Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticals

THE SOCIETY FOR APPLIED BACTERIOLOGY TECHNICAL SERIES NO. 25 Rapid Microbiological Methods for Foods, Beverages and Pha...

Author: C. J. Stannard | S. B. Petitt | F. A. Skinner

183 downloads 1063 Views 7MB 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

THE SOCIETY FOR APPLIED BACTERIOLOGY TECHNICAL SERIES NO. 25

Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticals kv C.]. STANNARD Edited

Applied Researdl Department, Pedigree PeJjoods, iHelton Mowbray, Leicestershire LEl3 lBE

S. B. PETITT U.B. (Ross Youngs) Ltd, Ross House, Wickham Road, Grimsby, South Humberside DN31 3SW'

F. A. SKINNER Harpenden, /fens

BLACKWELL SCIENTIFIC PUBLICATIONS OXFORD LONDON EDINBURGH BOSTON MELBOURNE

RAPID MICROBIOLOGICAL METHODS FOR FOODS, BEVERAGES AND PHARMACEUTICALS

A complete list of titles in the Society for Applied Bacteriology Technical Series appears at the end of this volume

THE SOCIETY FOR APPLIED BACTERIOLOGY TECHNICAL SERIES NO. 25

Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticals kv C.]. STANNARD Edited

Applied Researdl Department, Pedigree PeJjoods, iHelton Mowbray, Leicestershire LEl3 lBE

S. B. PETITT U.B. (Ross Youngs) Ltd, Ross House, Wickham Road, Grimsby, South Humberside DN31 3SW'

F. A. SKINNER Harpenden, /fens

BLACKWELL SCIENTIFIC PUBLICATIONS OXFORD LONDON EDINBURGH BOSTON MELBOURNE

©

1989 by the Society for Applied Bacteriology and published for them by Blad.-wcll Scientific Publications Editorial offices: Osney Mead, Oxford OX2 OEL 8 John Street, London WCIN 2ES 23 Ainslie Place, Edinburgh EH3 6AJ 3 Cambridge Center, Suite 208 Cambridge, Massachusetts 02142, USA 107 Barry Street, Carlton Victoria 3053, Australia

British Library Cataloguing in Publication Data

AU rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior permission of the copyright owner.

ISBN 0·-632-02629-4

First published 1989

Rapid microbiological methods for foods, beverages and pharmaceuticals. L Industrial microbiology. Laboratory techniques I. Stannard, CJ. II. Petitt, S.B. III. Skinner, F.A. (Frederick Arthur, 1919~) 1I1l. Series

660'.62'028

Library of Congress Cataloguing-in-Publication Data Rapid :\1icrobiological methods for foods, beverages, and pharmaceuticals/edited by C. J. StaIUlard, S. B. Petitt, F. A. Skinner. p. em. - (Technical series; flO. 25) Papers presented at the Society for Applied Bacteriology Demonstration Meeting, University of Bath, 30th September 1987. ISBN 0-632-02629-4

DISTRIBUTORS

Marston Book Services Ltd PO Box 87 Oxford OX2 ODT

(Orders: Tel: 0865 791155 Fax: 0865 791927 Telex: 837515)

USA Publishers' Business Services PO Box 447 Brookline Village Massachusetts 02147 (Orders: Tel: (617) SH~7678) Canada Oxford University Press 70 Wynford Drive Don ~1ills Ontario M3C IJ9 (Orders: Tel: (416) 441- 2941) Australia Blackwell Scientific Publications (Australia) Pty Ltd 107 Barry Street Carlton, Victoria 3053

(Orders: Tel: (03) 347-0300)

1. Food - .1\1icrobiology - Technique Congresses. 2. Beverages - .Mkrobiology Technique - Congresses. 3. Drugs Microbiology - Technique - Congresses. I. Stannard, c.]. (Catherine].) II. Petitt, S. B. Ill. Skinner, F. A. (Frederick Arthur), 1919- . IV. Society for Applied Bacteriology. Demonstration Meeting (1987: University of Bath) V. Series: Technical series (Society for Applied Bacteriology) no. 25. QRl15.R36 1990 664'.07 - dc20 89-17715

eIP

Contents

Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

x

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xiii

The Use of ATP Bioluminescence for the Analysis of Beer in Polyethylene Terephthalate (PEl) Bottles and Associated Plant. ]. W. AVIS AND P. SMITH

I

Materials and methods, 2 Results, 6 Discussion, 9 Acknowledgements, 10 References, 10

Rapid Assessment of the Bacterial Content of Milk by Bioluminescent Techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. W. GRIFFITHS AND]. D. PHILLIPS

13

lv1aterials and methods, 14 Results, 17 Discussion, 26 References, 29

DEFT': Recent Developments for Food and Beverages. . . . . . . . . . L. PETTIPHER, R. G. KROLL, L. J. FARR AND R. P. BETTS

G.

General principles, 33 Apparatus, reagents and methodology, 33 Selective pre-incubation for the detection of low levels of spoilage bacteria by the DEFT, 36 Osomotolerant yeasts in confectionery products, 38 Selective enumeration of bacteria by the DEFT, 39 Use of DEFT for irradiated foods, 42 References, 44

v

33

vi

CONTENTS

The Rapid Estimation of Bacterial Counts on Meat and Poultry by the Direct Epifluorescent Filter Technique. . . . . . . . . . . . . . . . . B. G. SHAW AND L. J. FARR

47

DEFT methodology for meat and poultry, 47 Comparison of DEFT and plate counts, 52 Applicability of DEFT to meat and poultry, 5S Acknowledgement, 56 References, 56

Medical and Pharmaceutical Applications of the Direct Epifluorescent Filter Technique (DEFD

S. P.

DENYER,

R. A. P.

LYNN AND Urine examination by the DEFT, 59 Analysis of intravenous fluids by the DEFT, 64 Conclusion, 69 References, 70

P. S.

,..

59

POVER

The Use of Image Analysis for MIC Determination and Bioassay B. J. BROOKS AND K. COLEMAN

73

What is image analysis?, 73 Image analysis at Brockham Park, 74 Application to bioassay, 75 Application to MIC determination, 78 References, 85

Optimization of Automated Electrometric Methods. . . . . . . . . . . .

D. M.

87

GIBSON

General hints, 89 Curve quality, 89 Comments on some conventional assays, 90 Temperature, 91 Calibration curves, 91 Practical experiments on fish, 93 Pathogen detection, 96 Inorganic constituents, 96 Conclusions, 98 References, 98

Conductance Techniques for the Detection of Contaminants in Beer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. L. KYRIAKIDES AND P. A. THURSTON General procedures, 102 Screening of non-selective media, 103 Detection routine evaluation, 104 Development of selective media, 105

101

CONTENTS

VII

Field trial, 109 Discussion, 115 References, 117

Electrical Methods for Water Quality Testing T. E. IRVING, G. STANFIELD AND B. W. T.

..........

119

HEPBURN

Conventional water tests, 120 Use of the Bactometer M123 for environmental and recreational water samples, 120 Use of the Bactometer M123 for potable water samples, 123 Use of the Malthus microbial growth analyser for potable water analysis, 125 Acknowledgement, 130 References, 130

A Conductance Screen for Enterobacteriaceae in Foods S. B. PETITT

131

Rapid tests for the Enterobacteriaceae, 132 Materials and methods, 133 Result~, 137 Discussion, 139 Acknowledgements, 140 References, 140

Electrical Screening of Powdered Dairy Products . . . . . . . . . . . . . . SHEILA M. FRYER AND KATE Materials and methods, 144 Result~, 147 Discussion, 151 Acknowledgement, 153 References, 153

143

FORDE

An Inter-Laboratory Evaluation of an Electrometric Method for Detection of Salmonellas in Milk Powders. . . ... . . . . . . . . . . ... . G. A. PRENTICE, P. NEAVES, D. I. JERVIS AND M. C. EASTER Participating laboratories, 156 Materials and methods, 157 Comparison of the British Standard method v.ith the Easter-Gibson method, 159 Results and discussion, 160 Acknowledgements, 164 References, 164

155

viii

CONTENTS

Rapid Salmonella Detection by a Combination of Conductance and Immunological Techniques JULIE A. BIRD, M. C. EASTER, S. GAYE IIADFIELD, E. MAY AND M. F. STRINGER

165

Materials and methods, 166 Results, 172 Discussion, 179 Acknowledgement, 182 References, 182

Automated Conductimetric Detection of Salmonellas in Confectionery Products. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S.]. PUGH AND M. L. ARNOTT

185

Description and principles, 186 Method evaluation, ] 95 Conclusion, 199 Acknowledgement, 200 References, 200

A Medium for Detection of Lancefield Group D Cocci in Skimmed Milk Powder by Electrometric Methods ..... . . . . . . . . P. NEAVES, M. J. WADDELL AND G. A. PRENTICE

203

Preparation of skimmed milk powder contaminated with Lancefield Group D cocci, 203 Use of conventional selective media in elcctrometric instruments, 205 Medium development for electrometric detection, 206 Effect of medium composition on curve quality, 207 Calibration of Malthus and Bactometer instruments, 209 Acknowledgements, 211 References, 211

BIOCHECK - a Mediated Amperometric Microbial Activity Monitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. S\VAIN, M. ALLEN, B. H. SCHNEIDER, F. TAYLOR AND A. P. F. TURNER Requirements for amperometric biomass sensing, 214 Evaluation of suitable mediators, 217 Assessment of electrodes, 219 Prototype development, 220 Considerations for the examination of real samples, 224 Acknowledgements, 225 References, 225

213

CONTENTS

Detection of Electron Transfer for the Assessment of Bacterial Contamination. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. G. KROLL, R. A. PATCHETT, STEPHANIE E. LEAROYD AND C. F. THURSTON

~

227

Materials and methods, 229 Results and discussion, 233 References, 239

Computer-Assisted Identification of Moulds. . . . . . . . . . . . . . . . . .

241

A. P. WILLIAMS AND ANIA BIALKOWSKA The evolution of mould identification, 242 Methods, 243 Conclusion, 248 References, 248

Immunological Detection Methods for Salmonellas in Foods. . . . C. DE W. BLACKBURN AND CATHERINE J. STANNARD

249

Bio-Enzabead Screen Kit, 249 TECRA Salmonella Visual Immunoassay, 253 Kirkegaard and Perry Salmonella ELISA, 256 Salmonella 1-2 Test, 259 Discussion, 262 References, 263

Immunoassay Kits for the Detection of Toxins Associated with Foodbome Illness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SALLY A. ROSE, N. P. PATEL, A. O. SCOTT AND

265

M. F. STRINGER Materials and methods, 266 Results and discussion, 267 Concluding remarks, 280 Acknowledgements, 280 References, 280

Rapid Detection of Viruses in Water and the Water Environment H. MERRETT AND C. E. STACKHOUSE

283

lVlaterials and methods, 284 Results, 289 Discussion, 289 References, 290,

Index

293

Contributors

M. ALLEN, Bioelectronics Division, Biotechnology Centre, Cranfield Institute of Technology, Cranfield, Bedfordshire MK43 OAL, UK M. L. ARNOTT, Cadbury Schweppes PLe, Group Research, The Lord Zuckerman Research Centre, The University of Reading, Reading RG6 2LA, UK J, W. AVI5, Research and Development Laboratory, Allied Breweries Ltd, 107 Station Road, Burton-on-Trent, Derbyshire DEJ4 1HZ, UK R, P. BE TT 5, Campden Pood (5 Dn'nk Research Association, Chipping Campden, Gloucestershire GL55 6LD, UK ANIA BIALKOWSKA, Leatherhead Food RA" Randal/s Road, Leatherhead, Surrey KT22 7RY, UK JULIE A. BIRD, Microbiology Department, Unilever Research and Engineering, Colworth Laboratory, Co/worth House, Sharnbrook, BedjOrdshire MK44 1LQ UK C. DE W. BLACKBURN, Applied Microbiology Section, Leatherhead Food RA., Randalls Road, Leatherhead, Surrey KT22 7RY, UK B. J. BROOKS, Beecham Phannaceuticals Research Division, Chemotherapeutic Research Centre, Brockham Park, Betchworth, Surrey RH3 7A], UK K. COLEMAN, Beecham Phannaceuticals Research Division, Chemotherapeutic Research Centre, Brockham Park, Betehworth, Surrey RH3 7AJ, UK S. P. DENYER, Department ofPh annaceuticaI Sciences, University ofNottingham, University Park, Nottingham NG 7 2RD, UK M. C. EASTER, Express Foods Group LtdJ 430 Viaoria Road, South Ruislip, Middlesex I/A4 OHF, UK L. J. F ARR, Foss Elean'c (UK) Ltd, The ChantryJ Bishopsthorpe, York Y02 IQf, UK KATE FORDE, Unigate Foods Ltd, Station Road, Wincanton, Somerset BA99ED UK SHEILA M. FRYER, St Ivel Technical Centre, Abbey J/ouse, Church Street, BradfOrd-on-Avon, Wiltshire BAIS JDH, UK D, M. GIBSON, Ministry of Agriculture, Fisheries and Food, Torry Research Station, J35 Abbey Road, Aberdeen AB9 8DG, Scotland, UK M, W. GRIFFITHS, Hannah Research Institute, Ayr KA6 SHL, Scotland, UK S. GA YE HADFIELD, Wellcome Research Laboraton'es, Langle,.y Court, Beckenham, Kent BR3 3BS, UK J

x

CONTRIBUTORS

xi

B. W. T. HEPBURN, Wessex Water, Regional Scientific Centre, Mead Lane, SaltfOrd, Bristol BS18 3ER, UK T. E. IRVING, Water Research Centre, Henley Road, Medmenham, PO Box 16, Marlow, Buckinghamshire SL7 2l/D, UK D. I. JERVIS, St Ivel Technical Centre, Abbey House, Church Street, Bradfordon-Avon, Wiltshire BA15 1DH, UK R. G. KROLL, Department of Microbiology, AFRC Institute of Food Research, Reading Laboratory, Shinfield, Reading RG2 9AT, UK A. L. KYRIAKIDES, Grand Metropolitan Brewing Ltd, Stag Brewery, 91 Brick lane, London E1 6QN, UK STEPHANIE E. LEAROYD, Department of Microbiology, King's College, University of London, Campden Hill Road, London W8 7AH, UK R. A. P. LYNN, Department of Pharmaceutical Sciences, Universi~v of Nottingham, Universi~y Park, Nottingham NG7 2RD, UK E. MAY, School of Biological Sciences, Portsmouth Po~ytechnic, King Henry I Street, Portsmouth, Hampshire POI 2DY, UK H. MERRETT, Virology Unit, Welsh Water PLC, Engineering and Environment Ltd, Bridgend Ojfice and Laboratory, Tremains House, Tremains Court, Bridgend, Mid Glamorgan CF31 2AR, Wales, UK P. NEAVES, Technical Division, Milk Marketing Board, Thames Ditton, Surrey KT70EL, UK R. A. PATCHETT, Department ojMicrobiology,AFRC Institute ofFood Research, Reading Laboratory, Shinfield, Reading RG2 9AT, UK N. P. PATEL, Campden Food (5 Drink Researth Association, Chipping Campden, Gloucestershire G155 6LD, UK S. B. PETITT, U.B. (Ross Youngs) Ltd, Ross House, Witkham Road, Gn'ms~v, South Humberside DN31 3SW, UK G. L. PETTIPHER, CadburySchweppesPLC, Group Research, TheLordZuckerman Research Centre, The University of Reading, Reading RG6 2LA, UK J, 0, PHILLIPS, Hannah Research Institute, 4vr KA6 SHL, Scotland, UK P. S. POVER, Ana{ytical Measun'ng s.ystems, London Road, Pampis.fOrd, Cambridge, CB2 4EF, UK G. A. PRENTICE, Milk Marketing Board, Thames Ditton, Surrey KT7 OEL, UK S. J. PUG II, Cadbury Ltd, Technical Laboratories, Bournville, Birmingham B30 2LU, UK SALL Y A. ROSE, Campden Food (5 Drink Research Association, Chipping Campden, Gloucestershire GLSS 6LD, UK B. H. S C H N EID ER, Bioelectronics Division, Biotechnology Centre, Cranfield Institute of Technology, Cranfield, Bedjordshire MK43 OAL, UK A. O. SCOTT, CampdenFood (5 Drink Research Association, ChippingCampden, GlouteStershire GL55 6LD, UK

xii

CONTRIBUTORS

B. G. S HA W, AFRC Institute of Food Research Bristol Laboratory, LangfOrd, Bristol BS18 7DY, UK P. SMITH, Research and Development Laboratory, Allied Breweries Ltd, 107 Station Street, Burton-an-Trent, Derbyshire DE14 1BZ, UK C. E. STACKHOUSE, Virology Unit, Welsh Water PLC, Engineen'ng and Environment Ltd, Bridgend Office and Laboratory, Tremains House, Tremains Court, Bridgend, Mid Glamorgan CF3 J 2AR, Wales, UK G. STANFIELD, Water Research Centre, Henley Road, jWedmenham, PO Box 16, Marlow, Buckinghamshire SL 7 2HD, UK C. ]. STANNARD, Applied Research Department, Pedigree Petfoods, Me/ton Mowbray, Leicestershire LE13 7BR, UK ~1. F. S T RING ER, Department ofA1icrobiology, Campden Food (5 Drink Research Assoa"ation, Chipping Campden, Gloucestershire GLSS 6LD, UK A. SWAIN, Bioelectronics Division, Biotechnology Centre, Cranfield Institute of Technology, Cranfield, Bedfordshire MK43 OAL, UK F. TAYLOR, Bioelectronics Division, Biotechnology Centre, Cranfield Institute of Technology, Cranfield, BedjOrdshire MK43 OAL, UK C. F. THURS TON, Department of Microbiology, King's College, Um'versity of London, Camden Hill Road, LontkJn SW14 7ET, UK P. A. THURSTON, Grand Metropolitan Brewing Ltd, Mortlake, London SW14 7ET, UK A. P. F. TURNER, Bioelectronics Division, Biotechnology Centre, Cranfield Institute of Technology, Cranfield, BedjOrdshire MK43 OAL, UK M. J. WADDELL, Technical Division, Jl1ilk Marketing Board, Thames Ditton, Surrey K17 OEL, UK A. P. WILLIAMS, Leatherhead Food R.A., Randal/s Road, Leatherhead, Surrey KT22 7RY, UK

Preface

This book is the 25th in the Technical Series of the Society for Applied Bacteriology. Each chapter is the written version of a practical contribution given at the Demonstration Meeting of the Society, held at the University of Bath on 30 September, 1987. For many years, more rapid and labour-saving methods have been sought as alternatives to conventional microbiological techniques. Many of the pioneering studies have taken place in clinical laboratories. For foods, beverages and pharmaceuticals, the materials tested and the organisms sought are more varied. The contributions to this book illustrate the wide variety of approaches that workers in these industries have taken in order to solve the particular problems associated with their own products. The methods described in this book include electrometric techniques, ATP assay, and immunological methods for a wide range of organisms from salmonellas to viruses. We feel that it is apparent that the choice of a rapid method for industry depends upon the equipment available and the accuracy required. This book should be useful to those in the food, beverage and pharmaceutical industries, or in research or teaching, who require a practical guide to the use of rapid microbiological methods. We should like to thank the contributors for all their hard work in preparing the demonstrations and contributions for the book, and Dr Ron Board and his staff at the University of Bath for the organization of the meeting.

J. Stannard S. B. Petitt F. A. Skinner

Catherine

Xlll

The Use of ATP Bioluminescence for the Analysis of Beer in Polyethylene Terephthalate (pET) Bottles

and Associated Plant J. W. AVIS" AND P. SMITH Research and Development Laboratory, Allied Breweries Ltd, 107 Station Street, Burton-on-Trent, Der~yshire DEl4 IBZ, UK

The growing demand for products packaged in polyethylene terephthalate (PET) bottles (which cannot be pasteurized) has led to increased emphasis on plant hygiene and end-product quality assurance. Rapid methods of detecting microbial contamination are especially useful for these products and this work concentrates on the rapid membrane filtration of products directly from PET bottles and the analysis of membrane filters and production plant swabs by adenosine triphosphate (ATP) bioluminescence. Many changes are currently taking place in microbiology. Micro-organisms are being studied increasingly as part of their natural environment rather than in isolation; automation has allowed large numbers of 'routine' samples to be examined and there is increased use and development of rapid methods of detection. In the brewing industry rapid methods would be particularly helpful to detect process failure, assess the quality of pitching yeast and the microbiological status of packaged products (Hope & Tubb 1985). Two electrical methods which have shown promise are impedence measurement as a rapid forcing test for beer (Evans 1982) and conductance measurement for the rapid detection of both lactobacilli in beer (Evans 1985) and Obesumbacterium proteus in pitching yeast (Kilgour & Day 1983). The sensitivity of the method is low, however, because cells are detected only when they reach a level of 105 -1 0 6/ ml and incubation times can be in excess of 48 h for some organisms.

" Present address: Group Quality Control Laboratory, Bass Brewing Ltd, 137 High Street, Burton-on-Trent, Derbyshire DE14 IJZ, UK. Copyright Rapid fvlicrobiological .~lethods for Foods, Beverages and Pharmaceuticals

©

1989 bJ' the Society jiJr Applied /JaaerioloJ{)' All rights of rtpmdlletirffl in any jimn reserved

0-632-02629-4

2

J.

W. AVIS AND P. SMITH

The Direct Epifluorescent Filtration Technique (DEFT) has been used with success for milk (Pettipher et al. 1980) and to a more limited extent for beer (Kilgour & Day 1983) where the method for heat-treated samples was improved by counterstaining with methylene blue. Counterstaining with Janus Green B was found to give more consistent results than methylene blue (Rodrigues & Kroll 1986) but DEFT preparations on heat-treated beverages were still found to be unreliable in differentiating between viable and nonviable yeast cells. Adenosine triphosphate bioluminescence has been used to detect microbial contamination in carbonated beverages (Littel & La Rocco 1986), wine (Lonvaud-Funel & Joyeux 1982) and beer (Hyserl et al. 1976; Kilgour & Day 1983; Dick et al. 1986). The methods detect only viable organisms at levels of the order of 100 yeast cells or 1000 bacterial cells/ml suspension. In the work to be described ATP bioluminescence was used specifically for the problems of PET bottled beer. The PET bottle is a popular package in the UK for many bottled drinks and sales arc continuing to increase. The package cannot be pasteurized or heat-treated, however, so plant hygiene is particularly important. The ATP method has a role in assuring the quality of both packaging plant and finished product. In developing the method we have concentrated on the detection of yeasts rather than bacteria, following our findings that yeasts were the cause of over 95 % of contamination problems examined over an 8-month period (Avis 1988).

Materials and Methods

ATP analysis The work was carried out using the Lumac Biocounter 2010 (Lumac BV, Schaesberg, The Netherlands). Samples are contained in disposable plastic cuvettes which are inserted into a light-tight chamber for reading by a sensitive photomultiplier tube. The amount of light emitted is displayed as a digital readout and is expressed in Relative Light Units (RLUs). The absolute amount of ATP contained in a sample may be found by injecting an ATP standard into the sample and taking a second reading. To be accurate the RLU value of the standard should be two to five times that of the sample so a range of standards are made up from a stock solution and the RLU values of these are checked before samples arc analysed. The reagents and standards required for the analyses were supplied by Lumac, and were: 'Somase' (a non-microbial ATP-ase), 'F-NRS' (used here as a buffer), 'NRB' (a nucleotide-releasing agent for the extraction of ATP from microbial cells) and 'Lumit-PM' (luciferin-Iuciferase reagent). ATP standards were made up from a stock solution of 1.65 x 10- 6 moll I ATP. All

ATP ANALYSIS FOR PET BOTTLED BEER

3

powdered reagents were stored between 0° and 2°C and, when reconstituted, Lumit-PM, Somase and ATP standard stock solution were split into 1- or 2-ml aliquots which were stored for a maximum of 4 weeks at -18°C. Temperature-sensitive reagents were kept on ice during analysis.

PET bottle filtration device Filtration methods currently used for PET bottled products suffer from two disadvantages. Firstly, the majority of commercial membrane filtration units have only a 250-ml reservoir for beer, which is inconvenient if beer has to be continually poured into the unit, particularly if a laminar flow cabinet is not available. Secondly, there exists the possibility of contamination of the beer from the outside of the neck of the bottle as the beer is poured into the filter funnel. This has led to a piercing technique in which a portion of the bottle is swabbed with 70% methylated spirit and then pierced with a hot needle; the beer is then poured into the filter. This technique is also inconvenient. The device which will be described here was developed to avoid these problems by allowing the entire contents of the bottle to be taken directly from an upright bottle without the necessity of pouring or piercing. The device is shown in Fig. 1 and is in two parts: a quick-release clamp and the gas inlet and beer outlet tube assembly. The 1/4 -inch National Pipe Thread fitting on the beer outlet tube accepts a conventional 47-mm membrane filter holder (e.g. Swinnex) and the whole device can be autoc1aved with the membrane in place, after which it can be attached very quickly to the bottle to be sampled. All the bottle contents are then passed through the membrane by applying a top pressure of gas which is filtered in-line. The membrane is then removed for plating on agar or ATP analysis.

Filters Conventional cellulose acetate (0.45 [lm) and Vltipore (1.2, 0.8, 0.65 [lm) nylon membrane filters (Pall Process Filtration Ltd, Portsmouth) were used. The latter type are electrostatically charged, the magnitude and polarity of the zeta potential being dependent on the pH. At pH 4 (approximately the pH of beer) the filter has a positive charge.

Swabs Plain, cotton wool sterile swabs (Northern Media Supply Ltd, Hessle) were used to assess the cleanliness of plant associated with PET bottling. Charcoalimpregnated swabs must not be used as these affect the results of ATP analysis.

4

J. W. AVIS AND P. SMITH

FIG. 1. PET filtration device. A, quick release damp; B, bras inlet and beer outlet tubes; C, NPT fitting to accept Swinnex-typc filter holder; D, in-line filter for top pressure gas.

1/4 -in

il1edia Cotton wool swabs wefC incubated after usc in 'Lumaculf, a pre-sterilized, low-ATP growth medium, at pI! 7.0 (Lumac BY). Yeast E>..1ract, Nialt Extract

Broth (Difco) adjusted to plI 4.5 was used for all other membrane incubations involving ATP analysis. \Ve have found that this broth is particularly suitable for ATP work because it has low background ATP and does not quench the luminescence too greatly. Batches do vary, however, and it is advisable to test the nledium before use. Plate counts were performed either on \Vallcrstein Laboratories Nutrient (\VLN) Agar (Difeo) for yeast spp. or on Raka Ray No. 3 Agar (Difeo) tor Pediococcus spp. and Lattohacillus spp.


5

Organisms The organisms used in this work were taken either from the Allied Breweries culture collection, where they are stored as lyophilized cultures, or from contaminated products. In both cases the organisms were sub-cultured on Malt Extract, Yeast Extract Glucose Peptone (MYGP) Agar (Difco) and into MYGP broth (Difco) to prepare a working culture for dilutions. Ana~ysis

of swabs

Before swabbing, 1 ml of Lumacult was dispensed into disposable plastic, sterile, 30-ml screw-capped bottles. These were found to be most suitable because of their conical bases which allowed maximum coverage of a swab tip with the minimum volume of medium. Swabs for ATP analysis were broken olf into these bottles for incubation whilst swabs for conventional analysis were broken off into 10 ml of 1/4 - strength, sterile Ringer solution. The Ringer solution was agitated on a vortex mixer for 30 s before membrane filtration. The membrane was placed on WLN agar plates. The swabs in Lumacult were incubated overnight (16 h minimum) at 27°C before examination by the following method: tubes were agitated for 30 s on a vortex mixer and a 100 I-ll sample of the Lumacult transferred to a cuvette. 100 I-li of NRB were added to the cuvette and after 60 s, 100 I-ll of Lumit PM were added and, without further mixing, the cuvette was placed in the Lumac Biocounter. The number of RLUs was read after a lO-s integration period. The whole procedure took place at ambient temperature.

Product

ana~ysL~

PET-bottled beer was membrane-filtered using either the conventional apparatus or the PET filtration device and the membranes examined by conventional plating procedures or by ATP analysis. The ATP analysis was carried out as follows: after membrane filtration, the membrane was removed from the holder and placed in a 50-mm Petri dish. Yeast-malt extract broth (YMB), 500 I-lI, at pH 4.5 was added to the membrane which was incubated at 27°C for 20 h. FNRS buffer and 20 I-ll V4 strength Somase were added and the mn-lure allowed to stand for 30 min at room temperature. ATP was extracted with 500 I-ll NRB for 60 sand 200 I-ll of the extract was transferred to a sample cuvette which was placed in the Lumac Biocounter. Lumit-PM (100 ~ll) was added and integration started. The sensitivity of the method was tested by spiking l00-ml volumes of beer with 1 ml of several serial dilutions of contaminant yeasts. The beer

6


was membrane-filtered and the membrane examined by the bioluminescence method above. The original serial dilutions were also spread on WLN agar for conventional plate counts.

Detection level A blank or control RLU value was obtained by testing an uninoculated broth or membrane before each series of tests. Both swab and membrane cultures were considered to be positive when the RLU value of the sample exceeded the blank value plus three times the standard deviation for this value. 'I'his threshold value was arrived at from a consideration of normal distribution theory. The standard deviation for blank tests was found to be c. 8 RLU. For example, a blank value of 15 RLU would give a detection level of 40 RLU.

Filtration times Filtration times of different membranes were found by timing the filtration of 2 I of beer by the PET filtration device. Top gas pressures from 15 to 50 Ibl in2 gauge pressure (psig) were provided from a nitrogen cylinder.

Filtration efficiencies The effidency of filters was tested by filtering 100 ml of beer containing

different numbers of organisms through the filter of interest. The filtrate was then passed through a second filter of 0.22 ~m pore size. Filters were plated on appropriate media. The range of organisms used for these tests was chosen for their beer spoilage potential and were as follows: Acetobacter rancens NCIB 6429; Laaobacillus brevis NCIB 8847; Pediococcus acidilactici NCIB 6990; Pediococcus cerevisiae NeIB 8066; Obesumbaaerium proteus AB68; Zymomonas anaerobia NCIB 8227; Zymomonas mobilis NeIB 3938; Saccharomyces cerevisiae strains AB7, AB70, AB140; and wild yeasts XY69, XY70, XY71, XY72, XY73, XY74, XY75.

Results Swabs Of 400 swabs taken routinely from different parts of a working bottling machine 66 % of ATP results agreed with the plate count taken at the same time, 27% were falsely positive (positive result by ATP testing without a corresponding plate count), and 9% were falsely negative (negative result by ATP testing with a corresponding plate count).

7


Product

An example of filtration rates for various filters is seen in Fig. 2 for 2-1 PET bottles of lager (original gravity I032°Sacch.). The efficiencies of the filters are shown in Table 1. The mean values for retention at four different levels of scven ycast strains and seven bacterial species were greatcr than 95% and there was no significant difference between filter types at the 5% level for yeast or bactcria. Figure 3 shows the log molar concentration of ATP plotted against log number of colony forming units for inocula of three yeasts and two filter types.

.

10.0

.~

~~

8.0

~

E 6.0 E .;:; c

4.0

~

2.0

0 .;:;

LL

;~

~.•

~---. t -::t==-

!

!

20

30

!

10

c:i

40

50

Pressure (psig)

2. Filtration times for a 2-1 PET bottle of lager expressed in min/I. Membrane filter types: ., 0.45 11m cellulose acetate; A, 0.65 11m Ultipore N66; ., 0.80 11m Ultipore N66; +, 1.20 [Tm Ultipore N66.

FIG.

TABLE

I. Efficiency of various membrane filters fOr different levels of bacteria and yeast % Retention" (membrane type and pore size)

Nominal loading (cells/membrane)

I 10 100 1000

Sartorius (0.45 [tm) 100 97.6 98.1 96.4

(I DO)t (97.0) (94.0) (93.3)

Ultipore (0.65 fUn) 91.7 96.9 96.1 95.3

(100) (99.9) (95.2) (92.9)

Ultipore (0.8 f!.m) 93.2 94.6 97.8 96.7

(100) (94.1) (96.6) (97.1)

" Mean retentions for four replicates of seven different strains or species. t Figures in brackets indicate efliciencies for yeast.

Ultipore (1.2 f!.m) 100 96.8 97.1 97.5

(100) (97.3) (97.6) (94.8)

8

J.

W. AVIS AND P. SMITH

-4 c

2

';='

1

0

~

C m ()

c 0

-6

4 3

u Cl.

I0::(

m

"0

E

+

-8

*

~

en

X 0

0 ....J

-10 0

0.5

1.5

2

2.5

Log 10 plate count/ml FIG. 3. Correlation curves between ATP concentration and number of CFUs for different filters and for inocula of three strains of contaminant yeast. Strain XY70: 0,0.45 !-lm filter; *, 1.20 11m filter (combined for regression line 2, r = -0.99). Strain XY71: 6, 0.45-(!m filter; +, 1.20-jJ.m filter (combined for regression line 1, r = -0.97). Strain XY72: 0, 0.45-flm filter (regression line 3, r = -0.91); x, 1.20-jJ.m filter (regression line 4, r = -0.97).

TABLE

2. Detection ofyeasts in 250-ml volumes of beer* (Yo Detection

Yeast

!VIean no. cells

Lumac

Plate

Difference

XY73

5

100 29

100 57

n.s. n.s.

7 1

100

57

100 29

n.s. n.s.

5 1

100

100

0

43

n.s. n.s.

2

XY74 XY75

n.s., not significant. '*' Based on a comparison of seven replicates at each level.

For two of the yeasts shown there was no significant difference between the slopes of the lines, so a single regression line has been drawn in this case. The difference between the intercepts for each yeast can be ascribed to differences in their ATP content per cell and growth rate. Table 2 compares the sensitivity for three contaminant yeasts with detection by conventional membrane filtration and plating. It was possible to detect one cell per 250 ml by ATP biolumi-


9

nescence but correlation with the plate method was not 100% until a mean level of five cells per 250 ml was reached. At levels lower than this there was no significant difference between the two methods. Discussion ATP bioluminescence combined with positively charged membrane filtration can provide an integrated approach to the rapid monitoring of PET-bottled beer and associated plant. One should exercise some caution, however, when comparing results with conventional plating procedures which are notorious for their poor correlation with other procedures (Sharpe 1980). The comparison of two methods will often give rise to false positive and false negative results. When considering a false positive by the ATP method one should allow for the possibility that it is a false negative by the plate method and vice versa. For example, if the results for swabs were examined in this light then the actual situation existing on the plant could be that 91 % (66 % + 25 %) of swabs were a true reflection of the state of the plant by the ATP method compared with 75% (66% + 9%) of swabs being a true reflection by the plate count method. We must also be cautious in comparing the results obtained from swabs taken from adjacent areas on the machine because they could be considered as two separate samples. There is little doubt, however, that the ability to obtain a result after 20 h led to improved cleaning and attention to hygiene. Rapid testing and reporting also allowed trends in contamination to be built up more quickly with persistent problem areas being identified and receiving closer attention. The effectiveness of conventional cellulosic and polycarbonate filters has been compared for brewery bacteria by Lin (1976) and the use of various large-porosity membrane filters has been described by Zierdt (1979) who ascribed their effectiveness to electrostatic charge. Kroll (1985) used electropositively-charged filters to concentrate bacteria from foods and subsequently to elute them from the membrane. The increasing availability of charged membranes is likely to increase their use for samples which take a long time to filter. In previous work, positively charged 'Posidyne' filters (Pall Process Filtration Ltd) were found to inhibit the growth of some brewery contaminants (Avis 1988) and the Ultipore filters were sought as an alternative. The inhibition was ascribed to the presence of the quaternary ammonium compounds which are used to enhance the electrostatic charge that is present on nylon filters. The results presented here have shown that large porosity filters can be used for conventional and bioluminescence procedures \vithout compromising on the numbers of micro-organisms which are retained on the filter. When used in combination with the PET filtration device the filters allow more beer t() be examined with, consequently, a greater inoculum for


10

ATP analysis. The lowest flow rate consistent with a reasonable filtration time should be sought because filtration can be a source of physiological stress which may be exacerbated at high pressures and flow rates. The rapid detection of microbial contamination in beer is not an easy task when the levels are low, typically only a few cells/I, particularly if cells have been sub-lethally damaged or stressed by pasteurization. The results here show that it is possible to detect cells at a level of 4 cells/l in 20 h provided that concentration and pre-incubation steps are included in the test. The combination of large-porosity filters with bioluminescence has an advantage over the current method for PET bottles because the whole package can be assessed rather than a 2S0-ml portion. \\'here organisms afe unevenly distributed in the bottle, which is often the case with flocculant yeast, there is a greatcr probability of detecting contamination if the levels arc also low. One must recognize, however, that the detection of any organism by ATP bioluminescence ultimately depends on its ATP content, its specific growth rate and the lag time before exponential growth begins. For example, with wild yeast cells that have doubling times of 1.02 to 1.83 h and ATP contents of 103 to 320 fg per cell we calculated that the time to detection for an inoculum of one cell before enrichment could range from 11 to 19 h under ideal conditions (Avis 1988). Alternatively) Dick et ai. (1986) concluded that after an enrichment time of 48 h and before concentration, Sacch. cerevisiae must be present at c. 2 cells/l in order to detect contamination with certainty.

Acknowledgements We wish to thank the directors of Allied Breweries for permission to publish this paper.

References

J.W. 1988. The use of ATP bioluminescence for the quality assurance of PET bottled beers. In Confermce proceedings of tlte Associaziotle Sodeta Italiatla di A1icrobiologia Applicata, ]\,letodi rapidi cd automatizziati nella nticrobiologia applicata, pp. 39-45, Societa Editoriale Farmaceutica Milan. DICK, E., WrEDMAN, R., LEMPART, K. & HAI\lMES, W.P. 1986. SchncUnachweiss mikrobeiller lnfektionen im Bier. Clmnie A1ikrohiologie Teclmologie der LebensmitteI 10, 37-41. EVANS, H.A.V. 1982. A note on two uses for impedimetry in brewing microbiology. Joumal of Applied Bacteriology 53, 423-426. EVANS, II.A.V. 1985. A note on the use of conductimetry in brewery microbiological controL Food Microbiology 2, 19-22. f lOPE, C.F.A. & TUBE, R.S. 1985. Approaches to rapid microbial monitoring in brev.ing. Journal AVIS,

of the Institute oj Brewing 91, 12-15. D.W., KOVECSES, F. & NloRR1SON, N.M. 1976. A firefly bioluminescence ATP assay method for rapid detection and enumeration of brewery microorganisms. Journal of the ,1merican Socie~y of Brewing Chemists 34, 145 - 150.

HYSERT,


11

KILGOUR, W.J. & DAY, A. 1983. The application of new techniques for the rapid detennination of microbial contamination in brewing. In Proceedings of the 19th Congress of the European Bmvery Convmtion, pp. 177 -184. Oxford: IRL Press Ltd. KROLL, R.G. 1985. Electropositively charged filters for the concentration of bacteria from foods.

Food Microbiolot,'Y 2, 183-186. LIN, Y. 1976. Use of various brands of membrane filters for the detection of brewery bacteria.

]oun!al of the American Society of Brcwing Chemists 34, 141-144. LITTEL, K.J. & LA Rocco, K.A. 1986. ATP screening method for presumptive detection of microbiolob>1Cally contaminated carbonated beverages. ]oumal of Food Scimce 51,474-476. LO[\'Vl\UD-FuNEL, A. & ]OYEUX, A. 1982. Application de la bioluminescence au denombrement des microorganismes vivants dans les "ins. Connaissance Vigue Vin 16, 241- 256. PETTlI'HER, G.L., MANSELL, R., McKINNON, C.H. & COUSINS, C.M. 1980. Rapid membrane filtration epilluorescent microscopy technique for direct enumeration of bacteria in raw milk.

Applied aud Enviroumental Microbiolot,'Y 39,423-429. RODRIGUES, U.M. & KROLL, R.G. 1986. Use of direct epilluorescent filter technique for the enumeration of yeasts. Joumal ofApplied Bacteriology 61, 139-144. SIIARI'E, A.N. 1980. Food Microbiolo&'Y - A Framework jilr the Future. Springfield, Illinois: Charles C. Thomas. ZIERDT, CJ I. 1979. Adherence of bacteria, yeast, hlood cells and latex spheres to large-porosity membrane filters. Applied and Euviro1l1nClltai Microbiology 38, 1166-1172.

Rapid Assessment of the Bacterial Content of Milk by Bioluminescent Techniques M. W. GRIFFITIIS AND J. D. PHILLIPS Hannah Research Institute, Ayr KA6 5HL, Scotland, UK

The use of bioluminescent ATP assay for the rapid detection of bacteria in foods has recendy been reviewed (Stannard & Gibbs 1986). The method relies on the fact that all living cells contain adenosinc 5-triphosphate (ATP). This can be readily assayed by measurement of the light emitted when ATP is reacted with the luciferin -Iuciferase enzyme complex. The amount of light generated is proportional to the ATP concentration which, in turn, is proportional to the number of cclls present. The detection of bacteria in milk by this method is complicated by the presence of free ATP associated with the colloidal calcium phosphate-citrate complex of the casein micelle (Richardson et al. 1980) and also presence of somatic cell ATP, both of which must be removed before meaningful estimates of bacterial ATP can be obtained. Bossuyt (1981) described a method for the estimation of bacterial numbers in milk using a bioluminescent assay system which involved extraction and removal of non-bacterial ATP prior to quantification of bacterial ATP. The test in this form takes c. 45 min to complete. For some time, the dairy industry has required a rapid bacterial counting procedure to monitor tanker milks arriving at processing sites. Ideally, this test should take less than 10 min to provide results before the tanker unloads. Bossuyt (1982) described a modification of his original test which allowed completion within 5-1 0 min: this revised procedure involved the addition of a Ca 2 + sequestrant to aid release of free ATP and an increase in the concentration of the apyrase enzyme required to hydrolyse the non-bacterial ATP. Using this system allowed the detection of milks with a bacterial count greater than 1 x 106 cfu/ml with 90% accuracy (Bossuyt 1982). However, with the introduction of payment schemes to farmers based on the total bacterial count (TIlC) of milks, the hygienic quality of farm bulk tank milks has improved dramatically (Harding 1987). The average TBC for milks produced in the

Cop)"ight Rapid lVlicrobiological Methods for Foods, Beverages and Pharmaceuticals

13

©

/989 by the Sodel)' jiJr Applied Btuuri%!'J' All righls of reproduction in any jimn reserved 0-632-02629-4

14

M. W. GRIFFITHS AND ]. D. PHILLIPS

UK is now about 2 X 104 cful m!. Thus, in order to be efIective, the sensitivity of the bioluminescent bacterial assay system must be improved. Techniques for achieving this have been investigated. These have included modifications to the assay procedure and removing and concentrating bacterial cells from raw milks. If dle bacteria from a relatively large volume of milk can be removed, a greater concentration of bacterial ATP can be presented to the assay system. Also, the quenching effect of milk on the light emission by luciferase will be absent. This should improve the sensitivity of the test. Various methods have been investigated for the removal of bacteria from milk and include electrostatic charge and affinity binding techniques. The use of the bacterial ATP assay for the assessment of the quality of pasteurized milks and for the monitoring of processing plant hygiene will be described.

Materials and Methods Samples Samples of ex-farm bulk tank milks wefe obtained from the Scottish Milk Marketing Board. In some cases where high bacterial counts were required these were incubated at 10°C for 48 h. Retail packs of freshly pasteurized milks were obtained from two creameries in south-west Scotland and transported to the laboratory in insulated containers. Samples were transferred to sterile containers (Sterilin Ltd, Hounslow, Middlesex) and stored in a thermostatic water bath (Grant Instruments (Cambridge) Ltd) at 6°C. Sub-samples were removed at intervals and the psychrotrophic count determined. The shelf-life of the samples was defined as the time in days for the psychrotrophic count in the milk to reach 1 x 107 cful m!. This is the count at which organoleptic changes in IDe product are usually detected (Muir & Phillips 1984). Samples were inoculated on Milk Agar plates (Oxoid) using a Spiral Plate Maker (Don Whitley Scientific Shipley, Yorks). When required, dilutions were made with Maximum Recovery Diluent (Oxoid CM733). Psychrotrophic counts were determined after incubation of the plates at 21°C for 25 h (Griffiths et af. 1980) and mesophile (total bacterial) counts after incubation at 30°C for 3 days.

Bacterial ATP assay in raw milk The reagents used were supplied as a Milk Bacteria Kit (Lumac BV, Schaesberg, The Netherlands). To NRS reagent (Lumac BY) were added

BACTERIAL ATP IN MILK

15

Somase (0.5 units/ml) (Lumac BV) and a chelating agent. Somase is the trademark for the ATP-hydrolysing enzyme, apyrase. In some experiments a different source of apyrase (Sigma Chemical Co. Ltd, Poole, Dorset) was used. The chelating agents studied were EDTA (ethylenediaminetetraacetic acid, disodium salt; 10 mmo1/I), EGTA (ethyleneglycolaminoacetylether tetraacetic acid; 80 mmol/l) and trisodium citrate (40 mmol/l). In one experiment, Triton X-lOO at a final concentration of 0.1 % was added with the citrate. NRS containing apyrase (500 !AI) and chelator were added to milk (500 !AI). Following incubation at room temperature for 5 min, 50 !AI were removed to a Lumacuvette (Lumac BV) and L-NRB (150 !!I; Lumac BV) were added. The cuvette was placed in a Lumac Biocounter M2010 and after 30 s, Lumit PM (100 !AI; luciferin-luciferase, Lumac BV) reconstituted in Lumit buffer (0.025 mo1/1 HEPES, pH 7.75; Lumac BV) according to the manufacturer's instructions, was added manually or using the automatic dispensing system of the instrument. The light emitted after lOs integration was read from the digital display as relative light units (RLU). The use of different luminometers to measure light output was also assessed. The machines used included Lumac Biocounter models M2010 and M2500, an LKB 1250 and a Turner Designs model 20-000. The wearing of disposable gloves was essential throughout the assay procedure to avoid contamination with ATP from the operator's skin. It was also important that reagents were allowed to reach room temperature slowly before use. Using cold reagents led to artificially low ATP levels being recorded. The modified procedure was compared with the method described by Bossuyt (1982). In this case 50 I-tl of milk sample were treated with 100 !AI NRS containing EDTA and apyrase at the concentrations described above. After 5 min continuous shaking at room temperature 150 !AI of L-NRB were added and after a further 15 s, 100 !AI of luciferin-Iuciferase solution were added. The generated light signal was integrated for lOs by the Biocounter and the result expressed in RLU.

Removal of bacteria from milk

By electrostatic charge Charged matrix (1.0 g) was placed in a chromatography column (1.5 X 30 em; Biorad Laboratories Ltd, Walford, Herts) and milk (10 ml) was added with the column outlet closed. In some experiments, nutrient broth cultures of bacteria were substituted for milk. The column was shaken at room temperature for 5 min using a wrist action shaker (Stuart Scientific Ltd, Croydon, Surrey) at maximum speed. After shaking, the column outlet was opened and the milk

16

M. W. GRIFFITHS AND J. D. PHILLIPS

removed. The column contents were washed with 10 ml of freshly distilled water, The compounds studied were magnetite (iron 2,3 oxide) and Celite (diatomaceous silica; Koch Light Ltd, Haverhill, Suffolk). Alternatively, the removal of bacteria from milk using magnetite was carried out in a test-tube. The magnetite was retrieved by placing a magnet at the bottom of the tube and the fluid decanted off. Being magnetic, the magnetite was retained in the tube. For removal of cells using Zeta plus filters (05S grade, 45 mm diameter; Gelman Sciences Ltd, Brackmills, Northarts), the filters were held in a 47-mm Swinnex filter holder (Millipore Ltd) and milk or nutrient broth cultures of bacteria (:0::;50 ml) were passed through the filter using a 60-ml syringe. The filter was washed with 10 ml of freshly distilled water.

By affinity techniques The lectins used in this study were Concanavalin A, Helix pomatia and Triticum vulgaris, immobilized on cross-linked 4% beaded agarose (Sigma Chemical Co. Ltd). Care was exercised in the choice of lectins used as some are extremely hazardous. Lectin (equivalent to I mg) was added to a chromatography column (l.5 x 30 cm) and milk (l0 ml) added with the column outlet closed. The column was shaken for 5 min at room temperature as described previously and subsequently the milk was removed from the column via the column outlet tap. The matrix was washed with 10 ml of freshly distilled water.

Bacterial ATP estimation on concentrated cells The matrix-bound cells obtained by the above procedure were washed with NRS (500 f.tl) with shaking for c. 5 min. The NRS was drained from the column and the matrix washed through with 10 ml of freshly distilled water. With the column outlet closed, L-NRB (200 f.tl) was added and the column contents shaken for 30 s. The L-NRB was collected and a sample (usually 100 f.tl) assayed for ATP with luciferin -luciferase in the manner already described.

Pasteurized milk testing Samples of freshly pasteurized milk (l0 ml) were pre-incubated at 21°C for 25 h in the presence of a sterile solution (0.1 ml) containing crystal violet (2 mg/ml), penicillin (20000 U/mi) and nisin (40000 U/mi) to prevent Grampositive bacterial growth (Griffiths et al. 1984a, b; Phillips et al. 1984). Following the pre-incubation procedure, the milk was assayed for bacterial ATP using the Milk Bacteria Kit (Lumac BY). NRS (500 ftl) was added to


17

milk (500 Ill) with the subsequent addition of Somase (20 Ill; one vial of somase reconstituted with 1 ml Lumit buffer). After incubation at room temperature for 45 min a sample (50 Ill) was removed and added to a Lumacuvette. L-NRB (150 Ill) was added and, after 30 s incubation, Lumit PM (100 Ill) was injected into the cuvette. The light emission following a lO-s integration time was noted.

Variation in pasteurized milk processing A series of experiments was performed using the processing facilities at the Institute which allowed production of pasteurized milks on plant with different degrees of sanitization. Milks were pasteurized using an APV Junior paraflow cream pasteurizer (10-15 gallons/h) (APV Co. Ltd, Derby). Inadequate cleaning was obtained when cold ODC (Reddish Savilles Ltd, Cheadle, Cheshire) (1 % w/v) was circulated through the plant using the product pump at a maximum flow rate of 20 gallons/h. A more effective cleaning regime was achieved by circulation of ODC (1 % w/v) at 75°C and 60 gallons/h for 0.5 h by means of a centrifugal pump (Parsilac type ZMH No.1). Pasteurized milk samples were collected into sterile containers and subjected to the pre-incubation-ATP assay procedure outlined above. Shelf-life of the milks was determined by following bacterial growth in the product at 6°C and lO°C. Results

Removal offree AT? by use ofsuquestrants The effect of EDTA, citrate, EGTA and citrate - Triton X-I 00 additions prior to extraction of nucleotide with NRS are shown in Fig. 1. There was little difference in the results obtained with all four chelating systems. The background readings appeared to be lowest in the presence of citrate but maximum light output was achieved with the assay system containing EDTA. The sensitivity of the assay remained the same regardless of the nature of the sequestrant.

Comparison of assay s.ystems jar use with raw milks There was a significant difference in results achieved using the 5-min bacterial ATP assay as described by Bossuyt (1982) and the modification in which an aliquot was removed before L-NRB extraction (Fig. 2). Substantial scatter of points was obtained using the original method, but, with the modification, a significant curvilinear relation (r = 0.84) was obtained between TBC and the bacterial count by bioluminescence. This may in part be due to the influence

18

M. W. GRIFFITHS AND ]. D. PHILLIPS

4.5

::> ...J

3.5

~ iJ

c:

:1 0

• •

U

a.. f-

« ~

0)

0

....J

log,o total bacterial count Icfu/ml)

I. Effect of chelating agents on the bacterial ATP assay in milk. The non-bacterial ATP was extracted with NRS containing: trisodium citrate, 40 mmol/l (0); EGTA, 80 mmol/l (.); EDTA disodium salt, 10 mmol/l (-); trisodium citrate (40 mmol/J) together with Triton X-100, 0.1% (A).

FIG.

3.2 ::> ...J

~

E :::)

0

3.0 2.8

u

a..

f-

« ~

01 0 ...J

2.6 2.4 4

FIG.

••

~

2.2 a

•• . • • ••• • •• • . .. ,.• • • I~

••

5

•• •

••

•

6

7

:3

3.6

!; +-'

c:

:::)

0

u

3.2

a.. f-

«

o 2.8

0

2.4 4 b

, . .. ....... • ••• • •"

..~

Q)

...J

Log lO total bacterial count (cfu/mll

• •• • • ••

5

6

7

Log 10 total bacterial count (cfu/ml)

2. Comparison of the 5-min bacterial ATP assay procedure: (a) according to Bossuyt

(1982); (b) the modification of that procedure whereby a sample of the reaction mixture was removed following extraction with NRS, for rea.ction with luciferase-luciferin.


19

of shaking during incubation with NRS which was found in some instances to reduce the light output of the reaction mixture. However, there was a good correlation between results obtained when the assay mixture was shaken and unshaken (r = 0.89; n = 55). There was some variation in results obtained using the modified 5-min ATP test on different batches of milk. The main differences were observed for milks with counts below 1 x 105 cful mI. The variation \vas due to changes in background counts noted with these milks. The variation in ATP readings obtained for milks with counts in excess of 1 x 10 6 /ml was acceptable. There was no increase in sensitivity of the assay regardless of the photometer used. Comparison of luciftrin -luciftrase reagents

As well as the luciferin -luciferase supplied as Lumit PM (Lumac BV), an alternative source of the enzyme (Sigma Chemical Co. Ltd) was also studied. The light output in the assay system containing Sigma enzyme was c. 20 times that of the Lumit luciferase. However, there was little or no improvement in the sensitivity of the assay using Sigma enzyme (Fig. 3). The second-degree polynomial regression coefficients were 0.78 and 0.72 for the Sigma and Lumit enzymes, respectively. There was a strong correlation (r = 0.92) between results obtained using the luciferase supplied by Sigma and Lumac. Removal of bacteria from milk

~y

electrostatic interaaion

Bacteria usually carry a net negative charge on their surface, and use can be made of this fact to remove bacteria from suspension (Daniels 1972; Wood 1980). A number of charged matrices have been examined for their ability to remove bacteria from suspension. The efficiency of removal of some bacteria commonly found in milk is shown in Table 1. In general, Gram-positive bacteria were more readily adsorbed on to the charged matrices tested than Gram-negative organisms. However, certain species of Gram-negative bacteria were effectively adsorbed. Bacteria could be removed from milk by electrostatic adsorption with an efficiency of between 25 and 90'10 depending on the milk sample and matrix type (Table 2). Zeta plus 05S grade filters were most effective at removing bacteria from milk. Celite also proved reasonably effective. Both the Zeta plus filters and Celite removed cells by a combination of electrostatic interaction and entrapment. Removal of bacteria from milk ~y affini~y techniques

Lectins are proteins which selectively bind carbohydrates but do not exhibit enzymic activity. They have been shown to react with a wide range of bacteria including species commonly found in milk (Pistole 1981).

20

,•

M.W. GRIFFITHS AND ]. D. PHILLIPS 5.2

• • ••• •

•

4.7

• 4.2 ~

...J

~

C ~

0

()

a..

~

3.7

I-

0

0 0)

0 ...J

0

0

3.2

0

0

0

0 0

2.7

0

2.2

3

4

5

6

7

8

Log 10 total bacterial count (cfu/ml) FIG. 3. Comparison of luciferase-luciferin obtained from Lumac BV (0) and Sigma (e) for estimating the bacterial ATP content of raw milk.

Immobilized lectins were studied to ascertain whether they could bind bacteria present in milk (Table 3). Of the lectins studied, Concanavalin A (Con A) proved the most effective and was able to remove a significant proportion of the milk microflora. The Triticum vulgaris lectin was much less effective than Con A.

ATP assay of bacterial cells concentrated by electrostatic interac#on The use of Celite to entrap bacterial cells in milk followed by a bioluminescent assay may be a way of improving the sensitivity of the 5-min assay procedure. Figure 4 shows a good correlation (r = 0.90) between the RLU output during

21

BACTERIAL ATP IN "vIILK TABLE

1. EjfiCiL7llJ oj'removal oj' baclaia Fom sUSf!L7lSif!l1 by electrostatic dlarge '7,. Removal of bacteria by

Zeta plus filter

Magnetite

Celite

Gram-positive Bad!lus brevis B. ({nus B. drtulal/S B. lichL7lijimllis B. pumilils B. subtilis

11.1 92.6 48.9 57.2 63.6 65.8

0 0 0 0 0 0

17.6 86.5 73.8 13.8 63.6 0

La(/obadlills Cl/seii var. cascii L pla11tarum

86.4 0

0 0

97.0 0

JJ icrococcus roseus Micro(()ccus sp.

20.3 82.9

0 14.1

32.8 n.d.

Staphylococcus sp. Staph. aureus Staph. warnerii

72.5 75.6 92.9

76.4 74.6 70.5

82.9 96.8 94.2

Streptococcus jill'calis Strep. lactis Sirep. lactis var. maltigou's

88.1 85.4 71.8

33.0 18.3 20.5

94.1 95.7 66.0

Gram-negative Achromobacta sp.

50.0

Organism

Ad11etobclcter sp.

84.3

/!eromm/as sp.

89.4

Alcaligem's jaecalis

60.6

Chromobacterium sp.

78.7

0 0

60.3 17.5 20.0

Citrobacta FClmdii

71.3

E11terobacter agglomera11s

79.5

escherichia coli

65.4

Serratia liqucfilciL7ls Serr. marceSCL7lS

66.3 36.2

50.4

Pseudmlll!l1as fluorescL7ls Ps. Fagi Ps. lemOll11ieri Ps. putida Ps. stutzeri

47.8 78.5 67.2 60.8 70.4

()

n.d., not determined.

19.8 0 0 ()

0 0 0 ()

58.9 63.2 93.5 91.3 36.9 82.6 61.1 84.5 49.4 51.3 86.9 22.2 84.4 93.8 n.d.

22

J.

M. W. GRIFFITHS AND

D. PHILLIPS

2. Removal of bacteria from milk by electrostatic charge

TABLE

Average

% removal f\1atrix

Number of e",:periments

of bacteria from milk

Range (%)

6 6

40.0 29.6 70.9

4-93.3 0-65.6 29.4-99.3

Celite Magnetite Zeta plus filter

TABLE

22

3. Removal of bacteria from milk by immobilized leetins % Removal of bacteria from milk

Lectin Concanavalin A Helix pomatia lectin Triticum vulgaris lectin

93.0 45.0 25.0

Values shown are means of three milk samples.

.

4

5-oJ

/

~ ::J

(.)

/

0...

f-

/

/

/

/

/

/

/

0

«

. • .. • .... /,...

C

/

/

• ..

3

••

9

en

/

/

/

/

/

/'/

/

0

~/.

-oJ

//

./

•

2

3

4

5

6

7

8

Log 10 total bacterial count (cfu/mll FIG. 4. Effect of removal of bacterial cells from raw milk by adsorption on Celite before ATP assay.


23

ATP assay of entrapped cells and the TBC of the original raw milk. The relation was linear over the TBC range 3.8 x 103 cfu/rnl to 1.7 x 108 cful ml. A weaker correlation existed when magnetite was used as the adsorbing matrix (r = 0.53; n = 101) for bacterial cells in milks with counts ranging from 1.8 x 10 2 to 2.1 X 108 cfu/ml. Initial results using the NRS/L-NRB e},.1:raction system for nucleotides from bacterial cells entrapped in Zeta plus filters have not been encouraging. This was mainly due to the physical nature of the membrane which made it difficult to retrieve the small volumes of reagents added. The cost of the extractants would make the use of large volumes uneconomical for routine testing. The use of other, cheaper, methods of ATP extraction such as trichloroacetic acid (Lundin & Thore 1975) are under investigation. AT? assay of bacterial cells concentrated by affinity techniques

Following reaction of the bacterial cells in milk with immobilized Con A their numbers were assessed using luciferin-luciferase. There was a weak correlation (r = 0.59) between the bacterial ATP present in the adsorbed cells and the TBC of the original milks. However, the milks examined had bacterial counts in the range 3.3 X 103 to 1.5 X 105 cfu/ml. This was well below the level at which bacterial cells could be detected using the conventional bioluminescence assay technique (Bossuyt 1981, 1982). Thus affinity binding techniques of this kind have an application in improving the sensitivity of the bioluminescence assay for the bacterial content of milks. Shelf-lift test.for pasteurized milk

Previous work described the use of bioluminescence in combination with a pre-incubation procedure for evaluating the shelf-life of pasteurized milk from retail outlets (Phillips & Griffiths 1985). The application of the method to freshly pasteurized milk has also been studied (Griffiths & Phillips 1986). A curvilinear relation existed between the shelf-life of the freshly pasteurized milk and the bacterial ATP content of pre-incubated samples (Fig. 5). The corrc1ation coefficient was 0.64 using the derived equation: shelf-life (days)

=

10.63 - Y(41.67 x log PINe ATP count - 83.79)

The shelf-life of milk samples with an expected shelf-life ofless than 6 days at 6°C could be predicted with an accuracy of between 80 and 90% using this method (Table 4). The reliability of the method fell when used to predict milks of better keeping quality.

24

M. W. GRIFFITHS AND ]. D. PHILLIPS 5

o

5

4

o

...J

0::

c:;,

o u a..

3

I-

1000 or < 1000 respectively). RLU, relative light units. TABLE

5. HygiC11ic quali!:y

Creamery

A Milk ex-pasteurizer Milk ex-homogenized milk tank Milk before filler Cartoned milk B Milk ex-pasteurizer Milk ex-storage tank Cartoned milk

0/ in-line samples Average shelf-life (days)

Average ATP pre-incubation count (log RLU)

7 7 7 8

11.8 12.1 11.7 6.5

2.11 2.04 2.08 2.33

10 10 10

14.7 I\.4 I\.6

2.07 2.17 2.16

No. of samples

RLU, relative light units.

either in terms of shelf-life or pre-incubated ATP count. This was due to the use of a Tetra Brik machine for filling which reduced contamination of product to a minimum. A separate e:\'])eriment to examine the usefulness of the bioluminescence technique coupled with pre-incubation to detect unclean plant gave similar results. Milk was processed at the Institute using inadequately cleaned plant and this resulted in a short shelf-life and high pre-incubated ATP count (Table 6). \\Then improvements were made to the cleaning operations there was a concomitant increase in shelf-life and lowering of pre-incubated ATP count.

26

M. W. GRIFFITHS AND TABl.E

J.

O. PHILLIPS

6. De/eelion of illadequate plant cleaming operatiollS Average shelf-life (days)

Cleansing procedure Inadequate (cold ODe, slow circulation) Adequate (hot ODe, rapid circulation)

Average pre-incubated ATP count (log RLU)

8.8

5.1

3.21

16.7

7.1

2.10

RLU, relative light units.

Discussion Raw milk testing

The estimation of bacterial numbers in milk by bioluminescent ATP assay is complicated by the presence of free ATP associated with the calciumphosphate -citrate complex of the casein micelles (Richardson et al. 1980). Free ATP has been found at levels of 0.13-0.31 ~mol/l (average 0.23 ~mol/l or c. 130 ng/ml) in milk. The total ATP present in milk containing somatic cells (4.5 x 105 /ml) and bacteria (2 x 104 /011) has been given as 210 ng/ml (Theron et al. 1986a). \Vith ATP at 0.4 fg/cfu (D'Eustachio & Johnson 1968) or 1 fgl cfu (Sharpe et al. 1970) for bacteria, the free ATP concentration in milk is equivalent to that contained in c. 1 X 108 bacterial cful ml. Poor correlations of ATP levels with bacterial numbers in milk samples with fewer than 1 x 105 cful ml may result mainly from ATP associated with casein micelles (Packard & l\1arth 1983). Casein micelles may be dissociated by the use of chelating agents (Lin et at. 1971). This results in improved access to the calcium phosphate bound ATP for the apyrase present'in the assay nlixture. Incorporation of EDTA into the NRS - somase reagent resulted in a decrease in background ATP levels (Theron et al. 1986a). The concentration of EDTA present in the reaction mixture is crucial and may account for some variation between results produced when using different batches of reagents (Neaves, personal communication). An improvement in the correlation of bacterial counts, as estimated by plate assay, with ATP determination was found when a small sample was removed after treannent with NRS - EDTA- somase reagent for reaction with L-NRB and luciferase. This was due primarily to decreased quenching of the light output by the milk present in the assay mixture. This quenching is


27

mostly caused by optical interference (Bossuyt 1978; Theron et al. 1986b) but chemical inhibition of the luciferase enzyme may also contribute (Aledort et al. 1966; Denbury & McElroy 1970; Thore 1979). Thus, using the protocols and reagents currently available for ATP assay, only bacterial levels in raw milk greater than about 5 X 105 cfu/ml can be detected. The reason does not depend solely on the sensitivity of the luminometers available. In this study, a luminometer (Lumac Biocounter M2500) was evaluated: this could detect 0.1 pg of ATP and would be expected to detect c. 5 X 103 bacterial cfu/ml by the assay procedure described. ATP can also be derived from somatic cells in milk: this ATP may approach 300 fg/cell (Bossuyt 1978). Theron et al. (l986a) have shown that the NRS-EDTA-somase treatment does not eliminate all non-bacterial ATP. The use of these extractants may also lower the ATP content of bacterial cells in milk. (Theron et al. 1986a). In addition, somatic cells can produce ATP-hydrolysing enzymes when present in large numbers (2 X 106 cells/ml) and these enzymes can hydrolyse extracted ATP (Botha et al. 1986). Pseudomonas fluorescens was also found to synthesize ATP-ase although bacterial counts of 1 X 108 cfulml were required before enzyme activity was detected (Botha et al. 1986). In order to improve the sensitivity of the bacterial ATP assay for use with milk, either the methods for destroying non-bacterial ATP must be improved or techniques to remove bacteria from milk before assay must be sought. Methods have been described for removal of bacteria from suspension. Most micro-organisms are negatively charged at pH values greater than 5.0 and will adsorb to positively charged matrices (Wood & Gibbs 1982). Many ion exchange matrices are available for use (Daniels 1972; Wood 1980) and methods for the removal of bacteria from water by electro-adsorption have been described (Oren et al. 1983). Magnetite, the magnetic oxide of iron, has been used eJl.'})erimentally to adsorb bacteria (De Latour & Kolm 1976) and has the advantage of being easily removed by magnetism. MacRae & Evans (1983) showed that more than 95% of cells of Escherichia coli, Streptococcus Iaecalis and Pseudomonas stutzeri could be adsorbed by magnetite at pH 6.0. The use of clectropositively charged filters for removal of bacteria from foods, including milk, has also been described (Kroll 1985). In the present study, magnetite, Celite (a commercially available diatomaceous silica) and Zeta plus filters were investigated for their ability to adsorb bacteria. For magnetite, 12 of the 31 strains of bacteria studied could be removed from suspension with efficiencies ranging from about 14 to 76% removal. The Gram-positive bacteria seemed to be adsorbed by magnetite more effectively than were the Gram-negative organisms studied. Streptococcus faecalis was adsorbed but none of the Bacillus spp. tested could be removed from suspension by magnetite. None of the Pseudomonas spp. tested, nor E.

28

M. W. GRIFFITHS AND

J.

D. PHILLIPS

coli, was adsorbed by magnetite and this makes its usefulness for removing bacteria from milk questionable. Such was the case, with the magnetite effectively removing bacteria from milk in only about 33 (~o of samples. This may have reflected the age of the milk as the microflora of freshly drawn raw milk consists mainly of Gram-positive bacteria whilst Gram-negative bacteria predominate in stored milks. Celite and Zeta plus filters were equally effective in removing bacterial cells from suspension in nutrient broth. There appeared to be little difference in the selectivity of the binding of Gram-positive or Gram-negative organisms. When the systems were applied to milk samples Zeta plus filters were the more effective in adsorbing bacterial cells. This was probably due to the physical nature of the filter which led to the retention of more bacteria by entrapment. The physical nature of the Zeta plus filter made it less amenable to use in conjunction with ATP methodology. Reasonable results, however, were obtained by using the bacterial ATP assay on cells removed from milk by Celite. Results indicate that the sensitivity of the method can be increased by at least one log cycle when this bacterial concentration step is incorporated in the assay system. Lecrins bind with polysaccharide residues of the cell wall of Gram-positive bacteria and the lipopolysaccharides of the cell envelope of Gram-negative bacteria. Among the bacteria that react ffith leetins are some commonly found in milk, such as streptococci (Hamada et ai. 1977) and members of the Enterobacteriaceae (Le Minor et al. 1973). Lectins with different specificities can be obtained and this present study investigated the use of Concanavalin A which is specific for (Y-D-mannose and (Y-D-glueose residues (Reeke et al. 1974), Helix pomatia lectin specific for N-acetyl-D-galactosamine (Hammarstrom el al. 1972b) and Triticum vulgaris lectin which reacts with sialic acid and N cetyl-D-glucosamine (Allen et al. 1973; Nagata & Burger 1974). Gram-positive bacteria including Bacilmus subtilis, Lactobacillus plantarum and Streptococcus faecalis were agglutinated by Concanavalin A because their cell walls contained teichoic acids substituted by (X- D-glucopyranosyl residues (Doyle & Birdsell 1972; Archibald & Coapes 1971; Bauer et al. 1974) whilst both 'Con A and llelix pomatia lectin have been shown to interact with the lipopolysaccharides of Gram-negative bacteria (Doyle et al. 1968; Hammarstrom & Kabat 1971; Hammarstrom et al. 1972a; Pistole 1981). Immobilized lecrins could, in fact, be used to remove bacteria from milk. Con A was the most effective in removing bacteria from raw milk with the Triticum vulgaris lectin the least effective. Subsequent assay of bacteria retnoved by inlmobilized Con A using the ATP technique gave a significant correlation with total bacterial counts in milks containing less than 1 x 105 cful ml.


29

Pasteurized milk testing

It is well documented that the main factor governing the shelf-life ofpasteurized products is the growth of psychrotrophic, Gram-negative bacteria introduced as the result of post-heat-treatment contamination (Griffiths et al. 1985). Although these organisms grow well at refrigeration temperatures their optimum temperature for growth is about 20- 22°C (Lawton & Nelson 1954). Indeed, a good correlation has been obtained between the growth of psychrotrophs in milk at 6°C for 14 days and at 21°C for 25 h (Oliveria & Parmelee 1976). Thus, they may be selectively grown at a temperature of 21°C for 25 h with incorporation of an inhibitor system which prevents Gram-positive bacterial growth. Of the inhibitors studied, crystal violet-penicillin-nisin had least inhibitory effect on Gram-negative growth whilst still preventing growth of Gram-positive bacteria (Phillips & Griffiths 1986). The bacteria that grew under these pre-incubation conditions could be estimated using the ATP assay. This provided a good indication of the quality of pasteurized products including milk (Phillips & Griffiths 1985; Griffiths & Phillips 1986) and cream (Griffiths et al. 1984a; Phillips & Griffiths 1985; Griffiths & Phillips 1986) within 26 h of production. A similar test for predicting the keeping quality of pasteurized milk has been described (Wacs & Bossuyt 1981, 1982). Pre-incubation conditions of 30°C for 24 h in the presence of benzalkoncrystal violet to inhibit Gram-positive bacterial growth were used prior to bacterial ATP estimation. Benzalkon was, however, more inhibitory towards Gram-negative bacteria than the crystal violet-penicillin-nisin mixture and many psychrotrophs did not grow well at 30°C (Langeveld et al. 1976). The use of the pre-incubation test in conjunction with in-line sampling at the processing site allowed the monitoring of the efficacy of plant cleaning operations. The source of contamination could be identified within a useful time frame and remedial measures taken. References ALWORT, L.M.R., WEW, R. & TROUP, S.B. 1966. Ionic effects of firelly bioluminescence assay of RBC ATl'. Analytical Biochemistry 17, 268-277. Al.LEN, A.K., NEUBERGER, A. & SHARON, N. 1973. The purification, composition and specificity of wheatgerm agglutinin. Biod1l'mimt ]ounJal131, 155~ 162. ARCHII.lALD, A.R. & COAPES, I I.E. 1971. The interaction of concanavalin A with teichoic acids and bacterial walls. Biodlemiml]ounJal 123, 665-667. EMlER, 11., E\RR, D.R. & IIORISBERGER, M. 1974. Ultrastructural localization of cell wall teichoic acids in Streptococcus jaew/z'· by mcans of concanavalin A. Archives o/Miaobiolo!..'J! 97, 17-26. BOSSUYT, R. 1978. Usefulness of an ATl' assay technique in evaluating the somatic cell content of milk. Milduvissetlsc!llIji 33, 11-13. BOSSUYT, R. 1981. Determination of bacteriological quality of raw milk by an ATP assay

30

M. W. GRIFFITHS AND

J. D. PHILLIPS

technique. Mi/cllWisSt'l1schafi 36, 257- 260. BOSSUYT, R. 1982. AS-minute ATP platform test for judging the bacteriological quality of raw milk. Netherlands Milk and Dairy Journal 36, 355 - 364. BOTHA, W.e., LUCK, H. & JOOST£. P.J. 1986. Detennination of bacterial ATP in milk - the influence of adenosine triphosphate-hydrolysing enzymes from somatic cells and Pseudomonas jiuoresce1ls. Journal of Food Protectioll 49, S22-82S. DANIELS, S.L. 1972. Adsorption of microorganisms onto solid surfaces. Review. Developments in Industrial Microbiolog)1 13, 211- 253. , DE LATOUR, C. & KOLM, H.B. 1976. High-gradient magnetic separation a water treatment alternative. Journal of the American Water Workers Association 68, 325 - 327. DENBURY, J.L. & lVlcELROY, \V.D. 1970. Anion inhibition of firefly luciferase. Archives of Biochemistry aud Bioph,ysics 141, 668-675. D'EUSTACHIO, A.]. & JOHNSON, n.R. 1968. Adenosine triphosphate content of bacteria. Federalio11

Proceedings of the Federation ofAmerican Soaeties of Expen'mC11tal Biology 1968, p. 761. DOYl.E, R]. & HIRDSE!.L, D.C. 1972. Interaction of concanavalin A with the cell wall of Bacillus

subtilis. Journal of Bacteriology 109, 652-658. DOYLE, R.j., WOODSIlJE, E.E. & FISHEL, C.\V. 1968. Protein and polyelectrolyte interactions:

The concanavalin A precipitin reaction with polyelectrolytes and polysaccharide derivatives. Biochemical}ourna/106, 35-40. GRIFFITHS, M.W. & PHILLlI'S, J.D. 1986. The application of the pre-incubation test in commercial dairies. Australian Journal of Dairy Technology 41, 71-79. GRIFFITHS, Nl.W., PHILLIPS, J.D. & Mum., D.O. 1980. Rapid plate counting techniques fc)r the enumeration of psychrotrophic bacteria in pasteurized double cream. Joumal of the Societ;y oj Dairy Tedmology 33, 8-10. GRIFFITHS, M.\V., P~IILl.IPS, J.D. & NluIR, D.O. 1984a. 1\1ethods for rapid detection of postpasteurization contamination in cream. Joumal oj tilt' Society of Dairy Teclmology 37, 22 - 26. GRIFFITHS, M."V., PHILLIPS, J.D. & lVlulR, D.O. 1984b. Rapid detection of post-pasteurization contamination. Hannah Research Institute, Bulletin No. 10. GRIFFITHS, M.\\I., PHILLIPS, J.D. & MUIR, D.O. 1985. Post-pasteurization contamination - the major cause of failure of fresh dairy products. Hannah Research 1984, 77-87. IIAMAOA, 5., GILL, K. & SL:\DE) H.D. 1977. Binding of lectins to Streptococcus mutans cells and type-specific polysaccharides and effect on adherence. Injec/ion and Immunio' 18, 708 - 716. lIAMMARSTR()M, S. & KAUtH, E.A. 1971. Studies on specificity and binding properties of the blood group A reacth;e hemagglutinin from Helix pomatia. Biochemistry 10, 1684-1692. lIAMMARSTROM, S., LINDBERG, A.A. & ROI-:lERTSSON. £.S. 1972a. Precipitation of lipopolysaccharides from rough mutants of Salmonella typhimurium by an A-hemagglutinin from Helix

pomatia. European ]oi/mal

(~r Biochemistry

25, 274- 282.

IIAJ\\MARSTROM, S., "VESTHOO, A. & BJORK, I. 1972b. Subunit structure of Hehr poma/ia A hemagglutinin. Scandinavian Journal oflmmunolof!J' I, 295-309. I lARDING, F. 1987. The impact of central testing on milk quality Dairy Industries Intematiollal 52(1), 17-19. KROLL, R.G. 1985. Elcctropositivcly charged filters for the concentration of bacteria from foods. Food A1icrobiology 2, 183-186. LA.NGEVELD, L.P.M., CUPERUS, F.) VAN BREEMEN, P. & DIjKERS, J. 1976. A rapid method for the detection of post-pasteurization contamination in HTST pasteurized milk. NetherJilfuls Milk alld Dairy .Joumal 30, 157-173. L'\ \\TON, \V.e. & NEI.SON, F.E. 1954. The effect of storage temperatures on the gro\\.1h of psychrophilic organisms in sterile and laboratory pasteurized skim milks. ]oumal of Dairy S'o'e11ce 37, 1164-1172.


31

LE MINOR, L., TOURNIER, P. & CHALON, A.M. 1973. The agglutination by concanavalin A of certain Gram negative bacilli: A study into the relationship with the somatic antigens of Salmonella serotypes. Annals ofivIicrohiology, Paris 124A, 467-476. LIN, S.l I.e., DEWAN, R.K., BLOOMFIELD, V.A. & MORR, CV. 1971. Inelastic light scattering studies of the size distribution of bovine milk casein micelles. Biochemistry 10, 4788-4793. LCNDlN, A. & THORE, A. 1975. Comparison of methods for extraction of bacterial adenine nucleotides determined by firefly assay. Applied Microhiology 30, 713 -721. MM:RAE, I.C & EVANS, S.K. 1983. Factors influencing the adsorption of bacteria to magnetite in water and wastewater. Water Research 17, 271-277. j\,IUIR, D.o. & PHILLIPS, ].D. 1984. Prediction of the shelf-life of raw milk during refrigerated storage. Mildlll'issetlsdza{i 39, 7 -11. N.AGAT.A, Y. & BURGER, lv1.M. 1974. Wheat geml agglutinin. Molecular characteristics and specificity fi.Jr sugar binding. ]oumal ofBiologiwl Chemistry 249, 3116-3122. Ol.lVERLA, j.S. & PARA1EI.EE, CE. 1976. Rapid enumeration of psychrotrophic bacteria in raw and pasteurized milk. ]oumal oj'Milk (lIId le/JOtI Tl'Chnolob.1' 39,269-272. ORFN, Y., Tom.As, II. & SOFFER, A. 1983. Removal of bacteria from water by e1ectroadsorption on porous carbon electrodes. Bioell'Ctrodle1nistry and Bioenergetics 11, 347-351. PAcKMm, R.A. & MARTH, E.II. 1983. JVlodifying the ATP bacterial test for use with raw milk. }oumal of Dairy Science 66, Supp!. I, 67. PHILLIPS, ].D. & GRIFFITHS, ,\tW. 1985. Bioluminescence and impedimetric methods for assessing shelt~life of pasteurized milk and cream. Food Microbiology 2, 39- 51. PHII.I.IPS, ].D. & GRIFFITHS, M.W. 1986. Estimation of Gram-negative bacteria in milk: A comparison of inhibitor systems for preventing Gram-positive bacterial growth. ]oumal of Applied Bal'lerioloKJ! 60, 491- 500. PHlI.I.lPS, ].D., GRIFFITHS, M.W. & MUIR, D.o. 1984. Preincubation test to rapidly identifY post-pasteurization contamination in milk and single cream. ]oumal of Food Protection 47, 391-393. PISTOLE, T.G. 1981. Interaction of bacteria and fungi with lectins and lectin-like substances. Annual Rf'vjews oj'lvlicrobiology 35, 85 -112. RI.I·:KE, G.N., BECKER, ].W., CUNNINt;IHM, B.A., GUNTHER, G.R., WANG, ].L. & EDELMAN, GAl. 1974. Relations between thc structure and activities of concanavalin A. Annals of the New York Aau/i'1I1y o{Scie1lce 234,369-382. RICILARDSON, T., McG:iNN, T.CA. & KEARNEY, R.D. 1980. Levels and locations of adenosine 5-triphosphate in bovine milk. Journal o{Dajry ResClmh 47, 91-96. SCHRODER, MJ.A. & BLAND, M.A. 1983. Post-pasteurization contamination and shelf-life of IITST-pasteurized milk when filled in Liqui-Pak conventional or Model 820A cartoning machine. ]ounwl oj'the Society o{Dairy Tedll/oloKJ! 36, 43-49. SHi\RI'E, A.N., WOODROW, M.N. & JACKSON, A.K. 1970. Adenosine triphosphate (ATP) levels in foods contaminated by bacteria. ]ourl/al o{Applied 8a(faiolo[.,'JI 33, 758-767. STi\NN.ARD, C]. & GmBs, P.A. 1986. Rapid microbiology: Applications of bioluminescence in the food industry - a review. Jouml/l of Biolu1l1i1lescfllu a1ld Che1l1ilumi1lcscetlee 1, 3-10. THERON, D.P., PRIOR, B.A. & LHEGAN, P.M. 1986a. Determination of bacterial levels in raw milk: Selecthity of non-bacterial ATP hydrolysis. Joumal of Food Protaljlm 49, 4-7. THERON, D.P., PRIOR, B.A. & L.ATEGAN, P..VI. 1986b. Sensitivity and precision of bioluminescent techniques ft)r enumeration of bacteria in skim milk. ]oul71al of Food ProtatiOlI 49, 8-11. TIIORE, A. 1979. Technical aspects of the hioluminescent fireflv luciferase assay of ATP. Scimce 71lOls 26, 30-34. ' \V·\\.s, G. & Bossun, R. 1981. :\ rapid method to detect postcontamination in pasteurized milk. Nli/chwissl'llsdwji 36, 548-552.

32

M. W. GRII"FITHS AND ]. D. PHILLIPS

G. & BOSSUYT, R. 1982. Usefulness of the Benzalkon-crystal violet-ATP method for predicting the keeping quality of pasteurized milk. Journal of Food Protection 45, 928- 931. WOOD, ].M. 1980. The imeraction of microorganisms 'With ion exchange resins. In Miaobial Adhesion to Surfaces, ed. Berkeley, R.C.W., Lynch, lM., Melling, l, Rutter, P.R. & Vincent, B. pp. 163-185. Chichester: Ellis Horwood Ltd. WOOD, J.M. & Gums, P.A. 1982. New developments in the rapid estimation of microbial populations in foods. In Developmetlts in Food Microbiology, ed. Davies, R. pp. 183-214. London: Applied Science Publishers. WAE5,

DEFT: Recent Developments for Foods and Beverages G. L. PETTIPHER 1 , R. G. KROLL 2 , L.J. FARR 3 AND R. P. BETTS 4 I Cadbury Schweppes PLC, Group Research, The Lord Zuckennan Research Centre, The University ofReading, Reading RG6 2LA, UK; 2 Department ofMicrobiology, AFRC Institute ofFood Research, Reading Laboratory, Shinjield, Reading RG2 9AT, UK; 3 Foss Electric (UK) Ltd, The Chantry, Bishopsthorpe, York Y02 lQf, UK; and 4 Campden Food 0 Drink Research Association, Chipping Campden, Gloucestershire GL55 6LD, UK General Principles The direct epifluorescent filter technique (DEFT) was originally developed for counting bacteria in raw milk (Pettipher et al. 1980; Pettipher 1983). This method, which uses membrane filtration and epifluorescence microscopy, takes less than 30 min to complete and does not suffer from many of the disadvantages of other microscopical methods. Pre-treatment of the milk with lyses somatic cells and makes fat a proteolytic enzyme and surfactant at globules sufficiently fluid to enable 2 ml of milk to be filtered easily. Filtration concentrates and distributes the bacteria in a manner that makes counting easier, and the technique is about 100 times more sensitive than the Breed smear. The use of a fluorescent stain and an epifluorescence microscope produces well-stained micro-organisms that are easily distinguishable from the small amounts of fluorescent debris. A polycarbonate membrane filter is used, the flat surface of which is better suited to microscopy than the uneven surface of cellulose acetate membranes. Micro-organisms on DEFT slides can be counted by automated methods, thereby considerably reducing operator fatigue. DEFT can be used for the rapid enumeration of micro-organisms in urines, pharmaceutical products, beverages, and, with additional sample pre-treatment, a variety of foods, in addition to milk.

sooe

Apparatus, Reagents and Methodology Details of the apparatus, reagents and methods are given in the chapter by Shaw and Farr (this volume). Cflpj'righl © 1989 It), Ih,' SO';,1J' jrJr Applied BarlennlnK)' All rights ({ rcprodlU:lion in

Rapid Microbiological \lethods for Foods, !leverages and Pharmaceuticals

an)1

fium rl'Scrved

0-632-02629-4

33

34

G. L. PETTIPIIER ET AL.

In addition to the count/ ml of sample, a presumptive identification of the types of micro-organisms present on DEFT slides may be made. Generally the following types of bacteria are discernible by morphology; streptococci, micrococci/staphylococci, and some bacilli and coryneforms. Fungi, yeasts and spores can also be distinguished. A tentative identification of the types of micro-organisms present in the sample may be useful in locating their source, e.g. a high bacterial count in refrigerated farm milk, where most organisms are streptococci, is usually indicative of mastitis in the herd.

Automated counting The manual DEFT count is suitable only for 30-40 samples per day per operator as prolonged use of the microscope results in operator fatigue. Micro-organisms on DEFT slides can be counted automatically using the 40-10 image analyser (Foss UK Ltd, York). A closed circuit TV camera (scanner) attached to the microscope camera port detects the microscopic image and the resulting video signal is analysed to obtain quantitative data. With the image analyser an operator can count about 50 slides per hour. The image analyser can be controlled by a microcomputer which processes the generated data. The relevant information, e.g. sample code, count/ml and grade, is printed out and can be recorded simultaneously on to a magnetic disc for storage purposes and report writing. The semi-automated count of bacteria on DEFT slides agrees well with the corresponding visual DEFT count with a correlation of 0.94 (Pettipher & Rodrigues 1982a). The counting system could be fully automated by including a cassene loading system for slides and focusing under microcomputer control. The technology for these modifications is already available. Automation of DEFT preparations (which would involve sampling, pre-treatment, filtration, staining, and rinsing stages) combined with automated counting, would provide a system capable of testing hundreds of samples each day.

Cost of equipment and disposable items The cost of equipping a laboratory for the DEFT is approximately £5000 (in 1988). The major capital cost is the epifluoresccnce microscope which retails for about £4000. The other e},.'Pensive items are the filter manifold and water bath. The additional cost of the instrumentation for semi-automated counting is about £7800. This includes the TV image analyser system and the microcomputer which, although not essential, greatly facilitates data logging, record keeping and report writing.

DEl'T: RECENT DEVELOPMENTS

35

The current cost of disposable items used in the DEFT is about £0.55 per test, of which the Nuclepore membrane filter is the most expensive single item. Alternative membranes are currently being evaluated. Reproducibility of the DEFT count seems to be similar to that of the plate colony count. The DEFT count correlates well with the plate colony count for farm, tanker and silo milks, with an overall correlation coefficient (r) of 0.91. The DEFT count is both sufficiently rapid for monitoring tanker and silo milk and sensitive enough for grading farm milk on the basis of bacteriological quality.

Foods The DEFT method was found to be unsuitable for counting micro-organisms in suspensions of food prepared in a stomacher because of the presence of food debris. Pre-filtration of the suspension through nylon filters (mesh size 5 [tm) removed most of the food debris and only slightly reduced the recovery of micro-organisms, as determined by the plate colony count (Pettipher & Rodrigues 1982b). For milk, 0.6 [Lm pore size polycarbonate membrane filters are used in the DEFT but for some 'cleaner' liquids, such as certain pre-filtered food suspensions, beverages and water, a smaller pore size (0.4 or 0.2 [Lffi). membrane filter is recommended. Depending upon the nature of the food, 4-15 ml of food suspensions (10 g food plus 90 ml diluent) could be filtered through a single nylon prefilter and 3 - 10 ml of each pre-filtered suspension could be filtered in the DEFT apparatus. A sample size of 2 ml was used routinely in the DEFT, together with a microscope factor of 57000. For fresh meat and fish, the pre-filtered DEFT count agreed well with the plate colony count over the range 10 4 -10 10/ g, with a correlation coefficient of 0.91. For counts on frozen meat and frozen fish, the pre-filtered DEFT count also agreed well with the plate colony count over the range 5 X 104 - 5 X 1Q7/ g. Of the other foods tested, there was good agreement between the prefiltered DEFT count and plate colony count for cooked meats, cream doughnuts and spices (Pettipher & Rodrigues 1982b). Beverages Methods based on the same principles as the DEFT are used to estimate viable and non-viable yeasts and bacteria in wines and on winery equipment surfaces (Cootes & Johnson 1980). The method for wine takes only 15 min to complete and has a sensitivity of less than 1 micro-organism/m\.

36

G. L. PETTIPHER ET AL.

Selective Pre-incubation for the Detection of Low Levels of Spoilage Bacteria by the DEFT The microbiology of pasteurized milk and cream is complex. During the summer months spoilage due to the growth of psychrotrophic spore-forming bacteria can be important (McKinnon & Pettipher 1983; Phillips & Griffiths 1986a). It is now generally held that low-level post-pasteurization contamination by Gram-negative rods is the most significant factor in reducing the keeping quality (KQ) of these products (Phillips et al. 1981; Schroder et al. 1982). The counts of the initial microbial flora of pasteurized milk and cream are thus of little value for predicting their KQ (Bishop et al. 1984), because the vast majority of viable bacteria present initially after pasteurization are thermoduric non-psychrotrophs which do not contribute to the eventual spoilage of these products. The definitive method of assessing the psychrotrophic contamination is by incubating plates at 7°C for 10 days. Such retrospective results are obviously of little practical value and more rapid estimates of this contamination would be most beneficial for monitoring production standard and product quality and to identifY sources of contamination. The numbers of these Gramnegative psychrotrophic contaminants are well below the current detection limits of the truly rapid methods of estimating microbial numbers, e.g. the assay of ATP or the DEFT. One approach has been to use a pre-incubation stage to increase the concentration of these organisms to detectable levels, the

premise being that the final concentration of organisms after pre-incubation is proportional to the original number of organisms present. The DEFT and ATP methods cannot, at present, easily distinguish between the types of organisms in the sample (but see Rodrigues & Kroll 1985), so the preincubation stage needs to be selective against the Gram-positive organisms (generally non-psychrotrophic). The outer membrane of Gram-negative bacteria has a restricted permeability to many compounds, including several inhibitors, which renders these bacteria more resistant to some inhibitors. This, together with different selective modes of action on Gram-positive bacteria, has been exploited to achieve selective pre-incubation stages. The procedure, in practice, is simple. Samples for testing are incubated at a controlled temperature for a set time with a mixture of inhibitors and the normal procedures of the assay (ATP or DEFT) are then performed on the samples. Wacs & Bossuyt (1982) successfully used a mixture of benzalkonium chloride (0.03% w/v) and crystal violet (0.002% w/v) as the selective inhibitors in the pre-incubation stage before assay of ATP. Rodrigues & Pettipher (1984) also used this combination and demonstrated that the DEFT could be used successfully to predict, within 24 h, the KQ of >80% of samples of

37

DEFT: RECENT DEVELOPMENTS T,\ULE I. CllIssijimlioll o/pasleurized milk samples 1111 Ihe basis 0/ KQ al 5°C alld 1JOC kv Ihe DEFT wUIII folllllvillg pre-illcubalioll al ](J'C iiJr 18 h

Storage temperature eC)

DEFT count/ml after pre-incubation

Corresponding KQ (days)

o/.. Correctly

10 4

10.6

106

7.6

J(f

88 80 80

lOx

6.0 4.5

95 91

10' 10;

3.8 2.85

82 88

106

1.9 0.95

92 90

10;

11°

classified

)07

9.1

From Rodrit,'ues & Pettipher (1984).

pasteurized milk stored under conditions of satisfactory (5°C) or unsatisfactory (10°C) refrigeration (Table 1). Griffiths et at. (1984) used a different mixture of inhibitors (crystal violet, 2 mg/ml; nisin, 40000 units/ml; penicillin, 20000 units/ml) and compared the ability of the ATP assay and the DEFT to detect post-pasteurization growth in pasteurized cream stored at 6°C. They found that both methods correctly identified samples as unacceptable or acceptable (DEFT, 91%; ATP, 94%). Kroll & Rodrigues (1986a) also studied the ability of tlle DEFT for predicting the KQ of pasteurized cream stored at 5° or lO°e. They found that the DEFT classified >81 % of the samples correctly, but only when undiluted cream was filtered (using the 20-ml 0.1 'Yo Triton X100 modification of the Pettipher & Rodrigues (1981) method). In this work the pre-incubation stage included only benzalkonium chloride at 0.05% (w/v) as inhibitor. All the pre-incubation systems described appear to function correctly and the DEFT (and assay of ATP) can be used to quantifY the low-level contamination present in these products. The ideal selective system that inhibits all the Gram-positive bacteria without any inhibitory effect on the Gram-negative bacteria probably docs not exist. Phillips & Griffiths (1986b) have recently made a useful study of seven possible inhibition systems that could be used for this purpose. They showed that the two mixtures that have already been used for this purpose appear to be the most effective at inhibiting Grampositive organisms and not inhibiting Gram-negative organisms in skimmed milk. A variety of pre-incubation temperatures have been used. Although the

38

G. L. PETTIPHER £1' AL.

Gram-negative post-pasteurization contaminants are psychrotrophic they are not psychrophilic and in our experience will grow well at least up to 30°C. as pre-incubation Rodrigues & Pettipher (1984) used 15°, 20° and 300 temperatures for 18 h and they found that the DEFT gave the best prediction of the KQ of pasteurized milk after pre-incubation at 30o e. Griffiths et al. (1984) pre-incubated samples at 21 DC for 25 h and found that the assay of ATP and DEFT gave satisfactory results at this temperature. Kroll & Rodrigues (1986a) found that the DEFT gave a good prediction of the KQ of pasteurized cream after pre-incubation of samples at 20 0 or 25°C for 18 h. It is not yet clear which time/temperature combination is optimal. The temperature must be chosen to give the maximum final yields to bring the cell density above the detection limits of the chosen method (especially with samples with low levels of contamination) while not being so high as to inhibit the growth of the psychrotrophs. Further work is needed to clarify the position, but when Kroll & Rodrigues (1986a,b) used the cytochrome c oxidase test for this purpose they encountered many false negatives with pre-incubation tenlperatures above 20o e. It should be emphasized, however, that this was not due to lack of growth in these samples so this should not affect assay by ATP or DEFT. A cautionary note must be made about the estimates of KQ in these studies. The end of KQ has been taken to be when the cell density reaches the arbitary value of 5 x 106 bacterialml (g). In practice, some milk samples can have higher cell densities than this although the product may not have spoiled (Rodrigues & Pettipher 1984). In principle, this approach could be used to detect low numbers of other types of spoilage or pathogenic micro-organisms in foods.

e

Osmotolerant Yeasts in Confectionery Products The presence of even low numbers of osnlOtolerant yeasts in creme fondant, a confectionery ingredient, is undesirable as growth during storage and transportation may cause spoilage of the products. In view of the potential cost of spoilage it is essential that rapid, sensitive and reliable methods are used to assess the quality of this ingredient. Results obtained using naturally spoiled creme fondant showed that, in the DEFT, 2-~m pore size Nucleporc nlembrane filters retained similar numbers of yeasts as 1.O-~m or O.6-~m pore size filters. For routine usc, 5 g of fondant and pre-treatment with trypsin and Triton X-IOO was necessary for efficient filtration. The DEFT could be used to detect high numbers (>200/g) of osmotolerant yeasts immediately or low numbers after pre-incubation in a medium containing 10% (\v/v) sucrose. The detection limits for osmophilic yeasts in the pre-incubated DEFT count, as extrapolated from the initial plate count, were c. 1/g after 25 hand 1/(10 g) after 49 h (Pettipher 1987). A pre-incubated plate count takes 4-5

DEFT: RECENT DEVELOPMENTS

39

days to give a result. The DEFT may also be of considerable value in plant hygiene surveys to pinpoint areas of yeast proliferation. Selective Enumeration of Bacteria by the DEFT At present, a major limitation to the truly rapid methods of microbial detection and enumeration (i.e. the DEFT and assay of ATP) is that they cannot specifically identify and quantitY or even give an indication of the types of organisms present in a sample, whereas this can be achieved with methods relying on changes in the impedance/conductance of the medium, although the results are obtained more slowly. If such advances could be made to the truly rapid methods, this would greatly increase the application and acceptance of these methods in microbiology. An advantage of the DEFT is that in some cases tentative identifications of the types of organisms present can be made in a sample of raw milk. With bacteria, however, identification on the basis of morphology alone is not valid. One modification has been made which allows the DEFT to give a total count of the Gram-negative bacteria present (Rodrigues & Kroll 1985). This method uses the normal Gram staining procedure of bacteria on the polycarbonate membrane, except that 95% ethanol is used as tile decolorizing agent and acridine orange (125 I-tg/m!) is used as the counter-stain. In a trial of this modification the active Gram-negative bacteria, but not the Gram-positive bacteria or inactive Gram-negative bacteria, fluoresced orange and the method was shown to give a good relationship between Gram-negative DEFT count and Gram-negative plate count (crystal violet agar) with samples of raw milk (r = 0.94). Occasional problems were encountered with clumps of streptococci which tluoresced orange, but these were distinguishable from the Gramnegative organisms by transmitted visible light illumination of the sample. Although this modification introduces a degree of selective counting of bacteria in the DEFT, there are many applications where a better differentiation between the types of bacteria present needs to be made. One approach to improve this aspect of the method has been investigated recently (Rodrigues & Kroll 1988): this is to eA'})loit the selectivity of commercially available microbiological media to grow microcolonies of the selected bacterial type. In essence, samples of meat, fish and vegetables are homogenized in a stomacher in 0.1 'Yo peptone water and the samples pre-filtered through 5.0-I-tm nylon mesh to remove the large particles of food in the normal way (Pettipher & Rodrigues 1982b). The samples are not pre-treated with enzyme or detergent as in the normal DEFT, but are immediately filtered under vacuum through 0.6-~lm Nuclepore filters held in filter towers. In some cases this reduces the volume of sample that can be filtered (typically 1-5 ml for meat, 2-10 ml for vegetables) and slightly reduces the sensitivity of the method. It

40

G. L. PETTIPIIER ET AI.

is essential, however, that the detergent/enzyme pre-treatment stage of the normal DEFT is omitted, to avoid damage to the bacteria. The filters are then removed from the filter towers and placed on the surface of the required agar medium in Petri dishes. The Petri dishes are inverted and incubated at 30°C for different periods of time. In our experience 3 and 5 h gives satisfactory microcolony formation by Gram-negative and Gram-positive bacteria, respectively. After incubation, the filters are replaced in the filter towers and stained with acridine orange by the normal DEFT procedure (Rodrigues & Kroll 1985). The mcmbrancs arc then mounted on slides with non-fluorescent immersion oil and examined by epifluorescent microscopy using a x 50 or x 100 objective lens. The microcolonies which have formed stain bright orange and are easily counted. Individual cells or small clumps «4 cells) are commonly observed. These are thought to be non-viable cells or, more likely, cells which have broken off from the microcolonies; in any event, these cells are not included in the microcolony count. In initial experiments microcolony formation by pure cultures was studied on filters incubated on nutrient agar and compared with the total viable count on nutrient agar. These experiments demonstrated that the microcolony counts in DEFT were very closely related to the number of colony-forming units in the standard plate count. A total of 26 Gram-negative and 17 Gram-positive cultures was studied and it was interesting to find that the morphology of the microcolonies was sometimes distinctive to that genus, often with spectacular formations on the membrane. These cultures were then tested on eight

selective media for their ability to form microcolonies of the correct type and in accordance with the selectivity of the medium. Only four of the media appeared to exhibit the correct properties and these were selected for further study. The natural and introduced flora of a variety of fresh and frozen fish, meat and vegetables was then examined for its ability to form microcolonies on these media (Table 2). The microcolony counts related very well to the normal plate counts but the results were available in only a few hours. The selective media for coliforms and pseudomonads gave the best results; the medium for streptococci, staphylococci and micrococci was less satisfactory, having lower correlation coefficients and higher intercept values (Table 2). This medium was also less satisfactory in that the staphylococci and micrococci microcolonies could not be properly differentiated. Despite these extremely encouraging results, the method is subject to the following restrictions: 1 The maximum sensitivity of the method with real food samples is in the range 103 - 104 bacteria/g. As coliforms or pseudomonads often form only a small proportion of the total viable count the method will be able to quantifY these organisms only in badly contaminated samples. 2 Injured cells may not have sufficient time to form microcolonies. The results with frozen fish and meat (Table 2) demonstrate slightly more scatter

41

DEFT: RECENT DEVELOPMENTS TA[JLE 2. Correlation of microcolony Dl::.rJ· counts with viable cOimls offood Medium for viable count

n

m

,.

r

Lowest count (logJ() reliably detected

Raw meat NA ABA PSA FJ,SAB

24 19 22 22

0.85 0.44 0.90 0.76

0.52 2.85 0.12 1.23

0.90 0.63 0.93 0.94

4.8 3.8 3.1 3.3

Frozl'11 meal or fish NA ABA PSA ELSAB

29 16 25 26

0.85 0.52 0.82 0.76

0.25 2.9 0.55 0.86

0.88 0.64 0.92 0.83

4.3 4.5 3.6 4.1

Frozen vegetables NA ABA I'A ELSAB

19 33 6 7

0.79 0.76 0.81 0.55

0.58 0.98 0.38 2.01

0.84 0.80 0.93 0.90

4.0 3.2 2.8 3.1

Regression data

n, the number of individual samples; m and c, values of the regression lines ofy = mx + c of plots oflogJ()cfu plate count/ml (x axis) and logJ() microcolony DEFT counts/ml (y axis); r, correlation coeflicient. NA, nutrient agar (total aerobic plate count); ABA, azide-blood agar (staphylococci/streptococci/ micrococci); PSA, pseudomonad supplement agar (pseudomonads); ELSAB, lauryl sulphateaniline blue agar (coliforms). Filters placed on NA, PSA and ELSAH were incubated fin 3 h and those on ABA for 5 h at 30°C.

than with raw foods and inadequate resuscitation of injured cells is suspected. Indeed, this is to be eJl.'Pected with the highly selective nature of the media used, which contain antibiotics or inhibitors that can be injurious to sublethally damaged cells. The results demonstrate the potential of the method with foods, however, and a detailed study of the effects of sub-lethal heat treatments on the efficiency of microcolony formation is being undertaken. 3 The selectivity of the method can only be as good as the selectivity of the medium. Few media appear to be totally selective for microcolony formation as they do not inhibit all the non-selected organisms. Many rely on differential colony morphology or changes in the medium surrounding the colony, e.g. pI I, to be diagnostic of the required organism, and many do not support the growth of microcolonies of all the strains or species of their target organisms.

42

G. L. PETTIPIIER £1' AL.

These results demonstrate that the microcolony/DEFT can be used to count specific groups of bacteria. The results are available well within a working day and this should allow better management of food supplies. The modifications also demonstrate that the full potential of the nlethod(s) has yet to be realized. There is no reason why the DEFT cannot be similarly modified specifically to detect other groups of organisms and, if used in conjunction with fluorescent polyclonal or monoclonal antibodies, this differentiation could be improved still further.

Use of DEFT for Irradiated Foods There is increasing interest in the use of gamma radiation in food preservation. Reports have suggested that the irradiation of foods with a dose of 10 kGy (1 Mrad) of gamma radiation presents no toxicological hazard and introduces no special nutritional or microbiological problems (Anon. 1986). One of the major points of concern about the use of radiation as a food preservative has been the lack of a simple, reliable method for detecting when the process has been used. Some methods for detecting irradiated foods have been proposed but most of them are elaborate and require ex])ert skills. The DEFT has the potential to be a relatively simple, inexpensive way to indicate whether a food has been irradiated. In addition, it will provide an estimate of the microbiological quality of an irradiated food prior to the process (Betts et al. 1988). This method for indicating the presence of an irradiated food relies on the comparison of the aerobic plate count (APe) and the DEFT count of the same sample. The APe will enumerate only those organisms remaining viable after the irradiating process. The DEFT, however, will show not only the organisms remaining viable but also those killed by the process. The APe of an irradiated food is therefore generally 2- 3 log cycles lower than the DEFT count, the DEFT count correlating well with the APe of the sample before irradiation (Fig. 1). Irradiation of minced beef, steak, bacon, ham, gammon, ox liver, pate and milk, at the suggested maximum dose of 10 kGy, reduced the APe from 107 -1 0 9 I g to 10 4 -10 6 / g. Milk with an initial count of 107 cful ml had an APe of zero after being irradiated at a dose of 5 kGy. In all cases the DEFT count of food samples after irradiation correlated well with the APe before irradiation. Results obtained for pure cultures are shown in Fig. 2. Initial DEFT and APe counts were in dose agreement but after irradiation at 25 kGy the APC was reduced by 6-8 log cycles, to zero, whilst the DEFT count was reduced by only 0.5-3 log cycles. The following method can be used to screen raw materials which may have been previously irradiated. The DEFT is carried out as described by

43


10

-g' c:

9

:J

10

-----_L---..'-----.

0

u Q>

9

...

!!'

8 §

8

o

ro

0u

:.a

e

Q>

'"

u

7

7

6

6

0

C; 0

--'

rt::i

o

o OJ

o

--'

5

~5

Radiation dose (kGy) FIG. 1. Effect of irradiation on the aerobic plate count (APe) (0), and DEFT count (e), of minced beef samples.

o kGy 0 kGy 0 kGy 0 kGy 0 kGy 0 kGy 25kGy 25kGy 25kGy 25kGy 25kGy 25kGy Bacillus subtilis

Escherichia Salmonella Bacillus coli typhimurium stearothermophilus Staphylococcus Pseudomonas aureus pero/ens

2. Effect of irradiation on the aerobic plate count (APC) (shaded), and DEFT count (black) of bacterial cultures in nutrient broth.

FIG.

44

G. L. PETTIPHER ET AI.

Pettipher (1983). Triton X-IOO and trypsin are used in the pre-treatment usually with 3-6 ml of homogenized sample, depending on the sensitivity required. For meat samples, 10 g of meat is macerated by stomacher in 90 ml of maximum recovery diluent or 0.050/0 peptone (not Ringer solution). The resulting homogenate is pre-filtered through a 5.0-,.un pore size filter (Shaw et al. 1987). Milk can be used without pre-filtration. The pre-treatment, filtration, staining and counting stages of the DEFT are as described in the chapter by Shaw and Farr (this volume). After the DEFT count of the sample has been established, an APC can be set up. If the DEFT count and the APe are similar then the sample has not been processed. If the DEFT count is considerably greater than the APC, then this suggests that the sample could have been irradiated. Heating is the only other treatment which is known to give results similar to this when a joint DEFT/APe is carried out. Heat-processed food in most cases will have a DEFT count higher than their APC but they usually show visual changes indicative of such a treatment. A joint DEFT/APe procedure therefore has potential as a method for the indication of an irradiated food. It also allows the estimation of the viable count of an irradiated sample before irradiation.

References ANON. 1986. Report on the Safety and Who/esom1less oIIrradiated and Novel Foods. London: HMSO. R.P., FARR, L., BANKES, P. & STRINGER, Nl.F. 1988. The detection of irradiated foods using the direct epifluorescent filter technique (DEFn. Jounzal of Applied Bacteriology 64,

BETTS,

329-335. ).R., WHITE, CM. & FIRSTENBURG-EDEN, R. 1984. Rapid impedimetric method for determining the potential shelf-life of pasteurised whole milk. Jounzal of Food Protection 46 1

BISHOP,

622-624. R.L. & JOHNSON, R. 1980. A fluorescent staining technique for determination of viable and non-viable yeasts and bacteria in wineries. Food Technology in Australia 32, 522-523. GRIFFITHS, t\1.W., PHILLIPS, J.D. & MUIR 1 D.O. 1984. Methods for rapid detection of postpasteurisation contamination in cream. Journal of the Society of Dairy Technology 37, 22 - 26. KROLL, R.G. & RODRIGUES, U.M. 1986a. The direct epifluorescent filter technique, cytochrome c oxidase test and plate count method for predicting the keeping quality of pasteurised cream. Food Microbiology 3, 185-194. KROLL, R.G. & RODRIGUES, U.~1. 1986b. Prediction of tIle keeping quality of pasteurised milk by the detection of cytochrome c oxidase. Journal ofApplied Bacten'ology 60, 21-27. McKINNON, CII. & PETTIPHER, G.L. 1983. A survey of the sources of heat-resistant hacteria in

COOTES,

milk with particular reference to psychrotrophic spore forming bacteria. Journal of Dairy Research 50, 163 -170. PETTIPHER, G.L. 1983. The Direa Epifluoresccnt Filter Technique for the Rapid Enumeration oIMicroorganisms. Innovation in Microbiology Series, 1. Letchworth: Research Studies Press. PETTIPHER, G.L. 1987. Detection of low numbers of osmophilic yeasts in creme fondant within 25 h using a pre-incubated DEFT count. Letters in Applied Microbiology 4, 95-98. PETTIPHER, G.L., l\1ANSELL, R., McKINNON, C.H. & COUSINS, C.M. 1980. Rapid membrane


45

filtration-epifluorescent microscopy technique for direct enumeration of bacteria in raw milk. Applied and Ellvironme1ltal Microbiolo&'V 39, 423-429. PETTIPHER, G.L. & RODRIGUES, U.M. 1981. Rapid enumeration of bacteria in heat-treated milk and milk products using a membrane filtration-epifluorescence microscopy technique. Journal ofApplied Bacteriology 50, 157-166. PETTIPHER, G.L. & RODRIGUES, U.l\l. 1982a. Semi-automated counting of bacteria and somatic ceUs in milk using epifluorescence microscopy and television image analysis. Journal of Applied Baeteriology 53, 323-329. PETTIPHER, G.L. & RODRIGUES, U.M. 1982b. Rapid enumeration of micro-organisms in foods by the Direct Epifluorescent Filter Technique. Applied and Environmental Microbiology 44, 809-813. PHILLIPS, J.D. & GRIFFITHS, !\'l.W. 1986a. Factors contributing to the seasonal variation of Badllus spp. in pasteurised dairy products. Journal ofApplied Baeteriology 61, 275 - 286. PHILLIPS, J.D. & GRIFFITHS, M.W. I 986b. Estimation of Gram-negative bacteria in milk: A comparison of inhibitor systems for preventing Gram-positive bacterial growth. Journal of Applied Baderiology 60, 491- 500. PHILl.lPS, J.D., GRIFFITHS, lVl.W. & MUIR, D.O. 1981. Factors affecting the shelf-life of pasteurised double cream. Journal of the Sode~v of Dairy Technology 34, 109-112. RODRIGUES, U.M. & KROLL, R.G. 1985. Increased selectivity, sensitivity and rapidity in the Direct Epifluorescent Filter Technique (DEFT). Journal o/Applied Bacteriology 59, 493-499. RODRIGUES, U.M. & KROLL, R.G. 1988. Rapid selective enumeration of bacteria in foods using a microcolony epifluorescence microscopy technique. .loumal 0/Applied Baderiology 64, 65- 78. RODRIGUES, U.M. & PETTIPHER, G.L. 1984. Use of the Direct Epifluorescent Filter Technique for predicting the keeping quality of pasteurised milk within 24 hours. Journal of Applied Bactenology 57. 125-130. SCHRODER. M.J.A., COUSINS, CM. & McKINNON, CII. 1982. Effects of psychrotrophic postpasteurisation contamination on the keeping quality at I I° and 5°C of I ITST-pasteurised milk in the U.K. Journal of Dairy Research 49, 619-630. SIHW, B.G., lIARDlNG, CD., HUDSON, W.l1. & FARR, L. 1987. The estimation of microbial numbers on meat and poultry by the direct epifluorescent filter technique. Journal of Food Protection 50, 652-657. WAES, G. & BOSSUYT, R. 1982. Usefulness of the benzalkonium-crystal violet ATP method for predicting the keeping quality of pasteurised milk. JOllmal of Food Pro/atioll 10, 928-931.

The Rapid Estimation of Bacterial Counts on Meat and Poultry by the Direct Epifluorescent Filter Technique B. G. SHAWl AND L. J. FARR 2 I AFRC Institute of Food Research, Bristol Laboratory, LangfOrd, Bristol BS18 7DY; and 2 Foss Electric (UK) Ltd, The Chantry, Bishopsthorpe, York, UK

Conventional culture methods for determining total viable counts on meat and poultry take at least 1 day to produce a result. In meat factories this is a considerable drawback in ingredient and raw product testing where results are often needed within hours (or even minutes) rather than days. Direct microscopic enumeration of bacteria using the direct epifluorescent filter technique (DEFT) (Pettipher 1983) offers a solution to this problem as results may be obtained in 35-45 min. The DEFT was originally developed for the rapid enumeration of bacteria in milk (Pettipher et al. 1980) but was later modified for application to other foods including meat (Pettipher & Rodrigues 1982b). Subsequent studies (Boisen 1983; Qvist & Jakobsen 1985; Shaw et al. 1987) have established its suitability for the rapid estimation of total bacterial numbers in a variety of red meat and poultry sample types. This article describes the DEFT procedure as applied to poultry and fresh, frozen and comminuted meat. The DEFT has been evaluated by comparison with the plate count on a variety of meat and poultry sample types to demonstrate the capabilities of the technique. DEFT Methodology for Meat and Poultry The DEFT procedure described in this article is based on that developed by Pettipher & Rodrigues (1982b) for food suspensions. The technique essentially involves the counting of fluorochrome-stained bacteria on membrane filters on which they have been concentrated from the sample suspension following removal of food debris by pre-filtration and enzyme treatment. The procedure may be conveniently divided into five stages as shown in Fig. I. Cr;pyright

© 1989 by the Society for Applied Bacteriology All rights of reproduction in any for", reserved 0-632 -02629-4

Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticals

47

48

B. G. SHAW AND L. J. FARR Preparation of sample suspension

! 1 1 1

Prefiltration

Enzyme and surfactant treatment

Filtration and staining

Counting FIG.

1. Outline of DEFT procedure for meat and poultry samples.

Preparation of sample suspension The sample suspension may be prepared by any of the sampling techniques in current use (e.g. scraping, swabbing, stomachering). Specific examples of sample suspension preparation techniques which we have used successfully with the DEFT are shown in Table 1. Rubber bungs or bottIe cap sealers must not be used on any sample container or equipment used in the DEFT as this will inevitably lead to orange fluorochrome-stained debris and incorrect counting. The diluent used in the preparation of the sample suspension must contain 0.1 % (w/v) peptone, otherwise cells may not take up the fluorochrome effectively at the later staining stage. A solution of 1% (wIv) peptone (pH 7.0) is recommended but peptone salt dilution fluid (0.1 % (w/v) peptone, 0.85% (w/v) NaCI; pH 7.0) has proved satisfactory. Tween 80 should be added at a concentration of 1 % (v Iv) prior to stomachering of samples with a relatively high fat content (e.g. sausages, mechanically recovered meat) - this disperses the fat, aiding pre-filtration and improving the quality of the final stained DEFT preparations by minimizing interference from fat particles. As recommended by Boisen (1983), comminuted meats should be placed in an inner nylon-lint bag (Foss Electric (UK) Ltd, York) prior to addition of diluent and stomachering. This retains a large proportion of the solids, whilst still permitting full recovery of organisms.

Pre-filtration Pre-filtration aims to remove coarse particles from the sample suspension and should be applied to all types of preparation including swab suspensions. For

49

DEFT FOR MEAT AND POULTRY TABLE

1. Meat alld poultry sample suspmsioll procedures appropriate to the DEFT

Sample type Joints of meat

Sampling method

3 x 7 cm l areas

Diluent volume (mI)

Dispersal methods

100

Stomachering for 1 min

cut from surface Carcasses and joints of meat

50 or 100 cnl area swabbed

20

Swab compressed repeatedly \\-ith pipette

Chicken carcasses or joints

10 g skin and muscle excised

90

Stomachering for I min

Chicken carcasses or joints

7 cm 2 area scraped

10

Surface scraped in presence of diluent

Frozen blocks of meat

100-200 g thawed in microwave oven (230 W, 3 mill)

Comminuted meat (e.g. sausages, mechanically recovered meat)

10 g sample examined

Equal to weight of sample

Stomachering for 2 min

90

Sample placed ill nylon-lint bag and stomachered for 1 min

prefiltration of suspensions, 25-mm-diameter 5 f!m polypropylene filter discs (Foss Electric (UK) Ltd) are mounted in Swinnex filter holders (Millipore Ltd, London) and autoclavcd before usc. After allowing large food particles in thc homogenate to settle for 2-3 min, a 10-15 ml volume of the sample suspension is drawn into a sterile disposable plastic syringe and forced through the Swinnex filter unit into a sterile McCartney or Universal bottle. Sample suspensions with an expectcd count > 10° cful ml should be decimally diluted below this level prior to pre-filtration, as should any sample which blocks the pre-filter prior to the passage of 3 ml (the normal volume used). Enzyme and surfactant treatment

Treatment with trypsin and Triton X-IOO disperses protein, lyses muscle cells and makes fat globules more fluid, aiding the final filtration stage and improving the quality of the stained DEFT preparations (Pettipher & Rodrigues 1982b). Reagents for use in this and the latcr filtration and staining stage may be prepared and dispensed using filters and pipetters obtained as individual items. Alternatively the bio-Foss Reagent System (Foss Electric (UK) Ltd) may be used. This is purpose-designed for the DEFT including all necessary filter units, containers, tubing and pipeners and greatly simplifies and speeds up procedures.

50

B. G. SHAW AND L.

J.

FARR

Trypsin is prepared by reconstituting the freeze-dried enzyme (Difco

8103 - 97) in chilled (...... 7°C) distilled water and sterilizing by passage through a 0.2-~m disposable filter (Foss Electric (UK) Ltd). The enzyme is dispensed aseptically in 0.5 ml amounts into sterile capped test-tubes, excess of requirement being stored in these tubes deep-frozen at - 20 0 (maximum 3 months). Triton X-IOO (BDII Chemicals Ltd or Difca 8106-97) is prepared as a filter-sterilized 0.5% (v/v) solution made up in warm distilled water. For treatment 3-9 ml of the pre-filtered sample suspension (the volume dependent on the count sensitivity required) is thoroughly mixed in the testtubes using a vortex mixer with the trypsin and 2 ml of added 0.5% (v/v) Triton X-IOO solution and incubated at 50°C for 10 min in a water bath.

e

Filtration and staining This stage is vital to the success of the DEFT. A good reliable, efficient membrane filtration system is essential if consistent DEFT slides are to be prepared suitable for accurate counting. This should include membrane filtration units having stainless steel screen membrane filter supports (sintered glass bases are not suitable), filter towers, and a 5-way filter manifold preferably fitted with a vacuum gauge (a vacuum of 15 - 20 mmHg is optimal excessive vacuum may distort the membranes). The manifold, which allows processing of five samples at a time, should be connected via a 3-way valve (one vent to the atmosphere) to a side-arm flask (to collect the filtrate) and then to the vacuum source for which an electric vacuum pump is preferred to a water jet. A suitable filtration system is commercially available (Foss Electric (UK) Ltd). All reagents used in flitration and staining should be pre- filtered through 0.22 (.tm membrane filters to remove bacteria and particulate matter. The fluorescent stain is acridine orange prepared from DEFT buffered acridine orange concentrate (Difco 8101-97). The staining solution should be freshly made up every 2- 3 days and stored at 100 clumps/ field occur. When an automatic counting system (see below) is used, it is convenient to count a set number of fields ,of view (usually 20) whatever the count per field, with the proviso again that the sample should be diluted and repeated when > 100 clumps per field occur. Cells fluorescing green are not included in the count. The DEFT count/ml of sample suspension is obtained by multiplying the

52

B. G. SHAW AND 1.. J. FARR

average number of clumps/microscope field by the microscope factor (MF). The microscope factor is calculated as follows:

MF

=

Area of membrane through which sample is filtered (mm 2) Microscope field area (mm 2) X Sample volume (ml)

The area of the filter used is calculated from the internal radius of the filter tower. The area of the microscope field of view is calculated from the radius of the field which is measured using a stage micrometer. Manual counting of DEFT preparations is cAtremely laborious and may cause eye fatigue. Automatic counting is possible using a television scanner linked to a suitable image analysis system as described by Pettipher & Rodrigues (1982a). A commercially available automatic counting system (Foss Electric (UK) Ltd), including microprocessor controlled image analyser and automatic slide stepper, requires the operator merely to focus the microscope and operate a hand switch. Counts/ml of sample suspension are automatically calculated and may be produced as a printout or transferred to a computer to give counts/g or / cm 2 and for storage of results in a file. Comparison of DEFT and Plate Counts The DEFT as described in this article aims to produce the equivalent of the total viable count. To evaluate the technique, Shaw et ai. (1987) have compared DEFT counts with plate counts on a variety of uncooked red meat and poultry samples examined as shown in Table 1, covering common sampling techniques in current usc. The procedure used for the DEFT was as described in this article except that Tween 80 was not added at the stomachering stage. The apparatus used was the bio-Foss Automated Microbiology Systeln (Foss Electric (UK) Ltd) shown in Fig. 2 which includes the purpose-built reagent and filtration equipment and automatic counting system. Total counts in these comparisons were made by surface plating on plate count agar (PCA, Oxoid) + 1% NaCI with incubation for 5 days at 20o e. The summarized results (Table 2) show that the DEFT count correlated well (correlation coefficient (r) = 0.95 -0.96) with the plate count for frozen or chill-stored lean red meat and for red meat carcasses (beef, pork, lamb). For frozen meat and swabbed samples the DEFT gave good agreement with the plate count over a range adequate tor routine quality evaluation (5 x 103 /g or /cm 2 and above). For samples taken by surface excision with plate counts of I04/cm 2 or less, DEFT counts were >0.5 lOglO unit higher, indicating a loss of sensitivity. If required, precision at these lower count levels could be enhanced either by use of a larger sample volume (6 ml were used for red meat samples) or by increasing the number of microscope fields counted. For stored and unstored raw chicken sampled by skin scraping or

DEfT FOR "IEAT AND POULTRY

53

FIG. 2. The hio-Foss Automated .\licrobiology System equipment fiJr the DEFT.

stomachering good overall agreement (r = 0.88-0.89) was obtained between the DEFT count and the plate count in the ranges 1.1 X 103 to 1.3 X 107 1 cm 2 (scraping) and 1 X 10 4 to 9.5 x Hl'/g (stomachering) even thoug'h the DEFT overcounted by >0.5 10glO unit on samples on which little or no growth had occurred (plate count --.J

3

/

o

o

0/ S

4

/

3

1

a

Log 10 plate count

b

6

I

": /0 /

o

345

~!/(/

r /

0>

o

--.J

o

2

/I

5

o

/~

l

o

C :J

o

234

5

Log 1D plate count

6

C 5 :J o o

~

~4 0>

/

o

--.J

/

l

i

/

§

5

o o

~

;;=

S 4 0>

o

--.J

3·

3

o

1

c

Log 1D plate count

234

5

o

1

d

Log lO plate count

234

5

FIG. 4. Correlation hetween IOglO memhrane tilter fluorescence count (l') and IOglO initial pour plate colony count (x) for Esdlerichia wli (a), PSl'udollllJ1laS aeruKillosa (h), Badllus sublilis (c) and Bacillus pumilus spores (d). 0, no incuhation before DEFT; ., 3 h incuhation before DEFT; 6, 5 h incubation before DEFT; titted regression lines given hy .y = a + bx. (Reproduced with permission from .toun/al o/PaYl7lleral Sciellce alld Tedll1ology.)

membrane filtration sterility test Crable 2). A full statistical treatment of these results is presented in the original papers (Denyer & Ward 1983; Denyer & Lynn 1987).

Conclusion The DEFT has been successfully applied to the analysis of unne for 'significant bacteriuria' and for the detection of contaminants In

70

S. P. DENYER ET AL.

aqueous infusion fluids. In its conventional form it is capable of providing a workable detection sensitivity (Le. 1 cell/field of view) in the region of 5 X 10.1 organisms/sample, dependent upon the microscope factor employed. '''ben used following a 5 h pre-incubation stage, as few as six viable organisms can be detected, irrespective of sample volume processed. The DEFT offers the benefits of rapid enumeration, discrimination between cell and particle types, and the opportunity for tentative identification by cell morphology. In intravenous fluids, comparisons between pre- and postincubation DEFT counts afford confirmation of cell viability. At present, the technique is limited in its routine medical and pharmaceutical laboratory applications by the labour involved in sample preparation and manual Iuicroscopy. It is now feasible, however, for the staining procedure and microscopic examination to be automated (Pettipher & Rodrigues 1982), and suitable image processing hardware and computer-controlled XYZ microscope stages are available, otTering the potential for adaptation to suit Individual applications.

References ANDERSON,

R.L.,

HKiHS,," 11TH,

A.K. &

HOLl.AND,

RW. 1986. Comparison of standard pour

plate procedure and the ATP ami Limulus Amoebocyte Lysate procedures for the detection of microbial contamination in intravenous fluids. ]()umaJ of Clillim/ /\!1icrobio/01.TJ' 23, 465~468.

A.E., BETTELHEI.\I, K.A., AI.-S,\UHI, S. & SH.-\\\, E.). 1978. The enumeration of bacteria in culture media and clinical specimens of urine by microcalorimetry. Science Tools

BtTZER,

25,6-8. EOI>'" CA. &

\V.\CHS.\'IITH, I.K. 1981. Ludferase assay to detect bacterial contamination of intravenous fluids. Amer;can ]ounw! ({Hospital Plwnluuy 38, 1747-·-1749. CADY, P., DUFOUR, S.\V., L\\\LESS, P., NUNKE, B. & KRAEt,;\R, S.]. 1978. Impedimerric scn:ening {or bacteriuria. ]ounwl of Clinical iHiaobjology 7, 273~278. DENYER, S.P. ]982. In-use contamination in intravenous therapy - the scale of the problem. In Injitsiolls and Infix/ions. The Ha::Alrds of Iu-use Coutaminalion in Illlmvellous Therapv, cd. D'Arcy, P.F. pp. 1-15. Oxford: Tbe Medicine Publishing Foundation. DENYER, S.P. & LYN~, R. 1987. A SCnSillye method for the rapid detection of bacterial contaminants in intravenous fluids . .7oumal 0/ Parmlt'raI Sdnw! and Tedln%b'Y 41, 60- 66. DENYER, S.P & \V,-\RJ), K.I 1. ] 983. A rapid method for the detection of bacterial conlaminants in intrayenous fluids using membrane filtration and epitluorescence microscopy. .7ounJtll 4

Parmleral Scient!.' awl TedwoJof...'}137, ]56-158. JE., CAMPUEI.I., ].E., DONNELLY, C.B., PEELER, j.T. & DU.ANEY, ).'\1. 1973. Spiral plate method for bacterial determination. Applied Jlicrobiolo!:.1J' 25, 244- 252. IIFI\,lRICHSEN, C. & J\Io\'Es, A. 1987. A semi-automated method for the culture, identification and susceptibility testing for hacteria direct from urine specimens. iWet/fca! Labora{ory Sdow:s

GII.CHRIST,

44,50-58. NF,\\;'Vl:\.!'\, D.E. & ST:\NECk, J.L. 1985. Bacteriuria screening by use of acridine orange stained smears. Journal of Chuiwl Alitrobiolob.1J' 21, 513- 516. jOl1' Ihe S",iuy Jin /l/J/Jlicd

Ba{/crilJllJ/,~'

All n~hls ({ rcpmduflifm ill atq' ./flnn rfscr;}('t/

Rapid Microbiological ;\ Iethods I"r Foods, Beverages and Pharmaceuticals

0-632-02629-4

165

166

J.

A. BIRD ET AI..

and for the induction of TMAO reductase enzymes. An evaluation of this method using artificially contaminated milk powders (see Prentice et al., this volume) found that the incubation period ofthe selective broth in the instrument needed to be increased from 24 to 30 h to allow for late detections. This lengthened the minimum total test time of the conductance method to 2 days. Detection of Salmonella by the conductance method is based on changes in conductance-with-time curves, which show typical Salmonella characteristics. These are produced when there are levels of 104 -10 5 salmonellas/ml EasterGibson medium. The time taken to reach this threshold is related to the number of salmonellas inoculated into the conductance medium and their subsequent growth. Immobilization of salmonellas from pre-enrichment broth cultures on to the antibody-coated solid supports used in the ELISA procedures could result in a concentrated inoculum such that higher numbers of Salmonella would be transferred to the conductance medium. This would reduce the instrumental incubation time required, Le. time taken to reach the 104 -105 cells/ml threshold of the instrument. Concentration of Salmonella could also decrease the pre-enrichment broth incubation period, resulting in a shorter total test time. The development of the ImmunoModified Electrical Technique (lMET) is described. Hadfield et al. (1987) described a rapid confirmatory test for Salmonella using a simple coloured latex agglutination technique based on serology that gives a result in 5 min. The coloured latex test was evaluated for its ability to detect Salmonella in naturally and artificially contaminated foodstuffs and environmental samples.

Materials and Methods

Media and chemicals Commercially prepared media used were: Bismuth Sulphite Agar (BSA, 007301-1, Difco); Brain- Heart Infusion Broth (BIll, CM225, Oxoid); Brilliant green agar (BGA, CM263, Oxoid); Buffered Peptone \Vater (BP\V, CMS09, Oxoid); Muller-Kauffman Tetrathionate Broth (TET, CM343, Oxoid); nutrient Agar (NA, CM3, Oxoid); Nutrient Broth (NB, CM1, Oxoid); Selenite Cystine broth (SEL, CM699 + LI21, Oxoid) and Xylose Lysine DesoAycholate Medium (XLD, CM469, Oxoid). :Hall's resuscitation agar (HRA) and minimal medium (MM) with filtersterilized glucose (0.2% w/v, BDH Ltd, Poole, Dorset) were prepared according to Hall (1979) and Davis (1980) respectively. The pre-enrichment broth used for all the methods consisted of BHI or BPW containing 0.5°10 (w/v) dulcitol (44590, Fluorochem, Glossop, Derbyshire) and 0.1 0/0 (w/v) trime-

SALMONELLA IMMUNO/CONDUCTANCE DETECTION

167

thylamine oxide (TMAO·2H zO, 92277, Fluorochem). The Easter-Gibson conductance medium was prepared according to Easter & Gibson (1985). A protein solution was prepared by diluting Bovine Serum Albumin (BSA, 44155, BOB) in Phosphate Buffered Saline (PBS, BR14a, Oxoid) to give a 1% (v/v) solution. Binding buffer solution was prepared at time of use by similarly diluting BSA in PBS to give a final concentration of 0.1 % (vIv) BSA. The wash solution consisted of PBS containing 0.05% (v/v) Tween 20 (66368, BOH). P-nitrophenylphosphate (N2765, Sigma Ltd, Poole) was diluted in I molll- 1 diethanolamine (08885, Sigma) at pH 10.0 to give a 1 mg/ml substrate solution.

Bacterial cultures Several salmonellas and closely related non-salmonellas were used in this study, many of which were isolates from naturally contaminated foods and environmental samples. Other bacteria used were obtained from the Public Health Laboratory Service, Colindale (strains labelled BR), National Collection of Type Cultures, London (strains labelled NCTC), Colworth Culture Collection, Unilever Research, Shambrook (strains labelled CCC) and Malthus Instruments Culture Collection, Crawley (strains labelled M). The cultures used were Salmonella agona (two strains including BR 1786), Salm. anatum (two strains including NCTC 5779), Salm. enteritidis (NCTC 4444), Salm. eastbourne, Salm. hadar (NCTC 9877), Salm. indiana (NCTC 11304), Salm. inflmtis, Salm. kedougou (two strains), Salm. kurbacha, Salm. montevideo, Salm. napoli (BR 3427), Salm. stanlc.y (NCTC 92), Salm typhimurium (two strains including phage type 10, CCC 1(6), Salrn. virchow, Citrobaaer freundii (three strains including M 120), Enterobacter cloacae, Escherichia coli, Hafnia alvei (MI26) and Proteus mirabilis. Stock bacteria were streaked on NA slopes and after incubation for 24 h at 3TC were stored at 4°C. Stock cultures were sub-cultured at monthly intervals until required.

Test materials Log phase cultures of Salm. agona, Sa1m. eastbourne, Salm. infimtis, Salm. kedougou, 5'alm. kurbacha, Salm. montevideo, Salm. ~yphimurium, Salm. virchow, Ent. cloacae and P. mirabilis grown in skimmed milk concentrate (40% w/v, Milk Marketing Board (MMB), Thames Oitton, Surrey) for 6 h at 37°C were indiYiduallY spraY-drinl with a laboratory spray drier (Model 190, Biichi Ltd, Flawil, Switzerland). A control powder was also prepared by spray-drying uninoculated skimmed milk concentrate. These powders were then diluted in skimmed milk powder (SMP), whey powder (WP) and whole milk powder

168

J. A. BIRD ET AL.

(WMP) to give estimated levels of Salmonella, Ent. cloacae or P. mirabilis of I, 4, 4 colony-forming units (cfu)/g powder, respectively. The uninoculated control powder was similarly diluted in SMP, WP and WMP. The milk powders containing the salmonellas and the uninoculated control powders were mixed in equal quantities with SMP, WP and WMP which were then inoculated with the Ent. cloacae and P. mirabilis Gram-negative competing microflora or left uninoculated. The ratio of competitors to salmonellas was 8:1, and the final levels of Salmonella in the powders were in the range < 3 cfu/IOO g to 1100 cfu/g powder (see Prentice et al., this volume). These samples (n = 354) were used for the evaluation of the coloured latex agglutination test only. Milk-based powder and environmental samples (n = 104) taken from a commercial spray drier were also examined by the traditional method and the coloured latex agglutination test.

Salmonella detection methods Cultural method Following overnight incubation in BPW at 37°C, pre-enrichment cultures were inoculated into SEL and TET selective enrichment broths, and incubated at 37°C and 43°C, respectively. After incubation for 24 and 48 h both selective broths were streaked on BGA, BSA and XLD selective agars and incubated at 3rC for 24 h. Presumptive colonies were confirmed by agglutination tests using polyvalent 0 and B antisera (9620- 53, 9630- 53, Difco).

Conductance method The method of Easter & Gibson (1985) was employed, using a 128B Malthus Microbial Growth Analyser (Malthus Instruments Ltd, Crawley) or 120SC Bactometer (Bactomatic Ltd, Henley on Thames). Malthus cells, each containing 10 ml of Easter-Gibson medium, were inoculated with 0.25 ml preenrichment culture. Bactometer wells containing 2.0 ml Easter-Gibson medium were inoculated with 0.05 ml pre-enrichment culture. The cells and modules were incubated at 3rC and conductance measurements recorded for 48 h. Cultures producing typical Salmonella conductance curves were streaked on XLD and incubated at 37°C for 24 h. Presumptive colonies were confirmed by agglutination tests.

Interpretation of conductance curves Conductance curves obtained from the Malthus and Bactometcr instruments were considered Salmonella positive if they showed the following characteristics:

SALMONELLA IMMUKO/CONDUCTANCE DETECTION

169

(1) a flat base-line and slight plateau immediately prior to the main detection point; (2) a sharp elbow or 'take-oir point at tJle point of detection; (3) a fast rate ofconductance change (100 ftS/h Malthus, step size ~ 6 units Bactometer); (4) a large total conductance change (> 300 f.tS Malthus) and (5) a flat top to curve (similar to stationary phase).

ImmunoModijied Electrical Technique (IME1) The Bio-Enzabead Salmonella Screen Kit (Organon Teknika UK Ltd, Cambridge) was used as a source of ferrous-coated polystyrene beads precoated with monoclonal antibodies specific to Salmonella. A lO-ml sample of a pre-enrichment broth culture incubated at 3TC was placed in a sterile capped lO-ml Malthus tube, Bio-Enzabeads were added, and tJ1e MaltJ1us tube placed horizontally in a shaking incubator (Dynatech Ltd, Billingshurst, Surrey) for 20 min at 37°C with sufficient agitation to gently move the beads around in the broth. Following incubation a magnet positioned at the base of tJ1e Malthus tube retained the beads whilst tJ1e preenrichment culture was decanted off. The beads were washed gently by dipping once in sterile BPW before 10 ml of Easter-Gibson medium and a sterile 10 ml electrode were added. The Malthus tube and contents were placed in the Malthus instrument and conductance changes were monitored for 48 h at 37°C. Broth cultures from selected tubes were streaked on XLD and incubated at 3TC for 24 h. Presumptive colonies were confirmed by agglutination tests.

Preparation offreeze-injured Salmonella cultures An overnight culture of Salmonella grown in NB at 3TC was diluted lO-fold in fresh medium and 0.5 ml transferred to a test-tube containing 9.5 ml NB. Growth at 30°C was monitored by optical density measurements. Mid-log phase cultures (0.1 ml) were inoculated into 10 ml MM with a 0.2% (v/v) glucose supplement. Injury was induced by two cycles of freezing to - 20°C for 24 h followed by overnight thawing at 5°C. Numerical differences in colony-forming units (cfu/ml) on BRA and XLD gave an estimation of the proportion of sub-letllally injured salmonellas in tJ1e culture.

Determination of minimum pre-enrichment incubation time Injured cells and overnight cultures of Salm. enteritidis, grown in NB at 3TC, were individually inoculated into flasks containing 100 ml pre-enrichment broth such that levels of 1-10 cells/ml were attained. The pre-enrichment cultures were incubated at 37°C in a water bath. Each culture was analysed at hourly intervals by plate counts on I IRA and XLD using a spiral plate maker

170

J. A. BIRD

ET AL.

(Spiral Systems Inc., USA) and plates were incubated for 48 h at 30°C and 24

h at 37°C, respectively. Aliquots (0.25 mt) of the culture were also inoculated in duplicate into Malthus tubes containing 10 ml Easter- Gibson medium and 10 ml pre-enrichment broth. The conductance method protocol was then followed for these rubes. The IMET method was carried out on the remaining portion of the pre-enrichment culture sample. These eX'j)eriments were repeated using other Salmonella serotypes and a pre-enrichment incubation time of 7 h. The other serotypes used were: Sa/m. agona (BR 1786), Sa/m. anatum (NCTC 5779), Sa/m. hadar, Sa/m. indiana, Sa/m. napoli, Sa/m. stan/e.Y and Salm. ~yphimun'um (phage type 10 CCC 166).

Effect oj competing bacteria on the detection of Salmonella conductance methods

~y

the IMET and

Overnight cultures of Sa/m. enteritidis, C. freundii and H. alvei, grown in NB at 37°C, were inoculated into 100 ml pre-enrichment broth to give ratios of Salmonella to competing bacteria ranging from 1: 100 to 1000:1. The preenrichment cultures were incubated in a water bath at 37°C for 7 h. Aliquots (0.25 ml) were used to inoculate the selective Easter- Gibson medium used in the conductance method. The IMET procedure was also performed on these pre-enrichment cultures.

Comparison of IMET, conductance and cultural methods in the deteaion of Salmonella from artificial(y and natural(y contaminated foods

A variety of foods from different sources naturally and artificially contaminated with a range of Salmonella scrotypes at differing levels were examined using the traditional cultural, conductance and IMET methods. Antibod,y immobilization on Dynal beads Eppendorf tubes (Sterilin Ltd Hourslow) were filled with protein solution, capped and incubated at 37°C in a waterbath for 3 h. The protein solution was discarded and the tubes washed ten times by filling, emptying and rinsing each tube with the wash solution. The protein-coated tubes were then airdried and stored at 4°C until required. One millilitre of Dynal beads (14001, Dynal UK Ltd, Wirral) was added to a protein-coated Eppendorf tube and washed three times in the wash solution by use of a magnetic particle concentrator (MPC, 12001, [)yoal). The Dynal beads were re-suspended in 2.5 ml of binding buffer. Aliquots (0.7 ml) of the beads in buffer solution were dispensed into five proteincoated Eppendorf tubes. Different volumes of polyclonal '0' antibody solution (9620- 53, Difco), 0, 0.05, 0.1, 0.2 and 0.5 mt were added to each tube and

S/ILMONELIA IMMUNO/CONDUCTANCE DETECTION

171

the total volume made up to 3.5 ml with binding buffer solution. Each Eppendorf tube was wrapped in tissue paper, placed on a rotating wheel and mixed end-over-end for 24 h. Following the binding period the MPC was used to retain the Dynal beads in the tubes while the binding buffer/antibody solution was discarded. One millilitre of wash solution was added to each tube and mixed as before for 45 min at 4°C. This wash stage was repeated a further three times. The antibody-coated Dynal beads were finally suspended in 0.2 ml binding buffer containing sodium azide (0.0001 % w/v) added as a preservative, and stored at 4°C until required. The coated beads were washed several times in wash solution to remove the preservative at time of use.

Determination of amount of antibody bound on Dynal beads Goat-anti-rabbit IgG antibodies conjugated with alkaline phosphatase (Sigma Ltd, Poole) were diluted 1 in 1000 in binding buffer solution. Aliquots (0.01 ml) of antibody-coated Dynal beads were dispensed into Eppendorf tubes and 0.5 ml binding buffer and 0.1 ml conjugated antibody solution were added to each tube and thoroughly mixed. After incubation for 3 h at 37°C in a water bath, the beads were washed five times in wash solution using the MPC. The p-nitrophenol phosphate substrate solution (0.8 ml) was added and the tubes incubated at 3rC for 30 min. The enzyme-substrate reaction was stopped by adding 0.2 ml of 1 mollI NaOH (Sigma) to each tube. The supernatant of each tube (0.2 ml) was transferred to a microtitre plate (Immulon 129A, Dynatech Laboratories Ltd, Billingshurst, Sussex) and the optical density (OD) measured at 405 nm, using a plate reader (Dynatech Laboratories Ltd).

EjJect of solid support size on the minimum pre-enrichment incubation time The experiments previously described to determine the minimum preenrichment incubation time were repeated using cultures of injured and uninjured Salm. enteritidis. The IMET method was modified by dividing the remaining portion of each pre-enrichment culture sample into one subsample of 5 ml and two of approximately 2 m( Two Bio-Enzabeads were added to the 5-ml sub-sample whilst 0.2 fll of antibody-coated Dynal beads were added to each of the 2-ml samples. Additionally, at the 7-h preenrichment sample time, a control of uncoated Dynal beads was incorporated.

Evaluation of the coloured latex agglutination test The method described by Hadfield et al. (1987) was followed; reagents and equipment were supplied by Wellcome Diagnostics, Department of Research and Development (Beckenham, Kent). Artificially contaminated milk powders

172

J.

A. BIRD ET AL.

and naturally contaminated milk-based powder and environmental samples were examined, using the cultural method. Selective broth cultures and presumptive colonies on BSA, BGA and XLD agars were tested for the presence of Salmonella with the coloured latex agglutination test.

Results

Determination of minimum pre-enri[hment inmbatioll time Typical Salmonella conductance curves were obtained using the IMET method for all cultures of uninjured Salm. enteritidis that had undergone pre-enrichnlent incubations of 5 h (Table 1). Frozen sub-lethally injured Sa 1m. enteritidis cells produced similar conductance curves, but the minimum pre-enrichment incubation time was extended to 6 and 7 h for the pre-enrichment broth and Easter- Gibson medium, respectively (Table 2). Cell counts of the uninjured and injured salmonellas at these times were in the range 103 - 104 cells/ml. The bacterial counts on XLD immediately after inoculation of sub-lethally injured Salm. enteritidis cells into the pre-enrichment broth were up to 3 log cycles lower than corresponding counts on HRA. The counts on the two media became similar after 4 h incubation of the pre-enrichment culture (Table 2). The seven other Salmonella serotypes tested all produced typical conductance curves after a 7-h pre-enrichment incubation and immobilization stage. The mean detection times for these serotypes ranged from 13.8 h for Salm. agona to 35.4 h for Salm. stanley (Table 3). The average detection time for the eight scrotypes was 20.6 h.

Effict of competing baaeria on the detection of Salmonella by the [MET and conduaance methods The IMET method detected Salm. enteritidis in the presence of competing bacteria in five of the six inoculation level combinations (Table 4). Presumptive Salmonella colonies were not produced on XLD selective agar for the IMET negative cultures although salmonellas were initially inoculated into this broth. Salmonella was detected in both of the duplicate tubes by the conductance method for five of the seven combinations. The remaining two samples were found to be positive in only one of the pair of Malthus tuhes. The conductance method gave detection times up to 11.4 h shorter than those of the IMET procedure (Table 4). However, the IMET conductance curves showed greater rates of change of conductance more typical of Salmonella. The total change in conductance values were similar for both methods (Fig. 1).

SALtlONELLA I\1MUNO/CONDUCTANCE DETECTION

173

1. Delee/iou ~r Salmonella ~y lite I;HET melhod ("siug pre-eIln'dl1nelll brollt (/1) and EasterGibsoll medium (B)) (lIId baclerial UlUIl/S OIl HR~ al various iumba/ioll limes of a pre-mridltllelll brolh ill/Kulaled lI'ith uuilljured Salm. enteritidis

T\IlI.E

Pre-enrichment incubation time (h)

Number of typical Salmonel/a conductance curves t

Bacterial count* (cfu/ml)

liRA 6.0 4.2 6.0 2.1 1.3 5.2 1.3 6.9 9.6 9.1

()

I 2 3 4 5 6 8 18 24

X

10 2

X

]02 2

X X

10 103

x IO~ x IO~ x lOs X 10" x 10K x 10K

A

B

0 0 0 0 0 2 2 2 2 2

0 0 0 0 0 2 2 2 2 2

du, colony-fomling units. * Mean of two results. t Two curves examined for each sample.

2. DeIeelioll of Salmonella b)' /he IMET method (using pre-e11ricltmenl bro/h (A) alld EaslerGibson medium (B)), and ba{len'al UJlwts OIl llM alld )(LD al various illcubalioll limes ofa pre-l'1lrichmeul brulh ;'lOculaled wilh /reeze-illjured Salm. enteritidis

TMILE

Pre-enrichment incubation time (h)

0 2 4 5 6 7

8

Number of typical Salmollella conductance curves t

Bacterial count* (cfu/ml) lIRA

2.1 x 10' I.Ix 10 2 2.5 x 102 6.1 X lO2 2.1 X 101 4.9 X 101 2.1 x IO~

* Mean of four results. t Four curves examined for each sample. NT, not tested; cfu, colony-forming units.

XLD

2 6.2 x L7x 3.3 X 1.1 X 1.1 X 1.5 x

W 102 102 IO J IOJ 10~

A

B

NT NT

NT NT

0 0 2 3 4

0 0 0 2 4

174

J. A. BIRD TABLE

ET AL.

3. Deteetio11 times fOr a variety of salmonel/as

~y

IMET using the

Easler-Gibs011 medium

Serotype Salm. Salm. Salm. Salm. Salm. Salm. Salm. Salm.

Range of detection times (h)

l\'1ean detection time* (h)

11.4-16.2 15.0-20.4

13.8 17.4 24.7 15.7 14.7 30.3 35.4 17.2

agona anaturn enteritidis htUlitr indiana napoli stanley typhimun'um

23.2-26.2 13.4-17.6

11.6-18.6 26.2-34.0 t

14.8-19.4

'*' Mean of four replicates. t One value only.

TABLE

4. Effie! of competing bacten'a on the detection of Sa1m. enteritidis conductance (OW) methods

Salm. ent I 1

Cit

Hal

1

I

10

1 100

100

100 10

1

10

100

10 1

1

1000 1

Number of typical Salmonella conductance cunrcst

Bacterial count'*' after 7 h at 37°e (cfu/ml)

Ratio of inoculated bacteria

0

1 0

lIRA 5.3 x lOs

4.5 X ION 5.7 X 108 8.0 x lOs 7.6 X lOs 7.2 X 108

5.6

X

108

XLD 8.0 1.2

X

8.6 1.9 8.0 4.9

X

X

103 10('

0 X X

X

~y

107 108 107 108

the IMET and

Detection times! (h)

(MET

eM

IMET

eM

2 2

2 1

12.0 6.8

5.0 4.4 15.2 2.8 2.6 2.2 2,2

0

]

NO

2 2 2 2

2

14.2

2 2 2

12.6 13.8

13.4

Salm. ent, Salm. enten'tidis; Cil) C freundii; HaI, H. alvei. ND, not detected; cfu, colony-forming units. '*' Mean of two replicates. t Two curves examined for each sample. r Mean of two replicates.

Compart'son o/IMET, conduaance and cultural methods in the detection of Salmonella from artificially and naturally contaminated fOods Salmonella was detected by both the culntral and conductance methods for 15 of the 22 contaminated food samples, representing an agreement of 68%

SALMONELLA IMMUNO/CONDUCTANCE DETECTION

175

~_-::c

(i) 600

A B

3(I)

()

C ro

tl 400 ::l

"t:)

c:

o

U

200 I

8.0

16.0

24.0

Time (h) FIG. 1. Typical conductance curves for Salmo11ella e11/e11dus in the presence of competing bacteria by IMET (A), conductance method (8) and Salmrmella only control (C).

(Table 5). The same detection results were obtained for both the cultural method and IMET procedure in 19 of the 22 samples giving an agreement of 86%. The minced beefsamples shown as Salmonella positive by the conductance method only were subsequently found to be contaminated with Citrobaaer freundii. Only five and four of the eight whey powder samples were Salmonella positive by the IMET and conductance methods, respectively. However, salmonellas were isolated and confirmed by agglutination reactions from all of the Malthus tubes containing whey powder cultures.

Determination oj the proportion oj po(yclonal 0 antibodies bound to Dynal beads The absorbance values, at 405 nm, of the double antibody ELISA performed on the antibody-coated Dynal beads shows that the lowest antibody volume (0.05 m!) added to the uncoated Dynal beads was sufficient to produce maximum binding (Table 6).

Effict of solid support size on the minimum pre-enrichment incubation time The minimum pre-enrichment incubation time for the IMET method was reduced from 5 and 6 h for uninjured and sub-lethally injured cultures of Salm. enteritidis, respectively, when using Bio-Enzabeads of 3 mm diameter) to 2 h with the use of smaller (4.5 /lm diameter) antibody-coated Dynal beads (1' abies 7 and 8). The bacterial counts of tlle uninjured and injured Salmonella preenrichment cultures corresponding to these Salmonella-positive Dynal results

J.

176 TABLE

A. BIRD ET AL.

5. Detection of Salmonella from naturally and artificially contaminated foods by the traditional cultural (TC), conductance (C#) and IMET methods

Food

Contamination

Serotype

A N N N A

Salm. typhimurium

Cottage cheese Minced beef \Vhey powder Malaysian prawns Chocolate crumb

A Cocoa powder Chocolate Desiccated coconut Linseed

A A A N

N N

* Salm. Salm. Salm. Salm. Sa/m. Sa/m. Salm,

virchow weltvereden panama enten'tidis

Total no. Level of (cells/g) samples

No. of Salmollellilpositive samples

TC

NK

2

2

2

NK "C ",,,,

o 1 ml host salmonella

!d IItJ lj I

II

"T~

l 8 lj r

~

!!

1r

1r

j j

0 0 () 0 0

BHIA plates Ibase layer) 10·

5

10

6

10 7

10 8

10 9

Control

Incubate at 37°C from 6 h to overnight

FIG. 6. Determination of the titre of phage lysate. MRD, Maximal Recovery Diluent; BI 11i\, brain - heart infusion agar; O.7'j(, BI IIi\, BI ILl,. containing 7 gil agar ('sloppy' agar). .'\fter incubation, plaques on suitablt· plates are counted and multiplied hy the dilution factor to give the titre in pfu/ml.

194

S.

J.

PUGH AND M. L. ARNOTT

FIG. 7. ~Sloppy' BHIA plate of host Salmonella with added diluted phage, as prepared during the dctemlination of lysate titre. Lysis of the bacterial cells in a plaque is seen as a dear zone 1- 2 ml11 in diameter.

produced separately and combined 9 final filtcr-steriliz3 tion.

pa11s

0-1 to 1 part G·47 just before the

Inoculum dilution During protocol development we stopped using lhe 1:40 inoculum-ta-medium

ratio specified for SC/T ID by Easter & Gibson (1985). vVith a 50-~tl inoculum this had ~quated to a 2-ml volume of medium in the wells and this frequently caused leakage during handling. The largest inoculum possible has to be used to ensure detection even when pre-enrichment leads to abnormally low cell concentrations but interference from pre-enrichment broths must be avoided. A sinlple protocol which did this and also coped wit.h the normal high cell concentration was desired. \Ve found that optimal results are obtained by diluting the BCilVIB culture 1: lOin MRI) for the SC/T/D and fvlLD wells, and 1:1000 for the phage test.

CONDUCTlMETRIC DETECTION OF SALMONELLAS

FIG.

195

8. Ultrafiltration cell ready lor use with magnetic stirrer and nitrogen cylinder.

Method Evaluation

Pure wIll/res A total of 81 salmonella strains and 39 non-salmonellas were tested in the protocol, as shown in Tables 2 and 3. In general, the negative reactions were evenly scattered amongst the salmonella cultures, but Sainz. guinea was an exception. All five strains tested were phage-negative and this adversely distorts the results. Although one sub-genus 4 su'ain (Sainz. houten) was detected only in one of the three tests (MLD) the protocol as a whole did not miss any salmonellas. Blackburn & Ellis (1973) reported that 15.6% of 552 salmonellas isolated from dried milk products and milk-drying plants were able to ferment lactose. These organisms are easily missed by conventional methods, particularly if they are unable to produce hydrogen sulphide. The irrelevance of these two

196

s. J.

PUGH

2. Reactions of salmonellas

TABLE

Organism

Salmonella sp. alachua anatum anatum arizonae braenderup campinense choleraesuis choleraesuis eastbourne enteritidis enteritidis ftrlac gallinarum guinea hatfield houten infantis italy johannesbu rg kentucky leeuwarden litchfield litchfield litchfield livingstone london montevideo montevideo napoli newport panama paratyphi para~yphi

saintpaul schwarzl71grund senflenberg senfienberg tennessee letmessee thompson ~yphimurium

(yphimurium

AND M. L. ARNOTT

No. of strains*

SC/T/D

MLD

Phage test

13 I 1 1

+ + + +

+ +

+

2 1 1 1 I 1 3

1 1 1

5 1 1 3

I 1 1

I I 1 1 1 1

4 1

1 1 3 I

1 2

+ + + + + + + + + +

+ + + + +

+ +

+

+

+

+

+ + +

+ +

+ +

+ +

+ +

+

+ + + +

+ + + + +

+ + + + + + + + + + +

+ +

+

+

+

+

+ + + + + +

+ + +

1 2

+

7 2

+

3 1

+

+ + + + + + + + + + + +

+ + + + + +

1 2

+ + +

+ + +

+

+ + + + + + +

"* ,",bich gave the same pattern of results. SC/T/D, selenite-cystine-trimethylamine oxide-dulcitol medium. MLD, modified lysine decarboxylase broth. All cultures: Cadbury Ltd Culture Collection.

CONDUCTIMETRIC DETECTION Of SALMONELLAS T.'IIlI.E

Organism

Badlllls eerells Cilmba{/er allla/mlii/ims diverslls ji'CIl11dii jYeliudii E/llt'robacler c!oawt saka;:;akii Fsclierieliia wli

afrugiuosa

Serratia

3. Rm{/io/ls o( /lO/l-sa/ullmd/as !\o. of strains*

SCiT/D

martc.,\(.:fllS

Shif!,d/a jlexneri SlapIiY/lJam'llS allrells epidennidis

[,,[l.D

Phage test

I I I

5 5 2

+

I ')

3

w!i K/cbsid/a p/lell/lll!lliat Pmltlls sr. vil/}!.aris Pselldm/lonas aerll}!.iulJsa

197

2 I I I I 2 I

+ +

+

I I

* Which gaye the same pattern of results. Abbreyiations: sec f()()tnote to Table 2. All cultures: Cadbury [.td Culture Collection.

characteristics in the Bactometcr proIocol means that such organisms are detected without problem. A lactose-positive IlzS-negative strain of Saltlt. 1011dol1 was included in the pure culture screen and was positive in all three tests. Citrobacter freundii gives false-positive reactions in many conventional salmonella tests and indeed five strains did so in SC/T10, but a further five did not. The false-positive reactions in MLD were given by E. coli (three strains), Proteus vulgaris (I) and Pseudomonas aeruginosa (1).

I/loculated samples After pre-enrichment in BGMB, the inoculated samples were tested by conventional and Bactometer methods. The former involved selective enrichment in tetrathionate and mannitol-selenite broths followed by streaking-out on XLO and Hektoen Enteric agars (all Oxoid). The results are given in Table 4.

Confirmatory procedures Presumptive positive reactions were confirmed by plating the contents of wells on XLO and IIektoen Enteric agars. This can be done before the 24-h test is

198

s. J.

PUGH AND M.L.

TABLE

ARNOTT

4. Results from inoculated samples* Bactometer tests

Sample and inoculum Skimmed milk powder Salmonella eastbourne illfimtis euteritidis napoli ~yphimurium

Conventional test

+ +

SC/T/D

MLD

Phage test

+

+ + + + +

+ + + +

+

+ + + +

+

+ +

+

+

+

Uninoculated control Egg powder Salmonella cas/bourne in/antis enteritidis napoli typhimurium Uninoculated control Milk chocolate erumb Salmonella eastbourne il1fimtis enteritidis napoli ~yphimurium

+

+ + + +

+ + + + +

+ + + +

+ +

+

+

+

+

+

+ +

+

+

+

+ + +

+ +

+

+ + +

+

+

Uninoculated control Milk chocolate Salmonella easlbourne itlfimtis enteritidis napoli ~yphimun'um

+ + + +

+

+

+

+

+

+

+

+

+

+

+

+

t-

+

Uninoculated control Dark chocolate Salmonella eastboume ill/antis enteritidis napoli typhimurium Uninoculated control Fondant-filled chocolate egg Sal11l01u:llil eas/boume ;njil11tis ttlterilidis

199

CONDUCTIMETRIC DETECTION OF SALMONELLAS

Ilapoli (yphimurium Uninoculated control Cocoa powder Salm(mella illfanlis e1l1erilidis llapoli lyphimurium Uninoculated control

+

+

+

+ + +

* Bactometer tests which gave positive curves but which did not contain recoverable salmonellas are not included. Abbreviations: see footnote to Table 2.

over if the module is briefly removed from the Bactometer processing unit. If this is done within a few hours of the test starting then the plates are available for examination at the same time as the finished Bactometer curves. Otherwise an extra day is added to the test duration. Confirmation of suspicious colonies was by serology and the use of biochemical test strips (API-20E, API BioMcrieux, Basingstoke or Micro-ID, Organon Technica, Cambridge. On some occasions, positive MLD curves were produced but no colonies appeared on the confirmatory plates - the salmonellas appeared to die out relatively quickly. To avoid this it was suggested that all three tests (excluding the phage-containing well) should be plated if any were positive. However, as usually only one well is presumptive positive, it would now appear sensible to do the following: Presumptive positive SC/TID Presumptive positive MLD Presumptive positive phage test

plate out SC/TID and phage-free DULE well plate out MLD and phage-free DULE well plate out SC/TID and phage-free DULE well

Should more than one test be presumptive positive then all three should be confirmed. We are keen to reduce the time taken by confirmation; novel rapid methods, such as ELISA and modified antisera, are currently under evaluation. These would reduce the test time to 2 days. Conclusion We have found that the Bactometer protocol is a reliable, effective replacement fl1r conventional methods of detecting salmonellas. It significantly reduces the test time and so allows earlier release of raw materials and finished goods.

200

S.

J.

PUGH AND M. L. ARNOTT

The method has a higher sensitivity but a lower cost (disposables amounting to about £1 per test) than the ELISA and DNA - DNA hybridization techniques. Laboratories that test a large number of samples can on this basis justifY the high capital cost of the Bactomcter for salmonella testing alone, but the instrument is becoming increasingly popular because it can be used for many other purposes as well.

Acknowledgement Mr C. Davda and Mrs Y. Watts arc gratefully acknowledged for their technical assistance; thanks are also extended to Dr G. Meier, University of Bern (G.t? phage and host), to Bactomatic Inc. (0 - 1 phage, original phage methodology and some strains) and to the many people who have donated strains to our culture collection.

References RD. & ELLIS, £.M. 1973. Lactose-fermenting Salmonella from dried milk and milk-drying plants. Applied /l1icrob;olob.'Y 26, 672-674. CHERRY, W.o., DAVIS, R.B., ED\VARDS, P.R. & HOGAN, R.B. 1954. A simple procedure for the identification of the genus Salmonella by means of a specific bacteriophage. ]ounlal vf Lahortltory and Clinical Medicine 44, 51- 55. BLACKBURN,

CRAVEN,

P.C,

MACKEL,

D.C.,

BAINE, W.B., BARKER,

\V.H.,

GANGi\ROSA,

Ej.,

GOlDFIELD,

N1.,

11., ALT'\1AN, R., LACHAPELU:, G., DAVIES, ].W. & SWANSON, R.C. 1975. International outbreak of Salmot/ella castboume infection traced to contaminated chocolate. Lancet I, 788-793. D'AOUST, j.Y. 1977. Salmonella and the chocolate industry. A review. ]oumal of Food ProJection ROSENFIELD,

40, 718-727. j.Y. & PIVNICK, H. 1976. Small infectious doses of Salmouella. Lancet 1, 866. EASTER M.C. & GIBSON, D.M. 1985. Rapid and automated detection of Salmonella by electrical measurements. Joumal of Hygiene, Cambn'dge 94, 245-262. FAR.i\1ER, J.J., DA\'IS, B.R., HICKMAN-BRENNER, F.W., MCWHORTER, A., }-IUNTLEY-CARTER, G.P., ASBURY, .M.A., RIDDLE, c., WATHEN-GRADY, ILG., ELIAS, c., F.t\NNING, G.R., STEIGERWALT, A.G., O'HARA, C.M., MORRIS, G.K., S.\1ITH, P.B. & BRENNER, Dj. 1985.

D'AOUST,

Biochemical identification of new species and biogroups of Enterobacteriaceae isolated from clinical specimens. Journal of Clil11l:aJ A1icrobiology 21, 46-76. FELIX, A. & C,\LLO\V, B.R. 1943. Typing of paratyphoid B bacilli by means of Vi bacteriophage. Bn'tish Medical }ouma[ 2, 127-130. FEY, I-1., BURGI, E., MARGADANT, A. & BOLLER, E. 1978. An economic and rapid diagnostic procedure for the detection of Salmonella and ,Shigella using the polyvalent salmonella phage

0-1. Zmlrallblilll ./iir Bakler7oJogie, A1ikrobioJogie ulld Hygiene, J Ableilrmg Or;ginale A 240, 7-15. F1RSTENBERG-Em':N, R. & EDEN, G. 1984. Impedalla: }1'1iaubio!v!fY. Letchworth: Research Studies Press. BARTIJ'TT, C.L.R., VAILE, M.S.B., ROWE, B., GILBERT, R.j., fl.C. & SAl.lv1ASO, S. 1983. Outbreak of Salmonella napoli infection caused by contaminated chocolate bars. Lancel 1, 574-577.

GILL,

a.N.,

DULAKE,

SOCKETT,

c.,

P.N.,

MURRELL,

CONDUCTIMETRIC DETECTION OF SALMONELLAS

201

GREENWOOD, M.H. & HOOPER, W.L. 1983. Chocolate bars contaminated with Salmouella uapoli: an infectivity study. Brilish Medical Joomal 286, 1394. GOIlEl., K. & FEY, I I. 1981. Improvement of the polyvalent salmonella phage's 0- I diagnostic value hy addition of a phage specific for the 0 groups E 1 - E 4 • Zmtrallblall fur Bakleriologie,

Mikrobiologie and Hygime, I Ableilong Origillale A 249,220-224. G.F. & FI.EET, G.H. 1985. Detection of salmonellae using accelerated methods. IlIlemalional Joomal ofFood /'vlicrobiolo/,,'Y 2, 259-272. ODEN-JOHANSON, B. 1972. An epidemic of Salmonella dorham caused hy contaminated cocoa. Liikartidingm 69, 5335 - 5338. PUGH, S.]. 1987. Rapid Microbiological Ivlethods. In Food Tedl1lolo/,,'Y IUlematiOllal Europe, cd. Turner, A. pp. 259-261. London: Sterling Publications Ltd. ROWE, R, BEGG, NT., HUTCHINSON, D.N., DAWKINS, I I.e., GILBERT, R.]., JACOB, M., lIAI.ES, B.H. RAE, F.A. & JEPSON, M. 1987. Salmouella ealiug infections associated with consumption of infant dried milk. Lancel 2, 900-903. . SI\10NSEN, B., BRYAN, F.L., CHRISTIA"J, ].H.N., ROBERTS, T.A., TOMPKIN, R.B. & SIl.LIKER, J .II. 1987. Prevention and control of food-borne salmonellosis through application of Hazard Analysis Critical Control Point (IIACCP). IlIlematiolla! Joomal of Food MiL'robioiogy 4, 227-247. STANNARD, c:.]. 1984. Development and usc of rapid microhiological methods in food quality assessment. PhD lhesis, University of Surrey. STANNi\RD, e.]. 1985. Development of a Rapid EICl1rical Detectioll Melhod for ~aI11lollella. Research Report 511. Leatherhead: British Food Manufacturing Industries Research IIJRAHI~I,

Association.

A Medium for Detection of Lancefield Group D Cocci in Skimmed Milk Powder by Electrometric Methods P. NEAVES, M. J. WADDELL" AND G. A. PRENTICE Techniral Division, Milk Marketing Board, Thames Ditton, Surrey KT70EL, UK

In the food industry, there is a movement away from traditional microbiological techniques towards instrumental methods for the detection of micro-organisms. This change has occurred largely because developments in the application of computers have made easier the automation of microbiological investigations. During the last decade, electrometric instruments have become relatively sophisticated and increasingly abundant in factory laboratories. The Bactometer and Malthus instruments, which are the most common, have both found applications for monitoring production processes and for testing finished products. Their use is restricted, however, because of a lack of suitable selective media. Apart from coliform media, there are few commercially-available media for indicator organisms and pathogens. The commercial production of skimmed milk powder requires strict microbiological control. After pasteurization, milk is concentrated by evaporation and spray- or roller-dried. Under poor production conditions, Lancefield Group D cocci may grow during the manufacture of dried milk powders. A number of major food companies include tests for them in their specification for skimmed milk powders. This work was therefore undertaken to develop a medium for the detection of Lancefield Group D cocci in skimmed milk powder with elcctrometric instruments. Preparation of Skimmed Milk Powder Contaminated with Lancefield Group D Cocci Commercially-produced skimmed milk powder was re-constituted with sterile distilled water to give 30% (w/v) total solids. Actively-growing cultures in Nutrient Broth (Oxoid), of Enterococcus Jaecalis (NCIB 775), Ent. Jaecium " Present address: Dale Farm Foods, Progress Drive, Flash Lane, Bramley, Rotherham, S. Yorkshire 566 OTU, UK. Copyright Rapid l'vlicrobiological Methods for Foods, Beverages and Pharmaceuticals

© 1989 by the Society fiJr Applied BtUurioloJ(Y All rights '1 reproduction in a".y jimn resaved 0-632 -02629-4

203

204

p . NEAVES E1' AL.

(NCIB 662), Streptococcus bovis (NCDO 597) and Strep. equinus (NCDO 1037) were separately inoculated into 1-1 batches of the reconstituted milk. After inoculation, the numbers of micro-organisms were c. 103 cfulml. After incubation at 37°C for 7-10 h, the numbers had increased to c. 108 cfu/ml. The cultures were then spray-dried using a laboratory-scale spray-drier (Biichi Ltd, Flawil, Switzerland). The inlet temperature was 182 - 185°C and the flow ratc was c. 1 I/h. Colony counts of Lancefield Group D cocci from the concentrates after incubation and from the powders were made using KF Streptococcus Agar (Difco) according to the British Standard method (Anon. 1984). From these, the death rate during spray-drying was determined. Survival of Lancefield Group D cocci during spray-drying

The survival of Lancefield GTOUp D cocci during spray-drying is shown in Table 1. The expected levels of Lanccfield Group 0 cocci in powder were derived by correcting the observed colony counts in skimmed milk concentrate for the reduction in volume due to evaporation of water. The differences between actual and expected levels in powder thus represent the death during spray-drying. Since all of the species were fairly resistant to spray-drying, powders containing high levels of contamination were prepared. These results suggest that spray-drying has little effect on Lancefield Group 0 cocci. This may be attributable to the short residence time or to the product temperature which remains relatively low due to cooling by evaporation of water. TABLE

1. Suroival of Lancejidd Group D cocci in pmvder spray-dried in the labora/ot)' Colony count during processing (loglO cfu/g or ml)* Ent. ftetalis

Skimmed milk concentrate (30% w/v total solids) Expected level in powder Level detennined in powder Net loss during spray-drying (log (.'}"des)

£11/.

faeciurn

S/rep. b(fVis

Strep. equinus

8.11

8.04

8.11

8.00

8.63

8.56

8.63

8.52

5.45

738

7.23

6.97

3.19

1.18

1.40

1.55

'*' Colony counts of Lancefield Group D coed for artificially inoculated powders.

INSTRUMENTAL DETECTION

or

ENTEROCOCCI

205

Use of Conventional Selective Media in Electrometric Instruments A variety of media formulations, containing various selective agents, has been proposed for detecting Lancefield Group 0 cocci by enrichment or colony count techniques. Broth versions of four of these media were chosen as a basis for an electrometric medium. The initial assessment and development were undertaken using the Bactomeler instrument (model MI23). The media employed were: KF Streptococcus Broth (Difco); Kanamycin AesculinAzide Broth (Oxoid); Slanetz & Bartley Medium (Oxoid); and thallous acetate broth (Barnes 1956; modified). The thallous acetate broth was modified by omitting the tetrazolium salt and agar. It contained (gil) : Proteose peptone (Difco), 10; Lab-Lemco (Oxoid), 10; glucose, 10; thallous acetate, 1.0; pH 6.0. The sterilized media were dispensed aseptically into Bactometer modules (2 ml/well) and separate wells were inoculated with actively-growing pure cultures of the Lanceficld Group D cocci. Conductance and capacitance signals were monitored during incubation at 37°C.

Quali~y

of response curves with conventional media

Since all lour species grew on KF streptococcus agar, all would be detected if present in a sample. Their responses in the electrometric medium must therefore be similar. The detection times for the same inoculum level and the magnitude and rate of conductance change should be similar for all species, otherwise only that organism which produces the strongest or fastest response may be detected. This may occur even when it is not the predominant species. In each of the four media, capacitance responses for the different species varied considerably; Ent. jaecalis produced strong capacitance curves in all formulations with 'Step' values (which indicate the scale of the .)I-axis) in the range 9-18 whilst the remaining three species produced much poorer curves. Thus, these would probably not be detected if Enl. faecalis were present in a sample. The conductance responses were generally similar lor all four species although 'Step' values were lower and in the range 3 -4. Streptococcus equinus failed to produce a conductance response in either Sianetz & Bartley medium or kanamycin-aesculin-azide broth. Thallous acetate broth showed the shortest detection times, strongest accelerations and least variations between the four species. These curves are shown in Fig. 1. It was decided that the conductance signal in thallous acetate broth was the most appropriate for further development.

206

P. NEAVES ET AL. -170 (J)

... ..... -..

Base = 1992 Step:::: 3.27

+'"

'§ 150

.....

>-

~ 130 :0

.e 110

, , .,-' t

I

90

,• ,,,

co

"5

70

,,

Q,)

g co

50

I

+oJ

U

.g c o

~

,

Q,)

g>

...

30 ~ ~........._...,--

U

2

4

6

8 10 12 14 16 18 20 22 Incubation time (h I

FIG. 1. Conductance responses for broth cultures of Streptococcus bovis (--), Stnptococcus equinus ( ), Enterucoccus fteaum (- - - -) and Enterococcus jaecalis ( ) in thallous acetate broth.

Medium Development for Electrometric Detection The effect of medium composition on conductance curve profiles was investigated for all species. Initial experiments employed double-strength media inoculated (50:50) with pure cultures in reconstituted skimmed milk. Subsequent investigations used powders containing Lancefield Group D cocci and spray-dried in the laboratory. A 10·,·,1 dilution of each powder was prepared directly in the conductance medium by adding 1 g of powder to 9 ml of broth, incubating at 45°C for 15 min and shaking vigorously to dissolve the sample. Bactometer modules were then inoculated with 2 mllwelL. Three aspects of the medium composition were investigated. Because the major conductance responses may be the result of the metabolism of nitrogen compounds, the peptone and yeast extract content of the medium was modified to promote conductance curves showing a rapid change with a large amplitude. The effect of pH on detection time was investigated because this affected the rates of detection for the different species. The concentration and types of selective agent were adjusted to maintain selectivity without destroying the curve quality. Details of the experimental procedures are shown in Table 2. Colony counts of Lancefield Group D cocci were also made on the same sample dilutions, to enable detection time and inoculum level to be compared. From the results of this series of experiments, a medium was derived which produced conductance curves of a high quality.

INSTRUMENTAL DETECTION OF ENTEROCOCCI TABLE

207

2. Concentrations 0/ companmts employed in modifying tlte conductana: medium

Component Nitrogen source Proteose peptone (Oxoid) Special peptone (Oxoid) Trypticase peptone (BBL) Skimmed milk powder Urea Yeast extract Selective agent Sodium azide ThaUous acetate pH value

Concentration range (0;;, w/v)

1-7 1-3 5

10 1

1-10 0.02-0.04 0.1-0.4 6.0-9.0

Effect of Medium Composition on Curve Quality In pure culture studies the final concentration of milk solids was 5% (w/v), the selective agent was thallous acetate (0.1 % w/v) and the pH before adding the sample was 6.0. Increasing the concentration of proteose peptone failed to affect either the gradient or the magnitude change. The addition of yeast extract did not increase the height of the curve. Substituting proteose peptone for trypticase gave poorer curves, showing increased base-line drift and a lower gradient, whilst substituting with special peptone failed to change the curve profile. In subsequent experiments, where Lancefield Group 0 cocci spray-dried in the laboratory were used, the final concentration of milk solids in the medium was 10% (w/v). This resulted in improved conductance curves for Eut. jaecium, Strep. bovis and Strep. equiuus which showed higher overall conductance changes and steeper gradients (Fig. 2), 'Step' values for these species increasing to 6. This suggested that a component of milk might be responsible for the curve quality. The lab-Iemco and glucose components of the medium were eliminated without affecting the conductance curves. The effect of pH on detection times for all species is shown in Fig. 3. All times increased with decreasing pH over the range pH 6.6-7.0. At pH 7.07.8 detection times generally remained constant except for that of Strep. bovis which showed a reduction with increasing pH value. The colony counts for powders containing Eut. jaecium, Strep. bovis and Strep. equiuus were similar whilst that of Ellt. jaecalis was lower. The colony counts correlated most

208

P. NEAVES ET AL. (i)

Base = 2537 Step = 6.16

170

+-'

'§

150

•• "." ••••••••••••••••• •••••••

--,

>

~ 130 .....

:.c

~ 110 Q,) 0)

c

90

\NKLIN, A. & I liGGINS, I.J. 1987 RapiJ methods for the dett'etion of li\ing microorganisms. Proceedings oflhe lsi IFAC Con[erClltt: Oil Ihe Modellillg alld COlltrol ofBiotedmo!ogica! Processes, Nordwikjerhollt, December 1985, ed. Johnson, A. Oxford: Pergamon Press. TURNIOI{, A.P.F., C'\RDOSI, M.F., FR·\"KI.lN, A. & R·\\IS,W, G. 1986 ,IVlediators filr bioelectrochemical cells. Brilish Palmi Applimlioll 8606831. TURNIR, A.P.F., CARDOSI, M.F., R.\\ISilY, G., SUINE1i>ER, l.UI. & S'L\IN, A. 1987 Biosensors Illr use in the fllod industry: a new rapid bioactidty monitor. In Bioledmo!(jgy ill Ihe Food I/lduslry. Pinner: Online Publications. TURNER, A.P.F., R.\\IS.'\Y, G. & IlK,GINS, I.]. 1983 Applications of electron transfer between biological systems and electrodes. Biodll'l/liw! SoddY "I'm/1sMlioIlS 11, 445~448.

Detection of Electron Transfer for the Assessment of Bacterial Contamination R. G. KROLL,l R. A. PATCHETT, I STEPHANIE E. LEAROY02 AND C. F. THURS'f01\'2 Department ofMicrobiology, AFRC Institute of Food Research, Reading Laboratory, Shinfield, Reading RG2 9AT, UK; and 2 Department ofMicrobiology, King's College, Universi~y ofLondon, Campden Hill Road, Londml W87AH, UK J

The microbial contamination of perishable items needs to be frequently and reliably assessed. Ideally, the methods used should be cheap, simple to perform and interpret, and provide the necessary information within defined limits of precision and repeatability. More important, such methods need to be rapid to allow the efficient management of food supplies. To achieve this, new methods must target some unique structural, biochemical or functional properties of micro-organisms that distinguish them from their surroundings. One possible approach is to e),:ploit the electron-carrying ability of cellular redox components. Indeed, the dye reduction tests, used for decades for assessing the hygienic quality of raw and pasteurized milk, fit this general description but they are now of little value for a variety of reasons (Luck 1982). Throughout this century, and especially Witll the oil crisis, alternative methods of producing electricity have been sought. Of particular interest has been the development of microbial fuel cells to convert organic wastes directly into electricity (Benetto 1987). Such cells were found to be efficient and capable of producing sustained currents but only if soluble mediator compounds were included (Delaney et al. 1984; Roller et al. 1984; Benetto et al. 1986). The principles of operation of microbial fuel cells is that chemoheterotrophic organisms derive their energy (and carbon source) to drive biosynthetic processes by oxidizing reduced carbon substrates. Energy is conserved by controlling the flow of electrons from these substrates to an oxidant, such as ox')'gen, and coupling the energy released to the synthesis of ATP. In microbial

Copyri/f,ht

Rapid .\licrobiological Methods telf Foods. Bc,'cragcs and Pharmaceuticals

t. irology II, 133 -135. MORRIS, R. & WAITE, W.M. 1981. Environmental virology and its problems. Journal of the Institution of Water Engineers and Scimtists 35, 232-245. MORTON, H.]. 1970. A survey of commercially available tissue culture media. In vitro 6, 89-108. SMITH, E.M. & GERllA, c.P. 1982. Development of a method of detection of human rotavirus in water and sewage. Applied and Ellvironmmtal Microbiology 43, 1440~ 1450. THOULESS, M.E., BRYDEN, A.S. & FLEWETT, T.H. 1977. Rotavirus neutralisation by human milk. British Medical Journal 2, 1390-1393. WYATT, R.G. & JAMES, W.o. 1980. Human rota,irus Type 2: cultivation ill vitro. Sdmce 207, 189-191.

Index

Acetohacter spp., 104 Acinetobacter, 237 Adenosine triphosphate, see ATP bioluminescence Aerobic plate count, 42, 90 Aeromonas spp., 128 Aeromonas hydrophila, 127 Affinity techniques for removal of bacteria from milk, 16, 19, 23 Aflasure kit, 277 Aflatest kit, 207, 277, 278 Aflatoxins, 265 assays for, 265, 267, 276-278 Amperometric biomass sensing, 214- 224 AMS image analysers, 74, 75, 77, 80 Antibody immobilization, 170, 171 APC, see Aerobic plate count Apyrase enzyme, 13, 15 ATP bioluminescence, for analysis of beer, I - 10 for analysis of milk, 13-32 reliability of, 9 Automated counting, for DEFT, 34, 52, 56 Automated electrometric methods, optimization of, 87-98

147, 149 for testing water samples, 120-125 protocol, cost for, 200 BEC, see Bioelectrochemical cell Beer, amperometric testing of, 225 detection of contaminants in, IOI-II7 PET bottled, analysis of, I-II use of DEFT with, 2 Benzalkonium chloride, as selective inhibitor,

36,37 Benzoquinone, 216 Beverages, use of DEFT on, 35 BlOCHECK, 213-225 Bioelectrochemical cell, 214 - 225 Bio-Enzabead Salmonella Screen kit, 169, 175,

179-181,249-254,262-263 Biokit, 279 Bioluminescent techniques for milk, 13-29 Breed smear, sensiti~ity of, 33 Brewery bacteria, 106-108 Brewing industry, use of electrometric techniques in, 10 1- II7 Brewing yeasts, 105, 107, 117 British Standard method, for detecting salmonellas in milk powders, 156-164

Bacillus spp., 237 Bacillus pumilus, 67 Bacteriological Analytical Manual, Food and Drug Administration, 251, 254 Bacteriological examination of drinking water supplies, 119 Bacteriuria, 59-64, 67 Bactomatic method, for TVCs of powdered products, 144, 147, 148 Bactometer, 32, 132, 134, 156, 159, 160, 168, 213 for conductimetric detection, 186-189, 197, 199, 200 for enterococci detection, 203, 205, 206, 209-211 for screening powdered products, 143, 144,

Ca Z+ sequestrant, 13 Cadmium, in Bio-Enzabeads, 180 Calcium phosphate-citrate complex, 13-26 Calibration curves, of standard and electrometric tests, 90, 91, 98, 121,

122, 144, 147, 149, 152,209,210,225 Calibration factor, in image analysis, 75 Calibration of Malthus and Bactometer instruments, 209 - 211 Carbon materials, use as an electrode, 215 Casein micelle, 13-26 Celite, use of in removal of bacteria from milk,

16, 19,27,28 Chelating agents, use of in dissociating casein micelles, 26

293

294

INDEX

Chelating systems, in removal of free ATP, 17 Chemafla kit, 277 Chicken, DEFT counts of, 53 detection of salmonellas in, 255, 256, 259 Chilli-eon-carne, detection of salmonellas in, 255 ELISA method for detection of salmonellas, 258 Chocolate, salmonellas in, 185, 187 Chromatography, for mycotoxin measurement, 266 citrate, in removal of free ATP, 17 CitroblUter spp., 128 Citroba

Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticals

Rapid Microbiological Methods in the Pharmaceutical Industry

Microbiological Methods

Compendium of the Microbiological Spoilage of Foods and Beverages (Food Microbiology and Food Safety)

Fermented Foods and Beverages of the World

Collins and Lyne's Microbiological Methods

Oxidation in foods and beverages and antioxidant applications, Volume 2

HPLC Methods for Recently Approved Pharmaceuticals

Oxidation in foods and beverages and antioxidant applications, Volume 1

Collins and Lyne's Microbiological Methods, 8th edition

Microbiological Methods for Assessing Soil Quality (Cabi Publishing)

Quality Systems and Controls for Pharmaceuticals

HPLC Method Development for Pharmaceuticals

Methods of Analysis for Functional Foods And Nutraceuticals

Methods of Analysis for Functional Foods and Nutraceuticals

Quality Systems and Controls for Pharmaceuticals

An Evaluation of the Role of Microbiological Criteria for Foods and Food Ingredients

Recipes - Hot and Cold Beverages

Beverages in Nutrition and Health

Rapid detection assays for food and water

Rapid Infectious Diseases and Tropical Medicine (Rapid)

Guide to Microbiological Control in Pharmaceuticals and Medical Devices, Second Edition

Handbook of Solubility Data for Pharmaceuticals

Radioionidation Reactions for Pharmaceuticals: Compendium for Effective Synthesis Strategies

Impurities Evaluation of Pharmaceuticals

Origin of Beverages

PCR Methods in Foods (Food Microbiology and Food Safety)

Rapid Fire

Microbiological Analysis of Food and Water

Rapid Psychiatry

Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticals

Rapid Microbiological Methods in the Pharmaceutical Industry

Microbiological Methods

Compendium of the Microbiological Spoilage of Foods and Beverages (Food Microbiology and Food Safety)

Fermented Foods and Beverages of the World

Collins and Lyne's Microbiological Methods

Oxidation in foods and beverages and antioxidant applications, Volume 2

HPLC Methods for Recently Approved Pharmaceuticals

Oxidation in foods and beverages and antioxidant applications, Volume 1

Collins and Lyne's Microbiological Methods, 8th edition

Microbiological Methods for Assessing Soil Quality (Cabi Publishing)

Quality Systems and Controls for Pharmaceuticals

HPLC Method Development for Pharmaceuticals

Methods of Analysis for Functional Foods And Nutraceuticals

Methods of Analysis for Functional Foods and Nutraceuticals

Quality Systems and Controls for Pharmaceuticals

An Evaluation of the Role of Microbiological Criteria for Foods and Food Ingredients

Recipes - Hot and Cold Beverages

Beverages in Nutrition and Health

Rapid detection assays for food and water

Rapid Infectious Diseases and Tropical Medicine (Rapid)

Guide to Microbiological Control in Pharmaceuticals and Medical Devices, Second Edition

Handbook of Solubility Data for Pharmaceuticals

Radioionidation Reactions for Pharmaceuticals: Compendium for Effective Synthesis Strategies

Impurities Evaluation of Pharmaceuticals

Origin of Beverages

PCR Methods in Foods (Food Microbiology and Food Safety)

Rapid Fire

Microbiological Analysis of Food and Water

Rapid Psychiatry

Recommend Documents