel, W. J. J., Harder, A., Buitelaar, R., eds.) 497- 504, Elsevier, London [5] Van den Burg, B., Eijsink, V. G. H., Stulp, B. K., and Venema, G. (1989) J. Biochem. Biophys. Methods 18, 202- 220 [6] Mansfeld, J., Vriend, G., Dijkstra, B. W., Veltman, O. R., Van den Burg, B., Venema, G., Ulbrich-Hofmann, R., and Eijsink, V. G. H. (1997) J. Biol. Chem. 272, 11152-11156 [7] Dahlquist, F. W., Long, J. W., and Bigbee, W. L. (1976) Biochemistry, 15, 1103 1111 [8] Veltman, O. R., Vriend, G., Hardy, F., Mansfeld, J., Van den Burg, B., Venema, G., and Eijsink, V. G. H. (1997) FEBS Lett. 405, 241 - 244


517

I m m o b i l i z a t i o n o f h y d a n t o i n cleaving e n z y m e s f r o m A r t h r o b a c t e r a u r e s c e n s D S M 3747 - Effect o f the coupling m e t h o d on the stability o f the L-Ncarbamoylase M. Pietzsch ~ H. Oberreuter, B. Petrovska, K. Ragnitz and C. Syldatk Institute of Biochemical Engineering, University of Stuttgart,Allmandring 31, D-70569 Stuttgart, Germany

Different coupling methods were tested for the immobilization of the N-carbamoyl-Laminoacid amidohydrolase (L-N-carbamoylase) partially purified from Arthrobacter aurescens DSM 3747. The operational stability of the immobilized biocatalyst was measured using both consecutive batch reactions and continuously operated fixed bed reactors, while the stability of the flee L-N-carbamoylase was investigated using an enzyme membrane reactor. The long term stability of the enzyme was markedly enhanced by all immobilization methods and carriers tested. In consecutive batch reactions the operational stability remained relatively low and significant differences in biocatalysts stability were not observed between the different immobilization methods used. In contrast there were significant differences in the stability when the biotransformations were carded out using fixed bed reactors. As a result of this comparison the determination of the operational stability of the air-sensitive L-Ncarbamoylase on a batch-to-batch basis seems not to be useful. 1. INTRODUCTION Optically pure D- and L-amino acids can be produced with a theoretical yield of 100% and with high enantioselectivities using the "hydantoinase method" [ 1]. The instability of most of the enzymes involved in the hydrolysis of D,L-5-monosubstituted hydantoins and Ncarbamoylamino acids is the major hindrance for an economical industrial process. Even in resting cell biotransformations the hydantoinase and especially the L-N-carbamoylase are inactivated under reaction conditions and the biocatalysts are therefore not reusable. Hydantoinase and L-N-carbamoylase from Arthrobacter aurescens DSM 3747 have been already purified to homogenity by chromatographic methods and further characterized [2]. The free enzymes are inactivated significantly faster than in whole cells. On the other hand it is not possible to use the hydantoinase in combination with a chemical decarbamoylation step as reported for the D-hydantoinase-process [3], because the hydantoinase is not absolutely stereospecific for a range of interesting products [4]. Therefore the L-N-carbamoylase is essential for the production of enantiomerically pure L-amino acids. In the present paper we report on recent results obtained in the covalent immobilization of the "bottleneck-enzyme" L-N-carbamoylase using watersoluble carbodiimide [5].

"To whomcorrespondenceshouldbe addressed.

518 Investigations on the long term stability of the immobilized biocatalyst in a fixed bed reactor with the enzyme bound either covalently or adsorptively are reported as well. Consecutive batch reactions are often used for a first characterization of immobilized biocatalysts, therefore the half life time obtained in batch reactions is compared with those of continuously operated fixed bed reactors. 2. MATERIALS AND METHODS

General: All chemicals were of analytical grade and were purchased from Fluka Chemie AG, Buchs, Switzerland. Immobilization media were a gilt from R6hm GmbH, Darmstadt, FRG (Eupergit C) or purchased from Pharmacia Biotech, Freiburg, FRG (Q-Sepharose FF, EAH-Sepharose). Enzyme preparation: A. aurescens DSM 3747 was cultivated in a 300-1-bioreactor under conditions as reported previously [6] using 0.3 g/1 N-3-methylene-D,L-indolylmethylenehydantoin as inducer. Cell disruption was carried out continuously under optimized conditions according to the results presented elsewhere [7] using a cooled (-20~ 600 ml cell and the Dyno-Mill KDL (Willy A. Bachofen, Basel, Switzerland). Unleaded glass beads (480 ml, diameter 0.3 mm) were agitated at an agitator speed of 2500 rpm. An icecooled suspension of 738 g bio-wet-mass (BWM) suspended in 1722 ml of 0.3 M phosphate-buffer, containing 2.5 mM MnC12, pH 6.5 (30% BWM (w/v)) was pumped four times through the cell with a flow rate of 80 ml/min (crude extract). The enzyme fraction used for the immobilization experiments was obtained by expanded bed chromatography which was carried out as described before [4] with the following modifications. The crude extract (2320 ml) was diluted with 6000 ml 0.05 M K2HPO4-solution, 960 ml of 0.05 M phosphate buffer, and finally the pH was adjusted to pH 8.0 by adding carefully about 80 ml 12% NaOH. The diluted crude extract (8600 ml) was applied to a Streamline 50 column equilibrated with buffer A (0.05 M phosphate buffer, pH 8.0) and eluted by applying a linear gradient between buffer A and buffer B (0.05 M phosphate-buffer, 1.0 M NaC1, pH 8.0) within 6000 ml. The active fractions (1900 ml) were collected and stored at -20~ (50 ml portions; Streamlinefraction, purification factor: 1.3). Immobilization of the L-N-carbamoylase to Eupergit C: Under stirring 3.45 g solid phosphate buffer (lyophilized 0.05 M phosphate buffer, pH 8.5) were added to 20 ml of the thawed Streamline fraction and the pH was adjusted to pH 8.5 by adding 5 ml of a 1.0 M K~HPO4-solution. Precipitated protein was removed by centrifugation at 48000 x g (4~ 15 min). 5 g of dry Eupergit C were suspended in the supematant and the suspension was gently shaken for 24 h at 4~ The immobilized L-N-carbamoylase was filtered off and washed with 250 ml of 0.1 M Tris-buffer, pH 8.5, using a glas funnel, before using it for bioconversion experiments. Immobilization of the L-N-carbamoylase to EAH-Sepharose 4B: 14 ml of 0.05 M Trisbuffer, pH 6.0 (buffer C) were added to 1 ml of the thawed streamline fraction and the pH was adjusted to pH 6.0 by carefully adding 1% HC1. Eleven ml of EAH-Sepharose 4B were washed with buffer C and added to the diluted enzyme solution. A solution of 307.6 mg EDC in 1 ml distilled water (1.6 M, pH 6.0, adjusted with 1% HC1) was added to this suspension and shaken at 4~ for 4 h. To hydrolyze excess activated esters, the immobilizate was filtered off and washed with 100 ml 0.2 M Tris-buffer, pH 8.5.

519

Immobilization of the L-N-carbamoylase to Q-Sepharose Fast Flow: Eighty ml of the thawed streamline fraction were diluted with 160 ml 0.05 M Tris-buffer, pH 8.5. After addition of 16 ml Q-Sepharose FF (equilibrated with the same buffer), this suspension was shaken for 20 min at 4~ The immobilizate was filtered off and whashed with 200 ml of 0.05 M Tris-buffer, pH 8.5. The immobilizate was stored in buffer at 4~ Analytical methods: Protein concentration was determined according to the method of Bradford [8]. The test-kit was purchased from Biorad, Munich, FRG. HPLC-analysis was carried out as described elsewere [4]. Retention times: Tryptophan, 17.77 min; N-carbamoyltryptophan (CTrp), 23.11 min. Determination of the activity offree enzymes: Fifby Ixl of the enzyme solution were added to 800 ~tl of the preincubated (37~ substrate solution (0.4 g/1 L-CTrp in 0.05 M Tris-buffer, pH 8.5). The reaction was stopped by adding of 450 ~tl trichloroacetic acid (TCA, 12%) and analyzed by HPLC after centrifugation. Determination of the stability of the free enzyme using an enzyme membrane reactor: Thirteen ml of prewarmed substrate solution were filled into an ultrafiltration cell (Omegacell, Filtron, Northborough, MA, USA) equipped with an 30 kDa ultrafiltration membrane and 350 ~tl (0.61 U) of the thawed Streamline fraction were added. The ultrafiltration cell was fixed in a water bath (37~ Using a hose pump the substrate solution was pumped through the stirred cell with a flow rate of 1.6 ml/min equivalent to a mean residence time of 8.5 min. Samples were taken from the filtrate and directly analyzed by HPLC. Determination of the activity of immobilized enzymes in batch experiments: Eight hundred ~tl of the preincubated substrate solution (see above) were added to 100 mg of wet immobilizate and shaken for 10 min at 37~ The reaction was stopped by adding of 400 ~tl TCA and analyzed by HPLC after centrifugation. Determination of the operational stability: The operational stability of the immobilized LN-carbamoylase was determined by repeated batch experiments and continuous reactions in fixed bed reactors. For repeated batch experiments approximately 2.5 g of the wet immobilizate were added to 40 ml substrate solution (see above) and shaken for 90 min at 37~ The immobilizate was filtered off, washed with 50 ml 0.05 M Tris-buffer, pH 8.5, and reused after weighing. For continuous reactions the immobilized L-N-carbamoylase was packed into XK 16/10 columns (Pharrnacia, Freiburg, FRG). Substrate solution (see above) was pumped through the column at 37~ at different flow rates (see Results). Samples were taken and directly analyzed by HPLC. 3. RESULTS AND DISCUSSION For the immobilization of proteins different adsorptive, covalent and inclusion techniques have already been investigated and published [9]. Covalent immobilization is of major interest because the bonds formed between the protein and the carrier are more stable. On the other hand, adsorptive immobilization, e. g. to ion exchangers, sometimes is favourable, even if the half-life time of the biocatalyst is lower, because the support can usually be reused after regeneration [ 10]. For a rapid comparison of two covalent (oxirane, carbodiimide) and one electrostatic immobilization procedures, consecutive batch reactions have been carried out.

520 The activity of the L-N-carbamoylase of A. a u r e s c e n s DSM 3747 obtained after covalent attachment on Eupergit C via oxirane coupling and on EAH-Sepharose by watersoluble carbodiimide after repeated batch reactions is shown in Figure 1.

Figure 1: Operational stability of L-N-carbamoylase after immobilization to Eupergit C (a, oxirane method), and to EAH-Scpharose using carbodiimide (b). Repeated batch conversions with reusagc of the biocatalyst, initial activity was calculated from concentrations determined after 15 rain). The immobilization via epoxide groups (oxirane method) resulted in a dramatic loss of activity within the first 3 cycles when reusing the biocatalyst (Figure l a: Batch 1 to 4), whereas the remaining activity was almost maintained (batch 5 to 7). The reason for the decrease may be a slow conformational change or even a dissociation of the covalently bound subunits of the L-N-carbamoylase due to the decrease of salt concentration (no protein could be determined in the supernatant; the immobilization to Eupergit C was carried out in 1.0 M phosphate buffer whereas the reaction was carried out in 0.1 M Tris/HCl-buffer). A strong hint for this hypothesis is the observation that the free L-N-carbamoylase is stabilized by high salt concentrations during purification and is immediately inactivated if the salt concentration is reduced (results not shown). Table I- Half-lifetimes of free and immobilized L-N-carbamoylase

Support

Immobilization method

Half-lifetime [h] batch 0.5

Q-Sepharose FF

electrostatic interactions

14.7 c)

EAH-Sepharose

carbodiimide coupling

14.5 d)

Eupergit C oxirane coupling 9.4 ") a) Enzymemembrane reactor b)Fixed bed reactor c)(batch 2-5) d)(batch 2-6)

naifqifctime [hi continuous 2.7 a) 570 b) 73 b) 3000

b)

c) (batch 4-7)

521 The loss of activity was lower in the case of immobilization via carbodiimide (Figure lb). However the half-life time calculated from the batch experiments is comparable for the different coupling methods investigated (Table 1). The operational stabilities of the L-N-carbamoylase immobilized to different supports were also determined in continuous processes using fixed bed reactors. For a comparison electrostatic (Q-Sepharose) and covalently bound enzyme (Eupergit C, EAH-Sepharose) were run for several hundred hours in fixed bed reactors. As can be seen from Figure 2 and Table 1 the operational stability of the oxirane-immobilized enzyme is significantly higher than the carbodiimide-coupled enzyme. Eupergit C contains about 160 ~tmol oxirane groups per g wet carrier while EAH-Sepharose contains only 17 p.mol/g. The higher stability of the oxirane coupled enzyme therefore may be explained by a higher extent of covalent bonds formed between the carder and the enzyme. The differences in the stability observed using Figure 2: Operational stability of L-N-earbamoylase after consecutive batch reactions and immobilization to different supports (continuous operation fixed bed reactors may be in a fixed bed reactor, 0.4 g/1 L-N-carbamoyltryptophan in explained by the different 0.1 M Tris/HC1, pH 8.5). handling of the biocatalysts. In consecutive batch reactions the immobilizate is filtered off from the substrate solution after each batch and washed with buffer without substrate. Afterwards the weight was measured and the humid biocatalyst was reused. In contrast, the catalyst in fixed bed reactions was neither in contact with air nor with substrateless buffer. From the characterization of the enzyme it is known that it is sensitive against oxidation and requires metal ions, which can be removed by addition of EDTA [2]. On the other hand, the free enzyme was stabilized by the substrate when the reaction was carded out using an enzyme membrane reactor, as can be seen from Table 1. Therefore it is not astonishing that the operational stability measured in fixed bed reactors is higher than in batch experiments. Ks can be seen from Figure 2, in the case of the oxirane-coupled enzyme there is a two step deactivation of the immobilizate which may be explained as described above for the batch experiments. It is obvious that the interpretation of the results of the batch reactions is highly dependent on the number of batch reactions atter which the half-life time is calculated, which is therefore not sufficient for a comparison of different coupling methods and supports in the case of the L-N-carbamoylase.

522 4. CONCLUSIONS It has been shown that the immobilization of the L-N-carbamoylase from A. aurescens DSM 3747 results in significant increase of operational stability. Although the absolute specific activities obtained so far are too low to fulfill the requirements of an industrial biocatalyst, it could be shown that covalent coupling by oxirane groups resulted a highly stable biocatalyst. For economical reasons an adsorptive binding to an reusable anion exchange support may be favorable. The operational stability measured in continuous operation using fixed bed reactors is markedly higher than in batch reactions, which can be explained by the sensitivity of the enzyme against oxidation. For unstable enzymes as the L-N-carbamoylase used in these investigations, batch reactions are not sufficient for a comparison of different coupling methods. If batch reactions have to be used for the characterization one should avoid washing with substrateless buffers and exposition to air (e. g. by nitrogen atmosphere). Because of the relatively low activity of the immobilized L-N-carbamoylase further experiments have to be carried out with the enriched enzyme, and the influence of the density of the coupling groups of the support as well as the protein concentration has to be investigated.

REFERENCES

1. Pietzsch, M., and Syldatk, C. (1995), in Enzyme Catalysis in Organic Synthesis (Drauz, K. and Waldmann, H., eds.), VCH-Verlag, Weinheim, Germany, 409-431. 2. Syldatk, C., Mtiller, R., Pietzsch, M., and Wagner, F. (1992), in Biocatalytic Production of Amino Acids & Derivatives (Rozzell, D. and Wagner, F., eds.), Hanser Publishers, Munich, Germany, 131-176. 3. Takahashi, S., Ohashi, T., Kii, Y., Kumagai, H., and Yamada, H. (1979), J. Ferment. Techol., 57, 328-332. 4. May, O., Siemann, M., Pietzsch, M., Kiess, M., Mattes, R., and Syldatk, C. (1998), J. Biotechnol., accepted. 5. Verhoeven, M., Cahalan, P., Cahalan, L., Hendriks, M., and Foache, B. (1994), European Patent Application, EP 0 608 095 A1. 6. Syldatk, C., Mackowiak, V., H~ike, H., Gross, C., Dombach, G., and Wagner, F. (1990), J. Biotechnol., 14, 345-362. 7. Bunge, F., Pietzsch, M., Miiller, R., and Syldatk, C. (1992), Chemical Engineering Science, 47, 225-232. 8. Bradford, M. M. (1976), Anal. Biochem., 72, 248-254. 9. Tischer, W. (1995), in Enzyme Catalysis in Organic Synthesis (Drauz, K. and Waldmann, H., exls.), VCH-Verlag, Weinheim, Germany, 73-87. 10. Ruttloff, H. (1994)Industrielle Enzyme, Behr's Verlag, Hamburg, Germany.


523

T h e r m o i n a c t i v a t i o n o f 13-xylosidase i m m o b i l i z e d on n y l o n M. J. Duefias and P. Estrada Departamento de Bioquimica y Biologia Molecular I. Facultad de Biologia. Universidad Complutense. 28040 Madrid. Spain.

ABSTRACT fl-xylosidase was isolated and partially purified from Trichoderma reesei QM 9414 grown on wheat straw as the sole carbon source. The enzyme was attached covalently to nylon powder and thermoinactivation was carried out at temperatures between 55 and 75* C. The activation energy for the decay was 114,36 kJ/mol for the immobilized and 193,11 kJ/mol for the free enzyme. These high values are consistent with a eonformational process as mainly responsible for thermal inactivation of 13-xylosidase. Thermostability of 13-xylosidase was enhanced by the addition of ligands such as the substrate and product of the reaction. Polyols increased also the half-life of the enzyme and xylitol was the best protector although it causes inhibition of enzyme activity.

1. INTRODUCTION The synergistic action of xylanase ( EC 3.2.1.8) and 13-xylosidase ( EC 3.2.1.37) accomplishes the breakdown of xylan. 13-Xylosidase takes account of the breakdown of xylobiose to xylose. Trichoderma sp. growing on lignocellulosics produce extracellular cellulases and hemicellulases of industrial interest. Immobilization of 13-xylosidase is necessary in order to reuse it. Moreover, operational stability is advisable since high temperatures are desirable, thus a major goal is to achieve thermal stabilization of etm3anes. Several methods have been tested to improve the thermostability of enzymes, including immobilization and chemical modification of the biocatalyst and the addition of ligands or polyols (1,2). During several years we have grown Trichoderma reesei QM 9414 on wheat straw as the sole carbon source (3) and purified and characterized cellulases and hemicellulases (4,5). We have immobilized 13-xylosidase on nylon powder by generating amino groups in nylon, which are activated then by glutaraldehyde. The aim of our work is to study the thermoinactivation of the immobilized enzyme as well as the effect of ligands and polyols on it and the operational stability.

2. MATERIALS AND METHODS DEAE-Sepharose CL-6B was from Pharmacia. Ultrogel AcA 44 from LKB. p-Nitrophenyl [3D-xylopyranoside, ethylene glycol, xylitol, sorbitol and erythritol from Sigma. p-Nitrophenol, citric acid and glycerol from Fluka. Nylon (polyamide 11), D-xylose and glutaraldehyde from Merck.

524 All the experiments were carried triplicate and the mean value is reported throughout. [3Xylosidase was partially purified from cultures of Trichoderma reesei QM 9414 grown on wheat straw (4,5) by precipitation from culture supematants with (NI-L)2SO4 at 4 ~ followed by cromatographies on DEAE-Sepharose CL-6B and on Ultrogel AcA 44. SDS-PAGE of the pool collected after the last purification step showed that the enzyme was not purify to homogeneity (5) and this pool was used for immobilization. The protein content was 0.76 mg/mL and the 13xylosidase activity was 152 nmol/min/mg. The immobilization of the enzyme was carried out as follows: Nylon powder was incubated with 3.6 N HC1 for 60 min, vacuum-filtered and washed until neutrality. The precipitate was incubated with 5 % glum'aldehyde in 0.1 M citrate buffer pH 4 for 60 min. After vacuum-filtering and several washings, the activated support was incubated with [3-xylosidase at a ratio 2 mg protcin/g nylon for 60 min at 20 oC, centrifuged at 3000 rpm and washed twice. Both supematants were analyzed for protein and activity and the pellet dried in vacuo (4.4 % water, w/w, determined by Karl Fischer titration) was the immobilized derivative. The protein content in solution was determined according with Lowry et al. (6) with bovine serum albumin as standard and the protein immobilized was calculated as the difference between the protein employed in the immobilization process and the remaining free in supernatants. Standard assay for free 13-xylosidase was carried out by incubating 1 mL of 2 mM pnitrophenyl [3-D-xylopyranoside (pNPX) in 0.1 M sodium citrate buffer pH 4 with 0.9 mL of the same buffer and 0.1 mL of the enzyme (76 p.g protein) at 50 oC for 10 min in a water-bath. The reaction was halted by addition of 3 mL 0.2 M NaOH and the absorbance of the p-nitrophenol (pNP) released was recorded at 420 nm in a Beckman DU-70 spectrophotometer. The standard assay for immobilized [~-xylosidase was carried out by incubating 0.1 g of the immobilized derivative with 1 mL of 2 mM pNPX in 0.1 M citrate buffer pH 4 and 1 mL of the same buffer at 50 oC for 15 min. The reaction was stopped as described for the free enzyme but centrifugation at 3000 rpm was required before the absorbance at 420 nm was recorded. The thermoinactivation was carried out by preincubating 76 }~gof free enzyme in 0.1 M citrate buffer pH 4, at 60-75 ~ or 0.1 g immobilized derivative (168 ~g protein) at 55-75 ~ Triplicate samples were taken periodically and cooled in ice. Enzymatic assay was carded out in standard conditions after 2 mM substrate addition. Protection against inactivation was carried out at 60~ C with free or 55~ C with immobilized enzyme by adding to the preincubating mixture either 3 M xylose, 2 M polyol or 2 mM pNPX. Samples of enzyme were cooled in ice and the assay started at 50~ after the addition of 2 mM pNPX. When ligand was pNPX, control was carried out to check enzymatic activity during the preincubation.


3.1. [~-Xylosidase immobilization The immobilized derivative showed 75-95 % of protein and 60 % of [~-xylosidase retention. The polymer-supported enzyme has the same optimum pH (4.0) but decreases its optimum temperature from 55 to 50 oC (Fig. 1) with regard to free enzyme. 3.2. Effect of temperature on activity Figure 1 shows the Arrhenius plot of the experimental data when the enzymatic activity vs temperature is studied, i.e., the reaction rate (v, in gmol/min/mg) and the deactivation rate

525

2

> m 1

103/ T (*K1)

dependence with increasing temperatures. Straight lines are obtained for both enzyme forms but with different slopes. The decrease of the slope in the reaction rate with the immobilized with regard to the free enzyme (up to the optimum temperature), can be due to diffusion control in the reaction with the immobilized enzyme. A similar decrease in the slope is shown at temperatures higher than the optimum pointing to the difficultty the protein has to unfold due to attachent to nylon in the presence of the substrate. If this is true, and since the immobilized enzyme is less stable to temperature than the free one when they are incubated in absence of substrate (Fig. 2), then substrate must stabilize the immobilized form.

Figure 1. Arrhenius plot of free (O) and immobilized (O) [3-xylosidase The activation energies (EA) w e r e obtained from data in Figure 1, being 59.2 kJ/mol for free enzyme, which is in the range of values reported previously: 50 kJ/mol (7). Moreover, the EA of the immobilized enzyme (25,8 kJ/mol) is almost half of the EA for free enzyme. Inactivation energies (Ei) calculated at temperatures higher than the optimum were: 116,4 kJ/mol for free and 73,2 kJ/mol for immobilized enzyme, since at these temperatures, points in Figure 1 are not true activity but the result of combined reaction plus inactivation during assay time. 3.3. T h e r m o i n a c t i v a t i o n

studies

Inactivation of free and immobilized enzyme was followed at 60, 65, 70 and 750C. Plotting residual activity against preincubation time 3 showed first order kinetics with correlation coefficients higher than 0.98 which points to the o 9 2 inactivation of either free or immobilized enzyme by a monomolecular process (1). The 1 polymer-supported enzyme is inactivated in a _= shorter time than the free enzyme at any 0 temperature tested since immobilization causes a decrease in thermal stability when the enzyme -1 is preincubated in absence of the substrate. The half-lives of either free or immobilized enzyme 2.~ 2.~2 ~.~ 3.~, 3.~, were calculated according with the exponential lit x 10 3 decay model and the dependence of the enzyme Figure 2 Temperature dependence of half-life half-life with temperature is depicted in Figure 9 2. of free (O) and immobilized (It) enzyme.

526 The fitting of the data by linear regression to an Arrhenius-type equation, gave an activation energy for the decay (Ed) for free and immobilized enzyme of 193 kJ/mol and 114.4 kJ/mol respectively. These are high values and agree with the existence of protein unfolding followed by refolding into a new thermodynamically stable but catalytically inactive conformation (1). Although we found first-order kinetics in the thermoinactivation of the free enzyme, suggesting that protein aggregation was not rate limiting, we checked it by preincubating at 60 *C several protein concentrations 90.019, 0.038, 0.076 and 0.152 mg/ml. Since the slopes in the logarithmic plot were about the same and the mean half-life (6.98 min) has an standard deviation of+ 0.9, we can conclude that aggregation does not take place as was expected since we did not observe protein precipitation. Relative to immobilized enzyme, we did not check inactivation at several protein concentrations because of the hindering of polymolecular processes such as agreggation through immobilization.

3.4. Protection against thermoinactivation 3.4.1. With iigands If unfolding is a critical step in the thermoinactivation of 13-xylosidase the addition of ligands must have an stabilizing effect by shifting the equilibrium between the native and the thermounfolded form of the enzyme towards the former. In order to check this assesment we used the substrate (2mM pNPX) or the product of the reaction (3 M D-xylose) as protectants. Table 1 Effect of ligands (pNPX and Xylose) on the thermoinactivation constants of 13-xylosidase Ligand

Immobilized Enzyme (55 ~ C)

Free Enzyme (60~ C)

kd (min-~) 95.6•

tla (min)

P.E.

kd (min~) 85.8•

t~a (min) .3

P.E

.3

6.98

4.75

+ pNPX

58• 10"3

14.24

2.0

7.3• 103

92.87

19.5

+ Xylose

2x 103

344

49

15.2• 103

48.68

10.2

tla is the half-life of the enzyme and kd is the decay constant in the exponential decay model. P.E (protective effect) is the ratio of half-life obtained with and without ligand. Table l summarizes the half-lives obtained with and without ligands and we see that free enzyme is better protected by xylose (it increases about 50times the half-life) than by substrate (about twice). However, the immobilized enzyme is best protected by pNPX and this protection is far higher (about 20times) than that of the free enzyme. The mechanism of pNPX protection is clear since it is the substrate of the reaction. The mechanism of xylose protection is not so well stablished since, as product, it is known to be a competitive inhibitor, but also it can act as an acceptor for the xylosyl residue in a transfer reaction (7). It is thougth that the hydroxyls of sugars

527 and polyols accept xylosyl residues better than water, increasing the rate at which the enzyme is regenerated and neutralizing the inhibitor effect of the sugar as a competitive inhibitor.

3.4.2.With polyols Additives other than ligands were also studied as protective agents with free and immobilized enzyme. First we tested several polyols from two to six hydroxyl groups or carbon atoms (Cn) such as ethylene glycol (C2), glycerol (C3), erythrytol (C4), xylytol (C5) and sorbitol (C6) at concentrations up to 2 M in enzymatic assays to check their effect on activity and we found that only sorbitol has an activating effect on immobilized enzyme while it inhibits free enzyme at concentrations above 0.5 M whereas all other polyol tested inhibit both enzyme forms to a degree that dependeded on the concentrationand the nature of the polyol, being erythrytol the strongest inhibitor followed by ethylene glycol (data not shown). Transfer reactions to polyols have been previously described for 13-xylosidase (7) with the consequence of increasing the enzyme activity at 3-10% (v/v) polyol, higher concentrations leading to enzyme inhibition. We employed 2M polyol in the preincubation of soluble and nylon attached 13-xylosidase at 60 and 55~ respectively being 1M final concentration (inhibitory with all polyol except sorbitol) when the E enzymatic assay was carried out at 50~ atter cooling the preincubated so samples. Results are depicted in o Figure 3 and we see that polyols with more than three carbons have 4o a noteworthy protective effect (the 3o ratio of half-life with polyol to the half-life without it) which is higher 2o for free than for the immobilized lO enzyme. This result was expected o since additives interacting more 0 1 2 3 4 5 6 strongly with water than with the Cn enzyme will favour an increase in the degree of water organization. Figure 3. Effect of polyols on the half-life of free (O) and immobilized ( 9 enzyme. A

v

Polyols stabilize proteins by forming strong hydrogen bonds with water and reducing the water activity of the medium and causing a reduction of "free" water (8). Our results (Fig. 3) indicate that xylitol is the best protector for both enzyme forms and it was not expected since it has been reported that the depressing effect of polyols on water activity increases with the carbon number (9). So, we must look for an additional effect ofxylitol, perhaps on the enzyme itself, in order to explain its behaviour. With regard to this point, we must consider that xylitol is structurally related to the reaction product, xylose, and could act as a competitive inhibitor stabilizing the enzyme.

528 3.5. Operational stability of immobilized [~-xylosidase in batch. The operational stability was checked by reusing six samples. After each enzymatic assay, the samples were centrifuged and the pellets washed twice with 9 mL of buffer to eliminate the reaction product. Each day, 5 or 6 assays were carried out but the days were not always consecutive (in the meanwhile, pellets were kept at 4 ~ C). Results in Fig. 4 are percent activity over the first assay. It can be seen that reusing the enzyme the first day has an effect of increasing activity. However, this tendence is not mantained the following days, since activity decreases gradually with regard to the first assay of the same day and thermoinactivation may account for it. Another striking effect is the increase in activity from one day last assay and next day first assay, just as if the immobilized enzyme had a "rest". After 34 reuses, the nylon-13-xylosidase only retains 15 % of the initial activity. Figure 4. Residual activity of immobilized 13-xylosidase during operation in batch. REFERENCES 1. Klibanov A. M., In Advances in Appl. Mierobiol., ed. A.I. Laskin, Academic Press, 29 (1983). 2. Graber M. and Combes D.,Enzyme Mierob. Teehnol., 11 (1989) 673-677. 3. Acebal C., Castillon P., Estrada P, Mata I., Costa E., Aguado J., Romero D. and Jimenez J. Appl. Mierobiol. Biotechnol. 24 (1986) 218-223. 4. Estrada P., Mata I., Dominguez J.M., Castillon M.P. and Aeebal C., Biochim. Biophys. Acta 1033 (1990) 298-304. 5. Mata I., Dominguez J.M., Macarron R., Castillon M.P. and Estrada P., In Biomass for Energy and Industry, ed. by Grassi, Elsevier., 2 (1990) 2283-2287. 6. Lowry O.H., Rosebroug N.J., Farr A.L. and Randall R.T., J. Biol. Chem., 193 (1951) 256-265. 7. Oguntimein G.B. and Reilly P.J., Biotechnol. and Bioeng., 22 (1980) 1143-1154. 8. Graber M. and Combes D., Enzyme Microb. Technol., 11 (1989) 673-677. 9. Lozano P., Carlo J., Iborra J.R. and Manjon A.,Enzyme Microb. Technol., 15 (1993) 868-873. Aknowlegdments: This work was supported by a grant of the CICYT (AMB95-0689-C02-01). We thank G. Orellana for helpful discussion and comments.


529

Increase of thermal stability of tannin acyl hydrolase by covalent immobilization through its carbohydrate side chain : application in tea cream solubilisation P. Nicolas, J.L. Sauvageat, S. Reymond and E. Raetz* Nestl~ Research Center, P.O. Box 44, 1000 Lausanne 26, Switzerland

Cooling black tea extracts causes the precipitation of "tea cream", a complex between caffeine and tea polyphenols, which can be solubilised by tannase (tannin acyl hydrolase EC 3.1.1.20). Immobilization of tannase through its carbohydrate moiety on oxirane acrylic beads was found to increase by 20~ the thermal stability of the enzyme. This made possible the treatment of black tea extracts prior to precipitation of tea cream to avoid inherent mass transfer problems in solid-solid reactions. Excellent operational stability of the covalently bound enzyme was demonstrated in multicycle stirred-batch reactor operation using methyl gallate as substrate. The level of hydrolysis in black tea extracts can be kept constant by progressive increase of the reaction time in automated stirred t a n k reactors. Analysis of progress curves indicated that the reaction obeys a first-order rate under usual operating conditions. 1. I N T R O D U C T I O N The presence of "tea cream", a cold water insoluble precipitate that forms naturally in black tea extracts when they are allowed to stand at temperatures below 40~ [1], is a major problem in the manufacture of instant tea for the preparation of iced beverages. Tea cream is a complex between caffeine and the tea polyphenols, theaflavins and thearubigins, which can be solubilised by tannase (tannin acyl hydrolase EC 3.1.1.20). The enzyme specifically hydrolyses *Present address: Laboratoire Cantonal VD, Les Croisettes, CH-1066 Epalinges.

530

esterified gallic acid present in tea polyphenols under mild conditions preserving the aroma and flavour of the original liquor. Available technical tannases are mainly isolated from Aspergillus sp. and consist of glycoproteins, which are inactivated above 40~ a temperature too low to maintain tea cream in a soluble form. Although immobilization on solid carriers is advantageous for industrial use, the low thermal resistance of the enzyme represents a serious drawback since it is difficult to efficiently react an insoluble enzyme with an insoluble substrate. In this paper we show that thermal stability of tannase is increased by 20~ by immobilization to oxirane acrylic beads through the carbohydrate side chain [2]. The glycosyl moiety was oxidised to provide aldehyde groups and the carrier was modified with bifunctional agents. In order to evaluate process feasibility, the kinetics and operational stability of the immobilized biocatalyst were investigated. 2. MATERIALS AND METHODS Eupergit C| ( RShm GmbH, Germany) was used as carrier in all experiments. Tannase 50000 from A. oryzae was purchased from Kikkoman (Japan). Black tea (Darjeeling FOP) was extracted in water (1/15) at 95~ for 15 min. Gallic acid was determined using HPLC (column, KS 250/6/4 Nucleosil 100-5).

i=............ t !repulses" Ii

i [ pHmeter,J"i ===NaOH~ I

]

i

1

'

I I I

1 1 II I I

i I I

I

Feed

Delivery burette

(~ Draining

Figure 1. Automated reactors used for evaluation of operational stability. Operational stability was measured using automated batch reactors fitted with a stainless steel sieve at the bottom (Figure 1). They can ensure either a constant reaction time (CRT) or a constant hydrolysis level (CHL).

531

3. RESULTS 3.1. Immobilization The immobilization of tannase via amino groups on porous glass has been previously described [3]. However, our attempts to immobilize this enzyme through amino groups on Eupergit C according to the usual procedure failed. Therefore, the glycosyl moiety of tannase was oxidised to generate carbonyl groups able to bind to the matrix using bifunctional agents. A similar principle has been described for the immobilization of antibodies on Eupergit C [4]. Tannase was oxidised with 0.1 M sodium periodate at 0~ and pH 5.5 for 60 rain. The activity of tannase remained unchanged after oxidation. Eupergit C was modified using three different agents: adipic dihydrazide, ethylene diamine and hexamethylene diamine. Coupling of oxidized tannase to hexamethylenemodified carrier resulted in very poor yield. Satisfactory load was obtained using Eupergit C modified with either adipic dihydrazide or ethylene diamine. Owing to its toxicity adipic dihydrazide coupling was abandoned. All the results presented here were obtained with tannase immobilized on ethylene diaminemodified Eupergit C. This immobilization was very reproducible yielding a stable biocatalyst containing about 30 mg protein per gram of dry carrier. As the enzyme is probably bound to the matrix by multipoint attachment, further stabilisation by reduction of the Schiffs base was found unnecessary. 3.2. T h e r m a l stability The thermal stability of immobilized tannase was compared with that of free (soluble) tannase by measuring residual activities after incubation of both preparations in 50 mM sodium acetate buffer pH 5.0 for 30 min. Figure 2 shows that the immobilization process substantially increased the thermal stability of the enzyme.

100

c ~

80

> ~

40

~

30

">

20 ,

,,.

,.

0

|

"

0.5

.

=

1

1.5

2

ammonium sulphate concentration (M)

Figure 2: Relative activity of glucose oxidase immobilised on CIM epoxy disk vs. concentration of (NH4)~SO4 in the coupling buffer To evaluate enzyme characteristics, long-term stability, linear response and reproducibilitywere investigated.For the latterwe performed 100 consecutive injectionson a single disk using a FIA system. Highly reproducible results were obtained with the relative standard deviation of less than 1.6% (Figure 3). This is very important when the C I M disk with immobilised enzyme is to be used as a biosensor.

82

72

52

42

I

0

20

I

4O

I

6O

I

80

No. of injections Figure 3: Reproducibility of the response vs. number of injections

IO0

545 To verify the linear response of the CIM disk with immobilised glucose oxidase we performed activity measurements on a FIA system with different glucose concentrations (Figure 4). The results show linear response in the range between 10 and 100 mg/1 of glucose. Finally, long-term stability was checked (Figure 5). We found that the activity slightly decreased with time. Moreover, although the slope of calibration curve slightly decreased, the linear range of response remained practically unchanged. We concluded, that CIM disk can be used as a suitable matrix for glucose oxidase immobilisation and further application as a biosensor.

100 y = 0.8783x + 0.1278

80 rc-

"--

.

60

O

.~ 40 20

0

l

!

I

20

40

60

'

d

I

80

100

120

glucose concentration (mg/I) Figure 4: Calibration curve for glucose oxidase immobilised on CIM epoxy disk

80 60 .~ tO (~.

v

40-

20-

I

0

5

....

I

10

"

I

15

20

time (days) Figure 5" Stability of glucose oxidase immobilised on CIM epoxy disk (1 M (NH4)2SO4 was used for imrnobilisation)

546 REFERENCES 1. 2. 3. 4.

R.W. Min, Ph.D. Thesis, The Technical University of Denmark, Lyngby, 1995 A. ~trancar, Ph.D. Thesis, BIA d.o.o, Ljubljana, 1997 A. Strancar, P. Koselj, H. Schwirm and Dj. Josi6, Anal. Chem., 68 (1996) 3483. Dj. Josid, H. Schwinn, A. ~trancar, A. Podgornik, M. Barut, Y-P. Lira and M. Vodopivec, J. Chromatogr. A, (1998), in press 5. O.H. Lowry, N.J. Rosebrough, A.L. Farr, R.J:Randall, J.Biol.Chem. 193, (1958) 265. 6. P.K. Smith, R.I. Krohn, G.T. Hermanson, A.K. Mallia, F.H. Gartner, M.D. Provenzano, E.K Fujimoto, N.M. Goeke, B.J. Olson, D.C. Klenk, Anal. Biochem., 150 (1985) 76. 7. T.M. Stich, Anal. Bioehem. 191 (1990) 343. 8. Affinity chromatography: Principles and methods, (1993), Pharmacia, Uppsala 9. D. Keilin, E.F. Hartree, Biochem. J., 42 (1948) 221. 10. J.B. Wheatley, D.E. Schrnidt, J. Chromatogr., 644 (1993) 11.


547

Immobilization o f spinach leaf hydroperoxide lyase L. M. Simona, Sz. J. Mhrczyb, M. Kotormhn", Sz. A. N6methb and B. Szajhni" "Departmem of Biochemistry, J6zsef Attila University, P.O. Box 533,6701 Szeged, Hungary bgesearch Institute of Chemical Process Engineering, Pannon University, P.O. Box 125, 8201 Veszpr6m, Hungary

Spinach leaf hydroperoxide lyase (I~LS) was immobilized on polyethylene terephthalate (Sorsilen), silica (Siloehrome) and polyacrylamide (Akrilex) supports. The highest immobilized activity (360 mU g-i dry gel) was achieved with the HPLS bound to Akrilex C. The Akrilex- and Siloehrome-bound enzymes were studied with respect to their thermal, pH and operational stabilities relative to those of the soluble HPLS. The Akrilex-bound enzyme had the highest stability. 1. INTRODUCTION In plants, the fatty acid hydroperoxides are produced from polyunsaturated fatty acids such as linoleie or linolenie acids by lipoxygenase catalysed oxygenation. Two major pathways are known for the metabolism of fatty acid hydroperoxides: in one, hydroperoxide lyase (HPLS) cleaves 13-hydroperoxylinoleie acid into 12-oxododecenoic acid and hexanal; in the other, hydroperoxide dehydrase catalyses the formation of allene dioxide, a key intermediate in the biosynthesis ofjasmonic acid, which has growth-regulating properties. There are two forms of HPLS, a membrane-bound form in higher plants and a soluble form in microalgae [1,2]. The oxocarboxylic acids obtained from hydroperoxides can be further oxidized to dicarboxylic acids, which can be used to produce a synthetic polymer similar to nylon or further derivatized [3]. To explore these possibilities, a stable form of HPLS is needed. The stability of solubilized HPLS is relatively poor: the half life of soybean leaf chloroplast HPLS stored on ice is 1.5 day [4]. The aim of the present work was to stabilize spinach leaf HPLS by immobilization and to characterize the stabilities of the immobilized forms. 2. MATERIALS AND METHODS 2.1. Chemicals

Linoleic acid, Tween 20 and 1-cyclohexyl-3-(2-morpholinoethyl)carbodiimide-p-toluene This workwas supportedby OTKAgrantT 023294.

548 sulphonate were purchased from Sigma Chemical Co. (St Louis, Mo, USA). Akrilex C-100, a polyacrylamide bead (100-320 l~m) polymer containing carboxylic functional groups (6.4 meq.g'l), and Akrilex P-100 activated with p-benzoquinone, were obtained from Reanal Factory of Laboratory Chemicals (Budapest, Hungary). Silochrome aldehyde and Silochrome activated with p-benzoquinone were obtained from NPO Biolar (Riga-Olaine, Latvia). Other chemicals were commercial, reagent grade products of Reanal. 2.2. Plant material I-IPLS was isolated from the leaves of spinach (Spinacia oleracea L.) according to Gardner et al. [2]. For solubilization of the enzyme, Tween 20 was used in 0.5 % concentration. 13-Hydroperoxylinoleic acid as substrate was prepared by reacting linoleic acid solution with soybean lipoxygenase in 0.5 M borate buffer, pH 9.0 [5]. The concentration of the hydroperoxide was determined as described by Ames and King [6]. 2.3. Immobilization of HPLS The adsorptive binding of the enzyme to a polyethylene terephthalate support (Sorsilen) was carried out as described earlier [7]. Silochrome and Akrilex P-100 preactivatext with pbenzoquinone, a polyacrylamide derivative possessing carboxylic functional groups (Akrilex C 100) activated with water-soluble carbodiimide, a silica derivative (Silochrome aldehyde) were also used for covalent immobilization of the HPLS [8]. 2.4. Measurement of protein Protein determinations were performed according to Lowry et al. [9]. The amount of bound protein was determined indirectly from the difference between the amount of protein introduced into the coupling reaction mixture and the amount of protein present in the supernatant and washing solutions after immobilization. 2.5. Assay of HPLS activity The HPLS activity was measured at 20 ~ by following the disappearance of the conjugated diene at 234 nm [10]. The assay mixture consisted of 0.1 M borate buffer (pH 9.0; final volume 3 ml) comaining 65 I~M hydroperoxide. The reaction was initiated by the addition of 5 units of HPLS. One unit of enzyme activity was defined as the amount of enzyme that catalysed the cleavage of one l~mol of hydroperoxide per min at pH 9 at 20 ~ The activities of the immobilized enzymes were measured with 100-200 !~1 of immobilized enzyme in the same reaction mixture as was used for the soluble enzyme. The reaction mixture was stirred for 10 min at 20 ~ the immobilized HPLS was filtered off quickly (a few minutes) and the amount of linoleic acid hydroperoxide was determined as for the soluble HPLS. 2.6. Stability tests Thermal inactivation experiments (40-55 ~ were carded out in 0.1 M phosphate buffer (pH 7) containing 1 mM dithiothreitol (DTT) and 1 mM EDTA. In the pH stability tests, 0.1 M phosphate (pH 6, 7 and 8) and Tds~CI (pH 9) buffers were used, containing 1 mM EDTA and 1 mM dithiothreitol (DTT). The operational stabilities of the immobilized enzymes were measured in 0.1 M Tris/HCl (pH 8).

549 3. RESULTS Different types of supports and coupling methods were used for the immobilization of HPLS. The results relating to immobilization and activity are listed in Table 1. Most protein was bound to Akrilex P- 100, but in respect of immobilized activity the Akrilex C- 100 proved to be the best. Both Siloehrome supports bound about 10% of the enzyme activity originally introduced for immobilization, but they had the highest activities per unit volume. The adsorption of HPLS on Sorsilen resulted in a catalytically active, immobilized enzyme. As concerns storage stability, HPLS bound to Akrilex C-100 and to Siloehrome aldehyde proved the most advantageous. In the further studies reported in this paper, these immobilized HPLS forms were characterized.

Table 1 Balance sheet for the immobilization ofHPLS on different supports HPLS used for immobilization Support

Immobilized

Quantity Total activity Protein Activity (nag g'ldry gel) (mU gldry gel) (mU ml"1) (nag) (mU)

Sorsilen

497

2920

24.3

12

2.3

Akrilex P- 100 (activated with p-benzoquinone)

994

5840

188.0

87

5.8

Akrilex C- 100 (activated with water-soluble carbodiimide)

994

5840

151.7

360

4.0

Silochrome aldehyde

99.4

584

19.0

51.7

28.7

Silochrome (activated with p-benzoquinone)

99.4

584

33.7

59.8

21.4

3.1. Thermal stability The rates of thermal inactivation of the soluble and Siloehrome-bound and Akrilex Cbound enzymes were studied in the temperature range between 40 and 55 ~ at pH 7.0 in 0.1 M phosphate buffer containing 1 mM EDTA and 1 mM DTT. At 50 ~ the soluble enzyme 10st 90% of its activity during a 1-hour incubation, whereas the Akrilex-HPLS lost only 10% and the Siloehrome-I-IPLS about 40% of the initial activity (Figure 1)

550 120 o --o,100 "~ 80 .--,,,

r16260

,=,=.

-o ..=,

40

o

a:: 20 0

!

!

!

50

100

150

200

Time (min)

120 . . . . . . . . . .

120 I

glO0

_~~

o

m

.~_ 40 ~ ID

c

~100 ~

.

i1:20

#. 2o

0 0

50

100

150

Time (rain)

200

0 250

0

,

,

=

|

50

100

150

200

=

250

Time (min)

Figure 1. Thermal inactivation of soluble (a) and immobilized (Silochrome (b) and Akrilex C (c)) I-IPLS at 40 ~ (A), 45 ~ (o), 50 ~ (El), 55 ~ (x). Experiments were carried out in 0.1 M phosphate buffer pH 7.0.

3.2. pH stability The stabilities of the soluble and immobilized enzymes were compared in 0.1 M phosphate buffer pH 6-8 and in Tris/HCl buffer pH 8, 9 at 50 ~ The residual activities were measured with the standard method. Figure 2 presents the progress curves of the inactivation at the different pH values. The soluble HPLS and the Silochrome-HPLS displayed the highest stability at pH 6. The inactivation of the soluble HPLS and the Silochrome-HPLS was fast at pH 9. For the Akrilex'HPLS, there were no marked changes in the actvity in the first hour of incubation, at the different pH values, but after a longer incubation the highest residual activity was observed at pH 9.

3.3. Operational stability The reusability of the immobilized HPLS forms was studied at pH 9 in Tris/HCl (Table 2). The Akrilex-enzyme exhibited the higher stability. After 5 cycles the Silochrome-HPLS had lost about 70% and the Akrilex-HPLS about 36% of its activity.

551 120 o'-7100 .~

&

~ 8o"6 r

60 40

(!)

n, 2O 0 0

I

I

I

20

40 Time (min)

60

80

120

120

o"9-100

o'-e,100"~ .>, .~ 8 0 -

80 6o

60

~9 40

~:

n,. 20

"1 o

'

0

i

!

50

100

20

0 150

0

Time (rain)

!

!

50

100

150

Time (min)

Figure 2. pH stability of soluble (a) and immobilized (Silochrome (b) and Akrilex C (c)) I-IPLS at pH 6 (A), pH 7 (o), pH 8 ([3) and pH 9 (x). Experiments were carried out in 0.2 M phosphate (pH 6-8) and 0.2 M Tfis/l-ICl (pH 9) buffer at 50 ~ For both the soluble and the immobilized enzyme, the activity at zero time was assigned a value of 100 %. Table 2 Operational stability of immobilized HPLS Relative activity % Cycle number* 1

2 3 4 5

Silochrome-HPLS 100 75.6 58.9 38.9 29.4

* The duration of the cycles was 10 rain each.

Akrilex-HPLS 100 84.6 72.7 66.6 63.8

552 4. DISCUSSION Hydroperoxide lyase is a membrane-bound enzyme in spinach leaf chloroplasts. In the natural microcnvironment it is stabilized in a lipoprotein complex. For practical purposes, HPLS is needed in an isolated form. During the isolation process, the interactions and secondary bondings stabilizing the enzyme structure are destroyed. The HPLS therefore becomes unstable which is disadvantageous from a technological point of view. One possible way to stabilize enzymes is immobilization. Adsorption onto polymers such as polyethylene terephthalate (Sorsilen) is a simple and rapid means of immobilization, but the adsorption complex is sensitive to changes in temperature, ionic strength, pH, etc. Covalent bonding onto solid supports results in more stable immobilized I-IPLS. Silica-based supports (Siloehrome) are advantageous as regards the volume activity, but HPLS immobilized on polyacrylamide matrices has a higher conformational stability, owing to the strong hydrophilie character. This is confirmed by the data concerning the operational stabilities. Our observations are in accord with the results of Nunez et al. [ 11] who immobilized HPLS from Chlorellapyrenoidosa and C. fusca on different supports. The best immobilized HPLS was that on Affi-Gr 501. This matrix-bound enzyme could be used during 5 cycles without loss of activity. 5. CONCLUSIONS Our results suggest that the stability of HPLS isolated from spinach leaves could be improved by immobilization on Silochrome or Akrilex supports. For a practical point of view the Akrilex-bound enzyme possessed the better stability properties. REFERENCES

1. 2. 3. 4. 5.

H.W. Gardner, Biochim. Biophys. Acta, 1084 (1991) 221. E. Bl6e and J. Joyard, Plant Physiol., 110 (1996) 445. D.L. Van Dyme and M.G. Blase, Biotechnol. Prog., 6 (1990) 273. H.W. Gardner, D. Weisleder and R.O. Plattner, Plant Physiol., 97 (1991) 1059. G. Iacazio, G. Langrand, J. Baratti, G. Buono and G. Triantaphylides, J. Org. Chem., 55 (1990) 1690. 6. G.R. Ames and T.A. King, J. Sci. Food Agr., 17 (1966) 301. 7. L.M. Simon, K. Heinriehova, I. Veszelka and B. Szajhni, Acta Bioehim. Biophys. Hung., 25 (1990) 1. 8. L.M. Simon, M. Kotorm~n, B. Szaj~ni and L. Boross, Enzyme Mierob. Technol., 8 (1986) 222. 9. O.H. Lowry, N.J. Rosebrough, A.L. Fm'r and R.J. Randall, J. Biol. Chem., 184 (1951) 313. 10. B.A. Viek, and D.C. Zimmerman, Plant Physiol., 85 (1987) 1073. 11. A. Nunez, G. St. Armand, T.A. Foglia and G.J. Piazza, Biotechnol. Appl. Biochem., 25 (1997) 75.


553

A New Immobilization System For Candida rugosa Lipase: Characterization and Applications K. Carbone and M. Casarci INN-NUMA Dept., C.R. ENEA Casaccia, Via Anguillarese, 301 - 00060 S. Maria di Galeria- Roma (Italy); sp 061

ABSTRACT A commercial crude lipase from Candida rugosa was immobilized onto special hydrophobic polymers synthetized by us (PVA-CL-Cn) and diatomaceus earth (Celite 545) using buffer (20 mM Hepes and 2 mM EDTA, pH 7.2) as reaction medium. PVACL-Cn are polymers made of fatty acids with different carbon chain length, ranging from 4 to 14 carbon atoms. Preoviously, they were employed as column-stationary phases for lipase purification and iso-forms separations. Enzyme preparations were tested in hydrolysis of different tryglicerides. The influence of temperature and chemical nature of supports on enzyme activity and stability were investigated. Results have shown not only a good stability of the biocatalysts (versus time and temperature) but also the efficency of PVA-CL-Cn polymers as lipase immobilization supports (better than Celite 545). 1. INTRODUCTION Lipases (EC 3.1.1.3) are enzymes whose biological function is to catalyze the hydrolysis of triacylglicerols, although some will degrade a fairly broad range of compounds containing an ester linkage13. Among them, lipase from Candida rugosa is a good candidate for industrial applications, because it is commercial available, requires no co-factors, it is relatively inexpensive and has a broad substrate specificity. This enzyme is used in oil hydrolysis 4"6 as well as in organic synthesis 7"j~ so it may be important to immobilize it. Recently several reports have dealt with advantages in using immobilized lipases as biocatalysts in organic media 11'12, but we do not forget the importance of using an insoluble biocatalyst in hydrolysis reactions as free enzyme is water-soluble and from an industrial point of view reusing of the protein is essential. A wide range of supports has been used for CRL i m m o b i l i z a t i o n 13'14. In this study we have tried to adsorbe this microbial lipase onto hydrophobic polymers synthetized by us, PVA-CL-Cn 15. We have compared the efficency of these polymers with Celite 545 as supports for the studied enzyme. Moreover, the influence of temperature on prepared biocatalysts stability was investigated.

554 2. MATERIALS AND METHODS 2.1. Protein and other materials A microbial lipase from Candida rugosa (CRL) obtained from Sigma-Aldrich Chemical Company (Gallarate, Milan) was used as test enzyme (lot. 12807 EQ), to produce immobilized lipase preparations. The crude enzyme powder had a nominal specific lipolytic activity of 1438 units/mg, based on olive oil hydrolysis and 30% lactose as extender. Celite 545 was purchased from BDH Laboratory Supplies Poole (England), while polyvinyl alcohol (PVA) was a commercial product of Merck (Darmstadt, Germany), used without further purification (MW 72000, saponification degree > 98%). All other chemicals were reagent grade. 2.2 Lipase immobilization Candida rugosa lipase was immobilized on Celite 545 and PVA-CL-Cn polymers according to Mustranta et al. ~7, with some modifications. PVA-CL-Cn matrixes are made according to Battinelli et al. 15, using linear fatty acids. First, poly(vinyl) alcohol has been cross-linked with epichlorohydrin in a basic medium (7M NaOH); then the obtained powder (PVA-CL, 1 mole) is esterified adding an excess of the appropriate linear acyl chloride. In a typical immobilization experiment, a 20 mg/ml lipase solution (used buffer: 20 mM Hepes and 2 mM EDTA - pH 7.2) was centrifugated at 3000 g for 30 min in a thermostated centrifuge (15~ After centrifugation, the supematant was carefully removed and used for immobilization without further purification. Lipase solution (30 ml) and the support (5 g dry solid) were mixed together at room temperature for three hours, under magnetic stirring conditions. Then, the suspension was filtrated, lyophilized for 48 h and stored at 4~ until use. 2.3. Hydrolysis reactions Esterase activities of immobilized enzyme toward triacetin and tributyrin were carried out in 20 mM Hepes and 2 mM EDTA buffer (pH 7.2) at 37~ In a typical test, the assay mixture, containing 2.5 ml of HEPES/EDTA pH 7.2 buffer solution (20 mM), 0.5 ml of tributyrin and 100 ~tl of the enzymatic sample, was shaken for 30 s and incubated at 37~ under magnetic stirring. After 30 rain the reaction was stopped with 2.5 ml of ethanol-acetone mixture (1:1). Lipase activities were assayed by using olive oil emulsion as substrate at 37~ and pH 7.2, according to Sigma quality control test procedure. In both cases (esterase and lipolytic assays) the liberated fatty acids were titrated with 0.1 N NaOH solution with an automatic titrator (Metrohom AG 645 multidosimat) using phenolphtalein (timolphtalein for olive oil assay) as an indicator. 3. RESULTS AND DISCUSSION 3.1 Preparations of biocatalysts Candida rugosa lipase was immobilized on various hydrophobic supports by physical adsorption.

555 The highest immobilized activity was achieved using PVA-CL-C12, but it is comparable with PVA-CL-C 10 and PVA-CL-C8. In Fig.1 is shown the effect of these different supports on the specific activity of lipase

Figure 1. CRL activity immobilized on different supports. Tributyrin was used as standard substrate. As we can see, the most suitable supports have a carbon chain lentgh ranging from 8 to 12, as was already observed during the purification process 16. Moreover, PVACL-Cn appear to be better than Celite 545 for lipase adsorption. In further studies, lipase immobilized onto PVA-CL-C8 and PVA-CL-C14 was characterized. We chose these matrixes to compare the influence of polymer hydrophobicity with biocatalyst performances, under a prefixed set of parameters.

3.2. Effect of temperature on activity and stability of the biocatalyst The temperature dependence of the immobilized lipase was investigated in 22 raM Hepes/EDTA buffer (pH 7.2) in the temperature range 30-60~ using olive oil as substrate. Data are shown in Fig.2.

556

70000 60000

8

50000

.,~

40000 30000

9~

20000

~'

10000

CRL/C14

0 3o

40

50

60

70

Temperature(~ Figure 2. Temperature dependence of CRL specific activity immobilized onto PVACL-C8 and PVA-CL-C 14. Olive oil was used as standard substrate. The apparent temperature optimum was about 40~ using PVA-CL-C 14 as support for CRL immobilization, while that for CRL onto PVA-CL-C8 was about 50~ This might be due to a different conformational state of lipase onto different matrixes or probably to the ability of these polymers to discriminate between different lipase isoforms. Thermal stability data were withdrawn in the same buffer solution used to study the temperature's influence on enzyme preparations. The incubation time was l h. Experimental results are shown in Fig. 3. Both enzymatic preparations (CRL/C8 and CRL/C14) showed the same trend as CRL immobilized onto PVA-CL-C8 support appeared to be a little more resistant to thermal deactivation. It's interesting that biocatalysts are active also at high temperatures after incubation period (about 20% of residual activity has found).

3.3. Enzymatic hydrolysis of triglycerides using prepared biocatalysts Released fatty acids by hydrolysis of different triglycerides using Candida rugosa lipase on PVA-CL-C8 and PVA-CL-C 14 are shown in Fig.4. Results pointed out that PVA-CL-C8 was the best support for the enzyme in all studied reactions, expecially when tributyrin was used as substrate. Moreover, prepared biocatalysts showed the highest activity towards triolein

557

Figure 3. Thermal inactivation of different lipase preparations after an incubation time of one hour. Olive oil was used as standard substrate. Residual activities were calculated with respect of not incubated enzyme preparations at the same temperatures.

Figure 4. Hydrolysis of different triglycerides with immobilized CRL. Yaxis were in logaritmic scale due to great differences in specific activity values between triolein and both triacetin and tributyrin.

4. CONCLUSIONS In this paper we present a new class of supports for Candida rugosa lipase immobilization. It's well known that hydrophobic matrixes (polypropylene, latex, nylon, etc.) are suitable supports for microbial (and not only) lipases immobilization. In this study we used some polymers with a different hydrophobic grade (due to the length of the carbon chain used to functionalize cross-linked poly(vynil) alcohol). These matrixes (PVA-CL-Cn), synthetized by us, were previously used with success as column-stationary phases for lipase purification and iso-forms separation. The results show that these polymers are good supports for lipase immobilization by physical adsorption, better than Celite 545. In fact, under the same quantity of immobilized protein, CRL onto Celite 545 shows the lowest specific activity using tributyrin as standard substrate. In this paper we only have reported diatomaceus earth efficiency as a support for CRL immobilization, comparing it with the PVA-CL-Cn ones. Experiments on the stability of this biocatalyst are in progress. CRL immobilized on PVA-CL-C8 gave the best results in terms of hydrolytic activity towards chosen triglycerides. We suppose that the reasons may be or a different kind of protein-polymer interaction which leads to a different

558 enzyme conformational state, i.e. more active in the case of PVA-CL-C8, or to the immobilization of only an iso-form (between the six ones present in the crude), which results more active in the hydrolysis of olive oil. Instead, the highest activity versus triolein as substrate is probably due to a more favourable hydrophobic interactions between substrate and supports. These results are confirmed by the study of the influence of temperature on enzyme activity. In fact, in the case of CRL on PVA-CLC8 the optimum is shifted at higher temperature that for PVA-CL-C14 and free enzyme (37~ Moreover, CRL on PVA-CL-C14 seems to be less affected from temperature showing a quite linear trend. All this was confirmed by thermal tests. In fact, the incubation period has a more pronounced effect on CRL on PVA-CL-C8, especially in the temperature range 40-50~ (the same of maximum activity of the biocatalyst). So we can say that CRL on PVA-CL-C 14 is more stable from a thermal point of view, but less active than CRL on PVA-CL-C8. However, prepared biocatalysts show a good thermal stability, with a residual activity of about 20 % for both preparations after 1 h incubation at 60~ in a buffer solution. In conclusion, to our opinion, the study of lipase immobilization on these supports might give the possibility to clarify protein-polymer interactions and relate them to enzyme (or iso-forms) performances. REFERENCES

1. Macrae, A.C. and Hammond, R.C., Biotec. and Genetic Engin. Reviews, 1985, 3,193-217 2. Kazlauskas, R.J., TIBTECH, 1994, 12, 464-472; 3. Bjorkling, F., Godtfredson, S.E., and Kirk, O., TIBTECH, 1991, 9, 360-363; 4. Albasi, C., and Riba, J.P., J. Chem. Tech. Biotechnol., 1997, 69, 329-336; 5. Hoq, M.M., and Yamane, T., JAOCS, 1985, 62, 1016-1020; 6. Virto, M.D., et al., Enzyme Microb. Techn., 1994, 16, 61-65; 7. Persichetti, A.R. et al., Tetrahed. Letters, 1996, 37, 6507-6509; 8. Tsai, S.W., et al., Biotechn. Prog., 1997, 13, 82-88; 9. Battistel, E., et al., Biotech. Bioeng., 1991, 38, 659-664; 10.Therisod, M., and Klibanov, M., J. Am. Chem. Sot., 1987, 109, 3977-3981; 11.Rosevear, A., et al., I.O.P. Publishing Ltd, Bristol, Philadelphia, 1987; 12.Malcata, F.X., et al., JAOCS, 1990, 67, 890-910; 13.Montero, S., et al., Enzyme Microb. Technol., 1993, 15, 239-247; 14.Carta, G., et al., Biotechnol. Bioeng., 1991, 37, 1004-1009. 15.BattineUi, L., et al., Joum. Chromat. A, 1996, 753, 47-55. 16.Carbone, K., and Casarci, M., ECB8-poster presentation, Budapest, August, 1997. 17.Mustranta, A., et al., Enzyme Microb. Technol., 1993, 15, 133-139.


559

Soybean Lipoxygenases: Purification and Stability of the Free and Immobilized Enzymes Anthony Chikere*, Boris Galunsky, Volker Kasche Technical University of Hamburg-Harburg, Department of Biotechnology II, Denickestrasse 15, 21071 Hamburg, Germany 1. INTRODUCTION Soybean lipoxygenases (EC 1.13.11.12) catalyse the addition of molecular oxygen across the double bonds of polyunsaturated fatty acids forming their corresponding hydroperoxides. The hydroperoxide products have for long been recognised as versatile reaction intermediates in the production of different fine chemicals 1. Lipoxygenases are getting increasingly important in the study of asthma and related allergies. Some lipoxygenase inhibitors such as Zeleuton have been shown to have positive therapeutic effects especially on some allergic and asthmatic conditions 2. The potential biocatalytic applications of lipoxygenase is however limited by its poor stability. Rapid inactivation of the enzyme takes place in both purified and crude forms. We have demonstrated in a previous work that soybean lipoxygenases can be stabilized on CNBr Sepharose 4B (a bead-formed agarose gel) without significant loss in activity3. However, although CNBr Sepharose 4B is stable under a wide range of experimental conditions such as high and low pH, detergents and dissociating agents, it is not the support of choice in biotransformations that require organic solvents, high temperature and high pressure. The Fractogel EMD tentacle supports may be better suited for all of these cases since they combine structural stability, high specific binding capacity and long spacers that greatly reduce stearic interactions. We report in this paper the successful immobilization of two soybean lipoxygenase isoenzymes: lipoxygenase-1 (LOX-1) and lipoxygenase-2 (LOX-2) on Fractogel EMD Azlactone. 2. MATERIALS AND METHODS 2.1. Materials Soybeans and flour were supplied by Edel Soya, Hamburg Germany. Q Sepharose fast flow, Mono S beads and CNBr Sepharose 4B were from Pharmacia Biotech, Freiburg Germany. Eupergit C was from Rrhm, Darmstadt and Fractogel EMD Epoxy and Fractogel EMD Azlactone were from Merck, Darmstadt Germany. All other chemicals and reagents were of analytical grade. 2.2. Purification of soybean iipoxygenase Crude soybean lipoxygenase was extracted from soybean flour in 50 mM potassium phosphate buffer pH 6.5 using a modified Surray methodL LOX-1 and LOX-2 were separated by anion exchange chromatography followed by cation exchange chromatography. LOX-1 and LOX-2 were first separated on Q Sepharose fast flow at pH 8.0 and room temperature. Bound LOX isoenzymes were eluted with a linear gradient (0-50% for 160 minutes) of 0.5 M NaC1. Pooled factions from each active peak were concentrated and further purified on l~:ono *Anthony Chikere is suported by the Deutscher Akademischer Austauschdienst (DAAD)

560 S beads at p H 4.4 with 50 m M citratebuffer. The purity of the two isoforms was established using isoelectric focussing and SDS-PAGE.

2.3. Enzyme assay Initial rates of enzyme activity were measured spectrophotometrically by an increase of absorbance of the reaction product, hydroperoxide at 234 nm (Shimadzu DR. 3, Kyoto, Japan). A l ml reaction mixture contained 900 ~tl buffer, 50 ~tl enzyme solution and 50 ~tl substrate. Substrate was 0.1ram linoleie acid. LOX-1 and LOX-2 were assayed at pH 9.0 (200 raM borate bufer) and pH 6.5 (200 mM potassium phosphate buffer) respectively. Activity measurement of immobilized LOX was done using the method for measuring enzyme activity in heterogeneous systems 5. One unit of LOX activity is defined as the amount of substrate (~tmole linoleic acid) oxidised per mg protein per minute at 25~ at pH 9.0 (LOX-1) or pH 6.5 (LOX-2) respectively. Protein concentration was determined using modified Bradford method and by absorbance measurement at 280 nm. 2.4. Immobilization of soybean iipoxygenase Immobilization of soybean lipoxygenase was performed by covalent linkage to the activated supports at slightly alkaline pH. LOX-1 and LOX-2 were immobilized on CNBr Sepharose 4B on Eupergit C on Fractogel EMD Azlactone and Fractogel EMD Epoxy following the manufacturer's instructions. LOX was coupled to Eupergit C and CNBr Sepharose 4B at pH 7.5 and on the Fractogel tentacle supports at pH 8.0. 250 mg of swollen support was incubated with 3ml of enzyme solution (2 mg protein/ml) in coupling buffer at room temperature for times ranging from 24 hours for CNBr Sepharose 4B to as long as 96 hours for Eupergit C. Non specifically bound proteins were removed by washing with 200 mM ethanolamine. This was followed by alternated washing steps with acidic and basic buffers. Enzyme activity of both free and immobilized enzyme was determined as described in 2.2. Stability of immobilized LOX was determined by comparing the residual activity of the immobilized LOX with the initial activity at time zero (to). Stability of immobilized LOX over time was also compared with that of the free enzyme. 3. RESULTS AND DISCUSSION

3.1. Purification of soybean lipoxygenase Two isoenzymes of soybean lipoxygenase were purified by a combination of anion and cation exchange chromatography. Figure 1 shows the separation of LOX-1 and LOX-2 on Q Sepharose fast flow at pH 8.0 and room temperature. Unbound substances were eluted in the first protein peak while the last protein peak represents other bound proteins lacking LOX activity. LOX-2 was eluted with approximately 150 mM NaCI and LOX-lwith 200 mM NaC1. Peaks 2 and 3 contained two active lipoxygenase isoenzymes which differed by their pH of optimum activity. LOX-1 had a pH optimum at 9.0 while LOX-2 had optimum activivity at pH 6.5. Since protein desorption from an ion exchanger is a function of binding strength, the elution pattern suggests that both isoforms may share very similar binding properties. Purified forms gave single bands on isoelectric focussing gel suggesting that the isoenzymes have been purified to apparent homogeneity. Results from isoelectric focussing (data not shown) and SDS-PAGE (Fig. 2) of the purified isoenzymes confirm this. The isoelectric points of LOX-1 and LOX-2 were determined to be approximately 5.8 and 6.2 and their molecular weights are hardly distinguishable by SDS-PAGE. Molecular weights LOX-1 and LOX-2 determined to be about 94 kDa on a 10% SDS-PAGE are in agreement with published results 6 and with amino acid sequence studies 7. This combination of anion and cation exchange chromatography seems to provide a rapid and efficient procedure for purifying soybean lipoxygenase

561

1.5

300

I

A 8O.m

I ~ L O X - I activity ~'~[ L- -O- X - 2 a c t i v i ~

(I

1

200 ~

0.5

A

~

0 0

20

40

60

80

Elution volume

100

100

' 120

0 140

% 0.5M NaC! 1oo

o

(ml)

Figure 1. Separation of soybean lipoxygenase isoenzymes (LOX-1 and LOX-2) on Q Sepharose fast flow at pH 8.0 and at room temperature. Bound LOX isoenzymes were eluted with a linear gradient (0-50%) of 0.5M NaC1. Pooled factions from each active peak were rechromatographed and further purified on Mono S cation exchanger.

Figure 2. 10% SDS-PAGE of crude LOX and purified LOX-1 and LOX-2. Lanes 1 and 8 - LMW makers, lanes 2 and 9 -crude LOX, Lanes 3 and 4 -anion exchange chromatography fractions of LOX-1 and LOX-2, lanes 5 and 6 - cation exchange chromatography fractions of LOX- 1 and LOX-2.

562

3.2. Stability of free and immobilized lipoxygenase Stability of imrnobilzed LOX was studied at pH 6.5 (LOX-2) and pH 9.0 (LOX-1) for at least 25 days. Figures 3-6 show the stability profiles of LOX-1 and LOX-2 immobilized on different supports compared with the free isoenzymes. Free soybean LOX was much less stable both at room temperature and 4~ LOX-2 immobilized on CNBr Sepharose 4B retained more than 90% of its initial activity after 25 days at room temperature (Figure 3). Free LOX-2 was inactivated after 9 days under similar conditions. Similar results were obtained for LOX-1 immobilized on CNBr Sepharose 4B (data not shown). Immobilized LOX-1 on Fractogel EMD Azlactone showed better stability than immobilized LOX-2, retaining about 90% of its initial activity after 40 days at room temperature. Immobilized LOX-2 retained only 40% of initial activity after 40 days. 100

2000

Immob. LOX-2 (4~ -o-Immob. LOX-2 (Room Temp.) Free LOX-2 (4~ Free LOX-2 (Room Temp.)

.~

1000

w

~

*.. e~

0 0

5

10

1S

20

25

Time (Days) Figure 3. Stability of LOX-2 immobilized on CNBr Sepharose 4B compared with the free enzyme at room temperature and at 4~ Coupling of LOX on Fractogel EMD epoxy and Eupergit C gave poor stability results. LOX-1 residual activity on Fractogel EMD Epoxy dropped to 50% after 25 days and 40% after 40 days. Residual activity of LOX-2 immobilized on Fractogel EMD Epoxy was 30% after 25 days and 18% after 40 days. Coupling of LOX-1 and LOX-2 to Eupergit C resulted in a rapid inactivation of both isoenzymes. Residual activity dropped to zero after 9 days. But on addition of 50-100 ~l 1 mM FeCl3 to the reaction mixture the enzyme was reactivated indicating that the inactivation process is reversible and may be associated with Fe, a cofactor important for LC)X catalysis. LOX activity however, dropped on prolonged incubation with Fe (results not shown). Atomic absorption spectroscopy studies with free and immobilized on Eupergit C-LOX showed that both forms contained 1 mole of Fe per mole of enzyme. The cause of rapid inactivation of Eupergit C-LOX is not very clear. More information may be obtained by determining the state of Fe in the immobilized enzyme. Inactive LOX is known to exist in

563 the Fe 2§ state and is oxidised to the high spin Fe 3§ by one equivalent of hydroperoxide product s. Potato tuber 5-1ipoxygenase was recentl~v reported to be successfully immobilized on oxirane acrylic beads without such inactivation'. Work is still going on in our laboratory to explain the rapid inactivation of soybean LOX observed when immobilized on epoxyactivated supports. 1200

100 -"an

Free LOX-I (4~

- .x-

Free LOX-2 (4~

-9, o - - I m m o b . LOX-I (40C) --

Immob. LOX-2 (4oC)

Addition of

600

so

FeCI 3

o~

0

0

5

10

15

20

Time (Days)

25

Figure 4. Stability of of LOX-1 and LOX-2 immobilized on Eupergit C. Immobilized LOX-1 a n d LOX-2 on Eupergit C showed rapid inactivation. After complete inactivation, the enzyme was reactivated by addition of 1 mM FeC13 during the activity measurement. On prolonged incubation with FeC13 the activity was lost (Data not shown). The arrow shows the jump in activity after FeC13 addition. 100 EMD Azlactone-LOX-I

80

60 EMD Epoxy-LOX-I 40 0

20

0

9

ee LOX-1

0

0

10

20

|

30

40

Time(days) Figure 5. Stability of LOX-1 immobilized on Fractogel EMD Azlactone and EMD Epoxy. Immobilization was done at pH 8.0 in 100 mM phosphate buffer containing 1.5 M NaC1. Stability we~ studied at room temperature for 40 days.

564 100

80

60

EMD Azlactone-LOX-2

40

Epoxy-LOX-2 20 Free LOX-2 0

10

20

30

40

Time(days)

Figure 6. Stability of LOX-2 immobilized on Fraetogel EMD Azlactone and Fractogel EMD Epoxy. LOX-2 was immobilized at pH 8.0 in 100 mM phosphate buffer containing 1.5 M NaC1. CONCLUSIONS Our work demonstrates that besides CNBr Sepharose 4B, soybean lipoxygenase can also be stabilised by immobilization on Fractogel EMD tentacle supports and successfully used as a biocatalyst in this form. Immobilization on these supports especially on EMD Azlactone resulted in a much better stabilization of the biocatalyst. This may be due to removal of proteolytic enzymes during washing and protection of the immobilized enzyme by the matrix. LOX-1 showed an overall better stability on Fractogel EMD tentacle supports than LOX-2. The successful immobilization of LOX on this mechanically stable supports can have some interesting technical implications for biotransformations where mechanical stability of supports is a desirable characteristic. Coupling of LOX to Eupergit C resulted in an apparent reversible inactivation of the enzyme. Atomic absorption spectroscopic studies of both the immobilized and free enzyme showed that there was no loss of Fe due to immobilization. Futher studies are necessary to explain this phenomena REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9.

H.W. Gardner, J. Am. Oils Chem. Soc., 73 (1996)1347-1357. B. Galunsky, R. Schlothauer, B. B6ckle, V. Kasche, Analytical Biochemistry 221 (1994), 213-214. E. Israel, J. Cohn, L. Dube, J.M. Drazen, J. Am. Med. Assoc. 275 (1996) (931-936). A.C. Chikere, B. Galunsky, V. Kasehe, Proceedings of the workshop Biokonversion nachwaehsender Rohstoffe 10 (1997) 68-72) Detmold, Germany. P.K. Surrey, Plant Physiology, 30 (1964) 65-70. J Steezko, G.P. Donoho, J.C. Clemens, J. E.Dixon, B. Axelrod, Biochemistry, 31(1992) 40534057. D. Shibata, J. Steezko, J.E Dixon, M. Hermodson, R. Yazdanparast, B. Axelrod, J. Biol. Chem. 262 (1987) 10080-10085. J.N. Siedow, Annual Rev. Plant Phys. Mol. Biol. 42 (1991) 42-145. M. Pinto, J. Luis Gata, P. Macias, Biotechnology Progress 13 (1997) 394-398.


565

Stabilization o f lipase B f r o m C a n d i d a antarctica b y i m m o b i l i z a t i o n on different s u p p o r t s M. Arroyo; J. M. S~inchez-Montero and J.V. Sinisterra Department of Organic and Pharmaceutical Chemistry, Faculty of Pharmacy, Universidad Complutense, 28040 Madrid, Spain

Summary Several immobilization methods and supports have been tested for the covalent bonding of pure lipase B from Candida antarctica. We have also compared the thermal stability of our covalent immobilized derivatives with those obtained by Novo Nordisk by absorption of the same lipase on different polymers. At 50~ native enzyme and Novozym 435 follow a deactivation model E--->EI, whereas covalently immobilized derivatives and SP435A follow the model E--->E1--->E2.This different behaviour is related to the nature of the support and to the immobilization methodology. 1. INTRODUCTION Lipase B from Candida antarctica (1) is an interesting lipase with potential application in a number of industrial processes and in organic synthesis of fine chemicals and pharmaceutical raw materials (2). Therefore, the immobilization of this lipase can offer several advantages, including repeated usage and enhanced stability. In the present paper we show the results obtained in its immobilization and stabilization using different methodologies and supports. 2. MATERIALS AND METHODS

2.1. Materials Native lipase B from Candida antarctica (SP525), and the same lipase immobilized on Lewatit OC 1600 (SP435A) and Lewatit E (Novozym 435) were kindly supplied by Novo Nordisk Bioindustrias (Madrid, Spain). Tresylated sepharose 4B and epoxy-activated sepharose 6B were purchased from Pharmacia (Sweden). Silica (Kiesegel 60, size 0.0150.040 nm, average pore diameter = 95 A, surface area = 239 m2/g), alumina (Aluminum 60, size 0.063-0.200 nm, average pore diameter = 60 A, surface area = 166 m2/g) and isooctane (analytical grade) were obtained from Merck (Darmstadt, Germany). 2,4,6 trichloro 1,3,5triazine was supplied by Aldrich (Steinhiem, Germany). Pure triacetin was purchased from Sigma Chemical Co. (St. Louis, MO).

566

2.2. Protein Determination The protein content of SP525 (0.1 mg of protein/mg of derivative) was determined by the Biuret method (3). In a typical experiment, 0.1 mL of a 125 mg/mL SP525 solution was mixed with the Biuret reagent and the protein concentration was determined spectrophotometrically at a wavelength of 545 nm, using a calibration curve of seroalbumin. 2.3. Covalent immobilization on inorganic supports The activation of silica and alumina was carried out according to the 2,4,6-trichloro-1,3,5triazine (TCT) method, previously described in the immobilization of Candida rugosa lipase (4). The immobilization of lipase B from C. antarctica was performed at 4~ for 6 hours, with low stirring. One gram of each support was added to different concentrations of enzyme in 10 ml of standard buffer (0.1 M Tris/HCl buffer, pH=8.0). After the desired contact time, the insoluble enzyme derivative was filtered and washed with standard buffer. The percentage of immobilized enzyme was determined by the difference between the initial activity of the native enzyme solution and the activity of the filtrate after the immobilization process. 2.4. Covalent immobilization on organic supports Epoxy-activated sepharose 6B (2g) was mixed with 10 rnL of the enzymatic solution of native CALB in 0.1 M Tris/HC1 (pH = 8.0) buffer. The mixture was stirred at 40C for 6 hours and then filtered and washed with 3 x 10 mL of buffer solution. The immobilized biocatalyst on tresyl-activated sepharose 4B was prepared by an experimental methodology equivalent to that described above. 2.5. Hydrolysis assay All the biocatalysts were tested in the hydrolysis of triacetin due to the low activity of this lipase in the hydrolysis of triglycerides with long chain fatty acids (5). As standard assay, the hydrolysis of pure triacetin was performed in 1 mM Tris/HC1 buffer (pH=7.0) at 37~ The acetic acid released was continuously titrated to constant pH with the help of a pHstat (Crison model microTT 2022). Several NaOH solutions (1 to 10 mM) were used as titrating agents. The catalytic efficiency of immobilized derivatives was determined as the ratio between the enzymatic activity of 3 ~tg of native lipase and the activity of the amount of immobilized derivative which contained 3 ~tg of enzyme, taking into the account the percentage of immobilized enzyme. 2.6. Thermal stability assays The thermal stability assays were performed with the same amount of lipase: native or immobilized. The storage stability of native and insolubilized enzymes was carried out at 50~ in 0.1 M Tris/HCl buffer, pH=8.0. These experimental conditions were selected as the extreme conditions. After incubation for different times, the remaining activity was measured in the hydrolysis of triacetin as described above. 2.7. Electron microphotographs Electron microphotographs were taken with a 7~ISS DSM 940 scanning electron microscope.

567 3. RESULTS AND DISCUSSION 3.1. Covalent immobilization of pure lipase B from Candida antarctica The results of covalent immobilization of CALB on different activated organic and inorganic supports are shown in Table 1. We may conclude that the length of the spacer arm in the sepharoses, does not affect the amount of immobilized lipase (640-660 ~tg CALB/g support, 42-44% immobilization). However, there is a significant effect on the catalytic efficiency of the immobilized lipase. We expected higher catalytic efficiency in CALB-ES-1 (49%) where the enzyme should have larger conformational freedom due to the longer spacer arm of the epoxy-activated sepharose (12 carbons). Instead, CALB-TS-1 where the enzyme is closely attached to the matrix, showed higher catalytic efficiency (72%). We tried to explain this behaviour taking into account the difference between the chemical bonds on different sites of the enzyme. Epoxy-activated sepharose provides a method for coupling proteins through hydroxyl, amino or thiol groups of their amino acids, whereas tresylated sepharose only links through amino and thiol groups. An interaction between the oxirane group of the sepharose and the hydroxyl group of serine 105 of the catalytic triad (1) could be involved during the immobilization, then diminishing the catalytic efficiency of CALB-ES derivatives respect to CALB-TS ones. Table 1 Immobilization of lipase B from Candida antarctica.

Derivative

Activated support

~tg lipase added/g. support

Immobilization

CALB-ES- 1

Epoxyactivated sepharose a

1500

44

CALB-ES-2

Epoxyactivated sepharose a

5000

CALB-TS- 1

Tresylated sepharose b

CALB-TS-2

C'E'f (%)

S'E'A'g

660

49

32

18.4

920

49

45

1500

42.5

640

72

47

Tresylated sepharose b

5000

27.4

1370

72

100

CALB-S-1

Silica-TCT c

2000

90.2

1810

38

62

CALB-S-2

Silica-TCT c

6700

48.8

3250

31

86

Alumina-TCT d

6700

9.3

620

27

14

CALB-AL- 1 ,

,

,,,

.,..,,

,

,

(%)

(~tglipase /g.support

_.

ai9-40 ~leq of oxyrane groups/ml wet gel; bGrade of activation not supplied by Pharmacia LKB; c0,24 g TCT/g silica; d0,19 g TCT/g alumina; ereferred to the amount of CALB added in the immobilization process; fCatalytic Efficiency in the hydrolysis of triacetin; gSpecific Enzymatic Activity expressed as p,moles of acetic acid released/min.g of dry derivative. If the amount of added lipase is increased in the immobilization on both activated sepharoses, the ~g of enzyme bonded to the support is increased as well, but the percentage of immobilization decreases (Table 1), and the catalytic efficiency remains the same (72% in CALB-TS derivatives and 49% in CALB-ES ones). Similar results have also been observed in the immobilization of C. rugosa lipase on agarose activated by tosylation methodology (6).

568 The enzyme loading in the derivatives on silica is increased with the amount of lipase added in the immobilization process (CALB-S-1 and CALB-S-2). Nevertheless, lipase molecules form multilayers on the support surface (Figure 1) and, as a consequence, the catalytic efficiency diminishes (31-38%, Table 1). Similar results were reported in the immobilization of lipase from C. rugosa on TCT-activated silica (4). Finally, we can assume that alumina is not an adequate support for the immobilization of CALB, according to the Figure 1. Electron microphotograph of poor catalytic efficiency value of CALBCALB-S-2 (original magnitude 1000x). AL- 1 (Table 1). The hindering effect of the pore diameter of the support (95 A in silica, and 60 A in alumina) can explain how the lipase loading is higher in CALB-S-2 (3250 lag/g support) than in CALB-AL-1 (620 ~tg/g support). This indicates that the enzyme, whose size is 30x40x50 A (1), is located in the external surface of alumina, and therefore, a poor enzyme loading is observed. Lipase multilayers are also formed and the catalytic efficiency of both derivatives is similar (31% and 27% respectively). 3.2. Thermal Stability

The storage stability of native and covalently immobilized lipase B from C. antarctica was studied in wet conditions at 50"C (Table 2). The thermal deactivation curves have been explained following the deactivation model proposed by Henley and Sadana (7). This model involves enzymatic states (E, E1 and E2), where kl and k2 are first-order deactivation rate coefficients and oq and or2 are the ratios of specific activities E1/E and E2/E respectively (Equation 1). The experimental plots of residual activity versus storage time were adjusted to exponential decays (Equation 2) (single or double, with or without offset) with the help of Simfit program developed by Dr. Bardsley (8). From the data of the data adjusted equations and using the Equation 3, we could calculate all the parameters (kl, k2, al, ct2), the half-life of the biocatalyst and the stabilization factor (F), considered as the ratio between soluble and derivatives half-lives. oq

(~2

E k--~l E1 k~2 E2

(1)

A = Al.e -kIt + A 2.e "k2t+ A3

(2)

A=

1

Oqkl _ (X2k2 e-kit + k2- kl]

100+ k2-kl

k2-kl

k2-kl

le,t.o

(3)

569 Table 2 Thermal deactivation of native and immobilized lipase B from C. antarctica at 50~ ,.

Derivative

Native CALB CALB-ES- 1 CALB-ES-2 CALB-TS- 1 CALB-TS-2 CALB-S- 1 CALB-S-2 SP435A e Novozym 435 f

Al a

,.i

98.9 43.6 57.8 52.8 27.6 12.5 74.0 46.7 101

1.30 1.80 0.70 1.01 0.56 0.85 0.42 4.10 0.05

,

A2a

k2(h-l) a

0 56.0 43.3 47.2 73.3 87.8 24.6 0 0

0 0.03 0.06 0.01 0.08 0.02 0.01 0 0

kl(hn) a

Otlb

,, tl/2 (h) c

(x2b (A3)

0 55 40 47 63 86 23 53 0

0 0 0 0 0 0 0 52.6 0

0.5 4 2.5 2.5 4.5 24 2.5 0.5 14

d -

1 8 5 5 9 48 5 1 28

aParameters (in %) from the t~tted exponential decay equation; btxl and ct2 are the ratio (in %) of specific activities E1/E and E2/E respectively; Chalf-life; astabilization factor; ~S.E.A.= 147 ~tmoles acetic acid/min.g of SP435A; fS.E.A.= 238 l.tmoles acetic acid/min.g of Novozym 435. At 50~ native CALB deactivation followed a single exponential decay which belongs to the classical first-order deactivation pattern (Table 2), in which cxl=0, tx2=0 and k2=0. The deactivation of covalently bonded CALB followed a double exponential decay when the derivatives were stored at 50~ (Figure 2). In all cases, kl > k2, t~l< 100% and t~2=0 (Table 2). Residual activity (%) 100~_ t ~ 80

I--] /,,

Native

(~

CALB-TS-I

CAL,-~.S.1 . .

Residual activity (%) 100 ~~xc_~ D J 80~

60

60

40

40

20 i

20

i i" i I 10 15 20 25 Time (hours) Figure 2. Thermal stability of covalent immobilized lipase B from C. antarctica, 0

0

i 5

~,~ .~

/~ (~

Native SP435A Novozym435

.

0 0

5

10 15 20 25 Time (hours) Figure 3. Thermal stability of immobilized derivatives from Novo Nordisk.

The small values of k2 mean a good stabilization of enzymatic state E~. The hydrophilic nature of sepharose protects the enzyme, independently from the length of spacer arm, giving similar al, half-life values and deactivation profile. In the derivatives prepared with silica, the

570 activity of the intermediate state El may be affected by the lipase loading. The intermediate of CALB-S-1 has higher activity (0L1=86%)than the intermediate of CALB-S-2 (tx1=23%), due to the formation of lipase multilayers in the second derivative as mentioned above. As a consequence of this enzymatic aggregation, the lipase is weakly linked in the outer layers, and it is quickly deactivated. We can conclude that covalent immobilization of pure lipase B from Candida antarctica produces an appreciable stabilization of the biocatalyst, changing its deactivation profile. This change from a single exponential decay (native lipase) to a double one (immobilized lipase), was also observed in pure lipases from Candida rugosa (9). On the other hand, Novozym 435 and SP435A showed a very different deactivation pattern (Figure 3) compared to our covalently immobilized derivatives (Figure 2). After a quick deactivation, SP435A keeps its residual activity (oq= 53%) for a long time. This stabilization could be explained by the lipase location inside the micropores of the support (Figure 4), where the enzyme is protected against alterations of the microenvironment. On the contrary, CALB is completely exposed to the medium in the surface of the small beads of Novozym 435 (Figure 5), so its deactivation model is similar to the native lipase.

Figure 4. Electron microphotograph of SP435A (original magnification 50x).

Figure 5. Electron microphotograph of Novozym 435 (original magnification 20x).

REFERENCES 1. Uppenberg, J.; Hansen, M.T.; Patkar, S. and Jones, T.A. Structure, 2 (1994) 293-308. 2. Uppenberg, J.; Ohrner, N.; Norin, M.; Hult, K.; Kleywegt, G.J.; Patkar, S.; Waagen, V.; Anthosen, T. and Jones, T.A. Biochemistry, 34 (1995) 16838-16851. 3. Gornall, A.G.; Bardawill, C.S.; David, M.M.J. Biol. Chem., 177 (1949) 751-766. 4. Moreno, J.M. and Sinisterra, J.V.J. Mol. Catal., 93 (1994) 357-369. 5. Rogalska, K.; Cudrey, C.; Ferrato, F.; Verger, R. Chirality, 5 (1993) 24-30. 6. Arroyo, M. ; Moreno, J.M. and Sinisterra, J.V.J. Mol. Catal., 83 (1993) 261-271. 7. Henley, J.P. and Sadana, A. Enzyme Microb. Technol., 7 (1985) 50-60. 8. Bardsley, W.G. SIMFIT: a computer package for simulation, curve-fitting, graph-plotting and statistical analysis using life science models. University of Manchester, U.K., 1995. 9. Moreno, J.M.; Arroyo, M.; Hermiiz, M.J.; Sinisterra, J.V. Enzyme Microb. Technol., 21 (1997) 552-558.


571

C o v a l e n t i m m o b i l i z a t i o n of crude and partially-purified lipases onto inorganic supports: stability and hyperactivation. A. R. Alcfintara, I. Borreguero, M. T. L6pez-Belmonte and J. V. Sinisterra. Departamento de Quimica Org~nica y Farmac6utica. Facultad de Farmacia Universidad Complutense de Madrid. Ciudad Universitaria, sin. E-2840 Madrid (Espafia). The covalent binding of crude and partially purified lipases from Rhizomucor miehei (RML) and Humicola lanuginosa (HLL) on inorganic supports (silica and alumina particles) was carried out via an activation step using 2,4,6-trichloro- 1,3,5-triazine (TCT), rendering very stable biocatalysts and showing an hyperactivated performance for HLL-derivatives. 1. INTRODUCTION Covalent binding of an enzyme to a support is probably the most interesting immobilization method from an industrial point of view: In this methodology, the previously activated groups of the carrier react with some external functional groups of the protein (generally, amino or carboxyl groups). In order to provide the best mechanical properties to the immobilized derivative, inorganic carriers should be used. Many procedures for activating these supports have been described [1], and many lipases have been covalently attached to them [2, 3]. In our research group we have described the immobilization of crude [4] and pure isoenzymes [5] of Candida rugosa lipase onto inorganic supports (alumina, silica or controlled pore glass) by means of an activation step using 2,4,6-trichloro- 1,3,5-triazine (TCT), testing these immobilized derivatives in the kinetic resolution of chiral compounds [6-8]. In these paper we present the results obtained in the covalent immobilization of crude lipases from Rhizomucor miehei and Humicola lanuginosa onto silica and alumina using the above-mentioned methodology. On the other hand, crude Rhizomucor miehei lipase was partially purified by ammonium sulphate precipitation, following a methodology slightly different from that recently reported [9], and immobilized on activated silica. 2. MATERIALS AND METHODS 2.1. Enzymes and chemicals Crude lipases from Rhizomucor miehei and Humicola lanuginosa (which would be denoted as CRML and CHLL, respectively), as well as an immobilized derivative of Rhizomucor miehei adsorbed onto a macroporous anion-exchange resin (Lipozyme IM) were kindly given by Novo Nordisk Bioindustrial, Spain. Silica-40 (particle diameter = 0.063-0.200 mm (70-230 mesh, ASTM); surface area = 239 m2/g; pore size = 95 ,~; pore volume = 0.25 cm3/g), Alumina-60 (particle diameter = 0.063-0.200 mm (70-230 mesh, ASTM); surface area = 166 m2/g; pore size = 60 A; pore volume = 0.25 cm3/g) and 2,4,6-trichloro-l,3,5-triazine (TCT) were from Merck (Darmstad, Germany).

572 Tributyrin, sodium chloride, glycerol, arabic gum, potassium dihydrogen phosphate, and nbutanol were of reagent grade from Sigma (Alcobendas, Spain). Ibuprofen (rac-2-(4isobutylphenyl)propionic acid) was kindly given by Boots (Nottingham, UK). Aqueous solutions were made with distilled/deionized water (Barsntead).

2.2. Partial purification of Rh. miehei iipase The crude enzyme (8.1:L-0.3 mg/ml, according to the Bradford procedure [ 10]) was diluted (1/10) with 50 mM Tris buffer pH=7.5.25 ml of this solution, together with the corresponding amount of (NH4)2SO4 to render the optimum saturation percentage ( 9.15 g, 60%) were gently stirred at 4~ for 1 h. The resulting precipitate (denoted as SPRML was collected by centrifugation at 5000 rpm at 4 ~ for 40 min, dissolved in Tris buffer pH=8.0 and kept at 4 ~

2.3. Immobilization procedure 2.3.1 Support activation Silica and alumina were previously activated according to the method described elsewhere [4]: the support (25 g) was placed in a round-bottom flask, and toluene (250 ml), TCT (7.5 g) and Et3N (15 ml) were added. This mixture was kept under stirring (700 rpm) for 4 h at 60~ After this time, the product was filtered and washed with toluene (250 ml) and acetone (2 x 250 ml). The activated support was dried under vacuum for 1 h and stored at 4 ~ The activation degree of the carriers was quantified by titration of the CI ions produced by alkaline hydrolysis (0.1 M NaOH) of the activated support. 0.01 M AgNO 3 was used as titrating agent and fluorescein as indicator [4].

2.3.2. Immobilization process The immobilization of crude lipases (CRML and CHI~) or semipurified Rh. miehei lipase (SPRML) on activated silica was carried out at 40C for 22 h (CHLL) or 49 h (CRML and SPRML), by gently stirring of, respectively: i) 10 grams of activated silica, 150 ml of a 1.5 mg/ml solution of CHLL in Tris buffer 0.1 M, pH=8.0; ii) 11 grams of activated silica, 100 ml of a 5.2 mg/ml solution of CRML or SPRML in Tris buffer 0.1 M, pH=8.0. After this time, the biocatalysts obtained (that would be denoted as SiI-CHLL, SiI-CRML and SiI-SPRML, respectively), were filtered and washed with Tris buffer 0.1 M, pH=8.0 (150 ml) and distilleddeionized water until the eluents were free of proteins. The immobilization of CHLL on activated alumina was carded out following a similar procedure, but using Tris buffer 0.1 M, pH=9.0, obtaining the derivative denoted as AI-CHLL. Subsequently, the percentage of immobilized enzyme was determined by the difference between the protein amounts present in the starting material and the filtrates obtained during the immobilization processes.

2.4 Enzymatic assays The hydrolysis of tributyrin was carried according to the Novo method [ 11], and the acid released was continuously titrated to pH=7.0 with the automatic addition of 0.05N NaOH, quantifying the activity by measuring the initial rate. All the data were average of triplicate samples and reproducible within __.5%experimental error.

2.5. Stability measurements In order to measure the thermal stability of the derivatives obtained, they were incubated at different temperatures, in the same buffer used for measuring its enzymatic activity. At different times, aliquots were taken and assayed for activity against tributyrin; the data obtained were fitted to exponential decay curves by using the EXFIT programme of the SIMFIT package and

573 analyzed according to the series-type mechanism proposed by Henley and Sadana [ 12].

2.6. Enzymatic esterification and transesterification i)Esterification of ibuprofen. The standard reaction mixture was composed by cyclohexane (10 ml), ibuprofen (0.125 M) and n-butanol (0.125 M). The reaction was carried out at 37 ~ by stirring in a 25 ml flask for a specified time, by adding the proper amount of biocatalyst (CRML, SPRML, SiI-CRML or SiI-SPRML). Then, aliquots of 0.1 ml were taken from the solution at different times and added to 1.4 ml of cyclohexane; after microfiltration, they were analyzed (yield and enantiomeric excess) by HPLC on a Water-Millipore apparatus, Model 590 equipped with a Chiracel OD chiral column of cellulose carbamate (25 cm x 0.46 cm, Daicel Chemical Industries Ltd, Tokyo, Japan) using a mobile phase composed by hexane/2-propanol/ trifluoroacetic acid (100/10/1, v/v/v), flow rate = 0.8 ml/min, detecting the products at 254 nm. ii)Transesterification of rac-l-chloro-3-(1-naphtoxy)-2-propanol. The standard reaction mixture was composed by isooctane (15 ml), rac-l-chloro-3-(1-naphtoxy)-2-propanol (0.42 mmol, ) and vinyl acetate (1.26 mmol). The reaction was carried out at 30~ by stirring in a 25 ml flask for a specified time, by adding the proper amount of biocatalyst (CHLL or SiI-HLL, 106 mg of protein). Then, aliquots of 0.1 ml were taken at different times and added to 1.4 ml of isooctane and analyzed as above mentioned, but now using a mobile phase composed by hexane/ 2-propanol/diethylamine (80/20/0.1, v/v/v), flow rate = 0.5 ml/min, and UV detection at 254 nm. 3. Results and discussion 3.1 Biocatalysts characterization The properties of the different catalysts are presented in Table 1. Table 1 Characterization of the Rhizomucor miehei and Humicola lanuginosa catalysts. Catalyst

KM,

kcat, units/mg a

Activity (units/mg) a

mM ~

CRML

0.75_-+0.04

60•

1.10_-_+0.08

SPRML

2.0_-+0.1

60•

3.8_+0.3

115•

0.014_-+0.001

Lipozyme IM (3.8_-+O.4)x10.3 SiI-CRML

(4. l•

2

95•

Sil-SPRML

0.25_-+0.01

210•

CHLL

3.20-M).05

13•

Sil-CHLL

0.016_+0.004

90+_36

(7.5•

103

0.5:fl9.1 62• 0.017•

Loading b

Yield d Ret. activitye (%) (%)

IMMOBILIZF~D CATALYSTS 120c

.....

0.013_-+0.002

28+_2

59•

9__.1

1.3•

29.-!:2

12.5•

IMMOBILIZED CATALYSTS 15

>99

0.5_+0.1

A1-CHLL 0.14_-+0.02 7_+3 0.071 • 15 >99 4.3_-+0.7 aTributyrin hydrolysis (1 unit= 1 mmol H + releasedxmint), bmg proteinxg of catalyst. r data [ 13]. dlmmobilization yield, eCompared to the soluble enzyme. As can be seen, of all the Rh. miehei catalysts, the ammonium sulphate precipitate (SPRML) is that one presenting the best activities in tributyrin hydrolysis, 3-fold increase, as reported in

574 literature [9]. Upon immobilization on activated silica, the crude enzyme losses a great activity percentage, although this activity diminution is smaller for the partially-purified enzyme. As expected, the less active catalyst was Lipozyme IM, because this adsorbed derivativeis not appropriate for aqueous media. We can also observe that the immobilization of crude H. lanuginosa lipase on activated alumina allows the obtention of a catalyst (AI-CHLL) better than that obtained with silica. This fact may be caused by the more alkaline character of alumina, that would favor the enzymatic activity, because the optimum pH of H. lanuginosa lipase is about 10-11 [14]. 3.2. Thermal stability The deactivation profiles of the catalysts are shown in Figures 1 to 5.

o~100 'd'~-~ I ~' '

100.

(r=,3~,ob] . '' ' '

~ 8o

~

8o

~

60

-

"

!

"

I

Ir=s~176

"

1

"

I

'

-....

~ 60 = 40 E 20 Q:: 0

= 40

cRii

E 20

SPRML i "

I

0

'

I

'

I

"

I

Q:

"

10 20 30 40 50

-

9. . . . . .

9CHLL" "

"~ 80 60

SiI-HLL ~

0

75 50

~

E 20 (b

0

9

~

40 r

10 20 30 40 50

100~' ' ' ' ' ' 'iC'H~' i

...... .e...,

~

0

Time (h) Fig.2.- Thermal deactivation of Rh. miehei derivatives at 50~

Tim e (h)

Fig.l.- Thermal deactivation of Rh. miehei derivatives at 37~

~100

0

I

~"" 9

I

25

-%'"I

9

100 200 300 400

Time (d)

9

400-

.) . , .

[A/'CHL~J~T

300

/

200

~@.=.

100

" I

500

=37~ .

,T=25oc

,,,.~,

"

-,r

"

-

=

-

-

~-''"

0 ~T-50.,.,, -

0 0

50

100

Time (h)

150

oC

0 15 30 45 60 75 90

Time (h)

Fig.3.- Storage stability (4~ of Fig.4.- Thermal deactivation Fig.5.- Thermal deactivation CHLL, SiI-CHLLandAI-CHLL of CHLL at different of AI-CHLL at different temperatures temperatures

Table 2 shows the parameters that characterize the thermal deactivation of the different biocatalysts, according to the series-type deactivation mechanisms proposed by Henley and Sadana [12]. As can be seen, the partially-purified Rh. miehei lipase shows an enhanced thermostability compared to the crude enzyme (8.7 and 1.8-fold, considering the half-life time, at 37 and 50~ respectively), while the immobilization on activated silica renders even much

575 more thermostable derivatives both for the crude and partially-purified Rh. miehei lipase, with increases up to more than 100-fold (SiI-CRML versus C R M L at 37~ Nevertheless, even at 4 ~ (storage conditions) the silica derivative ofH. lanuginosa (SiI-CHLL) gets deactivated faster than the crude enzyme, which retains a 95% of its initial activity after more than 400 days Table 2 Thermal deactivation parameters

T (oc)

kl

CHLL

4

7.9x10 6

CHLL

25

7.7x10 ~ 3.6x10 1 6.8x10 .3

CHLL

37

3.6x10 l 7.1xlO l

CHLL

50

1.6

A1-CHLL

4

-6.7x10 5

AI-CHLL

25

7.1xlO 3

AI-CHLL

37 -3,3x10 z

A1-CHLL

50

SiI-CHLL

4

CRML

37

CRML

2

r O

1.4x 10l

k2

tt 2 tit2 (h)

....

.... 87000 0

1.6xlO "3 0 2.7x 10.3

2.2 215.6

0

0.5

....

n.d

9.9

2,2x 10.2 > 1

n.d.

8

1.9x10 2 >1

n.d.

1.4x10 ~ 2.8x10 -2 0

0.4

1.7xlO ~ 3.6x10 ~ 1.8xlO 3

0

818.4

4.6x10 1 1.2x10 .2

0

1.8

50

6.2x 10.2 1.6x 10 "! 1.9x10 .2

0

1.4

SPRML

37

4.4x10 -2

0

....

SPRML

50

2.7x10 2

0

....

.... ....

15.7 2.6

SiI-CRML

37

7.1x10 "4 4.0x10 "l

4.6

0

SiI-CRML

50

3.7x10 .2 5.7x10 1 1.8x10 -2 0

257.9 10.7

SiI-SPRML

37

7.1 x 10-3

98.2

SiI-SPRML

50

7.1x10 1 8.0x10 ~ 8.7x10 .3

1.3

0

........ 0

5.4.2

(Fig. 3). Increasing the temperature, we can observe how the crude enzyme presents higher resistance to the thermal deactivation at 37~ than at 25~ (Fig. 4). This effect could be understood assuming a stabilizing effect of the cations Na § and K § which are present in the medium, similar to the effects described for Candida rugosa lipase with the s a m e cations, also maximum at 37~ [15]. On the other hand, the alumina derivative (AI-CHLL) displays an hyperactivation upon storage, although this effect is more dramatic at higher temperatures (25 and 37~ and disappearing at 50~ (Fig. 5). This important activity increase observed could be explained attending to a more alkaline micropH in the enzyme microenvirontment, due to the basic surface properties of the alumina; another reason could be the

hydrophobic adsorption of the covalently-linked enzyme molecules to a neighbor support surface (which upon TCT activation would increase its hydrophobicity)geometrically congruent, stabilizating the "open" form, as recently described for this same enzyme [ 16].

3.3. Enzymatic esterification and transesterification Table 3 shows the efficiency of the catalysts obtained in two reactions in organic solvents. The immobilization does not alter the enantioselectivity, and good activity retentions can be obtained. Therefore, these derivatives are very attractive for biocatalytical purposes.

576 Table 3 Kinetic resolution of chiral compounds in 0rl~anic solvents Biocatalyst

Specific activity Yield (t, h) (mMxh-~xmg prot. "l)

ee (%) remnant substrate

Relative Eficciency

ibuprofen + n-butanol, cyclohexane 137",C CRML

3.5x 10.2

65 (72)

68

4.1

100

SPRML

1.19x 101

21 (72)

15.5

4.3

2332

Lipozyme IM

3.4x 10.3

66 (72)

85

6.4

10

SiI-CRML

9.5x10 3

4.7 (72)

3.5

7.2

27

Sil-SPRML

.1.7x10 2

.2.9 (72)

1.1

4.5

49

..

rac-l-chloro-3-(17naphtoxv)-2-orooanol + vinyl acetate, (iPr)70, 25~ CHLL

1.5x 104

30.3 (547)

31.7

> 100

100

Sil-CHLL

6.5x 10.5

2.9 (839)

4.0

> 100

40

AI-CHLL

7.4x 10-5

3.3 (839)

4.6

> 100

49

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

14. 15. 16.

G.F. Bickerstaff (ed.), Immobilization of Enzymes and Cells, Humana Press, Totowa, New Jersey, 1997. T. Aydemir and A. Telefoncu, Indian J. Chem., 33B (1994) 387. A. Ivanov and M. P. Schneider, J. Mol. Catal. B: Enzymatic, 3 (1997) 303. J.M. Moreno and J. V. Sinisterra, J. Mol. Catal., 93 (1994) 357. J.M. Moreno, M. Arroyo, M. J. Hern~iiz and J. V. Sinisterra, Enzyme Microb. Technol., 21 (1997) 552. J.M. Moreno and J. V. Sinisterra, J. Mol. Catal. A: Chemical, 98 (1995) 171. J.M. Moreno, A. Samoza, C. del Carnpo, E. F. Llama and J. V. Sinisterra. J. Mol. Catal. A: Chemical, 98 (1995) 179. E.M. S~chez, J. F. Bello, M. G. Roig, F. J. Burguillo, J. M. Moreno and J. V. S inisterra, Enzyme Microb. Technol., 18 (1996) 468. X.Y. Wu, S. Ja~tskelainen and Y-Y Linko, Appl. Biochem. Biotechnol., 59 (1996) 145. M.M. Bradford Anal. Biochem., 72 (1976) 248 Novo Nordisk Analytical method AF95. Enzyme Process Division. Bagsveard, Denmark. J.P. Henley and A. Sadana, Enzyme Microb. Technol., 7 (1985) 50. F. V~quez-Lima, D. L. Pyle and J. A. Asenjo, Biotechnol. Bioeng., 46 (1995) 69. LipolaseTM. Technical Characteristics. Detergent Enzyme Division. Novo-Nordisk. Bagsveard, Denmark. M.J. Hernfiz, M. Rtia, B. Celda, P. Medina, J. V. Sinisterra and J. M. S,Snchez-Montero, Appl. Biochem. Biotech., 44 (1994) 213. A. Bastida, P. Sabuquillo, P. Armisen, R. Fern~dez-Lafuente, J. Hughet and J. M. Guisfin, Biotecchnol. Bioeng., in press (1998).


577

Immobilization o f alkaline p h o s p h a t a s e s on various supports B.Surinenaite, V.Bendikiene and B.Juodka Department of Biochemistry and Biophysics, Vilnius University, Ciurlionio 21, LT-2009 Vilnius, Lithuania

Alkaline phosphatase from E.coli RY 13 strain was purified from one fraction obtained during purification of the inorganic pyrophosphatase. The enzyme was covalently bound on aminopropylsilochrome and aminoalkylsilylated magnetite, both activated by glutaraldehyde and was fixed on the "naked" ferromagnetite by direct adsorption followed by the stabilization of the adsorbed enzyme with glutaraldehyde as well. Two other alkaline phosphatases (from E.coli strain C4 and from chicken intestine) were immobilized on magnetic derivatives of chitin and chitosan, magnetite and its aminoalkylsilylated analogue. Pure chitin, chitosan and aminopropylsilochrome were used as model nonmagnetic supports in the case. Those enzymes were bound covalently to matrix with reactive groups on their surface after the activation by glutaraldehyde and were fixed by direct adsorption on all supports mentioned. The efficiency of the immobilization depended on: carriers nature and preparation; method of the enzyme binding; concentration of protein and activating agent; volume of the reaction mixture; duration of the reaction etc.

1. INTRODUCTION The stabilization of biocatalysts remains one of the most actual problems in various fields of their application. The immobilization of enzymes is one of common methods used to that purpose. Recent developments in immobilization lead to commercially viable application of preparations in industry and many other spheres [1]. Such interest is connected with advantages of fixed biocatalysts, i.e. increased stability, less complicated separation of reaction products, possibility of enzymes regeneration etc. [2]. Magnetic carriers attracted the attention of researchers as magnetic particles with immobilized biocatalysts or whole cells could be easily removed in the presence of the magnetic field providing another simple method of separation [3-6]. In this report, immobilization of alkaline phosphatases from E.coli RY 13 and C4 strains as well as chicken intestine on various magnetic supports by direct adsorption and covalent binding is discussed.

578 2. MATERIALS AND METHODS 2.1. Materials E.coli alkaline phosphatases (AP) from strains RY 13 and C4 (E.C. 3.1.3.1) were obtained from MBI Fermentas (Lithuania). Chicken intestinal alkaline phosphatase (AP) was purchased from Reanal (Hungary). Reagents used were from: chitin- Reachim (Russia), TRIS- Sigma (USA), p-nitrophenylphosphate (p-NPP)Koch-Light (Great Britain), Taminopropyltriethoxysilan (T-APTES)- Serva (Germany), glutaraldehyde (GA)- Fluka (Switzerland), aminopropylsilochrome (APS)- Biolar (Latvia). Other salts, alkalies, acids, ethanol, toluene, acetone, salicylaldehyde used were from Reachim (Russia) of the highest degree of purity available. 2.2. Preparation of carriers for enzymes immobilization Chitin was prepared by melting, demineralization, deproteinization and depigrnentation according to the method of [7]. Chitosan was prepared by heating of demineralized chitin in 40% (w/w) NaOH solution in the presence of 1% (w/w) NaBH4 [8]. "Magnetization" of both chitin and chitosan was carded out in the mixture of Fe(III) and Fe(II) salts as described in [9]. Ferromagnetite (F%O4) was prepared by precipitation from Fe(III)/Fe(II) salts mixture (molar ratio 2/1) in 25% (v/v) NH4OH solution. Ferromagnetite was modified by T-APTES. A quantity of 18 g of magnetite washed with toluene was added to 18 g of y-APTES solution in toluene (final volume 180 ml). Reaction was carded out by stirring during 20-25 h at boiling temperature of toluene (110~ After reaction magnetite was washed with toluene, acetone and dried at 100-105~ APS was used as purchased, without additional preparation. 2.3. Immobilization of enzymes by direct adsorption AP from E.coli RY 13 strain was adsorbed on magnetite. Enzymes from E.coli C4 strain and chicken intestine on chitin, chitosan, their magnetic derivatives, magnetite, its modified analogue, and APS. A quantity of 0,1 g of each carder was washed with 0.5 M TRIS-HC1 buffer, pH 8.0, then 1 ml of enzymes solution in the same buffer was added and the mixture was stirred for 2 h at room ternperature. Then, the solution was poured, the carrier washed with buffer, and the protein concentration and the phosphatasic activity in all washing fractions and on the support determined. 2.4. Immobilization of enzymes by covalent binding AP from E.coli RY 13 strain was eovalently bound on aminoalkylsilylated magnetite and APS. The enzymes from E.coliC4 strain and chicken intestine were also immobilized on these supports and on pure and magnetic chitosans as well. All carriers were activated by GA using 25-fold excess with respect to 1 tool of amino groups on the surface of the carriers. The amount of primary NH2 groups was determined using salicylaldehyde as described [7]. A quantity of 0.1 g of each carrier was washed as before adsorption and stirred 2 h at room temperature in the presence of 2 ml aqueous GA solution in final concentration including the excess of GA. Then the carrier was thoroughly washed with bidistilled water, kept with 0.5 M TRIS-HC1 buffer, pH 8.0 during 30 rain and after pouring of the buffer, 1 ml of enzyme solution was added. Other procedures were the same as in the case of the adsorption. The stabilization of E.coli RY 13 AP adsorbed on pure magnetite was carried out in the presence of GA excess as in the case of modified magnetite during covalent binding. The

579 bifunctional agent was added after the adsorption and other procedures were as in covalent fixation (see above).

2.5. Evaluation of the immobilization efficiency (rl) and the yield (T) These characteristics could be expressed as rI=PJP,

(1)

where Pi- is the amount of the product released by the fixed enzyme under practical conditions, P,- is that for soluble enzyme. T=a'I3=Bi/Bs'Ei/E,

(2)

where B s- is the initial amount of enzyme before immobilization, B i- the amount of the immobilized enzyme, Es- the specific activity of soluble enzyme, and Ei- that of immobilized enzyme [ 10].

2.6. Alkaline phosphatase activity assay Phosphatasic activity was determined using p-NPP as substrate during 10 min at 37~ and pH 8.0 according to the method described in [ 11 ]. 2.7. Determination of protein concentration Protein concentration was detected by spectrophotometric absorption at 280 nm. 2.8. Statistical calculations All experiments were carried out twice independently, and all measurements were repeated twice as well.

3. RESULTS AND DISCUSSION AF are known as important enzymes of nucleic acids metabolism. Investigation of them could be a tool for examining of conversion of genome structural elements. Stabilization of AF is actual from this standpoint. E.coli AF has been studied in detail [ 12, 13] but there has been no such interest into the chicken intestinal enzyme [ 14]. We examined the stabilization of three AF (two from E.coli, strains RY 13 and C4, and one from chicken intestine) by immobilization on solid supports. Seven different carders were used for immobilization by two methods in order to evaluate the role of matrix nature and method of fixation on the immobilization efficiency and the yield. Results are presented in Table I. It is evident that the immobilization efficiency and the yield of E.coli C4 and chicken intestinal AF were determined to be very low for adsorption on chitin and magnetic chitosan, although the yield was rather low in most of E.coli C4 AP preparations as the remaining catalytic activity was reduced.

580 Table 1 Immobilization efficiency (rl) and yield (~,) of E.coli RY 13 and C4, and chicken intestinal AP immobilization. Initial activities of enzymes were: E.coli RY 13 0.320 U/mg of protein; E.coli C4 0.187 U/mg of protein; chicken intestinal 0.687 U/mg of protein. Other details described in Materials and methods. Carrier

Chitin Magnetic chitin Chitosan Chitosan Magnetic chitosan Magnetic chitosan Magnetite Magnetite

Silylated magnetite Silylated magnetite APS APS

Method of immobilization

E.coli RY 13

E.coli C4

Adsorption Adsorption

rI -

rI 0,006 0,600

3' 0,002 0,044

rl 0,060 0,070

0,036 0,037

0,130 0,150

0,039 0,038

0,110 0,220

0,064 0,047

0,030

0,011

0,080

0,035

"/ -

Adsorption Covalent binding Adsorption

Chicken intestinal

Covalent binding Adsorption Adsorption followed by stabilization with GA Adsorption

-

-

0,170

0,057

0,970

0,696

0,250

0,115

0,640

0,405

0,380

0,257

0,480

0,320

.

-

-

0,170

0,054

0,830

0,375

Covalent binding Adsorption Covalent binding

0,310

0,050

0,320

0,044

0,750

0,323

0,200

0,121

0,130 0,220

0,059 0,125

0,560 0,360

0,276 0,225

.

.

.

Magnetic chitin seemed to be a poor matrix for chicken intestinal enzyme but the adsorption efficiency of E.coli C4 one was as high as on the magnetite. It was interesting that covalent binding of E.coli C4 AP to silylated magnetite and APS was more efficient than adsorption, but the reverse dependence was observed for chicken intestinal enzyme (Table 1). The r I and ~, parameters for E.coli RY 13 AP covalent binding on modified magnetite and APS were similar to those for E.coli C4. Moreover, the first enzyme was fixed efficiently and with higher yield on pure magnetite after stabilization with GA in comparison with the adsorption only (Table 1). The reason of this effect is unclear. It is possible that after addition of GA the enzyme was more tightly fixed to the support as there was a rather little amount of protein found in washing fractions in comparison with this measurement for magnetite without GA. Summarizing, it is worth to note that the best results were obtained for enzymes adsorption on magnetite and covalent binding on APS as both rl and ~/were rather high on those carriers. But the most active preparations obtained were those of chicken intestinal AP

581 covalently bound on magnetic chitosan and silylated magnetite and also adsorbed on the latter one (Table 1). All preparations of fixed chicken intestinal enzyme were relatively more stable in comparison with those of both E.coli AP during three months storage at 4~ Chicken intestinal enzyme adsorbed on pure and silylated magnetite and APS lost 15, 26 and 35% of initial catalytic activity, respectively, while E.coli C4 preparations on those supports lost 30, 45 and 50% of the activity, respectively. E.coli RY 13 AP adsorbed on magnetite and treated by GA lost 18% of the activity under conditions mentioned but only adsorbed enzyme lost even 38% of that. Covalently bound alkaline phosphatases were more stable during the storage than adsorbed ones, and chicken intestinal enzyme preparations showed higher remaining activity in comparison with microbial ones as well. Chicken intestinal AP bound to magnetic chitosan was the most stable and showed even 98% of the initial activity. Chicken intestinal, E.coli C4 and RY 13 AP covalently bound to silylated magnetite lost 10, 31 and 36% of catalytic activity, respectively, and bound to APS 28, 40 and 47%, respectively. So, covalent binding seemed to be more suitable for AP storage stabilization. The reduced phosphatasic activity after three months in all preparations was not related to the loss of the immobilized enzyme as only negligible amount of protein was found in the reaction mixture (results not shown). The remaining activity of the three soluble enzymes after the storage not exceeded 10%, so the stabilizing effect of the immobilization was evident. Operational stability of few samples was also examined during one month operating immobilized preparations eight times (two times every week). Chicken intestinal enzyme covalently bound to magnetic chitosan showed the best results again as even 85% of initial activity was observed after a series of the experiment. The enzyme adsorbed on silylated magnetite and APS had 45 and 35% of initial activity, respectively, while covalently bound on the same supports 60 and 50% maintained, respectively. E.coli C4 AP showed lower operational stability on the supports mentioned (remaining activity of adsorbed on silylated magnetite and APS enzyme was 20 and 15%, respectively, and covalently bound on those supports was 40 and 20%, respectively). The most stable preparation of E.coli RY 13 AP was that adsorbed on magnetite with GA (remaining activity 43%). Rather high remaining activity and storage as well as operational stability of immobilized AP was noted by other authors [ 1519]. For example, the enzyme from calf intestine, covalently bound on glass activated by ),APTES, using azide method, showed more than 50% of initial activity during more than ten months at 4~ [15]. AP entrapped into chitosan gels had 50% of initial activity and lost only 5% of it when used four times during five days [ 17]. The optimum initial protein concentration determined to be I0 mg/ml for both E.coli AP and 20 mg/ml for chicken intestinal one. A quantity of 0.1 g of carrier and 1 ml of enzyme was determined to be optimal for reaction. Finally, 10 min was found to be the most suitable duration of reaction. Some properties of immobilized chicken intestinal AP were described earlier [20]. More detailed examination of the immobilization of enzymes mentioned is necessary for the practical approach.

4. CONCLUSIONS The immobilization of AP was determined to be influenced by the origin of enzyme, the nature and preparation of carder, and the method of fixation. Magnetic chitosan was better fit for covalent binding of chicken intestinal AP while pure and modified magnetite as well as

582 APS were suitable for all three enzymes examined. Covalently bound enzymes were relatively more stable during prolonged storage and repeated operation than adsorbed ones and chicken intestinal AP was more stabilized than both E. coli ones. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

V.Bales, Appl. Biochem. Biotechnol., 48 (1994) 5. V.Bringi, CEW Chem. World, 17 (1982) 64. P.L.Kronick, Meth. Cell Separation, 3(1980) 115. A.Garcia III, S.Oh. and C.R.Engler, Bioteehnol. Bioeng., 33 (1989) 321. V.Goetz, M.Remaud and D.J.Graves, Biotechnol. Bioeng., 37 (1991) 614. Z.J.A1-Hassan,J. Ferm. Bioeng., 71 (1991) 114. V.G.Bendikiene, I.G.J.Pesliakas and V.S.Vesa, Appl. Biochem. Microbiol. (Russia), 17 (1981)441. E.P.Feofilova, Ibid, 20 (1984) 147. V.G.Bendikiene, B.A.Juodka, R.M.Kazlauskas, S.S.Tautkus, E.L.Matulionis and A.A.Sudaviehyus, Appl. Biochem. Mierobiol. (Russia), 31 (1995) 335. A.I.Kestner, Uspechi chimii (Russia), 43 (1977) 1480. R.F.Schleifand, P.C.Wensink, Practical Methods in Molecular Biology, Springer-Verlag, New York, Heidelberg, Berlin, 1981. A.D.Hall and A.Williams, Biochemistry, 25 (1986) 4784. X.Xu and E.R.Kantrowitz, Biochemistry, 30 (1991) 7789. M.J.Kunitz, Gen. Physiol, 43 (1960) 1149. F.Pittner, Biochem. Biotechnol., 3 (1982) 105. G.G.Brownlle and F.Sanger, Eur. J. Biochem., 11 (1969) 395. J.Wiley, Biotechnol. Bioeng., 21 (1979) 711. L.Grasset, D.Cordier, R.Couturier and A.Ville, Biotechnol. Bioeng., 25 (1983) 1423. K.Jiro, J.Kenji and G.Shinichi, Bioteehnol. Bioeng., 1 (1984) 100. B.R.Surinenaite, V.G.Bendikiene and B.A.Juodka, Appl. Biochem. Microbiol. (Russia), 32 (1996) 547.


583

Stabilization o f serine proteases by immobilization V.Bendikiene and B.Juodka Department of Biochemistry and Biophysics, Vilnius University, Ciurlionio 21, 2009 Vilnius, Lithuania

Water- insoluble serine proteases derivatives from Acremonium chrysogenum, Bac. subtilis, and trypsin were prepared by immobilizing onto the surface of magnetic chitin and chitosan by adsorption and covalent fixation. Magnetic carriers were prepared by direct posmagnetization with ferrofluid and/or deposition of fine magnetite particles from Fe3+/Fe2+ salts mixture. Qualitative and quantitative investigations have been provided with the aid of AAS, IR and Mossbauer spectroscopy, SEM methods. Total iron amount in magnetic products of chitin and/or chitosan depended on the nature of the preparation. The stabilization of trypsin and serine proteases I and II from Acremonium chrysogenum against thermal inactivation and autolysis was achieved by optimization of the immobilization conditions. The pH, thermal and storage stability of the immobilized proteases were higher than those of free ones.

1. INTRODUCTION Magnetic supports for enzymes immobilization were first used by Robinson et al. [ 1]. The various types of magnetic supports and immobilization methods that have been used are described in review [2]. The interest in the immobilized enzymes and their application to bioprocessing, analytical system [3] and enzyme therapy [4] has steadily grown in the last decade. In order to recover enzymes magnetically, they have been immobilized on a magnetite body [5] or on a support containing magnetite [6]. In this regard, chitin and its deacetylated product chitosan are considered to be appropriate supports for immobilizing both magnetite and enzyme. In this paper stabilization of various serine proteases by immobilization on magnetic derivatives of chitin and/or chitosan was studied.


2.1. Reagents Chitin (TY6-09-05-141-79 "Reakhim") and ferrofluid (Krasnodar Institute of Gas and Oil) were purchased from Russia; glutaraldehyde and Na-benzoil-DL-arginine-p-nitroanilide (BAPNA) were from Merck (Darmstadt, Germany); crystallized trypsin was purchased from Spofa. Bac. subtilis serine protease was from Fermentas (Vilnius, Lithuania) and Acremonium chrysogenum serine proteases were from Moscow Antibiotics Institute (Russia). All other reagents were of analytical grade. 2.2. Equipment Qualitative and quantitative studies of magnetic derivatives of chitin and chitosan including the determination of chemical composition of magnetic particles, Fe3+/Fe2+ ratio, mean magnetite size, its magnetic moment, concentration and distribution on the chitosan surface and in the deeper layers of porous chitin were estimated with the aid of atomic absorption spectrometry, IR and Mossbauer spectroscopy, light and scaning electron microscopy (SEM) and magnetometry methods [7,8]. 2.3. Magnetizable particles Chitin was purified by demineralization and deproteinization steps according to [8]. Chitosan was prepared by heating pure chitin in 5N KOH solution (ratio 1:10) in presence of 1% NaBI-I4 [7]. The Fe304 microcrystals were prepared by coprecipitation of a 2:1 molar mixture of a ferric and ferrous salt in aqueous solution (solution A) under oxygen-free conditions. To produce magnetizable particles 2 methods were used: 1) postmagnetization of chitin and chitosan with ferrofluid, 2) precipitation of magnetite particles from Fe /Fe salts solution mixture on the surface of solid polisaccharides [7,8]. 9

3+

2+

9

-

2.4. Determination of esterase activity The esterase activity of trypsin or immobilized trypsin preparations was determined by the method of Erlanger et al. [9] with 0.05M Tris buffer, pH 8.2, containing 0.02M CaC12 and 5% dimethylsulphoxide. One unit (U) of enzyme activity was defined as that amount of enzyme which liberated l~tmole ofp-nitroaniline in 1 min at 37~ 2.5. Determination of caseinolytic activity The caseinolyticactivityof soluble and immobilized proteaseswas determined using 2% casein solutionaccording to Bergrneyer [l0]. 2.6. Immobilization of serine proteases A typical immobilization procedure is as follows: 100 nag of support was suspended in 5 ml of buffer at different pH. A given amount of enzymes was added to the suspension under sharing or stirring with magnetic stirrer for nonmagnetic and with mechanical for magnetic supports. A 25% glutaraldehyde solution (0.01-0.1 ml) was added and the stirring was continued for various periods of time (15-240 min). A 60 min reaction time was found to give rise to maximum esterase activity of immobilized trypsin and 120 min reaction time was found to

585 give rise to maximum proteolytic activity of immobilized serine proteases from Acremonium

chrysogenum. Trypsin was covalently bound on supports activated using 25-30 fold excess of the GA with respect to 1 mole of N H 2 groups on carriers surface to avoid the "crosslinking" of two carriers or enzymes molecules. GA- activated beads were contacted with trypsin solutions (10-50 mg/ml). The suspension was mechanically stirred for 10-24 h at 4~ and/or at 20~ The carrier was washed with buffer until the supernatant was free of esterase activity. 2.7. Stability measurements The thermal stability of the immobilized proteases was evaluated by measuring the residual activity of proteases treated at various temperatures (60~176 in buffer for various periods of time (15-60 min). After heating, the samples were rapidly cooled and its enzymatic activity was assayed at 37 ~ immediately or after storage. To determine the pH stability, the soluble and immobilized proteases were incubated with the substrate solution in buffers with appropriate pH. For determination of the temperature influence on enzyme activity, duplicate samples of the dissolved and immobilized enzyme, respectively, were added to the substrate solution which had been equilibrated at the desired temperature, and the initial rate of hydrolysis was measured.

3. RESULTS AND DISCUSSION We attempted the immobilization of serine proteases on magnetite (Fe304) and five supports based on chitin and two supports based on chitosan. Magnetic polisaccharide carriers varied in amount of magnetite (Fe%) and size of particles. Magnetic derivatives of chitin and chitosan used: magnetic chitin (I)- amount of Fe 3.75%, size of particles (0.6-0.4 mm); magnetic chitin (II)- Fe 9.0%, (~

100

;>

80

~

60

N

40

40

~

20

20

~

,~ 80 t~

"3 60 ~D 12h t/2

.>. ~9

0

0 0

20 40 60 Incubation time (h)

0

20 40 60 Incubation time (h)

80

Figure 1. Sitosterol side-chain cleavage activity decay in free (a) and immobilized (b)

Mycobacterium sp. cells incubated at 30~ with no added medium (A), in pH 7 phosphate buffer (Tq) or BEHP (A,II,O). Cells were harvested from the incubation medium by centrifugation ([:],l) or filtration (A) and assays for activity in a 5 gL -1 solution of sitosterol in BEHP. Alternatively, the activity assay was carried out in the incubation medium by adding sitosterol powder (O). These trials used a free cell load equivalent to 960 mgprotein/L was used, with an initial specific activity of 0.042 ~tmolAD/(h.mg protein), and an immobilized cell load equivalent to 900 mgprotein/L, with an initial specific activity of 0.080 ~tmolAD/(h.mg protein). Avoiding the cell harvesting step between incubation and the activity assay produced no change in cell catalytic stability (Figure l a). Parallel tests were carded out with Celite-

628 adsorbed cells (Figure l b). As with free cells, incubation in BEHP led to a slightly higher activity retention than incubation in the phosphate buffer. No stability difference was 120.~

100

.~

80= o L~

"3

60-

o .5

40 200-

I

0

20

'

I

40

'

~

I

60

~

I

80

~r:

100

Incubation time (h) Figure 2. Effect of cell load on the support on sitosterol side-chain cleavage activity decay of

Mycobacterium sp. cells immobilized on Celite, incubated in BEHP. Cell loads equivalent to 6.0 mgprotein/gdrysupport ([-7) and 12.6 mgprotein/gdry support (~k) were tested. observed between cells incubated in buffer and in the absence of added liquid, indicating that excess water in itself does not contribute to activity retention. In both forms, the whole-cell biocatalyst stability was not negatively affected by the organic solvent, suggesting that a possible solvent toxicity effect is not the main deactivation factor in the present conditions. However, it may be playing a role, since morphological changes have been previously reported with this same biocatalytic system after cell exposure to the solvent [9]. It should also be noted that in BEHP, immobilized cell activity retention is higher than with free cells in the first 20 hours of incubation. However, this stabilization cannot be attributed to a diffusional effect, since initial activity levels are also higher in the immobilized system. Supporting this observation, a decrease in the cell load on the immobilization support from 12.6 to 6.0 mgprotein/gdry support did not lead to a reduction in the catalytic stability upon incubation in BEHP (Figure 2).The catalytic stability of free and immobilized Mycobacterium cells was also evaluated in BEHP medium, in the presence of the substrate (sitosterol, 5 gL "1) (Figure 3). Stability levels were significantly increased when sitosterol was present in the media, particularly for longer incubation periods. This difference could have been due to the formation of ATP resulting from the side-chain cleavage of sitosterol [12], thus partially preventing depletion of the oxidative potential of the cells, as opposite to pure BEHP medium, which is, in principle, non-nutritive. In order to evaluate the possibility of reactivation of the sterol-degradation capacity of the whole-cell biocatalyst, free Mycobacterium cells were used in repeated, batch biotransformations of sitosterol (1 gL -l) in BEHP, being washed with a complex nutrient solution (yeast extract, 10 gL -l) between batches. A similar washing procedure with phosphate buffer was carried out in parallel runs, as a blank (Figure 4). The higher residual activities observed when yeast extract was used for the intermediate washings, as compared to phosphate buffer,

629 could be attributed to the retention of nutrients by the cells, partly fulfilling cell maintenance requirements, since the duration of the intermediate incubations does not allow for cell

growth.

120

120

a

100

L

b

100

~"

80

80

~

60

60

u~

~

40

40

~

N.

2o

20

-~

~

=',~

0 0

20

40

60

Incubation time (h)

0

20

40

60

80

Incubation time (h)

Figure 3. Sitosterol side-chain cleavage activity decay of free (a) and immobilized (b)

Mycobacterium sp. cells incubated in BEHP (II,A) or in a solution of sitosterol (5 gL -1) in BEHP (A,D). Cell loads and initial specific activities as in figure 1.

Figure 4. Repeated, batch biotransformations of sitosterol to AD in BEHP, with free

Mycobacterium sp. cells, either washed with yeast extract solution (filled bars) or with phosphate buffer solution (open bars). Cells were washed before the first and between biotransformations Co) or only between biotransformations (a). Each biotransformation lasted 24 hours. Complete bioconversion would correspond to 2.4 mM AD produced.

630 4. CONCLUSIONS Low catalytic stability levels were observed in sitosterol side-chain cleavage with either free or Celite-adsorbed Mycobacterium sp cells. Deactivation could not be attributed to defficient hydration of the biocatalyst or the presence of the organic solvent (BEHP). In organic media, immobilization produced a slight stabilization effect, apparently not related to diffusional limitations. Stability was improved both in the presence of the substrate and when nutrients were supplied to the free cells between successive batch bioconversions in organic medium, suggesting that the depletion of the cellular oxidative potential could be the cause for the inactivation of the sitosterol side-chain cleavage pathway. 5. ACKNOWLEDGMENTS P. Femandes gratefully acknowledges grant BD/9662/96 from Funda@~io para a Ci6ncia e Tecnologia, Portugal. This work was supported by grant no. Praxis/2/2.1/BIO/37/94, from the PRAXIS XXI programme, MCT, Portugal. REFERENCES

I. Cabral, J.M.S., Aires-Barros, M.R., Pinheiro, H. and Prazeres, D.M.F., J. Biotechnol. 59 (1997) 133. 2. Green, K.D., Gill, I.S., Khan, J.A. and Vulfson, E.N., Biotechnol. Bioeng., 49 (1996) 535. 3. Kannan, T.R., Sangiliyiandi, G. and Gunasekaram, P., Enzyme Microb. Technol., 22 (1998) 179. 4. Goetschel, R. and Bar, R., Enzyme Microb. Technol., 14 (1992) 462. 5. Salter, G.J. and Kell, D.B., Critical Rev. Biotechnol., 15 (1995) 139. 6. Mahato, S.B. and Garai, S., Steroids, 62 (1997) 332. 7. Pinheiro, H.M. and Cabral, J.M.S., Enzyme Microb. Technol., 14 (1992) 619. 8. Hocknull, M.D. and Lilly, M.D., Enzyme Microb. Technol., 10 (1988) 669. 9. Dias, A.C.P., Cabral, J.M.S. and Pinheiro, H.M., Enzyme Microb. Technol., 16 (1994) 708. 10. Lowry, O.H., Rosenbrough, N.J., Farr, A.L. and Randall, R.J., J. Biol. Chem., 193 (1953) 265. 11. Gyiire, I., Lenkey, B. and Szentirmai, A., Biotechnol. Lett., 15 (1993) 925. 12. Szentirmai, A., J. Ind. Microbiol., 6 (1990) 101.


631

Operational stability of immobilized C. reinhardtii cells: an approach to its potential as biocatalyst for N-consuming processes I. Garbayo and C. Vilchez Dpt. Quimica, E.P.S. La R~bida, Universidad de Huelva 21819 Huelva, Spain

1. INTRODUCTION It is well known that the presence of inorganic nitrogen in any of its different forms (nitrate, nitrite, ammonium) decreases the quality of drinking water (1). However, this problem cannot be easily avoided since these chemicals use to be present in the composition of compounds used in agriculture, activity that sustains the economy of many countries. The development of both cheap and efficient biologic treatments leading to remove them is a matter currently being investigated (2,3). Many microorganisms are able to consume, or to modify, a sort of chemicals which appear in drinking water, including nitrate, phosphate and sulphate (4,5,6). Whilst some of these microorganisms can modify the form of many contaminant molecules, another type, microalgae, may utilize nitrate, nitrite or ammonium as nitrogen source, then incorporated into carbon skeletons as glutamine and glutamate, used to biosynthesize other aminoacids and proteins. This assimilation occurs by the succesive action of several enzyme activities (nitrate reductase, nitrite reductase, glutamine synthetase and glutamate synthase) which constitute the assimilatory pathway of inorganic nitrogen in microalgae (5,7). This ability could be of special interest for designing new cheap and clean biologic systems to remove nitrogen compounds from water. Under the practical point of view, the use of freely suspended cells in ponds could result in many problems ranging from that only discontinuous systems can be run to the low control of the biologic system, which is continuously modified because of the cell growth and the subsequent increase in the biomass (8). Immobilization provides an useful tool that allows both to work with high biomass content and to maintain stable operational conditions (9). Depending on the type of process, cells could be immobilized mainly by either, entrapment or adsorption. At lab scale, most frequently used polymers are, alginate and carrageenan for entrapment, and different foams (PU,PV) for adsorption (10,11). Immobilization provides additional stability to microalga cells for nitrogen consuming. An approach to this through operation conditions and kinetic considerations is reviewed in this paper from our experimental data in nitrate and nitrite lab-scale consuming processes by using C. reinhardtii cells immobilized in alginate.

632 2. METHODS

2.1. Organisms and standard culture conditions Chlamydomonas reinhardtii, wild type, strain 21 gr, was grown at 25~ in 15 mM phosphate (pH 7.5) buffered culture medium containing 10 mM KNO3 as the sole nitrogen source. The standard cultures, in 250-rnl conical flasks, were bubbled with air contining 5 % (vol/vol) CO2 and continuously illuminated with white fluorescence lamps (250 mE m2s~ at the surface of the tube). The cells were harvested during the exponential growth phase (15 mg Chl mr 1) by centrifugation at 5,000 g 5 min. 2.2. Immobilization of C. reinhardtii cells by entrapment in alginate The cells were harvested, washed, and resuspended (0.5-1%, wt/vol) in 20 mM TricineNaOH (pH 8.0) buffered culture medium, and were thoroughly mixed with an equal volume of an alginate solution (3 %, wt/vol) prepared by mixing 2 g of alginic acid and 1 g of alginate sodium salt in 100 ml of medium, adding NaOH to reach pH 6.5-7.5. The final viscosity (7,000 cp) depended on the proportion of alginic acid and alginate mixed. Beads of about 3ram in diameter were obtained by dropping the alginate cell mixture into a solution of 0.1 M CaC12, and after half an hour they were rinsed with flesh culture medium and ready for using. 2.3. Nitrate and nitrite uptake conditions Nitrate and nitrite uptake experiments were carried out at 25~ in small reactors (1.3 1 batch reactor and 2.3 l airlift reactor) of transparent glass (Fig. 1), operating in either, discontinuous or continuous mode. The cells were continuously illuminated with white light (250 mE.m2.s~), and the bed was fluidized with air only. The medium supplied to the reactor contained Tricine-NaOH buffer supplemented with nitrate or nitrite and 5-10% (wt/vol) of immobilized cells (depending on the experiment).

2.4. Analytical determinations Chlorophyll (Chl) was determined by extracting the free cells with acetone. For immobilized cells, the beads were extracted with methanol or with acetone overnight. After removing the non-extracted material by centrifugation, the absorbance at 652 nm was determined in the supernatant (e = 34.5 mgI ml crnl). More details in Vilchez et al. (1995). Nitrate in the medium was determined according to the method of Cawse, and nitrite following the method described by Snell and Snell (12).

633

Reactor

Figure 1. Bioreactorsused for nitrate and nitrite uptake extxriments in continuousand discontinuousprocesses. Reactor 1: airlift loop (2,31.) Reactor 2: fluidizedbed (1,31.)

Reactor

Pump (n~um)

3. RESULTS AND DISCUSSION Although the ability of C. reinhardtii to utilize inorganic nitrogen has been well established in the literature (7), however only a minor number of papers considering the biotechnologic utilization of this capacity as a valid tool in water treatment have been published (11). The purpose of this paper is presenting no an useful and satisfactory Nremoval microalga system, but the chance to make a cell system stable for long time processes and under different conditions, which would be useful for N-consumption in this case. The choice of alginate as material for cell entrapment is widely supported by several authors, since this polymer combines all the requirements needed to immobilize microalgae cells: it is transparent, non-toxic, porous and rarely reactive (13). 3.1. Optimal conditions for nitrate and nitrite uptake Before studying the N-consumption in either, continuous or discontinuous system, the optimal conditions for a maximum uptake rate by C. reinhardtii immobilized in alginate were determined. They are shown in Table 1.

634 In general, alginate concentration can not be lower than 3 % because beads could get easily disrupted after one or two days of being shaken in a reactor. Higher concentrations would lead to lower consumption due to diverse limitations (13). However, although as for lab work alginate is an adequate matrix, the development of new cheap and nonbiodegradable polymers would be of interest, and is being currently studied (14).

Table 1 Optimization of parameters for maximum nitrate and nitrite uptake by immobilized C. reinhardtii cells Parameter

Optimum nitrate uptake

Optimum nitrite uptake

Alginate

3%

3%

Initial Cell loading

50 mg Chl.g~gel

50 mg Chl.gagel

Temperature

30~

35 ~

pH

7

7.5

ks (estimated)

200

80

Optimal cell loading is very low because in the experiments to characterize the optimal conditions for N-consumption, cells were used just immobilized, not grown inside the polymer. When increasing the immobilized cell loading, a considerable part of the biomass inside the beads is not efficient for nitrogen consuming, due to the shading effect of external cells and diffusion limitations for substrates (9). This problem is avoided when the immobilized cells are used in long time processes. They grow inside the beads, and a maximum cell loading in the stationary growth phase is reached, maintaining its stability over long time (see below, Cell growth and N-consumption in reactors)(12). Temperature and pH become critical to determine the advantages of immobilized cell systems. The intervals within which the cells maintain more than 80% of the maximum consumption rates are extense, ranging from 20-40~ for temperature and 6-8,5 for pH. This means that there is a significant difference between using either, freely suspended or immobilized cells. For the first ones, the consumption rates become very low when temperature or pH are not the optimal values (9). The aff'mity of the immobilized cells for nitrate and nitrite clearly diminishes in comparison to those freely suspended ones (100-200 fold). However, ks values for nitrate and nitrite are still very low if the nitrogen concentration present in contamined waters are taken in account (usually much higher than these ks values). The highest affinity for nitrite could be explained in terms of the electric interaction between nitrate and the charged

635 polymer, CaE+-alginate (13). It could be stronger than that between nitrite and alginate, due to the presence of an additional oxygen, resulting in a major difficulty for nitrate availability by the immobilized cells. On the other hand, both nitrate and nitrite uptake are inhibited by the presence of ammonium, which is preferently consumed. If nitrate or nitrite are the contaminants, this competition would reduce the potential use of immobilized C. reinhardtii cells if ammonium is also present.

3.2. Operational stability: cell growth and N-consumption in reactors Although maximum nitrate and nitrite uptake rates were found at 30 and 35~ respectively, these would not be the best conditions to operate long time processes. At the indicated temperatures, photosynthetic activity reachs maximum values, and the corresponding nitrogen assimilatory enzymes (nitrate and nitrite reductase) show usually high activities, but the cell state can not be maintained stable for long periods of time. Since the uptake rates in immobilized cells are not significantly modified in a wide range of temperature (see above), for practical purposes, it is recomended to operate at lower values (e.g. 25~ Since one of the advantages in using immobilized cells is also the wide range of pH that allows the cells to operate at high consumption rate, similar considerations could be done. Most of water flows show pH ranging between 5 and 7, having the immobilized cells high capacity of nitrogen consuming within this interval (9,12).

Table 2 Operation conditions for stable long-term N-consuming discontinuous processes" a comparison between free and immobilized cells Parameter

Free cells

Immobilized cells

Cell concentration (as Chl)

30 ~tg.ml1 (a)

300 ~tg.g~ (2 x a)

Temperature

22-30 ~C

20-40 ~C

pH

6,5-7,5

6-8,5

Activity after storaging

0%, 3 days

90%, 1 week

Concerning the cell loading, it is important to note that for either continuous or discontinuous processes, the cells have been previously grown in the beads. Then, the biomass profile shows a maximum concentration of colonies in a thin shell of polymer close to the bead surface. Both nutrient and light diffusion limitations determine that cell distribution. A maximum cell loading ranging from 100-140 mg Chl.g 1 gel was obtained for lab-scale experiments, but since it depends on the initial load in the beads, higher values could be obtained for practical purposes (4,11).

636 A comparison between the results obtained in lab-scale N-consuming processes by using both freely suspended and immobilized cells is shown in Table 2. It shows the additional stability that the immobilization provides to C. reinhardtii cells.

Table 3 Operational stability of C. reinhardtii immobilized cells in N-consuming processes

Nitrate uptake

Nitrite uptake

Parameter

Discontinuous

Continuous

Discontinuous

Continuous

Operation time N~ Initial rate (~tmol/mg Chl.h) Final rate/ Initial rate Dilution rate Inhib. by NH4 § Inhib. by NO2 N-conc. Intl. N-conc. Eft.

35 days 5 2,5

20 days

21 days

3

10 days 1 6,9

0.7

1

0,8

0.9

Yes Yes 3 mM

20 h -1 Yes Yes 3 mM 1,5 mM

Yes

20 h -1 Yes

6 mM

7,6

2 mM 0,6 mM

Some parameters led to determine the operational stability of C. reinhardtii immobilized cells were followed in both continuous and discontinuous processes for nitrate and nitrite consumption. As it is inferred from Table 3, discontinuous processes were carried out at least for three weeks. After this period, nitrogen conversion rate (final rate~initial rate, ratio of consumption) hardly changed compared to that showed at the begining of the experiments. Those discontinuous experiments for nitrate consumption were performed along six cycles of five-six days. Each cycle f'mished when nitrate concentration decreased down to 1,5 mM. In doing so, immobilized cells operated at the highest uptake rate along the experiment. At lower concentrations, the nitrate uptake rate decreased significantly. Continuous processes were carried out along three weeks, maintaining the same consumption rates at the end of this period. Ammonium inhibited nitrate and nitrite consumption as it was observed in freely suspended cells, but in a less strong way (data not shown). However, there is an important point that has to be noticed in these kinetic experiments carried out within reactors: the concept of nitrogen removal changes when

637 immobilized cells are used for nitrogen consuming processes. In continuous processes, the flow of medium can be set to keep N-concentration in the effluent as high as to get maximum uptake rates. Table 3 shows that nitrate concentration in the effluent was 1,5 mM for 3 mM in the influent, in nitrate consumption continuous processes. Nitrite concentration in the effluent was 0,6 mM for 2 mM in the influent, in nitrite consumption continuous processes. These results illustrate the ability of immobilized microalgae to sustain stable N-consuming continuous processes, but also indicate that they are not able to eliminate completely the N-contaminant from the influent, since the affinity of the immobilized cells for nitrate and nitrite decreases compared to those showed by freely suspended cells (around l l.tM), and the N-concentration within the reactor has to be maintained around the values above discussed to sustain maximum uptake rates.

4. CONCLUSIONS With all these considerations, from the characterization of N-consumption conditions, and from these discontinuous and continuous systems performed at the lab for nitrate and nitrite removal by immobilized C. reinhartii cells, the following general could be extracted: 1. Immobilized cells maintained stable at least over a period of three weeks. The biologic system does not need modifying or replacing by another new one within this period. 2. The differences observed between nitrite and nitrate consumption could be explained as a function of experimental conditions and the electric interaction between nitrate and the negatively charged polymer (alginate). 3. The immobilized cells operate with high consumption rates within a wide range of pH and temperature, which theoretically allow great improvement in the control of a real system. Then, the process stability becomes rather major than for freely suspended cells. 4. The immobilization provides a stable system for C. reinhardtii cells, able to consume nitrate/nitrite, at similar rates, after seven days of storage at 4~ 5. Since the affinity of the immobilized cells for nitrate and nitrite diminishes clearly in comparison with freely suspended cells, a certain N-concentration should remain in the effluent in long-term processes to allow the cells getting the maximum uptake rates. The removal concept has to be changed by contaminant consumption, getting it down to the maximum levels allowed by law.

5. REFERENCES 1. C.W. Mackerness et al. M.J. Hill (ed.). In: Nitrates and nitrites in food and water, Ellis Horwood, England, 1991. 2. W.M. Rostron et al. In: R.H. Wijffels et al. (eds.), Immobilized Cells: Basics and Applications, Elsevier Science B.V., Netherland, 1996. 3. K. Tanaka. In: R.H. Wijffels et al. (eds.), Immobilized Cells: Basics and Applications, Elsevier Science B.V., Netherland, 1996.

638 4. J. de la Notie et al. In: R.D. Tyagi et al. (eds.), Wastewater treatment by immobilized cells, CRC Press, USA, 1990. 5. J.M. Vega et al., Trends Photochem. Photobiol., 2 (1991) 69. 6. R.H. Wijffels et al., Enzyme Microb. Teclmol., 17 (1995) 482. 7. J.M. Vega et al. In: W.R.Ullrich et al. (eds.), Inorganic nitrogen in plants and microorganisms, Springer-Verlag, Germany, 1990. 8. D. Proulx et al. In: M. Moo-Young (ed.), Bioreactor immobilized enzymes and cells. Fundamentals and applications, Elsevier Applied Science, England, 1990. 9. C. Vilchez et al., Enzyme Microb. Technol., 17 (1995) 386. 10. C. Garbisu et al., J. Appl. Phycol., 3 (1991) 221. 11. C. Vilchez et al., Enzyme Microb. Technol., 20 (1997) 562. 12. I. Garbayo et al. In: R.H. Wijffels et al. (eds.), Immobilized Cells: Basics and Applications, Elsevier Science B.V., Netherland, 1996. 13. O. Smidsrod et al., Trends Biotech., 8 (1990) 71. 14. A. Muscat et al. In: R.H. Wijffels et al. (eds.), Immobilized Cells: Basics and Applications, Elsevier Science B.V., Netherland, 1996.


Induction and stability of cholesterol oxidase Rhodococcus

639

from cells of a

*

Joseph Kreit at and Pierre Germain b a Science Faculty, D e p a r t m e n t of Biology, B P 1014, Rabat, Morocco b LFBI, ENSAIA, BP 172, Vandoeuvre l~s Nancy, France Cholesterol oxidase (COX; EC 1.1.3.6) is sterol-inducible enzyme. In cells of

Rhodococcus sp. GK1, a soil-isolated strain, the sterol lateral chain is necessary for the induction process. Depending on growth media, this strain produced the enzyme in a cell surface-linked form together with a secreted form. COX forms are monomeric with a molecular mass around 60 kDas. They are thermostable a t t e m p e r a t u r e up to 50 ~ C. 1. I N T R O D U C T I O N

In the sterol degradation pathway by the nocardioforms, cholesterol oxidase (COX; EC 1.1.3.6) is the first enzyme acting on the ring A, converting the 3[~-ol-5en structure into the corresponding 4-en-3-keto derivative. Apart from this transformation, there is no other reaction t h a t is compulsory for the microbial a t t a c k on either end of the sterol molecule (1-2). The ring system and the side chain of sterols are catabolized independently but simultaneously. The same steps (but not necessarily the same enzymes) are involved in the degradation of the ring system of either sterols or steroids of the androstane or pregnane series. Obviously, steroid catabolism affects COX synthesis. COX of Arthrobacter simplex (3) and t h a t of Rhodococcus erythropolis (4) were induced with cholesterol. Enzyme induction in species of Nocardia was obtained with sitosterol, cholesterol or with cholesterol esters (5). Yet, the precise chemical structure of steroids responsible of COX induction had been unknown. During our work on steroid biodegradation, a strain of the genus Rhodococcus was screened from soil for its capability to degrade plant sterols. It was given the name R. sp. GK1, and deposited with the Pasteur-Institute Collection (Paris) under the n u m b e r CIP105335. A secreted and a cell-bound COXs were isolated from this bacterium and characterized (6-7). Another enzyme with industrial potential, a secondary alcohol NAD-dependent dehych-ogenase, was obtained from the strain cells (8). The present paper is aimed to present data concerned with the kinetics of COX production by R. sp. GK1 cells, and the enzyme stability.

* This article is dedicated to the memory of professor G~rard Lefebvre (ENSAIA, Nancy, France), born on April 25th, 1939 and died on January 2nd, 1997. t Correspondence to J. Kreit.

640 2. R E S U L T S AND DISCUSSION 2.1. C h o l e s t e r o l o x i d a s e i n d u c t i o n The kinetics of COX production from R. sp. GK1 cells, were investigated by culturing it in a medium of mineral salts with ammonium sulfate as the nitrogen source (6) and different steroids (fig.l). These steroids were used each as the sole carbon and energy source, unless otherwise noted. Microbial growth was allowed aerobically at 28-30~ in Erlenmeyer-flasks under mechanical shaking. Other growth conditions, and methods were cited elsewhere (7-9). The phytosterols, cholesterol, 5-pregnen-3~-ol-20-one and 5-androsten-3~-ol-17one were demonstrated to be substrates for COX (fig.l). A representative example of the data obtained for cells grown on the phytosterols or on 5-androsten-3~-ol-17one is given in figure 2. As shown, the enzyme amount (u/L) increased with the microbial growth on the phytosterols, and reached a maximum in the beginning of the stationary phase. The profile of the specific activity of COX (u/g dc) exhibited a maximum in the late growth and another maximum sited in the beginning of the stationary phase. Cells grown on 5-androsten-31~-ol-17-one started to produce COX only from the late growth phase. The specific activity of enzyme, estimated at the maximal growth, was by 10 times lower than that of the phytosterolsgrown cells. The profiles of growth and of enzyme production observed for cholesterol-grown cells were similar to those of the phytosterols-grown cells, except that COX amounts were by 2 times lower. Cell gowth on 5-pregnen-3~-ol20-one resulted in the same observations as those described for the growth on 5androsten-3~-ol-17-one. COX amounts estimated at the maximal growth of cells on the phytosterols, cholesterol, 5-pregnen-3~-ol-20-one and 5-androsten-3~-ol-17one were respectively: 170, 65, 22, and 20 u/g dc. These results supported that enzyme synthesis is inducible by sterols t h a t have a side chain moiety, and the 3B-ol-5-en group alone is not sufficient for induction. On the other hand, the side chain complexity seemed to increase the level of COX production as the enzyme amounts induced with the phytosterols were higher than those induced with cholesterol. Growth of the microorganism on either 4-androsten-3,17-dione, testosterone (steroid nucleus) or 4-cholesten-3-one does not need COX activity. Then, an appropriate way for studying the effect of the sterol side chain on enzyme production is to use these keto-steroids for culturing. The strain grew well on 4androsten-3,17-dione or on testosterone. However, no COX activity was observed. COX production from 4-cholesten-3-one-grown cells started since the beginning of growth, and the profiles of the enzyme production were comparable to those obtained for cholesterol-grown cells. This enzyme production can be related to the side chain, for the reason that the structural difference between 4-cholesten-3-one and 4-androsten-3,17-dione, or testosterone, is only this chain. Besides, catabolism of the steroid nucleus seemed to repress COX synthesis. The enzyme system that accomplishes the first microbial attack on the cholate nucleus is different from that of the sterols, and catabolism of the whole molecule of cholic acid does not involve COX (6,10-11). h~ contrast, microbial catabolism of bile acids and that of sterols have many common reactions comprising those of the B-oxidation mechanism, responsible for cleavage of their side chains. Considering these facts, cholic acid appeared to be an interesting steroid molecule to demonstrate the effect of the alkyl chain on COX induction.

641 CH3

~T ~CH3

OH Cholesterol (1 00%) 0

~0~

5-Androsten-3~-ol17-one (20%)

CH3

o.

o

5-Pregnen-3~-ol20-one (25%)

4-Cholesten-3-one

OH

0

CH3

-

0

Testosterone 0

COOH

O H . . . ~

OH--Cholic acid

4-Androsten-3,17-dione

5-Androstan-3 (~-ol-17-one

Figure 1. Steroids used for COX induction in R. sp. GK1 cells. Phytosterols" a mixed product consisted of 56% I~-sitosterol (R=C2Hs), 27% campesterol (R=CH 3) and 10% stigmasterol (R=C2Hs, A22). Values in parentheses are the relative activity of CO~ 101 Growth:

u/L or u/g dc

A

9

B

.200

[]

1

100

O.1

o

z'o

4'0

6o

Growth time (h)

2O

O 6O

Figure 2. Kinetics of COX synthesis by R. sp. GK1 grown on sterols. A mineral salt medium was used. It contained: A, 2.5 g phytosterols/L. B, 3 g 5-androsten-3~-ol17-one/L. o, o, enzyme unit (u)/L. A, A, enzyme u/g dry cells (dc). m, n growth. One enzyme unit catalyses 1 ~mol substrate/min.

642 The enzyme production from cells grown on cholate took place during the phase of slowing growth, and rose up to the maximum in the stationary phase (fig. 3 A). The amount of the produced COX (110-120 u/g dc) was relatively high. Catabolism of cholic acid, that occurred during the exponential growth, seemed to repress enzyme synthesis. In other experiments, the transcription inhibitor rifampicin was added (50-150 rag/L) into cultures of cholate-grown cells at a point where the growth and the enzyme level attained the maximum (fig. 3 A). Both parameters were afterwards followed for a further incubation time. In the presence of the antibiotic, the enzyme level decreased gradually, while the biomass remained unchanged. These results suggested that COX level was maintained constant during the stationary phase under the effect of two opposite phenomena, a possible degradation, and transcription leadingto a new biosynthesis. Further, the microorganism was cultured on cholate together with phytosterols (fig. 3 B). The initial growth proceeded with an equal rate in either the assay or the control culture, and lasted, under the applied conditions, 16 hours. COX, which is necessary for the first transformation of sterols, was produced at the end of this period in both media. Also, the additional growth in the assay culture, started at this end, due to the degradation of phytosterols. The inducing effect of sterols manifested in the second growth, causing an increase in the enzyme amount (u/L). These data supported occurrence of the repressive effect on COX synthesis during the early degradation of cholate. Actually, species of Mycobacterium , Corynebacteriu m and Arthrobacter degrade the ring system of cholic acid before shortening the side chain, as evidenced from the isolation of intermediates containing this chain (10-11). The present microorganism is taxonomically related to these species, and presumably it could degrade cholic acid by a similar way. Thus, synthesis of COX could be as a result of two facts occurring since the phase of the slowing growth: Derepression because the repressing structure in the cholate molecule had been degraded, and induction due to some intermediate that may contain the side chain. Although this prediction is mainly based on the literature data, it may be important for fttrther investigation. Interestingly, growth of R. sp. GK1 on 3a-hydroxy-androstan-17one, a steroid resembling the cholate nucleus, resulted in no production of CO~ 2. 2. P r o d u c t i o n of multiple forms of c h o l e s t e r o l oxidase Culturing the bacterium in the minimal medium on either cholesterol, phytosterols or 4-cholesten-3-one (2.5-5 g/L), as a sole carbon and energy source, resulted in the production of the cell-bound enzyme only. Extracellular COX was only detected in the stationary phase, at a level of 2-4 % of final production. Secreted COX was obtained (together with the cell-bound COX) whenever the swain grew on a carbon compound different from the sterols (or from 4-cholesten3-one) or in sterol-media enriched with another carbon source, such as acetate, cholate, or yeast extract. The kinetics of specific activity of the secreted COX compared to those of the the cell-bound COX (not shown) suggested that they are two forms of the same enzyme. This idea is also supported by the similarity of both forms in the molecular mass (see below), the substrate specificity and Km value (7). The mechanism of enzyme secretion remains unknown. Catalysis by the GK1 cell-linked COX occurs externally to the cytoplasmic membrane (9). This form was solubilized by cell treatment with nonionic detergents includingTriton X-100 and Lubrol PX (7).The detergent-solubilized COX was found to be complexed with orange-pink pigmented material that could not be

643 10

ulL

Growth: g dc/L

= = ~ 300

,200

-I00

0.1

o

"

{o

"

4'o

' 6o " Growth time (h)

i o'

40

9

60

0

Figure 3. Kinetics of COX synthesis by R. sp. GK1 grown on cholate. A mineral salt medium, containing 3 g cholic a c i d / L , was used. [], u, Growth; o,o, enzyme activity. Open symbols are for the assay cultures: A, with rifampicin (150 m g ~ ) added at the arrow. B, with phytosterols (2.35 g / L ) added at the beginning of cultivation. separated from the enzyme by Sepharose CL-6B or Sephadex gel filtration. This enzyme complex aggregated in the absence of detergent without activity loss. In all media allowing COX secretion, a part of the estimated enzyme in cells became soluble in phosphate buffer (0.05 M pH 7.0) by stirringfor 30 rain at room temperature. The residual enzyme in cells was subsquently extracted by agitation in the presence of nonionic detergent. Although the exact topology of the cell-linked COX of R. sp. GK1 has yet to be determined, it seemed to have no relation with the cytoplasmic membrane. This is in contrast to what was previously believed (7). The buffer-extracted and detergent-extracted COXs are probably the same enzyme. But they are differently linked to the cell surface. The linkage of the former is weak. The extracellular and the buffer-extracted COXs were purified by affinity chromatography on a column packed with kieselguhr and cholesterol (unpublished). Then" apparent molecular mass was around 60 kDas as estimated by SDS-PAGE. 2.3. C h o l e s t e r o l o x i d a s e stability In a comparative study, thermostability of the extracellular and the cell-bound COXs from R. sp. GK1 was investigated at 50~ or 55~ Because the cell-bound enzyme extracts contained Lubrol PX, the detergent was either discarded with Amberlite XAD-2 or parallely added to samples of the secreted enzyme. In all cases, COX denaturation followed the first order process (2.3 log E t / Eo = - k.t). Lubrol PX (0.2-1%) had no effect on the thermostability of both enzyme forms. An example of the kinetics is shown in figure 4. The secreted COX was less thermostable than the cell-bound enzyme. The difference between the two enzyme forms may be because the pigmented material complexed with the detergentsolubilized COX seemed to increase its thermostability.

644

0 ,

I

200 I

I

,

Time (min) 400 0 i,

O~ r'o

I

i ~ , . u

u

"\

200 I

k=9 0 . 2 1 9m"""

I

400

62 min-1 o 0.64 x 10.2 min-1 x 1

m -20~ 0 0

-40 0.04 • 10-2- rain-1 " 0.35 x 10-2 rain-1

55~

a-

Figure 4. Kinetics of COX denaturation. 2 ml-Samples contained each 0.05 M phosphate pH 7.0, 0.85 mg protein/ml, and 0.8% Lubrol PX. A, [], cell-bound enzyme: crude extract (0.9 u/mg). [3,0, secreted enzyme (0.15 u/mg, prepared by ammonium sulfate fractionation at 60% saturation and dialysis). Purified extracellular COX from R. sp. GK1 was found to be stable up to 50~ The half-life of its activity (0.1 mg/ml, 0.05 M phosphate pH 7.0) at 50~ was around 70 min. But, it lost 75 % of its initial activity on heating at 60~ for 10 min. This thermal stability was similar to those of the purified extracellular COXs of Corynebacterium cholesterolicum (12) and R. equi 23 (13). However, purified extracellular COX of Pseudomonas sp. (14) appeared to be more stable than the preceding ones. The half-life of its original activity (5 ~g/ml, 0.1 M phosphate pH 7.0) on heating at 70~ was 60 rain. REFERENCES

1. C.I~A. Martin, Biotechnology, H.j. Rehm and G. Reed (eds.) vol. 6a (K. Kieslich ed.) Biotransformations, P.79, Verlag Chemie, Weinheim, Gelmaany, 1984. 2. T. Murohisa and M. Iida, J. Ferment. Bioeng., 76 (1993) 174. 3. W.-h. Liu, M.-h.MengandK.-s. Chen, Agric. Biol. Chem., 52 (1988) 413. 4. M. Sojo, R. Bru, D. Lopez-Molina, F. Garcia-Carmona and J.-c. Argtielles, Appl. Microbiol. Biotechnol., 47 (1997) 583. 5. M.G. Halpern (ed.), Industrial enzymes from microbial sources (Chem. Technol. Rev., no 186), P.3, Noyes Data Corporation, Park-Ridge, N.J. 1981. 6. J. Kreit, P. Germain and G. Lefebvre, J. Biotechnol., 24 (1992) 177. 7. J. Kreit, G. Lefebvre and P. Germain, J. Biotechnol., 33 (1994) 271. 8. F. I~-ier, J. I~'eit and J.B. Milli~re, Lett. Appl. Microbiol.,26 (1998) 285. 9. J. Kreit, G. Lefebvre, A. Elhichami, P. Germain and M. Saghi, Lipids, 27 (1992) 458. 10. S. Hayakawa, Adv. LipidRes., 11 (1973) 143. 11. R.A. Leppik, Biochem. J., 202 (1982) 747. 12. Y. Shirokane, K. N a k a m u r a and I~ Mizusawa, J.Ferment.Technol.,55 (1977) 337. 13. I~ Watanabe, H. Aihara, Y. Nakagawa, R. Nakamura, and T. Sasaki, J. Agric. Food Chem., (37) (1989) 1178. 14. S.-y. Lee, H.-i Rhee, W.-c. Tae, J.-c. Shin and B.-k. Park, Appl. Microbiol. Biotechnol., 31 (1989) 542.


645

Stability o f Functionally Active Fusion Proteins D u r i n g Their B i o s y n t h e s i s and Isolation f r o m E x p r e s s i n g Bacterial Cells. L.M. Vinokurov ~, A.V. Yakhnin a, T.V. Ivashkina b, V.N. Ksenzenko b, Yu.B. Alakhov a . a Branch of the Institute of Bioorganic Chemistry, Russian Academy of Science, 142292, Pushchino, Russia. b Institute of Protein Research, Russian Academy of Science, 142292, Pushchino, Russia. 1. INTRODUCTION One of the general aims of biotechnology is the large scale production of proteins important for medicine and technology. The remarkable progress in gene engineering has primarily promoted the production of proteins requiring overexpression of corresponding genes in various types of cells, mainly i.n E.coli cells. Moreover, gene engineering allows to design and produce the so-called "fusion" proteins, i.e., polypeptide chains containing primary sequences of several proteins with quite different functional properties coded within one gene. The applications of such proteins are very wide and include, in particular, the purification of newly synthesized proteins with affinity chromatography, addressed delivery of enzymes and biologically active peptides to its functioning sites, as well as the development of various diagnostic systems for medicine, ecology and biotechnology. Despite the attraction of fusion proteins for elaborating various scientific and applied tasks, little success has been achieved in this direction. The little progress in this field seems to be mainly due to the difficulties connected with the folding of newly synthesized fusion proteins and to their lability to proteolytic cleavage during synthesis and isolation. To express fusion proteins, maltose-binding protein [ 1], glutation-S-transferase [2] and thioredoxin[3] were suggested as one of the functional components. Green fluorescence proteins [4] and Ca2+-activating photoproteins, in particular, obelin [5] were suitable candidates as the reporter part. In the present work several genetic constructs were created for the in vivo expression in E.coli cells of the following bimodular fusion proteins: glutation-S-transferase-sarcotoxin (GST-Sarc), sarcotoxin-obelin (Sarc-Ob), dihydrofolate reductase-obelin (DHFR-Ob), green fluorescence protein-obelin (GFP-Ob), barstar-obelin (Star-Ob). These proteins were used to demonstrate the effect of spacers and their nearest surrounding on the stability and functionality of individual modules in the fusion proteins.

646 2. MATERIALS AND METHODS. 2.1 Genetic constructs and their expression in E. coli cells The cDNA with the complete obelin coding sequence was provided by Dr. E.S.Vysotski (Kxasnoyarsk, Institute of Biophysics, Russian Academy of Sciences). The sarcotoxin coding gene was chemically synthetized by Dr. Dobrynin (Moscow, Institute of Bioorganic Chemistry, Russian Academy of Sciences). The gene coding for green fluorescence protein (GFP) has been described in [6]. The genes coding for barnase and bastar were provided by Dr. R.V. Hartley (Bethesda, NIH). The dehydrofolatreductase (DHFR) with the 6xHis epitome at the N-terminal was taken from pQE-41 (Qiagen). The genes coding for the fusion proteins sarcotoxin-obelin (Sarc-Ob), dehydrofolate reductase-obelin (DHFR-Ob), barstar-obelin (Star-Ob), and green fluorescence protein-obelin (GFP-Ob) were obtained from PCR fragments of the respective genes. The gene of the fusion protein glutation-S-transferase, sarcotoxin (GST-Sarc) was obtained by cloning the sarcotoxin gene at vector pGEX-2T (Pharmacia) under the control of the tac promoter in E.coli JM109. The gene of the fusion protein Sarc-Ob was cloned in the vector on the basis of pQE31 under the control of the phage T5 promotor and the duplicated lactose operator in JM109 E.coli cells. To express the fusion proteins containing sarcotoxin the cells were grown in the Terrific Broth media with 100 ~tg/ml ampicilin to A600=0.5, then induced by adding IPTG to a final concentration of 0.5 mM. The cells were collected after 4 h growth at 37~ C after induction. The gene of the fusion protein DHFR-Ob was cloned in vector pT7-5 under the control of the phage T7 promoter and the phage X operator. To express recombinant protein we used strain C600/pCP1-2 containing in the compatible plasmid pCP-2 the gene of the phage T-7 RNApolymerase. The cells were grown in an LB medium at 28 ~ C to A600=0,4 and then induced by heating to 42 ~ C for 15 min. The cells were collected within 2-3 h growth at 280 C after induction. The GFP-obelin and Star-Ob were cloned in vector pET11 cjoe under control of the O7/lac promotor. All three plasmids were introduced into the !ysogenic strain BL 21/DE3. The cells were grown in an LB medium to A600=0,4, then induced by adding IPTG to a final concentration of 0.4 mM and collected within 4 h growth at 37* C after induction. 2.2 Isolation of recombinant proteins E. col~ cells containing recombinant proteins were disrupted by ultrasound and centrifuged at 30,000g for 20 min. The precipitate of the undissolved inclusion bodies were subjected to further purification as described in [5]. The recombinant proteins contained in the inclusion body were dissolved in 6 M urea. Further purification of all the fusion proteins containing the obelin module was done on Ni-NTA agarose (unpublished data). The fusion protein glutation-S-transferase-sarcotoxin was expressed in the solubilized form and isolated from the 30,000g supernatant on glutation-Sepharose (Pharmacia) according to the producer protocol. 2.3 Determination of luminescence activity of obelin The luminescence activity of the obelin module fusion proteins was determined as described in [8]. Aliquots of 2-5 I.tl of the protein sample were added to 100 ~1 of the starting buffer containing 40 mM Na-MOPS, pH 7.1, 500 mM NaC1, 5 mM EDTA, 10 mM ~-ME, 1 mg/mL BSA, 80 ~tM coelentarazine. After overnight incubation at 4* C, 1-5 lal of the starting buffer were mixed in the luminometer cuvette with 200 ~1 of the buffer for measurement containing 100 mM Tris-HC1, pH 8.8, and 10 mM EDTA. Luminescence was initiated by adding 50 ~tl 200 mM CaC12. fixing the signal on a recorder. Determination of

647 fluorescence activity of green fluorescence protein was done using a Jasco 821 FP spectrofluorimeter at excitation and emission wavelengths of 395 and 510 nm, respectively. 2.4 Renaturation of proteins from inclusion bodies The clarified (by 10 min centrifugation at 30,000g) solution of inclusion bodies in 6 M urea were added drop-wise with intensive stirring to 20 volumes of renaturating solution containing 20 mM Tris-HC1, pH 7.8, 5 mM EDTA, 500 mM NaC1, 5 mM [3-ME. The solution of renatured protein was concentrated by ultrafiltration through a membrane with 30,000 Da pores.

3. RESULTS AND DISCUSSION The chimeric gene constructs were expressed in E. coli cells. The results of expression are presented in Fig. 1. The fusion protein GST-Sarc is expressed mainly in the dissolved form but the level of its expression is sufficiently low. However, the major band on the electrophoregram is that corresponding in molecular mass to glutation-S-transferase (Fig.lA, line 1).

Fig. 1. Analysis of expressed in E. coli fusion proteins by SDS-gradient polyacrylamide (1020%) gel electrophoresis. Gels are stained by Coomassie Blue G250. Molecular mass standards 90, 66, 44, 30, 20 and 14 kDa are shown as "M". (Panel A) Glutathione-S-transferase - Sarcotoxin (GST-Sarc). Line 1, 30,000 g supernatant after sonication; 2, undigested, and 3, digested (with thrombin) proteins purified by chromatography on Glutathione Sepharose (Panel B) Sarcotoxin-Obelin (Sarc-Ob). Line 1, inclusion bodies; 2, purified protein after Ni 2+NTA chromatography (Panel C) 6His-Dihydrofolate reductase-Obelin (DHFR-Ob). Line 1, inclusion bodies; 2, purified proteins after Ni2+-NTA chromatography; 3, proteins from line 2 purified after gel chromatography with Sephacryl $300 (Panel D) Green fluorescent protein-Obelins (GFP-Ob I and GFP-Ob II). Line 1, inclusion bodies of GFP-Ob I in 6 M urea; 2, renatured GFP-Ob I; 3, purified proteins from line 2 after Ni~+-NTA chromatography; 4, renatured from inclusion bodies GFP-Ob II. Fluorescence of GFP module is associated with the major bands of 30 kDa on line 2 and of 35 kDa on line 4. In contrast to 30 kDa polypeptide, the 35 kDa polypeptide from line 4 is copurified with 50 kDa major protein during Ni2+-NTA chromatography (not shown)

648 After isolation of the fusion protein a further decrease of the amount of full-sized protein takes place in the column with glutation-Sepharose, while the amount of glutation-S-transferase increases (Fig.lA, line 2). Treatment of the isolated GST-Sarc and GST with thrombin results in complete cleavage of the fusion protein with the formation of GST and, apparently, in complete degradation of sarcotoxin. Sareotoxin is a short polypeptide chain consisting of 41 amino acid residues without a rigid spatial structure and, as a result, easily undergoes proteolytical cleavage. Glutation-S-transferase under thrombin treatment fully retains its activity and is resistant to thrombin action (Fig.lA line 3). The obtained data show that endogenous proteases present in the cell lysate primarily cleave the polypeptide chain at the region connecting the functional parts of the fusion protein to the parts without a rigid structure, resulting from the character of the amino acid sequence (impossibility of folding). For the fusion sarcotoxin-obelin we created a gene construct on the basis of photoprotein obelin. It was previously shown [7] that apoobelin is synthesized in E. coli cells with the formation of inclusion bodies and that, after its isolation from these bodies, apoobelin, despite the absence of a rigid structure, can be transformed into the active form by incubation with the substrate [8]. Thus the formation of inclusion bodies prevents proteolytic degradation of the non-structured regions of the polypeptide chain. Figure 1 B (lines 1 and 2) shows the level of synthesis and purification of the fusion protein sarcotoxin-obelin. It is seen that Sarc-Ob is synthesized in the cells with a high yield and can be isolated on Ni2+-NTAagarose with a purity over 90% directly from the cell lysate in one step. However, determination of the N-terminal amino acid sequence in the fusion protein shows that the polypeptide chain is heterogeneous and shortened by 4 and 5 amino acid residues. Cleavage occurs at the Lys4 and Lys5 residues in the sarcotoxin polypeptide chain. Thus, in the given case, the formation of inclusion bodies and the rapid isolation of the product of expression of the chimeric gene from the cell lysate also does not prevent proteolytic cleavage of the nonstructured region of the polypeptide chain. Evidently, fusion proteins containing comparatively small non-structured functional parts can be protected from proteolytic degradation by introducing into its polypeptide chain an additional amino acid sequence which can form a rigid spatial structure, including a target sequence. In creating fusion proteins on the basis of obelin, it should be noted that obelin retains luminescence activity in fusion protein only upon preservation of its free C-terminal residue (Pro). On the other hand, both apoobelin and the fusion protein, containing the apoobelin chain, have a high affinity to metalchelating sorbents (e.g., Ni2§ which allows their isolation from cell lysates virtually in one step (our unpublished data). The chimeric construct coding for fusion protein DHFR-Ob was created with minimal insert between the functional parts of the molecule (KGSR), and the product of its expression mainly forms inclusion bodies (Fig. 1C). Isolation of DHFR-Ob on Ni2+-NTA-agarose and SDS electrophoretic analysis shows that together with full-sized protein fusion there are products of its disruption, in molecular mass corresponding to the mass of individual modules. At subsequent gel-filtration on a column with Sephacryl-300 the fusion protein undergoes degradation with the formation of individual modules (Fig. 1C, lines 2 and 3) possessing functional activity. In the given case, apparently the amino acid content of the insert (LysGly-Ser-Arg) containing two regions for the action of the trypsin-like protease exerted the deciding effect on the stability of the fusion protein.

649

Fig. 2. Amino acids sequences of regions connecting functional parts of the fusion proteins. Chimeric gene constructs coding fusion protein GFP-Ob were created in two variants differing in the structure of the insert between functional parts (Fig.2). In the first case the insert had the sequence -Leu-Ile-Asp-Met(GFP-ObI) and -Pro-Gly-Thr-Gly- (GFP-OblI). Since GFP contains the lysine residue at the C-terminal the fusion protein GFP-ObI can be cleaved at this residue. In the case GFP-OblI this cleavage site is blocked by the proline residue. Both proteins accumulate in the cells as inclusion bodies even at low cultivation temperatures (28 o lq). After isolation and renaturation GFP-OblI possesses both luminescence and fluorescence activity and binds with Ni2+-NTA-agarose. GFP-ObI is easily cleaved into its component modules in the process of purification (Fig. 1D), and the fraction corresponding to GFP does not bind with Ni2+-NTA- agarose. In creating the gene constructs coding the fusion protein barstar-obelin we used the same insert as in the case of GFP-OblI. As a result we obtained a stable fusion protein with both the luminescence activity of obelin and the inhibiting activity ofbarstar. In conclusion it can be said that in creating fusion proteins both the amino acid sequence and the size of the insert between the functional modules must be taken into consideration. Furthermore, to ensure stability one must also take into account the structural features of the functional modules, of which fusion proteins are built. In some cases the instability of fusion proteins, resulting in the disruption of the molecule into its functional active parts, is a desirable property and this can be provided by creating corresponding constructs, taking into account the discussed structural features of the different parts of fusion proteins.

650 REFERENCES

1. P.Li.Guan, P.D.Rigss, H.Inouye, Gene 67 (1987) 21-30. 2. D.B. Smith, K.S. Johnson, Gene 4 (1993) 220-229. 3. E.R.La Vallie, E.A. Di Blasio, S.Kovacic, P.F. Schendel, J.M. Me Coy, BioTechnology 11 (1993) 187-193. 4. H.H. Gerdes, C.Kaether FEBS Letters 389 (1996) 44-47. 5. L.A. Frank, V.A. Illarionova, E.S. Vysotski. Biochem. Biophys. Res. Comm. 219 (1996) 475-479. 6. A. Crameri, E.A.Whitehorn, E.Tate, W.P.G. Stemmer, Nature Biotechnol. 14 (1996) 315319. 7. V.S Bondar, K.P. Trofimov and E.S. Vysotski, Biochimiya 57 (1992) 1481-1490. 8. S.V. Matveev, B.A. Illarionov, E.S. Vysotsky, V.S. Bondar, S.V. Markova and Y.B. Alakhov, Anal. Biochemistry 231 (1995) 34-39.

Miscellaneous



653

Production of Bio-Esters by Immobilized Lipases B~lafi-Bak6, K., Gubicza, L., Csfinyi, E., Gaifin, M.A.*, Moreno, J.A. * Research Insitute of Chemical and Process Engineering of the Pannon Agricultural University, Egyetem u. 2., 8200 Veszpr~m, Hungary *Institute of Chemical Engineering and Textile, Salamanca University, Plaza de los Caidos 1-5, 37008 Salamanca, Spain

1. INTRODUCTION Esterification of short chain alcohols with short and long chain carboxylic acids yields flavour esters and surface active agents, respectively. Using natural initial compounds and enzyme catalytic procedure under mild conditions, natural or "bio"-products can be obtained as it is written in [1-4]. In our work the purpose was to study the production of"bio"-esters in two different systems: in organic solvent and membrane bioreactor. The enzyme used for the esterification reactions is lipase (E.C. 3.1.1.3. Triacylglycerol acylhydrolase). The difficulty in all enzymatic esterification is the presence of water. Production of natural flavour esters was accomplished in organic solvent, while longer chain esters were produced in membrane bioreactor. In both processes water produced was removed continuously from the system maintaining a constant water level, that is water activity in the reaction mixture. Water removal was carried out by circulating the reaction mixture continuously and at controlled rate through a column filled with 4A zeolite in case of flavour ester production. The water produced during the esterification in the membrane bioreactor passed into the aqueous alcohol phase through the hydrophilic membrane wall. 2. MATERIALS AND METHODS

2.1. Fiavour ester production Flavour esters were produced in organic solvent (n-hexane) by Mucor miehei lipase immobilized onto macroporous acrylic resin support (particle size 250-500 ~m) The enzyme was obtained from NOVO Nordisk (Denmark). The initial components were natural acetic acid, propionic acid and butyric acid (from Daniel GmbH, Germany) and ethyl alcohol, isoamyl alcohol (from fusel oil, Gy6ri Szeszipari V/dlalat, Hungary) The process was followed by gas chromatography, using Hewlett Packard Model 5890A instrument equipped with a flame ionization detector, a 25 m FFAP fused silica capillary column (Macherey Nagel, Germany) and a capillary inlet system fitted with a split line that allows the nitrogen flow to be split at

654 60:1. Nitrogen flow through the column was 2.0 ml/min. Injection port and detector temperatures were 225 ~ 2.2. Esters of fatty acids Esterification of natural ethanol and fatty acids was carried out in a hollow fiber enzyme membrane bioreactor, where lipase (Mucor miehei) was immobilized by adsorption onto the inner wall of the fibres. The membrane used was cellulose acetate (cut off 40 000), made in the Central Food Industrial Research Institute (Budapest, Hungary). The surface area was 0.02 m 2. The membrane module was jacketed and equipped by manometers. For immobilization of the enzyme, diluted lipase solution was ultrafiltered through the membrane. The procedure was followed by determination of the activity both in the initial solution and permeate. During the reaction pure fatty acid and diluted (aqueous) ethanol solution were circulated in the lumen (enzyme side) and the shell side, respectively. Reaction conditions were as follows: temperature 37 ~ flow rate 4 ml/min for both phases. The process was followed by HPLC, consisted of Merck-Hitachi L-6000A pump, 2 Nucleogel GPC 50-5 columns sized 300 mm x 7.7 mm completed with a guard column, Merck-Hitachi AS-2000A autosampler and a a Merck Differential refractometer RI-71. The conditions were: THF eluent, flow rate 0.6 ml/min, temperature 70 ~ loop 20BI.

3. R E S U L T S 3.1. Fiavour esters Experiments to determine the effects of the different acids in various concentration on the conversion were carried out in incubator shaker. The alcohol used was isoamyl alcohol. The acids tested were acetic acid, propionic acid and butyric acid, the initial concentrations were between 0.1 and 0.8 mol/1. The alcohol/acid molar ratio was kept 3 : 1 in all experiments. Data obtained are presented in Figure 1, where ester yields after 6 hours reaction time are shown as a function of acid concentration. Using butyric and propionic acids as substrates, high yields were achieved up to 0.4-0.5 mol/l acid concentration. Beyond this point the yield

Figure 1: Effect of acid concentration on ester yield (acid-alcohol molar ratio 1:3; t=6 h; T=40 ~ solvent: n-hexane)

655 decreased. On the other hand, isoamyl acetate synthesis had low yields above 0.2 mol/l acetic acid concentration. These results also show that acid substrate chain length may alter the degree of ester synthesis, particularly for lower molecular weight substrates. The effect of water content on the reaction was studied in the production of isoamyl acetate. Experiment were carried out in incubator shaker, using n-hexane as solvent. The initial water concentration of the reaction mixture was altered by adding water from 0.025 up to 2.0 %. In the practically anhydrous medium the reaction rate was very low. Certain amount of water is essential in organic solvent to maintain the native, catalytically active conformation of the enzyme. Increasing the initial water content, ester yield increased significantly up to a certain level. However further increase in the water content resulted in a considerable decrease. The low reaction rate measured at reduced initial water content could be explained by the fact that the presence of water necessary for the activity of the enzyme is missing. The water uptaking capacity of the hexane as well as the reaction mixture is finite. Thus at high initial water content, water molecules are agglomerated around the enzyme. The water produced in the reaction is added to it and under these circumstances the thermodynamic equilibrium is shifted towards hydrolysis. In the reaction used it was possible to remove water produced from the reaction mixture and to work at a constant water content. Figure 2 shows the results achieved at altering (without water removal) and at constant (0.2 and 0.4 %) water level. The ester yield at 0.4 % constant water content after 6 hours reaction time was about 15 % higher than that of without water removal.

Figure 2: Effect of water content on isoamyl acetate yield (acid-alcohol molar ratio 1:3; t=6 h; T=40 ~ solvent: n-hexane)

3.2. Fatty acid esters Experiments were carried out in membrane bioreactor to produce esters of fatty acids and ethanol. Time courses of the esterification reaction using different initial ethanol concentrations are compared in Figure 3. It can be seen that the ester yield was highest in case of 20 % ethyl alcohol content. In spite of the fact, that the water content was high in the fibres, remarkable conversion (60 %) was achieved. Thus the solvent-free process seems to be suitable for (online) alcohol removal from diluted solutions, broths, as well.

656

Figure 3: Esterification in a membrane bioreactor (T=37~ acid/alcohol molar ratio 1:1,3; 1:2,7 and 1:4)

4. CONCLUSION It is turned out from our experiments that the ester yield highly depends on the water produced during the reaction. Its removal needs to be solved. In the two systems studied different methods were found to be ideal. In the esterification in organic solvent water content was maintained at a constant level by circulating the mixture through a molecular sieve. After the saturation water can be desorbed from the zeolite. Using two parallel columns, the process can be continuous. In the enzyme membrane reactor, on the other hand, water produced passed through the membrane spontaneously because of the hydrophilic character of the aqueous ethanol solution circulated in the other side of the membrane.

REFERENCES 1. 2. 3. 4.

D.W. Armstrong and H. Yamazaki, TIBTECH, 10 (1986) 264. D. Moyler, Chemistry and Industry 7 (1991) 11. F.W. Wels, R.E. Williams and K.H. Dawson, J. Food Sci. 55 (1990) 1679. F. Molinari, G. Mmarianelli and F. Aragozzini, Appl. Microbiol. Biotechnol. 43 (1995) 967.

ACKNOWLEDGEMENTS The research work was supported by the Hungarian Research Fund for Fundamental Research, OTKA, grants No. T 15885 and F 16245; ERBCIPA No. 92-3011; and the SpanishHungarian Intergovernmental Technology and Scientific Co-operation, grant No. E-8/97.


657

An Aspergillusflavus strain promoting oleic acid esterification in isooctane V. Loscos a, G. Cariesb, B. Perpifiaa, M. Torres b, N. Salab and R. Canela a'r aChemistry Department (Universitat de Lleida), Rovira Roure, 177, 25198-Lleida, Spain. bFood Technology Department (Universitat de Lleida), Rovira Roure, 177, 25198-Lleida, Spain. CArea de Protecci6 de Conreus, Centre R+D de Lleida (UdL-IRTA), Rovira Roure, 177, 25198-Lleida, Spain.

1. INTRODUCTION The increasing emphasis on the use of biocatalysts because of their favorable properties, such as their mild and environmental friendly reactions and their high specifities, is again getting importance in the industry[1 ]. Enzymatic reactions performed in nearly anhydrous organic solvents is a particular interesting and intriguing field of research. Since pioneering work by Zaks and Klibanov, and Dordick, demonstrated that water-soluble enzymes retained their activity in organic solvents, the use of enzymes in nonaqueous media has been extended and has been used chiral synthesis or resolution, fats and oil modification, synthesis of sugar based polymers, terpene ester synthesis, polyesters production, among other examples[2-5]. Bioconversion in presence of an organic solvent is interesting since it improves the solubility of the water immiscible substrate and allows the separation of the biocatalyst from the reaction mixture. Biocatalyst separation is also improved by enzyme immobilization, this method implies several techniques such as covalent attachment of enzymes to solid supports, adsorption on solid supports, entrapments in polymeric gels, cross-linking with bifunctional reagents and encapsulation of enzymes [6]. Cell-bound enzymes of filamentous fungi have been considered as immobilized catalyst also[7-10]. Among enzymes, lipases (triacylglycerolacyl hydrolase EC 3.1.1.3) are recognized as one of the most important biocatalysts. Although animals, plants, and microorganism produce them, the majority of lipases used for biotechnological purposes have been isolated from fungi[ 11 ]. It has been demonstrated that cell-bound lipase of filamentous fungi can catalyze lipolytic reactions. Subsequently, acetone-dried fungus cells were used directly in hydrolysis and esterification reactions[8]. In this report, it is described the synthesis of propyl oleate from oleic acid and 1-propanol using acetone-dried Aspergillusflavus cells in an organic solvent. The effect of the batch sample and aw is studied.

Author to whom all correspondence should be addressed.


2.1. Materials Oleic acid was purchased from Merck. Methyl palmitate, 1-propanol and isooctane were obtained from Fluka. Propyl oleate to be used as internal standard was prepared from the same oleic acid and propanol by using a chemical conventional method. 2.2. Preparation of acetone-dried resting cells Aspergillus flavus cells were prepared in our laboratory. The synthetic liquid medium used for growing the microorganism contained 2.0 g asparagine, 1.0 g K2PO4, 0.5 g MgSO4, 2.0 g glucose, 5.0 mg thiamine hydrochloride, 1.45 mg Fe(NO3)3.9H20, 0,88 mg ZnSOa.7H20 and 0.31 mg MnSOa.4H20/liter water.The pH was adjusted to 6.0. 250 ml of medium was sterilised in 1 1 flask by autoclaving at 121 ~ and 2% of refined sunflower oil was added. Two and a half millilliter of a spore suspension (5xl 06 spores/ml) of Aspergillus flavus, grown on potato dextrose agar, was inoculated to the medium, and then the flask and its contents were shaken at 200 rpm on a rotatory shaker at 28 ~ for five days.. The harvested cells were washed extensively with tap water followed by acetone, and dried under a vacuum for 48 hr. Then, micellia was milled to powder consistence and the water activity of the powder (aw = 0.17) was determined using a Novasina apparatus. 2.3. Preequilibration of water activity Powdered cells were equilibrated with glycerol solutions at 25 ~ in separate containers. The solutions were 55.2 g glycerol/100 ml water (water activity, aw= 0.89) and 0.92 g glycerol/100 ml water (water activity, aw=0.99). Equilibration was achieved in 6 days. 2.4. Esterification reaction Half-a-gram (1.77 mmol) of oleic acid and 0.8 g (13.3 mmol ) of 1-propanol were dissolved in 20 ml isooctane in a capped 50 ml Erlenmeyer flask. 0.3 g of powdered cells at the desired water activity were then added and the flask was incubated in a rotatory shaker at 28 ~ and 200 rpm until the end of the experiment. A sample of 100 ~tl reaction mixture was removed at intervals adding 100 ~tl of methyl palmitate solution as internal standard for gas chromatography analysis. The reaction was carried out in triplicate. The yield was calculated based on the conversion of initial oleic acid to the corresponding oleate. 2.5. Gas chromatography A Fisons instrument (GC 8000) with flame ionisation detection was used and the glass column (i.d. 2 mm, length 2 m) was packed with FFAP 10% on 80/100 WAW. Separation was achieved using nitrogen as carrier gas (35 ml/min) with a temperature program between 200 and 230 ~ Standard curve of product was obtained to calculate response factor.

3. RESULTS AND DISCUSSION Figure 1 shows the incidence of batch origin on the catalytic activity of acetone-dried

Aspergillusflavus cells at a fixed initial water activity of 0.17. The esterification reactions were performed in isooctane at a concentration of 88 mM of the oleic acid. Important differences

659

100

-

90-

I-o-Al

80O

"~

70 -

o

60

~ B

~ C

50 40

I

I

~

12

D

24 Time

I

48 (h)

Figure 1. Effects of batch origin and time on the synthesis ofpropyl oleate from oleic acid and 1propanol. Each point represents the averages of three replicates, and error bars indicate one standard deviation of the mean. aw of the micellia were 0.17.

between distinct batches are found for the conversion rates after 12 hours of starting the reactions. However this differences almost disappear after 24 and 48 hours, when maximum conversion rate is almost reached. Thus, batch origin should be taken into account when comparing results obtained during the first 24 hours of the experiment, beyond this period batch source seems to have less influence on the results. Influence of the heterogeneity of fungal mass on the transformation rates has already been described by several authors [ 12 ]. Chen & Wang have pointed out that this heterogeneity comes from the variety of sizes and shapes of the dried filamentous fungal cells growth in freely suspended cultures[ 12]. Subsequently, catalytic activity of acetone-dried Aspergillus flavus cells was measured at fixed initial water activity ranging from 0.17 to 0.99 (Fig. 2). Although after two incubation days esterification is almost complete in all assayed conditions, the lowest assayed water activity shows the highest transformation rate in any of the time intervals studied. Curiously, conversion percentages show a slightly decrease from day 2 to day 8, last day studied, for all assayed conditions, likely equilibrium is not reached during this period. Although our results about aw influence are concordant with the ones described by Chen et al. [8] for wax esters production using acetone-dried Rhizopus niveous cells, we must take in account the influence of the ratio of alcohol/acid concentration on the reaction rate [8] as well as the results presented by Wehjte and Adlercreutz [13-14] about the influence of the substrate concentration when water activity effects want to be studied.

660

100

-

.--!1--0,17 ---.o--0,89

90

"~

80

I

2

I

I

5

I

0,99

I

8

Tim 9 (d)

Figure 2. Effects of water activity and time on the synthesis of propyl oleate fi'om oleic acid and 1-propanol. Each point represents the averages of three replicates, and error bars indicate one standard deviation of the mean. 4. ACKNOWLEDGEMENTS

This research was supported by the Local Govemment (Paeria de Lleida, grant X0051).

REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

T. Nagasawa and H. Yamada, Pure Appl. Chem., 67 (1995) 1241. K. Faber, Bio-transformation in Organic Chemistry, Springer-Verlag, Berlin, 1992. G. Carrea, A. Corcelli, G. Palmisano and S. Riva, Biotechnol.Bioeng., 52 (1997) 648. H. Yamada and S. Shimizu, Angew.Chem.Int.Ed., 27 (1988) 622. K. Drauz and H. Waldmann, Enzyme Catalysis in Organic Synthesis, VCH, Weinheim, 1995. A.M. Klibanov, Science, 219 (1983) 722. V. L6gier and L. C. Comeau, Appl.Microbiol.Biotechnol., 37 (1992) 732. J.P. Chen, J. B. Wang and H. S. Liu, Biotech.Lett., 17 (1995) 1177. K. Sode, I. Karube, R. Araki and Y. Mikami, Biotechnol.Bioeng., 33 (1989) 1191. K. Kawakami and S. Y. Furukawa, Appl.Biochem.Biotechnol., 67 (1997) 23. P. Commenil, L. Belingheri, M. Sancholle and B. Dehorter, Lipids, 30 (1995) 351. J.P. Chen and J. B. Wang, Enzyme Microb.Technol., 20 (1997) 615. E. Wehtje and P. Adlercreutz, Biotech.Lett., 19 (1997) 537. E. Wehtje and P. Adlercreutz, Biotechnol.Bioeng., 55 (1997) 798.


661

Covalent enzyme immobilization in different porous polymer membranes H.-G. Hicke, M. Becker, G. Malsch, M. Ulbricht GKSS Research Center Geesthacht GmbH, Institute of Chemistry, Dept. Membrane Research, Kantstr. 55, D-14513 Teltow, Germany 1. INTRODUCTION Porous enzyme membranes can act as an assembly of microreactors, where each individual membrane pore contains immobilized enzyme molecules. By convection through these porous membranes high transport rates can be realized. This is especially interesting: if large diffusional resistances cause a pH change due to enzymatic conversion and thus activity decay, or if substrate depletion or product inhibition are involved in the enzymatic reaction mechanism. It is our particular interest to synthesize enzymaticaUy special polysaccharides from sucrose using transferases. In order to realize a continuous process with the substrate feed perfused through the membrane(s) and the product obtained in the filtrate, the enzyme should be covalently immobilized in appropriate membrane pores (enzyme-membrane reactor, EMR). The aim of this work is to investigate the potential of the following two strategies: 1. formation of asymmetric ultrafiltration membranes (UFM) by phase inversion from special polymers with reactive epoxide groups for direct coupling of enzymes, 2. heterogeneous pore surface functionalization of commercial symmetric microfiltration membranes (MFM) from non-reactive polymers with carboxylic or amino groups and sequential activation/coupling immobilization o,f enzymes. Enzyme binding capacities and activities, membrane permeabilities and EMR performance were evaluated using amyloglucosidase (AG) and invertase (INV) as model systems. Inulin synthesis with covalently immobilized inulinsucrase (FTF) in the EMR was studied. 2. REACTIVE MEMBRANES FOR COVALENT ENZYME IMMOBILIZATION 2.1. Asymmetric ultrafiltration membranes Integral-asynunetric membranes fi'om reactive poly-(acrylonitrile-co-glycidyl methacrylate) (PAN-co-GMA) with different contents of GMA were prepared by phase inversion from dimethyl formamide solutions into water. By different casting solution concentrations, UFM with varied pore morphologies were obtained as indicated by the water permeabilities (see Tables 1 & 2) and electron micrographs (see Figure la). Oxirane ring opening on the membrane surface can be applied for direct enzyme immobilization (see equation 1). (~H 3 '--(~H-CH 2")~" ( ] - - C H CN

2--~y

enzyme_NH2 ~i,,,/.~ ~ __ O O - - CH ~ CH /CH 2

\o

~H 3 m

..--.(CH-CH 2 " ~ " I~ CN O~ C

CH 2~y O

H

NH

I

enzyme

(1)

662

Figure 1. Scanning electron micrographs (SEM) of the different membrane types:

above left PAN-co-GMA UFM (ca. 100 I.tm thick);

above right PP MFM

(dp,no m = 0.2 [.tm;

150 ~m thick); right PET capillary MFM ( d p = 0.2 I.tm; 23 p.m thick); an effect of the heterogeneous photo-functionalization was not detectable with SEM for the latter two membrane types at the selected DG values.

2.2. Symmetric microfiltration membranes MFM with different pore structure were selected: polypropylene (PP) with a sponge-like fibril-nodule morphology and polyethylene terephthalate (PET) with straight cylindrical capillary pores (see Figure l b,c). The PP MFM had a nominal pore size of 0.2 ~m but a large size distribution, while the PET MFM were almost isoporous with dimensions of 0.2, 0.4, 1.0 and 3.0 lain, respectively. These membranes were functionalized by heterogeneous photoinitiated graft copolymerization of either acrylic acid (AA) or 2-aminoethyl methacrylate (AEMA) from water solutions using a benzophenone coating on the membrane surface and selective UV irradiation (E > 300 nm) for initiation [1]. The degree of graft polymer modification (DG) was selected so that for all different membranes and functionalities the surface coverage of the membrane polymer was similar [2]. Covalent enzyme immobilization is achieved via activation

663 of carboxyl groups (g-PAA) with water'soluble carbodiimide and derivatization of amino groups (g-PAEMA) with glutaraldehyde. 3. ENZYME MEMBRANES 3.1. Asymmetric enzyme ultrafiltration membranes For immobilization, the enzyme solution (5 g/l AG; pH = 4.6, 25~ was forced through the UFM from the more porous supporting layer; the enzyme molecules had access to the entire pore system. Hydrolysis of starch (partially hydrolyzed, ace. to ZULKOWSKY, 20 g/l; pH = 4.6, 50~ produced glucose measured by a glucose sensor. The influences of different GMA contents and membrane morphologies onto enzyme binding capacity and activity are shown in Tables 1 and 2.

Table 1. Influence of GMA (glycidyl methacrylate) content in PAN-co-GMA UFM on water permeability and enzyme activity (amyloglucosidase) without flow through the membranes Membrane

Parent membrane GMA content Permeability y (cf. (1)) J/p (l/hm2bar)

Enzyme membrane Permeability AG activity J/p (l/hm2bar) z (U/cm 2)

UFM 1

0.13

1330

290

0.02

UFM 2

0.03

435

40

0.06

UFM 3

0.015

420

55

1.50

Table 2. Influence of PAN-co-GMA (y = 0.015) UFM morphology on enzyme activity without flow through the membranes (amyloglucosidase; specific activity in solution: zsr~ = 35 U/mg) Membrane

Permeability J/p (l/hm2bar)

Bound enzyme m (lag/crn2)

AG activity z (U/crn2)

Specific activity Zsp (U/mg)

UFM 3A

123

321

1.24

3.85

UFM 3B

417

211

1.55

7.0

UFM 3C

1067

18

0.37

20.0

The lowest GMA content gave by far the highest AG activities. This can be explained by multiple binding deactivation of the enzyme. On UFM with smaller pores (lower permeability), more enzyme was bound while the specific enzyme activity was low. The former effect is attributed to a higher specific surface area, the latter one may indicate increasing sterical hindrance for enzymatic conversion of the substrate. 3.2. Symmetric enzyme microfiltration membranes Enzyme immobilization was performed sequentially: First, reactive groups were activated (g-PAA: l0 g/1 N-dimethylaminopropyl N'-ethyl carbodiimide at p H - 4.6, 4~ g-PAEMA: 100 g/1 glutaraldehyde in water, 25~ After rinsing followed the coupling reaction with

664 enzyme (AG or INV; 20 g/1 at pH = 7.5, 50~ An example of the achieved enzyme distribution inside the pores of a functionalized PET capillary MFM is shown in Figure 2. Substrate hydrolysis (AG: starch or maltose; INV: sucrose; 20 g/1 at pH = 4.6, 50~ produced glucose measured by a glucose sensor. Representative results for enzyme binding capacities and enzyme activities without flow through the membranes are shown in Table 3. Permeabilities of both unmodified (not shown) and g-PAEMA membranes were almost identic; while the g-PAA membranes had significantly lower values. This is caused by the expansion of the g-PAA "tentacles" due to charge repulsion at a pH around and above the pK~ of the graft polymer [1]. Coupling of enzyme did slightly reduce the membrane permeabilities, but large perfusion rates were possible (of. 3.3., 3.4.). Figure 2: SEM of a PET-g-PAEMA / INV Decreasing enzyme binding capacities with membrane after Auimmunolabelling of the increasing MFM pore size were due to enzyme decreasing specific surface area. Specific enzyme activities were always lower for g-PAA membranes. This may be due to steric hindrance of enzyme in the "graft-tentacle" layer [2], while g-PAEMA membranes can be described as "grafted-coat" structures with enzyme bound to reactive groups on the surface of a hydrophilic, flexible layer. Table 3. Overview on heterogeneously photo-functionalized MFM for covalent enzyme immobifization Membrane

Pores Function. dp DG (~tm) (ktg/cm2)

Permeability pH = 4.6 J/p (l/m2hbar)

Bound enzyme m (lag/cm1)

Enzyme activity z (U/era2)

Specific activity z~p (U/mg)

PP-g-PAA PP-g-PAEMA

0.2 0.2

120 90

7200 8100

INV: 540 INV: 950 AG: 335

3.6 7.0 8.5

6.7 7.4 25.4

PET-g-PAA

0.2

30

15000

INV: 110

0.30

2.8

PET-g-PAEMA

0.2

30

20000

INV: 95

0.37

3.8

PET-g-PAEMA

1.0

6

120000

INV: 49

0.22

4.5

PET-g-PAEMA

3.0

5

200000

INV: 32

0.04

1.2

PP ... polypropylene; PET ... polyethylene terephthalate; g-PAA ... grafted poly(acrylic acid); g-PAEMA ... 2-aminoethyl methacrylate (AEMA)

665

3.3. Enzyme-membrane reactor performance Enzyme membranes were tested in different modules for continuous perfusion/bioconversion applications (EMR). For a "grafted-coat" enzyme MFM the effect of substrate flux through the membrane onto enzyme activity is shown in Figure 3. 12108-

/I Y

3-

IT_G2 G401

6-

-~

2.

4-

-- AG membrane

20

...... AG dissolved

r~.=-.---

0

I

500

I

I

1000 1500 Flow rate, J (I/m2h)

I

2000

0

0

t

200

"

I

I

400 600 Flow rate, J (I/m2h)

I

800

Figure 3. Effect of transmembrane flow (perfusion) rate on the AG activity of a PP-g-PAEMA MFM (DG = 80 gg/crn2; AG amount 335 gg/cm2); G2 = maltose, G40 = starch (cf. 3.1.). The AG activity for hydrolysis of both substrates increased significantly with rising flux through the membrane. The activity ratio which was low for the membrane simply immersed in a stirred substrate solution approached the value for the dissolved enzyme for perfusion rates above J = 100...200 1/m2h. The specific activity of immobiliTed AG at the highest flow rate was 80% compared with the free enzyme. Hence, internal transport limitations can be effectively minimized by convective flow through the enzyme membrane. Then (suited surface functionality provided), the immobiliTed enzyme acts similar to the free,'native one. Another remarkable result was obtained with PAN-co-GMA AG membranes (UFM 3; cf. Table 1): Within 500 hours cominuous operation of the EMR (UF/starch hydrolysis), no significant loss of enzyme activity was observed.

3.4. Enzyme-membrane reactor application for synthesis of polysaccharides Immobilization of FTF on PAN-co-GMA UFM (cs Table 2) and "grai~ed-coat" PET-MYM (cs Table 3) was done as for model enzymes (cf. 3.1., 3.2.). Table 4 shows results for perfusion/bioconversion processes (EMR: 200 g/1 sucrose, pH = 7.2, 28~ perfusion rate 32 l/m2h). Both products of FTF f~ctosyl transferase activity, inulin (13-2,1-fructosyl fructose polymer) and glucose, were detected. Discrepancies between amounts of these products indicate side reactions (e.g. oligosaccharide synthesis). The molecular weight of the product inulin is exceptionally high. Largest initial FTF activities were obtained for the UFM correlating with high enzyme binding capacity due to large specific surface area. However, these membranes quickly became impermeable, obviously due to internal pore blocking by the synthesized inulin. As a consequence, a drastic FTF activity decay, even for release of glucose, was observed. The PP MFM showed similar behavior with slower pore blocking (not shown). For PET capillary MFM, a gradually different blocking tendency in correlation with the pore size was observed. Remarkably, the very large pore size (3.0 gm) MFM provided inulin synthesis almost without a

666 side reaction. The process could be realized for several hours with only marginal pore blocking. However, due to low FTF loading, the EMR productivity was low. Table 4. Performance of different FTF membranes in EMR applications for inulin synthesis Parent membrane Type Pores Permeab. dp, lam UFM 3C 0.005-0.1 PET-gPAEMA PET-gPAEMA PET-gPAEMA

0.4

J/p,

Initial permeab.

Jin/P,

l/hm2bar l/hm2bar 1,000 40,000

196 19,000

1.0

1 4 0 , 0 0 0 200,000

3.0

500,000

330,000

E~e membrane Inulin: Final InitialFTF Initial Molecular permeab, a c t i v . FTF activ, weight Jn,/p,

zi, (Glue.), zi, (Inulin),

Mw,

l/hm2bar laglmincm2 ~tglmincm2 106Dalton 0 8

144 41

32.7 2

80 48

70,000

6.2

1.0

32

290,000

2.4

2.1

30

4. CONCLUSIONS All developed membrane surface activation chemistries are suited for covalent enzyme immobilization with retention of the biological activity. Thus, a variety of membranes with largely different pore morphologies is now available for EMR development. With larger specific membrane surface area more enzyme can be bound; highest enzyme activities result for the UFM and MFM. However, the advantageous'high (convective) trans-membrane flow can be realized much better with MFM. Heterogeneous graft copolymerization provides enhanced binding capacity remote from the solid surface, but the grafted layer can also act as a barrier lowering enzymatic activity. Therefore, amino-functional "grafted-coat" surfaces have an advantage compared with carboxyl-functional "grafted tentacle" ones. Release of a high molecular weight product into the filtrate without pore blocking seems to require large pores with low tortuosity. This is best realized with capillary pore MFMs having a "grafted-coat" surface. Using the presented results and conclusions as guide-lines, further work is in progress to explore and exploit the possibilities of novel EMR applications.

Acknowledgements. Financial support for the project "Bio-synthetic chemistry" by the Bundesministerium far Bildung, Forschung und Technologie (BMBF), Bonn, the SEM work by Prof. S. Nunes-Pereira and M. Schossig-Tiedemmm, the protein analyses by Dr. A. Oechel, the technical support by A. Pfeiffer, the supply of FTF by Dr. A.G. Heyer, MPI, Golm, and the inulin analyses by Dr. S. Radosta, FhG, Teltow, are gratefully acknowledged. REFERENCES

1. M. Ulbricht, React. Funct. Polym. 31 (1996) 165. 2. M. Ulbricht, Habilitation Thesis, Humboldt University, Berlin, 1996.


667

Functionalised Cross-Linked Polyvinyi Alcohol As New Matrix For Lipase Immobilization E. Cemia, G. Milana, G. Ortaggi, C. Palocci, S. Soro Department of Chemistry, University of Rome "La Sapienza" Piazzale Aldo Moro 5 - 00185 Rome - Italy

I. INTRODUCTION. One of the more important aspects of enzymes industrial applications are referred to their immobilization. In fact immobilized biocatalysts offer certain unique advantages in terms of better process control, enhanced stability, enzyme-free products, predictable decay rates and improved economics. [1] The methods used for immobilizing enzymes are varied in complexity and efficiency. Numerous methods are available and have been reviewed recently. [2] The support used for immobilizing enzymes should possess mechanical strength, microbial resistance, thermal stability, chemical durability, chemical functionality, low cost, hydrophobicity, regenerability and high capacity of enzymes. In connection with our interest in the synthesis and applications of new polymeric stationary phases for the immobilisation of lipolytic enzymes [3] as well as for their purification and isoenzymes separation [4-5], we have recently investigated the properties of polyvinyl alcohol cross-linked with epichloridrin (CL-PVA) and successively esterified with several terms of linear fatty acids, from hexanoic to stearic acids. The above polymers were found to be used successfully for the immobilisation and for the purification of lipase from Candida rugosa. It is of relevant interest to introduce in fatty acid chain structural changes in order to check their influence on the interaction forces between the polymer moiety and the enzymes. These changes involve the introduction of substituents with different polar and steric effects, and of stereocenter. In this work we report the preparation of new examples of cross-linked polyvinyl alcohol esters with polar substituents at the end of the chain.

2. RESULTS AND DISCUSSION The substituted fatty acids used for the esterification of the cross-linked polyvinyl alcohol are reported above.

Thiswork has been supportedby EC frameworkprogrammeIV

668

COOH

I

CH3

c%

~~----~---~--~ c%

~----C~H c%

I

H

C~---(C~)Io---C~H

CI-----(C~)I l--C~H

m

Br---(CH2) 11--COOH

IV

v

C%-----(CH~)7---CH=CH--tC%)7---COOH %71

1 2,2-dimethylpropanoic acid (pivalic acid); H (+)-3,7-dimethyl-6-octenoic acid ((+)citronellic acid); !II dodecanoic acid (lauric acid); IV 12-Cl-dodecanoic acid; V 12Br-dodecanoic acid; VI cis-9-octadecenoic acid (oleic acid)

The corresponding acyl chlorides were used for the esterification and the esters were characterised by infrared spectroscopy, and elemental analysis. An example of infrared spectrum is reported in Figure 1, which shows the absorption of the ester function near 1750 cm ~ (which is completely absent in the CL-PVA spectrum). Absorbance

1.2

L0

0.8

0.6

0.4

/ oa ~, 4000

i

i

,

3S00

3000

2S00

'

I 2000

Wavek.ngm(cm'5

Figure 1. CL-PVA oleate IR spectrum

i .... 1~

,

,

10~

SO0

669 This peak disappears when the polymer is hydrolysed in a isopropanolic solution of KOH. This hydrolytic treatment allows to calculated roughly the percentage of esterification of the CL- PVA. The values are reported in Table 1.

Table 1 Degree of functionalisation expressed as percentage esterification of CL-PVA matrices Ac),l residues of polymeric esters Esterification ~%) pivaloyl lauroyl 12-Cl-dodecanoyl 12-Br-dodecanoyl oleoyl

of

43.7 % 65.5 % 51.4 % 59.3 % 66.2 %

The data of Table 2 are referred to the determination of the equilibrium water content (EWC) and the degree of swelling (DS).

Table 2 Equilibrium water content (EWC) and degree of swelling (DS) of the CL-PVA esters Acyl residues of polymeric esters EWC (%) . DS i

CL-PVA lauroyl pivaloyl 12-Br-dodecanoyl 12-Cl-dodecanoyl (+)-citronelloyl oleoyl

i

83.1 5.3 13.2 48.1 67.7 54.0 . .

4.92 0.06 0.15 0.93 2.10 1.17 .

.

The trend is the expected one. While the CL-PVA interacts strongly with water, the interaction of oleate ester is quite negligible and that of laurate ester is low. However the introduction of the end of the chain of polar groups such as alogen strongly increase the degree of swelling. Also the presence of double bond (citroneUate ester) increase the degree of the interaction with water, but at less extent. The swelling data show unequivocally that the modification of the ester chain strongly affect the interaction with water, this influence being potentially extended to the interaction with the enzymes and Candida rugosa lipase in particular. The results of the immobilisation experiments are presented in Table 3.

670 Table 3 Adsorption of Candida ru~osa lipase on different CL-PVA esterified matrices Acyl residues of Concentration a Adsorbed ,, pol)rmeric esters ,,(mg/ml) enzyme (%) CL-PVA pivaloyl (+)-citronelloyl lauroyl 12-Br-dodecanoyl 12-Cl-dodecanoyl oleoyl

3.2 3.2 0.1 2.4 1.9 1.5 0.8

20 20 80 40 35 47 87

a Concentration of the residual solution after immobilisation; Starting solution Candida rugosa lipase concentration 20 mg/ml Data show that the amount of the adsorbed enzyme depends on the different polymers functionalisation. This trend shows that 12-Cl-dodecanoate, 12-Br-dodecanoate and laurate esters are not very different in their interaction with Candida rugosa lipase. This suggest that the presence of polar substituents at the end of the acyl chain is not relevant in the enzymepolymer interaction. The enzyme-polymer interaction may be neglected for CL-PVA, non functionalised polymer, and CL-PVA pivalate. For the latest one, the short chain length and the hindrance effect can play an important role on enzyme adsorption mechanism. Moreover the oleate and citronellate esters strongly selves distinguishing, thus suggesting that the alkylic chain length, the hydrophobicity and the presence of double bond may be markedly important. In order to neglect the influence of the different degree of esterification of various polymers on enzyme adsorption, the amount of immobilised enzyme is corrected by the percentage of esterification. Data are shown in Figure 2. However the general picture is not changed. The oleate ester is sharply distinguished from the other ones. The amount of lipase adsorbed on this polymer increase, increasing the enzyme solution content.

Figure 2. Adsorption of Candida rugosa lipase on different CL-PVA esterified resins. The amount of adsorbed enzyme (rag) is corrected by the different functionalysed CL-PVA polymers esterification percentage.

671

Candida rugosa lipase immobilised on citronellate, laurate and oleate CL-PVA esters is employed as biocatalyst in triglycerides hydrolysis. Specific activities of immobilised enzyme are reported in Table 4. Table 4 Specific activity of different immobilised systems Enzymea ~m~)

Specific activityb tlamol acid/min m~) CL-PVA citronellate CL-PVA laurate

100 200 300 400 500

0.86 0.50 0.60 0.90 1.66

-

0.73 0.50 1.60

CL-PVA oleate 0.68 0.40 0.32 0.73 0.21

i

a Enzymepresent in the starting solutionfor immobilisationprocedure b Hydrolysedacid per minuteper enzyme(mg) adsorbedon the polymer;tributyrinas substrate These results point out that the high activity is obtained with citronellate and laurate immobilising systems. This behaviour could be explained with a good compromise between the strength of hydrophobic interaction and the possible hindrance effect of stationary phase alkyl chain on the active site area of the enzyme (enzyme hydrophobic patch). [6] 3. CONCLUSIONS Preliminary experiments show that almost all the esters investigated, being able to adsorb Candida rugosa lipase, may be used as immobilising phases for this enzyme. Further experiments in this direction are in course. We have being using Canaqda rugosa lipase immobilised on CL-PVA oleate ester as biocatalyst in transesterification reactions in organic solvents. REFERENCES

1. "Enzymes in Industry: production and applications", edited by Wolfgang Gerhartz, VCH Germany (1990) 2. Malcata F.X., Reyes H.R., Garcia H.S., Hill C.G., Ammundson C.H., J.A~ Oil Chem.Soc., 67 (1990) 890-910 3. Cernia E., Ortaggi G., Soro S., Castagnola M., Tetrahedron Lett., 35, No. 48 (1994) 90519054 4. Battinelli L., Cernia E., Delbo m., Ortaggi O., Pala A., Soro S., J. Chromatogr. A, 753 (1996) 47-55 5. Battinelli L., Castagnola M., Cernia E., Delb6 M., Padula F., Pala A. Soro S., International Journal of Biochromatography, in press 6. Grochulski P., Bouthiller F., Kazlauskas R.J., Serreqi A.N., Schrag J.D., Ziomek E., Cygler M., Biochemistry, 33 (1994) 3494-3500



673

E n z y m a t i c b i p h a s i c m e m b r a n e reactors for the synthesis of chiral products H.A. Sousa, J.P.S.G. Crespo Chemistry Department, Faculdade de Ci~ncias e Tecnologia, Universidade Nova de Lisboa, 2825 Monte da Caparica, Portugal

This work reports the use of the enzyme Pig Liver Esterase ~ L E ) (EC 3.1.1.1) to catalyse the enantioselective hydrolysis of c/s-cyclohex-4-ene-l,2-dicarboxylate (c/S-DE) to the chiral product of pharmaceutical interest, methyl-hydrogen (1S-2R)-cyclohex-4-ene-l,2dicarboxylate ((1S-2R)-ME). For this purpose, a hollow fiber reactor is used, with enzyme immobilized in the membrane porous structure. The membrane module was operated as a contactor between an organic phase, where the substrate is dissolved, and an aqueous phase, to where the product diffuses. The results obtained with immobilized enzyme are compared with those obtained using a bulk biphasic reactor where the aqueous phase with the solubilized enzyme contacts with a non-miscible organic phase containing the substrate.

1. INTRODUCTION The development of efficient methodologies for the production of optically pure enantiomers is of fundamental importance, particularly in the synthesis of biologically active products. Its preparation by enzymatic processes has had a considerable development. Its application, under the form of isolated enzymes or incorporated in microorganisms, is mainly due to its versatility, efficiency, selectivity in relation to the structure, stereochemistry of the substrate and because it uses moderate reactional conditions when compared to those necessary when using chemical reagents. The use of isolated enzymes has the advantage of reducing the occurrence of secondary catalyses (due to other enzymes present in the microorganism) and the simplification of the purification process. Since most organic compounds of commercial interest have low solubility in aqueous solutions, biphasic membrane reactors have been developed to minimize the severe mass transfer limitations, product recovery and phase separation difficulties associated with heterogeneous emulsion systems. In these systems, the enzyme is immobilized in the membrane porous structure and the reactor is operated as a contactor between an organic and an aqueous phase. These membrane systems provide a high-surface area contact between both phases and guarantees phase separation during the process. Avoiding the necessity to separate the product from the enzyme, they minimize the purification costs and enzyme needs. In this work the enzyme Pig Liver Esterase (molecular weight: 180 kD) was used to catalyze the enantioselective hydrolysis of the substrate c/s-cyclohex-4-ene-l,2-dicarboxylate (c/s-DE) to the chiral product methyl-hydrogen (1S-2R)-cyclohex-4-ene-l,2-dicarboxylate ((1S-2R)-ME).

674

This product is an important chiral unit because it can be converted to natural biological active products such as 6-epi-PS-5 (c/s-carbapenems antibiotic), (-)-fortamine, prostaglandin, carbacyclin, brefeldinA and pentalenolactone (Figure 1).

,,o

I

",,.,"~c S

n.

,.[ i H~'"y~ I

/ ~

NHMe ,

/

/

O•

~r

..

\

\

~

~I~SiO Prostaglandinsynthon X OH

H. . . .

"~\

V-~r . - ~i

f

/

I

!

•

(-)-Fort,amine

,.,2

",,

/ ~.,'~..c S

,/ "

~"'

s~

H

HO....

Carbacyclin Ref. 4 ~O,H

' ' H .....J... 2 .....\... "~~J~j,O ~k~ Ref. 4

H

cis carbapencms Ref. 1

Brcfcldin Ref. 5

Figure 1. Enzymatic reaction for the production of methyl-hydrogen (1S-2R)-cyclohex-4-ene1,2-dicarboxylate and its main applications.

Two different systems were compared to carry out this reaction: a biphasic hollow fiber reactor in which the enzyme is physically entrapped in an ultrafiltration module; and a bulk biphasic reactor where the enzyme is solubilized in the aqueous phase. 2. METHODS 2.1. Activity measurements in the bulk biphasie reactor An enzyme solution in phosphate buffer was prepared and contacted with an organic phase in which the substrate was dissolved. Each phase was individually stirred so that the two phases would only contact at the interface, as shown in figure 2. The experiments were performed at 25~ and the pH was held constant at pH 7. Since the product diffuses to the aqueous phase, the reaction rate is determined by continuous titration

675 of this phase with sodium hydroxyde. An automatic titrator (Radiometer) was used for that purpose. 2.2. Enzyme immobilization a n d activity measurements in the m e m b r a n e r e a c t o r

- Physical entrapment- Pig liver esterase was physically entrapped in a polysulphone hollow fiber module (Fresenius) by dead-end filtration of an enzyme solution in phosphate buffer pH 7. After washing with buffer, the module was essayed for enzymatic activity at 25~ by contacting the organic solution with the substrate (in which the solvent is hexane) with the same volume of phosphate buffer pH 7, as shown in figure 3. The fibers used have a cut-off of 30 kD and the total membrane area is 0.2 m2.

Figure 2. Bulk biphasic reactor

Figure 3. Biphasic membrane reactor

- Analytical methods- Product and substrate concentrations, as well as the enantiomeric excess, were determined by HPLC, using a Chiralcei OJ-R Column and a UV detector at 225 rim.

3. RESULTS AND DISCUSSION 3.1. B u l k biphasic reactor

Solvents evaluated include hexane, heptane, toluene, and carbon tetrachloride. No significant difference was noticed between the rates of reaction of the enzyme in the presence of hexane and heptane, and they both proved to be a good choice, since the substrate is highly soluble in these solvents, and the product fairly insoluble. In addition, they are both highly hydrophobic, exhibit a high interracial tension with the aqueous phase and have a low viscosity. Toluene led to a significant enzyme deactivation in a short period of time and with carbon tetrachloride, the monitorization of pH in the aqueous phase revealed that no product was transferred to this phase. The results obtained with the bulk biphasic reactor will be discussed in 3.2 along with those obtained for the biphasic membrane reactor.

676

3.2. Membrane reactor Different immobilization procedures were essayed; they are represented in figure 4. The effect of the enzyme loading procedure was evaluated in terms of enzymatic activity and stereospecificity. The amount of immobilized enzyme and its axial and radial distribution in the fibers was also determined and related to the enzymatic activity.

t

(a)

(b)

t

t

(c)

(d)

Figure 4. Immobilization procedures evaluated when physically entrapping pig liver esterase. The arrows represent the direction of the recirculating enzyme solution.

Each of the represented techniques resulted in a different enzyme distribution, being the enzyme concentration in the fibers always higher close to the ends where the enzyme solution was loaded. The outer layers of the fibers also seem to immobilize more enzyme. It was found that the activity of the enzyme was dependent on its distribution along the fibers. The more even the enzyme distribution, the more active the enzyme remains. Procedures (a) and (b) yielded better axial and radial enzyme distribution and therefore presented an increased enzymatic activity. It was also observed that reducing the concentration of the recirculating enzyme solution favorably affects the enzyme distribution along the fibers. The enzyme tends to have a much more uniform distribution and therefore retain a higher activity. The enantiomeric excess of the desired product is not affected by the immobilization procedure, and was found to be no less than 97% in all cases. A typical evolution of the enantiomeric excess is shown in Figure 5. The membrane reactor was essayed for activity over a 3-week time period, to evaluate enzyme deactivation over time. Enzymatic activity measurements were given by initial reaction rates when recirculating a given susbtrate solution. In-between these measurements, the module was kept in working conditions, always being recirculated by an aqueous and an organic solution. A typical evolution of the activity of a physically entrapped enzyme is that shown in Figure 6 (activity measurements performed with a 0.224 M substrate solution in hexane for 6346 units of immobilized enzyme)). In the first 100 h of operation, the enzyme seems to be adapting to the new environment and its activity increases. After this period, the activity starts decreasing and appears to stabilize at a value slightly lower than its initial activity.

677

Figure 5. Evolution of the enantiomeric excess as the reaction proceeds

Figure 6. Evolution of the enzyme activity over time

Although, in this case, the activity of the immobilized enzyme is not directly comparable to that of the fa'ee enzyme, since we are not using an aqueous solution of the substrate, we can compare the performance of this biphasic membrane reactor with the performance of the bulk biphasic reactor where the enzyme is dissolved in the aqueous phase (Figure 7). It can be inferred from this figure that when the enzyme is immobilized the reaction proceeds at a higher rate, because all the enzyme present is in contact with the organic phase.

Figure 7. Comparison between the reaction rates for the fxee and physically entrapped enzyme when contacting with an organic solution of the substrate

678 4. CONCLUSIONS The immobilization of the esterase by physical entrapment proved to be a good method to immobilize the enzyme since the enzymatic activity was retained for a long period of time, yielding better results than those obtained by previous studies involving covalent binding of the enzyme to nylon membranes. The biphasie membrane reactor proved to be an efficient configuration for the production of the desired product: the system is very stable (no emulsion is formed), the reaction proceeds until substrate depletion, resulting in 100% conversion, the solvent used does not seem to affect the enzyme stability, and the enantioselectivity of the enzyme is preserved. ACKNOWLEDGEMENTS H. A. Sousa would like to acknowledge the doctoral grant conceded by the program PRAXIS XXI. The authors would like to thank Fresenius for providing the membrane modules, and the financial support of JNICT under the project PBIC/C/QUI/2360/95. REFERENCES 1. T. Tamura, Y. Kawano, Y. Matsushita, K. Yoshioka, M. Ochiai, Tetrahedron Lett. 1986, 27, 3749, and cited references. 2. K. Kamiyama, S. Kobayashi, M. Ohno, Chem. Lett., 1987, 29. 3. H.-J. Gais, K.L. Lukas, W.A. Ball, H.J. Lindner, T.Lied, H.Sliwa, B. Rosenstock, Justus Liebigs Ann. Chem., 1986, 1179, and cited references. 4. H.-J. Gais, K.L. Lukas, W.A. Ball, S. Braun, H.J. Lindner, Justus Liebigs Ann. Chem., 1986, 687, and cited references. 5. H.-J. Gais, T. Lied, Angew. Chem. Intt. Ed. Engl. 1984, 23, 145. 6. Hermanson, G.T.; Mallia, A.K.; Smith, P.K.; Immobilized affinity ligand techniques, Academic Press, 1992.


679

C o i m m o b i l i z a t i o n o f e n z y m e s and cells on chitosan and d e r v a t i v e s Martin, A. B., Picciolato, M. and Heras, A. ~ Unit of RMN. Department of Physical Chemistry. Faculty of Pharmacy. Complutense University. E-28040 Madrid, Spain. Fax 341-3943245.

1. INTRODUCTION Immobilization of enzymes and cells has the potential for future industrial and commercial use in many areas of food and fodder, pharmaceutical and chemical industries and chemical specialities in processes that have no current equivalents. For the many potential applications of immobilized enzymes and cells, the preparation technique should be easy and the cost low. The expansion of Biotechnology, especially genetic technology and pharmaceutical industry, needs new studies and methods of immobilization of enzymes and cells. In this sense, coimmobilization of cells and/or enzymes is performed for three main reasons : first, to enable cells to use nonmetabolizable substrates other than the natural ones of the corresponding strain (1); second, to enlarge the product spectrum by the utilization of the catalytic capabilities of the coimmobilized enzymes or cells ; and third, to simplify conventional two or more step processes. For enzymes acting in sequence, coimmobilization creates a favourable microenvironment for the second enzyme, by reducing the time of diffusion of the substrate to the enzyme, increasing, thus, the efficiency of the overall system. Chitin and chitosan have been used as supports in a previous work (2). Chitin is a constituent of the outer structure of various living forms including insects, fungi and crustaceans. Chitin is also significant because of its relationships to source components of food of animal and fungal origin, as well as by its medical and pharmaceutical potential (3). Thus, it makes an excellent supporting material for immobilization of enzymes and cells as it offers a high mechanical stability, appropriate density and a low solubility in most solvents.

*Corresponding author.

680 The use of chitin as support for immobilization of enzymes and cells has lately aroused increasing interest. In the literature, it is possible to find examples of enzyme immobilization on chitin (4), but not in which respects to cells. Recently, some studies using chitosan-complexes have appeared (5). Chitosanpolyanion complex coacervates are good immobilizing agents, because they show a good mechanical stability while increasing plant cell culture productivity ; so, they are attractive candidates for plant bioteehnology application. This work shows the preliminary results of a series of researchs on different supports, all of them derivatives of chitin from different natural sources: chitin and chitosan obtained and characterizated in our group, in order to get an immobilized multierm3rnaatic system and immobilized resting cells. The multienzimatic extract and resting cells studied came from Agrobacterium radiobacter. This microorganism is rich in D-hydantoinase and N-carbamoylase. Agrobacterium radiobacter was used for conversion of racemic DL-5-substituted hydantoin to corresponding D-amino acids (6). It is also reported that this bacterium, sp 1-671, possesses D-hydantoinase and N-carbamoylase activity (7). Resting cells and multienzymatic extract were immobilized on different supports in order to obtain active and stable derivatives. The production of D-p-hydroxyphenylglycine is compared between each derivative and their relationship is established. 2. MATERIALS AND METHODS 2.1. Materials

Chitin was prepared from Cuban lobsters, Polinurus vulgaris . Chitosan was obtained from lobster chitin and commercial chitin from Sigma. Commercially available chitin, glutaraldehyde, sodium dihydrogen phosphate, and sodium hydrogen phosphate were purchased from Sigma, while hydrochloric acid, acetone, sodium chloride, glutaraldehyde, methanol, phenol and cyclohexane were supplied by Merck. All of the above chemicals were of analytical reagent grade. Biological material, Agrobacterium radiobacter resting cells and enzymatic extracts from the same microorganism, were provided by DSMDERETIL. DL-p-hydroxyphenylhydantoin, D(-)-p-hydroxyphenylglycine, and N-carbamoyl of D(-)-p-hydroxyphenylglycine were also supplied by DSMDERETIL. 2.2. Methods

Isolation and physico-chemical caracterization of supports. a. Chitin. The procedure used for isolation and physieo-chemical characterization of chitin from natural sources have been described elsewhere (6). Aquaphilicity of lobster chitin was determined according with the Mattiasson method (2). b. Chitosan. b.1. Preparation of Chitosan. The samples of chitin from Cuban lobster and commercial chitin were put in contact with a solution of NaOH at 50% (w/v) at 40~ for 24 hours under orbital stirring (the solution of NaOH was changed 2 times). The precipitate was washed with distilled water and HC10.5 N until pH=7.

681 After that it was centrifuged at 10.000 rpm for 20 minutes at 15~ The precipitate was stored on a phosphate buffer pH=7.2 for 24 hours. Finally the precipitate was filtered and lyophilized, then a white chitosan powder was produced. b.2. Estimation of the degree of acetylation. The degree of acetylation of chitosan was measured by using a previously reported method (8). b.3. Viscosity measure. The viscosity was measured on a solution of chitosan on Acetic acid 0.1 M, at 1%, stirring for overnight at room temperature. The experiments were performed with an Ubbelohde capillary viscometer immersed in a Selecta water circulator .The temperature was 25~ The molecular mass was estimated from the relation [11]= 0.069x My~ (9).

Immobilization of biocatalysts. a. Coimmobilization of the multienzyme system. The multienzymatic system was supplied by DSMDERETIL. This one was obtained from Agrobacterium radiobacter by sonication. The system had D-hydantoinase and Ncarbamoylase. A solution of a multienzymatic system on potasium phosphate buffer pH=8 was put in contact with the support (different chitins and chitosan) and glutaraldehyde at 1% for overnight at 4~ Immobilized derivatives were washed with an aqueous solution, acetate sodium solution and finally they were stored on a sodium phosphate solution pH=7. The amount of the immobilized multienzymatic system in each case was calculated as the difference between that in the initial solution brought into contact with the support and that of the protein swept by washing, determined by the Lowry method. b. Immobilization of resting cells by covalent binding. An Agrobacterium radiobacter gift from the collection of DSMDERETIL was used. A solution of it on a phosphate buffer pH=8 was put in contact with different supports and glutaraldehyde at 1% for overnight at 4~ according to the Heras method (2). Immobilized cells were washed with an aqueous solution, acetate sodium solution, and were stored on a sodium phosphate solution pH=8. The yield of the resting cells immobilized wa~ calculated as the difference of O.D. between the initial solution brought into contact with the support and that of the resting cells swept by washing at ~.= 660nm.

3. RESULTS

3.1. Supports. The five immobilized derivatives were carried out on chitin and chitosan. The supporting material was obtained from shells of lobster and commercial chitin. Chitin from lobster was characterized on a previous paper (8), and it was used as a support for the enzyme (2). Chitosan was obtained from lobster chitin throughout an homogeneous basic hydrolysis. The yield of chitosan was 20 %. The degree of acetylation was determined spectrophotometrically using the method

682 of Marinas et al (10). Basic sites in chitin and chitosan, (-NH2) were titrated with phenol dissolved in cyclohexane. The amount of phenol adsorbed per gram of solid is relative to the deacetyled group of chitin (8). Table 1 shows the results of the three compounds. Table 1 Amount of phenol adsorbed in monolayer form by various chitin and chitosan samples. ........ Lobster chitin (LC) Commercial chitin (CC) Chitosan from Lobster chitin (CT)

X (toO1 g"chitin/chitosan ) x 10:~..... 5.5 3.7 100.24

As can be seen, chitosan has a hundred times more free amino groups than lobster chitin. This means that the degree of acetylation on chitosan keeps the same ratio; then, if chitin shows an acetylation degree between 1-8 %, the chitosan obtained by this procedure is practically deacetylated, close to 100%. In order to characterize chitosan, the viscosity of the solution of chitosan on acetic acid was determined at 25~ The value of viscosity (11) found was: rlspecific = 0.148426 (mPa.s). Average molecular mass was estimated from the relation" [11] = 0.069xMv ~ (9) where [11] is the intrinsic viscosity and Mv is the viscosimetric average molecular weight. The value of Mv was 3,500. 3.2. Immobilization of the multienzyme extract. Multienzymatic extract was immobilized in three different supports: Commercial chitin (CC), chitin from lobster (LC) and chitosan obtained from lobster chitin (CT). All of them were immobilized according to the method described before. Table 2 shows the result of the immobilization reaction. Table 2 Immobilization of multienzyme extracts from Agrobacterium radiobacter on support derivatives from chitin. Support CC LC CT

Ratio (multienzymatic extract volume)/g, of support 10 10 100

Coupling yield (%) 77 80 97

As can be seen in all supports, the multienzyme extract is immobilized, even when the ratio between it and the support is very different. So, chitosan looks like the best support because the coupling yield is the highest; however both chitins are also good supports since they offer a high coupling yield of the multienzyme extract. Chitosan has more active free -NH 2 groups, as it is shown in table 1, and it has also a better possibility of coupling with the enzymes.

683

3.3. Immobilization of resting cells. Agrobacterium radiobacter resting cells, were inmmobilized on commercial chitin (CC) and lobster chitin (LC). Table 3 shows the results. The resting cells showed high coupling on both chitins. The reason of this can be because the deacetylation degree of the two chitins is very similar; as shown in table I. However it is important the low ratio biomass/gr, support for derivatives LC used to get the biggest yield. It is possible to conclude that chitin from lobster shows the best conditions with respect to commercial chitin to be used as a support on Agrobacterium radiobacter resting cells. Table 3 Immobilization of resting cells from Agrobacterium radiobacter on different chitins. Support CC LC

Ratio ( resting cells )/g. of support 10 4

Coupling yield 80 73

CC : Comercial Chitin. LC :Lobster Chitin

3.4. Activity assays. All immobilized derivatives were tested with respect to D-hydantoinase and Ncarbamoylase activity. The substrate used was DL-p-hydroxyphenylhydantoin. The reaction was carried out in aqueous media, at high temperature. The concentration of the product was determined by HPLC, according to the method described by Kim (6). All samples showed a wide range of product yield.

4. CONCLUSION The chitin obtained by us from the shell of Cuban lobsters shows good properties to be used as support for immobilization of multienzymatic extract and resting cells. Chitosan from lobster chitin was obtained and characterized by our group in order to get a good support. This chitosan has a high deacetylation degree and a low molecular weight. Immobilized derivatives were obtained with a multienzymatic system from Agrobacterium radiobacter and resting cells from the same microorganism on chitin and chitosan. All of them show D-p-hydantoinase and N-carbamoylase activity in a wide range, retaining 90% of it for at least 1 week.

5. ACKNOWLEDGEMENTS The Spanish Secretariat of State of Universities and Research for a grant (APC19970047), DSMDERETIL for providing us with financial support, Prof. P. Galera for his viscosity measurements and the group of Prof. J.L. L6pez Lacomba for their help with the microorganisms, deserve our thanks.

684 6. R E F E R E N C E S

1. Hahn- H~igerdal, B., Biotechnol. Bioeng. 26, (1984), 771-774. 2. Heras A. and Acosta N., Biocatalysis, 11, (1994), 305-313. 3. Chitin Handbook. European Chitin Society. Edited by Muzzarelli, R.A.A. and Peter, M.G. ATEC Edizioni. Grottammare (AP)-Italy. (1997). 4. Itiyama, k., Tokura, S., and Hayashi, t., Biotechnol. Prog. 10, (1994), 225-229. 5. Dumitriu, S., Vidal, P. and Chomet, E. Methods in Biotechnology vol. 1: Immobilization of enzymes and cells. Eds : Humana Pres Inc. (1997), 229-236. 6. K.im, G.J. and Kim, H.S. Enzyme Microb. Tech. 17, (1995), 63-67. 7. Runser S, Ohleyer E, Biotechnol. Lett, 4, (1990), 259-264. 8. Acosta, N., Jim6nez, C., Borau, V. and Heras, A., Biomass and Bioenergy, 5(2), (1993), 145-153 9. Rinaudo, M., Milas, M., and Le dung, P. Ira. J. Bio. Macromol. 15,(1993), 285. 10. Aramendia, M.A., Borau, V., Jim6nez, C. Marinas, J. M. and Rodero, F., Colloid Surfaces, 12, (1984), 227-238.


685

P a r a m e t e r s affecting the activity o f M u c o r miehei esterase 30 000 in a solventfree s y s t e m M. Karra-Chaabouni, S. Pulvin, and D. Thomas Laboratoire de Technologie Enzymatique UPRES A 6022 du CNRS, Universit6 de Technologie de Compi/~gne, B.P. 20529, 60205 Compi~gne Cedex France. E-Mail : [email protected]

The stability and the capacity of esterase 30 000 from Mucor miehei to synthesize flavour esters in a solvent free-system were investigated. The studied esterification reaction is that of butyric acid and geraniol. The effects of substrates molar ratio, the initial addition of water and temperature on the stability of enzyme were studied. It is found that the enzyme activity increases with the increase of the ratio alcohol/acid, so the butyric acid has an inhibitor effect on the enzyme activity. It was noted that the initial addition of water permits to increase the initial rate, however it decreases the conversion yield obtained at 75 h. It is also noted that the temperature increasing activates the reaction during the first ten hours, then it provokes a progressive denaturation of the enzyme and the quantities of ester produced at the end of reaction decreased when the temperature passes from 37 ~ to 60 ~ This enzyme denaturation is due to the presence of reactants coupled to the temperature, because the incubation of the enzyme alone at 60 ~ did not provoke a decrease of the enzymatic activity. Finally, a study of the influence of acid chain length (C2-C6) and alcohol structure (primary, secondary or terpenic) on the activity of Mucor miehei esterase 30 000 were made, the results show that the enzyme activity increases with the increase of acid linear chain length and it is also affected by the alcohol structure.

1. INTRODUCTION The lipases or triacylglycerol acylhydrolases (EC 3.1.1.3) are biocatalysts widely used in the industry. They catalyze the reversible reaction of acylglycerides hydrolysis in the water/oil interface, and they are found in the animals, plants and micro-organisms [1 ]. The lipases were essentially used to produce emulsifiers [2], polyglycerol-fatt~ acid esters [3], detergents [4]. The lipases were used also to catalyze the biosynthesis of natural flavour components, such us geraniol and citronellol esters by direct esterification between terpenic alcohols and short chains acids (C2-C6) [5-8]. These esters are essential oil components used in food, cosmetic and pharmaceutical industries. Lipases used essentially in these applications are of microbial and fungi origins. Two types of liquid reaction mediums were usually used for the enzymatic

686 synthesis of esters. The micro aqueous medium containing an organic solvent in which the reactants were dissolved with a very low quantity of water [9] and the micellar medium which differs from the previous medium by the addition of a surfactant [ 10]. The lipases can work also in the presence of high reactant concentrations in a system without organic solvents. In this case, the reaction medium is biphasic, composed of the co-soluble substrates in the liquid phase and the enzymatic preparation in the solid phase [8]. In this work, we have studied the parameters affecting the activity of esterase 30 000 of Mucor miehei in a solvent-free-system. We have studied, in particular, the effects of the substrate molar ratio (alcohol/acid) while keeping the total mass of the reaction mixture constant, the stability of the enzymatic preparation at high temperature, and the effect of initial addition of water. Finally the effects of the acid chain length and alcohol structure on the enzyme activity were studied.

2. MATERIALS AND METHODS Enzymes : Lipase/esterase from Mucor miehei (esterase 30 000), in a powder form, was provided by Gist Brocades (France, Seclin), the enzyme was used without further purification. Chemicals : Butyric acid, valeric acid, propionic acid, geraniol, nerol, and hexanol were provided by Fluka chemie (France), 2-hexanol was obtained from Merck, citronellol, cis-3hexenol and caproic acid were obtained from Sigma (France), acetic acid was provided by Carlo Erba reagenti (Italy), geranyl butyrate was supplied by PCAS (La Vigne aux loups, France). All these chemicals were of analytical grade. Esterification : Esterification reactions were made in screw-capped flasks containing 1 g of alcohol-acid mixture in various substrate molar ratios and 0.08 g of crude enzyme, with and without initial addition of water, the flasks were incubated in a orbital shaker at 37 ~ and 250 rpm. At precise intervals of time, samples were taken and diluted in hexane containing 5 % of hexanol as an internal standard, then they were analyzed by gas chromatography. Analysis 9The reaction was followed by measuring of the quantity of acid removed and the ester formed by capillary gas chromatography Carlo Erbo equipped with a flame ionisation detector (FID), the column used is 25 QC2/.BP21-0.25 (SGE, France), the initial oven temperature was 50 ~ then it was increased until 200 ~ with a rate of 15 ~ The injector temperature was fixed at 280 ~ and detector temperature at 240 ~ The ester quantity was expressed in mmole/g of reaction mixture. 3. RESULTS AND DISCUSSION 3.1. Effect of alcohol/acid molar ratio R When the alcohol/acid molar ratio was increased while keeping the total mass of the reaction mixture constant, the initial activity of esterase 30 000 increased. Thus the butyric acid has an inhibitor effect on the enzyme, probably by acidification of the aqueous medium which surrounds the enzyme.

687 0.035 0.03 --~ 0.025 O

0.02 "~ 0.015 .,.

0.01

-= 0.005 0

0

I

I

I

I

1

2

3

4

5

Substrate molar ratio (alcohol/acide)

Figure 1. Effect of substrates molar ratio on the esterase 30 000 initial activity, 0.1 g of enzyme, at 37 ~ and 250 rpm 3.2. Effect of temperature Figure 2 shows the evolution of the ester quantity at different temperatures (37 ~ 45 ~ and 60~ as a function of time. A thermal activation of the reaction is visible during the first 10 hours, then a thermal deactivation takes place progressively and the quantities of ester produced at the end of the reaction decreases when the temperature passes from 37 ~ to 60 ~ The effect of the temperature depends on the experimental conditions such as time of heating, presence of effectors, reactants, products and water. 2.5 -----or----- ~ C

_~

2

eO~C

1.5

,.o

t~

o 0.5 i-

0

10

20

30

40

50

60

70

80

Time (h)

Figure 2. Effect of temperature on the synthesis of geranyl butyrate, 0.08 g of enzyme, R alcohol/acid = 1.4, at 37 ~ and 250 rpm Moreover, to study the effect of the reactant on the thermal stability of esterase 30 000, the following experiments were made. Three lots of the crude enzymatic preparation were incubated at 60 ~ during 24 hours. The first lot was incubated alone, the second was incubated in the presence of geraniol and the third was incubated in the presence of butyric

688 acid. Then the reaction mediums were completed by the addition of both substrates to the first lot, the butyric acid to the second lot and the geraniol to the third lot and the three reactions were followed at 37~ through 75 hours. The figure 3 shows the kinetics of geranyl butyrate obtained with these three lots, the enzyme incubated alone at 60 ~ showed a good thermic stability. However this stability was affected by the presence of high substrates concentrations, in particular the butyric acid which deactivated totally the enzyme, indeed, the conversion yield obtained in 75 hours was in the order of that obtained without enzyme (about 3 %). The inactivation of the enzyme is due to the presence of reactants coupled to the temperature. Indeed, the incubation of the enzyme with butyric acid at 37 ~ (data no shown) did not inhibit the enzyme totally, since the conversion yield obtained in 75 hours was about 8%.

3

Enzyme

--

-~ 2.5 "6

+ Geraniol

Enzyme+ butyric a c i d

.1.5 --

1

o 0.5 -

0

7

I

10

20

~"

I

I

30

40

9 I

50

I

I

60

70

~--

80

Time (h)

Figure 3. Effect of reactants on the thermic stability of esterase 30 000, 0.08 g of enzyme, R alcohol/acid = 1.4, at 37 ~ and 250 rpm 3.3. Effect of initial addition of water In the esterification reaction, the water constitutes an important factor, because it has an effect in the thermodynamic equilibrium of chemical reaction and since this is reversible, the water favours the hydrolysis and inhibits the esterification. On the other hand the water activates the enzyme and permits a good functioning of the catalyze [11]. In this work we have studied the influence of the initial addition of increasing quantities of water on the esterification reaction. We have added in the reaction medium 0.1%, 0.2%, 0.5% and 1% of

sodium phosphate buffer 10-2 M pH = 7,5 (W/W). It was noted that the addition of water increases the initial rate and decreases the conversion yield obtained at 75 h (see Table 1). Indeed, the initial rate obtained without the addition of buffer was 0.020 mmoles/g/h and the conversion yield was 71.4 %; however, when 1 % of buffer was added, the initial rate obtained was 0.046 mmoles/gha and the conversion yield was 40 %.

689 Table 1 Effect of initial addition of water % of water ~W/W)

0

0.1

0.2

0.5

1

Initial rate (mmole/g/h)

0.020

0.024

0.028

0.038

0.046

Conversion yield (%)

71.4

57.5

51.4

45.5

40

R alcohol/acid = 1.4, 0.08 g of enzyme, at 37 ~ and 250 rpm 3.4. Effect of acid carbon chain length The study of the influence of acid chain length on the activity of esterase 30 000 was made with geraniol and linear acids whose carbon chain length varied between C2 and C6. Table 4 shows the variation of conversion yield on ester at 75 hours, the conversion yield on geranyl acetate and geranyl propionate were low about 30 %, beyond C4 the conversion yield increases upto 60 % for geranyl butyrate, 74 % for geranyl valerate and 85 % for geranyl caproate, the activity of esterase 30 000 ofMucor miehei increases when the acid carbon chain length increases according to the study of Langrand et al [12] made in organic solvent. The inhibition of the enzyme by short chain acids can be due to the presence of acid functions in the water layer which surrounds the enzyme, and the molar concentrations of these acid functions decrease when the acid carbon chain length increases. However another work [13] showed that the activity of lipase of Mucor miehei immobilized on celite decreased when the acid or alcohol carbon chain length increased.

Table 2 Effect of acid carbon chain length Acids


Acetic acid

27

Propionic acid

34

Butyric acid

61

Valeric acid

74

Caproic acid

85

R alcohol/acid = 1, 0.1 g of enzyme, at 37 ~ and 250 rpm 3.5. Effect of alcohol structure Esterification reactions of butyric acid catalyzed by esterase 30 000 were made with a primary saturated alcohol (1-hexanol), a secondary alcohol (2-hexanol), a primary unsaturated alcohol (cis-3-hexenol) and terpenic alcohols (nerol, citronellol). Table 3 shows that the conversion yield on ester obtained with the secondary alcohol (2-hexanol) is lower than the one obtained with the corresponding primary alcohol (1-hexanol), this result is in accordance to the classical chemical reactions, and can be explained by the difference in the reactivity of these two classes of alcohol. The high conversion yield on ester obtained with citronellol (86 %), geraniol and nerol (75 %) contradicts the result obtained by Langrand et al [12], they

690 found a low activity of Mucor miehei lipase with terpenic alcohols. The low relative conversion yields obtained with unsaturated alcohol can be explained by a low mobility of alcohol groups. It seems that the isomer cis/trans, which exists between the geraniol and the nerol has no effects on the catalytic activity of the enzyme. Table 3 Effect of alcohol structure on the esterase Alcohol


Hexanol

86.4

2-Hexanol

22.6

Cis-3-hexenol

50

Geraniol

71.4

Nerol

75.9

CitroneUol

85.8

R alcohol/acid = 1, 0.1 g of enzyme, at 37 ~ and 250 rpm REFERENCES

1.

N.N. Gandhi, S. B. Sawant and J. B. Joshi, Biotechnol. Bioeng. 46 (1995) 1.

2.

B. Mymes, H. Barstad, R. 1. Olsen and E. O. Elvevoll, J. Am. Oil Chem. So. 72 (1995) 1339.

.

D. Charlemagne and M. D. Legoy, J. Am. Oil Chem. So. 72 (1995) 61.

4.

M. L~zns~i,A. Huhtala, Y. Y. Linko and P. Linko, Biotechnol. Techn. 8 (1994) 451.

5.

P. A. Claon and C. C. Akoh, J. Am. Oil Chem. So. 71 (1994) 575.

6.

G. Langrand, C. Triantaphylides and J. Baratti, Biotechnol. Lett. 10 (1988) 549.

7.

P. A. Claon and C. C. Akoh, Biotechnol. Lett. 15 (1993) 1211.

8.

M. Karra-Chaabotmi, S. Pulvin, D. Touraud and D. Thomas, Biotechnol. Lett. 18 (1996) 1083.

9.

L.N., Yee, C. C. Akoh and R. S. Phillips, J. Am. Oil Chem. Soc. 72 (1995) 1407.

10. M. Bello, D. Thomas and M. D. Legoy, Biochem. Biophys. Res. Commun., 146 (1987) 146. 11. A. Zaks and A. J Russell, J. Biotechnol. 8 (1988) 259. 12. G. Langrand, N. Rondot, C.Triantaphylides and J. Baratti, Biotechnol. Lett. 12 (1990) 581. 13. A. Manjon, J. L. Iborra and A. Arocas, Biotechnol. Lett. 13 (1991) 339.


691

R e c o m b i n a n t antigens by fusion of antigenic epitopes to a G S T partner J Molmir a, I. Marczinovits a, M. Kiss b, S. Husz b, G. T6th c, L. DorgaP and M. K~ilm~ind Departments of aMicrobiology, bDermatology and CMedical Chemistry of Albert SzentGyrrgyi Medical University, P.O. Box 8, 6720 Szeged, Hungary* dlnstitute for Biotechnology, Zolt~in Bay Foundation for Applied Resarch, Derkovits Fasor 2, 6726 Szeged, Hungary The native human bullous pemphigoid autoantigens with molecular masses of 230 kDa and 180 kDa (BPAG1 and BPAG2, respectively) are recently used in the clinical practice by Western blot analysis for the differential diagnosis of a large group of autoimmune blistering skin diseases. Our aim was to construct ELISA and rapid diagnostic test systems with the application of 1 or 2 characteristic antigenic epitopes in the form of recombinant fusion polypeptides as low-cost and easily applicable variants of the above autoantigens. With the help of computer programmes three antigenic epitopes of the BPAG1 and BPAG2 were predicted. The chemically synthesised antigenic epitope peptides were tested by the patients' sera. DNA sequences coding for the epitopes were chemically synthesised and introduced as monomer, homologous and/or heterologous dimer and multimer blocks into fusion-expression plasmids of pGEX-4T-2 (Pharmacia). The recombinant products following induction were purified from the lysates of Escherichia coli DH5ot cells by affinity chromatography (Glutathione Sepharose) and used for the detection of autoantibodies of patients' sera suffering from bullous pemphigoid in an ELISA and a rapid diagnostic test. Up to 90% of 51 patients showed positive reactions with the above methods and gave far better and reliable results than were achieved by the Westem blot technique with native autoantigens. We suppose that the "avidity effect" of the fusion products has greatly contributed to the high sensitivity and fidelity of the diagnostic systems developed in our laboratory. 1. INTRODUCTION Bullous pemphigoid (BP) is an autoimmune subepidermal blistering disease of elderly people. Direct immunofluorescence studies revealed preformed Ig and C3 linear deposition along the basement membrane zone (BMZ). About three-quarters of these patients have circulating autoantibodies to hemidesmosomal BMZ proteins. These are the major 230 kDa andthe 180 kDa minor autoantigens (BPAG1 and BPAG2, respectively), cDNAs of both BP antigens have been isolated and amino acid sequences for the human autoantigens were deduced [ 1, 2] "This work was supported by the Hungarian National Committee for Technical Development (OMFB), project No. 95-97-65-0985.

692 Immunofluorescence studies and Western blot techniques with authentic antigens extracted from cultured human keratinocytes are recently used for the differential diagnosis of autoimmune blistering skin diseases. We have now addressed the question of whether the demonstration of circulating antibodies against the BP antigens are possible by means of ELISA and/or rapid diagnostic tests with recombinant fusion antigens containing only short antigenic epitopes of the two BP autoantigens. 2. MATERIALS AND M E T H O D S

Prediction of antigenic epitopes and peptide synthesis: The sequences of the BP antigens were from Swiss-Prot and PIR data banks and were analysed with WISCONSIN Package, Version 8 (Genetics computer Group, Madison, USA) using the Peptidestructure and Plotstructure programs. For the BPAG1, epitopic fragments of positions from 1814 to 1834 and from 1793 to 1813 were chosen (designated as BP1 and BP3, respectively), and for the BPAG2 only the fragment with positions from 507 to 528 (designated as BP2). Peptide sequences were synthesised by solid-phase technique utilising tBoc chemistry [3]. Peptide chains were elongated on a p-methylhydrylamine resin [4] (0.48 mmol/g) and the syntheses were carried out using an ABI 430A automatic machine with certain minor modifications on the standard protocol. The completed peptide resins were treated with liquid HF/dimethyl sulphide/pcresol/p-thiocresol (86:6:4:2, v/v), at 0 ~ 1 h [5]. HF was removed and the resulted free peptides were solubilized in 10% aqueous acetic acid, filtered and lyophilized. The crude peptides were purified by reverse-phase HPLC. The appropriate fractions were pooled and lyophilized. The purified peptides were characterised by mass spectrometry using a Finnigan TSQ 7000 tandem quadrupol mass spectrometer equipped with electrospray ion source. Peptide purities were above 97% (HPLC) and the measured Mw values were in good agreement with the calculated values in all cases. DNA synthesis: Oligonucleotides were synthesised on a Crucachem PS250 synthesiser using phosphoramidite chemistry. Enzymes were purchased from Promega and Fermentas and used according to the instruction of the manufacturer. DNA was sequenced by the dideoxy chaintermination technique. Expression and purification of the recombinant fusion products were according to [6] and described here briefly. E. cob DH5tx cells harbouring the different fusion-expression plasmid constructions were grown in LB medium and induced with 0.5 mM isopropyl [~-D-thiogalactopyranoside (IPTG) for 3 hours. Bacteria were harvested by centrifugation, then treated with lysozyme and disrupted by sonication. Isolation of the recombinant products was performed directly from the cell lysate with the help of a Glutathione Sepharose 4B affinity gel (Pharmacia) in the presence of 1% Triton X 100. The fusion proteins were eluted by reduced glutathione. The recombinant fusion products were analysed by electrophoresis in 12.5 % SDSpolyacryl-amide gels (SDS-PAG) under reducing conditions and by immunological methods. Immunological techniques: Immunological reactivity of synthetic peptides and the recombinant fusion products with patients' and the control sera were investigated by ELISA technique, and the recombinant fusion peptides by immunoblotting, too. For the ELISA 0.1 gg/100 gl concentrations of synthetic peptides, and 1 gg/100 ktl concentrations of monomer

693 fusion peptides as antigens were used. In the case of dimer and multimer fusion products antigen concentrations were equalised to the peptide components of complexes. Our earlier studies indicated a positive result when A>X+2SD (A = absorbance, X = mean A of control subjects, and SD = standard deviation of control A). Immunobiottings were performed by dropping fusion products in solution on a nitrocellulose filter or electro-transferring to the filter the proteins separated with SDS-PAGE. The blots were blocked with BSA and probed with 1:200 dilutions of patients' or control sera. Bound primary antibodies were visualised by antispecies horseradish peroxidase conjugates at 1:1000 dilution and diaminobezidine [6]. Patients a n d sera: Fifty-one patients with proved BP were investigated. The diagnosis was established on the basis of the clinical picture and by histology, as well as by direct immunofluorescence investigations. In all patients, the sera were examined by indirect immunofluorescence and immunoblotting too, as described earlier [7]. Sera from 10 blood bank donors and 10 patients with other bullous diseases (drug allergy, porphyria cutanea tarda and 3 patients with pemphigus vulgaris) served as controls.


Construction of plasmids overexpressing the autoimmune epitopes as fusion proteins DNA, coding for the epitope peptides were chemically synthesised as two overlapping oligonucleotides as depicted in Fig. 1. The synthetic sequence also contained appropriate restriction sites for cloning and in the case of the "monomer" insert a translational stop codon was incorporated as well. The two oligonucleotides were annealed (step 1) and the full double stranded state was achieved by enzymatic synthesis with Klenow polymerase (step 2). The product was amplified (step 3) and digested with the appropriate restriction endonucleases (step 4). After gel-purification (step 5) the DNA was inserted into pGEX-4T-2 (Pharmacia) and transformed into E. coli DH5(x cells (step 6). Transformants containing the insert were screened I

II oligo 7 --~ BamH I l oligo 5 . . . . .,o . . . . . . . .

oligo 3 BamH I

stop

....... --

oligo 1 J

eO o

o e,,,I oligo 2

GST ,,

.

.

.

.

Bgl II

oligo 4

V p..--

steps 1-7

~

ptac

~

-IE m GST

_

I

steps 1-7

21 E

monomer

ptac ~

oligo 8

BP1801 BP230 monomer

pGX4T-2

BP1801 BP230

oligo 6

!

S a l l l EcoR I

-- uJ -tEt~ m m

ptac

.

,1[~ GST

-rE t~ -.

BP1801 BP230 ptac m o n o and ~ _ heteromultimers

GST

! ..~-,~-~ ----

Fig. 1. Schematic diagrams of the construction of monomer (I) and multimer (II) variants of the fusion-expression plasmids. For details see the text.

694 for, and the structure of the plasmids purified from those transformants were checked by sequencing (step 7). Plasmids containing multimer epitopes were constructed in a similar way except the insert did not contain stop codons, and it was inserted into the plasmid containing monomer epitope at the unique BamHI site located at the N-terminal end of the resident epitope. In the fight orientation of the insert the N-proximal BamHl site was regenerated allowing multiple cycles of insertion. In order to minimize possible interactions between the epitopes two prolin residues were introduced at the junction of epitope petides. As all of the three synthetic peptide epitopes reacted with patients' sera, DNA blocks encoding them were synthesised and inserted into the fusion-expression plasmid vectors. The BP 172 sequence proved to be unstable in monomeric construction, therefore, only its heterologous dimer variants could be inserted and maintained in fusion expression plasmids. Altogether, eight fusion constructions were developed.

Expression and purification of recombinant fusion peptides Recombinant products of the eight fusion-expression constructions are as follows: GST-BP 1, GST-BP2, GST-BP22, GST-BPlll, GST-BP221, GST-BPlll2, GST-BP31, and GST-BP32. Each of them is composed of the GST at the -NH~ end followed by the antigenic epitope(s) at the -COOH terminus. The numbers and the order of numbers 1, 2 or 3 represent the antigenic peptide epitope(s) (see in the Materials and Methods) and the order of the peptides in the construction, respectively. After induction E. coli DH5ct cells harbouring the recombinant plasmids expressed the fusion products at high level (Fig. 2). The yield of the expressed and purified fusion proteins was between 5 and 7 mg per 200 ml of cell culture after a 3 hours incubation.

Fig. 2. SDS-PAG electrophoresis of the lysates of E. coli cells harbouring plasmids of pGEX4T-2 (lanes 1 and 2) or its recombinant derivatives (lanes 3-8) before (lanes 1, 3, 5 and 8) and after induction (lanes 2, 4, 6 and 8) with IPTG for 3 h. The expressed recombinant products are as follows: lane 2 - GST; lane 4 - BP2; lane 6 - BP22; lane 8 - BP221. M = LMW markers (in kDa) on the right. 12.5 % PAG, stained with Coomassie brillant blue.

695 Fusion proteins purified by affinity chromatography (Fig. 3) showed positive immunological reactions with the authentic BP patients' sera (Fig.4).

Fig. 3. SDS-PAG elctrophoresis of affinity-purified recombinant products. Lanes 1 and 8 GST; lane 2 - GST-BP1; lane 3 - GST-BP2; lane 4 - GST-BP22; lane 5 - GST-BP111; lane 6 GST-BP221; lane 7 - GST-BPlll2. M: LMW markers in kDa. 12.5 % PAG, stained with Coomassie brillant blue. Each lane contains 5 ~tg of the relevant recombinant fusion product.

Fig. 4. Westem blot analysis of recombinant fusion products demonstrated in Fig. 3 with sera of patients suffering from BP. Panel A: Incubation with sera of patient Ro.I. reacted only with BPAG2. Panel B: Incubation with sera of patient D.L. reacted exclusively with BPAG1. Conditions of SDS-PAG electrophoresis were as described in Fig. 2 and 3. Electroblotted proteins were incubated with 200-fold dilutions of patients' sera for 4 h and anti-human IgG HRP (Bio Rad) for 1 h, respectively. Staining was performed with diaminobenzidine.

696 Utilization of the recombinant fusion peptides in test systems In ELISA system, comparing to other methods, the recognition capacity of the recombinant fusion products was far the best with patients' sera (Fig. 5) and sensitivity of the immune reactions i.e. the measured O.D. values always exceeded those which were obtained in the case of chemically synthesised peptides. Increased sensitivity of the immunological reactions was especially true for homologous or heterologous dimer and multimer products.

Fig. 5. Correlation of the results of ELISA and Western blot investigations according to different antigens. Reactivity of BP patients' sera (in %, axis of ordimates) with the three synthetic peptides (1), with the two recombinant monomer fusion variants (2), and with all of the recombinant fusion products (3). 4: Reactivity of the patients' sera by the Western blot technique with authentic autoantigens [7]. We concluded that the efficiency of our recombinant antigens in the test systems can not only explained by the avidity effect [8, 9], but measurable beneficial conformations of the fused pel6tides take place due to the GST partner which in the case of the solution of synthetic peptide epitopes can not be observed (I. Laczk6 et al., unpublished results). REFERENCES 1. D. Sawamura, K. Li, M-L Chu and J. Uitto, J. Biol. Chem., 266 (1991) 17784. 2. G.J. Giudice, D.J. Emery and L.A. Diaz, J. Invest. Dermatol., 99 (1992) 243. 3. R.B. Merrifield, J. Amer. Chem. Soc., 85 (1963) 2149. 4. G.R. Matsueda and J.M. Stewart, Peptides, 2 (1981) 45 5. S. Sakakibara, Y. Shimonishi, Y. Kishida, M. Okada and K. Sugihara, Bull. Chem Soc. Jpn., 40(1967)2164 6. I. Marczinovits, Cs. Somogyi, A. Patthy, P. N6meth and J. Molnhr, J. Biotechnol. 56, (1997) 81. 7. M. Kiss, S. Husz and A. Dobozy, A. Microbiol. Immunol. Hung., 43 (1996) 115. 8. D.M. Crothers and H. Metzger, Immunochemistry, 9 (1972) 341. 9. T. Tudika and A. Skerra, Protein Science, 6 (1997) 2180.

Stability and Stabilization of Biocatalysts A. Ballesteros, F.J. Plou, J.L. Iborra and P.J. Hallir~(Editors) 9 1998 Elsevier Science B.V. All rights reserved.

697

M o d i f i c a t i o n o f m e t a l s u b s t r a t e s a n d its a p p l i c a t i o n to the s t u d y o f r e d o x proteins T. Pineda, J. M. Sevilla, A. J. Roman, R. Maduefio and M. B lazquez. Departamento de Quimica Fisica y Termodin~imica Aplicada, Facultad de Ciencias, Universidad de C6rdoba, 14004 C6rdoba, Spain.

This work deals with a comparative study of the surface modification of metal substrates (gold, mercury and platinum) by chemisorption of 6-mercaptopurine (6MP). Experimental conditions for film formation were defined and tested in ex-situ voltammetric experiments. A self-assembled monolayer (SAM) seems to be formed by judging from desorptive reduction peaks and supression of hydrogen adsorption region. The 6MP-Pt substrate allows to observe promoted quasi-reversible stable electrochemistry of metalloprotein cyt c similar to that reported on 6MP-Au and 6MP-Hg electrodes. Differences in metal-S bonding strength are evidenced but lateral interactions of 6MP-adsorbed molecules seems to play a main role on the behavior of the modified substrate.

1. INTRODUCFION The 6-mercaptopurine (6MP) molecule is able to chemisorb to the metal atoms by covalent bond through the S atom. The strength of the bond between the sulfur and a gold atom is on the order of 40-50 kcal/mol [ 1]. This high affmity of the sulfur atom for the metal together with the favorable interactions between the close-packed groups makes possible a high organization in the monolayers (SAM, self-assembled monolayers) obtained by these thioderivatives. Therefore, the high affmity of gold and other metals towards sulfur adsorption allows a diverse range of functional groups to be incorporate into the SAM or onto the exposed surface of the SAM. The electrochemical reactivity of metalloproteins at bare metal electrodes is often highly irreversible and in some cases undetectable. The rapid growth in the research related to the direct electron transfer between metalloproteins and electrodes came after the first clear demonstrations of quasi-reversible and direct electrochemistry of cyt c at a 4,4'-bipyridinemodified gold electrode [2,3]. Different promoter molecules have been studied and, it has been concluded that at least two functional groups are required to promote direct electron transfer between metalloproteins and electrodes: one of them should be able to bind to the electrode surface and the other, to show a suitable orientation to allow a favourable interaction with the electron transfer domain of the protein [4,5]. On the course of our studies on characterization of adsorbed monolayers of 6MP on gold, platinum and mercury substrates, we have tested the electrochemical response of cyt c. The 6MP molecule has proved to be a good promoter for the electrochemistry of cyt c as it has

698 been reported by Taniguchi et al., which used a 6MP-modified gold electrode [6] and in characterizing the molten globule conformation in cyclic voltammetric studies [7]. In this work we will analyze the electrochemical response of cyt c obtained at the different substrates, gold, platinum and mercury, all of them modified with 6MP.

2. EXPERIMENTAL Horse heart cyt c (type VI) and 6MP were purchased from Sigma. KOH, semiconductor grade, was purchased form Aldrich. All other reagents were of Merck p.a. grade and were used without purification. As the supporting electrolyte, buffered solutions of 0.025 M phosphoric acid at pH 7.0 in 0.1 M NaCIO4 or 0.1 M KOH were used. Milli-Q purified water was used throughout to prepare solutions. All electrochemical measurements were performed at a temperature of 25+0.1~ using thermostated Metrohm cells. The reference electrode was a saturated calomel electrode (SCE) and a platinum wire served as a counter electrode. Mercury, gold and platinum were used as working electrode. The mercury electrode was a Metrohm EA 290 hanging mercury drop electrode (HMDE) with a surface area of 0.0139 cm 2. The gold electrode was a 2 mm diameter sphere made by melting a 0.5 mm diameter gold wire in a flame at natural gas/air, resulting a few facets visible on the surface. The platinum electrode was a disk of 1.6 mm diameter from BAS (Bioanalytical Systems). Voltammetric curves were recorded on an Autolab (Ecochemie model Pgstat 20) instrument attached to a PC computer with proper software (GPES) for totally control of the experiments and data acquisition and treatment. The modified electrodes were prepared using the following procedure: the gold and platinum electrodes were sequentially polished with 0.3 and 0.05 ~tm alumina + water slurries until a shiny mirror-like finish was obtained. The electrodes were then sonicated and washed with deionized water (milli-Q). In the case of the mercury electrode, a fresh mercury drop was used in each experiment. Surface modification of the electrodes was carried out by the film transfer method of dipping the clean electrode into a solution of 100 lxM 6MP for a determined time (1 and 15 min for mercury and gold and 24 hours for platinum, respectively) following by rinsing with distilled water. For the experiments with cyt c, a homemade mierocell with an optimum volume of 500 ~tl, was coupled to the conventional cell, as a separate compartment, and connected by a built-in frit terminal.

3. RESULTS AND DISCUSSION The chemisorption of 6MP on a metal surface changes the properties of the interface. The first step in this study is to find the potencial range where the interface remains modified. At high potential values, oxidative desorption of 6MP occurs coinciding in the case of gold and platinum with the oxide monolayer formation. On the other hand, at low potential values reductive desorption could take place. It is interesting to note that the reductive desorption could be used to evaluate the surface coverage of the electrode by integration of the

699 voltammetric peak obtained under these conditions. The following reaction is thought to occur for the desorption of these molecules: M-S-R + H § + e ~ M + HS-R. Figure 1 shows the voltammograms obtained for the three substrates under the conditions where a complete monolayer is thought to exist. In order to observe the reductive desorption for the substrate, is necessary to transfer the modified electrode to a basic solution (i.e., 0.1 M KOH). At lower pH, only the desorption from mercury is observed with a good shape in the voltammogram. In the case of gold the signal is very spread compared to that in alkaline media (Fig. 2).

|

|

"!

|

!

,

A

I

,

I

I

-400

-800 E/mV

|

!

B

-

- 1200

Figure 1. Cyclic voltammograms of reductive desorption of a 6MP monolayer from (A) mercury, (B) platinum and (C) gold electrodes in 0.1 M KOH solution. Scan rate 100 mV/s. Scale: (A), (B) 50 gA/cm 2, (C) 10 gA/cm 2.

0

-400 E/mV

-800

Figure 2. Cyclic voltammograms of reductive desorption of a 6MP monolayer from (A) mercury and (B) gold electrodes in 0.1 M acetic acid solution (pH 6). Scan rate 100 mV/s. Scale: (A) 50 ~tA/cm2, (B) 5 gA/cm 2.

One interesting aspect of this reductive desorption phenomena is that the potential depends strongly on the substrate nature. Then, in the case of mercury the process occurs at less negative potential and only one sharp peak is observed. This behaviour has been reported in acid medium as being due to the destruction of a condensed 2D phase [8]. Figure 1 also shows the voltammogram corresponding to the reductive desorption of the monolayer built over the gold electrode. Several peaks are observed, some of them assigned to the desorption of thio-compounds from a specific monocrystalline facet. Recent studies on reductive desorption of alkanethiolate on different gold substrates have showed that the peak potential is dependent on the surface crystallinity of the underlying gold substrate and on the differences in the binding strengths of adsorbates at terraces and step sites [9]. In the case of a platinum substrate, it seems that the reduction of the M-S bond takes place at potential close to the hydrogen evolution. A difference between the first and second sweep is observed but attempts to quantify the charge did not give any proper results. From these reduction peaks corresponding to the desorption of the 6MP molecule from the different substrates it can be concluded that the energy of the metal-sulfur bond increase in the sense HgII-o

E-CRL /

I

2P-cRt|

[]

CRL (S4))

,

60

:~ c~

9 &

loo

1P-CRL

;

s,r

|

/~

M-CRL

2

/"

'-~

ao~

/~

""~ /

,,.c,~/

~g,o

40

~ 40

orj

IAJ 2O

2O 0 0

40

80

120

160

200

240

280

0

80

TIME (h)

160

240

320

400

TIME (h)

Figure 1. Deactivation kinetics at 50~ of temperature of treated-CRL fractions and CRL. (A) CRL (L 54), 2P-CRL, 1P-CRL, E-CRL and SA-CRL. (B) CRL (L 85), A-CRL, M-CRL, 2B-CRL and 1B-CRL.

3.3. Synthetic activities of CRL and partially-purified fractions in esterification We tested our preparations in the synthesis of (S) and (R) ester of ketoprofen and we obtained the data shown in Table 3. Table 3 Esterification of R(-) and S(+)-Ketoprofen with 1-propanol, catalyzed by CRL and CRL-treated fraction

specific activity,v,"

specific activitys,"

R/S b

CRL (L 54)

9.27

12.1

1.3

SA-CRL

22.9

10.5

0.46

2P-CRL 1P-CRL E-CRL

20.3 25.3 13.9

15.3 20.4 30.0

0.75 1.19 1.47

CRL (L 85)

7.19

16.3

2.27

.

A-CRL 24.5 25.3 1.03 M-CRL 19.6 25.7 1.31 2B-CRL 21.1 15.6 0.74 1B-CRL 23.5 24.8 1.35 ' Specific ester synthetic activity (mM Umh~)x' 107. bRelation between specific synthetic activities obtained with each enantiomer. There is an enhancement in the synthetic specific activities in anhydrous media in all treated fractions versus crude CRL With both ammonium sulphate precipitation and organic solvents treatments, better biocatalysts are obtained regardless the reaction medium, because we also improved the activity in aqueous media as mentioned in section 3.1.

3.4. Synthetic activities of CRL and partially-purified fractions in transesterification We tested the transestcrification of aromatic secondary alcohols, (R,S)-1-phenylethanol and

746

(R,S)-1-phenylpropanol, with vinyl acetate, obtaining the data shown in Table 4. Table 4 Transesterification of (R,S)-l-phenylethanol and (R,S)-l-phenylpropanol with vinyl acetate, catalyzed by CRL and CRL-treated . . . . .

(R,S)- 1-phenylethanola CRL (L 54)

,

.

7.4

(R,s)- 1-phenylpropangP .......

SA-CRL

3.6

1.6

2P-CRL 1P-CRL E-CRL

11.8 15.4 13.8

7.5 16.7 5.7

CRL (L 85)

5.6

A-CRL M-CRL 2B-CRL IB-CRL Ester yield at final time (600 h).

7.7 4.4 12.2 7.8

,

,,

5.8 13.3 4.1 4.0 6.7

We can conclude that we are only improving the activity observed with the commercial enzyme in the preparations 2P-CRL, 1P-CRL, E-CRL, A-CRL and 1B-CRL with both substrates, and 2B-CRL only with (R,S)-l-phenylethanol. CRL-Treated fractions showed a Renantiopreference as described for crude CRL [ 15]. As the enantiomeric excess and yields were low (in the crude enzyme and in the preparations),a more detailed study is actually in progress. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

M.J.Hermiiz, J.M. S~inchez-Montero and J.V. Sinisterra, Tetrahedron, 36 (1994), 10749. M.J.HernS.iz,J.M. S~inchez-Montero and J.V. Sinisterra, J.Mol.Catal., 96 (1995), 317. M.L. Rtia, T. Diaz-Maurifio, V.M. Fern~dez, C. Otero and A. Ballesteros, Biochim. Biophysic. Acta, 1156 (1993), 181. I. Mingarro, C. Abad and L. Braco, Proc. Natl. Acad. Sci., 92 (1993) 3308. H. Gonz~.lez-Navarro and L. Braco, J.Mol.Catal., 3 (1997) 110. I.A. Colton, N.A. Sharmin and R.J. Kazlaukas, J. Org. Chem., 60 (1995) 212. C. Torres and C. Otero, Enzyme Microbiol. Technol., 19 (1997) 594. K. Lundell, T. Raijola and L.T. Kanerva, Enzyme Microbiol. Technol., 22 (1998) 86. M.W.Lee, F.B. Kraemer and D.L. Severson, Biochim. et Biophys.Acta., 1254 (1995) 311. T. Handelsman and Y. Shoham, J. Ger. Appl. Microbiol., 40 (1994) 435. P. Grochulski, L. Young, J.D. Scharg, F. Bouthillier, P.Smith, D. Harrison, B. Rubin and M. Cygler, J. Biol. Chem., 268 (1993) 12843. A. Radzcika, S.A. Acheson and R. Wolfender, Biorg. Chem. 20 (1992) 282. J.P. Henley and A. Sadana, Enzyme Microbiol. Technol., 7 (1985) 50. J.M. Moreno, M. Arroyo, M.J. Hermiiz and J.V. Sinisterra, Enzyme Microbiol. Technol., 21 (1997) 552. R.J. Kazlaukas, A.N.E. Weisfloch, A.T: Rappaport and L.A. Cuccia, J. Org. Chem., 56 (1991) 2656.


747

Effect o f pH and temperature on i m m o b i l i s e d Thiobacillus ferrooxidans cells over nickel alloy fibre Jos6 Manuel G6mez and Domingo Cantero Biological and Enzymatic Reactors Research Group Department of Chemical Engineering, Food Technology and Environmental Technology Faculty of Sciences, University of C~idiz (UCA) 11510 Puerto Real (Cfidiz) SPAIN E-mail: j [email protected]

1. INTRODUCTION

Acid mine drainage formation represents a major environmental problem that results from exposure of the pyritic minerals to the combined effects of atmospheric oxygen, moisture and a group of acidophilic iron-oxidising bacteria. These bacteria catalyse the oxidation of iron pyrite to ferric sulphate and sulphuric acid and increase the reaction rate against chemical via. At the step of iron oxidation in such a system, a high density of cells of iron-oxidising bacteria is essential for rapid iron oxidation. Immobilisation technology of various kinds of cells has rapidly developed and been used as an effective biocatalyst for various applications (1-4). Immobilisation of whole cells is characterised by stable maintenance of an extremely high density of living cells in and/or on the immobilised matrices. Therefore, immobilised cells are available effectively over a longer period in the systems having a small number of free living cells. Various matrices have been used for immobilisation of Thiobacillus ferrooxidans by adhesion to surfaces of glass beads (5), ion-exchange resin (6), activated carbon (7) or entrapment within calcium alginate, agar, K-carrageenan and gerlite (8,9). At the same time, it has been proposed other supports, as polyurethane foam (10) that combines the advantages of adhesion with those of entrapment. In our previous work (11) it has been investigated the immobilisation of Thiobacillus ferrooxidans in nickel alloy fibre and the kinetic study using freely suspended cells of Thiobacillus ferrooxidans (12), has shown that ferrous iron oxidation is influenced by temperature and pH. However, there is limited information available on the influence of pH and temperature on immobilisation process of Thiobacillusferrooxidans. The main aim of the present work was to investigate whether stability of Thiobacillus ferrooxidans is affected by temperature and pH values of medium during immobilisation process.


2.1. Microorganism and growth conditions The strain of Thiobacillus ferrooxidans used in this study was isolated in the Riotinto mines of Huelva (Spain) and kindly made available by the Biohydrometallurgy Group of the University of Seville (Spain). After that, this strain was purified by several subcultures on solid media. The bacteria were grown in a medium proposed by Silverman and Lundgren9: (NH4)2SO43.0 g/L; MgSO4 0.5 g/L; K2HPO4 0.5 g/L; KC1 0.1 g/L; Ca(NO3)2 0.01 g/L and a variable concentration of FeSO4, depending on the experiment to be performed. At the late logarithmic growth phase, 4 ml of sterile pyrite dissolution was added to the cultures in order to maintain the microorganism. Solid medium for subcultures was proposed by Johnson and McGinness 1~ and was called FeTSBo. This medium was prepared as follows: the three separately sterilised solutions (TSB/basal salts, ferrous sulphate and agarose) were combined, thoroughly mixed and the complete medium divided 2:1 by volume and held at 50~ To the large volume was added 2.5% (v/v) of an active culture of heterotrophic acidophile, the solution was again thoroughly mixed and 20 ml aliquots dispensed into sterile Petri plates. When the agarose had gelled, it was covered with a thin layer (10 ml) of sterile FeTSB (which had been molten al 50~

2.2. Support characteristics Nickel alloy fibre was used as an immobilisation matrix. This support is commercially provided by Scotch Brite 3M (Spain) and consists in an alloy of nickel and stainless steel as coiled metallic ribbons. Experimentally, was proven that nickel alloy fibre is a suitable matrix for immobilisation in an acidic environment (pH ~ 1.0). Support characteristics are showed in Table 1. Table 1 Matrix characteristics Real density (g/cm 3) Apparent density (g/cm 3) Porosity (%) Specific surface (m2/m3) Ribbon width (mm) Prime spiral diameter (mm) Ribbon thickness (mm)

7.83 0.58 92.6 8831.18 1 2.5 0.1

2.3. Immobilisation procedure

Thiobacillus ferrooxidans was immobilised on nickel alloy fibre according to the following procedure: a fixed amount of support was placed in a 1L flask with 600 ml of liquid medium (Fe+2 concentration over 2000 mg/L) and 10% (v/v) inoculum. To adjust the pH to 1.8 H2SO4 5M was added. Cultures were incubated at 30~ on a rotary shaker at 200 rpm.

749 When the ferrous iron concentration was minimum, the flask was drained and the medium was replaced without any intermediate inoculation. Several consecutive batches were run on a "draw and fill" basis until time steady-state biomass levels had been achieved.

2.4. Analytical methods Ferrous sulphate oxidation was monitored by determining the residual ferrous iron concentration at various intervals. The 1,10 phenanthroline method of Vogel 11 was used. In order to determine the ferrous iron concentration a 10-~tL sample was placed in a tube and diluted with 1.0 ml of distilled water. The pH was adjusted to between 3.0 and 6.0 with 2M sodium acetate, 0.8 ml of 1,10-phenanthroline solution was added and finally a further 10 ml distilled water. The absorbance at 515 nm was measured after 5-10 min. In order to determine total iron concentration, 1.0 ml hydroxylamine chloride was added to the sample instead of 1.0 ml distilled water and the same procedure followed. A calibration curve of known FeSO4 concentrations was used to calculate the iron concentrations. The concentration of iron (III) in solution was calculated by subtracting the average iron (II) concentration for the total iron concentration calculated at each point in time.

2.5. Biomass determination In order to measure the biomass content adhered to matrix support, a known amount of nickel alloy fibre was placed in a flask with 5 ml of oxalic acid 10% (w/v), at each "draw and fill" cycle. After 10 min., the support was rinsed with 5 ml of distilled water during 10 min. After that, the rinsings were added to previous cell suspension obtained. The biomass concentration was determined, then, by direct counting using a Neubauer chamber counter of 0.02 mm depth and 1/400 mm 2 area with optical microscope. In some cases, it was necessary dilute the samples with basal salts solutions by consequence of the high biomass concentrations. Each measurement was made in duplicate to avoid the experimental errors inherent in working with microbial populations.

2.6. Scanning electron microscopy Samples of the matrix material were removed during and at the termination of experiments for scanning electron microscopy (SEM). The samples were gently washed with 0.1N sulphuric acid and fixed with 2.5% glutaraldehyde solution (Sigma Chemical Co., St. Louis, MO, USA) for lh at 4~ After, the samples were washed three times with 0.1M cacodylic acid solution (pH 7.0) during 15 minutes. The fixation was made with 1% osmium tetroxide (pH 7.0) for 1 hour, followed by dehydration with acetone and critical-point drying. All samples were mounted on specimen stubs with a silver paint, gold coated, and examined by SEM under a Jeol JSM-820 model. All preparation steps, which involved washing, fixation, and dehydration, were carried out with minimum physical disturbance of the sample material.

750 3. RESULTS AND DISCUSSION Biomass immobilisation and colonisation process in nickel alloy fibre was studied following the modification in biomass concentration adhere to surface, after detached treatment with oxalic acid at different temperatures and pH values of medium. The evolution of immobilised biomass along the process is showed in Figures 1 and 2. In Figure 1, it is possible to see that adhered biomass increased rapidly in first step on cultivation and, after that, it reached a maximum in which the advance of biomass concentration was minimum. So, these maximums ofbiomass decrease with an increase of temperature value. This fact can be explained because there is a decline in growth rate of Thiobacillus ferrooxidans free cells and so, the probability of adhesion of cells to the support is smaller.

121 10-~1

O _~

~X---X--X--X

.im

9.

~

-o- ~

4 _

/

--.--35 --X--40

.~

2-

0

100

200

300

Time (h)

Figure 1. Evolution of immobilised biomass at different temperature values. At the same time, results obtained showed that immobilised Thiobacillusferrooxidans cells can oxidise ferrous iron up to the optimum temperature (31 ~ proposed for growth in submerged culture (13). So, the effect of temperature is weak when the bacteria are immobilised, like that when ferrous iron is oxidised in suspended culture. In this case, temperatures over the optimum force to decay the maximum specific growth rate to zero (13). This shows that these bacteria change their stability when they grow in a fixed state, but the mechanism and nature of this change are still unclear. Probably these changes are due to a certain physiological alteration of the microorganisms during fixation and provoke an stabilization of cells versus temperature changes. The experimental data, which demonstrate the effect of pH on colonisation process, are shown in Figure 2. It can be seen that there no significant differences between experiments at pH 2.0 and 1.7 for rate of immobilisation and maximum adhered biomass

751 concentration. Nevertheless, there are differences on formerly colonisation cycles at pH 1.4. In this case, rate of colonisation is smaller by consequence of difficulty for growth and reduction of viability (13) at this pH value. Furthermore, if we consider that biofilm consists of Thiobacillus ferrooxidans cells attached to the surface of the pores of an inert porous substance Garosite) and there is a sharp reduction of precipitates as the pH is reduced (13); a decrease of immobilised biomass per gram of support is obtained.

12 10 ~ 9 ~

8

~'~

6

--

t

..o

_

----

--

...-O-- pH 1,7 - - X - - p H 1,4

2 I 0

100

t

Time (h)

200

300

Figure 2. Evolution of immobilised biomass at different pH values. So, when Thiobacillus ferrooxidans grown in free culture, like most other bacteria, is affected by changes in the physicochemical parameters of the culture media. The dependence of the oxidation rate of ferrous iron on temperature and pH was studied earlier (16) and is in the form of a peak. Sharp optima at 31 ~ and pH 1.8 were determined. Contrary to these results, the same bacteria when grown as a biofilm was unaffected by changes of temperature between 30 and 40 ~ and a pH from 1.4 to 2.0. Olem et al. (14), Karamanev et al.(6) and Nakamura et al. (15) reported similar effects. Lastly, it is significant that immobilisation of Thiobacillusferrooxidans cells to nickel alloy fibre is considerable stable, and that they conserve their oxidative capability during storage under normal atmospheric conditions (13). It is thus possible to re-start a reactor using particles of nickel alloy fibre colonised by Thiobacillus ferrooxidans atter storage for 10 weeks, without addition of substrate. These results are important for industrial-scale operation; with respect to the practical operation of a reactor, this may be stopped and restarted at any time, without too much trouble.


We thank to the Comisi6n Interministerial de Ciencia y Tecnologia (Spain) for funding this research through Project AMB93-0353. Jos6 Manuel G6mez wish to express sincere gratitude to the Instituto de Cultura "Juan Gil-Albert" (Spain) for the provision of a studentship.

REFERENCES

1. Abbot, B.J. Advances in Applied Microbiology, Vol. 20, Perlman, G. (eds.). Academic Press, New York, 1976, p. 203-257. 2. Brodelius, P. Advances in Biochemical Engineering, Vol. 10, Ghose, T.K., Fietcher, A. and Blakebrough, N. (Eds.) Springer, Berlin, 1978, p. 75-129. 3. Chibata, I. and Tosa, T. Advances in Applied Microbiology, Vol. 20, Perlman, G. (eds.). Academic Press, New York, 1977, p. 1-25. 4. Colowick, S.P. and Kaplan, N.O. Methods in Enzymology, Vol. 44, Mosbach, K., Academic Press, New York, 1976, p. 1-530. 5. Grishin, S. and Tuovinen, O.H. Appl. Environ. Microbiol., 54 (1988) 3092-3100. 6. Karamanev, D.G. and Nikolov, L.N. Biotechnol. Bioeng., 25 (1988) 295-299. 7. Halmeier, H.; Schtifer-Treffenfeldt, W. and Reuss, M. Appl. Microbiol. Biotechnol. 40 (1993) 582-587. 8. Lancy, E.D. and Tuovinen, O.H. Appl. Microbiol. Biotechnol., 20 (1984) 94-99. 9. Wakao, N., Endo, K., Mino, K., Sakurai, Y. and Shiota, H. J. Gen. Microbiol., 40 (1994) 349-358. 10. Armentia, H. and Webb, C. Appl. Microbiol. Biotechnol., 36 (1992) 697-700. 11. G6mez, J.M.; Caro, I. and Cantero, D. Immobilised Cells: Basics and Applications, Wijffels, R.H., Buitelaar, R.M., Bucke, C. and Tramper, J. (Eds.) Elsevier Science, Amsterdam, 1996, p. 785-792. 12. G6mez, J.M.; Caro, I. and Cantero, D. J. Biotechnol., 48 (1996) 147-152. 13. G6mez, J.M. PhD Thesis. University of Cadiz. 1997. 14. Olem, H. and Unz, R.F.J. Wat. Pollut. Contr. Fed., 52 (1980) 257. 15. Nakamura, K., Noike, T. and Metsumoto, J. Wat. Res., 20 (1986) 73. 16. G6mez, J.M. and Cantero, D. IchemE Research Event. Newcastle, 1998, p. 16.

753

Index of authors

Acebal, C. 89, 719 Adler-Nissen, J. 441 Agerlin Olsen, A. 147 Aguirre, C. 95 Aillapfin, A. 27 Aires-Barros, M.R. 483 Alakhov, Yu.B. 645 Alcalde, M. 535 Alcfintara, A.R. 571,741 Aliwan, EO. 41 Almeida, M.C. 483, 487 Altamirano, C. 27 Aoike, M. 295 Arnold, U. 121 Arroyo, M. 565, 719 Ath6s, V. 205

Baeza, J. 95 Bakir, U. 151 Bfile~, V. 77 Ballesteros, A. 115, 535 Balny, C. 197 Baratti, J.C. 459 Barbotin, J.N. 591,709 Barone, G. 189, 211, 217 Barreiros, S. 483, 487 Barros, M. 731 Barut, M. 541 Bautista, EM. 505 Becker, M. 661 Bekers, M. 603 B61afi-Bak6, K. 653 Bell, G. 365 Bendikiene, V. 577, 583 Berzal, L. 157 Bhat, R. 263 Bielecki, S. 423 Bilen, J. 151 Blackwood, A.D. 399

Blazquez, M. 697 Bonet, C. 303 Boross, L. 477 Borreguero, I. 571 Boulaich, M.C. 247 Boy, M. 35 Bravo, M.C. 505 Bucke, C. 399 Busto, M.D. 157

Caanan-Haden, L. 405 Cabral, J.M.S. 483, 625 Cafaro, V. 211 Calvo, M.V. 115 Cameselle, C. 703 Campelo, J.M. 505 Canela, R. 657 Caries, G. 657 Cantero, D. 107, 747 Capasso, S. 189 Carbone, K. 553 Carvalho, M. 487 Casarci, M. 553 Castill6n, M.P. 89, 719 Catanzano, E 189, 211, 217 Caussette, M. 393 Cedano, J. 303 Cernia, E. 667 Chamorro, S. 741 Cherepanov, D.N. 65 Chikere, A. 559 Ciaramella, M. 311 Cobucci Ponzano, B. 311 Combes, D. 135, 205 Correia, A.C. 71 Cowan, D.A. 349 Craynest, M. 591 Crespo, J.P.S.G. 673 Csfinyi, E. 101,653

754 da Fonseca, M.M.R. 435 D'Alessio, G. 211 Daniel, R. 349 D'Auria, S. 83 de Diego, T. 411 De Filippis, V. 381 de Filippis, V. 277 de la Casa, R.M. 453, 735 delaMata, I. 89,719 de Vries, R.P. 41 Decagny, B. 709 Di Donato, A. 211 Dijkstra, B.W. 317 Dominik, A. 35 Dorgai, L. 691 Duarte, C.S. 435 Duefias, M.J. 523 Ebmeier, L. 223 Egorov, A.M. 65 Eijsink, V.G.H. 513 Engbersen, J.EJ. 429 Ergan, E 709 Estrada, P. 523 Falaschi, V. 435 Fatum, T.M. 147 Faulds, C.B. 41 Febbraio, E 325 Fernandes, P. 625 Fernfindez-Lafuente, R. 349, 405 Ferreira-Dias, S. 71,435 Ferrer, M. 115 Flaschel, E. 223 Fontana, A. 277, 381 Fontes, N. 483, 487 Fraaije, M.W. 141 Fuglsang, C.C. 147 Fiirlinger, M. 19 Fusi, P. 217 Gal~n, M.A. 653 Galunsky, B. 559 Gamse, T. 471 Gamulin, S. 53

Garbayo, I. 631 Garcia, A. 505 Garcia, D. 257 Garcia, E 303 Garcia, S. 483, 487 Gaunand, A. 393 Germain, P. 639 Giel3auf, A. 471 Gladilin, A. 417 G6mez, J.M. 747 Grave, J. 483 Graziano, G. 189, 211, 217 Greco Jr, G. 59 GriefJler, R. 83 Gubicza, L. 653 Guillfin, A. 611 Guisfin, J.M. 349, 405 Gusmao, J.H. 435 Haaker, H. 619 Hailing, P.J. 365, 373 Haltrich, D. 19 Hammou, H.O. 241 Hanan-Aharon, E. 183 Heras, A. 679 Hicke, H.-G. 661 Hirayama, N. 295 Hoy, C.-E. 441 Husz, S. 691 Ibarra Molero, B. 251 Iborra, J.L. 411,417 Illanes, A. 27, 95 Ivashkina, T.V. 645 Jaenicke, R. 165 Jan~ky, T. 269 Juodka, B. 577, 583 Kfilmfin, M. 691 Kapeluich, Yu.L. 65 Karra-Chaabouni, M. 685 Karsakevich, A. 603 Kasche, V. 559 Katzav-Gozanski, T. 183

755 Kele, M.Z. 269 Kim, J.R. 465 Kiss, M. 691 Klingsbichel, E. 471 Kock-van Dalen, A.C. 447 Kolisis, EN. 725 Koops, B.C. 127 Koppel, R. 183 Kos~iry, J. 477 Kotorman, M. 547 Kramhoft, A. 343 Kreit, J. 639 Ksenzenko, V.N. 645 Kula, M.-R. 331 Kulbe, K.D. 19 La Cara, E 83 Laane, C. 141,619 Lfiszl6, K. 713 Lee, H.J. 465 Lema, J.M. 611 Levitzky, V. 417 Lindet, B. 393 Llama, M.J. 229 Longo, M.A. 135, 703 L6pez-Belmonte, M.T. 571 Loscos, V. 657 Lozano, E 411,417 Lfi Chau, T. 611 Luna, D. 505 Maduefio, R. 697 Malsch, G. 661 Manco, G. 325 Mansfeld, J. 513 Marczinovits, I. 269, 691 M~irczy, Sz.J. 547 Marinas, J.M. 505 Markovi6, I. 53 Markovi6-Dev~,i6, B. 53 Marques, S.R. 435 Marr, R. 471 Martin, A.B. 679 Martin, M.T. 535 Marty, J.-L. 257

Mateo, C. 349 Mater, D. 591 Maycock, C.D. 483 McKeon, U. 47 Milana, G. 667 Misset, O. 3 Miura, T. 337 Moln~r, J. 269, 691 Monsan, P. 535 Moore, B.D. 373 Moracci, M. 311 Moreno, J.A. 653 Mozhaev, V.V. 355 Mozo-Villarias, A. 303 MiJller-Fembeck, B. 83 Murooka, Y. 295 Nair, U.B. 263 Nava Saucedo, J.E. 591 N6meth, Sz.A. 547 Nicolas, P. 529 Nidetzky, B. 19, 83 lqiguez, M.J. 411 Nishiya, Y. 295 Nogueiro, E. 487 Nomura, K. 337 Norde, W. 495 Nffiez, M.J. 611 Oberreuter, H. 517 Obreg6n, V. 89 O'Connor, B. 47 O'F~ig/tin, C. 47 Ortaggi, G. 667 Ort6ga, E 257 Ortega, N. 157 Otzen, D. 147 Palocci, C. 667 Pardo, M.A. 229 Parody-Morreale, A. Partridge, J. 373 Pastuhova, T.A. 65 Pavlovi6, N. 53 Pearl, L.-H. 311

247

756 Pencreac'h, G. 459 Peres, C. 483, 487 Perez-Mateos, M. 157 P~rez-Pons, J. 303 Perpifia, B. 657 Petrovska, B. 517 Picciolato, M. 679 Pickersgill, R.W. 41 Pietzsch, M. 517 Pineda, T. 697 Pinheiro, H.M. 625 Pires, E. 731 Pirozzi, D. 59 Planas, A. 303 Planche, H. 393 Plaza del Pino, I.M. 235, 241 Plou, EJ. 115, 535 PodgQrnik, A. 541 Podgornik, H. 541 Pohl, M. 331 Polakovi6, M. 77 Polverino de Laureto, P. 277, 381 Pons, J. 303 Pulvin, S. 685 Querol, E.

303

Raetz, E. 529 Ragnitz, K. 517 Ram6n, E 89 Reinhoudt, D.N. 429 Remaud, M. 535 Reymond, S. 529 Roblot, C. 709 Roca, E. 611 Rodes, L. 405 Rodriguez Couto, S. 703 Roman, A.J. 697 Romero, A.A. 505 Romero, L.E. 107 Rosell, C.M. 405 Ross, A.C. 365 Rossi, M. 311,325 Rubtsova, M.Yu. 65

Sala, N. 657 S~inchez, A. 735 S~inchez-Montero, J.M. 453, 565, 735, 741 Sanchez-Ruiz, J.M. 235, 241,247, 251 Sangeeta Devi, Y. 263 Sanromhn, A. 703 Sarazin, C. 709 Sarmento, A.C. 731 Sauvageat, J.L. 529 Scaramella, E. 277, 381 Schwab, H. 471 S~guin, J.P. 709 Sereti, V. 725 Serra, J.L. 229 Sevilla, J.M. 697 Sheldon, R.A. 447 Shibata, H. 337 Silva, A. 487 Simon, L.M. 547, 713 Sinisterra, J.V. 453, 565, 571,735, 741 Sisak, C. 477 Sisak, Cs. 101 Slotboom, A.J. 127 Slusarczyk, H. 331 Soares, C.M. 483 Solomon, B. 183 Somiari, R.I. 423 Soro, S. 667 Souhail, B. 235 Sousa, H.A. 673 Stamatis, H. 725 Stefanovits-B~inyai, E. 477 Strancar, A. 541 Surinenaite, B. 577 Svensson, B. 343 Syldatk, C. 517 Szab6, P.T. 269 Szaj~ni, B. 477, 547 Thomas, D. 591,685 Thompson, D. 349 Tomita, K. 337 Torres, J. 719 Tortes, M. 657

757 Torres, R. 719 Tortora, P. 217 T6th, G. 691 Toyama, M. 295 Truffaut, N. 591 Uitdehaag, J.C.M. 317 Ulbrich-Hofmann, R. 121,513 Ulbricht, M. 661 Vallmitjana, M. 303 van Berkel, W.J.H. 141 van den Ban, E. 619 Van den Burg, B. 513 van den Heuvel, R.H.H. 141 van Dijk, A. 3 van Rantwijk, F. 447 van Unen, D.J. 429 Venema, G. 513 Ventina, E. 603 Venyige, T. 101 Verheij, H.M. 127

Vilchez, C. 631 Vina, I. 603 Vinokurov, L.M. 645 Visser, J. 41 Vodopivec, M. 541 Voss, H. 35 Vrfibel, P. 77 Vriend, G. 513 Wassink, H. 619 Willemen, H. 619 Williamson, G. 41 Wilson, L. 27 Xu, X.

441

Yakhnin, A.V. 645 Yamashita, M. 295 Yoo, Y.J. 465 Zambonin, M. Zoungrana, T.

277, 381 495


Stability and Stabilization of Biocatalysts

Slope stability and stabilization methods

Stability and Stabilization of Nonlinear Systems

Stability and stabilization of nonlinear systems

Stability and Stabilization of Nonlinear Systems

Stability and Stabilization of Time-Delay Systems

Stability and Stabilization of Linear Systems with Saturating Actuators

Industrial Application of Immobilized Biocatalysts

Stabilization and Degradation of Polymers

Self-Stabilization

Stabilization of Polymers and Stabilizer Processes

Chemical Grouting and Soil Stabilization

Torrent control and streambed stabilization

Boundary stabilization of thin plates

Pharmaceutical Photostability and Stabilization Technology

Industrial Application of Immobilized Biocatalysts (Biotechnology and Bioprocessing Series)

Performance and Stability of Aircraft

A0-Stability and stiff stability of Browns multistep multiderivative methods

Stability

Global controllability and stabilization of nonlinear systems

Stability and Stabilization of Time-Delay Systems (Advances in Design & Control)

Stability

Stability

Stability

Stability

Boundary stabilization of thin plates

Fundamentals of Structural Stability

Structural Stability And Morphogenesis

Energy stability and convection

Fundamentals of Structural Stability

Stability of queueing networks

Stability and Stabilization of Biocatalysts

Slope stability and stabilization methods

Stability and Stabilization of Nonlinear Systems

Stability and stabilization of nonlinear systems

Stability and Stabilization of Nonlinear Systems

Stability and Stabilization of Time-Delay Systems

Stability and Stabilization of Linear Systems with Saturating Actuators

Industrial Application of Immobilized Biocatalysts

Stabilization and Degradation of Polymers

Self-Stabilization

Stabilization of Polymers and Stabilizer Processes

Chemical Grouting and Soil Stabilization

Torrent control and streambed stabilization

Boundary stabilization of thin plates

Pharmaceutical Photostability and Stabilization Technology

Industrial Application of Immobilized Biocatalysts (Biotechnology and Bioprocessing Series)

Performance and Stability of Aircraft

A0-Stability and stiff stability of Browns multistep multiderivative methods

Stability

Global controllability and stabilization of nonlinear systems

Stability and Stabilization of Time-Delay Systems (Advances in Design & Control)

Stability

Stability

Stability

Stability

Boundary stabilization of thin plates

Fundamentals of Structural Stability

Structural Stability And Morphogenesis

Energy stability and convection

Fundamentals of Structural Stability

Stability of queueing networks

Recommend Documents