Proteins, Cells and Materials: Festschrift in Honor of the 65th Birthday of Dr. John L. Brash

Proteins, Cells and Materials VSP Utrecht, The Netherlands, 2003 Proteins, Cells and Materials Dr. John L. Brash ...

Author: Heather Sheardown | Stuart L. Cooper

277 downloads 736 Views 6MB 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

Proteins, Cells and Materials

VSP Utrecht, The Netherlands, 2003

Proteins, Cells and Materials

Dr. John L. Brash

VSP BV P.O. Box 346 3700 AH Zeist The Netherlands

©VSP BV 2003

First published in 2003

ISBN 90-6764-381-5

All 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 prior permission of the copyright owner.

Printed in The Netherlands, on acid-free paper, by Ridderprint BV, Ridderkerk.

CONTENTS Foreword

ix

Introduction

xi

Letter

xiii

I. Proteins PEO-like plasma polymerized tetraglyme surface interactions with leukocytes and proteins: in vitro and in vivo studies M. Shen, L. Martinson, M.S. Wagner, D.G. Castner, B.D. Ratner and T. A. Horbett

3

Limits of detection for time of flight secondary ion mass spectrometry (ToF-SIMS) and X-ray photoelectron spectroscopy (XPS): detection of low amounts of adsorbed protein M.S. Wagner, S.L. McArthur, M. Shen, T.A. Horbett and D.G. Castner

27

Photoimmobilization of biomolecules within a 3-dimensional hydrogel matrix X. Cao and M.S. Shoichet

49

Fibrinogen adsorption by PS latex particles coated with various amounts of a PEO/PPO/ PEO triblock copolymer M. Bohner, T.A. Ring, N. Rapoport and K.D. Caldwell

63

Effects of base material, plasma proteins and FGF2 on endothelial cell adhesion and growth P.A. Underwood, J.M. Whitelock, P.A. Bean and J.G. Steele

77

Acoustics of blood plasma on solid surfaces M. Andersson, A. Sellborn, C. Fant, C. Gretzer and H. Elwing

95

Development of small alginate microcapsules for recombinant gene product delivery to the rodent brain C.J.D. Ross and P.L. Chang

107

II. Cells Integrin α2β1 on rat myeloma cells modulates interaction of α4β1 integrin with vascular cell adhesion molecule-1 but not fibronectin B.M.C. Chan, V.L. Morris, D. Hangan-Steinman, B. Jarvie, M. Cialacu, J. Laansoo, G. Hunter, W. Wan and S. Uniyal

119

vi

Contents

Interactions of corneal epithelial cells and surfaces modified with cell adhesion peptide combinations L. Aucoin, C.M. Griffith, G. Pleizier, Y. Deslandes and H. Sheardown

137

Polyelectrolyte multilayer films modulate cytoskeletal organization in chondrosarcoma cells D. Vautier, V. Karsten, C. Egles, J. Chluba, P. Schaaf , J.-C. Voegel and J. Ogier

153

Improved blood compatibility and decreased VSMC proliferation of surface-modified metal grafted with sulfonated PEG or heparin H.J. Lee, J.-K. Hong, H.C. Goo, W.K. Lee, K.D. Park, S.H. Kim, Y.M. Yoo and Y.H. Kim 173 Characterization of poly(ethylene oxide) brushes on glass surfaces and adhesion of Staphylococcus epidermidis H.J. Kaper, H.J. Busscher and W. Norde

187

Tissue-culture surfaces with mixtures of aminated and fluorinated functional groups. Part 2. Growth and function of transgenic rat insulinoma cells (βG I /17) J.R. Bain and A.S. Hoffman

199

III. Materials Elastomeric biodegradable polyurethane blends for soft tissue applications J.D. Fromstein and K.A. Woodhouse

229

Influence of surface morphology and chemistry on the enzyme catalyzed biodegradation of polycarbonate-urethanes Y.W. Tang, R.S. Labow, I. Revenko and J.P. Santerre

245

Cardiopulmonary bypass technology transfer: musings of a cardiac surgeon F.D. Rubens

267

Surface analysis methods for characterizing polymeric biomaterials K. Merrett, R.M. Cornelius, W.G. McClung, L.D. Unsworth and H. Sheardown

283

Glucose binding to molecularly imprinted polymers H. Seong , H.-B. Lee and K. Park

313

The effect of oxidation on the enzyme-catalyzed hydrolytic biodegradation of poly(urethane)s R.S. Labow, Y. Tang, C.B. McCloskey and J.P. Santerre

327

Novel dendrimer based polyurethanes for PEO incorporation X. Duan, C.M. Griffith, M.A. Dubé and H. Sheardown

343

Identification of biodegradation products formed by L-phenylalanine based segmented polyurethaneureas S.L. Elliott, J.D. Fromstein, J.P. Santerre and K.A. Woodhouse

367

Contents

vii

Bioresorbable polymeric stents: current status and future promise R.C. Eberhart, S.-H. Su, K.T. Nguyen, M. Zilberman, L. Tang, K.D. Nelson and P. Frenkel

389

Tissue-culture surfaces with mixtures of aminated and fluorinated functional groups. Part 1. Synthesis and characterization J.R. Bain and A.S. Hoffman

403

Deterioration of polyamino acid-coated alginate microcapsules in vivo J.M. van Raamsdonk, R.M. Cornelius, J.L. Brash and P.L. Chang

419

Water structure around enkephalin near a GeO2 surface: a molecular dynamics study A.M. Bujnowski and W.G. Pitt

441

Towards practical soft X-ray spectromicroscopy of biomaterials A.P. Hitchcock, C. Morin, Y.M. Heng, R.M. Cornelius and J.L. Brash

463

A new vascular polyester prosthesis impregnated with cross-linked dextran D. Machy, P. Carteron and J. Jozefonvicz

483

This page intentionally left blank

FOREWORD

This Festschrift celebrates the 65th birthday of John L. Brash. The field of biomaterials owes much to the contributions that he has made over a 40 year career that includes research done primarily at McMaster University, but also at E. I. DuPont de Nemours and Company and Stanford Research Institute. John has had a broad influence internationally as his work on cardiovascular biomaterials and plasma protein adsorption has received worldwide recognition. The editors, publisher and the many colleagues who are participating in this multi-issue Festschrift wish him well on this special occasion. We anticipate John Brash maintaining a leadership role in biomaterials research and we look forward to his continuing contributions to the field. These articles were first published in 2002– 2003 in several issues of Journal of Biomaterials Science, Polymer Edition. We are most pleased now to make the entire Festschrift available in book form. M. VERT T. TSURUTA S. L. COOPER


Introduction It is with a great deal of pleasure that I write an introduction in this special issue dedicated to Dr. John Brash on the occasion of his 65th birthday. To say that John’s work over the years has advanced the field of polymeric biomaterials would be an understatement. His sphere of influence and the respect that people have for him and for his work is global. He has always been willing to extend his expertise and to open his facilities to others and has had numerous collaborations over his career. Former students and post doctoral fellows have established strong research programs in academia and made significant industrial contributions. He is on editorial boards for a number of scientific journals, is the journal editor for Colloids and Surface B: Biointerfaces, and through membership on grants and other committees, has helped to bring biomaterials research to the forefront in Canada. To date, John has supervised 13 doctoral students and 18 master’s students, many of whom are represented in this journal. He has published more than 100 papers in refereed journals and is co-editor of two books on the behaviour of proteins at interfaces. His influence in the biomaterials community has been recognized by the Clemson Award for Basic Research from the US Biomaterials Society in 1994, an honorary degree, Docteur Honoris Causa, from the Univeristé Paris Nord in 1996 and by an appointment as University Professor, McMaster University’s highest research honour, in 2001. John received his BSc (1958) and PhD (1961) in the area of kinetics of free radical polymerization from the University of Glasgow. He spent two years as a postdoctoral fellow at the National Research Council of Canada, where he worked on photochemistry and chemical kinetics, and one year working with E. I. Du Pont de Nemours and Company. In 1964, he joined the Stanford Research Institute where he began working on a program on artificial organ research, which included the development of new membranes for hemodialysis and investigations of the mechanism of surface induced thrombogenesis. The former was supported by the Artificial Kidney Program of the NIH and the latter by the Artificial Heart Program, National Heart and Lung Institute. In subsequent work for the Medical Devices Applications Program, John directed the development of a system of biocompatible polyether urethane elastomers for use in circulatory assist devices. Fabrication technology for balloons and cups was also worked out. Another phase of the project was concerned with sulfonated polymers of controlled content and distribution of functional groups.

xii

Introduction

In 1972, John joined the faculty of McMaster University with a joint appointment in Chemical Engineering and Pathology, with teaching responsibilities in Chemical Engineering and a research program in biomedical engineering that bridges the two disciplines. At McMaster, John has conducted research in the field of polymeric biomaterials with an emphasis on cardiovascular implant materials. Much of his efforts have focused on detailed studies of blood-surface interactions, particularly those of plasma proteins. His multifaceted approach recognizes the importance of hemodynamics, transport and surface phenomena in the gross effects that result from blood surface contact. He has spent sabbaticals at the Centre de Recherches sur les Macromolécules, Centre Nationale de la Recherche Scientifique in Strasbourg France (1978– 1979), the University of Linkoping, Linkoping Sweden (1994– 1995?) and at CSIRO in Melbourne and Sydney Australia in 2001. In a communication to me, Allan Hoffman’s comment was that John is too young to retire. This echoes the sentiments of many of his colleagues and collaborators. I hope he does not really retire for a very long time to come. HEATHER SHEARDOWN

Letter Dear John: It is now ten years ago since you contributed a thirteen-page article to my Festschrift, so it is about time for me to contribute at least one letter to yours. In that article you introduced flow, and since then I spent the rest of my natural research life trying to do the same, with Ed Leonard and others. Now, more or less but not entirely away from it all, I sometimes wonder how much in this area we have contributed to the welfare of humanity, and how much humanity would care. I think I earned about one million dollars in my entire lifetime studying interface reactions, so I like to imagine myself going from door to door over the entire United States, telling people that for two cents I can tell them what happens to their fibrinogen when their blood touches glass. I would never have collected the needed funds, so that I would never have been able to tell even myself what happens. Obviously therefore, I am grateful to our government for the support that allowed me to enjoy myself all those years, but especially for the occasion to meet and work with you. Nobody else I know works with that precious combination of intensity and relaxation we should all work with. To me, one of the most enjoyable episodes was our study of ring formation by IgG between a lens and a slide. I had already published this, and asked you to try it out. You never succeeded. The problem appeared not to be the source or preparation of IgG or of the substrate, or you. When you visited us at Columbia I was — thank heavens! — able to demonstrate the phenomenon for you, you then took some of our IgG and I think an anodized tantalum coated slide as well with you to Toronto, and presto! it still did not work when you tried it in your lab. The whole thing would have been frustrating if experienced with almost anybody else, but for me, with you, it was just fun. Thank you, thank you. I was happy to see you have broadened your field of interest. In the name of that same humanity, I thank you. And I hope we will meet more often! Warmest regards, also to all those around you, LEO VROMAN Fort Worth, Texas, 7/4/01


Part I

Proteins


PEO-like plasma polymerized tetraglyme surface interactions with leukocytes and proteins: in vitro and in vivo studies MINGCHAO SHEN 1 , LAURA MARTINSON 1 , MATTHEW S. WAGNER 2 , DAVID G. CASTNER 1,2 , BUDDY D. RATNER 1,2 and THOMAS A. HORBETT 1,2,∗ 1 Department 2 Department

of Bioengineering, University of Washington, Seattle, WA 98195, USA of Chemical Engineering, University of Washington, Seattle, WA 98195, USA

Received 5 December 2001; accepted 18 March 2002 Abstract—Polyethylene oxide (PEO) surfaces reduce non-specific protein and cell interactions with implanted biomaterials and may improve their biocompatibility. PEO-like polymerized tetraglyme surfaces were made by glow discharge plasma deposition onto fluorinated ethylene propylene copolymer (FEP) substrates and were shown to adsorb less than 10 ng/ cm2 of fibrinogen in vitro. The ability of the polymerized tetraglyme surfaces to resist leukocyte adhesion was studied in vitro and in vivo. Polymerized tetraglyme and FEP were implanted subcutaneously in mice and removed after 1 day or 4 weeks. Histological analysis showed a similar degree of fibrous encapsulation around all of the 4-week implants. Darkly stained wells were present in the fibrous tissues at the tissue-material interface of both FEP and tetraglyme. Scanning electron micrographs showed that in vivo macrophage adhesion to polymerized tetraglyme was much higher than to FEP. After 2-hour contact with heparinized whole blood, polymorphonuclear leukocyte (PMN) adhesion to polymerized tetraglyme was much higher than to FEP, while platelet adhesion to polymerized tetraglyme was lower than to FEP. When PMNs isolated from blood were suspended in 10% autologous plasma, cell adhesion to polymerized tetraglyme was higher than to FEP; however when the cells were suspended in heat inactivated serum, cell adhesion to FEP was higher than to polymerized tetraglyme. The surface chemistry of polymerized tetraglyme did not change after 2-hour blood contact, but displayed nitrogen functional groups after 1-day implantation and became slightly degraded after 4-week implantation. The surface chemistry of FEP did not change significantly after blood contact or implantation. Loosely bound proteins such as fibrinogen on polymerized tetraglyme may contribute to the adhesion of PMNs and macrophages and ultimately to fibrous encapsulation (the foreign body response) around the implants. Key words: PEO; RF plasma deposition; surface modification; non-fouling; foreign body response; macrophage; protein adsorption. ∗ To

whom correspondence should be addressed. E-mail: [email protected]

4

M. Shen et al.

INTRODUCTION

Macrophages are considered the central mediators of the foreign body response to implanted biomaterials [1, 2]. Macrophages are found adherent to the surface of many implants [3, 4], where they can undergo cytokine release [5] and fusion to form foreign body giant cells [6]. The characteristic foreign body capsule that forms around implants is thought to be triggered by macrophage adhesion and activation [2]. Polymorphonuclear leukocytes (PMNs) may also be involved in the foreign body response since they mediate acute inflammatory response to injuries and can also adhere to implanted biomaterial surfaces [7, 8]. After contacting synthetic materials such as polyurethane and Dacron, PMNs can secrete reactive oxygen species such as superoxide ions that can induce surface degradation [9, 10]. Due to the highly adhesive nature of macrophages [11] and PMNs [12], it has been difficult to engineer biomaterial surfaces that can inhibit cellular attachment, fibrous encapsulation, and the foreign body response to implanted materials. Cell adhesion proteins that adsorb non-specifically to implanted biomaterials are believed to mediate leukocyte adhesion to surfaces [2, 11, 13]. Polyethylene oxide (PEO)-like surface coatings have been considered promising to prevent nonspecific protein adsorption and cell adhesion to biomaterial surfaces [14, 15]. Although some degree of reduced cell adhesion has been shown on PEO-containing surfaces [16, 17], others did not achieve decreases in leukocyte adhesion [18, 19], probably because protein adsorption to those surfaces was not low enough. Several in vivo studies on PEO-containing surfaces fail to demonstrate reduced cell-surface interactions. For example, in a cage implant system, macrophage cell density on triblock copolymers containing increasing amounts of PEO was much higher than the control surface [20]. Polyurethane polymers containing PEO units did not reduce cell adhesion when implanted intramuscularly in rats [5]. On poly(propylene fumarate-co-ethylene glycol) hydrogels, macrophage density was higher on surfaces with greater PEO content, and there was much higher neutrophil adhesion to PEO-containing materials than the poly(propylene fumarate) control [21]. Truly non-fouling surfaces that completely inhibit protein adsorption are needed to control biological interactions with biomaterials [22]. For example, in previous studies in our lab, we found that platelet adhesion still occurred on surfaces with only 10 ng/ cm2 of adsorbed fibrinogen [23]. We have also applied this criteria to monocytes and shown that greatly reduced short term monocyte adhesion in vitro on a PEO-like glow discharge plasma polymerized tetraglyme surface was only obtained when the surfaces reduced protein adsorption to less than 10 ng/ cm2 [24]. In this study we investigated whether such highly non-fouling plasma polymerized tetraglyme surfaces would resist protein adsorption and leukocyte adhesion in vivo. Cell adhesion to and fibrous encapsulation of subcutaneously implanted materials were measured. Cell adhesion to polymerized tetraglyme surfaces from whole blood and from washed PMNs in plasma or serum was also evaluated. In addition, surface analysis of the explanted materials was performed to determine if surface degradation contributed to the fouling of biomaterial surfaces.

PEO-like plasma polymerized tetraglyme surface interactions

5

MATERIALS AND METHODS

Buffers and reagents Phosphate buffered saline (PBS), May-Grünwald stain, Giemsa stain, Triton X-100, Histopaque 1077, heat-inactivated fetal bovine serum, trypan blue, RPMI 1640, and dextran were purchased from Sigma (St. Louis, MO). Glutaraldehyde (3%) solution for scanning electron microscopy (SEM) contained 0.1 M cacodylate, pH = 7.4. Tetraglyme (CH3 O(CH2 CH2 O)4 CH3 ) was purchased from Aldrich (Milwaukee, WI). Surfaces The FEP film, tetrafluoroethylene-hexafluoropropylene-copolymer (CF(CF3 )CF2 (CF2 -CF2 )n )m , was a gift from E. I. Du Pont de Nemours & Co., Inc. (Circleville, OH). The film was cut into 6.4 mm diameter disks and cleaned by successive 10-min ultra-sonications in methylene chloride (× 2), acetone (×2) and methanol (×2). Radio-frequency glow discharge (RFGD) plasma deposition of tetraglyme on FEP was done as previously described [24]. Tetraglyme was plasma polymerized on FEP for 1 min at 80 W then 10 min at 10 W. The polymerized tetraglyme surfaces were found to inhibit protein adsorption and monocyte adhesion [24]. Fibrinogen adsorption to polymerized tetraglyme from a 0.03 mg/ml fibrinogen solution was less than 10 ng/ cm2 . Prior to implantation, FEP and plasma deposited tetraglyme samples were soaked in 70% ethanol overnight, rinsed in sterile PBS, and screened for endotoxin contamination using a Pyrotell® Limulus Amebocyte Lysate (LAL) assay kit (Associates of Cape Cod Inc., Falmouth, MA) sensitive to 0.06 endotoxin Unit (EU)/ml. The samples were found to contain less than 0.06 EU /ml of endotoxin. Implantation All surgical instruments were cleaned and autoclaved prior to surgery and soaked in 70% ethanol between animal surgeries. Strict aseptic techniques were used for material implantation. Healthy, 6-week old C57Bl6 male mice (B&K Universal, Kent, WA) were used. The animals were anesthetized with isoflurane gas (Abbott Laboratories, North Chicago, IL) for the 4-week implantation study, or with a cocktail of ketamine (Fort Dodge Animal Health, Fort Dodge, IA) and xylazine (Phoenix Pharmaceutical, Inc., St Joseph, MO) for the 1-day implantation study. The 6.4 mm samples were surgically implanted subcutaneously on the backs of mice, each mouse implanted with two different materials, polymerized tetraglyme and FEP (one disk of each). The incision site was prepared by shaving and swabbing with Betadine. A single 1– 1.5 cm incision was made midline on the back of a mouse and two subcutaneous pockets were created by blunt dissection lateral to each side of the incision. One implant was placed in each pocket, and the incision was closed with sterile wound

6

M. Shen et al.

clips. Animals were allowed to recover prior to returning to housing cages. Animals were given food and water ad lib for the remainder of the study. Explantation After 1 day or 4 weeks, animals used for subcutaneous implant studies were sacrificed by CO2 asphyxiation. The wound clips were removed, and animal skin was cut open to access the implants. Fibrous capsules surrounded all of the 4-week implants. For histological analysis, the 4-week samples were retrieved en-block so as not to disturb the biomaterial/ host tissue fibrous capsule interface. The intact tissues were fixed in methyl Carnoy’s for 2 days at 4◦ C. The fixed tissues were prepared using standard paraffin-embedding and sectioning techniques and stained with hematoxylin and eosin (H&E) or Masson’s Trichrome. The cross section of the tissue-implant interface was viewed with an upright Nikon light microscope. For each cross section, the thickness of the fibrous capsule on both the skin and muscle sides of the implants was measured at 4 locations (1 mm apart) near the midportion of the material and the average capsule thickness was calculated for each sample. To determine the surface chemistry of implanted samples and adherent cell type and number, additional FEP and polymerized tetraglyme samples were implanted and removed after 1 day or 4 weeks. During explantation, the fibrous capsule tissues around the implants were nicked open with the tip of forceps to expose the edges of the samples. The implants were not integrated with the surrounding fibrous tissues and were easily pulled out of the capsule by the forceps. The implants were kept in sterile PBS until further analysis of the morphology of the adherent cells and material surface chemistry as described below. Cell adhesion from whole blood Whole blood from healthy human donors was drawn by venipucture into Vacutainers containing sodium heparin (Becton Dickinson, Franklin Lakes, NJ). Polymerized tetraglyme samples made at 10, 20, or 60 W of plasma power and untreated FEP samples were incubated in whole blood in a 48-well polystyrene plate (Corning Inc., Corning, NY) for 2 h at 37◦ C. The samples were dip-rinsed successively in three petri dishes containing PBS to wash away bulk blood cells. No red blood cells were visible on the samples after rinsing. The adherent cell number was characterized by measuring total lactate dehydrogenase (LDH) activity associated with the samples. The LDH method was previously used to measure platelet and leukocyte adhesion to surfaces [11, 23, 24]. Adherent cells were lysed with 1% Triton X-100 for 20 min and the total LDH activity was measured with a cytotoxicity kit (Roche, Indianapolis, IN) at optical density 490 nm with reference optical density at 650 nm [11]. The morphology of the adherent cells and material surface chemistry were also analyzed as described below.


7

PMN adhesion Whole blood from healthy human donors was drawn by venipuncture into Vacutainers containing EDTA (Becton Dickinson). Human PMNs were isolated according to a published method [25]. The cells were isolated with the following steps: density gradient centrifugation over Histopaque 1077, sedimentation in dextran, and hypotonic cell lysis of residual red cells. The cells were >95% viable as determined by Trypan blue exclusion. The isolated cells were suspended at 2 × 106 /ml in RPMI 1640 media containing 10% FBS or autologous plasma. Autologous plasma was prepared from heparinized blood from the same blood donor. PMN suspensions of 400 µl were added to the 48-well plate holding FEP or polymerized tetraglyme samples and incubated at 37◦ C in humidified 5% CO2 atmosphere for 1 h. The surfaces were washed by filling and aspirating the wells 3 times with RPMI 1640. Adherent cell number was determined with the LDH assay using a calibration curve based on PMNs suspended in PBS. Cellular image analysis The samples from the implant and whole blood studies were analyzed with microscopy to determine the type and amount of cells adherent on the surfaces. For SEM, the samples were dip-rinsed in PBS three times and fixed with glutaraldehyde for 24 h at 4◦ C. The samples were rinsed in deionized water three times, dehydrated in a graded series of ethanol-water solutions, critical point dried, sputter coated with gold-palladium, and visualized with a JEOL JSM-6300F scanning electron microscope. The acceleration voltage was 15 keV. For light microscopy, the samples were placed in a 24-well plate, fixed for 6 min with methanol and air-dried. Fixed cells were stained with May-Grünwald for 1 min, rinsed with PBS twice, stained with Giemsa for 5 min, and rinsed with deionized water twice. The samples were dried overnight and imaged with an inverted Nikon light microscope. Surface chemistry analysis Surface analysis was done at the National ESCA and Surface Analysis Center for Biomedical Problems (NESAC /BIO) at the University of Washington. The implants and the samples exposed to whole blood were dip-rinsed in PBS and sonicated for 20 min in PBS containing 1% Triton X-100, except that the 1-day implanted samples were exposed to Triton X-100 without sonication. The samples were further washed 3 times in deionized water and air-dried before being analyzed with ESCA and ToF-SIMS. Control FEP and polymerized tetraglyme samples were incubated with 0.03 mg/ ml of fibrinogen for 2 h at 37◦ C, rinsed with PBS and deionized water, and air-dried before surface analysis. ESCA analysis was done with an SSX-100 surface analysis instrument (Surface Science Instruments or SSI, Mountain View, CA) using a monochromatized AlKα

8

M. Shen et al.

X-ray source and an electron flood gun for charge neutralization. The photoelectron take-off angle for the analysis was 55◦ . The hydrocarbon peak for the ESCA spectra acquired in this study was referenced to a binding energy 285.0 eV. Surface elemental compositions and high resolution carbon functional group compositions were calculated using SSI software. For ToF-SIMS analysis, positive ion ToF-SIMS spectra were acquired using a PHI Model 7200 reflectron time of flight secondary ion mass Spectrometer (Physical Electronics, Eden Prairie, MN) with an 8 keV Cs+ primary ion source. The primary ion dose was maintained below 1012 ions/ cm2 to insure that all spectra were acquired under static SIMS conditions [26]. Spectra were acquired at three different spots for each surface. Spectra were acquired over a horizontal raster size + + of 200 × 200 µm. The positive spectra were calibrated to the CH+ 3 , C2 H3 , C3 H5 , + C7 H7 peaks. A pulsed electron flood gun was used for charge neutralization for all samples. The major peaks associated with polymerized tetraglyme and FEP were identified by their m/z ratios. Statistical analysis To determine the significance of the data, unpaired F -tests and 2-tail t-tests were performed with Microsoft® Excel 98 (Microsoft Corp., Redmond, WA). The level of significance was set at p < 0.05.

RESULTS

Histological tissue analysis Figure 1 shows the cross section view of the implanted materials and the surrounding fibrous tissues. The thickness of the implants is indicated with 2-headed arrows. Sample preparation for histology caused an artificial separation (the open space) between the tissues and the implanted material surface. A thin, dense layer of fibrous/ granulous tissues was observed around both FEP (Fig. 1a) and polymerized tetraglyme (Fig. 1b). The fibrous tissues ran parallel to the smooth implant surfaces. The implants were not integrated into the surrounding connective tissue. No new blood vessel growth from the fibrous tissues towards the implants was observed. For FEP, darkly stained cells were present at the material-tissue interface but not directly on the implant surface (Fig. 1a). For polymerized tetraglyme, darkly stained adherent cells were observed on one side of the tetraglyme sample (Fig. 1b). There were also stained cells within the fibrous tissues. Next to the thin fibrous tissue and away from the implants was more loosely arranged connective tissue. On the skin side, hair follicles were scattered among the collagen fibers. On the muscle side, adipose tissues were occasionally present. The collagenous fibrous capsules around implanted FEP and polymerized tetraglyme were about 30– 40 µm in thickness and were not significantly different from


Figure 1. Collagen tissues and cells around implanted (a) FEP (b) plasma polymerized tetraglyme shown by Masson’s Trichrome staining. The tissues on the top side of FEP and the bottom side of polymerized tetraglyme are the skin side tissues. The hair follicles are not shown. The thickness of the implants is indicated by the arrows. The spaces between the implants and tissues are artifacts from histology preparations. Scale bars represent 100 µm.

9

10

M. Shen et al.

Figure 2. The thickness of fibrous tissue capsules around 4-week implanted FEP (white bars) and polymerized tetraglyme (black bars) was measured with light microscopy. The capsule thickness from the skin and the muscle side was measured. The data represent mean ± S.D.; n = 4 FEP or polymerized tetraglyme samples that were implanted in 4 mice. No differences in capsule thickness were found.

Figure 3. Representative scanning electron micrographs of adherent cells on FEP and polymerized tetraglyme surfaces: Adherent cells on FEP (a) or polymerized tetraglyme (b) after 1-day subcutaneous implantation in mice. Adherent cells on FEP (c) or polymerized tetraglyme (d) after 4-week subcutaneous implantation in mice. Enlarged micrograph of adherent cells on polymerized tetraglyme (e) after 4-week subcutaneous implantation. Adherent cells on FEP (f) or polymerized tetraglyme (g) after 2-hour blood contact. Scale bars represent 10 µm in a, b, e, f, g, and 100 µm in c, d.


11

each other (Fig. 2). The fibrous capsules displayed no significant difference in thickness on either the skin or the muscle side of the implants. Leukocyte adhesion in vivo Cell adhesion to FEP and polymerized tetraglyme samples was observed with SEM (Fig. 3a– g) and light microscopy (Fig. 4a– f). Cell adhesion to 1-day subcutaneously implanted samples was lower on FEP than on polymerized tetraglyme, as shown by SEM (Fig. 3a and b) and light microscopy (Fig. 4a and b). The adherent cells on polymerized tetraglyme appeared to be mainly PMNs, indicated by stained cells that displayed multi-lobular polymorphic cell nuclei. No platelets or red blood cells were observed on either surface. Cell adhesion to 4-week implanted samples was also much higher on tetraglyme than on FEP (Fig. 3c and d). In fact, there were hardly any adherent cells on the FEP surface. The adherent cells on polymerized tetraglyme appeared to be mainly macrophages, as the cells were about 15– 20 µm in diameter, well spread, and displayed ruffled cell membrane characteristic of macrophages (Fig. 3e). No foreign body giant cells were observed on the surfaces. Leukocyte adhesion in vitro After contact with whole blood, the adherent cells on FEP were mainly platelets, while the adherent cells on polymerized tetraglyme were mainly PMNs. The

Figure 3. (Continued).

12

M. Shen et al.

Figure 4. Representative light microscopy images of Giemsa-May-Grunwald stained FEP and polymerized tetraglyme samples: Adherent cells on FEP (a) or tetraglyme (b) after 1-day subcutaneous implantation in mice. Adherent cells on FEP (c) or polymerized tetraglyme (d) after 2-hour blood contact. Adherent PMNs on FEP (e) or polymerized tetraglyme (f) after 1-hour adhesion from 10% autologous plasma. Scale bars represent 100 µm.


13

Figure 5. Cell adhesion to FEP or polymerized tetraglyme surfaces after 2-hour heparinized blood contact. Adherent cells were measured by the LDH method. The data represent mean ± S.D.; n = 4 samples. The LDH activity on 10 W or 60 W polymerized tetraglyme samples was significantly higher than that on FEP control (∗ p < 0.01). The LDH activity on 20 W polymerized tetraglyme was significantly higher than that on 10 W tetraglyme (∗∗ p < 0.01).

adherent cells on FEP were flat (Fig. 3f) compared to the cells on polymerized tetraglyme, which were fully spread and showed ruffled cell membrane (Fig. 3g). Leukocyte adhesion from whole blood to FEP was much lower than to polymerized tetraglyme (Fig. 4c and d). PMNs appeared to be the main type of cells adherent on polymerized tetraglyme, judging by the multi-lobular polymorphic nuclei staining and cell size of about 10 µm diameter (Fig. 4d). The adherent cells on FEP were much smaller and were mainly platelets. No platelets were found on polymerized tetraglyme. No red blood cells were observed on either surface. The LDH method was used to measure total adherent cell LDH content on FEP and three types of polymerized tetraglyme surfaces deposited at different powers. The amount of fibrinogen adsorbed from 0.03 mg/ml of fibrinogen to polymerized tetraglyme samples deposited at plasma powers of 10, 20 or 60 W was 3.4, 17, and 209 ng/ cm2 respectively. Total LDH activity was the lowest on FEP, and increased in order of 60 W, 10 W, and 20 W polymerized tetraglyme samples (Fig. 5). Due to the presence of various cell types, the LDH activity only approximated the total cell number on the surfaces. Since the LDH content in platelets is much lower than in leukocytes, the LDH data agreed with observations above that more leukocytes adhered to polymerized tetraglyme than to FEP.

14

M. Shen et al.

Figure 6. Adhesion of human PMNs in 10% heat-inactivated FBS to FEP (white bars) or plasma polymerized tetraglyme (black bars) after 1 hour. Cell adhesion was measured by the LDH method. The data represent mean ± S.D.; n = 5 samples.

PMN adhesion from washed cell preparations depended on the type of proteins in the media. When 10% autologous plasma was used in cell media, adhesion after 1-hour was higher to polymerized tetraglyme than to FEP (Fig. 4e and f). The corresponding LDH values that correlated to cell number on FEP and polymerized tetraglyme were 0.71 and 1.29 respectively. However, when the cells were in 10% heat inactivated FBS media, adhesion after 1-hour was slightly higher to FEP than to polymerized tetraglyme with (p = 0.067, 2-tail t-test) or without (p = 0.276, 2-tail t-test) preadsorbed fibrinogen (Fig. 6). However, these differences were not statistically significant. ESCA analysis FEP and polymerized tetraglyme surfaces were characterized by ESCA (Table 1a– b and Fig. 7a– d). FEP contained 34.5% carbon and 65.5% fluorine (Table 1a) and displayed fluorinated carbon groups [24]. Polymerized tetraglyme contained 70.5% carbon and 29.4% oxygen (Tables 1b) and displayed mainly ether carbon [24]. After 2-hour adsorption with 0.03 mg/ ml of fibrinogen, FEP displayed 10.9% nitrogen and 13.7% oxygen (Table 1a), showing that a large amount of proteins adsorbed to FEP. Polymerized tetraglyme resisted fibrinogen adsorption, as the surface chemistry remained the same after 2-hour exposure to the fibrinogen solution (Table 1b). After 2-hour contact with whole blood, sonication in Triton X-100 and rinsing, the surface chemistry of both FEP and polymerized tetraglyme was similar to untreated samples (Tables 1a and b). Very small amounts of fluorine


15

Table 1. ESCA elemental compositions of FEP (a) and polymerized tetraglyme (b). The control samples were not treated. The samples adsorbed with fibrinogen were rinsed with PBS and deionized water. The samples implanted or exposed to blood were sonicated in Triton X-100 and rinsed in deionized water, except that the 1-day implant samples were not sonicated when lysed with Triton X-100. The data represent mean ± SD; n = 4 samples except polymerized tetraglyme adsorbed with fibrinogen Materials

%C

%O

%F

%N

(a) FEP control 2-hr in fibrinogen 2-hr in blood 1-day implant 4-week implant

34.5 ± 0.5 57.4 ± 1.3 33.1 ± 0.1 40.3 ± 0.9 32.8 ± 0.3

nd 13.7 ± 0.5 nd 0.5 ± 0.9 nd

65.5 ± 0.6 18.0 ± 1.9 66.9 ± 0.1 58.7 ± 1.6 67.2 ± 0.3

nd 10.9 ± 0.2 nd 0.5 ± 0.1 nd

(b) Polymerized tetraglyme control 2-hr in fibrinogen 2-hr in blood 1-day implant 4-week implant

70.5 ± 0.9 70.5 70.1 ± 1.4 70.4 ± 2.1 64.1 ± 2.8

29.4 ± 0.9 29.5 28.5 ± 0.5 24.8 ± 3.9 24.7 ± 3.6

nd nd 0.8 ± 1.2 0.6 ± 1.2 5.0 ± 5.2

nd nd 0.6 ± 0.7 4.1 ± 3.4 6.3 ± 2.8

Note: nd means that this element was not detected during analysis.

and nitrogen were detected on polymerized tetraglyme (Table 1b). The ESCA C1s spectrum of blood-exposed polymerized tetraglyme surfaces was similar to untreated polymerized tetraglyme surfaces (Fig. 7a). After 1-day subcutaneous implantation of FEP and Triton X-100 treatment to remove any adherent cells, the carbon content increased and fluorine content decreased (Table 1a). There was only a slight increase of oxygen nitrogen on FEP. Compared to untreated FEP, the hydrocarbon peak in the ESCA C1s spectrum of 1-day implanted FEP increased from 0 to 25% (Fig. 7b). Implanted polymerized tetraglyme displayed a significant increase of nitrogen content on the surface (Table 1b), indicating adsorption of proteins to the polymerized tetraglyme surface. There was a small fluorine signal from the polymerized tetraglyme surface, similar to that of blood-contacting polymerized tetraglyme. The oxygen content of polymerized tetraglyme decreased from 29.4% to 24.8%. Compared to untreated polymerized tetraglyme that normally contained 65– 75% of ethercarbon content at 286.5 eV binding energy [24], the polymerized tetraglyme implanted for one day contained about 40% ethercarbon (Fig. 7c). After 4-week subcutaneous implantation of FEP followed by sonication and Triton X-100 lysis, the carbon and fluorine content remained the same as untreated FEP (Table 1a). There was no detectable nitrogen or oxygen on the implanted FEP. A small hydrocarbon peak was observed on implanted FEP (Fig. 7d). After 4 weeks of implantation, there was a significant increase of nitrogen and fluorine content on polymerized tetraglyme (Table 1b). The presence of 6.3% nitrogen on

16

M. Shen et al.

(a)

(b) Figure 7. The resolved ESCA C1s spectra of polymerized tetraglyme after 2-hour contact with whole blood (a), FEP after 1-day implantation (b), polymerized tetraglyme after 1-day or 4-week implantation (c), and FEP after 4-week implantation (d). The samples implanted or exposed to blood were sonicated in Triton X-100 and rinsed in deionized water, except that the 1-day implant samples were not sonicated when lysed with Triton X-100.

the polymerized tetraglyme surface indicated adsorption of proteins. The fluorine signal suggested slight degradation of the polymerized tetraglyme coating on FEP substrate, although the amount of fluorine was low compared to control FEP.


17

(c)

(d) Figure 7. (Continued).

The polymerized tetraglyme oxygen content decreased from 29.4% to 24.7% after implantation. The ethercarbon content was about 40%, similar to 1-day implanted samples (Fig. 7c).

18

M. Shen et al.

ToF-SIMS analysis The ToF-SIMS spectra of polymerized tetraglyme and FEP (Fig. 8a– d) contained a large number of peaks. The major peaks associated with polymerized tetraglyme and FEP were identified by their m/z ratios and are listed in Table 2a– b. Overall, the peaks from whole blood exposed or implanted polymerized tetraglyme include

(a)

(b) Figure 8. ToF-SIMS positive ion spectra: FEP (a) and polymerized tetraglyme (b) after whole blood contact for 2 h, 1-day implanted polymerized tetraglyme (c), and 4-week implanted polymerized tetraglyme (d). The samples implanted or exposed to blood were sonicated in Triton X-100 and rinsed in deionized water, except that the 1-day implant samples were not sonicated when lysed with Triton X-100.


19

chemical species from both polymerized tetraglyme and nitrogen-containing groups derived from adsorbed proteins (Table 2a). The nitrogen-containing peaks include m/z at 56 (C3 H6 N+ ), 70 (C4 H10 N+ ), and 86 (C5 H12 N+ ). The peaks from whole blood exposed or implanted FEP contained mainly fluorinated carbons and no nitrogen-containing groups (Table 2b). The ToF-SIMS spectrum of FEP after 2-hour blood contact displayed mainly fluorinated carbon peaks (Fig. 8a) that are typical of untreated FEP. The spectrum of polymerized tetraglyme after 2-hour blood contact (Fig. 8b) was similar to untreated

(c)


20

M. Shen et al.

polymerized tetraglyme, but a nitrogen-containing peak at 70 m/z (C4 H10 N+ ) was detected. For 1-day implanted samples, the ToF-SIMS spectrum of FEP was similar to the control FEP and displayed mainly fluorinated carbon peaks and no nitrogencontaining peaks. The 1-day implanted polymerized tetraglyme samples displayed nitrogen-containing peaks at m/z 70 and 86 (Fig. 8c). The relative peak intensity Table 2a. Identities of ToF-SIMS positive ion spectra major peaks from polymerized tetraglyme samples before or after implantation Peaks (m/z)

Chemical structure

15 27 29 31 41 43 45 55 56 59 70 71 86 89 101 103

CH+ 3 C 2 H+ 3 C 2 H+ 5 CH3 O+ C 3 H+ 5 C 2 H3 O+ CH3 -O-CH+ 2 C 4 H+ 7 C 3 H6 N+ CH3 -O-CH2 -CH+ 2 C4 H10 N+ C 4 H7 O+ C5 H12 N+ CH3 -O-CH2 -CH2 -O-CH+ 2 C 5 H9 O+ 2 CH3 -(O-CH2 -CH2 )+ 2

Table 2b. Identities of major ToF-SIMS positive ion spectra peaks from FEP samples before or after implantation Peaks (m/z)

Chemical structure

12 31 50 69 93 100 119 131 169 181

C+ CF+ CF+ 2 CF+ 3 C3 F + 3 C2 F + 4 C2 F + 5 C3 F + 5 C3 F + 7 C4 F + 7


21

at m/z 59 (CH3 OCH2 CH+ 2 ) associated with the tetraglyme monomer structure was still the highest. For 4-week implanted samples, the ToF-SIMS spectrum of FEP was similar to the control FEP and displayed mainly fluorinated carbon peaks and no nitrogencontaining peaks. Implanted polymerized tetraglyme displayed nitrogen-containing peaks at m/z 56, 70, and 86 (Fig. 8d). The relative peak intensity at m/z 59 associated with the tetraglyme monomer structure was lower than the nitrogen containing peak at m/z 70.

DISCUSSION

Although PEO-containing polymers that reduced protein adsorption have been shown to reduce platelet adhesion [16, 17, 27], reducing leukocyte adhesion to surfaces has been more difficult to achieve [19– 21]. In a recent study, we showed that plasma polymerized tetraglyme effectively reduced monocyte adhesion in vitro [24], while older studies from this lab showed that platelet adhesion to this type of surfaces was also greatly reduced [28]. However, despite ultra low uptake of fibrinogen (85◦ 0.98 for all of the peptides examined). There were, however, peptide specific differences with respect to surface reaction. It can be seen at the same concentration in the reaction solution all cases, GYRGDS was found on the surface at the lowest density and GYPDSGR was found at the highest density. These differences were significant at all concentrations examined (α < 0.01). Water contact angles Water contact angle measurements, summarized in Fig. 2, show several notable factors. The unmodified PDMS surfaces are quite hydrophobic, with a measured advancing water contact angle of 96.3 ± 3.2◦ (n > 10) and receding water contact angle of 80.1 ± 4.0◦ . Following hydroxylation of the surface by plasma polymerization of allyl alcohol, a significant (p < 0.005) decrease in both the advancing and receding water contact angles occurred to 77.4 ± 1.0 and 57.8 ± 1.5 (n > 6). Modification with the peptides also resulted in significant changes in the water contact angles compared with the hydroxylated surfaces in some but not all cases. However, there were significant differences in both the advancing

144

L. Aucoin et al.

Figure 2. Advancing and receding water contact angles measured on combination peptide modified surfaces. There are significant differences in both the advancing and receding water contact angles as a result of modification with the different peptide combinations. Fractions represent % mass in reaction solution YIGSR : RGDS : PDSGR : PHSRN.

(p 0.001) and receding (p < 0.05) water contact angles on the surfaces modified with different peptide combinations. X-ray photoelectron spectroscopy XPS analysis of several of the modified surfaces, summarized in Table 3, provides additional evidence that the surfaces are modified with the peptide combinations. Relative to the unmodified PDMS surface, the hydroxylated PDMS surface shows a slight increase in the C1s signal and a decrease in the Si2p signal. While the O1s signal decreased relative the control surface, this may be more representative of the hydrophobicity of the XPS environment than of the effectiveness of the plasma polymerization of allyl alcohol in light of the significant decreases in the water contact angles. The high-resolution C1s envelope showed the appearance of a peak at 285.8, indicative of a C O bond, as would be expected on a poly allyl alcohol surface. A significant amount of silicon remained on the hydroxylated surfaces, likely the result of the depth of XPS analysis and the fluidity of the polymer. Attachment of the peptides, alone and in combination resulted in significant changes in the surface as evaluated by XPS. A small but significant N1s signal was noted on all but one of the peptide-modified surfaces as would be expected. The nitrogen levels, on the order of 1%, were however considerably less than would be expected on surface containing a peptide monolayer which is likely a reflection of

Corneal epithelial cells and cell adhesion peptide combinations surfaces

145

Table 3. Summary of XPS results on combination peptide modified surfaces Sample

C1s

Unmodified Hydroxylated 33/ 33/ 33/ 0∗ 33/ 33/ 0/ 33∗ 33/ 0/ 33/ 33∗ 0/ 33/ 33/ 33∗

Total 47.7 47.2 35.5 41.3 42.0 46.3

285.0 42.2 43.9 27.7 33.3 37.0 39.1

285.7 5.5 3.3 4.5 5.5 3.9 5.0

287.3

2.4 2.5 1.1 1.8

O1s

Si2p

26.9 28.6 38.7 34.2 32.2 29.3

25.5 24.2 24.9 24.5 25.0 23.6

N1s

289.6

0.8

1.0 0.6 0.8

* YIGSR/RGDS/PDSGR/PHSRN.

the inevitable presence of a significant amount of underlying polymer. There were significant changes in the C1s high-resolution envelopes on the peptide modified surfaces compared with the hydroxylated surfaces. Specifically, on the peptide modified surfaces there was an increase in the peaks corresponding to carbon bonded to nitrogen or oxygen via a single bond (286.3 eV) and the appearance of the peak corresponding to carboxylic and amide functions (288.8 eV) as would be expected. While there were slight differences between the different peptide combinations, no trends were apparent as would be expected from the size of an analyzed area. In all cases the amount of silicon on these surfaces remains relatively constant at approximately 24.5 atom percent, likely a reflection of the relatively small peptide size, the surface morphology and the analysis depth. In vitro cell culture studies Fractional coverage of human corneal epithelial cells on the various modified surfaces is summarized in Fig. 3. Several of the samples were repeated, and in some cases there were multiple repeats. In all but one case, the fractional adhesion between identical surfaces did not differ by more the 10% despite the fact that many of the repeats were prepared and were cultured separately from the surfaces prepared for the main experimental design. These repeated results were included in the statistical analysis of the data. However due to the low number of repeats based on the experiment design parameters (n 4 in all cases and n = 1 for most samples) error bars are not shown on the figure. It should be noted that while there were small differences in the spreading of the cells on the different surfaces, these were generally unremarkable. There was no adhesion noted on the unmodified surface as would be expected. Surprisingly, even in the presence of serum (not shown), there was no adhesion on these surfaces. There were however differences in cell coverage on the various modified samples. Statistical analysis of the data using a mixture design suggests surface modified with combinations of peptides resulted in greater fractional coverage with cells than surfaces modified with single cell adhesion peptides (p 0.01). The YIGSR, RGDS and PHSRN fractions in the reaction mixture

146

L. Aucoin et al.

Figure 3. Summary of cell adhesion results to combination peptide modified surfaces. Differences in the level of corneal epithelial cell adhesion were noted on the various surfaces as a result of modification with cell adhesion and synergistic peptides in various combinations. Fractions represent % mass in reaction solution YIGSR : RGDS : PDSGR : PHSRN.

were determined to significantly affect the resultant cell coverage. While PDSGR concentration in the reaction mixture was not shown to have a significant effect on fractional cell coverage, there was a YIGSR-PDSGR interaction effect. This is not surprising since these two peptides have been shown to act synergistically [20]. However, the combination of RGDS and PHSRN, which have also been suggested to show synergy [18, 19], did not significantly affect fractional corneal epithelial cell coverage. It is also interesting to note that the peptides combinations resulting the greatest fractional cell coverage included three or four peptides with RGDS and/ or YIGSR. Modification of the surfaces with the peptides PDSGR-YIGSR and PHSRN-RGDS gave similar levels of adhesion to surfaces modified with the same ‘free’ synergistic peptides in a 50 : 50 ratio, although in both cases the fractional epithelial cell coverage with the long chain combination peptides was slightly less than that noted on the surfaces modified with free peptides in combination. These differences are not likely significant. The effect of replacing the fibronectin derived synergistic peptide PHSRN with an inactive control on fractional corneal epithelial cell coverage is summarized in Fig. 4. In most cases, surfaces modified with the active peptide showed greater cell adhesion as expected. It is not surprising that there are also cells on the surfaces modified with combinations including the inactive control however, since


147

Figure 4. Comparison of corneal epithelial cell adhesion to surfaces modified with peptide combinations including the RGDS synergistic peptide PHSRN and the non-synergistic control peptide PPSRN. In most cases, adhesion to the surfaces modified with the non-synergistic control was less than adhesion to the surfaces modified to include the synergistic peptide. Fractions represent % mass in reaction solution YIGSR : RGDS : PDSGR : P(P-H)SRN.

the presence of the other peptides will also affect the interactions of the surface with the cells. A very low fractional coverage was noted on the surface modified with 100% PPSRN as expected. However, there was significant adhesion on the PHSRN surface, which is somewhat surprising given the reported lack of cell binding activity of this peptide [19].

DISCUSSION

The importance of tissue integration and specifically an epithelial layer in the development of a corneal onlay or keratoprosthesis has been established in a number of studies [31]. Various surface modification techniques including plasma surface modification [3, 25, 26] and covalent attachment of cell adhesion proteins [9, 14– 16] and peptides [11, 12] have been used to promote the interactions of corneal epithelial cells with the surfaces in question. In previous studies, modification with collagen or collagen derivatives, fibronectin or the fibronectin binding peptide RGDS, as well as the laminin-derived peptide YIGSR has given promising in vitro results. Despite these results however, maintaining a confluent monolayer following in vivo implantation of various devices has not been successfully achieved. This is

148

L. Aucoin et al.

likely the result of a complex interplay of factors including the surface chemistry and topography [32– 35] as well as the porosity of the underlying substrate [36– 40], providing adequate nutrient flux to the cells. The goal of surface modification with cell adhesion peptides is to mimic as closely as possible the surface under natural conditions while avoiding the complications associated with the use of proteins derived from animal tissues including inflammation and nonspecificity of attachment. It is hoped that under these conditions, preseeded cells will ultimately generate an appropriate matrix and perform the functions of the natural cornea. Given that the composition of the corneal epithelial basement membrane is primarily laminin and Type IV collagen, while fibronectin dominates under wound healing conditions, it is reasonable to hypothesize that combinations of laminin derived, fibronectin derived and/ or collagen derived peptides may mimic the corneal surface more closely, resulting in epithelial cell interactions and in the generation of a stable and physiologically active epithelial layer. The peptides used in the present study were selected based on their presence in the extracellular matrix of a normal human cornea under various conditions. The synergism between of the fibronectin derived peptides RGDS and PHSRN and that between the laminin derived peptides YIGSR and PDSGR have been shown in various studies [18– 20]. The importance of this synergism in interactions with corneal epithelial cells on synthetic surfaces was examined in the current work. While we and others have demonstrated the potential for single peptide modified surfaces, there have been relatively few studies where combinations of peptides such as those used in the current work have been examined [21– 23]. Furthermore, we are aware of no studies examining the specific peptide combinations used in this work or studies in which the various synergistic peptides are attached to the surfaces alone rather than as ‘super peptide’ combinations (i.e. RGDS and PHSRN alone rather than as RGDSPHSRN or PHSRNRGDS with or without a glycine spacer molecule). The use of an experimental design also permits us to examine a range of combinations and concentrations not previously evaluated. Longer time point experiments rather than shortterm initial adhesion studies were used in this work in order to clearly define the differences as well as to provide information about the long-term interactions between the cells and the surfaces. While the peptide modified surfaces showed a poorer response than tissue culture polystyrene, which showed complete coverage at the end point in all cases, there was a clear difference between the surfaces modified with the single peptide versus the multi-peptide solutions. Statistically, the combination of laminin derived peptides YIGSR and PDSGR showed the greatest effect on surface coverage with corneal epithelial cells, in agreement with our previous results which suggest that the presence of the laminin derived YIGSR had a greater influence on the adhesion of corneal epithelial cells than the fibronectin derived RGDS [12]. YIGSR, RGDS and PHSRN were also found to significantly affect adhesion despite that fact that PHSRN has been shown not to have a cell adhesive effect on its own. Furthermore, in support of the hypothesis that combinations of cell adhesion peptides may better


149

mimic the natural extracellular matrix, the surfaces that resulted in the greatest levels of corneal epithelial cell adhesion included multiple peptides. Surfaces with all four of the peptides, with the either RGDS or YIGSR in the greatest quantity showed particularly good corneal epithelial cell adhesion. XPS, water contact angles and surface modification with radioactively labeled peptides demonstrate that the surfaces were modified with multiple peptides. Significant differences in water contact angles were noted on the surfaces modified with different peptide combinations. As well, differences between the different peptide combinations were noted by XPS. Examination of peptide surface interactions with radiolabeled peptides showed interesting patterns of peptide surface interactions. While the radioactively labeled peptides studies are not identical to those used in cell studies due to the need for a tyrosine for labeling, it is assumed that the behaviour of these peptides will be reasonably similar to their non-labeled counterparts, and that the results therefore give a reasonable indication of the composition of the modified surfaces. There was a strong linear correlation between the amount of peptide in the reaction solution and the amount of peptide on the surface. The amount of peptide on the surfaces was clearly peptide dependent suggesting differences in peptide reactivity as a function of the composition. It is difficult to predict the surface densities with the unlabeled peptides however due to the presence of the glycine spacer and tyrosine group. Surface peptide concentrations were on the order of pmol/ cm2 , consistent with peptide surface densities measured in other studies [41, 42] and above the minimum levels suggested to be necessary for cell adhesion [43]. It is therefore apparent that the plasma polymerization reaction is generating a sufficient surface hydroxyl concentration to give a surface with a high enough peptide density to support adhesion of corneal epithelial cells. The substrate selected for these studies, poly dimethyl siloxane (PDMS), was chosen based on its previously demonstrated ophthalmic compatibility, transparency and oxygen permeability. However the porosity and therefore nutrient permeability of the materials are such that it is unlikely that this will be suitable material for long-term implantation, and modifications will be necessary to improve these properties. The surface modification procedure, involving plasma polymerization of allyl alcohol and attachment of the peptides via tresyl chloride chemistry should yield relatively substrate independent surfaces and can also be applied to other substrates, although the interactions of the cells with surface modified PDMS are superior to the interactions with surface modified pHEMA from our previous studies [12]. In this study, epithelial cells were plated on the surfaces under serum-free conditions, in order to avoid exposure to adhesive proteins in the serum including fibronectin and vitronectin. While the presence of laminin and/ or fibronectin in the culture medium may have provided information about the nature of the binding between the cells and the substrates through interactions between cell surface receptors and the adhesion proteins blocking interactions with the surface, interactions between these receptors and proteins adsorbed on the surface cannot

150

L. Aucoin et al.

be ruled out. Therefore, the use of serum-free conditions only allowed us to study the interactions of corneal epithelial cells and surface-bound peptides without the interference of extraneous and possibly confounding peptide groups. From the results, it is therefore clear that combinations of appropriate cell adhesion and synergistic peptides can improve corneal epithelial cell adhesion to synthetic surfaces. However, while it seems likely in all cases that the peptides are mediating initial attachment of the cells, allowing them to secrete matrix proteins that further promote attachment. This has been previously shown with corneal epithelial cells [44]. While it is possible that this initial attachment is non-specific in all cases, the differences in coverage between the various combinations and the lack of coverage on the PDMS control make it seem reasonable that there is some specificity in the initial attachment imparted by the presence of the peptides in different combinations. Furthermore, the dramatic decreases in the attachment when PPSRN rather than PHSRN was used provide additional evidence that there is some specificity to the attachment in the presence of the peptides. In summary, PDMS surfaces modified with a combination of cell adhesion and synergistic peptides derived from laminin and fibronectin were found to improve adhesion of corneal epithelial cells compared to modification with single peptides only. The combination of YIGSR and PDSGR was shown to have a particularly significant effect on the fractional coverage of the surfaces with the cells. Statistically, the amount of YIGSR, RGDS and PHSRN alone were also shown to have a significant effect on cell coverage. Surface modification with the different peptides was confirmed by XPS, water contact angles and radiolabeling. The radiolabeled peptide results demonstrate that the surface density of peptide on the surface varies in a linear fashion with the solution concentration of peptide. However, the surface mass density was shown to be a function of the peptide in the solution. In general, surface concentrations were on the order of pmol/ cm2 . The use of combinations of peptides on surfaces therefore has the potential to improve cell surface interactions, potentially by more closely mimicking the composition of the extracellular matrix. Acknowledgements The authors acknowledge the technical assistance of Lulu Burstzyn, Manish Bharati and Rena Cornelius. We thank Dr. John Brash and Dr. John MacGregor for useful discussions. Funding from the Natural Sciences and Engineering Research Council of Canada is gratefully acknowledged.

REFERENCES 1. A. W. Lloyd, R. G. A. Faragher and S. P. Denyer, Biomaterials 22, 769 (2001). 2. M. Franco, P. F. Nealey, S. Campbell, A. I. Teixeira and C. J. Murphy, J. Biomed. Mater. Res. 52, 261 (2000). 3. S. D. Lee, G. H. Hsiue, C. Y. Kao and P. C. T. Chang, Biomaterials 17, 587 (1996).


151

4. H. M. Leibowitz, V. Trinkaus-Randall, A. Tsuk and C. Franzblau, Prog. Ret. Eye Res. 13, 605 (1994). 5. R. Latkany, A. Tsuk, M.-S. Sheu, I.-H. Loh and V. Trinkaus-Randall, J. Biomed. Mater. Res. 36, 29 (1997). 6. C. R. Hicks, J. H. Fitton, T. V. Chirila, G. J. Crawford and I. J. Constable, Surv. Ophthalmol. 42, 175 (1997). 7. C. R. Hicks, X. Lou, S. Platten, A. B. Clayton, S. Vijaayasekaran, H. J. Fitton, T. V. Chirila, G. J. Crawford and I. J. Constable, Australian and New Zealand Journal of Ophthalmology 25, S50 (1997). 8. J. M. Legeais, I. Drubaix, B. Briat, G. Renard and Y. Pouliquen, Journal Francais d’Ophthalmologie 20, 42 (1997). 9. H. Kobayashi and Y. Ikada, Curr. Eye Res. 10, 899 (1991). 10. V. Trinkaus-Randall, M. Vural, J. Capecchi, C. Franzblau and H. M. Leibowitz, Invest. Ophthalmol. Vis. Sci. (Suppl.) 32, 1983 (1991). 11. G. Johnson, M. Jenkins, K. M. McLean, H. J. Griesser, J. Kwak, M. Goodman and J. G. Steele, J. Biomed. Mater. Res. 51, 612 (2000). 12. K. Merrett, C. M. Griffith, Y. Deslandes, G. Pleizier and H. Sheardown, J. Biomater. Sci. Polymer Edn 12, 647 (2001). 13. V. Trinkaus-Randall, J. Capecchi, A. Newton, A. Vadasz, H. Leibowitz and C. Franzblau, Invest. Ophthalmol. Vis. Sci. 29, 393 (1988). 14. H. Kobayashi and Y. Ikada, Biomaterials 12, 747 (1991). 15. R. Z. Xie, D. F. Sweeny, G. J. Beumer, G. Johnson, H. J. Griesser and J. G. Steele, Aust. NZ J. Ophthalmol 25 (Suppl. 1), S46 (1997). 16. H. Thiessen, K. McLean, G. Johnson, J. G. Steele and H. J. Griesser, Society for Biomaterials, Meeting Transactions 22, 450 (1999). 17. E. Ruoslahti, Ann. Rev. Cell. Dev. Biol. 12, 697 (1996). 18. S. Aota, M. Nomizu and K. M. Yamada, J. Biol. Chem. 269, 24756 (1994). 19. S. D. Redick, D. L. Settles, G. Briscoe and H. P. Erikson, J. Cell. Biol. 149, 521 (2000). 20. H. K. Kleinman, J. Graf, Y. Iwamoto, M. Sasaki, C. S. Schasteen, Y. Yamada, G. R. Martin and F. A. Robey, Arch. Biochem. Biophys. 272, 39 (1989). 21. W. J. Kao, D. Lee, J. C. Schense and J. A. Hubbell, J. Biomed. Mater. Res. 55, 79 (2001). 22. D. J. Heath, S. Halstenberg, J. C. Schense, J. W. Kao, S. Downes and J. Hubbell, in: Society for Biomaterials 25th Annual Meeting, Abstracts 22, p. 140 (1999). 23. Y. W. Tong and M. S. Schoichet, Biomaterials 22, 1029 (2001). 24. D. L. Mooradian, J. B. McCarthy, J. D. Cameron, A. P. N. Skubitz and L. T. Furcht, Invest. Ophthalmol. Vis. Sci. 29, 393 (1988). 25. G.-H. Hsiue, S. D. Lee, P. C. T. Chang and C.-Y. Kao, J. Biomed. Mater. Res. 42, 134 (1998). 26. P. C. T. Chang, S. D. Lee and G. H. Hsiue, J. Biomed. Mater. Res. 39, 380 (1998). 27. B. M. Wickson and J. L. Brash, Colloids and Surfaces A: Physiochem and Eng Aspects 156, 201 (1999). 28. D. Yu and R. E. Marchant, Macromolecules 22, 2957 (1989). 29. S. Yuan and R. E. Marchant, J. Appl. Polym. Sci. Appl. Polym. Symp. 54, 77 (1994). 30. K. Araki-Sasaki, Y. Ohashi, T. Sasbe, K. Hayashi, H. Watanabe, Y. Tano and H. Handa, Invest. Ophthalmol. Vis. Sci. 36, 614 (1995). 31. T. V. Chirila, C. R. Hicks, P. D. Dalton, S. Vijaysekaran, X. Lou, Y. Hong, A. B. Clayton, B. W. Ziegelaar, J. H. Fitton, S. Platten, G. J. Crawford and I. J. Constable, Prog. Poly. Sci. 23, 447 (1998). 32. R. Z. Xie, D. F. Sweeney, H. J. Griesser and J. G. Steele, Aust. NZ J. Ophthalmol. 26, S47 (1998). 33. J. H. Fitton, B. A. Dalton, G. Beumer, G. Johnson, H. J. Griesser and J. G. Steele, J. Biomed. Mater. Res. 42, 245 (1998).

152

L. Aucoin et al.

34. B. A. Dalton, M. D. M. Evans, G. A. McFarland and J. G. Steele, J. Biomed. Mater. Res. 45, 384 (1999). 35. M. D. M. Evans, B. A. Dalton and J. G. Steele, J. Biomed. Mater. Res. 46, 485 (1999). 36. R. Turss, J. Friend and C. H. Dohlman, Exp. Eye Res. 9, 254 (1970). 37. S. L. Lane and R. L. Lindstrom, Int. Ophthalmol. Clin. 31, 37 (1991). 38. B. E. McCarey, Refract. Surg. 31, 87 (1991). 39. M. B. McDonald, B. E. McCarey, B. Storie, R. W. Beuerman, B. Salmeron, G. Van Rihj and P. M. Knight, J. Cataract. Refract. Surg. 19, 213 (1999). 40. D. F. Sweeney, R. Z. Xie, D. J. O’Leary, A. Vannas, R. O’Dell, K. Schindhelm, H. Cheng, J. G. Steele and B. A. Holden, Invest. Ophthalmol. Vis. Sci. 39, 284 (1997). 41. R. McMillan, B. Meeks, F. Bensebaa, Y. Deslandes and H. Sheardown, J. Biomed. Mater. Res. 54, 272 (2001). 42. S. Saneinejad and M. Shoichet, J. Biomed. Mater. Res. 42, 13 (1998). 43. S. P. Massia and J. A. Hubbell, J. Cell Biol. 114, 1089 (1991). 44. J. G. Steele, G. Johnson, H. J. Griesser and P. A. Underwood, Biomaterials 18, 1541 (1997).

Polyelectrolyte multilayer films modulate cytoskeletal organization in chondrosarcoma cells DOMINIQUE VAUTIER 1,∗ , VERONIQUE KARSTEN 1 , CHRISTOPHE EGLES 1 , JOHANNA CHLUBA 1 , PIERRE SCHAAF 2 , JEAN-CLAUDE VOEGEL 1 and JOËLLE OGIER 1 1 INSERM U 424, UFR Odontologie, 11 Rue Humann, 67085 Strasbourg Cedex, France 2 Institut Charles Sadron (CNRS/ULP), 6 Rue Boussingault, 67083 Strasbourg Cedex, France

Received 6 December 2001; accepted 18 March 2002 Abstract—The aim of this study was to evaluate polyelectrolyte multilayer films as interfaces for implants. Polyelectrolyte multilayers were built up with different terminating layers by alternate deposition of oppositely charged polyelectrolytes on which chondrosarcoma (HCS-2/8) cells were grown in the presence of serum. Films formed by an increasing number of layers were investigated. The terminating layer was made of one of the following polyelectrolytes: poly-sodium-4-styrenesulfonate (PSS), poly-L -glutamic acid (PGA), poly-allylamine hydrochloride (PAH), or poly(L -lysine) (PLL). Cell viability, inflammatory response, adherence, and cytoskeletal organization were studied. Induction of interleukin-8 (IL-8) secretion was detected on PAH and PLL ending polyelectrolyte films. Early cellular adherence was enhanced with PGA, PAH, PLL, and, to a lower extent, PSS terminating layers. Adherence was independent of the number of layers constituting the films. The presence of actin filaments and vinculin focal adhesion spots was observed on PSS or PAH ending films. They were respectively partially and totally absent on PGA and PLL terminating multilayer architectures. For PLL ending films, vinculin and actin organization was clearly dependent on the number of deposited layers. The results of this study suggest that PSS ending multilayered films constitute a good interfacial micro-environment at the material surface for HCS-2/8 cells. Key words: Polyelectrolyte multilayer films; cell adhesion; actin; vinculin.

INTRODUCTION

The chemical modification of biomaterial surfaces became a major challenge in the last decade, allowing broad medical applications for implant and tissue engineering [1– 3]. Bioactive molecules such as enzymatic peptides, for example, ∗ To whom correspondence

should be addressed. E-mail: [email protected]

154

D. Vautier et al.

have been incorporated into the bulk of biodegradable materials [4]. In other approaches, only surface properties are modified. Self-assembled monolayers (SAMs) or Langmuir– Blodgett techniques have been commonly employed to produce new interfaces [5, 6]. More recently, a new versatile method of self-assembled architectures based on the alternate deposition of polyanions and polycations has been developed for the build up of multilayered polyelectrolyte films [7]. Besides varying film thickness, roughness, and porosity, it is also possible to incorporate in the film architecture functionalized macromolecules, enzymes, and proteins embedded at different depths [8, 9]. This concept gives rise to complex nanoarchitectures with specific biomimetic properties. It has, for example, been shown that α-melanocortin peptide hormone covalently coupled to poly(L-lysine) conserves full biological activity when deeply embedded into poly(L-lysine)/ poly(Lglutamic acid) (PLL /PGA) multilayer architectures [10]. Aimed to be used in contact with biofluids and/ or tissues, interfaces need to be evaluated in terms of biocompatibility properties in in vivo or in vitro conditions and in terms of cellular interaction with the substratum [11, 12]. The cell adhesion process contributes strongly to the clinical success of the implanted biomaterials [13]. During the adhesion of cells to material, specific cellular processes, such as signal transduction responses [13], can be deeply influenced by the interface microenvironment. The responses are mediated through cell-surface integrin receptors, which bind the cell to surface-adsorbed extracellular matrix proteins (ECMs) and connect the actin filament cytoskeleton to signalling molecules, including among them the cytoskeletal proteins vinculin and paxillin [14– 16]. Successful formation of the cytoskeletal organization, in addition to other early adhesion-dependent events [17, 18], is a major indicator of efficient ‘outside– in’ signal occurring between the extracellular environment (e.g. ECMs adsorbed on the material surface) and the cell [19, 20]. Physiological fluids (e.g. serum) contain numerous soluble proteins that react spontaneously with solid surfaces in very dynamic ways and contribute to the attachment and spreading of the cells [21– 23]. Most of the investigations concerning cellular interactions with a polyelectrolyte have been performed on adsorbed polyelectrolyte monolayers. For example, the coating of PLL monolayer is one of the most commonly used methods to promote cellular adhesion. Up to now, only a few studies have been devoted to the behavior of cells on polyelectrolyte multilayers. In particular, the influence of polyelectrolyte multilayer films on cellular cytoskeletal organization has not yet been studied. Therefore, in the present work we propose to investigate cell cytoskeletal actin and vinculin organization in early adhesion events on polyelectrolyte multilayer films. Different polycations such as polyethyleneimine (PEI), poly-allylamine hydrochloride (PAH), and poly(L-lysine) (PLL) and polyanions such as poly-sodium 4-styrenesulfonate (PSS) and poly-L-glutamic acid (PGA) were used to build up these architectures. The chosen cellular model was a cell line of human chondrosarcoma (HCS-2/8) having a cartilage phenotype. HCS-2/8 cells possess the ability

Polyelectrolyte multilayer films modulate cytoskeletal organization

155

to synthesize cartilage-specific proteoglycans and type II collagen [24]. Here, we evaluated viability parameters such as cell nucleus morphology, cell inflammatory response, and adherence.


Cell culture HCS-2/8 cells, a human chondrosarcoma-derived chondrocyte-like cell line [24], were routinely grown in Gibco BRL’s minimum essential medium with Eagle’s Salts (MEM, Life Technologies), 10% fetal calf serum (FCS, Life Technologies), 50 U /ml penicillin, and 50 µg/ ml streptomycin (Biowhittaker) in an atmosphere of 5% CO2 and 95% air at 37 ◦ C. Polyelectrolyte film deposition The following polyelectrolytes were used for the build up of the multilayer films: polyethyleneimine (PEI; Mr 750 × 103 , Aldrich), polysodium 4-styrenesulfonate (PSS; Mr 70 × 103 , Aldrich), poly-allylamine hydrochloride (PAH; Mr 70 × 103 , Aldrich), poly-L-glutamic acid (PGA; Mr 54.8 × 103 , Sigma), and poly(Llysine) (PLL; Mr 23.4 × 103 , Sigma). PEI, PSS, and PAH solutions were prepared at 5 mg/ ml, whereas PGA and PLL solutions were used at 1 mg/ml. All polyelectrolytes were dissolved in 1 M NaCl. Glass coverslips (CML, France) were pretreated for 15 min at 100 ◦ C with 10−2 M SDS and 0.12 N HCl, followed by extensive doubly distilled H2 O rinsing. Glass coverslips were deposited in 24well plastic plates (NUNC) and immersed for 20 min in 300 µl of the appropriate polyelectrolyte solution. After each polyelectrolyte layer deposition, the wells were rinsed three times for 5 min with distilled water. A first set of film build-ups started by the deposition of a precursor film made up of PEI-(PSS /PAH)2 . The architecture was completed by further addition of either PSS, PGA /PLL /PGA, PGA /PAH, PGA /PLL, or PGA /PEI layers. In the following, these films will be described by the last deposited layer and denoted with the corresponding polyelectrolyte abbreviation (PSS, PGA, PAH, PLL or PEI) (type 1 architectures). As the control surface, a PAH or PLL monolayer and PLL /PSS or PLL /PGA bilayers, directly deposited on the solid surface, were made. A second set of polyelectrolyte multilayer architectures was prepared by changing the number of polyelectrolyte layers. These films were made up of (PAH /PSS)n or (PLL /PGA)n -PLL, with n = 1, 5, 10, and directly assembled on the glass surface (type 2 architectures). All of these different build-ups of polyelectrolyte multilayer architectures were controlled by optical wave-guide light-mode spectroscopy [10, 25]. Films were sterilized for 15 min by UV light (254 nm), stored at 4 ◦ C, and usually used within 1 week.

156

D. Vautier et al.

Test of cell viability HCS-2/8 cells (105 cells per well) were seeded on different polyelectrolyte multilayer architecture-coated coverslips, placed into 24-well plates, and maintained for 24 h. As a positive control, apoptosis was induced on PEI-terminating layers, which are known to be cytotoxic [26]. Cells were fixed with 3.7% paraformaldehyde (Boehringer), permeabilized with 0.1% Triton X-100 (Sigma), washed with phosphate-buffered saline (PBS), and finally stained with 2 µg/ ml Hoechst 33258 (Sigma) to detect condensed nuclei of apoptotic cells [27]. At the end of the procedure, coverslips were mounted in VectaShield (Vector) on glass slides and visualized using a Zeiss Axioplan 2 microscope with an MC 80 DX photographic attachment. Cytokine production The amount of IL-8 released after 24 h by HCS-2/8 cells (105 cells per well) in 24well plates coated with the different polyelectrolyte films was measured in culture supernatants by enzyme-linked immunoassay (ELISA, Biosource) according to the manufacturer’s recommended procedures. We used TNF-α preconditioned HCS-2/8 (10 ng of TNF-α per ml of cell culture, 24 h activation) as a positive control. These pretreated cells were washed out of the drug before seeding them into ELISA plates. In graphs, mean values with standard errors for one experiment as normalized data for an identical total cell number per well are given. Cell adhesion assay Quantification of attached cells labeled with horseradish peroxidase (HRP) was performed according to the method described by Löster et al. [28]. Briefly, HCS2/ 8 cells were washed with serum-free MEM, suspended subsequently in serumfree MEM containing 2 mg/ ml HRP, and incubated at 37 ◦ C for 30 min. After a further wash with serum-free MEM, HCS-2/8 cells, in 10% serum MEM, were plated at a density of 105 cells/ ml on the films. After 30 min of culture, the absorbance was measured. Unattached cells were removed after two washes with PBS at 37 ◦ C. Wells containing HRP-labeled attached cells on the different polyelectrolyte multilayer film-coated glass coverslips were incubated with 500 µl/ well of HRP-substrate buffer [10 mg OPD Sigma, 25 µl of 6% (v/ v) H2 O2 solution in 20 ml of 0.1 M sodium citrate buffer (pH 6.0), 0.5% (v/ v) Triton X-100]. After 15 min of reaction in the absence of light, the reaction was stopped by adding 125 µl of 1 M H2 SO4 per well and the OD of the reaction product was measured at 490 nm. The absorbance of the dye varied linearly with the number of adhering cells per well. Immunofluorescence The cells attached on the films were rinsed briefly in PBS and fixed in 3.7% formaldehyde diluted in PBS for 15 min at room temperature. The cells were then


157

permeabilized for 10 min with 0.1% Triton X-100 in 3.7% formaldehyde diluted in PBS at room temperature. Rinsing was performed with PBS, followed by two 15-min incubations with 10% fetal calf serum and 15 min staining by TRITCphalloidin (5 µg/ ml, P1951, Sigma) and a final wash in PBS. For detection of vinculin, the cells were incubated with the primary antibody anti-human vinculin (V9131, Sigma), 1/ 100 diluted in PBS, washed in PBSand incubated for 30 min with the secondary antibody conjugated to fluorescein (Nordic, Tilburg), diluted 1/ 40 in PBS. Finally, cells were rinsed in PBS and mounted in VectaShield (Vector). Samples were observed with a Nikon Microphot-FXA fluorescence microscope.

RESULTS

Cell viability on polyelectrolyte multilayer films When HCS-2/8 cells were observed after 24 h of culture on PSS (Fig. 1A), PGA (not shown), PAH (not shown) or PLL (Fig. 1B) terminating layers (type 1 architectures), the nucleoplasm of the cells was uniformly labeled with the specific staining (Hoechst 33258). The aspect of the nucleus was very similar to that observed on untreated glass coverslips (Fig. 1C). In contrast, apoptotic condensed HCS-2/8 cell nuclei were observed after 24 h of culture on PEI terminating layers (Fig. 1D). Cytokine production on polyelectrolyte multilayer films In order to check the behavior of the polyelectrolyte multilayer architectures towards pro-inflammatory processes, we tested their effect on IL-8 secretion by HCS-2/8 cells after 24 h contact. In the culture supernatant of HCS-2/8 cells grown on PLL or on PAH terminating films (type 1 architectures), the IL-8 level was increased by 2.5 and 3 times, respectively, in comparison with the IL-8 level measured in the supernatant of HCS-2/8 cells grown on control surfaces without a film (‘IL-8 control surface’) (Fig. 2). The IL-8 level measured with PGA or PSS terminating layers was not significantly different from the ‘IL-8 control surface’. Adhesion of HCS-2/ 8 cells on polyelectrolyte multilayer films The cellular adhesion of HCS-2/8 cells was quantified by a HRP adhesion assay (see the Materials and Methods section). After 30 min of contact, comparatively to an uncoated glass surface, the cellular adhesion on PAH, PGA, and PLL ending films was increased by about 4 times (Fig. 3) and by 2 times on type 1 architectures

158

D. Vautier et al.

(A)

(B) Figure 1. Assessment of apoptosis by staining with H33258. HCS-2/8 cells were cultured for 24 h in MEM 10% FCS on PSS (A), PLL (B), and PEI (D) ending polyelectrolyte films, or on glass coverslips (C) fixed and stained with H33258. (D) Apoptotic condensed nuclei (arrow). Bar = 4 µm.


(C)

(D) Figure 1. (Continued).

159

160

D. Vautier et al.

Figure 2. Effect of polyelectrolyte multilayer films (type 1 architectures) on IL-8 production by HCS-2 cells. Cells were seeded at a density of 105 cells per well in 96-well plates, pretreated with polyelectrolyte films as described in the Materials and Methods section. Cell culture supernatants were collected after 24 h of culture and analyzed by ELISA for the presence of IL-8. TNF-α-preconditioned HCS-2 cells (10 ng/ ml, 24 h) were used as a positive control for cytokine production. All films started by the precursor film PEI-(PSS/PAH)2 (wave symbol). The ending polyelectrolyte layers are indicated under the different columns. The cytokine level of each sample was normalized to the total cell number per well. Results are means of triplicate well (three measurements per well) determinations of one independent experiment. The error bars represent the standard errors of the means derived from independent experiments. *Statistically significant difference with respect to the other groups (n = 3, p < 0.001, ANOVA, Tukey test).

(Fig. 3). When the assay was monitored after 2 h, the measurements of the absorbance reached 2.3 OD on untreated surface, 2.1 OD on PSS terminating films, 2.29 OD on PGA terminating films, 2.3 OD on PAH terminating films, and 2.4 OD on PLL terminating films (type 1 architectures). These data suggested that the total number of attached cells was similar for all substrates. These results also indicate that the ending polyelectrolyte layer of the architecture modulates only early adhesion events. The influence on cellular adhesion of the number of deposited polyelectrolyte layers was next studied using PSS and PLL terminating layers. In this evaluation, (PLL /PGA)n -PLL or (PAH /PSS)n multilayer architectures were built up with n = 1, 5 or 10 (type 2 architectures). For both kinds of architecture, no significant differences were found, whatever the number of deposited layers (Fig. 4A, PSS ending films; Fig. 4B, PLL ending films).


161

Figure 3. Cell adhesion on various polyelectrolyte ending multilayer films (type 1 architectures). All build-ups started by a precursor PEI-(PSS/PAH)2 film (wave symbol). The ending polyelectrolyte layers are indicated under the different columns. Cell attachment was monitored after 30 min postseeding. Results are means of triplicate well (four measurements per well) determinations of two independent experiments. The error bars represent the standard error of the two means derived from the two independent experiments. *Statistically significant difference with respect to the other groups (n = 24, p < 0.001, ANOVA, Student Newman– Keuls method).

Cytoskeletal organization The interactions of HCS-2/8 cells with their substratum were also evaluated by means of the cytoskeletal organization of vinculin and actin proteins. HCS-2/8 cells were cultured for 4 h on surfaces coated with monolayers (PAH or PLL), bilayers (PLL /PSS or PLL /PGA), or multilayer architectures (see the Materials and Methods section) in which the ending polyelectrolyte was varied (PSS, PGA, PAH or PLL) (type 1 architecture). On uncoated surfaces, cells presented a large variety of morphologies with predominant rounded aspects. A punctate distribution of vinculin is known to be associated with focal adhesion complexes [14]. As can be seen in Fig. 5A, HSC-2/8 cells newly attached on glass coverslips contain focal adhesion spots of vinculin localized at the base of the cell body. These structures were also observed when cells attached on PLL /PSS (not shown) or PLL /PGA bilayered films (Fig. 5B) and on multilayered PSS ending architectures (Fig. 5C). In contrast, the spots of vinculin were considerably reduced in size in HCS-2/8 cells attached on multilayered architectures ending with PGA (not shown). With PAH monolayer or multilayered architectures ending with PAH (not shown), the

162

D. Vautier et al.

Figure 4. Cell adhesion on polyelectrolyte multilayer films [varying the number of layers and ending with PSS (A) or PLL (B)]. The polyelectrolyte multilayer architectures are indicated under the different columns. Cell attachment was monitored after 30 min post-seeding. Results are means of triplicate well (four measurements per well) determinations for two independent experiments. The error bars represent the standard error of the two means derived from the two independent experiments. *No significant difference with respect to the other groups (n = 24, p < 0.001 ANOVA, Student Newman– Keuls method).


163

(A) Figure 5. Vinculin immunostaining in HCS-2/ 8 cells adhered to polyelectrolyte monolayer, bilayer or multilayer films, varying the ending polyelectrolyte (type 1 architectures). HCS-2/8 cells were plated on (A) glass coverslips, or on (B) PLL/PGA and (D) PLL, or on polyelectrolyte multilayer films starting with the PEI-(PSS/PAH)2 precursor film and ending with (C) PSS and (E) PGA/PLL. HCS2/ 8 cells were cultured for 4 h at 37 ◦ C, fixed, and stained for vinculin. (A– D) Focal adhesion spots of vinculin (arrowhead). Comparable cellular observations were made in two independent experiments. Bar = 5 µm.

vinculin complexes were also similar to those described for cells adhering on glass coverslips. On PLL monolayers, several cells still presented characteristic vinculin focal adhesion (Fig. 5D), whereas for PLL ending multilayered architectures this cytoskeletal organization was not detected (Fig. 5E). Cells in contact with uncoated surfaces exhibited a fine parallel actin filament network (Fig. 6A). Cellular morphology and actin filament organization remained quite similar when cells attached on PLL /PSS (not shown) or on PLL /PGA bilayers (Fig. 6B). On PSS ending multilayer architectures, cells had actin cables localized at the periphery of the cell bodies together with an actin punctate distribution inside the cell body (Fig. 6C). However, on PGA ending multilayer architectures, little polymerized actin was seen (not shown). An actin punctate distribution was also clearly seen on PAH (not shown) or PLL monolayers (Fig. 6D), or with PAH (not shown) and PLL (Fig. 6E) ending multilayered architectures.

164

D. Vautier et al.

(B)

(C) Figure 5. (Continued).


(D)

(E) Figure 5. (Continued).

165

166

D. Vautier et al.

(A) Figure 6. Actin staining in HCS-2/ 8 cells adhered to polyelectrolyte monolayer, bilayer or multilayer films, varying the ending polyelectrolyte (type 1 architectures). HCS-2/8 cells were plated on (A) glass coverslips, or on (B) PLL/PGA and (D) PLL, or on multilayered polyelectrolyte starting with the PEI-(PSS /PAH)2 precursor film and ending with (C) PSS and (E) PGA/ PLL. HCS-2/8 cells were cultured for 4 h at 37 ◦ C and then fixed. Actin was stained with TRITC-phalloidin. (A– C) Actin filaments (arrow); (D, E) actin dots (arrowhead). Comparable cellular observations were made in two independent experiments. Bar = 5 µm.

DISCUSSION

Since biomaterial/ cell interactions are determinant for cellular behavior such as cell adhesion [29], proliferation [30], and differentiation [31], we evaluated the viability, adhesion, and cytoskeleton organization of HCS-2/8 human chondrocyte-like cells with interfaces made of different polyelectrolyte multilayer films in the presence of serum. The structural and physico-chemical properties of multilayered polyelectrolyte architectures have been extensively investigated in the literature in the last decade. As a general rule for such films, each new polyanion (respectively polycation) addition renders the surface alternatively negatively (respectively positively) charged. This also holds for the PSS /PAH system, with zeta potential values alternating between −20 mV and +20 mV [32], and between −50 mV and +50 mV for the PGA /PLL system [33], after each new polyanion or polycation addition. The film thickness or amount of polyelectrolyte deposited per layer is largely dependent on various parameters such as ionic strength, the pH for weak polyacids or polybases, and film


(B)

(C) Figure 6. (Continued).

167

168

D. Vautier et al.

(D)

(E) Figure 6. (Continued).


169

growth mechanisms (linear or super increase regimes). Thus, for the PSS /PAH system, under comparable experimental conditions, the thicknesses determined by means of optical wave-guide light-mode spectroscopy (OWLS) were about 40 nm for PEI(PSS /PAH)5 and 70 nm for PEI(PSS /PAH)10. For the PGA /PLL system growing in a linear manner, the thicknesses obtained for PEI(PGA /PLL)5 and PEI(PGA /PLL)8 were about 100 and 280 nm, respectively [33]. Film roughnesses were also estimated by means of atomic force microscopy (AFM) and the maximal height amplitudes were of the order of 10– 12 nm for films having thicknesses of about 40 and 70 nm as estimated by OWLS. For the PGA /PLL system, the height amplitudes of the film profile increase with the number of deposited layers and then reach a constant value of about 25 nm for PEI(PGA /PLL)5 and PEI(PGA /PLL)10 [33]. Cell morphology and inflammatory responses were investigated to assess cell viability in the presence of such films. As previously observed for other cell models [26], on PSS, PAH, PGA, and PLL ending layers, HCS-2/8 cells did not exhibit an apoptotic morphology. However, cells secreted a pro-inflammatory cytokine, IL-8, when grown on PAH or PLL ending films. We observed that compared with adhesion to glass, early cell adhesion was enhanced by interfaces made of polyelectrolyte films ending with PAH, PGA, and PLL, and to a lower extent with PSS ending films. These results confirm that early adhesion is promoted on positively charged surfaces [26]. The cell attachment assay (HRP assay) used here was based on the total number of attached cells on a given surface. In these conditions, we found that the number of layers (n = 1, 5 or 10) did not influence cell adhesion, either with (PAH /PSS)n or with (PLL /PGA)n /PLL polyelectrolyte multilayer architectures. These results suggest that cells detect principally the ending layer and do not sense the underlying oppositely charged layers; they also suggest that cells were relatively insensitive to the surface roughness or porosity changes related to the number of deposited polyelectrolyte layers (Lavalle, personal communication). However, the HRP adhesion assay did not necessarily reflect attachment forces, which have been shown, in our laboratory, to be dependent on the number of layers as well as on the charge of the ending layer. The micro-peeling approach allows quantitative estimation of the cellular adhesion onto polyelectrolyte multilayer architectures. In accordance with the present work, stronger adhesion was found on positively charged ending multilayer films than on negatively charged ones, but differences in the cellular adhesion responses by changing the number of layers in the film were also measured. Decreased rupture forces with increased layer numbers for the positively charged PLL and no variation for the negatively charged PGA ending multilayers were measured (Richert, personal communication). The mechanism of cell attachment, spreading, and migration on surfaces with various physico-chemical properties, surface characteristics, and in the presence of adsorbed protein has been extensively studied [34]. Cell adhesion likely occurs via extracellular matrix proteins [fibronectin (FN), vitronectin, laminin, and albumin]

170

D. Vautier et al.

adsorbed onto the surface [35]. Previous studies have shown that most of the cell adhesion activity on synthetic surfaces was generated by surface-bound proteins containing serum [36]. A recent study by McClary et al. showed that the surface of carboxyl-terminated alkylthiol SAMs yielded higher levels of FN adsorption that correlated with high degrees of cell attachment spreading than methyl-terminated SAMs [20]. For osteoblasts, phosphorylation of paxillin, a cytoskeletal adaptor protein, could be induced by serum proteins, FN, and vitronectin adsorbed on a charged polystyrene surface [37]. In this study, analysis of the cytoskeletal actin and vinculin focal adhesion organization demonstrated differences in response with the charge of the ending layers and also with the number of deposited layers. Negatively charged ending layers preserved both actin filaments and vinculin spots for PLL /PSS or PLL /PGA bilayer and PSS ending multilayer build-ups, but not for PGA ending multilayer films. In this latter condition, the spots of vinculin were considerably reduced. PAH and PLL ending layers were less favorable for the polymerized form of actin. Interestingly, the most striking difference in the cytoskeletal organization, related to the number of layers constituting the films, was observed for PLL ending films. Cells contained vinculin adhesion spots and also some actin filaments after binding on a PLL monolayer. In contrast, they showed neither vinculin adhesion spots nor actin polymeric forms on PLL ending multilayer architectures. It has been reported that the focal adhesion kinase (FAK) [38, 39], the tensin protein [40], and the mitogen-activated protein kinase (MAPK) [41] were not phosphorylated in cells plated onto PLL. Tensin and FAK are important components of the focal adhesion sites, which become tyrosine-phosphorylated upon integrin-mediated attachment to ECM proteins. MAPK is activated and translocated to the nucleus in integrinmediated signal transduction pathways. Therefore, Krause et al. [41] proposed that cell adhesion to PLL does not involve integrins, but ionic charges. In accordance with these data, our results suggest that the adhesion mechanism is rather chargedependent on PLL multilayered ending films. PLL could exert an inhibitory effect on vinculin adhesion spots and actin filament organization. It will be interesting to determine whether PLL and PSS multilayered ending films involve an integrindependent adhesion mechanism in the presence of serum. In this case, contributions of serum-derived adhesion proteins such as fibronectin, vitronectin, laminin, and albumin would need to be evaluated. During cell adhesion, ECM produced by HCS2/ 8 cells could also participate in the adhesion mechanism, as well as ECM proteins present in serum. Considering this possibility, one cannot exclude a possible influence of the polyelectrolyte architecture on ECM production. On PLL monolayers, some cells presented spots of vinculin. This difference could be explained by the adsorption process of monolayers compared with multilayer films. Indeed, by AFM investigations, it was found that after deposition of PEI(PGA /PLL)1 on a silica surface, the surface was not completely coated with polyelectrolytes [33]. However, the same authors found that the polyelectrolyte multilayer architecture completely covered the silica surface after deposition of the


171

fifth bilayer. Consistent with these data, our results suggest that cells adsorbed on a PLL monolayer could sense that silica was not completely covered with PLL and form focal adhesion sites. In contrast, on multilayer PLL ending films, cells were unable to sense the silica surface and did not develop focal adhesion sites.

CONCLUSION

We have examined chondrosarcoma cell interactions with polyelectrolyte multilayered surfaces, evaluating the biocompatibility of the films and the cytoskeletal organization of the HCS-2/8 cells in an in vitro cell culture system with serum. The results of this study demonstrate that the different polyelectrolyte ending multilayers tested did not induce a nucleus apoptotic morphology. However, PAH and PLL ending films activated IL-8 secretion. Early cell adhesion was optimal on positively charged (PAH and PLL) compared with negatively charged ending polyelectrolytes (PSS), and was not influenced by the number of polyelectrolyte layers deposited. In contrast, the number of layers could modulate cytoskeletal actin and vinculin. Finally, PLL ending multilayer films appear to be less biocompatible in terms of cytoskeletal organization than PSS. PSS ending architectures could be new and interesting interfaces for implants. Acknowledgements We thank Dr. C. Picart (University Louis Pasteur, Strasbourg, France) for stimulating discussions. We acknowledge Mrs N. Fournier for developing photographs and Mrs C. Affolter for technical support (University Louis Pasteur, Strasbourg, France). This study was supported by grants from Ligue Nationale Contre le Cancer (Comité du Bas-Rhin), from the program INSERM –CNRS ‘Ingéniérie Tissulaire’, and from the IFRO (Institut Français pour la Recherche Odontologique).

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

R. Langer and J. Vacanti, Science 260, 920 (1993). N. Peppas and R. Langer, Science 263, 1715 (1994). N. Angelova and D. Hunkeler, Trends Biotechnol. 17, 409 (1999). J. C. Schense, J. Bloch, P. Aebischer and A. Hubbell, Nature Biotech. 18, 415 (2000). M. Mrksich, Curr. Opin. Colloid Interface Sci. 2, 83 (1997). M. Lösche, Curr. Opin. Solid State Mater. Sci. 2, 546 (1997). G. Decher, Science 277, 1232 (1997). F. Caruso, K. Niikura, D. N. Furlong and Y. Okahata, Langmuir 13, 3427 (1997). T. Cassier, K. Lowack and G. Decher, Supramol. Sci. 5, 309 (1998). J. Chluba, J.-C. Voegel, G. Decher, P. Erbacher, P. Schaaf and J. Ogier, Biomacromolecules 2, 800 (2001). 11. B. Groessner-Schreiber and R. S. Tuan, J. Cell Sci. 101, 209 (1992). 12. L. Chou, J. D. Firth, V. J. Uitto and D. M. Brunette, J. Cell Sci. 108, 1563 (1995).

172 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41.

D. Vautier et al. J. A. Hubbell, Biotechnology 13, 565 (1995). C. H. Damsky and Z. Werb, Curr. Opin Cell Biol. 4, 772 (1992). S. Miyamoto, S. K. Akiyama and K. M. Yamada, Science 267, 883 (1995). D. A. Calderwood, S. J. Shattil and M. H. Ginsberg, J. Biol. Chem. 275, 22610 (2000). L. Kornberg, H. S. Earp, J. T. Parsons, M. Schaller and R. L. Juliano, J. Biol. Chem. 267, 23439 (1992). M. D. Schaller and J. T. Parsons, Curr. Opin. Cell Biol. 6, 705 (1994). E. A. Clark and J. S. Brugge, Science 268, 233 (1995). K. B. McClary, T. Ugarova and D. W. Grainger, J. Biomed. Mater. Res. 50, 439 (2000). J. Andrade (Ed.), Surface and Interfacial Aspects of Biomedical Polymers. Plenum Press, New York (1985). M. Mrksich, C. S. Chen, Y. Xia, L. E. Dike, D. E. Ingber and G. M. Whitesides, Proc. Natl. Acad. Sci. USA 93, 10775 (1996). A. S. Curtis and J. V. Forrester, J. Cell Sci. 71, 17 (1984); C. H. Damsky and Z. Werb, Curr. Opin. Cell Biol. 4, 772 (1992). M. Takigawa, K. Tajima, H. O. Pan, M. Enomoto, A. Kinoshita, F. Suzuki, Y. Takano and Y. Mori, Cancer Res. 49, 3996 (1989). C. Picart, G. Ladam, B. Senger, J.-C. Voegel, P. Schaaf, F. J. G. Cuisinier and C. Gergely, J. Chem. Phys. 115, 1086 (2001). P. Tryoen-Toth, D. Vautier, Y. Haikel, J.-C. Voegel, P. Schaaf, J. Chluba and J. Ogier, J. Biomed. Mater. Res. (in press). Z. Darzynkiewicz, G. Juan, X. Li, W. Gorczyca, T. Murakami and F. Traganos, Cytometry 27, 1 (1997). K. Löster, C. Schüler, C. Heidrich, R. Horstkorte and W. Reutter, Anal. Biochem. 244, 96 (1997). G. Rainaldi, P. Filippini, A. Ferrante, P. L. Indovina and M. T. Santini, J. Biomed. Mater. Res. 55, 104 (2001). Y. Mitsui, K. Yamamoto, M. Yamamoto and K. Matuoka, Adv. Exp. Med. Biol. 190, 567 (1985). R. Herrero, M. R. Moncelli, R. Guidelli, M. Carla, A. Arcangeli and M. Olivotto, Biochim. Biophys. Acta 1466, 278 (2000). G. Ladam, P. Schaad, J. C. Voegel, P. Schaaf, G. Decher and F. Cuisinier, Langmuir 16, 1249 (2000). Ph. Lavalle, C. Gergely, F. J. G. Cuisinier, G. Decher, P. Schaaf, J. C. Voegel and C. Picart, Macromolecules (in press). K. Anselms, Biomaterials 21, 667 (2000). T. A. Horbett, Cardiovasc. Pathol. 2, 137 (1993). P. A. Underwood and F. A. Bennett, J. Cell Sci. 4, 641 (1989). D. W. Sommerfeldt, K. J. McLeod, C. T. Rubin and M. Hadjiargyrou, Biochem. Biophys. Res. Commun. 285, 255 (2001). K. Burridge, C. E. Turner and L. H. Romer, J. Cell Biol. 119, 893 (1992). M. Ouwens, H. M. M. Mikkers, G. C. M. Van der Zon, M. Stein-Gerlach, A. Ullrich and A. Maassen, Biochem. J. 318, 609 (1996). S. M. Bockholt and K. Burridge, J. Biol. Chem. 268, 14565 (1993). A. Krause, E. A. Cowles and G. Gronowicz, J. Biomed. Mater. Res. 52, 738 (2000).

Improved blood compatibility and decreased VSMC proliferation of surface-modified metal grafted with sulfonated PEG or heparin HEE JUNG LEE 1 , JONG-KYU HONG 1 , HYUN CHUL GOO 1 , WON KYU LEE 1 , KI DONG PARK 2 , SOO HYUN KIM 1 , YOUNG MI YOO 1 and YOUNG HA KIM 1,∗ 1 Biomaterials

Research Center, Korea Institute of Science and Technology, P.O. Box 131, Cheongnyang, Seoul 130-650, Korea 2 Department of Molecular Science and Technology, University of Ajou, Suwon 442-749, Korea Received 29 October 2001; accepted 28 May 2002 Abstract—Although the technique of coronary stenting has remarkably improved long-term results in recent years, (sub)acute thrombosis and late restenosis still remain problems to be solved. Metallic surfaces were regarded as thrombogenic, due to their positive surface charges, and stenosis resulted from the activation and proliferation of vascular smooth muscle cells (VSMCs). In this study, a unique surface modification method for metallic surfaces was studied using a self-assembled monolayer (SAM) technique. The method included the deposition of thin gold layers, the chemisorption of disulfides containing functional groups, and the subsequent coupling of PEG derivatives or heparin utilizing the functional groups of the disulfides. All the reactions were confirmed by ATR-FTIR and XPS. The surface modified with sulfonated PEG (Au-S-PEG-SO3) or heparinized PEG (Au-S-PEGHep) exhibited decreased static contact angles and therefore increased hydrophilicity to a great extent, which resulted from the coupling of PEG and the ionic groups attached. In vitro fibrinogen adsorption and platelet adhesion onto the Au-S-PEG-SO3 or Au-S-PEG-Hep surfaces decreased to a great extent, indicating enhanced blood compatibility. This decreased interaction of the modified surfaces should be attributed to the non-adhesive property of PEG and the synergistic effect of sulfonated PEG. The effect of the surface modification on the adhesion and proliferation of VSMCs was also investigated. The modified Au-S-PEG-SO3 or Au-S-PEG-Hep surfaces also exhibited decreased adhesion of VSMCs, while the deposited gold layer itself was effective. The enhanced blood compatibility and the decreased adhesion of VSMCs on the modified metallic surfaces may help to decrease thrombus formation and suppress restenosis. It would therefore be very useful to apply these modified surfaces to stents for improved functions. A long-term in vivo study using animal models is currently under way. Key words: Metallic surface modification; self-assembled monolayer technology; sulfonated PEG; heparin; protein/ platelet adhesion; smooth muscle cell adhesion. ∗ To


174

H. J. Lee et al.

INTRODUCTION

The implantation of metallic stents is widely accepted as post-operative therapy to percutaneous transluminal coronary angioplasty (PTCA) [1– 4]. Although the technique of coronary stenting has remarkably improved long-term results in recent years, (sub)acute thrombosis and late restenosis still remain problems to be solved. Generally, metallic materials are regarded as thrombogenic due to their positive surface charge and/ or the high surface free energy [3, 4]. In addition, late restenosis might result mainly from the activation and proliferation of vascular smooth muscle cells (VSMCs) [1– 4]. Surface-induced thrombosis is initiated by plasma protein adsorption and platelet activation. First the adsorbed proteins may serve as sites for platelet adhesion and activation by interactions with platelet membrane receptors. At the same time, coagulation factors in the plasma are activated to form a fibrin network. Heparin is a well-established anticoagulant used clinically to inhibit the formation of the fibrin network. There have been a number of approaches to utilize heparin by either slow-releasing systems or immobilization methods [5– 10]. Heparin has actually been applied to coronary stents, which have been successful, to some extent, in decreasing the associated (sub)acute thrombosis [8– 10]. Heparin has been also reported to decrease the proliferation of VSMCs [11, 12]. Surface modification using hydrophilic polymers, especially poly(ethylene glycol) (PEG), has been shown to decrease protein adsorption and platelet/ cell adhesion on biomaterials that make contact with blood [13– 30]. The role of PEG was explained by its unique properties, such as the excluded volume on the surface and the flexible hydrophilic chain motion to expel proteins/ cells, in addition to its nontoxicity and non-immunogenecity [14, 15, 23]. Methods for the surface modification by PEG have included simple physical adsorption, a self-assembled monolayer (SAM) [16– 18, 21, 28– 30], and chemical bond formation, such as chemical coupling [20, 22, 26] or graft polymerization [14, 24]. SAM techniques using silanes or gold/ sulfur compounds have been applied to modify glass, metal, and polymers for investigation as models in several areas [31– 33]. Whitesides et al. demonstrated that a SAM composed of oligo-PEG decreased protein adsorption [16– 18]. The surface grafting of PEG by SAM techniques has been investigated not only on polymers [14, 21, 27, 39], but also on metals and glass [20, 28– 30]. All of these studies demonstrated more or less decreased adsorption of proteins and adhesion of platelets, and therefore improved blood compatibilty. Furthermore, in our previous studies, sulfonated PEG (PEG-SO3) grafted polyurethane demonstrated enhanced blood compatibility and biostability to a larger extent than PEG alone; this may be attributed to a synergistic effect of the non-adhesive and mobile PEG chain motions combined with anticoagulant active sulfonate groups [25, 26, 34– 38]. In this study, PEG or sulfonated PEG was coupled on gold surfaces by SAM technology using functional sulfur compounds. Heparin was also immobilized

Improved blood compatibility and decreased VSMC proliferation

175

onto PEG chains containing functional groups. The surfaces coupled with PEG or heparin were characterized by static contact angles, attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy, X-ray photoelectron spectroscopy (XPS), and atomic force microscopy (AFM) [28]. In vitro protein adsorption, platelet adhesion, and VSMC proliferation were investigated to evaluate blood compatibility and the effect on restenosis.

EXPERIMENTS

Materials Stainless steel plates (SS, 1 × 3 cm or 0.5 × 0.5 cm; Kistech Co., Korea) were soaked in chromic acid and washed in distilled water three times before use. Gold-coated SS (Au) was prepared as follows: gold was deposited to a thickness of about 100 µm onto SS with an ion sputter (Hitachi E-1030, Japan), using an argon plasma at about 6 Pa under 15 mA with a deposition rate of about 11 nm / min. Au was soaked in a chromic acid solution for 3 days and then washed three times in distilled water in an ultrasonic bath. Dithiobutyric acid [TBA, (SCH2 CH2 COOH)2], cystamine dihydrochloride [cystamine, (SCH2 CH2 NH2 )2 · 2 HCl], taurine (2-aminoethanesulfonic acid), EDC [1-(3-dimethylaminopropyl)-3ethylcarbodiimide hydrochloride], and triethylamine (TEA) were purchased from Aldrich Chemical Co., USA. Heparin (sodium salt, grade I-A, from porcine intestinal mucosa), bovine serum fibrinogen (Fib, type I-S), fetal bovine serum (FBS), penicillin, and streptomycin were bought from Sigma, USA. Platelet-rich plasma (PRP) was donated by the Central Red Blood Center, Korea and vascular smooth muscle cells (VSMCs) from rat artery were provided by Dr. K. H. Chung, Cardiovascular Research Institute, Yonsei University College of Medicine, Korea. Bicinochroninic acid (BCA; Pierce Chemicals, USA) protein assay kit, Dulbeco’s modified Eagle’s medium F12 (DMEM /F12; Gibco-BRL, USA), bovine serum albumin (Gibco-BRL, USA), and Denacol EX-861 (PEG 1000 with epoxy groups at both ends; Nagase Chemical Ltd., Japan) were used. Diamino-terminated PEG (H2 N-PEG-NH2, molecular weight of PEG 1000; NOF Co., Japan) was dissolved in chloroform for purification, precipitated into diethyl ether, and dried in a vacuum. Then only one amino end-group of H2 N-PEG-NH2 was sulfonated by 1,3-propane sultone as described previously [25]. Briefly, 10% w /v propane sultone in tetrahydrofuran (THF) was added dropwise to 10% w /v H2 N-PEG-NH2 solution in THF and reacted at 50 ◦ C for 6 h. The resulting product precipitated as the reaction proceeded. The filtered product was washed with cold THF and dried overnight at room temperature. The structure of sulfonated PEG (H2 N-PEG-SO3) was confirmed by a nuclear magnetic resonance spectrometer (1 H-NMR, 13 C-NMR, Varian Gemini 200 MHz).

176

H. J. Lee et al.

Immobilizing sulfonated PEG or heparin on the gold surface Figure 1 shows a schematic illustration of the surface modification using dithiobutyric acid (TBA) or cystamine. In method (a) using TBA, Au was immersed in a solution of 100 mM TBA in methanol for 12 h to obtain Au-S-COOH. The mixture was immediately placed into a 20 wt% aq. solution of H2 N-PEG-SO3 containing EDC in a glass tube and the glass tube was sealed under N2 and kept at 60 ◦ C for given periods of time. The H2 N-PEG-SO3 immobilized gold (Au-S-PEGA-SO3) was washed with distilled water three times to remove unreacted PEG derivatives and dried in N2 . In method (b) using cystamine, Au was soaked in a solution of 100 m M cystamine in methanol at room temperature for 12 h. The cystamine-adsorbed gold (Au-SNH2 ) was then rinsed thoroughly with methanol and 0.1 mM KOH to remove the physically adsorbed cystamine and HCl, respectively. After rinsing and drying at 50 ◦ C, Au-S-NH2 was treated with a 100 mM solution of Denacol EX-861 in THF with a few ml of TEA at 50 ◦ C for 48 h. The Denacol-grafted gold plate (Au-S-PEG) was rinsed with THF and soaked in a solution of 50 mM taurine in formamide at 50 ◦ C for 48 h so that the residual epoxy group on Au-S-PEG reacted with taurine to introduce a sulfonate group (Au-S-PEG-SO3). In a similar way, Au-S-PEG having residual epoxy groups was immersed in a solution of 2 m M heparin in formamide to immobilize heparin (Au-S-PEG-Hep). The resulting gold plates, Au-S-PEGSO3 and Au-S-PEG-Hep were rinsed several times with formamide and ethanol and dried in N2 . Static contact angle (θS ) measurements The static contact angles were measured by an apparatus (Digidrop, GBX Scientific Instrument, France) fitted with a video camera (Nikon, Japan). Water drops of 10 µl prepared with a micro-syringe (Perfektum® , Popper & Sons Inc., Japan) were dropped onto three different points of each modified gold surface with a contact time t = 30 s. Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy and atomic force microscopy (AFM) ATR-FTIR (Bruker FT-IR, IFS 66, Germany) was used to analyze the modified gold surfaces, using a KRS-5 crystal. The modified gold surfaces were observed by AFM (Park Scientific Instruments, CP, USA) in the tapping mode. X-ray photoelectron spectroscopy (XPS) XPS spectra of the modified gold surfaces were obtained on an ESCA 2803-S spectrometer (SSI, USA) with Al Kα X-rays. In order to determine O /C, N /C, S /C, and Na/C stoichiometries, collecting factors of 2.50, 1.68, 1.80, and 8.5 were


177

used for C1s, O1s, N1s, and S2p3, respectively. Binding energies were referenced to the C H group of the C1s peak components at 285 eV. In vitro fibrinogen adsorption test The modified gold plates were placed into 5 ml plastic disposable syringes and equilibrated with 2 ml of phosphate buffer solution (PBS) overnight. The buffer was replaced by 0.15 mg/ ml bovine serum fibrinogen (Fib), and the syringes were tapped to remove air bubbles, sealed and rotated in a shaking incubator (Daeil Engineering Co., Korea) at 37 ◦ C. By this method, the modified gold surfaces were constantly exposed to the protein solution. After 15 and 30 min, Fib-adsorbed gold plates were rinsed with PBS three times and sonicated in 1% SDS solution and centrifuged. The adsorbed protein concentration was determined using a BCA protein assay kit as described by the manufacturer. In brief, BCA working reagent was prepared by combining 50 parts of buffer solution A, 48 parts of BCA stock solution B (4% w /v BCA in water), and two parts of 4% w /v CuSO4 · H2 O. A set of protein standards (bovine serum albumin) was prepared using PBS as a diluent. A 100 µl volume of each standard sample was pipetted into a 96-well plate. The well plate was incubated at 60 ◦ C for 1 h. The absorbance of the water-soluble purple product was measured at 562 nm by an ELISA (enzyme-linked immunosorbent assay) apparatus (Spectra Max 340, Molecular Device Inc., USA). In vitro platelet adhesion test The surface-modified gold plates were incubated first with PBS and then with PRP by the same method as the above fibrinogen adsorption test. The number of platelets in PRP was 5 × 104 cells/ µl. The depleted platelets in the PRP were counted with a hemacytometer (Bright-Line® , Reichert, USA). The morphologies of the surfaces to which the platelets adhered were observed with a scanning electron microscope (SEM, HI-TACHI S-510). In vitro VSMC proliferation study Vascular smooth muscle cells (VSMCs) between the third and fifth passages were incubated at 37 ◦ C for 3 days in DMEM /F12 supplemented with 10% FBS, 100 U /ml penicillin, and 100 µg/ ml streptomycin. Meanwhile, the modified gold surfaces were sterilized with 70% ethyl alcohol for 24 h and rinsed with PBS three times before seeding. SMCs (number of cells 1×105 ) were gently seeded onto each plate located in a 24-well plate. After 1, 2, and 4 days, the cells on the surfaces were detached by trypsin– EDTA treatment and counted by a hemacytometer. As a control, a tissue culture dish (24-well plate, Falcon, USA) was used. The morphologies of the cells adhered on the surfaces were observed by SEM.

178

H. J. Lee et al.

RESULTS AND DISCUSSION

Characterization of the modified surfaces A number of methods have been studied for the incorporation of PEG derivatives onto glass, metallic, or polymeric surfaces. As glass and metals do not have functional groups, several approaches were studied: plasma coating or plasma polymerization onto the surfaces; coupling of various silane compounds; and chemisorption of sulfur compounds onto deposited gold layers. Self-assembled monolayers (SAMs) using silanes or sulfur compounds and their applications have been reviewed in detail by Ulman [33]. Silane coupling agents are well known to modify the surface of glass or metal, and recently several groups have demonstrated the incorporation of PEG, using silanes [20, 29, 46]. Gold and silver are very stable against oxidation and other chemical reactions, but they adsorb sulfur compounds very strongly to form quite stable coupling with thiolates, sulfides, or disulfides. Whitesides [16– 18] applied SAMs using thiolated oligo-PEG on gold surfaces to demonstrate the effect of PEG in decreasing protein adsorption; this technology was extended to several studies [28, 30, 39, 40]. Figure 1 illustrates the surface modification scheme. In method (a), dithiobutyric acid (TBA) was adsorbed onto gold layers to yield Au-S-COOH. The COOH

Figure 1. Schematic illustration of the surface modification process.


179

group introduced was utilized to react with the NH2 group of H2 N-PEG-SO3 to incorporate the sulfonated PEG (Au-S-PEGA-SO3), where EDC was used as a catalyst. The adsorption and reaction were confirmed by FTIR: a COOH peak at 1740 cm−1 or/ and SO3 group at 1342– 1352 cm−1 (asymmetric SO2 stretch), 1165– 1150 cm−1 (symmetric SO2 stretch) and 910– 895 cm−1 (S – O stretch), respectively, as shown in Figs 2a and 2b. In method (b), cystamine containing NH2 groups was adsorbed to obtain Au-S-NH2 , and PEG containing epoxy groups at both ends (Denacol EX-861) was coupled to yield Au-S-PEG via a reaction between NH2 and epoxy groups, which was catalyzed by an amine (TEA). In addition, the other residual epoxy group of PEG was further reacted with taurine to introduce sulfonate groups (Au-S-PEG-SO3 ), or to immobilize heparin (Au-SPEG-Hep). All the above reactions were confirmed again by ESCA; the surface atomic compositions are presented in Table 1. In Table 1, it can be seen that Au-SCOOH and Au-S-NH2 before coupling of the PEG derivatives exhibited relatively large S contents, but the values decreased substantially after the coupling of PEG. The C, O, and S concentrations of the Au-S-COOH surface were 66.8%, 23.7%, and 9.50%, respectively; those of the Au-S-PEGA-SO3 surface were 63.5%, 31.1%, and 5.40%. The grafted sulfonated PEG decreased the C and S concentrations, but increased the O content. These changes indicated the surface coverage of TBA or sulfonated PEG. In the case of Au-S-NH2 , Au-S-PEG, and Au-S-PEG-Hep, the changes in the surface atomic concentrations were similar, but N was not detected on the Au-S-PEG surface. It can be concluded from the FTIR and ESCA data that all the above reactions proceeded as planned.

Figure 2. ATR-FTIR spectra of the modified surfaces; (a) Au-S-COOH, (b) Au-S-PEG-SO3 .

180

H. J. Lee et al. Table 1. Elemental compositions of the modified gold surfaces with sulfonated PEG and heparinized PEG Atomic %a

Samples

Au-S-COOH Au-S-PEGA-SO3 Au-S-NH2 Au-S-PEG Au-S-PEG-SO3 Au-S-PEG-Hep

C

O

S

N

66.8 63.5 62.6 54.0 58.7 54.1

23.7 31.1 22.3 43.8 32.3 31.6

9.50 5.40 9.4 2.2 3.5 5.2

— — 5.8 — 5.5 9.1

O/Cb

S/Cb

0.36 0.49 0.36 0.81 0.55 0.58

0.14 0.085 0.15 0.04 0.094 0.097

a Analyzed b Surface

from survey scan spectra. atomic ratio. Table 2. Static contact angle (θS ) data of the modified gold surfacesa Samples

θS

SS Au Au-S-COOH Au-S-PEGA-SO3 SS Au Au-S-NH2 Au-S-PEG Au-S-PEG-SO3 Au-S-PEG-Hep

56.3 ± 1.5 26.1 ± 1.3 43.0 ± 1.9 wetting 56.3 ± 1.5 47.6 ± 2.6 51.5 ± 2.2 40.0 ± 2.3 13.9 ± 1.0 wetting

a Unit

degree, Average ± standard deviation, n = 5.

Such changes of the surface atomic concentrations described above were directly reflected by the water contact angles, which resulted from the change in surface hydrophilicity. Table 2 lists the static contact angles of the modified gold surfaces. The contact angle of SS was 56.3◦ and those of Au were 26.1◦ and 47.6◦ , indicating that the Au surface may be very sensitive to dust and impurities. Au-S-COOH and Au-S-NH2 showed a moderate decrease of the contact angles, similar to the case of Au-S-PEG containing epoxy end-groups. On the contrary, the Au-S-PEGASO3 , Au-S-PEG-SO3, and Au-S-PEG-Hep surfaces demonstrated very low contact angles, indicating the high hydrophilicity due to the PEG chains and attached ionic end-groups. Furthermore, Au-S-PEGA-SO3 and Au-S-PEG-Hep were completely wetted. Such an increased hydrophilicity of the modified gold surfaces again confirmed the actual surface modification processes. The topographies of the modified gold surfaces were examined by AFM. The control (Au) surface was relatively smooth and had a specific texture (Fig. 3a).


181

Figure 3. AFM images of the modified surfaces; (a) Au, (b) Au-S-PEG-SO3 , (c) Au-S-PEG-Hep: RMS roughness; (a) 2.231, (b) 1.070, (c) 1.129 nm.

Figures 3b and 3c show the images of the Au-S-PEG-SO3 and Au-S-PEG-Hep surfaces, respectively. They were even smoother than the control. The root-meansquare (RMS) value of the control Au surface was 2.231 nm, but those of Au-SPEG-SO3 and Au-S-PEG-Hep decreased to 1.070 and 1.129 nm, respectively. The RMS data revealed that the modified gold surfaces became even smoother due to the coverage of the incorporated PEG chains. SAM technology using the chemisorption of sulfur compounds onto gold layers provides various methods of introducing a wide range of functional groups on inert metallic surfaces. In this study, a unique method to couple PEG and/ or heparin onto a gold layer deposited on SS was investigated. The PEG or heparin compound grafted on the surface was quite stable during the various characterizations or measurements. Recently, several groups have investigated the stability of the coupling between gold and sulfur compounds [41– 43]. Wirde reported that the coupling was not so stable against oxidation in particular [41]. However, Lee observed that it was stable for a month at room temperature [43]. A study on the long-term stability should be carried out. In vitro protein adsorption and platelet adhesion test When biomaterials make contact with blood, proteins are first adsorbed instantaneously onto the surfaces and deformed. Then platelets adhere to the surfaces, are activated and aggregate, so that platelets may play a major role in thrombus formation. Therefore, a study on protein adsorption and platelet adhesion is the first step in evaluating the blood compatibility of biomaterials [23]. The effect of the modified gold surfaces on protein adsorption was examined using fibrinogen (Fib) as a model protein. The results are presented in Fig. 4. The concentrations of adsorbed Fib were measured after 15 min and 30 min, although the values were not very different. The amount of Fib adsorbed after 30 min onto the Au surface was 0.50 µg/ cm2 (89% of that on SS), slightly smaller than the value on SS (0.56 µg/ cm2 , 100%).

182

H. J. Lee et al.

Figure 4. Fibrinogen adsorption on the modified surfaces.

Figure 5. Platelet adhesion on the modified surfaces.

The Fib concentrations of Au-S-PEG-SO3 and Au-PEG-Hep surface decreased further to 0.23 (41%) and 0.20 µg/ cm2 (39%), respectively, demonstrating the effect of the introduced PEG-SO3 or PEG-Hep. Figure 5 shows the results of in vitro platelet adhesion onto the modified gold surfaces for 15 and 30 min. The percentage of adhered platelets for the 30 min period is higher than that for 15 min. However, the tendency of platelet adhesion


183

was very similar to protein adsorption. The Au surface presented a smaller percentage of adhered platelets (31% for 30 min) than SS (42%). The Au-SPEG-SO3 (20%) and Au-S-PEG-Hep (13%) surfaces exhibited much less platelet adhesion than the Au surface for 30 min. The Fib adsorption and platelet adhesion on the modified surfaces decreased in the order SS > Au > Au-S-PEG-SO3 > Au-PEG-Hep. Furthermore, the surface modifications using both sulfonated PEG and heparinized PEG were very effective in preventing protein adsorption and platelet adhesion onto the modified gold surface. This effect of PEG or heparin should be attributed to the specific function of PEG or hydrophilic heparin. Hydrophilic PEG chains grafted on surfaces inhibit the attachment of proteins or platelets by an excluded volume effect and flexible dynamic chain motions, as explained in other studies [14, 15]. However, the extent of the decrease of platelet adhesion, or, especially, protein adsorption, may be regarded as not very much compared with other studies. In our previous study on the protein adsorption of modified PU, the amounts of Fib adsorbed on both PU-PEG and PU-PEG-SO3 were much smaller than on PU; PU-PEG-SO3 exhibited a smaller Fib value but a higher amount of adsorbed albumin than those on PU-PEG [35], although PU-PEG-SO3 exhibited better in vivo blood compatibility than PU-PEG [34]. The advantage of the PEG-SO3 grafted system compared with only PEG may be the heparin-like anticoagulant activity of PEG-SO3 itself (14% of free heparin) [36], further decreased platelet adhesion, and increased hydrophilicity, indicating a synergistic effect of PEG chains and SO3 groups [34– 36]. In a different previous study on the platelet adhesion of modified PU, PU-PEG-SO3 always showed a smaller value than PUPEG for various MWs of PEG [25]. Heparin inhibits the function of thrombin and at the same time decreases platelet adhesion, possibly due to hydrophilicity first of all [7]. In vitro vascular smooth muscle cell (VSMC) proliferation study It was investigated whether the late restenosis associated with coronary stents resulted from the stimulation and proliferation of VSMCs after stenting in vivo [2, 3]. Heparin has been reported not only to act as an anticoagulant, but also to decrease VSMC proliferation in vitro and in vivo [11, 12], as heparin may interfere with the action of growth factors in addition to clotting factors or it may act directly on cells to inhibit their division. In addition, PEG and especially PEG-SO3 exhibited decreased interactions with cells and bacteria [25, 37, 44]. Therefore, in vitro VSMC adhesion and proliferation on the modified surfaces were investigated and the results are summarized in Fig. 6. The numbers of VSMCs that adhered to and grew on the modified gold surfaces were compared. When VSMCs (number of cells 1 × 105 ) were seeded on the 24-well culture plate (control), the number of SMCs was 1 × 105 at day 1, but increased to 1.5 × 105 at day 2 and 2.8 × 105

184

H. J. Lee et al.

Figure 6. Vascular smooth muscle cell adhesion and proliferation on the modified surfaces.

at day 4. However, when the same number of SMCs was seeded onto the surface of Au, Au-S-NH2 , Au-S-PEG-SO3, and Au-S-PEG-Hep, respectively, only about 10% of the total SMCs adhered to each surface and they did not proliferate at all. Actually, it was expected that Au-S-PEG-SO3 or Au-S-PEG-Hep might exhibit decreased adhesion of SMCs. We have previously reported that surfaces grafted with PEG or PEG-SO3 decreased the adhesion of corneal stromal cells [44] or bacteria [25], due to the non-adhesive property of PEG or PEG-SO3. Heparin has also been reported to decrease VSMC proliferation as discussed above [11, 12]. However, in this study VSMCs did not adhere on grow so much either on Au or on Au-S-NH2 , indicating a specific effect of the gold surface itself on the adhesion of SMCs. Actually, Sheardown reported that deposited gold surfaces revealed the adhesion of few cells [40]. In addition, gold-coated stents were claimed to be effective against allergies [45]. Therefore, the decreased adhesion of SMCs to Au or Au-S-NH2 might be due to the specific character of gold. The interaction of VSMCs with deposited gold, PEG-SO3 or introduced heparin should be further investigated in detail, maybe by animal models. The surface areas of the metallic sheets applied here might be too small to investigate differences in cell adhesion and proliferation. This unique surface modification method for metallic materials presented a successful incorporation of PEG and/ or heparin demonstrating decreased protein adsorption and platelet adhesion. In addition, the modified surfaces revealed decreased VSMC adhesion and maybe proliferation. Therefore, this method would be very useful to apply to stent coating. A long-term in vivo study using animal models is currently under way.


185

CONCLUSION

Although the technique of coronary stenting has remarkably improved long-term results in recent years, (sub)acute thrombosis and late restenosis still remain problems to be solved. Restenosis was suggested, due to the activation and proliferation of vascular smooth muscle cells. In this study, a unique surface modification method for metallic surfaces was studied using a SAM technique. The method included the deposition of thin gold layers, the chemisorption of disulfides containing functional groups, and the subsequent coupling of PEG derivatives or heparin utilizing the functional groups of the disulfides. All the reactions were confirmed by ATR-FTIR and XPS. The modified surface with sulfonated PEG (Au-S-PEG-SO3) or heparinized PEG (Au-S-PEG-Hep) presented very low contact angles or complete wetting, and therefore high hydrophilicity. In vitro fibrinogen adsorption and platelet adhesion onto the Au-S-PEG-SO3 or Au-S-PEGHep surfaces decreased to a great extent, indicating enhanced blood compatibility. Such a decreased interaction of the modified surfaces can be attributed to the nonadhesive property of PEG and the synergistic effect of the sulfonated PEG. The effect of the surface modification on the adhesion and proliferation of SMCs was also investigated. The modified Au-S-PEG-SO3 or Au-S-PEG-Hep surfaces also exhibited decreased adhesion of VSMCs, while the deposited gold layer itself was effective. The enhanced blood compatibility and less adhesion and maybe proliferation of VSMC on the modified metallic surfaces might help to decrease thrombus formation and suppress restenosis; therefore it would be very useful to apply to stents with an improved function. A long-term in vivo study using animal models is currently under way. Acknowledgement This work was supported by the Korean Ministry of Science and Technology National Research Laboratory Program No. N19610 and N21820. The authors appreciate Dr. K. H. Chung, Cadiovascular Research Institute, Yonsei University College of Medicine, Korea for supplying vascular smooth muscle cells. We thank also to Nagase Chemical Ltd., Japan for providing Denacol EX-861. REFERENCES 1. 2. 3. 4. 5. 6. 7.

T. Peng, P. Gibula, K. Yao and M. F. A. Goosen, Biomaterials 176, 685 (1996). D. L. Fischman, M. B. Lean, D. S. Daim et al., New England J. Med. 331, 496 (1994). J. C. Palmaz, Am. J. Roentgenol. 160, 613 (1993). P. A. Ribeiro, R. Gallo, J. Antonius et al., Am. Heart J. 125, 501 (1993). K. D. Park, T. Okano, C. Nojiri and S. W. Kim, J. Biomed. Mater. Sci. 22, 977 (1988). M. Pekna, R. Larsson, B. Formgren, U. R. Nilsson and B. Nilsson, Biomaterials 14, 189 (1993). D. K. Han, S. Y. Jeong and Y. H. Kim, J. Biomed. Mater. Res.; Appl. Biomaterials 23 (A2), 211 (1989). 8. W. J. van der Giessen, H. M. M. van Beusekom et al., Semin. Intervent. Cardiol. 3, 173 (1998). 9. R. Blezer, L. Cahalan, P. T. Cahalan and T. Lindhout, Blood Coagulation & Fibronolysis 9, 435 (1998).

186

H. J. Lee et al.

10. P. W. Serruys, H. Emanuelsson, W. J. van der Giessen et al., Circulation 93, 412 (1996). 11. A. D. Snow, R. P. Bolender, T. N. Wight and A. W. Clowes, Am. J. Pathol. 137, 313 (1990). 12. L. A. Pukac, G. M. Hirsch, J. C. Lormeau, M. Petitou, J. Choay and M. J. Karnovsky, Am. J. Pathol. 139, 1501 (1991). 13. E. W. Merill and E. W. Salzman, Am. Soc. Artif. Intern. Org. J. 6, 60 (1983). 14. Y. Mori, S. Nagaoka, H. Takiuchi, T. Kikuchi, N. Noguchi, H. Tanzawa and Y. Noshiki, Trans. Am. Soc. Artif. Intern. Org. 28, 459 (1982). 15. J. H. Lee, H. B. Lee and J. D. Andrade, Prog. Polym. Sci. 20, 1043 (1995). 16. C. Pale-Grosdemage, E. S. Simon, K. L. Prime and G. M. Whitesides, J. Am. Chem. Soc. 113, 12 (1991). 17. K. L. Prime and G. M. Whitesides, J. Am. Chem. Soc. 115, 10714 (1993). 18. P. Harder, M. Grunze, R. Dahint, G. M. Whitesides and P. E. Laibinis, J. Phys. Chem. B 102, 426 (1998). 19. P. Claesson, Colloids Surf. A 77, 109 (1993). 20. T. B. McPherson, H. S. Shim and K. Park, J. Biomed. Mater. Res. Appl. Biomater. 38, 553 (1997). 21. G. Mao, D. W. Grainger and D. Castner, Polym. Prep. Am. Chem. Soc. Div. Polym. Chem. 38, 289 (1997). 22. K. Holmberg, K. Bergstrom, C. Brink, E. Osterberg, F. Tiberg and J. M. Harris, J. Adhesion Sci. Technol. 7, 503 (1993). 23. W. R. Gombotz, W. Guanghui, T. A. Horbett and A. S. Hoffman, J. Biomed. Mater. Res. 25, 1547 (1991). 24. Y. X. Qui, D. Klee, W. Pluster, B. Severich and H. Hocker, J. Appl. Polym. Sci. 61, 2373 (1996). 25. H. J. Lee, K. D. Park, Y. H. Kim et al., Colloids Surf. B 18, 355 (2000). 26. Y. G. Ko, Y. H. Kim et al., Biomaterials 22, 2115 (2001). 27. F. Lim, X.-H. Yu and S. L. Cooper, Biomaterials 14, 537 (1993). 28. Y. H. Kim, H. J. Lee et al., Trans. Soc. Biomaterials 24, 7 (2001). 29. F. Zhang, E. T. Kang, K. G. Neoh and W. Huang, J. Biomater. Sci. Polymer Edn 12, 515 (2001). 30. F. Zhang, E. T. Kang, K. G. Neoh, P. Wang and K. L. Tan, Biomaterials 22, 1541 (2001). 31. C. D. Bain and G. M. Whitesides, Science 240, 62 (1988). 32. C. D. Bain and G. M. Whitesides, J. Am. Chem. Soc. 111, 7164 (1989). 33. A. Ulman, Chem. Rev. 96, 1533 (1996). 34. D. K. Han, S. Y. Jeong, Y. H. Kim, B. K. Min and H. I. Cho, J. Biomed. Mater. Res. 25, 561 (1991). 35. D. K. Han, G. H. Ryu, K. D. Park, U. Y. Kim, B. G. Min and Y. H. Kim, J. Biomed. Mater. Res. 30, 23 (1996). 36. D. K. Han, N. Y. Lee, K. D. Park, S. Y. Jeong, Y. H. Kim, H. I. Cho and B. G. Min, Biomaterials 16, 467 (1995). 37. D. K. Han, K. D. Park, S. Y. Jeong, Y. H. Kim, U. Y. Kim and B. G. Min, J. Biomed. Mat. Res. 27, 1063 (1993). 38. K. D, Park, Y. H. Kim et al., Biomaterials 19, 851 (1998). 39. Y. J. Du and J. L. Brash, Trans. Soc. Biomaterials 22, 71 (1999). 40. R. McMillan, B. Meeks, F. Bensebaa, Y. Deslandes and H. Sheardown, J. Biomed. Mater. Res. 54, 272 (2001). 41. M. Wirde, U. Gelius and L. Nyholm, Langmuir 15, 6370 (1999). 42. R. Voicu, T. H. Ellis, H. Ju and D. Leech, Langmuir 15, 8170 (1999). 43. Y. S. Shon and T. R. Lee, J. Phys. Chem. B 104, 8192 (2000). 44. M. K. Kim, I. S. Park, H. D. Park, W. R. Wee, J. H. Lee and Y. H. Kim, J. Cataract Refract. Surg. 27, 766– 774 (2001). 45. E. Alt, US Patent No. 5,824,045 (Oct. 20, 1998). 46. S. Jo and K. Park, Biomaterials 21, 605 (2000).

Characterization of poly(ethylene oxide) brushes on glass surfaces and adhesion of Staphylococcus epidermidis HANS J. KAPER 1 , HENK J. BUSSCHER 1,∗ and WILLEM NORDE 1,2 1 Department

of Biomedical Engineering, University of Groningen, Antonius Deusinglaan 1, 9713 AV Groningen, The Netherlands 2 Laboratory of Physical Chemistry and Colloid Science, Wageningen University, Dreijenplein 6, 6703 HB Wageningen, The Netherlands Received 28 January 2002; accepted 15 July 2002 Abstract—Poly(ethylene oxide) brushes have been covalently bound to glass surfaces and their presence was demonstrated by an increase in water contact angles from fully wettable on glass to advancing contact angles of 54◦ , with a hysteresis of 32◦ . In addition, electrophoretic mobilities of glass and brush-coated glass were determined using streaming potential measurements. The dependence of the electrophoretic mobilities on the ionic strength was analyzed in terms of a softlayer model, yielding an electrophoretic softness and fixed charge density of the layer. Brush-coated glass could be distinguished from glass by a 2– 3-fold decrease in fixed charge density, while both surfaces were about equally soft. Adhesion of Staphylococcus epidermidis HBH276 to glass in a parallel plate flow chamber was extremely high and after 4 h, 19.0 × 106 bacteria were adhering per cm2 . In contrast, the organisms did not adhere to brush-coated glass, with numbers below the detection limit, i.e. 0.1 × 106 per cm2 . These results attest to the great potential of polymer brushes in preventing bacterial adhesion to surfaces. Key words: Polymer brushes; poly(ethylene oxide); electrophoretic mobility; Staphylococcus epidermidis; bacterial adhesion.

INTRODUCTION

Biomaterials are widely used in modern medicine or the production of artificial organs and in a variety of intra- and extra-corporeal prostheses. However, their application can give rise to biomaterial-centered infections (BCIs) that defy treatment with antibiotics. Therefore, the only effective remedy appears to be the removal of the infected device [1]. The development of biofilms causing BCIs occurs in several ∗ To

whom correspondence should be addressed. Tel.: (31-50) 363-3140; Fax: (31-50) 363-3159; e-mail: [email protected]

188

H. J. Kaper et al.

steps: firstly, a substratum surface is covered with a conditioning film of surfaceactive molecules, in particular proteins. Secondly, bacteria are transported towards the substratum and adhere. Initial adhesion is through long-range interactions, acting over distances up to a few tens of nanometers between the bacterium and the substratum, after which short-range interactions acting on a sub-nanometer scale become operative. The long-range interactions in bacterial adhesion may be calculated by applying concepts from colloid and interface science [2]. Thus, the DLVO theory, originally formulated to describe the adhesion of inanimate colloidal particles, has been applied to bacterial adhesion with varying success. According to the DLVO theory, the overall bacterium – substratum interaction is governed by contributions from Lifshitz– Van der Waals forces and forces resulting from overlapping electrical double layers. The description of short-range interactions (e.g. hydrogen bonding, ion pairing, hydrophobic interaction) depends on detailed knowledge of the chemistry for each surface involved and such information is usually not available for biological surfaces. Two strategies may be followed to reduce the risk of BCIs and to prevent or delay the adhesion of infectious bacteria to the substratum surface: (i) modifying the surface with (charged) groups that make the surface less attractive for the bacteria [3]; and (ii) introducing steric hindrance that keeps the bacteria at a distance from the surface where long-range attractive interaction forces are reduced to an ineffective magnitude [4, 5]. Steric hindrance may be achieved by decorating the surface with polymer molecules that are attached through an anchor to the surface, whereas the other part (the buoy) is moving freely in the surrounding medium. When the density of the polymer is high enough, the polymer molecules are forced to stretch out and the resulting layer is called a ‘molecular brush’ (see Fig. 1). The brush is essentially penetrable for solvent and low-molecular-weight ions, but depending on its packing density σ and extension L0 , may prevent the deposition of larger components such as protein molecules and bacteria [6]. So far, most of the research, both experimental and theoretical, on biomedical applications of molecular brushes has focused on the prevention of protein adsorp-

Figure 1. Schematic representation of polymer brush molecules. Upon increasing the grafting density, the polymer molecules are forced to stretch out.

Characterization of poly(ethylene oxide) brushes on glass surfaces

189

tion using poly(ethylene oxide) (PEO) as the polymer material. Work on bacterial adhesion includes the experiments by Park et al. [7], who grafted PEO molecules with different lengths and end-groups to a polyurethane surface, and Vacheethasanee and Marchant [5], who modified polyethylene and pyrolytic graphite substrata with a surfactant consisting of a poly(vinyl amine) backbone with grafted PEO and hexanal. In both experiments, surface modification with PEO molecules of sufficient length and grafting density was effective in preventing bacterial adhesion. Several theoretical models exist to predict the interaction of a tethered layer with incoming particles as a function of its parameters, such as the grafting density and polymer chain length [8– 11]. Jeon et al. [8, 9] assumed that the polymer chain is uniformly stretched with the end points located at the layer– solvent boundary. This corresponds to long and densely grafted polymer molecules. Szleifer [10] has pointed out that these assumptions do not hold in many practical situations. He has developed a model based on a self-consistent mean field approach. In this approach, interactions such as the intramolecular interactions between the monomer elements and the interactions between the brush polymer and the protein are taken into account. The model has been worked out for a PEO –lysozyme system. It was found that protein adsorption depends mainly on the grafting density, whereas the thickness of the grafted layer influences the kinetics of the adsorption process. It is doubtful whether this conclusion also holds for bacterial adhesion. Proteins are small enough to reach the substratum by diffusing through the layer. Bacteria, however, can adhere only by compressing the brush layer. Halperin [11] adopted a simple model for the polymer molecule, which was described as a string of non-interacting monomer elements. According to this model, an incoming particle may penetrate the brush and adsorb in the (absolute) primary minimum at the substratum, or it may be trapped in the secondary minimum at the aqueous edge of the brush. Various techniques, offering different degrees of control over the brush lengths versus densities, have been applied to attach the polymer molecules to a surface [12]. These techniques may be divided into non-covalent attachment, where the anchor has a high physico-chemical affinity for the substratum surface, and chemical grafting, where the anchor forms a chemical bond with surface groups. As the binding energy of a chemical bond exceeds that of a physical bond by approximately an order of magnitude, chemically grafted brushes are likely to be more robust. A technique that is often used for chemical grafting is the ‘grafting from’ method, where polymer molecules are grown from the substratum. This can be accomplished, for instance, by covering the surface with covalently linked initiators from which polymer chains grow in a solution containing the monomers, resulting in a polydisperse brush [13], or by using radio-frequency glow discharge plasma deposition to coat the surface with a thin, covalently bound polymer layer [14]. In the latter case, however, depending on the reaction conditions, the polymer molecules will form many cross-links and the resulting layer is sometimes referred to as a ‘surface gel’.

190

H. J. Kaper et al.

Figure 2. Vinyl-terminated poly(ethylene oxide) covalently bound to a hydrated silica surface.

Polymer molecules can be end-grafted to silicon and silica surfaces using an ether bond. This is an example of the ‘grafting to’ technique. Zhu et al. [15] employed a two-step procedure to link the hydroxyl end-group of poly(ethylene oxide) molecules with an ether bond to a silicon surface. Using an even simpler method, Maas et al. [16] linked vinyl-terminated polystyrene molecules to a silica surface by letting the vinyl groups react with the hydroxyl groups at a silica surface that appear as a result of the chemisorption of water to form ether bonds. Grafting densities, σ −1 , of 0.47 and 0.33 nm−2 were obtained using polystyrene of 2000 and 20 000 Da molar mass, respectively. These densities are high enough (i.e. σ −1 > RF2 , where RF is the Flory radius) to make the polystyrene chains overlap so that they stretch out to form a brush. The aim of this study was to graft methacryl-terminated monodisperse poly(ethylene oxide) to glass by linking its vinyl end-group to the hydroxyl groups on the glass surface (see Fig. 2), using the method described by Maas et al. [16], and to assess the usefulness of the resulting brush in preventing initial bacterial adhesion as the onset of infection [1]. The grafted polymer layers were physico-chemically characterized by water contact angles and streaming potentials. The adhesion experiments were carried out in a parallel plate flow chamber using S. epidermidis HBH 276. This bacterial species was chosen because it is among the most relevant strains found in BCIs.


Poly(ethylene oxide) (PEO) brushes Methacryl-terminated poly(ethylene oxide) with a molar mass of 9800 Da, corresponding to approximately 250 monomer units, and a polydispersity index less than 1.03 was purchased from Polymer Source (Dorval, Quebec, Canada) and used as received. Microscope glass slides of size 76 × 26 × 1 mm (Menzel-Gläser) were used as a substratum surface. The slides were first sonicated in 2% RBS detergent (Omnilabo International, Breda, The Netherlands), rinsed in warm tap water and


191

demineralized water, sonicated again in methanol, and rinsed in demineralized water, to remove oil contamination and fingerprints. Next, possible metallic oxides on the glass surface were removed by submersing the slides in hot (95 ◦ C) nitric acid (65%; Merck, Darmstadt, Germany) for 45 min. Finally, the slides were extensively rinsed with demineralized water and Millipore-Q water and dried in a heat box at 80 ◦ C for 5 h. To graft the PEO chains on the glass surfaces, the slides were covered with a solution of the methacryl-terminated PEO in chloroform (4 mg/ ml). The solvent was evaporated in a stream of nitrogen, after which the slides were annealed overnight in a vacuum at 145 ◦ C. Prior to experiments, excess material was removed by washing the slide with Millipore-Q water and the slides were dried in a stream of nitrogen. Water contact angle measurements Water contact angles were measured at room temperature with a home-made contour monitor using the sessile drop technique. Advancing and receding water contact angles were obtained by keeping the syringe needle in the water droplet (1– 1.5 µl) after positioning it on the surface and by carefully moving the sample until the advancing angle was maximal. Contact angles with water droplets at rest will be referred to as equilibrium contact angles. On each sample, at least ten droplets were placed at different positions. Streaming potential measurements For a solid surface in contact with a liquid, streaming potentials Estr , arising from a forced flow of the liquid under the influence of a pressure p, depend on the electrokinetic potential ζ at the solid– liquid interface according to Estr εε0 = ζ, p ηκsp

(1)

where εε0 is the dielectric permittivity, η is the viscosity and κsp is the specific conductivity of the liquid. By comparison with the electrokinetic model of colloidal particles, the electrophoretic mobility µ for a flat surface is related to the electrokinetic potential ζ by [17] εε0 ζ. (2) µ= η Combining equations (1) and (2) yields the connection between µ and Estr : µ = κsp

Estr . p

(3)

The pressure dependence of the streaming potentials was measured employing a parallel plate flow chamber [18]. The walls of the flow chamber were brush-coated

192

H. J. Kaper et al.

microscope slides (26 × 76 mm) separated by a 0.2 mm Teflon gasket, while two rectangular platinum electrodes (5.0 × 25.0 mm) were located at both ends of the parallel plate flow chamber. Streaming potentials were measured in KCl solutions with ionic strengths of 5, 10, 15, 25, 50, 75 and 100 m M at ten different pressures ranging from 50 to 200 hPa. Each pressure was applied for 10 s in both directions. Ohshima et al. [19] have proposed a model for the electrophoretic mobility of particles with fixed surface charges distributed across an ion-penetrable, porous layer. It unites the theory of electrophoresis of coiled structures of highly solvated polyelectrolytes with that of rigid spheres where the surface charge is located in an infinitesimal thin layer at the surface. In the Ohshima model, the ion-penetrable layer is characterized by its fixed charge density ρ and a parameter 1/λ, referred to as the electrophoretic ‘softness’ of the ion-penetrable layer, which depends on the frictional force exerted on the water when it flows through the ion-penetrable layer. For planar surfaces, under the conditions that (a) the charge densities are relatively low, (b) 1/λ is less than the thickness over which the liquid flow penetrates the soft surface layer and (c) the Debye length, κ −1 , is less than the thickness of the ion-penetrable layer (all being fair assumptions for PEO-grafted surfaces and to a somewhat lesser extent also for glass having a porous, jelly surface in aqueous media with a wide range of ionic strengths), the electrophoretic mobility as a function of the reciprocal Debye length κ is approximated by 2 λ 1 + λ/2κ ρ . (4) µ= 2 1+ ηλ κ 1 + λ/κ For symmetrical electrolytes, κ is related to the ionic strength as κ2 =

2F 2 ci zi2 , εε0 RT

(5)

where F is the Faraday constant, T is the absolute temperature, R is the gas constant, zi is the valency and ci is the concentration of ion i. The most salient feature of equation (4) is the fact that in contrast to the rigid surface model, the electrophoretic mobility does not approach zero as the electrolyte concentration increases. A least-squares fit of electrophoretic mobilities measured as a function of the ionic strength to equation (4) allows the evaluation of the softness of the polymer layer and the space charge density in the soft part of the layer. This is based on the assumption that the values for 1/λ and ρ are invariant with ionic strength. All electrophoretic mobilities and softness values reported are the mean values of three different measurements with separately prepared brushes. Bacterial strain and adhesion experiments Staphylococcus epidermidis HBH276 was cultured in tryptone soya broth (OXOID, Basingstoke, UK). First, a strain was streaked from a frozen stock and grown overnight at 37 ◦ C on a blood agar plate. A colony was used to inoculate 5 ml of


193

Figure 3. Schematic view of the parallel plate flow chamber used in this study.

growth medium, which was incubated at 37 ◦ C in ambient air for 24 h and used to inoculate a second culture in 150 ml of growth medium that was grown for 17 h. The bacteria from the second culture were harvested by centrifugation (5 min, 5000 g) and washed twice with Millipore-Q water. Subsequently, the bacteria were resuspended in PBS (pH 6.8). The suspension was sonicated on ice (10 s) to disrupt aggregates. The concentration of bacteria was determined using a Bürker-Türk counting chamber and adjusted to 3 × 108 bacteria/ ml. Adhesion experiments were carried out using a flow chamber and an image analysis system. Figure 3 shows an exploded view of the flow chamber used. The Teflon gasket between the upper and the lower part of the flow chamber determines the dimensions of the flow channel (175 × 17 × 0.75 mm). The top and bottom collector plates have the dimensions of a common microsope slide: 76×26×1 mm. The top slide is made out of glass. The bottom slide is covered with the surface under study and microbial adhesion can be directly observed using a phase-contrast microscope equipped with a 40× ultra-long working distance objective. A pulse free flow (0.0325 ml/s) was created by hydrostatic pressure and the suspension was recirculated using a peristaltic pump. By means of a valve system, the flasks containing buffer or bacterial suspension can be connected with the flow chamber. Prior to an experiment, all air bubbles were removed from the tubing and the flow chamber and the system was perfused for 60 min with buffer. Subsequently, flow was switched to the bacterial suspension. During deposition, images of the bottom plate were recorded using a 512 × 512 pixel CCD camera and the bacteria present on the surface were counted by dedicated image analysis software [20]. On the one hand, the lower detection limit of the system is determined by the number of CCD pixels per bacterial cell that are necessary for the automatic counting of the bacteria, and on the other hand, by the statistical error in the counting

194

H. J. Kaper et al.

of the bacteria that one is willing to accept. For the current system, a statistical error of 10% results in a lower detection limit of approximately 0.1 × 106 bacteria per cm2 . The initial adhesion experiments were carried out for 4 h, with bacteria suspended in PBS. The initial increase in the number of adhering bacteria was linearly extrapolated to t = 0 to obtain the initial bacterial deposition rate j0 , which represents the number of bacteria transported by convection and diffusion towards the substratum surface that subsequently have been able to adhere. RESULTS AND DISCUSSION

Physico-chemical characterization of the polymer brush Cleaned glass, i.e. the glass surface prior to application of the polymer, was fully wettable with water, while after application of the polymer, the surface was more hydrophobic (see Table 1). The equilibrium water contact angle increased to 43◦ and the advancing and receding water contact angles were 54◦ and 22◦ , respectively. This contact angle hysteresis either suggests [21] partial coverage of the glass surface by the brush, or attests to the mobility of the polymer chains in the brush. The advancing angle and the hysteresis are somewhat higher than the values measured by Park et al. [7], who reported advancing and receding angles of 44.6◦ and 30.2◦ on a polyurethene surface covered PEO with a molecular mass of 1000 Da. Harder et al. [22] found advancing water contact angles of 30– 35◦ for oligo(ethylene oxide). Our PEO-coated surfaces are probably more hydrophobic than coatings made of smaller molecules, because small molecules possess a relatively higher fraction of hydroxyl to carbon groups, creating a more hydrophilic coating. Electrophoretic mobilities Figure 4 shows the electrophoretic mobilities of clean and brush-covered glass as a function of the ionic strength. Both surfaces have a finite electrophoretic Table 1. Physico-chemical characteristics, including water contact angles and electrophoretic properties according to a soft-layer model of polymer brushes on glass, together with adhesion of Staphylococcus epidermidis in a parallel plate flow chamber Water contact angles

Glass Brush

Electrophoretic properties

Bacterial adhesion

θw,eq (degrees)

θw,adv (degrees)

θw,rec (degrees)

ρ (106 C m−3 )

λ−1 (nm)

j0 (cm−2 s−1 )

n4h (106 cm−2 )

Wetting 43 ± 3

Wetting 54 ± 4

Wetting 22 ± 4

−1.5 ± 0.3 −0.6 ± 0.03

3.3 ± 0.3 3.6 ± 0.1

1860 ± 120 bd

19.0 ± 3 bd

± indicates the SD over three separately prepared substratum surfaces, while in the case of bacterial adhesion, also three separately grown cultures were employed. bd, below detection.


195

softness, as can be inferred from the fact that both curves tend to a non-zero electrophoretic mobility at elevated ionic strength. The electrophoretic softness of glass was calculated to be 3.3 nm and that of the brush 3.6 nm (Table 1). The results indicate that glass has a water- and ion-penetrable surface, which is in line with earlier measurements reporting thicknesses of ion-penetrable layers on glass of about 0.7 nm [23], 1.9 nm [24] and 4 nm [25]. Surprisingly, the softness of the brush hardly differs from that of the glass surface, which may point to strong immobilization of water in the polymer matrix of the brush. The soft outer region of the brush has a much lower charge density (ρbrush = (−0.6 ± 0.03) × 106 C /m3 ) than in the soft glass layer (ρglass = (−1.5 ± 0.3) × 106 C /m3 ). This result is expected because PEO is essentially uncharged. The thickness L0 of the brush layer can be estimated using the equation [11] 1/3 , (6) L0 ≈ aN σ a 2 where a is the length of a monomer unit and N is the number of monomer units. The applicability of this equation to approximate the thickness of PEO brushes has recently been demonstrated by Efremova et al. [26]. As we used the same grafting technique as and comparable materials to Maas et al. [16], we assume a similar grafting density. Adopting the grafting density that Maas et al. found for Mw = 20 000 (i.e. σ −1 = 0.33 nm−2 ) and inserting N = 250 and a = 0.278 nm [27], L0

Figure 4. Electrophoretic mobilities of glass and brush-coated glass as a function of the ionic strength of a KCl solution. The lines indicate the best fit to equation (4), i.e. the Ohshima soft-layer model.

196

H. J. Kaper et al.

is calculated to be 21.6 nm. Hence, every polymer molecule in the brush occupies a volume of L0 /σ equal to 65 nm3 . The molar volume of an EO segment is equal to 38.9 cm3 [28]. Therefore, the volume taken by the monomer segments of the PEO molecule is 16 nm3 , corresponding to approximately 25% of the total brush volume. This fraction is high enough to ensure that virtually every water molecule in the brush layer is located in the vicinity of a polymer molecule. Since ethylene oxide is known for its strong hydration, it is thus plausible that the water is strongly immobilized in the soft brush layer. Bacterial adhesion In Fig. 5, a representative example of the results of a bacterial adhesion experiment is given. From these plots one can determine the numbers of S. epidermidis adhering after 4 h, as well as the initial deposition rates j0 , which are calculated by linear regression of adhesion numbers during the first 30 min of desposition. Table 1 summarizes the results of the bacterial adhesion experiments quantitatively. The numbers of S. epidermidis adhering after 4 h to pristine glass are several orders of magnitude higher than to the brush-coated glass. In fact, the brush effectively decreases bacterial adhesion to below the detection limit (0.1 × 106 per cm2 ) for direct counting in the parallel plate flow chamber. The results appear to be consistent with other experiments on bacterial adhesion to ‘brush-like’ layers, although the experiments differ in surface modification techniques and/ or experimental methods to assess the efficacy of the surface layer, making a fair comparison difficult. Hendricks et al. [14] tested a plasma-deposited PEO coating

Figure 5. Example of the deposition kinetics of S. epidermidis HBH276 to glass with (circles) and without (triangles) a PEO brush.


197

Figure 6. Micrograph of S. epidermidis HBH276 adhering to a glass slide, partly covered with a PEO brush. The bar marker corresponds to 10 µm (micrograph by courtesy of A. Roosjen).

on polyetherurethane under flowing conditions and reported reductions of over 99% in the initial adhesion of Pseudomonas aeruginosa with respect to a control, as well as over 90% reductions after 18 h of growth. Park et al. [7] also reported reductions of 90% in the adhesion of S. epidermidis on a polyurethane surface with PEG3.4 kDa modification after allowing the bacteria to grow for 24 h. Surprisingly, a PEG-1 kDa-modified substrata performed significantly worse for S. epidermidis. Finally, in order to illustrate the effectiveness of the brush layer in reducing bacterial adhesion, Fig. 6 shows a micrograph of a partly brush-coated glass slide after 4 h of exposure to a bacterial suspension in the parallel plate flow chamber. Clearly, far less bacteria have adhered to the brush-covered surface than to the bare hydrophilic glass.

CONCLUSIONS

Based on this study, the following conclusions can be drawn: (1) polymer brushes applied on a glass surface can be distinguished from untreated glass by an increased hydrophobicity and a decreased fixed charge density, as derived from an electrokinetic soft-layer model;

198

H. J. Kaper et al.

(2) poly(ethylene oxide) brushes on a glass substratum strongly discourage the adhesion of an S. epidermidis strain. Hence, it can be anticipated that polymer brushes will constitute effective non-adhesive coatings for the control of BCIs.

REFERENCES 1. A. G. Gristina, Science 237, 1587 (1987). 2. M. Hermansson, Colloid. Surf. B: Biointerfaces 14, 105 (1999). 3. B. Gottenbos, H. C. van der Mei, H. J. Busscher, D. W. Grijpma and J. Feijen J. Mater. Sci.: Mater. Med. 10, 853 (1999). 4. A. Razatos, Y.-L. Ong, F. Boulay, D. L. Elbert, J. A. Hubbell, M. M. Sharma and G. Giorgiou, Langmuir 16, 9155 (2000). 5. K. Vacheethasanee and R. E. Marchant, J. Biomed. Mater. Res. 50, 302 (2000). 6. D. Leckband, S. Sheth and A. Halperin, J. Biomater. Sci. Polymer Edn 10, 1125 (1999). 7. K. D. Park, Y. S. Kim, D. K. Han, Y. H. Kim, E. H. B. Lee and K. S. Choi, Biomaterials 19, 851 (1998). 8. S. I. Jeon, J. H. Lee, J. D. Andrade and P.-G. de Gennes, J. Colloid Interface Sci. 142, 149 (1991). 9. S. I. Jeon and J. D. Andrade, J. Colloid Interface Sci. 142, 159 (1991). 10. I. Szleifer, Biophys. J. 72, 595 (1997). 11. A. Halperin, Langmuir 15, 2525 (1999). 12. B. Zhao and W. J. Brittain, Prog. Polym. Sci. 25, 677 (2000). 13. O. Prucker and J. Rühe, Macromolecules 31, 592 (1998). 14. S. K. Hendricks, C. Kwok, M. Shen, Th. A. Horbett, B. D. Ratner and J. D. Bryers, J. Biomed. Mater. Res. 50, 160 (2000). 15. X.-Y. Zhu, Y. Jun, D. R. Staarup, R. C. Major, S. Danielson, V. Boiadjiev, W. L. Gladfelter, B. C. Bunker and A. Guo, Langmuir 17, 7798 (2001). 16. J. H. Maas, M. A. Cohen Stuart, A. B. Sieval, H. Zuilhof and E. J. R. Südhölter, Thin Solid Films (in press). 17. P. C. Hiemenz and R. Rajagopalan, in: Principles of Colloid and Surface Chemistry, 3rd edn, Ch. 12. Marcel Dekker, New York (1997). 18. R. J. van Wagenen and J. D. Andrade, J. Colloid Interface Sci. 76, 305 (1980). 19. H. Ohshima, Colloids Surfaces A: Physicochem. Eng. Aspects 103, 249 (1995). 20. J. M. Meinders, J. Noordmans and H. J. Busscher, J. Colloid Interface Sci. 152, 265 (1992). 21. C. W. Extrand, J. Colloid Interface Sci. 207, 11 (1998). 22. P. Harder, M. Grunze, R. Dahint, G. M. Whitesides and P. E. Laibinis, J. Phys. Chem. B 102, 426 (1998). 23. J. M. Kleijn, Colloids Surfaces 51, 371 (1990). 24. A. T. Poortinga, R. Bos and H. J. Busscher, Colloids Surfaces B: Biointerfaces 20, 105 (2001). 25. D. J. Shaw, Colloid & Surface Chemistry, 4th edn. Butterworth-Heinemann, Oxford (1992). 26. N. V. Efremova, S. R. Sheth and D. E. Leckband, Langmuir 17, 7628 (2001). 27. M. Morra, J. Biomater. Sci. Polymer Edn 11, 547 (2000). 28. D. W. van Krevelen, Properties of Polymers, 2nd edn. Elsevier Scientific Publishing, Amsterdam (1976).

Tissue-culture surfaces with mixtures of aminated and fluorinated functional groups. Part 2. Growth and function of transgenic rat insulinoma cells (βG I /17) JAMES R. BAIN 1,2,∗ and ALLAN S. HOFFMAN 3 1 Sarah

W. Stedman Center for Nutritional Studies, Duke University Medical Center, Durham, NC 27710, USA 2 Department of Pharmacology and Cancer Biology, Duke University Medical Center, Mail Stop DUMC 3813, LSRC Building, Room C348, Durham, NC 27710, USA 3 Department of Bioengineering, Box 352255, University of Washington, Seattle, WA 98195, USA Received 10 January 2002; accepted 12 December 2002 Abstract—Interactions of transplantable cells with synthetic polymers can influence the function of biohybrid artificial organs. This study explored growth and secretion of human insulin by βG I/17 cells cultured on surfaces bearing diamine groups (N2), trifluoropropyl groups (F3) and mixtures of the two. Cells cultured on high-F3 and high-N2 surfaces spread well, grew rapidly and produced >1.8 mol lactate per mol glucose consumed, closely resembling cells grown on the permissive control, glass. On one mixed surface, with a molar ratio of 33 N2 groups : 67 F3 groups, cells had a lower lactate/ glucose ratio, adopted a rounded form, grew slowly and were quick to form emergent aggregates, similar to cultures on the inhibitory control, untreated polystyrene. Cultures on surfaces with higher F3 content secreted the most insulin and, in the case of the highest-F3 surface, showed improved responsiveness to secretagogues. Hormone secretion was roughly 50% greater when cells were grown on F3 surfaces conditioned by earlier cultures of βG I/ 17. Incubation of conditioned surfaces with high concentrations of a polyclonal anti-laminin serum prior to re-plating partially abolished this improvement in secretory function. Polymers bearing trifluoropropyl groups appear to be attractive candidates for use in the artificial endocrine pancreas. Surface coatings that include laminin might promote function of transgenic insulinoma cells in vitro and in vivo. Key words: Artificial endocrine pancreas; transgenic insulinoma cells; cell culture; surface modification; insulin; glucose.

∗ To


200

J. R. Bain and A. S. Hoffman

INTRODUCTION

Surgeons might someday be able to restore function of diseased tissues by implanting composites of living cells and artificial materials. Since chemistry and morphology of biomaterials can influence the phenotype of transplantable cells, knowledge of these interactions is fundamental to the creation of successful biohybrid organs, including the artificial endocrine pancreas. Type I or insulin-dependent diabetes mellitus arises from autoimmune destruction of the β cells of the pancreatic islets of Langerhans. Millions of humans suffer from Type-I diabetes and the incidence is increasing in diverse societies worldwide [1– 4]. Despite massive investments in research in recent decades, the causes of this devastating disease remain obscure. Insulin-replacement therapy by injection has been the standard treatment for decades. Though injection therapies improve the quality and length of life, they are not able to provide the close control of blood glucose necessary to prevent diabetic complications in later life, including damage to the kidneys, nerves, blood vessels and eyes [3]. Treatment of human patients by transplantation of islets or whole pancreata from humans (allografts) and pigs (xenografts) has had variable success. Challenges include limited donor supply, immune reactions and the fastidious nature and limited growth potential of human β cells [5– 16]. To overcome these challenges, proliferative cells capable of processing and secreting human insulin are attracting interest as platforms for the creation of transplantable β-cell surrogates by genetic engineering or by manipulation of stem cells. Popular candidates include insulinoma cell lines, which are derived from rare β-cell tumors (Fig. 1), along with other cell types that share the insulinoma’s secretory phenotype [3, 5, 7, 10, 11, 13, 14, 16– 31]. Concomitantly, synthetic materials are being developed to serve as anchorage supports or immune-protection barriers to foster survival and function of β-cell surrogates in the human body. Hybrid artificial organs employing mitotically-expanded islets, insulin-secreting cell lines, or fragments of insulinoma tumors have been grown in vitro, implanted in laboratory animals and used to treat several cases of human diabetes [6, 8, 9, 12, 14, 18, 32– 40], but little is known about the effects of device chemistry and design on the phenotype of transplantable insulinoma cells [29, 38– 43]. In the present study, we examined the growth and phenotype of βG I/ 17, a line of genetically modified rat insulinoma cells (RINs), cultured on borosilicate glass modified with organosilane coupling agents to create surfaces rich in diamine groups, trifluoropropyl groups and mixtures of the two [44]. Portions of this report are reproduced from our preliminary presentation [45]. Note that certain control data on cell metabolism are common to Ref. [43] and the present study, since the two investigations were concomitant. MATERIALS AND METHODS

Culture surfaces As described in the accompanying paper [44], borosilicate glass discs were silanized to create culture surfaces bearing diamine groups (N2), trifluoropropyl groups

Insulinoma culture on aminofluorosiloxanes

201

Figure 1. Presumed interrelationships among rat insulinoma (RIN) cell lines discussed in the text. All derive from a single, radiation-induced tumor in the New England Deaconess Hospital (NEDH) rat. These and related RIN lines have become mainstays of diabetes research, worldwide. RIN line βG I/ 17 was used in the present study.

(F3) and mixtures of the two. Etched and silanized glass discs were mounted for cell culture in 24-well, surfaced-oxidized, tissue-culture plates (Falcon Multiwell 3047, Becton Dickinson Labware, Franklin Lakes, NJ, USA). Discs were held in place by fluoroelastomeric o-rings made of poly(vinylidene fluoride-cohexafluoropropylene), size 014 (Viton A, Kontes Glass Company, Vineland, NJ, USA), as described earlier [43]. Cell line and cell culture The lineage leading to the engineered insulinoma cell line βG I/ 17 began with a rat insulinoma (RIN) that developed in an NEDH rat (New England Deaconess

202


Hospital) after high-dose X-irradiation [46] (Fig. 1). The cell line, RINr, was established from this insulinoma after nine serial tumor transplants in NEDH rats [7, 47]. At low passage, a RINr clone, RINr1046-38, or RIN-38 for short, demonstrates glucose-induced secretion of native rat insulin [48, 49]. Genetic engineers at BetaGene (Dallas, TX, USA), created βG I/ 17 from RIN-38 by transfection with a single human proinsulin cDNA (Fig. 1), as confirmed by Southern-blot analysis [50]. Transgene expression is driven by the cytomegalovirus promoter/ enhancer element. βG I/ 17 contains approx. 410 ng insulin per 6 µg DNA, more than 10 times that of the parental RIN-38 line and is capable of processing human proinsulin to insulin [50, 51]. βG I/ 17 expresses transgenic aminoglycoside phosphotransferase, which confers neomycin resistance [50]. Its transgenic phenotype has been maintained after a year of continuous culture [50]. In the present study, we used βG I/ 17 used at the 53rd passage after it received the human proinsulin gene. We maintained it under selection with the neomycin analog, G418 sulfate (Geneticin, Gibco-BRL, Life Technologies, Grand Island, NY, USA), until the 27th passage. Phenotypic expression was stable at the 53rd passage, as demonstrated by background production of human insulin. Immediately prior to use in this study, βG I/ 17 was shown to be free of contamination with mycoplasmata by the bisbenzimide 33258 fluorochrome (Hoechst, Sigma, St. Louis, MO, USA). For this study, near-confluent cultures of βG I/17 on tissue-culture polystyrene were rinsed, trypsinized, re-suspended, counted and plated at a density of 80 000 cells/ cm2 , approximately one fifth of confluence, as described [43]. Cultures proceeded 7 days in a humidified incubator set to 95% air, 5% CO2 at 37◦ C, with daily changes of media. Conditioned media were frozen for later assays of metabolites. Basal and maximal insulin secretion at seven days After one week, constitutive insulin secretion was evaluated under 100 µM diazoxide (DZ; Sigma), a strong inhibitor of regulated insulin secretion [52]. Cultures were subsequently stimulated for 1 h with a ‘Swiss cocktail’ of secretagogues to assess their maximum potential for insulin secretion. As previously described [43], this cocktail included 3-isobutyl-1-methylxanthine or IBMX, Larginine, D-glucose, L-leucine, L-glutamine and carbamylcholine chloride or carbachol (an acetylcholine analogue). Finally, samples were fixed in 10% neutral, buffered formalin, stained with Harris’s hematoxylin and eosin, inverted and mounted on microscope slides. Assays of glucose, lactate and insulin Media conditioned by the βG I/17 cultures were frozen in polypropylene microcentrifuge tubes, thawed and assayed for glucose consumption and lactate production as described [43]. Insulin levels were assessed in the conditioned media from the seventh day of routine culture, the basal incubation under diazoxide and the Swiss-


203

cocktail stimulation, using the radioimmunoassay (RIA) kit from Incstar (Stillwater, MN, USA) [43]. In subsequent studies of surface conditioning (infra), insulin was assayed with a different RIA kit (Coat-A-Count , Diagnostic Products, Los Angeles, CA, USA). Both RIA kits have calibration histories traceable to the World Health Organization insulin standard 66/304. Where necessary, insulin secretion is expressed in molar units using a molecular mass of 5.808 kDa for fully processed human insulin. Surface conditioning by insulinoma cells After conducting the cell-culture studies described above, we investigated whether conditioning of surfaces by an initial culture of insulinoma cells had any effects on the behavior of subsequent cultures. For this second round of studies, the transgenic insulinoma cell line, βG I/17, was cultured on the pure F3 surface for seven days, as described above and then stripped from the surface without the use of enzymes. This was accomplished by rinsing the cultures twice with Hanks’ balanced-salt solution without divalent cations (HBSS; Gibco-BRL) and then incubating them under a proprietary, enzyme-free, chelating solution (Gibco-BRL, number 13151014). Two sequential chelations were performed, each followed by a firm smack of the plate onto the lab bench. This caused most cells to break free into the overlying liquid. Cells so liberated were removed by pipette and discarded. The very few cells remaining on the surface were then lysed with 20 mM NH4 OH (sterilefiltered) for 5 min at room temperature, a procedure that previous workers developed to remove cells while at the same time preserving extracellular matrix (ECM) molecules in a functional state [53, 54]. Finally, plates were rinsed extensively with HBSS, given fresh, dry lids and stored briefly under HBSS at 37◦ C in preparation for re-plating. Some plates were incubated with antibodies prior to re-plating (infra). Rationale for the selection of anti-laminin reagents to challenge re-plating onto conditioned surfaces To gain insight into the mechanisms of any differences observed between cultures growing on fresh F3 surfaces versus βG I/ 17-conditioned F3 surfaces, we sought to challenge the re-plating exercise with immune reagents raised against ECM proteins that might be involved. Which cell-adhesion ligands are used by native islet β cells and the insulinomas derived from them? In vivo, the connective tissue supporting pancreatic islets contains collagens, laminins and fibronectins. In rats and humans, islets and endocrine tumors of the pancreas express laminin and certain laminin receptors (viz., various combinations of integrin subunits α3, α6, β1 and β4 [22, 55– 60]). Taken together, these published observations, along with our preliminary immunofluorescence results (infra), suggested that laminin might be secreted onto F3 surfaces by βG I/ 17 rat insulinoma cells during the first week of culture. We therefore decided to challenge re-plating onto conditioned surfaces

204


with anti-laminin antibodies, rather than with antibodies to other pancreatic ECM proteins, such as fibronectin or collagen IV. Selection of anti-laminin reagents by indirect immunofluorescence First, we searched for antibodies with reactivity or cross-reactivity to rat laminins. Eight antibody preparations were procured and screened. Monoclonals 2E8, 3.1C12, C4, D5, D7 and D18 were obtained from Dr. David R. Soll, Developmental Studies Hybridoma Bank, National Institute of Child Health and Human Development, University of Iowa (Iowa City, IA, USA). All monoclonals were produced in mouse-mouse hybridomas. Two polyclonal antisera were evaluated: Z0097 (Dako, Glostrup, Denmark) and 5620-3019 (Biogenesis, Poole, UK). Negative controls for monoclonals and polyclonals were the mouse Ig1,KAPPA , MOPC-21 (number 03171D, Becton-Dickinson PharMingen, San Diego, CA, USA) and rabbit serum (number R-9133, Sigma), respectively. Phosphate-buffered saline (PBS, Gibco-BRL number 10010-023) with 1% (w /v) RIA-grade bovine-serum albumin (BSA; Sigma number A-7888) served as a diluent for antibodies. BSA was not heat-treated prior to use. BSA-PBS solution was sterilized by filtration. All monoclonals except D5 were evaluated at dilutions of 1 : 100, 1 : 200 and 1 : 300. D5, which had been received as a raw culture supernatant, was used full strength and at dilutions of 1 : 10 and 1 : 100. Dako polyclonal Z0097 was diluted 1 : 25, 1 : 50 and 1 : 100. Rabbit serum and Biogenesis polyclonal 5620-3019 were diluted 1 : 30, 1 : 100 and 1 : 300. βG I/17 cultures grown for one week on 3/3 F3 were rinsed thrice with 1 ml warm PBS, fixed in situ with ice-cold methanol for ten minutes, again rinsed thrice with 1 ml warm PBS, incubated for 1 h at 37◦ C under the primary antibodies, rinsed twice with PBS, then incubated under the appropriate, fluorescently labeled, secondary antibody, for 1 h at 37◦ C. Secondary antibodies were Texas-Red-X® conjugated goat-anti-mouse IgG for monoclonal preparations and Texas-Red-Xconjugated goat-anti-rabbit IgG for polyclonals (Molecular Probes, Eugene, OR, USA). Cultures were rinsed yet again, then examined with a Diaphot 300 inverted microscope (Nikon, Melville, NY, USA) equipped with an epifluorescent apparatus with a Texas Red filter set (Chiu Technical, Kings Park, NY, USA). Antibody challenge to re-plating Results of the indirect immunofluorescence studies (infra) suggested that Biogenesis polyclonal 5620-3019 might be able to block cellular adhesion to laminin deposited on the conditioned surfaces by the initial cultures of βG I/17. Insulinomaconditioned surfaces were incubated for 90 min under antiserum diluted to 1 : 30 and 1 : 300 with BSA-PBS. Controls included conditioned F3 discs incubated under rabbit serum (diluted 1 : 30) and F3 discs incubated under plain BSA-PBS.


205

Re-plating and secondary culture Cells propagated on tissue-culture polystyrene (TCPS) T-flasks were plated into sample wells and fed daily for 7 days, exactly as described above. Note that these cells were exposed to trypsin immediately prior to the plating experiment. Controls included TCPS (Falcon Multiwell 3047, supra) and fresh, unconditioned F3 wells. TCPS was used as received. Unconditioned F3 controls were incubated under BSAPBS prior to use. At the completion of the week of secondary culture, media conditioned during the final 24 h of cell growth were removed and frozen for later analysis of insulin. Terminal cell counting of re-plated cultures Plates were rinsed in HBSS and then trypsinized. Cells were suspended in their individual culture wells by repeatedly pipetting up and down with a 1 ml disposable pipette tip. Because we knew it would take many hours to count all the samples (allowing for the occasional aperture jams that are perhaps inevitable with such preparations), we lightly fixed each individual culture by adding 0.5 ml of 0.05% (v/ v) glutaraldehyde in Gibco-BRL cell-dissociation buffer (number 13151-014), for a total volume of 1 ml per well. Aliquots of the fixed cell suspensions were diluted in Coulter Isoton® for counting in a Coulter Counter (model Z1, Coulter Electronics, Hialeah, FL, USA). Statistical analysis Pair-wise comparisons of data were evaluated post hoc with the Tukey– Kramer honestly significantly different (HSD) test [61], using version 3.0 of the JMP software (SAS Institute, Cary, NC, USA). Differences were considered significant at P < 0.05.


Surface chemistry affects growth form of βG I/ 17: permissive surfaces βG I/ 17 displays a distinctive morphology when grown on planar substrata in vitro. Cells sprout emergent structures in late log phase when grown on such permissive surfaces as glass, as shown by white arrows in Fig. 2. Cell-material interactions appear to dominate early development of such βG I/ 17 cultures, while cell-cell interactions dominate late events. In this study, growth of βG I/ 17 cells to near confluence, followed by formation of emergent foci, was quite similar on alkalineetched borosilicate glass (Figs 2 and 3), the pure aminosiloxane, N2 (Fig. 3), surfaces with a mole fraction of 1/3 F3 (not shown) and the pure fluorosiloxane, F3 (Fig. 3).

206 J. R. Bain and A. S. Hoffman Figure 2. βG I/17 cells growing on alkaline-etched borosilicate glass, a permissive surface. The same morphologic progression is seen in cultures grown on permissive members of the N2-F3 silane series. Arrows show emergent colonies. Phase-contrast microscopy. Each image represents a sample area measuring 1.36 × 0.9 mm.

Insulinoma culture on aminofluorosiloxanes 207

Figure 3. Appearance of fixed, stained cultures of βG I/17 cells after stimulation with the ‘Swiss cocktail’ of secretagogues. One surface (2/3 F3) is inhibitory. The other three are permissive. Arrows show emergent colonies. Hematoxylin and eosin stain. Bright-field microscopy with Köhler illumination. Each image represent an area of 1.36 × 0.9 mm. This figure is published in colour on http://www.ingenta.com.

208


A generic description of the growth of βG I/17 on these permissive surfaces follows. At one day, βG I/ 17 cells attach and spread into irregular bipolar, threepointed, or stellate shapes, with pronounced cytoplasmic processes (Fig. 2). Cells are dispersed. Doublets, triplets and higher aggregates are rare at 24 h. During days three to five, βG I/ 17 cells spread further and grow rapidly (Fig. 2). The underlying surface is never fully covered. Groups of spread cells have scalloped margins, leaving bare lacunae demarcated by cells. By days five to seven, most cells appear to be joined by cytoplasmic processes, large areas of nearly-confluent cells are seen and it becomes increasingly difficult to distinguish individual cells (Figs 2 and 3). A similar sequence of morphologic events has been observed when disaggregated islet cells from normal rats are grown in plate culture [62]. Emergent, multicellular domes then begin to appear, progressing to oblate, tethered spheroids, apparently by combined processes of mitosis, aggregation and local delamination of cell sheets. Incipient supra-confluent structures are visible at seven days along the margins of a delaminating cell sheet on etched borosilicate glass in Fig. 3. Multicellular character of these emergent structures is evident after Swiss-cocktail stimulation, formalin fixation and staining with hematoxylin and eosin (Fig. 3). Occasionally, these large aggregates become lobulate, grow to macroscopic dimensions and develop arms hundreds of microns in length (e.g. Fig. 4, day 5, taken at low magnification). We have seen this formation of three-dimensional structures and a loss of strong adherence to the substratum in maturing cultures of βG I/ 17 cells and the related 832/13 cell line [19] (Fig. 1) grown on other permissive materials, including tissueculture polystyrene (TCPS, data not shown). Fong et al. [63] grew the closely related rat insulinoma line, RINr (Fig. 1), on TCPS. Morphology of their fiveday cultures was strikingly similar to our own. Clark and Chick [64] also noted delamination of maturing RINr clusters. RINr is ancestral to our βG I/17 [50] (Fig. 1). A similar morphologic progression occurs when diverse lines of insulinsecreting cells from mice are grown on permissive surfaces [65– 67]. These findings of adhesion to surfaces by freshly-plated cells, followed by later aggregation and partial delamination, contrast with most studies of two related insulinoma lines from rats, RINm5F and BRIN-BD11 (Fig. 1), in which cultures grow on TCPS as a monolayer of well-spread, stellate cells (Refs [7, 41, 68], but see Ref. [69]). When interpreting published accounts of the behavior of insulinoma cells in vitro, one must keep in mind that most insulin-secreting cell lines are unstable and change their phenotype with time [5, 7, 10, 16, 32, 70, 71]. Also confounding the interpretation of published studies is the apparent genetic heterogeneity of many insulin-secreting lines: βG I/ 17 is clonal [50], but many other insulinoma lines are not [7, 16, 19]. Cells adopt a distinct growth form on inhibitory culture surfaces To our surprise, some aminofluorosiloxane surfaces of mixed character inhibited growth of βG I/ 17 cells. Depression of cell growth was most pronounced at a mole fraction of 2/3 F3, where cells never approached confluence (Fig. 3). Sparse aggregates of poorly spread cells were visible at 24 h. Compared to cells grown on

Insulinoma culture on aminofluorosiloxanes 209

Figure 4. βG I/ 17 cells growing on untreated polystyrene or UPS, the negative control. Note the slow, incomplete coverage of the surface and the early formation of emergent masses (arrows). Phase-contrast microscopy. The five-day image was taken at low magnification to show formation of a macroscopic, emergent colony and represents an area 3.35 × 2.23 mm. The other images represent areas of 1.36 × 0.9 mm.

210


more permissive surfaces, cells on 2/3 F3 displayed an earlier tendency toward the formation of emergent domes and spheroids, often giving isolated small aggregates a ‘sunny-side-up, fried-egg’ appearance (Fig. 3). In all these characteristics, βG I/ 17 cells grown on 2/3 F3 resembled cultures grown on the negative control, untreated polystyrene (Fig. 4). Thivolet et al. [72] made similar observations when they cultured the rat insulinoma line, RINm5F (Fig. 1), on uncoated ‘plastic’ dishes. Phenotype of cells grown on mixed surfaces with a mole fraction of 5/6 F3 (not shown) was intermediate in appearance between the inhibitory 2/3 F3 and the permissive 100% F3. We have grown other types of mammalian cells on silanized surfaces of the N2F3 series and we have not yet found another case where the culture surface has such striking effects on cellular morphology. When grown under serum-containing medium, the human melanoma cell line, A-375 (American Type Culture Collection, Manassas, VA, USA), for example, spreads and grows equally well on all surfaces of the N2-F3 series (data not shown). The culture surface influences glucose consumption and lactate production Metabolic data from assays of conditioned media correlated well with the qualitative observations of cell growth described above. βG I/ 17 insulinoma cells have a highly active metabolism, as measured by their consumption of glucose and production of lactate (Figs 5– 7). Our plate cultures of βG I/17 cells depleted sugar so rapidly that by the end of the first week, they consumed more than half of the available glucose in 24 h (Fig. 5). Related rat insulinoma (RIN) cell lines exhibit a similar voracious appetite for metabolic fuels, which is much higher than that of the native β cells from which they were derived [7]. A plot of raw data of lactate and glucose concentration in one representative well shows the ‘saw-tooth’ profile (Fig. 5) characteristic of intermittently fed plate cultures. Papas et al. [9] reported a similar stoichiometry in glucose-lactate metabolism in cultures of the mouse insulinoma cell line, βTC3. Roughly exponential plots of glucose consumption (Fig. 6) and lactate production (Fig. 7) during the week of culture correlate with observations of morphology (Figs 2– 4). Culture surfaces of glass, pure N2 (0/3 F3), 1/3 mole fraction F3 and pure F3 are permissive. Untreated polystyrene and 2/3 mole fraction F3 are inhibitory and 5/6 mole fraction F3 is intermediate. The molar ratio of lactate produced to glucose consumed provides a useful measure of cell metabolism in vitro [73]. In the present study, the roughly 2 : 1 stoichiometry of production of the triose, lactate, to consumption of the hexose, glucose, remained relatively stable throughout the week of culture (Figs 5– 10). This suggests that catabolism of glucose to pyruvate to lactate was a significant energetic pathway in these rapidly growing cultures of insulinoma cells. Data on consumption of glucose and production of lactate during the seventh day of culture are plotted as a function of surface composition in Fig. 8. Analysis of lactate and glucose data (Fig. 8) with the Tukey– Kramer HSD statistic separates cultures into


211

Figure 5. Characteristic ‘saw tooth’ profile of raw data on glucose consumption and lactate production by a single culture. With appropriate scaling, the lactate profile would nearly mirror the glucose profile. Given the pronounced daily swings in composition of the aqueous phase, life in such a dish more closely resembles life in a tide pool than life in a stream. This culture support, a mixed aminofluorosiloxane surface with a mole fraction 1/3 silane F3, is permissive, allowing growth as rapid as alkaline-etched borosilicate glass, the positive control.

three exclusive groups, based on identity of their culture surfaces: Group A surfaces are permissive, those in Group B are inhibitory and surfaces with a mole fraction of 5/6 F3 are intermediate (Group C). Group separations are significant at P < 0.05. These correspond to the three classes defined by growth morphology (vide supra). βG I/ 17 cells grown on permissive surfaces converted more glucose to lactate than those grown on inhibitory or intermediate surfaces. Molar ratio of glucose consumed to lactate produced was >1.90 for cultures grown on the permissive aminofluorosiloxane surfaces (Fig. 9), indicative of a highly anaerobic metabolism. Cultures on the other permissive surface, alkaline-etched borosilicate glass, were similar and produced an average of 1.84 mol lactate per mol glucose (Fig. 9). Cultures on the inhibitory and intermediate surfaces averaged less than 1.80 mol lactate per mol glucose. Many paired comparisons were significant (Tukey– Kramer HSD test, P < 0.05, Fig. 9). Thus, in the present study, a picture emerges that rapidly growing cultures tended to be more anaerobic in their metabolism than cultures with slower growth. Confirming this, a cross plot of glucose consumption and the lactate/ glucose ratio (Fig. 10) shows a weak, positive correlation between these two variables (r 2 approx. 0.6) and a separation of data into two groups based on the permissive-

212


Figure 6. Glucose consumption by βG I/17 cells cultured on aminofluorosiloxanes and two reference materials. Solid lines show data for cultures grown on permissive surfaces, while broken lines show data for inhibitory or intermediate surfaces.

Figure 7. Lactate production by βG I/ 17 cells cultured on aminofluorosiloxanes and two reference materials. Solid lines show data for cultures grown on permissive surfaces, while broken lines show data for inhibitory or intermediate surfaces.


213

Figure 8. Glucose consumption and lactate production for the final 24-h culture period, plotted as a function of the mole fraction of silane F3 residues in the culture surface. UPS, untreated polystyrene.

Figure 9. Molar ratio of lactate produced to glucose consumed. Complete anaerobic conversion of glucose to lactate via pyruvate would give a ratio of 2.0. Complete aerobic conversion of glucose to carbon dioxide would give a ratio of zero. UPS, untreated polystyrene. ∗ P < 0.05, Tukey– Kramer HSD test.

214


Figure 10. Cross plot of the lactate/glucose ratio as a function of glucose consumption. Rapidly growing cultures on permissive surfaces (ellipse) are more anaerobic than slower, more rounded colonies growing on inhibitory and intermediate surfaces (circle).

ness of the growth surface. Constantinidis et al. [74] observed a similar relationship between growth morphology and glucose/ lactate metabolism when they compared gel-entrapped and plate cultures of the mouse insulinoma cell line, βTC3. Bulk secretion of human insulin is highest on fluorinated surfaces All cultures in this study secreted 17 ng or more of human insulin per well per h on the seventh and final day of culture (Fig. 11). Moreover, all cultures grown on aminofluorosiloxane surfaces showed a higher mean seventh-day and Swisscocktail-stimulated insulin secretion than cultures grown on the two reference materials, etched glass and untreated polystyrene. Insulin secretion was highest when cells were grown on the three surfaces with the highest fluorine content (Fig. 11 and Table 1). Surface composition can favorably influence sensitivity to secretagogue drugs Cell biologists evaluating responses of secretory cells to stimuli often examine the ratio of stimulated to basal secretion. In the present study, the ‘fold stimulation’ was similar for βG I/ 17 cells grown the two reference materials, etched glass and untreated polystyrene (2.3- and 2.4-fold, respectively; Fig. 12). This surprised us, since the insulinoma cells showed strikingly divergent morphology and glucoselactate metabolism on the two materials. In all cases, the mean responsiveness of βG I/17 cells grown on N2-F3 surfaces was about 3-fold or greater. Moreover,


215

Figure 11. Bullk insulin secretion by βG I/17 cells cultured on aminofluorosiloxanes N2-F3 and two reference materials. UPS, untreated polystyrene. ∗ Significantly different from UPS, P < 0.05; #significantly different from glass, P < 0.05.

Table 1. Summary of biological studies of silanized surfaces and two reference materials

Growth rate of βG I/17 insulinoma Growth form of βG I/ 17 insulinoma Formation of emergent cell aggregates Glucose consumed on day 7 Lactate produced on day 7 Mol lactate/ mol glucose on day 7 Insulin secreted on day 7 Secretagogue sensitivity (Swiss/basal) Insulin secreted per glucose consumed UPS, untreated polystyrene.

Etched glass

0/3 F3

2/3 F3

3/3 F3

UPS

rapid spread late high high 1.85 modest P1-4 > P2-4 or P3-1 > P1-1 > P2-1. This indicates that the interaction between glucose and functional monomers is influenced by the composition of the functional monomers. Figure 3 shows the amount of glucose remaining in MIPs after each extraction. MIPs were synthesized using the same cross-linkers, but in the presence of different amounts of glucose. The amount of glucose used during the imprinting process does not appear to affect the glucose binding to the MIPs. For example, no significant differences were observed between P1-5 and P1-8. The amount of glucose remaining in MIPs was in the order P3-8 > P1-8 > P2-8 or P3-5 > P1-5 > P2-5. These results suggest that monomers in P3 interact with glucose with a higher affinity than those in P1 or P2. The swelling behavior of MIPs was examined by measuring the swelling ratios (S) in PBS. The swelling behavior of MIPs prepared with different cross-linker amounts is shown in Fig. 4. While there were no significant differences in the S values among the highly cross-linked MIPs (P1-4, P2-4, and P3-4) irrespective of their monomer compositions, there was a pronounced difference in the S values among MIPs with

320

H. Seong et al.

Figure 3. Amount of glucose remaining in MIPs after repeated extraction in deionized distilled water at 25 ◦ C. Each extraction lasted for 24 h. MIPs were synthesized in the presence of different amounts of glucose as template. The cross-linking densities of the MIPs were the same (n = 3).

Figure 4. Swelling ratios (S) of MIPs having different cross-linking densities. Swelling was measured at 25 ◦ C in PBS (n = 3).

Glucose binding to molecularly imprinted polymers

321

lower cross-linking densities ((P1-1, P2-1, and P3-1). In addition, the degree of swelling was in the order P2 > P1 > P3, indicating that the physical stability of MIPs in aqueous solution was dependent on the degree of cross-linking [30]. Glucose-binding affinity to MIPs The binding affinities between glucose and MIPs were determined by equilibrium dialysis techniques. Each MIP was dialyzed against glucose and the amount of glucose bound to the MIP was calculated using the predetermined calibration curve of the glucose solution. Information on the equilibrium, glucose + MIP ↔ glucose– MIP, was obtained using the following Scatchard plot, a tool already applied in MIP work [25, 32]: B/[F] = Bmax /KD − B/KD , where B is the amount of glucose bound to the polymer, [F] is the concentration of free glucose (approximated by the analytical concentration of glucose), KD is the dissociation constant of the glucose– MIP complex, and Bmax is the apparent maximum number of binding sites. Figure 5 shows the amount of glucose bound to P1-4 polymer against free glucose (a) and the Scatchard plot of the data (b) after the equilibrium dialysis test. Linear regression of the Scatchard plot (R 2 = 0.997) gave a KD value of 1.94 mM [association constant (KA ) = 1/KD = 5.13 × 102 M−1 ]. The linearity of the Scatchard plot indicates that the binding sites are identical and independent. Con-A was also dialyzed against glucose to determine the KD value, which was then compared with those of the MIPs. The KD value of Con-A was found to be 1.57 mM (KA = 6.35 × 102 M−1 ), which is in agreement with the value reported in the literature [33]. The KD or KA value of P1-4 suggests that MIP has glucose-binding affinity and thus, MIPs having higher glucose-binding affinity than Con-A could be achieved by an optimized composition of monomers and cross-linker. On the other hand, the KD values of glucose-non-imprinted polymers were in the range of 32.6– 49.1 m M (KA = 30.7– 20.4 M−1 ) and no relationship between KD and the polymerization conditions was found. The results of Scatchard analyses for MIPs prepared using different amounts of cross-linker or glucose are summarized in Table 3. With an increase in the cross-linker amount (1– 4 of P1, P2, and P3 polymers), the KD value decreased in all series of polymers, indicating that glucose bound with higher affinity to MIPs with a higher cross-linking density at the fixed ratio of monomers. Once glucose-binding sites are formed in MIPs by non-covalent interactions between glucose and functional monomers, the stability of the binding sites would determine the subsequent binding affinity of MIPs. The stability of a binding site is dependent on the degree of cross-linking of the functional monomers. Thus, the higher glucose-binding affinity of highly crosslinked MIPs may be due to the increased stability of glucose-binding sites in the MIPs. In addition, the KD values of the MIPs decreased in all series of polymers as the amount of glucose for imprinting was increased (5– 8 of P1, P2, and P3

322

H. Seong et al.

Figure 5. Amount of glucose bound to P1-4 at pH 7.4 (a) and the Scatchard plot for the bound glucose (b). B is the amount of glucose bound to polymer and [F] is the concentration of free glucose (n = 3).


323

Table 3. Dissociation constants (KD ) of MIPs having different amounts of cross-linkera or glucose as a template moleculeb Polymer

KD ± SD (mM)

Polymer

KD ± SD (mM)

Polymer

KD ± SD (mM)

P1-1 P1-2 P1-3 P1-4 P1-5 P1-6 P1-7 P1-8

6.48 ± 0.07 4.21 ± 0.09 2.93 ± 0.11 1.94 ± 0.21 11.59 ± 0.01 8.24 ± 0.09 2.49 ± 0.01 1.94 ± 0.21

P2-1 P2-2 P2-3 P2-4 P2-5 P2-6 P2-7 P2-8

8.09 ± 0.38 5.93 ± 0.17 3.37 ± 0.04 2.32 ± 0.29 13.05 ± 0.05 10.37 ± 0.22 5.29 ± 0.06 2.32 ± 0.29

P3-1 P3-2 P3-3 P3-4 P3-5 P3-6 P3-7 P3-8

4.27 ± 1.31 2.84 ± 0.55 2.14 ± 0.19 1.66 ± 0.03 9.31 ± 0.02 6.64 ± 0.06 2.05 ± 0.16 1.66 ± 0.03

a 1– 4 b 5– 8

of P1, P2, and P3. of P1, P2, and P3.

polymers). In these polymers, the same amounts of functional monomers and crosslinker were used. Therefore, the higher glucose-binding affinity by a higher glucose concentration is most likely due to the fact that more functional monomers would participate in the formation of a binding site. When the KD values were compared with respect to the monomer composition, P1 polymers composed of VA, AAm, and AB had lower KD values than those of P2 polymers composed of VA, AAm, and PA. In addition, P3 polymers composed of VA, AAm, PA, and AB had lower KD values than those of P1 polymers. These results support the finding that interactions between glucose and functional monomers can be achieved by hydrogen bonding (P2) and that additional hydrophobic interaction can increase the glucose-binding affinity of MIPs. To compare the contribution of each monomer to glucose binding, MIPs having different monomer ratios (9– 11 of P1 and P2 polymers, or 9– 12 of P3 polymers) were dialyzed against glucose and the data were analyzed by a Scatchard plot. The results are summarized in Table 4. When the data of 9– 11 in P1 polymers were compared, the contribution of functional monomers to the glucose binding affinity was in the order VAA > AAm > AB. This implies that hydrogen bonding of the carboxyl group in VAA is stronger than that of the amine group in AAm and also that the hydrophobic interaction of AB is weaker than the hydrogen bonding of VAA or AAm. The KD values of 9– 11 in P2 polymers were in the range of 2.3– 4.1 mM, lower than those of P1 polymers. In addition, the highest KD value of P2-11 polymer, representing the lowest glucose-binding affinity, indicates that PA has weaker hydrogen bonding than that of VAA or AAm. As can be seen in Table 4, the KD values of 9– 12 of P3 polymers were in the range of 2.0– 2.9 m M. The lowest KD value of P3-9 indicates that the carboxyl group of VAA serves as the main functional group for glucose binding. In addition, the relatively low KD value of P3-12 suggests that hydrophobic interaction of AB with glucose can increase the glucose-binding affinity of MIPs which already have glucose-binding affinity by hydrogen bonding [34].

324

H. Seong et al.

Table 4. Dissociation constants (KD ) of MIPs having different molar ratios of monomers Polymer

KD ± SD (mM)

Polymer

KD ± SD (mM)

Polymer

KD ± SD (mM)

P1-9 P1-10 P1-11 —

1.99 ± 0.03 2.24 ± 0.01 3.58 ± 0.22 —

P2-9 P2-10 P2-11 —

2.26 ± 0.09 2.95 ± 0.02 4.11 ± 0.06 —

P3-9 P3-10 P3-11 P3-12

1.99 ± 0.11 2.23 ± 0.06 2.84 ± 0.01 2.35 ± 0.05

Table 5. Dissociation constants (KD ) of MIPs prepared by UV polymerization and thermal polymerization Polymer

KD ± SD (mM)a

KD ± SD (mM)b

P1-8 P2-9 P3-8

2.49 ± 0.06 3.83 ± 0.09 2.72 ± 0.01

1.94 ± 0.21 2.26 ± 0.09 1.66 ± 0.03

aK bK

D D

of UV-polymerized polymers. of thermally polymerized polymers.

UV polymerization is preferred since it has been demonstrated that polymers made at low temperature exhibit higher recognition abilities [35]. Therefore, the polymers having the highest glucose-binding affinity in each polymer series (P1-8, P2-9, and P3-8) were also prepared by UV polymerization at 0 ◦ C. The KD values of UV-polymerized polymers and those of thermally polymerized polymers are summarized in Table 5. In general, it is believed that weak non-covalent interactions, such as hydrogen bonding, essential for imprint formation and subsequent recognition, are stronger at lower temperatures due to a favorable entropy term [36]. However, the KD values of UV-polymerized polymers were higher than those of thermally polymerized polymers. The lower binding affinity of UV-polymerized MIPs may be explained by the lower degree of cross-linking, because of less initiation of AIBN at low temperature.

CONCLUSIONS

MIPs having glucose-binding affinity can be synthesized by copolymerization of mixtures of amino acid-mimicking functional monomers, excess cross-linker, and glucose as a template. Glucose-binding affinity evaluated by the Scatchard analysis was dependent on the composition of the functional monomers, as well as on the degree of cross-linking. The MIPs composed of vinyl acetic acid (VAA), acryl amide (AAm), 4-pentenoic acid (PA), and allyl benzene (AB) had the lowest KD value of 1.66 mM, which is comparable to the KD value of Con-A (1.84 mM). Furthermore, the results of the glucose-binding affinity of MIPs having different


325

molar ratios of functional monomers demonstrate that the strength of interaction with glucose is in the order VAA > AAm > AB > PA. Acknowledgement This study was supported in part by the National Institute of Health through grant DK54164.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

J. Heller, J. Control. Rel. 8, 111 (1988). K. Ishihara, M. Kobayashi and I. Shinohara, Makromol. Chem., Rapid Commun. 4, 327 (1983). M. Casolaro and R. Barbucci, Int. J. Artif. Organs 14, 732 (1991). F. Fischel-Ghodsian and J. M. Newton, J. Membr. Sci. 48, 33 (1990). S. Kitano, Y. Koyama, K. Kataoka, T. Okano and Y. Sakurai, J. Control. Rel. 19, 162 (1992). D. Shiino, Y. Murata, K. Kataoka, Y. Koyama, M. Yokoyama, T. Okano and Y. Sakurai, Biomaterials 15, 121 (1994). Y. Ito, D. J. Chung and Y. Imanishi, Bioconjugate Chem. 5, 84 (1994). M. Brownlee and A. Cerami, Science 206, 1190 (1979). C. M. Pai, Y. H. Bae, E. J. Mack, D. E. Wilson and S. W. Kim, J. Pharm. Sci. 81, 532 (1992). S. J. Lee and K. Park, J. Mol. Recognition 9, 549 (1996). A. A. Obaidat and K. Park, Pharm. Res. 13, 989 (1996). A. A. Obaidat and K. Park, Biomaterials 18, 801 (1997). W. H. Beckert and M. M. Sharkey, Int. Arch. Allergy Appl. Immunol. 39, 337 (1970). A. E. Powell and M. A. Leon, Exp. Cell Res. 62, 315 (1970). T. Li, H. Lee and K. Park, J. Biomater. Sci. Polymer Edn 9, 327 (1998). D. Kriz, C. B. Kriz, L. I. Anderson and K. Mosbach, Anal. Chem. 66, 2636 (1994). K. Mosbach, Trends Biochem. Sci. 19, 9 (1994). K. J. Shea, D. A. Spivak and B. Sellergren, J. Am. Chem. Soc. 115, 3368 (1993). G. Wulff and H. Haarer, Makromol. Chem. 192, 1329 (1991). G. Wulff and S. Schauhoff, J. Org. Chem. 56, 395 (1991). W. J. Wizeman and P. Kofinas, Biomaterials 22, 1485 (2001). G. Vlatakis, L. I. Anderson, R. Muller and K. Mosbach, Nature 361, 645 (1993). J. M. Lin, K. Uchiyama and T. Hobo, Chromatographia 47, 625 (1998). M. Kempe, Anal. Chem. 68, 1948 (1996). M. T. Muldoon and L. H. Stanker, Anal. Chem. 69, 803 (1997). C. Malitesta, I. Losito and P. G. Zambonin, Anal. Chem. 71, 1366 (1999). D. Kriz, O. Ramström and K. Mosbach, Anal. Chem. 69, 345A (1997). M. Dubois, K. A. Gilles, J. K. Hamilton, P. A. Rebers and F. Smith, Anal. Chem. 28, 350 (1956). B. Sellergren, Makromol. Chem. 190, 2703 (1989). S. A. Piletsky, E. V. Piletskaya, T. L. Pannasyuk, A. V. El’skaya, R. Levi, I. Karubo and G. Wulff, Macromolecules 31, 2137 (1998). D. J. Cram, Angew. Chem. Int. Ed. Engl. 25, 1039 (1986). J. Matsui, Y. Miyoshi, O. Doblhoff-Dier and T. Takeuchi, Anal. Chem. 67, 4404 (1995). L. L. So and I. J. Goldstein, Biochim. Biophys. Acta 165, 398 (1968). R. D. Whitaker, Y. Cho, J. Cha, H. L. Carrell, J. P. Glusker, P. A. Karplus and C. A. Batt, J. Biol. Chem. 270, 22895 (1995). D. J. O’Shannessy, B. Ekberg, L. I. Anderson and K. Mosbach, J. Chromatogr. 470, 391 (1989). P. Turkewitsch, B. Wandelt, G. D. Darling and W. S. Powell, Anal. Chem. 70, 2025 (1998).


The effect of oxidation on the enzyme-catalyzed hydrolytic biodegradation of poly(urethane)s ROSALIND S. LABOW 1,∗ , YIWEN TANG 2 , CHRISTOPHER B. MCCLOSKEY 2 and J. PAUL SANTERRE 2 1 University of Ottawa Heart Institute, 40 Ruskin Street, Ottawa, ON, K1Y 4W7, Canada 2 Faculty of Dentistry, University of Toronto, 124 Edward Street, Toronto, ON, M5G 1G6,

Canada

Received 30 November 2001; accepted 1 April 2002 Abstract—Although the biodegradation of polyurethanes (PU) by oxidative and hydrolytic agents has been studied extensively, few investigations have reported on the combination of their effects. Since neutrophils (PMN) arrive at an implanted device first and release HOCl, followed by monocytederived macrophages (MDM) which have potent esterase activities and oxidants of their own, the combined effect of oxidative and hydrolytic degradation on radiolabeled polycarbonate-polyurethanes (PCNU)s was investigated and compared to that of a polyester-PU (PESU) and a polyether-PU (PEU). The PCNUs were synthesized with PCN (MW = 1000), and butanediol (14 C-BD) and one of two diisocyanates, hexane-1,6-diisocyanate (14 C-HDI) or methylene bis-p-phenyl diisocyanate (MDI). The PESU and PEU were synthesized using toluene-diisocyanate (14 C-TDI), with polycaprolactone and polytetramethylene oxide as soft segments respectively, and ethylene diamine as the chain extender. The effect of pre-treatment with 0.1 mM HOCl for 1 week on the HDI-based PCNUs and both TDI-based PUs resulted in a significant inhibition of radiolabel release (RR) elicited by cholesterol esterase (CE), when compared to buffer alone, whereas the MDI-based PCNU showed a small but significant increase. When PMN were activated on the HDI-based PCNU surface with phorbol myristate acetate (PMA), HOCl was released for 3 h, and was almost completely abolished by sodium azide (AZ). Simultaneously, the PMN-elicited RR, shown previously to be due to the esterolytic cleavage by serine proteases, was inhibited ∼75% by PMA-activation of the cells, but significantly increased relative to the latter when AZ was added. Both in vitro oxidation by HOCl and the release of HOCl by PMN were associated with the inhibition of RR and suggest perturbations between oxidative and hydrolytic mechanisms of biodegradation. Key words: Polyurethanes; biodegradation; oxidation; hydrolysis; neutrophils.

∗ To

whom correspondence should be addressed. Phone: (613) 761-4010. Fax: (613) 761-5035. E-mail: [email protected]

328

R. S. Labow et al.

INTRODUCTION

Polycarbonate-based polyurethanes (PCNUs) are materials that are being promoted as having a lower susceptibility to hydrolysis relative to polyester-urethanes (PESU) and lower oxidation relative to polyether-urethanes (PEU) [1]. Biodegradation, which includes environmental stress cracking (ESC), is still observed clinically on long-term implanted polyurethane (PU) devices. The process is postulated to involve multiple components, of which the chemical interactions between the biomaterial and biological components are not clearly understood. Oxidation is hypothesized to be an important parameter in the process of ESC in vivo, since the end result of the latter process can be simulated in vitro on the surface of materials using oxidizing reagents such as hydrogen peroxide and hypochlorous acid (HOCl) [1]. Although monocyte-derived macrophages (MDM) eventually become the most abundant cell type at the material interface of implanted devices and release reactive oxygen species when activated during the inflammatory response, neutrophils (PMN) are the first to arrive during the early hours following implantation and may be recruited for the first few days following surgery [2]. PMN release one of the strongest oxidative biological compounds yet identified in these cells, HOCl [3]. Although the steroid dexamethasone, which inhibits the respiratory burst, was able to reduce the effect of MDMs on a polyetherurethane [4] and thus imply a role for oxidants in ESC, this agent will also shut down the remaining events in the inflammatory response, i.e. the release of lysosomal and other intracellular enzymes [5] which may be responsible for hydrolytic degradation. Hence, the MDM agents that are effectively responsible for the chemical breakdown of the PUs are still being elucidated. Previous studies have suggested that MDM are capable of promoting hydrolytic degradation of PU in addition to the potential degradation due to the release of reactive oxygen species [1, 6, 7]. Using a trypsinized activated mature MDM cell system, degradation of a 14 C-labeled-PESU [8] as well as a 14 C-labeled PCNU [9], has been demonstrated by measuring radiolabel release into the cell supernatant. In parallel to these findings, it was found that a significant increase in esterase activity was measured not only in the MDM after differentiation [10], but continued to increase while the activated MDM remained in culture on the PESU [9] and PCNU [10] surfaces. In several enzymatic and cell studies involving the hydrolytic degradation of PUs, esterolytic activities have been shown to be involved. Thus far, when commercial cholesterol esterase (CE) was incubated with several polyurethanes it has been shown to be the most destructive [11]. These include polyester- [12– 14] and polyether-PUs [15], and dental restorative composite materials [16]. The polyester PUs were primarily cleaved at the ester bonds in the soft segment and the products were identified as oligomers of TDI [17]. CE also cleaved polyether PUs at the most probable bond susceptible to hydrolytic cleavage, which is the urethane bond, resulting in the release of the free amine (toluene diamine) [18, 19]. Recently, PCNUs have been shown to be susceptible to cleavage by CE as well [20, 21]. Both

The effect of oxidation on hydrolysis of polyurethanes

329

the carbonate and urethane bonds were cleaved resulting in many products ranging in molecular weights from ∼150– 850 which were identified by GC-MS [22]. The degree of susceptibility varied considerably with the ratio of hard to soft segment as well as with different diisocyanates [20, 21]. It was possible to determine the susceptibility of each of the PCNUs to hydrolytic cleavage by CE and relate the extent of degradation to the amount of hard segment interaction within the polymer and at the surface. More specifically, the degree of phase separation and soft segment crystallinity were found to be less important in comparison to the hydrogen bonding among the carbonate and urethane linkages [21]. The rank of the different chemical groups’ susceptibility to hydrolysis was as follows: nonhydrogen bonded carbonate > non-hydrogen bonded urethane > hydrogen bonded carbonate > hydrogen bonded urethane. There is a strong belief that it is a synergistic effect between hydrolytic and oxidative activation pathways which are involved in the biodegradation processes leading to the clinically observed phenomenon of ESC [23]. Since PCNUs have been reported to be more stable to oxidation than PEU and more stable to hydrolysis than PESU, the combined effect of oxidation on the subsequent CE-mediated hydrolysis was of interest with respect to these PUs. This study attempted to partly simulate the biochemical effects of the in vivo condition, by pre-treating PUs with HOCl, the oxidative compound released from PMN, followed by incubation with CE which is secreted by MDM [24]. Although PMN have been shown previously to release serine proteases which are capable of hydrolytic cleavage of PUs, their degradative potential is much less than that of MDM [8]. In this study, an in vitro PMN cell system was used to measure HOCl and radiolabel release simultaneously, in order to determine if the oxidative effect of PMN had an impact on the subsequent hydrolytic biodegradation.


Materials Unless otherwise specified all reagents were purchased from the Sigma Chemical Company, St. Louis, MO. Polymer synthesis The diisocyanates used for the synthesis of the PUs were toluene diisocyanate (TDI), 1,6-hexane diisocyanate (HDI, Aldrich, Milwaukee, WI, USA) and 4,4 methylene bis-phenyl diisocyanate (MDI, Aldrich, Milwaukee, WI, USA). The same chain extender and soft segment were used for all the PCNUs and were 1,4-butanediol (BD, Aldrich, Milwaukee, WI, USA) and poly(1,6-hexyl 1,2-ethyl carbonate) diol (PCN, MW = 1000, received in kind from Corvita Corporation, Miami, FL, USA), respectively. The chain extender for the polycaprolactone

330

R. S. Labow et al.

Table 1. Polyurethanes synthesized Reagent

Stoichiometry

Acronym

MW

CPM/ 100 mg

MDI/ PCN/BD HDI/ PCN/BD HDI/ PCN/BD HDI/ PCN/BD HDI/ PCN/BD HDI/ PCN/BD TDI/PCL/ED TDI/PTMO/ ED

3:2:1 3:2:1 3:2:1 2:1:1 3:1:2 4:3:1 2 : 1 : 2.2 2:1:2

a MDI-321B

10 × 104 10 × 104 10 × 104 3.9 × 103 3.8 × 103 11 × 104 11 × 104 18 × 104

2.8 × 105 4.7 × 105 11 × 105 10 × 105 9.4 × 105 12 × 105 7.0 × 105 9.1 × 105

a HDI-321B b HDI-321 b HDI-211 b HDI-312 b HDI-431 c TDI/PCL/ED c TDI/PTMO/ ED

Abbreviations: Diisocyanate: Methylene bis-p-phenyl diisocyanate (MDI), 1,6-Hexane diisocyanate (HDI), toluene diisocyanate (TDI), Macroglycoldiol soft segment: Poly(1,6-hexyl 1,2-ethyl carbonate) (PCN) (MW 1000); polyether diol (PTMO) (MW 1000), polycaprolactone diol (PCL) (MW 1250) Diol chain extender: 1,4-butanediol (BD) (MW 90). a 14 C-BD used in the synthesis. b 14 C-HDI used in the synthesis. c 14 C-TDI used in the synthesis.

(PCL) (polyester) and polytetramethylene oxide (PTMO) (polyether) based PUs was ethylene diamine (ED). Table 1 lists the stoichiometry of the reagents used in the PU synthesis, the specific radioactivities and the location of the radiolabel, as well as the acronyms to which the polymers are referred. The 14 C-labeled PUs used in the biodegradation experiments were synthesized with either 14 C-BD, 14 C-HDI, 14 C-TDI (custom synthesis from NEN DuPont, Mississauga, Ontario). The details of the synthesis and characterization of the polymers were previously described [13, 14, 20, 21]. The specific radioactivity for each polymer was calculated by dissolving 1.0 mg of each polymer in dimethylacetamide (DMAC) (BDH Chemical, Mississauga, ON) (1.0 ml) and counting the solution in 10 ml of Formula 989 liquid scintillation cocktail (Packard Instruments, Inc., Meriden, CT) in a liquid scintillation conter (LKB Rackbeta, Gaithersburg, PA). The release of radioactivity (CPM) into incubation solutions from the PU substrate has proven to be a reliable and sensitive method for measuring the degree of material degradation [9, 13– 15, 18– 20, 25]. The radioactive products have been isolated and identified and related to the surface damage due to chemical bond cleavage [17]. The preparation of polymer samples was carried out using solutions of 10% (wt/vol) PU in DMAC. Any non-soluble material in the solution was removed from the polymer solutions by passing them through a 0.45 µm Teflon® filter (Chromatographic Specialties, Toronto, ON). Samples incubated in cell culture plates consisted of round glass coverslips (15 mm diameter) (Fisher Scientific, Ottawa, ON) coated with the PCNU solutions (100 µl) under sterile conditions in a laminar flow hood and dried overnight at 50 ◦ C, followed by purging at 50 ◦ C without vacuum for 24 h and then under vacuum for 72 h as described in detail previously [13]. The experiments with the polyester and polyether PUs were


331

carried out using hollow glass tubes coated with polymer and incubated in sealed vacutainers using an established method previously reported on [13– 15, 20, 21]. Assay of HOCl Hypochlorous acid (HOCl) (0.5 mM) solutions were prepared by adding 168 µl of sodium hypochorite (NaOCl) (6%) (VWR Scientific, VW 3248-1) to 25 ml of 0.05 M sodium phosphate buffer, pH 7.0 and diluting 1 : 10, giving a concentration of 0.5 mM. In order to determine the amount of HOCl consumed by the PUs or that was produced by the PMN, a standard curve was prepared by adding 0.01– 0.5 ml of the 0.5 mM solution of HOCl, with a final volume of 1.0 ml in water. The colorimetric assay was a modification of the procedure of Weiss et al. [26]. The following reagents were added to these different concentrations of HOCl: 0.5 ml of 0.05 M sodium phosphate buffer, pH 7.4, 0.15 ml of a taurine solution (0.2 g of taurine (f.w. 125.1) (Sigma T0625) into 25 ml of distilled deionized water) and 0.25 ml of a potassium iodide solution (1.99 g of potassium iodide (f.w. 166) (Sigma P8256) in 100 ml of distilled deionized water). Each sample was read at 350 nm in a spectrophotometer (Beckman Instruments). Assay of cholesterol esterase activity CE (Genzyme Diagnostics, Cambridge, MA) (0.003 g) was dissolved in 10.0 ml of 0.05 M sodium phosphate buffer and diluted before assaying according to the procedure of Moore et al. [27]. A 0.05 M sodium phosphate buffer, pH 7.0 (1.4 ml) and p-nitrophenylbutyrate (PNB) solution (4 mM in acetonitrile, 50 µl) were mixed with 25 µl of CE solution diluted 1 : 100. This solution was incubated at 37 ◦ C for 30 min and read at 400 nm. One unit of activity was defined as the release of 1 nmol of p-nitrophenol (ε = 16 300 cm−1 M−1 ) per minute at 37 ◦ C. Polyurethane degradation experiments All biodegradation experiments using a pre-incubation of HOCl utilized a 0.1 mM solution of the oxidative agent. The PU samples were pre-treated with either 0.1 mM HOCl solution or 0.05 M sodium phosphate buffer, pH 7.0 and incubated at 37 ◦ C. The HOCl solution (prepared fresh daily) was replaced every day for 5 days (a total of 5.0 ml). After 1 week, the pre-incubation solution was removed and replaced with either CE solution (100 units/ ml in sodium phosphate buffer, pH 7.0) or buffer and incubated for either 24 h or extended periods up to 18 weeks as described previously [12, 13]. The radiolabel release (0.8 ml of solution) was counted in 10 ml of Formula 989 liquid scintillation cocktail and counted in a liquid scintillation counter. The effect of HOCl or buffer pre-treatment was assessed by comparing data to PU-coated samples which had not been pre-treated but were incubated with buffer and CE solutions for the 24 h period. For comparative purposes the radiolabel release data among the PCNUs were normalized to HDI431 based on the

332

R. S. Labow et al.

specific radioactivity values for each polymer in Table 1 (e.g. HDI431 had a specific radioactivity of 12 × 105 CPM /100 mg and MDI 321’s was 2.8 × 105 CPM /100 mg. The value for the CPM for the latter polymer was multiplied by 4.8). Isolation of human neutrophils Human PMNs were isolated from whole blood using a density gradient centrifugation procedure as described in detail previously [25]. The freshly isolated PMNs were resuspended in DPBS (Dulbecco’s phosphate buffered saline) and seeded onto 14 C-HDI-431-coated glass slips (4 × 106 cells/ ml) or HDI-431 without a radiolabel for measurement of HOCl (2 × 106 cells/ ml) and allowed to adhere for 1 h. Following 1 h the DPBS was removed and the following reagents were dissolved in DPBS and added (phorbol myristate acetate (PMA), 10−7 M or sodium azide (AZ) 5×10−3 M or both) to the adherent cells which were then incubated at 37 ◦ C for 24 h. The use of PMA and AZ at these defined concentrations was established based on a previous report [28]. The influence of the cells on radiolabel release, i.e. material breakdown and release, was determined by counting the cell supernatant in a liquid scintillation counter. The HOCl released into the cell supernatant was determined using the assay described above with the following change, 0.18 g/ 100 ml of glucose was included with the DPBS according to the procedure of Schacter et al. [3]. Both the radiolabel release data and the HOCl production were related to the number of live adherent cells, determined by lysing the cells and assaying the lysate for lactic acid dehydrogenase (LDH) activity as described in detail previously [29]. Briefly, the lysate was added to pyruvate and the remaining pyruvate determined colorimetrically by the absorbance of the 2,4-dinitrophenylhydrazone derivative using a kit (Sigma Chemical Co., St. Louis, MO). A standard curve was generated by plotting the number of PMN seeded (103 – 106 cells, determined by counting the cells seeded in each well using a Baker cell counter (Serono Instruments)) against the LDH activity in cells following lysis with 0.05% Triton X-100. The correlation coefficient for the curve was r 2 = 0.993. Statistical analysis The data were analyzed by a 1 way ANOVA using a program written in Lotus 1-2-3. A significant difference was defined at p < 0.05.

RESULTS

Following contact with the PU samples, the disappearance of HOCl was monitored spectrophotometrically as described above and compared to samples not coated with PUs. Approximately 10% of the HOCl remained in the solutions in contact with glass slip controls after 24 h, however, no HOCl was detected in the PU-containing incubation solutions for the same time frame (data not shown). Based on these


333

Figure 1. Effect of HOCl on the cholesterol esterase (CE) hydrolysis of polycarbonate-urethanes (PCNUs): PCNUs-coated slips (described in Table 1) were treated with either 0.1 mM HOCl (closed bars) or 0.05 M sodium phosphate buffer, pH 7.0 (open bars) for 1 week, followed by treatment with CE for 24 h according to the procedure described in MATERIALS and METHODS. The radiolabel release in CPM/ml was plotted for each polymer and each condition.

findings, fresh HOCl was prepared daily and added to the PU samples in order to maintain oxidant activity levels. At the end of the one week incubation period, no HOCl could be measured prior to the addition of CE. From these observations it was concluded that the polymer was continuously consuming the HOCl via oxidation reactions. In this study a relatively simplistic in vitro model for the initial oxidative burst and the subsequent synthesis of esterase activity from the MDM system was carried out by first treating the PCNU surface with 0.1 mM HOCl for a period of 1 week and then with CE (100 units) for 24 h. No difference in radiolabel release after the treatment with either HOCl or buffer (341 ± 16 CPM) was detected prior to the addition of CE. The latter radiolabel count was subtracted from the data presented in Fig. 1 in order to facilitate quantifying the specific contribution to biodegradation from the subsequent CE treatment. The radiolabel release from the incubation of all the PCNU samples for 24 h in CE solution without any pre-treatment was not significantly different from the buffer pre-treated/ CE samples that were recently reported for HDI-431 [9]. When the PCNU-coated slips were treated with HOCl prior to incubation with CE, there was a significant reduction in radiolabel release for all the HDI-based polymers in comparison to control samples exposed to CE after only pre-treatment with buffer (Fig. 1). The CPM released were normalized based on the specific radioactivity values (Table 1) in order to compare the effect of HOCl and CE on each PCNU. There was no significant difference in the radiolabel release or the effect of HOCl when the position of the radiolabel in HDI-321 was changed between the HDI

334

R. S. Labow et al.

or the BD moiety (Fig. 1). The effect of diisocyanate chemistry (i.e. HDI vs MDI) had an important influence on the effect of HOCl treatment and the subsequent radiolabel release generated by CE. The MDI-321B was degraded to a lesser extent than HDI-321B which had a similar stoichiometry (Table 1) (Fig. 1). Moreover, the effect of the pre-treatment by HOCl enhanced the hydrolysis by CE rather than decreasing it as was observed for the HDI polymers. This was the only polymer to show this small but significant increase in degradation following oxidation (Fig. 1) (p < 0.05). The inhibitory effect of HOCl on PU degradation (Fig. 1) was not only observed on the PCNU materials. A biodegradation study using a similar concept to that used for acquiring data in Fig. 1, examined the effect of HOCl on the CE catalyzed degradation for TDI/PTMO /ED and TDI/PCL /ED, polyether and polyester-PUs respectively (Fig. 2). Despite the fact that the degradation study was extended to 18 weeks and that the samples had completely different chemistries than that of the PCNU based materials, it was observed that the HOCl had a significant influence on the polyester based PU (Fig. 2A) (p < 0.05). The effect was less pronounced for the polyether based PU (Fig. 2B) (p < 0.05). Again, there was no significant difference in radiolabel release during the first week of incubation when these PUs were treated with only HOCl or buffer (Fig. 2A). Similar results were obtained with the polyester-PU (Data not shown) (Fig. 2B). In a separate series of experiments, to assess if HOCl generation from PMN influenced the hydrolytic potential of these cells to release radiolabel from the polymers, PMN were seeded on HDI-431-coated glass slips in 4 groups: DPBS (control), PMA (protein kinase C activator), AZ (myeloperoxidase inhibitor) or PMA with AZ. HOCl production (Fig. 3) by the PMN was measured simultaneously with radiolabel release (Fig. 4) as well as cell viability by LDH activity (Fig. 5). Only a very small amount of HOCl could be measured if PMA was not included in the DPBS. At 1 h there was no difference in HOCl production between PMA and PMA with AZ. The rate of HOCl release increased with PMA up until 2 h (∼18 nmol/ h), but the generation of HOCl from the PMN with AZ present along with PMA decreased significantly by that time (∼2 nmol/ h). Between 3 and 24 h the rate of HOCl production from the PMA group had decreased to 0.5 nmol/ h (Fig. 3). In the absence of cells, the reaction of DPBS with HDI-431 gave a reading of zero in the assay. Radiolabel release data (Fig. 4) showed that although PMA elicited the greatest release of HOCl compared to the other 3 groups of PMN, at 24 h there was ∼75% inhibition of radiolabel release observed when compared to the PMN in DPBS. Moreover, at the same time point there was a significant increase of radiolabel release for the AZ treated group when compared to the DPBS control (Fig. 4). As was observed in Fig. 3, the effect of PMA over AZ, when combined together with the PMN, was dominant and produced an overall reduction in radiolabel release when compared with the DPBS control. At 2 h, where HOCl release was peaking


335

Figure 2. Effect of HOCl on the cholesterol esterase (CE) hydrolysis of polyether- and polyester urea urethanes. The polyester-urethane (PU) (Fig. 2A) and polyether-PU (Fig. 2B) were treated with either 0.1 mM HOCl (closed circles) or 0.05 M sodium phosphate buffer, pH 7.0 (open circles) for 7 days, followed by treatment with CE for a further 126 days (18 weeks) according to the procedure described in MATERIALS and METHODS. Control samples were treated with either 0.1 m M HOCl (closed triangles) or 0.05 m sodium phosphate buffer, pH 7.0 (open triangles) only. The cumulative radiolabel release in CPM/ml was plotted for each polymer and each condition.

336

R. S. Labow et al.

Figure 3. Effect of phorbol myristate acetate (PMA) and sodium azide (AZ) on neutrophil (PMN)mediated HOCl release: PMNs (2 × 106 ) were seeded onto HDI-431 coated slips in DPBS (closed circles), PMA (open circles), AZ (closed triangles), or PMA plus AZ (open triangles). HOCl production was followed from 1 h to 24 h as described in MATERIALS and METHODS.

(Fig. 3), there were no significant differences in radiolabel release (Fig. 4) between any of the groups. Both radiolabel release and HOCl production were normalized to the number of adherent PMN as determined by LDH activity. The freshly isolated PMN, at the time of seeding onto the HDI-coated glass slips were >95% viable. However, the cells in the 4 groups died at different rates over the course of the 24 h incubation. At 2 h, when HOCl production was at its maximum (see Fig. 3), there was no difference in the viability in the 4 groups as was previously reported [9]. At 24 h, the cells with PMA were almost completely dead whereas the cells with AZ and PMA were >20% viable. The cells in DPBS and AZ were not significantly different at 24 h than from time = 0 (Fig. 5).

DISCUSSION

Since previous studies have implied that esterase activities are most likely involved in the degradation of PU by MDM [9, 10], it was of interest to assess the combined contribution of both hydrolytic and oxidative processes in this degradation. In earlier work with the PCNU polymers, significant differences in their susceptibility to CE hydrolysis was found [20, 21]. These changes were quite pronounced after a


337

Figure 4. Effect of phorbol myristate acetate (PMA) and sodium azide (AZ) on neutrophil-mediated radiolabel release: Neutrophils (4×106 ) were seeded on to HDI-431 coated slips and radiolabel release was measured as described in MATERIALS and METHODS at 2 h and 24 h in DPBS (closed bars), PMA (hatched lines right), AZ (hatched lines left), PMA with AZ (crossed lines).

few weeks of incubation with HDI-312 showing high stability early on. Here again, HDI-312 established itself as the most stable of the HDI-based PCNUs. As well, it shows the least amount of change following oxidation (Fig. 1). The availability of CE sensitive cleavage groups in the PCNU polymers was previously shown to be dependent on the chemical nature of the diisocyanate contained in the PCNU [20]. Although phase separation of the polymers may play a role in the biostability, it is the chemical nature of cleavage sites and their immediate structural environment which was reported to allow the diisocyanate components to shield the polycarbonate soft segment from hydrolytic degradation. Among PCNUs with similar reagent stoichiometry, but different diisocyanates, MDI-321 was the least susceptible to hydrolytic cleavage. The most influential factor on the biodegradation of the PCNUs, catalyzed by CE, was demonstrated to be the degree of hydrogen bonding among the carbonate and urethane groups and inter-urethane groups [20, 21]. The enhanced stability of MDI-321 was again demonstrated in this study (Fig. 1). There are two oxidation mechanisms that are suspected of being involved in the degradation process of the PUs by HOCl. These include chlorination of the nitrogen in the urethane bond which would be similar to the chlorination of taurine, a reaction which has been well characterized [26]. The other reaction would be related to the oxidation of the soft segment hydrocarbon chain which may contribute to the

338

R. S. Labow et al.

Figure 5. Effect of phorbol myristate acetate (PMA) and sodium azide (AZ) on neutrophil viability. Neutrophils (2×106 ) were seeded onto HDI-431 coated slips and viability was measured by lactic acid dehydrogenase activity in the Triton X-100 lysate of the adherent cells in DPBS at 24 h as described in MATERIALS and METHODS.

cross-linking of the polymer chains [23]. There has been some documentation in the literature which specifically reports on the susceptibility of diisocyanates and their urethane linkages. Particularly, it is noted that aromatic systems can readily undergo oxidative degradation to yield quinone-imide formation and this results in the loss of the urethane hydrogen on the nitrogen [30]. The latter product is in part responsible for color changes in these PUs and can ultimately lead to chain scission of the urethane bond. Loss of the urethane bonds has been observed to be more prominent in phase mixed polymer domains than in segregated hard segment domains [31]. Furthermore, this breakdown can result in the loss of H bonding in these systems [32]. The above considerations may provide some very important insight towards explaining the dissimilar behaviour among the PCNU materials in the current study (Fig. 1). Very recent information [20, 21] has indicated that HDI-based PCNUs are much more phase separated than is the MDI-321 polymer. Furthermore, the surface of the MDI-321 material is made up primarily of the H-bonded urethane/ polycarbonate moities as opposed to segregated hard segment domains [20, 33]. These data were corroborated by ATR-FTIR and FTIR reported recently [20, 21]. Meijs et al. [34] have shown that oxidation is particularly prominant at hard/ soft segment interphases. If H-bonding can be lost through oxidation processes, as discussed above, then the protective shield of the carbonate groups against hydrolysis would be effectively compromised at the soft/ hard segment


339

interphase, and hydrolytic degradation of carbonate groups would be expected to increase. This would have an end result of more labeled products being released. On the other hand, the structural character of the HDI-based PCNUs is such that the surfaces are highly phase separated and the polycarbonate is reported to be in a crystalline state [21, 33]. Therefore, oxidation would be expected to be low in the phase separated hard segment domains. Oxidation of the PCNUs may possibly lead to crosslinking of the oligomeric soft segment, as well as chain scission. If the former is occurring, then there could be a reduction in the amount of released product from the surface. The alternative explanation is that the change in carbonate chemistry may have rendered the crystalline soft segment less vulnerable to the particular enzyme studied in this paper. However, these considerations remain to be further elucidated. The variation in hydrolytic behaviour of CE with respect to oxidation pretreatment is not restricted to the PCNU materials, because as seen in Fig. 2, both the polyether- and polyester-PUs have unique responses. Hence, it is concluded that if synergistic mechanisms are at play between hydrolytic and oxidative pathways of the biodegradation process then consideration must be given to both polymer chemistry and structure in order to access the potential direction that such mechanisms will take for a given implant material. The whole question of whether a synergistic mechanism does exist in vivo is further elaborated on in the PMN experiment. PMA stimulates the respiratory burst and should enhance the oxidative effect of PMN on the polymer, whereas AZ inhibits myeloperoxidase which catalyzes the formation of HOCl and therefore should minimize the oxidative effect of PMN on the polymer degradative process [26]. In the current study, the PMN cell system (Fig. 3) yielded a similar conclusion to that observed for the in vitro biodegradation of HDI-431 (Fig. 1). The PMA which caused the release of measurable amounts of HOCl (Fig. 3), also inhibited radiolabel release (Fig. 4). Although a similar mechanism may be operating in vivo, these findings should be taken as circumstantial evidence and not proof that such synergistic events are at play. In fact, it was previously found that although PMA significantly inhibited radiolabel release elicited by PMN, it was possible that this was due to the effect of PMA on cell survival and its ability to synthesize and/ or secrete the hydrolytic agents acting on the polymer [8]. In the latter study by Labow et al., the rate of radiolabel release was found to be directly proportional to cell viability and that as the PMN died, radiolabel release decreased [8]. Similar results were found in the current study. PMA greatly accelerated the rate of cell death (Fig. 5). The additional finding in this study was that AZ, which decreased the rate of cell death due to PMA, significantly increased the amount of radiolabel release, although there was no significant difference in the LDH activity or HOCl production by PMN treated with AZ and no PMA when compared to DPBS controls. The above results are interesting because they suggest that by inhibiting specific cellular activities, the cells’ hydrolytic potential may be increased. PMA stimulates the respiratory burst by activating protein kinase C [35]. This increases the

340

R. S. Labow et al.

production of hydrogen peroxide which is cytotoxic itself, in addition to being a substrate for myeloperoxidase which catalyzes the production of HOCl, another very cytotoxic agent, from chloride and hydrogen peroxide. AZ inhibits the production of hydrogen peroxide, thereby inhibiting HOCl production, and so decreases the rate of cell death [35], although not significantly different from that of the DPBS control in the current study. Future studies will need to probe the level of hydrolytic enzyme activity between these different conditions in order to gain a deeper understanding of the mechanisms at play. In summary, this study showed that rather than accelerating biodegradation, oxidation significantly decreased the release of hydrolysis degradation products from several PUs. The effect of surface and bulk chemistries on the formation of stabilizing forces such as inter- and intra-molecular hydrogen bonding, is believed to be important in terms of defining the effect of the oxidation on the PUs susceptibility to subsequent hydrolysis. The DSC and SAXS data along with the ATR-FTIR reported for the materials in other recent publications [20, 21, 33] support the hypothesis that hydrogen bonding is more important and possibly more dominant than crystallinity in stabilizing the PCNUs. The effect of oxidation by the release of HOCl from PMN may be a perturbation factor on hydrolytic cleavage, but cell survival may also be a factor. Therefore, further work remains to be done with regard to elucidating the biochemical reactions involved in the process of in vivo ESC. Acknowledgements The assistance of Janet Malowany, Samir Hazra, Sanjay Jacob, Erin Meek and Girija Waghray is gratefully acknowledged. This study was funded by grants from the Canadian Institutes of Health Research (CIHR) and Materials Manufacturing Ontario (MMO).

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

L. Pinchuk, J. Biomater. Sci. Polymer Edn 6 (3), 225 (1994). J. M. Anderson, Cardiovasc. Pathol. 4 (3), 33S (1993). E. Schacter, R. L. Lopez and S. Pati, Biochem. Pharmacol. 41, 975 (1991). J. Casas, Q. Zhao, M. Donovan, P. Schroeder, K. Stokes and K. Untereker, J. Biomed. Mater. Res. 46, 475 (1999). N. R. Ackerman and J. R. Beebe, J. Pharmacol. Exp. Ther. 193, 603 (1975). W. J. Kao, A. Hiltner, J. M. Anderson and G. A. Lodoen, J. Biomed. Mater. Res. 28, 819 (1994). J. A. Hunt, B. F. Flanagan, P. J. McLaughlin, I. Stickland and D. F. Williams, J. Biomed. Mater. Res. 31, 139 (1996). R. S. Labow, E. Meek and J. P. Santerre, J. Biomed. Mater. Res. 54, 189 (2001). R. S. Labow, E. Meek and J. P. Santerre, Biomaterials 22, 3025 (2001). R. S. Labow, E. Meek and J. P. Santerre, J. Biomed. Mater. Res. 39, 469 (1998). R. S. Labow, E. Meek and J. P. Santerre, J. Biomater. Sci. Polymer Edn 10, 790 (1999). J. P. Santerre, R. S. Labow, D. Erfle, D. Duguay and G. A. Adams, J. Biomed. Mater. Res. 28, 1187 (1994).

The effect of oxidation on hydrolysis of polyurethanes 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.

341

R. S. Labow, D. G. Duguay and J. P. Santerre, J. Biomater. Sci. Polymer Edn 6, 169 (1994). J. P. Santerre, R. S. Labow and G. A. Adams, J. Biomed. Mater. Res. 27, 97 (1993). J. P. Santerre and R. S. Labow, J. Biomed. Mater. Res. 36 (2), 223 (1997). J. P. Santerre, L. Shajii and H. Tsang, J. Dent. Res. 78, 1459 (1999). F. G. B. Wang, R. S. Labow and J. P. Santerre, J. Biomed. Mater. Res. 35, 207 (1997). F. G. B. Wang, R. S. Labow and J. P. Santerre, in: 23rd Amer. Soc. Biomater. Mtg., p. 451 (1997). F. G. B. Wang, J. P. Santerre and K. Solovey, in: 24th Amer. Soc. Biomater. Mtg., p. 393 (1997). Y. W. Tang, J. P. Santerre and R. S. Labow, J. Biomed. Mater. Res. 56, 516 (2001). Y. W. Tang, J. P. Santerre and R. S. Labow, J. Biomed. Mater. Res. 57, 597 (2001). Y. W. Tang, PhD Thesis, University of Toronto (2000). Y. Wu, C. Sellitti, J. M. Anderson, A. Hiltner, G. A. Lodoen and C. R. Payet, J. Appl. Polym. Sci. 46, 201 (1992). F. Li and D. Y. Hui, J. Biol. Chem. 272, 28666 (1997). R. S. Labow, D. J. Erfle and J. P. Santerre, Biomaterials 16, 51 (1995). S. J. Weiss, R. Klein, A. Slivka and M. Wei, J. Clin. Invest. 70, 598 (1982). A. N. J. Moore, H. Vakos, T. A. M. Neville, P. Tonge, P. Arya and G. W. Burton, Biotech. Tech. 10, 523 (1996). R. S. Labow, E. Meek and J. P. Santerre, J. Cell. Physiol. 186, 95 (2001). D. J. Erfle, J. P. Santerre and R. S. Labow, Cardiovasc. Path. 6, 333 (1997). V. Gajewski, Rubber World 15 (1990). M. A. Schubert, M. J. Wiggins, M. P. Schaefer, A. Hiltner and J. M. Anderson, J. Biomed. Mater. Res. 29, 337 (1995). S. Khatua and Y. L. Hsieh, J. Polym. Sci., Part A: Polym. Chem. 35, 3263 (1997). I. Revenko, Y. W. Tang and J. P. Santerre, Surf. Sci. 491, 346, (2001). G. Meijs, S. J. McCarthy, E. Rizzardo, Y.-C. and R. Chatelier, J. Biomed. Mater. Res. 27, 345 (1993). H. Bergstrand, T. Eriksson, A. Hallberg, B. Johansson, K. Karabelas, P. Michelsen and A. Nybom, J. Pharmacol. Exp. Ther. 262, 1334 (1992).


Novel dendrimer based polyurethanes for PEO incorporation X. DUAN 1 , C. M. GRIFFITH 2 , M. A. DUBÉ 1 and H. SHEARDOWN 3,∗ 1 Department

of Chemical Engineering, University of Ottawa, 161 Louis Pasteur St., Ottawa ON, K1N 6N5, Canada 2 University of Ottawa Eye Institute, 501 Smyth Rd., Ottawa ON, K1H 8L6, Canada 3 Departments of Chemical Engineering and Pathology and Molecular Medicine, McMaster University, 1280 Main St. W., Hamilton ON, L8S 4L7, Canada Received 2 January 2002; accepted 3 April 2002 Abstract—A series of segmented polyurethanes based on methylene diisocyanate/ poly (tetramethylene oxide) and chain extended with either ethylene diamine or butane diol in combination with a generation 2 polypropylenimine octaamine dendrimer were synthesized. For polymer synthesis, the dendrimers were protected with either t-boc or Fmoc groups and were incorporated into the polyurethane microstructure to permit further functionalization with biologically active groups. Following deprotection, the dendrimers were reacted with succinimidyl propionate polyethylene oxide (SPA-PEO) to improve the protein resistance of the polymers and to examine the potential of this technique for polymer functionalization. Different synthesis techniques were examined to optimize the incorporation of the PEO into the polymer microstructure. Incorporation of the dendrimers and the PEO were confirmed by NMR and FTIR. Gel permeation chromatography was used to examine the molecular weights of the various polyurethanes. The dendrimer incorporated polymers had significantly lower molecular weights than the ED or BDO chain extended controls, likely due to lower reactivity of the dendrimers as a result of steric factors. Following PEO reaction, the molecular weights of the resultant polymers were consistent with the levels of PEO incorporation noted by comparison of peak intensities in the NMR spectra. Due to the highly hydrophilic nature of the PEO, some migration to the polymer surface was expected. Water contact angles and XPS, used to characterize the surfaces, suggest that there was some PEO enrichment at the surface of the polymers. Adsorption of radiolabeled fibrinogen to the polymer surfaces was decreased by a factor of approximately 40% in some of the PEO incorporated polymers. There were also differences in the patterns of plasma protein adsorption on the various surfaces as evaluated by SDS PAGE and immunoblotting. Therefore, the use of dendrimers in biomaterials for incorporation of a large number of functional groups seems to be promising. Key words: Polyurethanes; PEO; dendrimers; protein adsorption.

∗ To

whom correspondence should be addressed.

344

X. Duan et al.

INTRODUCTION

Polyurethanes have seen widespread medical and biomedical use, particularly in blood contacting applications due to their superior mechanical properties and acceptable biocompatibility. Nevertheless, blood compatibility remains a significant limitation in the widespread application of these materials. A number of different methods for improving the blood and biocompatibility of these materials have been used, including functionalization of the bulk material [1– 5] and surface modification with various moieties [6– 12]. One of the more promising methods for improving blood compatibility of biomedical polymers involves modification with polyethylene oxide (PEO). The protein resistant properties of PEO are well known e.g. [13, 14] and as such, surface modification with PEO has been widely used to generate materials with superior blood compatibility. The protein resistant properties of PEO have been demonstrated in a number of studies to be dependent on the density of the polymer. As a result, star PEO [15] and comb-like PEO [16] have been examined as a means of improving the protein resistance of various surfaces. PEO incorporation into the polymer microstructure has also been used to improve biocompatibility. For example, Corneille et al. [17] incorporated PEO into the microstructure of a polyurethane backbone and demonstrated by surface plasmon resonance (SPR) that these polymers showed significantly reduced albumin adsorption. Santerre et al. [11] also demonstrated that PEO containing polyurethanes showed reduced protein adsorption. However, as expected, incorporation of PEO into a polyurethane backbone has been shown to decrease the mechanical properties of the polymer [18]. To overcome such limitations, Tan and Brash have developed novel polyurethane based pluronic equivalents, PEO-PU-PEO, which are coated onto the surface of polyurethanes. They have demonstrated significant reductions in the adsorption of plasma proteins with these coated polyurethanes, due to the presence of PEO at the polymer– biological interface [19]. Dendrimers are three-dimensional, highly ordered oligomeric and polymeric compounds formed by reiteration reaction sequences starting from smaller molecules or initiator cores [20]. These highly branched spherical polymers have well defined microstructures and a set number of terminal amine groups, which doubles (or triples or quadruples depending on the monomer selected) with each subsequent level of monomer addition. These molecules combine characteristics typical of small organic molecules, including monodispersity and definite composition with the attributes of traditional polymers including high molecular weight. These molecules have been examined in biomedical applications as in vitro gene transfer agents [21– 24], and in various therapeutic applications [25– 27]. Polyamidoamine dendrimers having poly(ethylene glycol) grafts were designed as a novel drug carrier with an interior suitable for the encapsulation of drugs and a biocompatible surface [28]. Dendrimers have also been shown to be useful in providing strong signal amplification in diagnostic applications [29]. Recently, fructose modified dendrimers have been used in biomaterials to generate surfaces with a high density of ligands for improving interactions between hepatocytes and surfaces [30]. There-

Novel dendrimer based polyurethanes for PEO incorporation

345

fore, as a result of their regular and multifunctional nature, these molecules seem promising in biomaterials applications. These molecules have not been widely examined for generating surfaces with higher levels of functional groups. Here we report on the development of methods for the incorporation of dendrimers into a polyurethane backbone via the chain extension step. This resulted in a polymer with a potentially large number of sites for further functionalization. Generation 2 poly (propylene imine) dendrimers were used in all studies. In the current work, these polymers were subsequently reacted with PEO and examined for plasma protein interactions. These results as well as extensive polymer characterizations are presented. MATERIALS AND METHODS

Polymer synthesis Control polyurethane synthesis. A conventional two step procedure was used in the synthesis of all polyurethanes used in the study [11]. To minimize the introduction of water, and the initiation of undesired side reactions, where possible, reactants were dried overnight under vacuum. The control polyurethanes were synthesized by reacting methylene diphenyl p isocyanate (MDI, Acros Organics, Nepean ON) with poly tetramethylene oxide (PTMO, Aldrich Chemical Co. Milwaukee WI, MW 650) dissolved in dry dimethyl sulfoxide (DMSO) (10%, w / v) in a 2 : 1 ratio in a dry nitrogen environment for a period of 90 min at 50◦ C. The polymers were chain extended with a 1 : 1 molar ratio (prepolymer : chain extender) of either ethylene diamine (ED, Adrich Chemical Co., Milwaukee WI) or butanediol (BDO, Aldrich Chemical Co., Milwaukee WI) in DMSO (2% w /v) for 2 h at room temperature in the case of ED or at 40◦ C in the case of BDO. Two to 3 drops of dibutyl tin dilaurate in tetrahydrofuran were added as catalyst when BDO was used as the chain extender. All polymers were purified by precipitation in water and extensive water washing prior to characterization. Several different techniques were used in an attempt to optimize the incorporation of the dendrimers into the polyurethane microstructure. The various methods are described below. PEO functionalization prior to dendrimer incorporation into polyurethane. A solution of polypropyleneimine octaamine dendrimer generation 2.0 (G2 ) (Aldrich Chemical Co., Milwaukee WI) and SPA-PEO with a reported molecular weight of 2000 (Shearwater Polymers, Huntsville, AL) in CH2 Cl2 in a 1 : 6 (partially modified) or 1 : 8 (fully modified) molar ratio was stirred at room temperature for 2 h to generate a statistical distribution of PEO modified dendrimers. Dendrimer modification was confirmed by GPC. These PEO modified dendrimers were then incorporated into the polyurethane backbone during the chain extension step by reacting the prepolymer in a 1 : 1 molar ratio with a mixture of standard chain extender and PEO modified dendrimer in a 9 : 1 molar ratio.

346

X. Duan et al.

Simultaneous chain extension of polyurethanes with dendrimers and reaction with PEO. Using a 6 : 1 molar ratio of SPA-PEO 2000 and dendrimers, and a 9 : 1 molar ratio of standard chain extender and dendrimer PEO mixture, the SPA-PEO 2000 and dendrimers were introduced into the polyurethane during the chain extension step in the following sequence : standard chain extender, PEO and dendrimer. This sequence was selected to minimize polymer crosslinking by the multifunctional dendrimers.

Dendrimer incorporation and subsequent PEO functionalization. In order to facilitate the incorporation of the dendrimers into the polymer backbone and their subsequent functionalization by PEO, the amine functional groups were first protected by either t-boc [31] or Fmoc chemistry. Briefly, for t-boc protection, the G2 polypropyleneimine dendrimer was reacted with N-t-Boc-L-Ala hydroxy succinimide ester (1.01 equivalent per NH2 end group at a molar ratio of 6 : 1 protecting group: dendrimer) overnight in CH2 Cl2 with triethylamine. Following the reaction, the protected dendrimers were washed with distilled water and saturated Na2 CO3 . The product was subsequently dried with H2 SO4 and the solvent removed by rotoevaporation to yield the protected dendrimers. Fmoc protection of the NH2 groups of the dendrimer was achieved by reaction of a G2 polypropyleneimine dendrimer with 9-fluorenylmethyloxycarbonyl chlorocarbonate (9-Fmoc) in CH2 Cl2 at a molar ratio of 6 : 1. The Fmoc was added slowly in 1– 2 ml aliquots and the mixture was reacted at 4◦ C for 4 h, followed by 8 h of reaction at room temperature. Following removal of the solvent by roto-evaporation, the product was dried under vacuum at 60◦ C to yield the Fmoc protected dendrimers. The t-Boc or Fmoc protected dendrimers were then incorporated into the polyurethane during the chain extension step in combination with either ED or BDO in a 9 : 1 molar ratio of standard chain extender to dendrimer. Following precipitation of the polymers in water and purification by extensive water washing and methanol extraction, the protection groups were removed. In the case of the t-boc protected dendrimer incorporated polyurethanes, stirring overnight in 99% formic acid facilitated deprotection. In the case of the Fmoc protected dendrimer incorporated polymers, deprotection occurred by reaction with diethylamine for a period of two hours. The deprotection step was followed by PEO incorporation. The deprotected dendrimer polyurethane was dissolved in DMF, and reacted with the theoretical maximum amount of SPAPEO 2000 for full dendrimer modification overnight. Following solvent removal, the final product was purified extensively prior to characterization. Films of the final polymers were cast from 10% solutions of the polymers in DMF for characterization. Several of the PEO incorporated polymers had poor mechanical properties. Films of these polymers were therefore cast onto the control polyurethanes for characterization. A summary of the various polyurethanes synthesized is presented in Table 1.


347

Table 1. Summary of polyurethanes synthesized and nomenclature Nomenclature

Chain extender

Dendrimer

PU-ED PU-G2 -ED2a PU-ED-PEO2a PU-ED-PEO2b PU-BDO PU-G2 -BDO2a PU-G2 -BDO2b PU-BDO-PEO2b PU-ED-PEO1

ED ED ED ED BDO BDO BDO BDO ED

No Yes Yes Yes No Yes Yes Yes Yes

1 Synthesized 2 Synthesized

Protection

Deprotection

t-Boc t-Boc Fmoc

Formic acid Diethylamine

t-Boc Fmoc Fmoc

Diethylamine

by serial addition of chain extender, followed by dendrimer and PEO. by reaction with protected dendrimers followed by deprotection and reaction with

PEO. a t-boc used for dendrimer protection. b Fmoc used for dendrimer protection.

Polymer characterization Gel permeation chromatography. Gel permeation chromatography was used to obtain molecular weights of the various polymers. The polyurethanes were dissolved in dimethyl formamide (DMF) at a concentration of 0.1% (w /v). The samples were analyzed using a Waters GPC system equipped with three Waters Styragel® HR4 GPC columns at room temperature. A solution of 0.1% (w /v) LiBr in DMF was used as the mobile phase. Polystyrene standards were used for calibration. Nuclear magnetic resonance. 1 H-NMR spectroscopy was performed to confirm the polymer microstructure using an AV-2000 spectrometer after dendrimer and using a Bruker AMX-500 after PEO incorporation. This was done using the peaks at 6.13 ppm and 8.37 ppm which are assigned to the protons in the urea groups formed by the reaction of NH2 with NCO of the prepolymer and the peak at 3.51 ppm ( CH2 O ) for PEO. All spectra were recorded in a mixture of deuterated DMSO and chloroform (v : v = 1 : 1). Fourier transform infrared spectroscopy. All polymers were analyzed using transmission FTIR spectroscopy with a Bio-Rad FTS-10 FTIR. Films were cast on a NaCl crystal window and dried under vacuum at 60◦ C for 24 h prior to recording. Water contact angles. Surface hydrophilicity was examined by measuring advancing water contact angles using a Rank Scherr Tumico 22-2000 series 14 inch horizontal beam bench comparator. A 10 µl water bubble was placed on a previ-

348

X. Duan et al.

ously dried and methanol rinsed polymer surface using a microsyringe. For each sample, 3– 5 separate drops were examined. X-ray photoelectron spectroscopy. XPS analysis was performed at the National Research Council of Canada. The surface of the samples was analyzed using a KRATOS AXIS HS X-ray photoelectron spectrophotometer (Kratos, Manchester UK). The size of the analyzed area was approximately 1 mm2 . Monochromatized AlKα radiation was used for excitation and a 180◦ hemispherical analyzer with a three channel detector was employed. The X-ray gun was operated at 15 kV and 20 mA. The spectrophotometer was operated in Fixed Analyzer Transmission (FAT) mode throughout the study using electrostatic magnification. Surface and highresolution spectra were collected using a 160 and 20 eV pass energy respectively. The pressure in the analyzer chamber was 10−8 to 10−9 torr. An electron flood gun was used to neutralize the charge during the experiment. Binding energies were referenced to the carbon-carbon bond that was assigned a binding energy of 285 eV. Atomic composition was estimated using standard software provided with the instrument using the following sensitivity factors: 0.25 for C1s, 0.66 for O1s and 0.42 for N1s relative to F1s at 1.00. Peak deconvolution was performed using the software provided with the instrument. Protein adsorption Adsorption of radiolabeled fibrinogen from buffer. Human fibrinogen (Enzyme Research Labs, South Bend IN) was labeled with 125I (Na125 I, 0.5 mCi, Amersham, Arlington Heights IL) using the iodogen method. Free iodide was removed by overnight dialysis at 4◦ C with three changes of the phosphate buffered saline (PBS) dialysate. Free iodide was determined by trichloroacetic acid precipitation of the protein. Fibrinogen concentration in the final solution was determined spectrophotometrically. A 0.25% labeled 1 mg/ ml fibrinogen solution in PBS was prepared and diluted for adsorption experiments. 6 mm diameter polyurethane discs were preincubated overnight in PBS at 4◦ C. The surfaces were then incubated with 250 µl of the radiolabeled fibrinogen solution for three hours. Following the adsorption, the surfaces were dip-rinsed three times in PBS, and the radioactivity determined using a gamma counter. The amount of adsorbed fibrinogen was determined by comparison of the results to known standards. SDS PAGE and immunoblotting. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting were used to examine the patterns of protein adsorption from plasma to the various surfaces. The surfaces were incubated in citrated pooled normal plasma (>25 donors) for two hours. Following 3 dip rinses in PBS to remove loosely bound proteins, the surfaces were incubated overnight in 2% (wt) SDS at 4◦ C to elute the adsorbed proteins. SDS-PAGE and


349

immunoblotting were performed as previously described [32]. Total protein adsorption was determined based on bovine serum albumin (BSA, Sigma Chemical Co., St. Louis MO) standards using a Bio-Rad DC total protein microassay.

RESULTS

Polymer synthesis In the current work, two approaches were used to incorporate dendrimers and additional functionality via PEO into a polyurethane microstructure. A summary of the polymers synthesized and the nomenclature used to describe them is shown in Table 1. In the first approach, a star PEO dendrimer was synthesized by reacting SPA-PEO 2000 with poly (propylene imine octaamine) G2 dendrimer at a molar ratio of 6 : 1 (PEO : dendrimer). This approach should leave, on average, two free amine groups for polyurethane chain extension. To overcome the steric hindrance problem, a modification was made to the method. SPA-PEO2000 and dendrimers were introduced in sequence during the chain extension step, with the standard chain extender (i.e. ED or BDO) first, SPAPEO2000 second, and the G2 dendrimer last. The sequence was selected in order to minimize crosslinking by the multifunctional dendrimers. A number of parameters, including the reaction temperature, and speed of reactant addition, affected the chain extension reaction. However, at the determined optimal conditions, summarized in Table 2, a DMSO and DMF soluble polyurethane was formed (denoted PU-EDPEO1). While actual tensile testing was not performed due to the need for large and regular films, this polymer appeared to be similar if not superior mechanically to the control polyurethanes, likely the result of some crosslinking during the reaction. An alternative approach, involving protection/ deprotection of the amine groups in the dendrimer for chain extension and subsequent PEO attachment was also examined. The dendrimers were partially protected with the amine protecting groups t-Boc or Fmoc at a molar ratio of 1 : 6, leaving on average two free Table 2. Optimal conditions for the synthesis of dendrimer incorporated polyurethane with PEO attached (PUED-PEO1 ) Prepolymerization temperature (◦ C) Prepolymerization duration (min) Chain extension temperature (◦ C) Chain extension duration (min) Standard chain extender concentration (%, w/v) Dendrimer G2 concentration (%, w/ v) Chain extension sequence Dropping speed of dendrimer solution (drops/ s) Post chain extension temperature (◦ C) Post chain extension duration (min)

50 120 25 60 2.0 2.0 ED/ BDO→SPA-PEO→dendrimer 1 90 60

350

X. Duan et al.

Table 3. Stability of ED chain extended polyurethanes to formic acid treatment overnight: GPC results Sample ED chain extended control BDO chain extended control Formic acid treated ED PU Formic acid treated BDO PU

Mn

Mw

82 800 49 400 68 900 30 300

133 100 64 700 110 000 48 900

amine groups for the chain extension reaction. The protected dendrimers were incorporated during the chain extension step with either ED or BDO at a ratio of 9 moles of standard chain extender to one mole dendrimer and the mixture was reacted as in the case of the controls. The resulting polyurethanes (denoted PU-G2 ED2a for the t-Boc protected polymers, which were ED chain extended and PU-G2 BDO2b for the Fmoc protected polymers which were BDO chain extended) showed no macroscopically apparent signs of crosslinking, suggesting that there were no significant amounts of unprotected dendrimer present in the chain extension reaction mixture. The PEO incorporated polyurethanes were synthesized following the appropriate deprotection technique. Upon precipitation in water, the previously t-Boc protected PEO — polyurethanes formed a stable emulsion, necessitating solvent removal by roto-evaporation. Since the acid treatment used for deprotection breaks urea bonds between the protecting group and the dendrimer, the effect of the acid treatment on ED and BDO chain extended polymer controls was examined. GPC analysis results for the untreated polymers and those treated with acid are summarized in Table 3. It is clear from these results that there is degradation of the polymer by the acid treatment, with a 17% decreased noted in the molecular weight of the ED chain extended polymers and a 39% decrease noted in the molecular weight of the BDO chain extended polymers. However, the molecular weights of the polymers remained significant, suggesting that a combination of factors, including polymer degradation and increased hydrophilicity as a result of the PEO incorporation likely resulted in the formation of the emulsion. No emulsion formed following PEO incorporation into the Fmoc protected polymers, providing additional evidence that the degradation of the polyurethane chains contributed to the formation of the emulsion. The final PEO polyurethanes did not possess the superior mechanical properties traditionally associated with polyurethanes, and it was difficult to obtain workable films for subsequent tests with these polymers. This hindered the absolute measurement of the mechanical strength of these polymers. Polymer characterization 1

H-NMR and FTIR. 1 H-NMR spectra were used to confirm the microstructure of the polymers based on peak assignments. Spectra for the ED chain extended polyurethanes before and after dendrimer and PEO modification are shown in Fig. 1.


351

Table 4. Amount of PEO incorporated into polyurethanes as determined by 1 H-NMR Polymer

PEO amount (g/ g of G2 dendrimer)

Theoretical maximum PEO amount (g/ g of G2 dendrimer)

PU-ED-PEO2a PU-ED-PEO1 PU-BDO-PEO2b

1.57 12.36 10.02

15.52 15.52 15.52

Of specific interest are the peaks at 6.13 and 8.37 ppm, which are assigned to protons in the urea groups formed by the reaction of NH2 (from ED or dendrimer) with NCO of the prepolymer. These were found to increase slightly with incorporation of the dendrimer prior to deprotection, as expected, due to the generation of a urea linkage by the protection reaction. However, the presence of existing urea groups in the polyurethane urea structure and the relatively low amount of dendrimer incorporated make it difficult to evaluate the level of incorporation. A PEO peak is clearly seen in Fig. 1c at 3.51 ppm following deprotection and reaction with PEO. The incorporation of the dendrimers into the polyurethane microstructure was shown more clearly with the BDO chain extended polymers due to the lack of urea groups in the base polymers. Quantitative analysis of the spectra, summarized in Table 4, show that relative to the dendrimer peaks at 6.13 and 8.37 ppm, the PU-ED-PEO2a had the least amount of incorporated PEO, while PU-ED-PEO1 had the most PEO chains. Similar results were noted in the FTIR spectra of the various polymers. Compared with the ED chain extended polyurethane control, the dendrimer incorporated polyurethane, had increases in peaks at 3315, 1708 and 1647 cm−1 , which are indicative of increases in hydrogen bonding of N H and C O of urethane and urea, probably due to the incorporation of dendrimers. Following PEO attachment, the FTIR spectrum for the ED chain extended polymer (PU-ED-PEO1), shows decreased peaks at 3315, 1708 and 1647cm−1 , indicative of the lower hydrogen bonding in these polymers. Gel permeation chromatography. GPC analysis of the modified dendrimers and the polymers, shown in Table 5, demonstrated that star PEO dendrimers had been formed and that the dendrimers modified at a 6 : 1 molar ratio were not completely modified compared to those modified at an 8 : 1 ratio. However, these measured molecular weights were lower than those expected based on complete reaction, likely due to a combination of steric factors, the use of polystyrene rather than PEO standards in the analysis, the fact that for branched polymers such as those in the current study, the measured molecular weights will be lower than the actual values or simple changes in the hydrodynamic volume based on the new chemical structure. Similarly, the polyurethane molecular weights were lower what would be expected for full PEO modification. Nonetheless, these measurements are useful to reveal important trends in the data. As expected, based on the reactivity of

352

X. Duan et al.

Table 5. Polymer molecular weights determined by GPC Polymer

Mn

Mw

PU-ED PU-G2 -ED2a PU-ED-PEO2a PU-BDO PU-G2 -BDO2a PU-G2 -BDO2b PU-BDO-PEO2b PU-ED-PEO1

82 800 67 800 62 600 49 400 30 400 76 400 86 000 138 700

133 100 106 300 100 600 64 700 48 800 117 000 135 900 214 800

Figure 1. Typical 1 H-NMR spectrum of the synthesized polymers. Shown is (a) the ED chain extended polyurethane control (PU-ED), (b) dendrimer modified ED chain extended polyurethane (PU-G2 -ED2a ) and (c) the PEO modified ED chain extended PU (PU-ED-PEO2a). Of interest are the peaks at 6.13 ppm and 8.37 ppm (as shown), indicative of the formation of urea bonds in both polymers but to a greater extent in the dendrimer modified polymer as expected and the presence of the PEO at 3.51 ppm as shown.


353

isocyanates and amine groups relative to hydroxyl groups, the molecular weights of the ED chain extended polymers were significantly greater than those of the BDO chain extended polymers. The dendrimer incorporated polyurethanes had significantly lower molecular weights than control polyurethanes when t-Boc was used (PU-G2 -ED2a compared to PU-ED, and PU-G2 -BDO2a to PU-BDO), which is likely the result of the lower reactivity of the dendrimers due to steric factors. The polymers synthesized using Fmoc protected dendrimers had higher molecular weights than those synthesized using the t-Boc protected dendrimers (PU-G2 BDO2b compared to PU-G2 -BDO2a ). This, we hypothesize, was due to slight crosslinking in the polymer by unprotected dendrimer when Fmoc was used, as the t-Boc protected dendrimers were extensively purified prior to chain extension. The GPC results also demonstrate that no unreacted dendrimers remained in the polyurethanes following reaction and purification, as there were no peaks other than the polyurethane peak present. Following PEO attachment, the t-Boc protected dendrimer modified polyurethanes showed a slight but unlikely significant decrease in molecular weight. (PU-EDPEO2a compared to PU-G2 -ED2a ). While the incorporation of PEO would be expected to result in an increase in the molecular weight, it seems likely that degradation of the polymer backbone by the acid treatment used for deprotection affected the resulting polymer molecular weights. When the Fmoc protection was used, the molecular weight increased after PEO attachment, as expected. Taken together, the NMR, FTIR and GPC results provide strong evidence that the dendrimers and the PEO have been incorporated into the polymer microstructure and that differences in the attachment methods and protection methods resulted in differences in the amount of PEO incorporated in the polymer microstructure.

Water contact angles. Advancing water contact angles on the various polymers are summarized in Fig. 2. It can be seen that water contact angles measured on the PU-ED and PU-BDO controls were similar at approximately 70◦ . Following incorporation of the dendrimer, there was a small decrease in the water contact angle relative to the respective control. In the case of the ED polymer (PU-G2 -ED2a compared to PU-ED), this decrease was not significant, as might be expected due to the similarity in the microstructure of the dendrimer and ED. However, in the case of the BDO chain extended polymers (PU-G2 -BDO2b compared to PU-BDO), this decrease was significant and suggests surface differences between these polymers. Following incorporation of the PEO into the polymer microstructure (i.e. PU-EDPEO2a and PU-BDO- PEO2b), the contact angles showed a significant (p > 0.99) decrease to approximately 45◦ with both chain extenders. While these advancing water contact angles are higher than those noted with PEO surface modification, which average between 15 and 30◦ depending on the underlying substrate, they are similar to the results obtained by others for PEO grafted polyurethanes [33].

354

X. Duan et al.

X-Ray photoelectron spectroscopy (XPS). XPS results, summarized in Table 6, provide additional evidence for dendrimer incorporation into the polymer microstructure and presence at the surface. Specifically, there was an increase in the N1s signal on the dendrimer incorporated polyurethanes relative to the control polymers. For the ED chain extended polymers, the high resolution C1s envelope showed a significant increase in the contribution at 289.2 eV, consistent with an

Figure 2. Advancing water contact angles measured on the various polyurethanes. The PEO polymers in this case were PU-ED-PEO2a and PU-BDO-PEO2b. Significant decreases in the water contact angles were noted on the surfaces following PEO incorporation. These differences did not seem to be a function of the amount of PEO incorporated based on the NMR data. Table 6. Summary of XPS results on modified polyurethanes Sample

PU-ED PU-G2 -ED2a PU-ED-PEO2a (TFA) PU-ED-PEO2a (FA) PU-BDO PU-G2 -BDO2b PU-BDO-PEO2b

C1s Total

284.8 C C

286.3 C N, C O

289.2 urea

289.6 COOR

76.3 76.3 78.0 75.7 77.2 77.8 76.8

50.6 50.5 51.8 46.1 57.9 56.2 47.7

24.6 24.5 22.1 26.8 17.6 20.1 28.1

0.4 0.62 4.02 2.02 1.65 0.90 0.48

0.68 0.72 0 0.73 0 0.66 0.98

O1s

N1s

21.4 21.1 18.2 20.3 21.3 19.4 19.7

2.28 2.65 3.86 4.04 1.50 2.83 3.44


355

increase in the relative atomic concentration of C-OR, which would be expected following PEO incorporation. However, a similar increase was not noted on the BDO chain extended polymers. Furthermore, there was a decrease in the O1s signal on both the ED and BDO chain extended polymers following PEO incorporation. As well, a consistent increase in the N1s signal was noted. These results suggest surface enrichment of the dendrimers and the hard segment of the polyurethane in these polymers, likely due to the highly hydrophilic PEO chains being buried in the bulk polyurethane matrix when in the hydrophobic XPS environment. It is difficult however, to obtain an accurate picture of the surface composition in a biologically relevant aqueous environment however. Protein adsorption Adsorption of 125 I fibrinogen from buffer. Fibrinogen adsorption results to the various surfaces are summarized in Figs 3 and 4. For all experiments, the free iodide was determined to be less than 0.5%, and therefore free iodide is not expected to contribute to the measured amount of adsorbed fibrinogen on these polymers. A slight decrease in the amount of adsorbed fibrinogen was consistently noted following incorporation of the dendrimer. Following PEO incorporation, the decrease in fibrinogen adsorption was more significant. Furthermore, the effect of dendrimer and PEO incorporation on fibrinogen adsorption seemed more apparent in the BDO chain extended polymers. This result correlates well with the water contact angle measurements, which show a significant decrease in the hydrophilicity of the surfaces following dendrimer and PEO incorporation. On the t-Boc protected, ED chain extended polyurethanes, there was no significant difference between the amounts of fibrinogen adsorbed with the different acids used for deprotection purposes, i.e. formic acid and trifluoroacetic acid. Consistent with the NMR results, which showed that the level of PEO incorporation was significantly less when

Figure 3. Fibrinogen adsorption from buffer to the various ED chain extended polyurethanes. A decrease in fibrinogen adsorption was noted following dendrimer incorporation into the polymer microstructure and again following PEO incorporation. Furthermore, the trend follows the NMR results for level of PEO incorporation into these polymers.

356

X. Duan et al.

Figure 4. Fibrinogen adsorption from buffer to the various BDO chain extended polymers. As with the ED chain extended polymers, a decrease in fibrinogen adsorption was noted following dendrimer incorporation and again following PEO incorporation. However, the reduction in fibrinogen adsorption was not as great as with the ED chain extended polymers.

(a) Figure 5. Immunoblot results for plasma protein adsorption to the various polymers. With both chain extenders, there was a slight decrease in protein adsorption overall with the incorporation of the dendrimers and an even greater decrease with the incorporation of PEO chains into the polymer with the qualitative trends again following the measured levels of PEO incorporation. (a) PU-ED, (b) PUG2 -ED, (c) PU-ED-PEO1, (d) PU-ED-PEO2a, (e) PU-ED-PEO2b, (f) PU-BDO, (g) PU-G2 -BDO, (h) PU-BDO-PEO2b. The two right hand lanes on the blots represent molecular weight standards and the acrylamide gel of all adsorbed proteins.


357

t-Boc was used for dendrimer protection (PU-ED-PEO2a), there was no significant difference in the level of protein adsorption between the dendrimer incorporated and the PEO incorporated polymers. However, the polyurethane synthesized by sequential addition of the various reagents (PU-ED-PEO1) showed the greatest decrease in fibrinogen adsorption relative to the control, with an almost 40% decrease in the amount of adsorbed fibrinogen. This result is again consistent with the GPC and NMR results that suggest higher levels of PEO incorporation in these polymers. SDS PAGE and immunoblotting. SDS PAGE and immunoblotting analyses of the eluates following protein adsorption from plasma were performed to determine the patterns of plasma protein adsorption on the various surfaces. Immunoblots are presented in Fig. 5. Consistent with the fibrinogen adsorption results, there was a decrease in the adsorption of most of the proteins on the surfaces following incorporation of the dendrimers and following PEO attachment. On both the ED and BDO chain extended polyurethanes, decreases were noted in the amounts of adsorbed fibrinogen, C3, transferrin, albumin, vitronectin, apolipoprotein A1 as well as the complement proteins. However, the banding patterns indicated that there were no changes in adsorption of some of the other proteins, possibly the result of the relatively low amounts of adsorption noted on the control polymers. These low levels of adsorption to the base polyurethanes are consistent with the results of others [19] and are therefore not unexpected. While optical scans of the gold stained

(b) Figure 5. (Continued).

358

X. Duan et al.

(c)



(e)

(f) Figure 5. (Continued).

359

360

X. Duan et al.

(g)

(h) Figure 5. (Continued).


361

gels and immunoblots were obtained and subsequently analyzed using Whole Band Analysis software (BioImage) to determine band molecular weights and intensities, gel/ blot data are at best semiquantitative and the intensities of the bands depend on the characteristics of the different antigen-antibody responses and colour reactions.

DISCUSSION

In the current work, we examined the potential of using dendrimers for incorporating higher levels of functional groups into the microstructure of polyurethanes for biomaterials applications. Since surface modification with PEO has been shown in numerous studies to significantly reduce protein adsorption, PEO has been widely used in biomaterials applications and its effects have been shown to be density dependent, we examined the potential of dendrimer incorporation into polymers and subsequent PEO modification to increase the protein resistance of the polymers. Several synthesis methods were used in order to maximize the conversion of reactive groups in both the dendrimers and in the polymers. Not surprisingly, modification of the dendrimers with the PEO and subsequent incorporation into the polyurethane microstructure during the chain extension step failed to result in high molecular weight products, likely a result of steric factors. However, sequential addition of the various reagents in an order derived to minimize polyurethane crosslinking while maximizing PEO incorporation resulted in mechanically strong yet solvent soluble polyurethanes. A variety of protection/ deprotection strategies were also used to develop PEO incorporated polyurethanes. With the protection/ deprotection approach, high molecular weight polyurethanes were synthesized in all cases. However, when t-Boc was used to protect the amine groups in the dendrimer, the acid deprotection step apparently affected the polyurethane backbone, decreasing the molecular weight of the resultant polymer. Based on these results, it is therefore preferable to use Fmoc for dendrimer protection. In all cases, the mechanical properties of the polyurethanes synthesized using the protection/ deprotection strategy followed by PEO reaction were poorer than the control polymers, despite the fact that a significant amount of the base chain extender was used in the synthesis of these polymers. While dendrimer incorporation did result in polyurethanes with marginally lower mechanical strength than that controls, the PEO incorporated polyurethanes could not be cast into workable films and coating with these polymers was necessary. It would therefore be expected that the incorporation of either higher relative amounts of dendrimer or higher dendrimer generations would have a further negative impact on these properties. However, based on the results of Tan and Brash [19], it is possible that these polyurethanes could be used with standard ED or BDO chain extended polyurethanes either as coatings or as blended films, thereby improving their mechanical integrity. NMR, FTIR and GPC confirmed incorporation of both the dendrimer and the PEO into the polyurethane microstructure. Furthermore, comparison of the amounts of PEO incorporated into the polyurethane chains as determined by peak ratios in

362

X. Duan et al.

NMR and by molecular weight changes determined by GPC demonstrates that the greatest amounts of PEO were incorporated by simultaneous reaction with the chain extender, PEO and dendrimer. Consistent with the degradation problems during the deprotection step in the t-Boc protected polymers, the BDO chain extended polymers reacted with the Fmoc protected dendrimers and subsequently modified with PEO showed significantly greater amounts of PEO incorporation than those using the t-Boc protected dendrimers. The amount of PEO incorporated obtained from the NMR spectra and that obtained from the GPC results were surprisingly consistent. Advancing water contact angles measured on the various polymer surfaces decreased significantly following PEO incorporation, although the measured angles remained somewhat higher than those noted in other studies following surface modification with PEO, suggesting that the mobility of the PEO is not adequate to generate surfaces with high PEO concentrations. However, it is also possible that with higher PEO concentrations obtained by using a higher dendrimer : standard chain extender ratio or higher dendrimer generations could also result in surfaces with higher PEO concentrations and therefore lower advancing water contact angles. XPS analysis of the various polyurethane surfaces demonstrates that some dendrimer and PEO were present on the surfaces. Specifically, relative to the control polyurethanes, there was an increase in the N1s signal following dendrimer incorporation, indicative of dendrimer enrichment at the polymer surface. A further increase in the N1s signal was noted following PEO incorporation, which may suggest enrichment of the dendrimer at the surface since this contribution was found to be predominantly from amine rather than isocyanate groups. For all of the ED chain extended polymers after PEO incorporation, there was an increase in the contribution of COR in the C1s envelope, which may be indicative of the presence of PEO. However, this was not accompanied by an increase in the O1s signal. It is likely that the relatively complex chemistry of these polymers, relatively low amounts of PEO incorporated and the hydrophobicity of the XPS environment contributed to the lack of clear evidence of PEO surface enrichment in these polymers. However, in agreement with the water contact angle measurements, it is clear that there is a significant amount of standard polyurethane present on the surfaces, suggesting that incorporation of higher amounts of PEO may be possible. The literature suggests that the levels of protein adsorption to PEO modified surfaces are dependent on such factors as the PEO density and molecular weight e.g. [15, 18]. Relative to the control polyurethanes, the levels of fibrinogen adsorption decreased following incorporation of the dendrimer in the ED series of polyurethanes. However, relative to the dendrimer incorporated polyurethanes, there was no decrease in the adsorption of fibrinogen following PEO incorporation. This is not surprising given that with the t-Boc protection method, both GPC and NMR analysis suggested that very low levels of PEO were present in these polymers. However, on the surface that showed the highest level of PEO incorporation, there was a significant decrease (approximately 40% relative to the control) in the


363

adsorption of fibrinogen from buffer. On the BDO series of polyurethanes, using Fmoc protection for the dendrimers, a similar decrease in the adsorption relative to the controls was noted following incorporation of the dendrimers. Furthermore, there was a small but significant decrease in fibrinogen adsorption following PEO incorporation. However, not surprisingly, the levels of fibrinogen adsorption remained higher than those noted on the ED-PEO-1 polymers. Therefore, the protein adsorption results correlate well with the NMR and GPC results, with lower levels of fibrinogen adsorption noted on the surfaces that showed higher amounts of attached PEO. SDS-PAGE and immunoblotting results were also consistent with the levels of PEO incorporation into the polymers, with decreases in plasma protein adsorption noted following dendrimer and PEO incorporation. While it would obviously be desirable to see a greater reduction in the fibrinogen adsorption since the levels noted in the current work would not likely significantly alter the response of the system to blood, the relatively low levels of dendrimer used and the low generation of dendrimer selected could potentially be altered to increase the protein resistance of these surfaces. It seems likely increasing the dendrimer and PEO content of these polymers, which in the current work was quite low, can further decrease that protein adsorption. Furthermore, only molecular weight of PEO was examined in the current study and it is possible that the mechanical and/ or biological properties of the polymers can be further enhanced through the use of different chain lengths of PEO.

CONCLUSIONS

In the current work, dendrimer incorporation into the microstructure of polyurethanes was examined, based on the hypothesis that these multifunctional polymers could be used to further functionalize with the polymers with large numbers of biologically relevant molecules. Modification of the dendrimers with PEO for improving protein resistance was examined in the current work. A number of strategies were used to incorporate the dendrimers and PEO into the polymer microstructure, including simultaneous reaction during polyurethane synthesis and the use of different amine protection groups. Relatively low levels of dendrimer to standard chain extender were examined in these studies in order to develop polymers with reasonable mechanical properties. NMR and FTIR confirmed dendrimer and PEO incorporation into the polymer microstructure. Gel permeation chromatography was used to confirm the molecular weights of the various polymers. The results suggested that a highest level of PEO in the polymers was achieved using the simultaneous reaction approach. Furthermore, physical observations and GPC analysis demonstrated that t-Boc protection of the dendrimers was not optimal as the deprotection step resulted in degradation of the polymer chains and relatively low levels of PEO incorporation. Surface analysis of the polymers using advancing water contact angles and XPS suggested some PEO enrichment at the polymer surfaces, with the potential for improvement through the use of greater amounts of dendrimers in the chain

364

X. Duan et al.

extension step. On the polymers with the highest levels of PEO incorporated, a 40% decrease in the adsorption of fibrinogen was noted. Furthermore, immunoblotting results suggest that there were decreases in the adsorption of a number of plasma proteins on these polymers. Therefore, the results of the current study demonstrate that dendrimers can be used in biomaterials applications in order to generate surfaces or bulk polymers with higher levels of functional groups, including PEO and other biologically relevant molecules such as cell adhesion peptides. Acknowledgements The authors gratefully acknowledge the technical assistance of J. Tan, S. Yonson and M. Bergeron. As well, we would like to thank Dr. Farid Bensebaa for XPS analysis. Funding from the Natural Sciences and Engineering Research Council and the University of Ottawa is gratefully acknowledged.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

L. Y. Yung and S. L. Cooper, Biomaterials 19, 31 (1998). J. N. Baumgartner, C. Z. Yang and S. L. Cooper, Biomaterials 18, 831 (1997). F. Lim, X. H. Yu and S. L. Cooper, Biomaterials 14, 537 (1993). J. H. Silver, H. B. Lin and S. L. Cooper, Biomaterials 14, 834 (1993). J. P. Santerre, P. ten Hove, N. H. VanderKamp and J. L. Brash, J. Biomed. Mater. Res. 26, 39 (1992). W. G. McClung, D. L. Clapper, S. P. Hu and J. L. Brash, Biomaterials 22, 1919 (2001). G. A. Skarja and J. L. Brash, J. Biomed. Mater. Res. 34, 439 (1997). H. B. Lin, W. Sun, D. F Mosher, C. Garcia-Echeverria, K. Schaufelberger, P. I. Lelkes and S. L. Cooper, J. Biomed. Mater. Res. 28, 329 (1994). H. B. Lin, K. B. Lewis, D. Leach-Scampavia, B. D. Ratner and S. L. Cooper, J. Biomater. Sci. Polymer Edn 4, 183 (1993). K. A. Woodhouse and J. L. Brash, Biomaterials 13, 1103 (1992). J. P. Santerre, P. ten Hove and J. L. Brash, J. Biomed. Mater. Res. 26, 1003 (1992). T. G. Grasel, D. G. Castner, B. D. Ratner and S. L. Cooper, J. Biomed. Mater. Res. 24, 605 (1990). J. H. Lee, J. Kopecek and J. D. Andrade, J. Biomed. Mater. Res. 23, 351 (1989). J. H. Lee, H. B. Lee and J. D. Andrade, Prog. Polym. Sci. 20, 1043 (1995). E. W. Merrill, J. Biomater. Sci. Polymer Edn 5, 1 (1993). P. Banerjee, D. J. Irvine, A. M. Mayes and L. G. Griffith, J. Biomed. Mater. Res. 50, 331 (2000). S. Corneillie, P. N. Lan, E. Schacht, M. Davies, A. Shard, R. Green, S. Denyer, M. Wassall, H. Whitfield and S. Choong, Polymer International 46, 251 (1998). J. H. Silver, C. W. Myers, F. Lim and S. L. Cooper, Biomaterials 15, 695 (1994). J. Tan and J. L. Brash, Trans. Soc. Biomaterials. 23, 499 (2001). D. Tomalia, M. Naylor and W. Goddard, Angew. Chem. Inter. Ed. 29, 138 (1990). J. Haensler and F. C. Szoka, Jr., Bioconj. Chem. 4, 372 (1993). A. Bielinska, J. F. Lakowska-Latallo, J. Johnson, D. A. Tomalia and J. R. J. Baker, Nucleic Acid Res. 24, 2176 (1996). J. F. Kukowska-Latallo, A. U. Belinska, J. Johnson, R. Splindler, D. A. Tomalia and J. R. J. Baker, Proc. Natl. Acad. Sci. USA 93, 4897 (1996). M. X. Tang, C. T. Refemann and F. C. Szoka, Jr., Bioconj. Chem. 7, 703 (1996).


365

25. J. P. Defoort, B. Nardelli, W. Huang, D. D. Ho and J. P. Tam, Proc. Natl. Acad. Sci. USA 89, 3879 (1992). 26. V. V. Martin, W. H. Ralston, M. R. Hynes and J. F. W. Keana, Bioconj. Chem. 6, 616 (1995). 27. R. Roy, D. Zanini, S. J. Meunier and A. Romanowska, J. Chem. Soc. Series Chem. Commun. 24, 1869 (1993). 28. C. Kojima, K. Kono, K. Maruyama and T. Takagishi, Bioconj. Chem. 11, 910 (2000). 29. S. Capaldi, R. C. Getts and S. D. Jayasena, Nucleic Acids Research 28, E21 (2000). 30. M. Kawase, T. Shiomi, H. Matsui, Y. Ouji, S. Higashiyama, T. Tsutsui and K. Yagi, J. Biomed. Mater. Res. 54, 519 (2001). 31. J. F. G. A. Jansen, E. M. M. de Brabander-van den Berg and E. W. Meijer, Science 266, 1226 (1994). 32. R. Cornelius and J. L. Brash, J. Biomater. Sci. Polymer Edn 4, 291 (1993). 33. K. D. Park, A. Z. Piao, H. Jacobs, T. Okano and S. W. Kim, J. Polymer Sci. 29, 1725 (1991).


Identification of biodegradation products formed by L -phenylalanine based segmented polyurethaneureas S. L. ELLIOTT 1 , J. D. FROMSTEIN 1 , J. P. SANTERRE 2 and K. A. WOODHOUSE 1,∗ 1 Department

of Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Ontario, M5S 3E5, Canada 2 Faculty of Dentistry, University of Toronto, 12H Edward St., Ontario M56 166, Canada Received 29 November 2001; accepted 2 April 2002 Abstract—The degradation of novel biodegradable segmented polyurethanes was investigated with a view to determining the cleavage points within the polymer backbones targeted by the enzyme chymotrypsin. While the materials were developed with specific enzyme cleavage sites designed into the polymer chains, the nature of their degradation had not yet been determined. In this work, two segmented polyurethaneureas containing L -phenylalanine residues in the chain extender and two control polymers were subjected to degradation in the presence of chymotrypsin. Samples were collected for analysis over a time period from 1 day to 8 weeks. The degradation products from these materials were isolated using solid phase extraction and reversed phase high pressure liquid chromatography, and identified using mass and tandem mass spectrometry. Three hard segment related degradation products were identified and provide important insight into the polyurethane backbone cleavage sites. Cleavage of urea, ester and urethane bonds were observed. The results confirmed that chymotrypsin was able to cleave ester bonds adjacent to phenylalanine residues contained within the novel chain extender. Key words: Polyurethaneurea; biodegradation; chymotrypsin; L -phenylalanine; mass spectrometry.

INTRODUCTION

Biodegradable polymeric materials are currently of great interest for use as medical implants and drug delivery vehicles. These materials, designed to degrade into non-toxic components over a pre-determined period can either degrade once they have completed their intended function, or may achieve their goal via degradation itself [1– 6]. Synthetic bioadsorbable sutures, invented in the 1960s are an example of a commonly used degradable polymeric material [7]. The biodegradable ∗ To


368

S. L. Elliott et al.

approach has the potential to alleviate a number of the problems associated with non-degradable implants, such as long-term safety and implant removal [1, 3, 5, 6]. Several recent candidates for use in biodegradable polymeric materials include poly(ortho esters), polyanhydrides, polyphosphazenes, poly-ε-caprolactone and its copolymers, polyglycolic acid (PGA), polylactic acid (PLA), polydioxane, and degradable segmented polyurethanes [8– 13]. Segmented polyurethanes are an important class of biomaterials with a long history of use as medical implants due to their excellent physical properties and relatively good biocompatibility [14]. A broad range of physical properties can be achieved by varying the chemistries and molecular weights of the various components, and through manipulation of the ratios in which they are reacted [14]. Material properties can range from very brittle and hard materials to soft, tacky, viscous ones [14]. While relatively few biodegradable segmented polyurethanes exist, due in part to concerns about the toxicity of potential degradation products, the development of an isocyanate based on the amino acid lysine opened up the possibility of synthesizing polyurethanes whose ultimate degradation products include lysine [4, 15, 16]. The residence time of a polymeric implant in an organism can be controlled by changing the quantity of cleavable links present in the polymeric chain [17]. The most common method of introducing hydrolysable linkages into polyurethanes has been to incorporate soft segments such as polylactides and ε-polycaprolactone into the backbone [18]. An alternative is to utilize hydrolysable groups in the hard segment, particularly in the chain extender. If hydrolysable hard segments are used, a wider variety of soft segment chemistries can be introduced [15]. This increases the range of material properties attainable without altering the degradable portion of the polyurethane [15]. In addition to simple hydrolysis, enzyme mediated hydrolysis has long been known to play a role in polyurethane biodegradation [19– 21]. Such enzymatic degradation may be encouraged through the incorporation of suitable amino acids into the polymer’s backbone [22]. Kartvelishvili et al. [23] synthesized degradable non-segmented poly(ester urethane)s containing L-phenylalanine, and found that these materials exhibited an enhanced susceptibility to chymotrypsin mediated degradation. Chymotrypsin-like serine proteinases are implicated in a wide variety of pathological states including inflammation (cathepsin G) and cardiomyopathy (mast cell chymase) [24, 25]. Chymase is also released in response to allergens or other challenges [26]. Chymotrypsin hydrolyzes peptide and ester bonds after phenylalanine residues [27, 28]. Pkhakadze et al. [29] synthesized polyurethanes containing a chain extender based on symmetric esters of phenylalanine and glycols. They found that chymotrypsin enhanced the degradation of their materials. Huang et al. [27] and Katsarava et al. [30] also found that chymotrypsin improved the degradation of L-phenylalanine containing polymers. Recently, we synthesized a family of segmented polyurethanes containing an L-phenylalanine based chain extender [15]. In these studies, the phenylalanine

Phenylalanine polyurethane degradation products

369

containing materials demonstrated what appeared to be preferential degradation in the presence of chymotrypsin [15]. In addition, these polyurethanes and their degradation products were not found to elicit a significant cytotoxic response, thereby demonstrating potential for use in biomedical applications [31]. Knowledge of the backbone cleavage sites will allow for the design of materials with controllable residence time in the body, based on the number of these sites present in the polyurethane. In the current work, two of the polyurethaneureas developed by Skarja and Woodhouse [32] were degraded in the presence of chymotrypsin. Degradation products were subsequently isolated using solid phase extraction and reversed-phase high pressure liquid chromatography (HPLC), and identified using tandem mass spectrometry (MS /MS). The identities of the polyurethane fragments were then used to determine the suspected cleavage sites in the polyurethane backbone.


Materials Polycaprolactone diol (molecular weight 530), polyethylene oxide (molecular weight 600) and 1,4 cyclohexane dimethanol were obtained from Aldrich, Milwaukee, WI, USA. The soft segment diols were placed in a vacuum oven at 60◦ C for 48 h to remove residual water prior to reaction. 2,6-diisocyanato methylcaproate (LDI) was supplied by Kyowa Hakko Kogyo Co. Ltd., Tokyo, Japan and was distilled under vacuum prior to use. Stannous 2-ethylhexanoate was obtained from Sigma, St. Louis, MO, USA. Polyurethane solvents N,N-dimethyl formamide (anhydrous grade DMF) and chloroform were obtained from Aldrich and ACP Chemicals Inc., Montreal, PQ, Canada, respectively. Synthesis of polyurethanes Polyurethanes were synthesized via the two-step reaction procedure of Skarja and Woodhouse [32], using LDI as the diisocyanate. One of two soft segments was used for each polyurethane: polycaprolactone diol of molecular weight 530 (PCL) or polyethylene oxide of molecular weight 600 (PEO). An L-phenylalanine based chain extender (PHE), developed and characterized by Skarja and Woodhouse [32], was also used. This PHE compound is comprised of a 1,4 cyclohexane dimethanol (CDM) group covalently linked on either side by an L-phenylalanine residue. CDM was used as a phenylalanine-free control chain extender. All reactions were carried out under nitrogen with a stoichiometry of 2 : 1 : 1 of LDI : soft segment : chain extender. The prepolymer reaction was catalyzed by stannous-2-ethylhexanoate and allowed to proceed for 2 h at 85◦ C. The chain extension reaction was allowed to continue for 18 h at 10 years) effects of metallic stents in humans are still unknown. Thus, other measures are required to resolve the restenosis issue. Two approaches to the management of residual stent-induced restenosis have emerged: stent polymer or ceramic coatings loaded with various pharmacologic agents [14– 18] and beta- or gamma-emitting radioisotopes, delivered via the stent itself or at stent implantation. Early clinical results suggest that pactitaxel, sirolimus, GPIIb/IIIa inhibitors, and other agents reduce short-term stent restenosis rates almost to zero (Table 1) [3, 13, 19]. These encouraging early results must be verified in longer-term trials. Low-dose radioisotope treatments with metal stents loaded with beta (90 Sr, 90 Y, 32 P) or gamma (192 Ir) emitters have also improved the

Bioresorbable polymeric stents

391

Table 1. Clinical trials and animal studies via coated stents and use of drugs with stenting Agents

TRIALS/studies

References

Paclitaxel, microtubular inhibitor Sirolimus, immunosuppressive agent Actinomycin D Metalloproteinase inhibitora

TAXUS I, ASPECT RAVEL, US trial Pig coronary arteries Porcine coronary arteries, porcine femoral arteries Porcine coronary model

[81, 82] [83, 84] [85] [86, 87]

Pig coronary arteries

[89, 90]

Porcine coronary model

[91]

TARGET, ADMIRAL CRUISE, ESPRIT Porcine coronary model Pig coronary arteries

[92, 93] [94– 96] [97] [98]

Pig coronary arteries NUGGET

[99] [100, 101]

Cytochalasin D, inhibitor of the actin microfilament formation Oligodeoxynucleotides to human transcription factor EGR-1 Antisense morpholino compound (AVI-4126) GPIIb/IIIa inhibitors: Abciximab Eptifibatide Methylprednisolone, anti-inflammatory Dextrose albumin microbubles containing c-Myc antisenses Cross-linked hyaluronan or chitosan Gold film coating a Opposing

[88]

responses were observed.

restenosis rate [19– 21]. Such stents have also been proposed for the local treatment of tumors and the prevention of excessive granulation tissue formation [22, 23]. Improvements in the restenosis rate notwithstanding, metal stents have other important limitations, including thrombogenicity, permanence, a limited potential for local drug delivery, and, for isotope-loaded stents, continuing radiation-induced damage [3, 12, 24]. Metal stent surfaces are moderately thrombogenic, requiring short-term antiplatelet or anticoagulant therapy. Metal stents are permanent implants. It is practically impossible to remove a metal stent, despite claims to the contrary for shape memory alloy stents that, in theory, can be narrowed in situ by application of heat or cold. Surgical revision of a stented vessel is also a practical impossibility, due to the difficulty of freeing the metal fiber impacted in the neointima. Coated metal stents have been introduced recently to provide controlled drug release, with very good short-term results (Table 1). Current practice is to use a bioresorbable phosphoryl choline polymer, or other polymer coating. However, the small reservoir capacity of the polymer film limits the amount of drug that can be loaded and eluted. Radiation-emitting stents have also been effective in reducing stent-induced restenosis; these, however, may induce radiation damage to the vessel wall. Finally, although not reported to date, there is the theoretical possibility of erosion of the arterial wall, due to compliance mismatch between the stent and arterial tissue. Given the difficulties with further development of metal stents, consideration of bioresorbable polymeric stents is attractive, as they may avoid the cited limitations of metal stents and offer other advantages as well.

392

R. C. Eberhart et al.

POLYMERIC STENTS

Several reports of resorbable and nonresorbable polymeric stents have recently appeared [12, 24– 29]. The rationale for the nondegradable stent is improved biocompatibility over the metal stent and convenient drug loading. Nonresorbable polymers being investigated for stent use include polyethylene terephthalate, polyurethane urea, and polydimethyl siloxane. The rationale for the bioresorbable stent is support of the arterial wall only during vessel healing, with gradual transfer of the mechanical load to the tissue as the stent mass and strength decrease over time, longer-term delivery of drug and/ or gene therapy to the vessel wall from an internal reservoir, and no need for a second surgery to remove the device. Bioresorbable polymers under investigation include aliphatic polyesters, polyorthoesters, and polyanhydrides. Recently, bioresorbable, linear, multiblock copolymers with shape memory capability have been introduced [30]. Controlled incremental heating of this thermoplastic material has been used to shrink sutures, making graded tissue approximations feasible in minimally invasive surgery applications. The same concept is also valuable for bioresorbable stent applications. Following balloon expansion, heat (approx. 5 ◦ C temperature change) applied to shape memory elements in the stent could reinforce designs that might not otherwise have sufficient recoil resistance. Poly-L-lactic acid (PLLA), poly-D,L-lactic acid (PDLA), poly ε-caprolactone (PCL) and polyglycolic acid (PGA), all aliphatic polyesters, are the most frequently used materials for bioresorbable stents [12, 24, 25]. PLLA and PDLA have a high tensile strength, permitting robust mechanical design, but requiring long degradation times. PGA and PCL have less strength, but faster degradation rates. Useful combinations of these materials (copolymers and blends) can be made to improve flexibility. These materials degrade principally by simple hydrolysis of the ester bond in the polymer backbone. Partial chain scission degrades the polymer to 10– 40 µm particles, capable of being phagocytosed and metabolized to carbon dioxide and water, which are of course fully resorbed. The degradation time is a function of the chemical structure of the polymer and its molecular weight. In typical formulations, PGA degrades over a time period of 6– 12 months, while PLLA degrades over months to years (Table 2). The long-term behavior of biodegradable polymers in blood vessels has not been well established. Van der Giessen et al. [31], testing strips of five different biodegradable polymers, PGA /PLA, PCL, polyhydroxybutyrate valerate, polyorthoester and polyethylene oxide/ polybutylene terephthalate, found extensive inflammatory responses within the coronary arterial wall. The observed tissue responses might be due the parent polymer compound, additives to the polymer, intermediate biodegradation products, the implant geometry, or combinations thereof. On the other hand, the authors noted that the implants were cleaned but not sterilized; therefore, bacterial or nonbacterial contamination might also have accounted for the inflammatory response. We have also observed a similar inflammatory response to sterilized PLLA stents implanted in the porcine femoral artery [32]. This


393

Table 2. Characteristics of typical bioresorbable polymersa Polymer

Melting point ( ◦ C)

Glass transition temperature ( ◦ C)

Modulus (Gpa)

Degradation time (months)

PGA PLLA PDLA PCL

225– 230 173– 178 Amorphous 58– 63

35– 40 60– 65 55– 60 −65 to −60

7.0 2.7 1.9 0.4

6– 12 >24 12– 16 >24

PGA = poly(glycolic acid); PLLA = poly(L -lactic acid); PDLA = poly(D, L -lactic acid); PCL = poly(ε-caprolactone). a Adapted from Ref. [102].

may have been due to the original polymer formulation, which was not intended for medical applications and contained an epoxide functionality of unknown quantity. More recent, purified formulations appear to induce a much less intense inflammatory response, as determined both by studies in the progress of PLLA fiber implantation into the rat aortic wall over 1– 4 weeks and by PLLA stent implantation in the pig femoral artery for 2 weeks. In the same vein, long-term study of polylactide copolymer, PLLA /PDLA (PLA96) stents in a rabbit abdominal aorta model found that the stents degraded with minimal tissue response within 24 months, with suitable encapsulation of polymer fragments in a thin neointima, leading the authors to suggest PLA96 as a promising stent core material [33]. Several early calls for expandable bioabsorbable stents have been published as alternatives for metallic stents [12, 24, 25]. These led to bioresorbable stents from Duke University [34], Tianjin/ Beijing University [35], Kyoto University [28], Igaki/ Tamai [36] and the University of Texas at Arlington/ Southwestern Medical School (UTA /SW) [32, 37]. The first biodegradable stent was developed and investigated at Duke University. This PLLA stent, based on a slotted polymer fiber design, was reported to withstand up to 1000 mmHg compression pressure; in vivo studies demonstrated minimal thrombosis and inflammatory responses, and moderate neointimal growth. The Tianjin/ Beijing stent, made of PDLA /PCL copolymer with an inner heparin layer, was deployed with a balloon catheter, employing heating and pressurization. This stent produced mild neointimal proliferation in swine carotid artery models at 2 months. The Kyoto University PGA coil stent exhibited thrombus deposition in canine implant studies, but no subacute closure. The Igaki/ Tamai stent, a bioresorbable PLLA zigzag coil thought to be derived from the Kyoto design, was studied in the first clinical report of a bioresorbable stent in the human coronary artery. This stent also required a combination of heating and pressurization for expansion. The preliminary (6-month) results suggest that this stent is safe and effective for human use. Long-term studies are anticipated.

394


THE MULTIPLE LOBE STENT

This PLLA stent is designed using a linear, continuous coil array principle, by which multiple furled lobes (four in the present design) convert to a single large lobe upon balloon expansion (Fig. 1). Melt-extruded PLLA fibers (drawn 6 : 1) with a diameter of 0.14 mm and an ultimate stress of 350 ± 40 MPa are woven continuously around a four-mandrel array (one central, three peripheral) into a fourlobe configuration. Three longitudinal fibers are interwoven and glued to the coil for mechanical support. After fabrication, a conventional angioplasty balloon catheter is inserted in the central lobe and the stent can be deployed at the target site. The structure of the fully expanded stent is that of a helical coil with three longitudinal reinforcing fibers. The initial and final diameters of stents are adjustable by various combinations in sizes of central and peripheral rod mandrels. Stents with furled diameters ranging from 1.6 to 2.4 mm were fully expanded by 3 atm pressure, to 2.3– 4.7 mm: the corresponding expansion ratios ranged from 1.4 to 1.9. Collapse pressure under radial compression was adequately high, ranging from 0.4 to 2.4 atm, depending on the fiber ply and other design parameters. Preliminary results from various in vitro and in vivo studies of this expandable bioresorbable stent suggested that the design principles and fabrication technique were sufficiently robust and versatile, thus meriting further investigation [37, 38].

Figure 1. Schematic diagram of the helical coil polymeric stent design (external version). The fiber is wound continuously over four mandrels to obtain the multiple lobes. The side lobes are flattened passively, or by use of a sheath during delivery. These lobes can also be wound inside the primary coil to reduce the profile during delivery. Both designs open readily with balloon expansion.


395

In 1- and 2-week implant studies in the porcine common femoral artery, stents did not migrate; however, the vessel lumen was markedly reduced at 2 weeks, due to a strong inflammatory response. Stents, like other implants, elicit a range of host responses, which interfere with the patency of the device [31, 39]. Various approaches have been investigated to improve the biocompatibility of these stents, including surface plasma treatment and drug incorporation. Pulsed RF plasma treatment with di(ethylene glycol) vinyl ether significantly reduced platelet adhesion in a 1 h porcine arteriovenous shunt model to less than 10% of untreated control values [37]. Curcumin (diferoyl methane), a non-steroidal anti-inflammatory drug, was melt-extruded with PLLA to generate curcumin-loaded PLLA fiber (C-PLLA). The curcumin was uniformly distributed within the fibers and a stable curcumin release rate for 36 days was observed. In vitro studies of mouse peritoneal phagocytes indicated significant reductions in the adhesion of these cells to C-PLLA compared with PLLA controls. These results suggested that C-PLLA has antiinflammatory properties, which may benefit the implants. Other non-steroidal antiinflammatory agents with sufficiently high melting points can be introduced into the polymer bulk in the same way. We have also investigated the bulk loading of aqueous drugs that cannot tolerate melt extrusion, using a wet spinning technique that permits the incorporation of large amounts of drug (up to 20 wt%) in the PLLA fiber. In addition, hollow PLLA fiber spinning processes that allow loading drugs, genetic vectors, or radioisotopes into PLLA accessories have been examined.

STENTS AS RESERVOIRS FOR LOCAL DRUG AND GENE THERAPY

Efforts have been directed towards developing stents coated with a biodegradable drug-impregnated polymer, capable of gradually releasing therapeutic agents into the vessel wall to reduce thrombosis and restenosis [2, 3, 39, 40]. The use of antithrombotic drugs such as heparin and hirudin is one strategy [41]. Other agents include prostacyclin analogy Iloprost [42], glycoprotein IIb/ IIIa receptor antibodies or inhibitors [43], and antiproliferative agents such as nitric oxide donors, corticosteroids and taxanes that inhibit neointima and local tumor proliferation [18, 39, 40]. Drugs or peptides contained within polymers can be in a nonchemically bonded configuration (physical entrapment) or chemically bonded to the polymer side-chains [26]. Stents coated with drug-eluting polymers such as hirudin, prostacyclin, and nitrosylated albumin were shown to reduce neointima formation [39, 42]. Decreased early thrombosis and neointima formations were also observed in stents loaded with glycoprotein IIb– IIIa inhibitors [44, 45] and with nitric oxide donors [46, 47]. Furthermore, intramural delivery of an antiproliferative agent, a specific tyrosine kinase inhibitor, using biodegradable stents has suppressed the restenotic changes of coronary arteries of treated pigs [28]. In addition to local drug delivery, stents can also serve as carriers for gene therapy delivery. Stents seeded with cells transfected with the desired gene, stents loaded

396


with recombinant adenovirus gene transfer vectors, and stents loaded with naked DNA impregnated in various matrices have been proposed [2, 29, 48– 50]. The introduction of an interested gene into the arterial wall can be achieved either by in vitro genetic manipulation of cells before their seeding onto stents or by direct in vivo gene transfer. Cell-based gene transfer using stents as platforms has been shown a major advantage in terms of site-specific gene expression. However, cell-based gene delivery has several limitations, including removal or injury of cells from the stent after balloon expansion and a significant time delay required for cell harvest, expansion, gene transfer, and subsequent selection prior to stent seeding. Yet seeding the stents with genetically engineered endothelial cells (ECs) to produce agents such as tPA has been shown to inhibit smooth muscle cell (SMC) proliferation [51]. A recent study has shown that a mesh stent coated with fibronectin is an excellent platform for adherent gene transduced SMCs [52]. Similar to the advantage of cell-based gene transfer, site-specific gene delivery, gene-stent therapy has been applied to reduce thrombosis and in-stent restenosis. Genes that encode enzymes of the prostacyclin synthetic pathway, nitric oxide synthase, the thrombin inhibitor, and thrombomodulin have been studied and demonstrated a significant reduction in thrombosis and restenosis in animal models [29, 50, 53– 55]. We successfully demonstrated local gene transfer and expression from PLLA stents impregnated with a recombinant adenovirus carrying a nuclear localizing βGal reporter gene into the carotid and renal arteries in the rabbit. Liver transfection was negligible in both cases, suggesting that gene delivery was local, not convected to remote sites to a significant degree [48]. In spite of promising results in animal models, to date no effective human gene therapy has been found to prevent restenosis [29, 50, 56]. In addition, potential side effects of the gene therapy approach such as subsequent malignant transformation due to oncogene activation with utilization of retroviral gene vectors and subsequent expression in other organs need to be further evaluated. In order to prevent potentially dangerous distal spread of viral vectors, a recent study has developed stent-based anti-viral antibody tethering of vectors onto the collagen coating surface of stents as a suitable platform for local gene delivery [57]. Another promising strategy for gene therapy delivery involves the introduction of antisense oligonucleotides into cells in order to inactivate the mRNA encoding proteins important in the restenotic process [58]. Uses of synthetic oligonucleotides to suppress proto-oncogenes including c-myb and c-mbc, proliferating cell nuclear antigen, and cell cycle-specific proteins cdc2 and cdk2 kinases were reduced protein expression and cell proliferation [12, 58].

NON-CORONARY USES OF STENTS

The range of stent applications has expanded with increases in experience and encouraging results in the treatment of vascular diseases. Stents have been used for the treatment of urethral obstruction from benign prostatic hyperplasia; for the treatment of tracheobronchial obstruction of benign or malignant origin; for the


397

treatment of benign and malignant strictures of the esophagus, the gastrointestinal (GI) tract, and the bile duct; and for the treatment (stents and stent-grafts) of arterial dissections, aneurysms and various neurovascular diseases.

STENTS IN UROLOGY

Stents have been used to prevent postoperative urine retention following thermal treatment of benign prostatic hyperplasia (BPH) by various means, including direct vision laser ablation of the prostate and transurethral microwave therapy. Several stent designs, including the Nissenkorn, Barnes, Finnish biodegradable selfreinforced polyglycolic acid (SR-PGA) spiral and Trestle, were shown to prevent obstruction of the prostatic urethra and restricture of the anterior urethra [27, 59, 60]. Biodegradable stents have been studied clinically in the treatment of BPH and are claimed to provide superior results to suprapubic catheters [61– 65]. Self-reinforced PLLA bioresorbable spiral stents are also undergoing evaluation for use in the anterior urethra, posterior urethra and upper urinary tract, to prevent urinary retention and repair of local ureteral trauma or defects [66, 67]. Surface modification of these biodegradable stents, by grafting with hydroxyethylmethacrylate or by incorporation of biologically active compounds, is claimed to be an efficient approach to improve biocompatibility and cell adhesion properties [68, 69].

STENTS FOR THE MANAGEMENT OF TRACHEOBRONCHIAL OBSTRUCTION

Tracheobronchial obstruction from either benign or malignant disease causes significant morbidity and mortality. Metal stents, developed originally for the vascular system, have been adapted for lesions involving the tracheobronchial tree. These include the Palmaz (Johnson & Johnson), Strecker (Boston Scientific), Gianturco-Z (William Cook Europe), Wallstent (Boston Scientific) and Ultraflex (Boston Scientific) stents [70]. These stents were successfully used to treat patients with inoperable bronchogenic cancer, esophageal tumors, primary tracheal tumors, and metastatic malignancy. Bioresorbable tracheal stents have been investigated in the setting of pediatric tracheal malacia, to solve the problem of limited tracheal growth in children with rigid external fixation and to avoid the necessity of a second procedure to remove the synthetic material [70– 72]. The general results from these studies suggest that stenting is a promising method to treat tracheal obstruction.

STENTS IN THE ESOPHAGUS AND GASTROINTESTINAL (GI) TRACT

Many malignant and benign strictures in the esophagus and GI tract can be treated by minimally invasive alternatives to surgery, including the use of stents. Most

398


commonly used in the esophagus and GI tract are the Wallstent (Boston Scientific), Ultraflex (Boston Scientific), Gianturco-Z (William Cook Europe), Esophacoil (Instent) and Flamingo stents (Boston Scientific). In general, these stents have been shown to be effective in relieving esophageal dysphagia [22, 73, 74]. This success has led to the employment of stents to manage lesions of the GI tract, including the stomach, pylorus, upper small intestine, duodenum, and colon [22, 73]. The use of bioresorbable material is currently being explored for the esophageal stent. First results in the placement of a PLLA stent (Instent) for the management of benign esophageal stricture suggest that a bioresorbable stent offers a new treatment modality [75]. In addition, bioresorbable stents were recently used in pancreaticojejunal anastomoses [76].

STENTS IN THE MANAGEMENT OF NEUROVASCULAR DISEASE

Stents and stent-grafts have been used for the management of arterial and venous sinus stenosis, arterial dissection, arterial aneurysm and arteriovenous fistulae [77, 78]. A number of case reports have been published describing the significant reduction in carotid stenosis with the use of stents in the treatment of carotid stenosis, recurrent carotid stenosis, vertebral artery stenosis, and venous sinus stenosis [78]. As cited in this review, Shawl et al. [79] reported a series of 124 stented vessels in which carotid stenosis was reduced from 86 ± 7% to 2 ± 2%; the major postprocedural stroke rate was 0.8% and the minor stroke rate was 2%. Three cases of basilar artery stenosis have been successfully treated with coronary stents at our institution [80]. Other unpublished reports from our institution have demonstrated the effectiveness of stents in bridging side-wall aneurysmal ostia, suggesting that stents are a promising means for the management of arterial dissection and pseudoaneurysm. Unfortunately, no large studies have yet been published, so the effectiveness of stents in this application remains to be determined.

CONCLUSION

Stents play an increasingly important role in percutaneous coronary interventions. Various metal stents have been shown to reduce the restenosis rate compared with angioplasty alone. This success has prompted the expansion of stent usage to peripheral arteries, the urethra, trachea, esophagus and GI tract. Stents do not eliminate the problem of coronary arterial restenosis and may contribute to it by inducing neointimal hyperplasia. Thus, isotope-loaded metal stents and polymer- and ceramic-coated metal stents, using the coating as a vehicle for local anti-inflammatory drug delivery, have been introduced, with promising results. Several bioresorbable stent designs are in development for temporary mechanical support combined, in some cases, with drug and/ or gene therapy delivery. Such temporary, bioresorbable stents that match the expandability and recoil resistance


399

of metal stents in the coronary arterial setting have been reported. These stents are theoretically superior for arterial wall healing, but face challenges in their application to smaller, more tortuous channels. Radioisotope-loaded metal or polymeric stents are also appealing for the local treatment of tumors and for the prevention of excessive granulation tissue formation in non-coronary settings. Stent design and development is currently a very active area of bioengineering practice. The expanding range of applications and new designs, materials, and surface treatments suggest that more effective, less invasive therapies may be anticipated in the near future. Acknowledgement This work was supported in part by USPHS grants RO1 HL /DE 53225 and F32HL010380.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

G. Belli, S. G. Ellis and E. J. Topol, Prog. Cardiovasc. Dis. 40, 159 (1997). H. Bult, Trends Pharmacol. Sci. 21, 274 (2000). R. Hoffmann and G. S. Mintz, Eur. Heart J. 21, 1739 (2000). A. J. Carter, D. Scott, D. Rahdert, L. Bailey, J. De Vries, K. Ayerdi, T. Turnlund, R. Jones, R. Virmani and T. A. Fischell, J Invasive Cardiol. 11, 127 (1999). J. M. Dumonceau and J. Deviere, Baillieres Best Pract. Res. Clin. 13, 109 (1999). Y. Ozaki, A. G. Violaris and P. W. Serruys, Prog. Cardiovasc. Dis. 39, 129(1996). A. G. Violaris, Y. Ozaki and P. W. Serruys, Int. J. Cardiac Imaging 13, 3 (1997). T. Nicholson, Hosp. Med. 60, 571 (1999). T. J. Cleveland and P. Gaines, Hosp. Med. 60, 630 (1999). C. D. J. Barras and K. A. Myers, Eur. J. Vasc. Endovasc. Surg. 19, 564 (2000). D. R. Ramsdale, Hosp. Med. 60, 624 (1999). J. F. Tanguay, J. P. Zidar, H. R. Phillips and R. S. Stack, Cardiol. Clin. 12, 699 (1994). A. H. Gershlick, Heart 86, 104 (2001). C. von Birgelen, M. Haude, J. Herrmann, C. Altmann, W. Klinkhart, D. Welge, H. Wieneke, D. Baumgart, S. Sack and R. Erbel, Catheter Cardiovasc. Interv. 47, 496 (1999). C. Stefanandis, E. Tsiamis, C. Vlachopoulos, K. Toutouzas, C. Stratos, I. Kallikazaros, M. Vavuranakis and P. Toutouzas, Cathet. Cardiovasc. Diagn. 40, 217 (1997). C. Herdeg, M. Oberhoff and K. R. Karsch, Semin. Interv. Cardiol. 3, 197 (1998). W. J. van der Giessen, H. M. van Beusekom, M. H. Eijgelshoven, M. A. Morel and P. W. Serruys, Semin. Interv. Cardiol. 3, 173 (1998). J. Gunn and D. Cumberland, Eur. Heart J. 20, 1693 (1999). A. W. Chan and D. J. Moliterno, Cleve. Clin. J. Med. 68, 796 (2001). J. Kotzerke, H. Hanke and M. Hoher, Eur. J. Nucl. Med. 27, 223 (2000). C. Hehrlein and W. Kubler, Semin. Interv. Cardiol. 2, 109 (1997). R. Lamber, Endoscopy 32, 322 (2000). S. Balter, Cathet. Cardiovasc. Diagn. 45, 292 (1998). M. Labinaz, J. P. Zidar, R. S. Stack and H. R. Phillips, J. Int. Cardiol. 8, 395 (1995). A. Colombo and E. Karvouni, Circulation 102, 371 (2000). T. Peng, P. Gibula, K. Yao and M. F. A. Goosen, Biomaterials 17, 685 (1996). M. Talja, T. Valimaa, T. Tammela, A. Petas and P. Tormala, J. Urol. 11, 391 (1997).

400


28. T. Yamawaki, H. Shimokawa, T. Kozai, K. Miyata, T. Higo, E. Tanaka, K. Egashira, T. Shiraishi, H. Tamai, K. Igaki and A. Takeshita J. Am. Coll. Cardiol. 32, 780 (1998). 29. C. Indolfi, C. Coppola, D. Torella, O. Arcucci and M. Chiariello, Cardiol. Rev. 7, 324 (1999). 30. A. Lendlein and R. Langer, Science 296, 1673 (2002). 31. W. J. van der Giessen, M. Lincoff, R. S. Schwartz, H. M. M. van Beusekom, P. W. Serruys, D. R. Holmes, S. G. Ellis and E. J. Topol, Circulation 94, 1690 (1996). 32. S. Su, Biomedical engineering, Doctoral dissertation, University of Texas at Arlington, Arlington (2000). 33. E. Hietala, U. Salminen, A. Stähls, T. Välimaa, P. Maasilta, P. Törmälä, M. S. Nieminen and A. L. J. Harjula, J. Vasc. Res. 38, 361 (2001). 34. C. M. Agrawal and H. G. Clark, Invest. Radiol. 27, 1020 (1992). 35. R. Gao, R. Shi, S. Qiao, L. Song and Y. Li, J. Am. Coll. Cardiol. 27, 85A (1996). 36. H. Tamai, K. Igaki, E. Kyo, K. Kosuga, A. Kawashima, S. Matsui, H. Komori, T. Tsuji, S. Motohara and H. Uehata, Circulation 102, 399 (2000). 37. S. Su, R. Y. Chao, C. L. Landau, K. D. Nelson, R. B. Timmons, R. S. Meidel and R. C. Eberhart, Ann. Biomed. Eng. (2003) (in press). 38. S. Su, C. L. Landau, R. Y. N. Chao, R. B. Timmons, R. S. Meidel, L. Tang and R. C. Eberhart, Circulation 104, II-507 (2001). 39. O. F. Bertrand, R. Sipehia, R. Mongrain, J. Rodés, J. Tardif, L. Bilodeau, G. Côté and M. G. Bourassa, J. Am. Coll. Cardiol. 32, 562 (1998). 40. D. Brieger and E. Topol, Cardiovasc. Res. 35, 405 (1997). 41. R. Herrmann, G. Schmidmaler, B. Märkl, A. Resch, I. Hähnel, A. Stemberger and E. Alt, Thromb. Haemost. 82, 51 (1999). 42. E. Alt, I. Haehnel, C. Beiharz, K. Prietzel, D. Preter, A. Stemberger, T. Fliedner, W. Erhardt and A. Schömig, Circulation 101, 1453 (2000). 43. A. Colombo and C. Briguori, Am. Heart J. 138, S153 (1999). 44. R. Aggarwal, D. Ireland, M. Azrin, D. De Bono and A. Gershlick, J. Am. Coll. Cardiol. 29, 353A (1997). 45. R. Aggarwal, D. Ireland, M. Azrin, M. Ezekowitz, D. De Bono and A. Gershlick, Circulation 94, 3311 (1996). 46. J. Folts, N. Maalej, J. Keaney and J. Loscalzo, Circulation 92, I-670 (1995). 47. J. Folts, N. Maalej, J. Keaney and J. Loscalzo, J. Am. Coll. Cardiol. 27, 86A (1996). 48. Y. Ye, C. Landau, J. E. Willard, G. Rajasubramanian, A. Moskowitz, S. Aziz, R. S. Meidell and R. C. Eberhart, Ann. Biomed. Eng. 26, 398 (1998). 49. B. D. Klugherz, P. L. Jones, X. Cui, W. Chen, N. F. Meneveau, S. DeFelice, J. Connolly, R. L. Wilensky and R. J. Levy, Nature Biotechnol. 18, 1181 (2000). 50. M. D. Feldman, B. Sun, B. J. Koci, C. C. Wu, J. R. Kneller, H. S. Borovetz, S. Watkins, A. Nadeem, L. E. Weiss, M. L. Reed, A. J. Smith and W. D. Rosenblum, J. Long Term Eff. Med. Implants 10, 47 (2000). 51. D. A. Dichek, R. F. Neville, J. A. Zwiebel, S. M. Freeman, M. B. Leon and W. F. Anderson, Circulation 80, 1347 (1989). 52. C. J. Panetta, K. Miyauchi, D. Berry, R. D. Simari, D. R. Holmes, R. S. Schwartz and N. M. Caplice, Hum. Gene Ther. 13, 433 (2002). 53. P. Vermeersch, Z. Nong, E. Stabile, O. Varenne, H. Gillijns, M. Pellens, N. Van Pelt, M. Hoylaerts, I. De Scheerder, D. Collen and S. Janssens, Arterioscler. Thromb. Vasc. Biol. 21, 1604 (2001). 54. L. M. Akyurek, S. Nallamshetty, K. Aoki, H. San, Z. Y. Yang, G. J. Nabel and E. G. Nabel, Mol. Ther. 3, 779 (2001). 55. O. Varenne and P. Sinnaeve, Curr. Interv. Cardiol. Rep. 2, 309 (2000). 56. L. Gruberg, R. Waksman, L. F. Satler, A. D. Pichard and K. M. Kent, Exp. Opin. Invest. Drugs 9, 2555 (2000).


401

57. B. D. Klugherz, C. Song, S. DeFelice, X. Cui, Z. Lu, J. Connolly, J. T. Hinson, R. L. Wilensky and R. J. Levy, Hum. Gene Ther. 13, 443 (2002). 58. N. Kipshidze, J. Moses, L. R. Shankar and M. Leon, Curr. Opin. Mol. Ther. 3, 265 (2001). 59. R. Kapoor, E. N. Liatsikos and G. Badlani, Curr. Opin. Urol. 10, 19 (2000). 60. B. A. Kletscher and J. E. Oesterling, Urol. Clin. 22, 423 (1995). 61. J. J. M. C. H. de la Rosette, H. P. Beerlage and F. M. J. Debruyne, J. Endourol. 11, 467 (1997). 62. E. Kemppainen, M. Talija, M. Riihelä, T. Pohjonen, P. Törmälä and O. Alfthan, Urol. Res. 21, 235 (1993). 63. A. Pétas, M. Talja, T. Tammela, K. Taari, K. Lehtoranta, T. Välimaa and P. Törmälä, J. Urol. 157, 173 (1997). 64. A. Pétas, T. Isotalo, M. Talja, T. L. J. Tammela, T. Välimaa and P. Törmälä, Scand. J. Urol. Nephrol. 34, 262 (2000). 65. A. Pétas, P. Kärkkäinen, M. Talja, K. Taari, M. Laato, T. Välimaa and P. Törmälä, Br. J. Urol. 80, 903 (1997). 66. J. Lumiaho, A. Heino, T. Pietiläinen, M. Ala-Opas, M. Talja, T. Välimaa and P. Törmälä, J. Urol. 164, 1360 (2000). 67. J. Lumiaho, A. Heino, V. Tunninen, M. Ala-Opas, M. Talja, T. Välimaa and P. Törmälä, J. Endourol. 13, 107 (1999). 68. A. Brauers, P. K. Jung, H. Thissen, O. Pfannschmidt, W. Michaeli, H. Hoecker and G. Jakse, Tech. Urol. 4, 214 (1998). 69. A. Brauers, H. Thissen, O. Pfannschmidt, H. Bienert, A. Foerster, D. Klee, W. Michaeli, H. Höcker and G. Jakse, J. Endourol. 11, 399 (1997). 70. A. L. Rafanan and A. C. Mehta, Int. Chest Radiol. 38, 395 (2000). 71. T. C. Robey, P. M. Eiselt, H. S. Murphy, D. J. Mooney and R. A. Weatherly, Laryngoscope 110, 1936 (2000). 72. H. Lochbihler, J. Hoelzi and H. Dietz, J. Pediatr. Surg. 32, 717 (1997). 73. R. Morgan and A. Adam, J. Vasc. Interv. Radiol. 12, 283 (2001). 74. T. Nicholson, Hosp. Med. 61, 97 (2000). 75. S. W. Fry and D. E. Fleischer, Gastrointest. Endosc. 45, 179 (1997). 76. M. Parviainen, J. Sand, A. Harmoinen, H. Kainulainen, T. Välimaa, P. Törmälä and I. Nordback, Pancreas 21, 14 (2000). 77. H. A. Gray and A. W. Crane, Am. J. Roentgenol. 175, 289 (2000). 78. M. B. Horowitz and P. D. Purdy, Neurosurgery 46, 1335 (2000). 79. F. Shawl, W. Kadro, M. J. Domanski, F. L. Lapetina, A. A. Iqbal, K. G. Dougherty, D. D. Weisher, J. F. Marquez and S. T. Shahab, J. Am. Coll. Cardiol. 35, 1721 (2000). 80. M. B. Horowitz, G. L. Pride, D. F. Graybeal and P. D. Purdy, Neurosurgery 45, 925 (1999). 81. E. Grube, S. M. Silber and K. E. Hauptmann, Circulation 104, II-463 (2001). 82. S. Park, W. H. Shim, D. S. Ho, A. E. Raizner, S. W. Park, J. J. Kim, M. K. Hong, C. W. Lee, S. Y. Cho, Y. S. Jang, D. H. Choi, C. P. Lau, R. Lam and Y. Wang, Circulation 104, II-464 (2001). 83. J. W. Moses, M. B. Leon, J. J. Popma and R. E. Kuntz, Circulation 104, II-464 (2001). 84. J. E. Sousa, M. Morice, P. W. Serruys, J. Fajadet, M. Perin, E. B. Hayashi, A. Colombo, G. Schuler, P. Barragan and C. Bode, Circulation 104, II-463 (2001). 85. K. A. Robinson, N. A. Chronos, J. Royal, L. Suh, G. D. Cipolla, R. Virmani and R. S. Stack, Circulation 104, II-506 (2001). 86. F. Gobeil, M. Laflamme, M. Bouchard, Campbell E., M. Wood, P. St-Jacques, J. Vincent and G. Leclerc, Circulation 104, II-388 (2001). 87. H. M. Van Beusekom, M. J. Post, B. J. De Smet, M. G. Put, M. S. Harteveld and W. J. Van der Giessen, Circulation 104, II-506 (2001). 88. K. J. Salu, H. Yanming, J. M. Bosmans, L. Xiaoshun, I. K. Scheerder and C. J. Vrints, Circulation 104, II-506 (2001).

402


89. H. C. Lowe, R. Fahmy, C. N. Chesterman and L. M. Khachigian, Circulation 104, II-265 (2001). 90. H. C. Lowe, R. G. Fahmy, M. M. Kavurma, A. Baker, C. N. Chesterman and L. M. Khachigian, Circ. Res. 89, 670 (2001). 91. H. Kim, H. A. Yazdi, R. Seaborn, P. Iversen, G. S. Roubin, S. S. Lyer, M. B. Leon, J. W. Moses and N. Kipshidze, Circulation 104, II-545 (2001). 92. T. A. Swierkosz, S. Kapoor, D. C. Tardiff, J. W. Hirshfeld, B. D. Klugherz, D. M. Kolansky, K. Magness, N. Valettas, R. L. Wilensky and H. C. Herrmann, Circulation 104, II-385 (2001). 93. G. Montalescot, P. Barragan, O. Wittemberg, P. Pinton, S. Elhadad, X. Deboisgelin, A. Akesbi, B. Jouve, M. Amor and F. Funck, Circulation 104, II-386 (2001). 94. G. Amoroso, A. J. van Boven, D. J. van Veldhuisen, R. A. Tio, C. P. Balje-Volkers, A. S. Petronio and W. van Oeveren, J. Cardiovasc. Pharmacol. 38, 633 (2001). 95. D. L. Bhatt and A. M. Lincoff, Circulation 104, II-384 (2001). 96. A. G. Rebeiz, K. S. Pieper, J. C. O’Shea, I. C. Gilchrist, B. Chandler, J. Slater, J. B. Muhlestein, T. J. Loenz, D. Joseph, M. M. Kitt and J. E. Tcheng, Circulation 104, II-386 (2001). 97. Y. Huang, L. Wang, I. Vermeire, E. Verbeken, E. Schacht and I. D. Scheerder, Circulation 104, II-665 (2001). 98. T. R. Porter, P. Iversen, F. Xie, D. Kricsfeld, S. J. Radio, J. Lof and S. Flynn, Circulation 104, II-589 (2001). 99. R. Rohde, B. Heublein, S. Ohse, E. Evagorou, K. Pethig and A. Haverich, Circulation 104, II-666 (2001). 100. G. B. Danzi, C. Capuano, M. Sesana, A. D. Blasi and D. Antoniucci, Circulation 104, II-625 (2001). 101. S. Silber, N. Reifart, M. Morice, J. V. Dahl, E. Benit, K. Hauptmann, J. E. Sousa, J. G. Webb, U. Kaul and C. Chan, Circulation 104, II-623 (2001). 102. J. C. Middleton and A. J. Tipton, Med. Plast. Biomater 30 (March) (1998).

Tissue-culture surfaces with mixtures of aminated and fluorinated functional groups. Part 1. Synthesis and characterization JAMES R. BAIN 1,2,∗ and ALLAN S. HOFFMAN 3 1 Sarah

W. Stedman Center for Nutritional Studies, Duke University Medical Center, Durham, NC 27710, USA 2 Department of Pharmacology and Cancer Biology, Duke University Medical Center, Mail Stop DUMC 3813, LSRC building, room C348, Durham, NC 27710, USA 3 Department of Bioengineering, Box 352255, University of Washington, Seattle, WA 98195, USA Received 10 January 2002; accepted 4 December 2002 Abstract—Surface chemistry of culture dishes can have profound effects on the phenotype of cultured cells. In the present study, chemisorption from aqueous, binary mixtures of organosilanes onto borosilicate glass created surfaces bearing diamine groups (N2), trifluoropropyl groups (F3) and mixtures of the two. Composition of N2-F3 surfaces was controlled by the ratio of monomers in the silanization bath, as confirmed by electron spectroscopy for chemical analysis and by conjugation of surface amines with fluorescein-5-isothiocyanate. Atomic-force microscopy revealed that silanized surfaces are patchy, though their root-mean-square roughnesses do not differ significantly from that of smooth glass (0.3 nm). Surfaces richest in diamine residues were the most hydrophilic, with advancing water-contact angles 90◦ . The accompanying paper (the next article in this issue) describes the effects of these surface chemistries on the phenotype of transgenic insulinoma cells in vitro. We conclude that chemisorption from the N2-F3 system provides a simple, one-pot method for tailoring the chemistry of glass culture surfaces. Key words: Silanes; surface modification; tissue culture; electron spectroscopy for chemical analysis; atomic-force microscopy; wettability.

INTRODUCTION

Chemistry of culture surfaces can affect the behavior and phenotype of cultured mammalian cells [1– 4]. In the present study, we sought to create novel culture substrates on borosilicate glass by chemisorption of organosilanes. Resulting ∗ To


404


surfaces bear diamine groups, trifluoropropyl groups and mixtures of the two. We chose silanized borosilicate glass as a model system because we were able to finetune the surface chemistry by competitive chemisorption from binary, aqueous solutions of silane monomers. Silanization of glass surfaces results from the hydrolysis and subsequent condensation of chloro- or alkoxysilanes with surface silanols (SiOH, Fig. 1). Lateral condensation polymerization of silanes on the surface creates thin deposits of polysiloxanes [5, 6]. Numerous investigators have employed silanes and the related silazanes to create chemically defined surfaces for the study of cell-material interactions [4, 7– 34]. Swalec [20] created cell-culture surfaces of mixed character by competitive, vapor-phase deposition of methyl- and chloropropyl silanes. We hoped that competitive chemisorption from aqueous solutions of mixed silanes would offer a practical, controllable and inexpensive alternative to vapor-phase deposition. Competitive chemisorption of organosilanes from water and other solvents has previously been used to create surfaces of mixed character [5, 6, 35– 37], but to our knowledge such surfaces have not yet been evaluated in cell culture. Portions of this report are reproduced from a preliminary presentation [38].


Selection of silane monomers for competitive chemisorption A priori, we expected that amine-rich surfaces would favor cell growth and that fluorinated surfaces would be less suitable for anchorage-dependent mammalian cells, so we selected silane monomers bearing these functional groups. We sought a competitive pair of monomers that are stable for a reasonable time in a common

Figure 1. Silanization of etched borosilicate glass with monomers N2 and F3. A monolayer is shown for simplicity. These and most other silanized surfaces are in fact patchy (see text).

Aminofluorosiloxane culture surfaces

405

aqueous solvent and in which the pendant chains of both moieties are grossly similar in length. Diamine monomer N2: 3-(2-aminoethylamino) propyltrimethoxysilane Many anchorage-dependent vertebrate cells spread and grow well on surfaces rich in amines, amides and other nitrogen-containing groups, including surfaces rendered nitrogenous by organosilanization [8– 10, 13, 16, 17, 21– 25, 28, 32– 34, 39, 40]. A principle focus of the present investigation was to create novel surfaces for culture of insulinoma cells [1]. Limited data indicate that some insulin-secreting endocrine cell lines will grow and function on amide- and amine-rich surfaces [41– 44]. We chose silane N2, with its primary and secondary amine groups (Fig. 1), because it has been used in previous studies to create surfaces acceptable to a wide variety of mammalian cells, including such fastidious cells as neurons and endothelial cells [10– 14, 18, 19, 26, 27, 29– 31]. For the promotion of cell attachment, spreading and growth, N2’s diamine structure might be preferable to monoamines and other common aminosilane structures [29]. The residue remaining after aminosilanization with N2 is shown in Fig. 1. Several hypotheses have been put forth to account for the ‘cell-friendly’ nature of N2 residues (i.e. 3-(2-aminoethylamino) propylpendant groups) and other solid-phase amines, including direct association of cellsurface proteoglycans with synthetic cations by electrostatic or hydrogen-bonding interactions [12, 19, 21, 24, 26, 29, 31, 39, 45– 48] and the preferential adsorption of soluble cell-attachment proteins, such as fibronectin and vitronectin, to nitrogenrich surfaces [13, 30, 31, 39, 40, 49]. Trifluorinated monomer F3: (3,3,3-trifluoropropyl) trimethoxysilane We chose F3 because we expected fluorinated surfaces to be relatively poor growth substrata for our insulinoma cells [1]. In general, fluorinated materials have large advancing contact angles with water (θA 85◦ ), low critical surface tensions (Zisman’s γc 25 dyne/ cm) and do not support vigorous cell growth [50– 55]. Poor cell growth seen on many fluorinated materials might be due to low binding strengths or unfavorable binding conformations adopted by physisorbed celladhesion proteins [56]. We were unable to find evidence that surfaces modified with silane F3 have previously been evaluated in cell culture. The C3 residue resulting from its use is shown in Fig. 1. Silanized surfaces bearing longer, partially fluorinated, aliphatic side chains have been shown to inhibit growth of mammalian cells. Side-chain lengths in past studies have included C8 [19], C10 [34] and C18 [23]. Silanes with such long fluoroalkyl side chains were not considered for use in the present study, because they are insoluble in the aqueous solutions we wished to use for competitive deposition with the diamine co-monomer, N2. In selecting the hydrophobic member of the monomer pair, we chose the fluoroalkyl silane F3 over nonfluorinated alkylsilanes because it bears an element (F) not present in N2. Extreme electronegativity of fluorine makes it especially easy to detect at low degrees

406


of incorporation; F3’s fluorines provided a strongly reporting atomic marker for surface characterization by electron spectroscopy for chemical analysis, infra [57]. Glass selection, cleaning and etching Conventional borosilicate cover glasses for microscopy were purchased from Carolina Biological Supply (Burlington, NC, USA). According to the manufacturer (Glaswarenfabrik Karl Hecht, Sondheim / Rhön, Germany), the bulk formulation, in mass percentages, is 65% SiO2 , 12% (Na2 O + K2 O), 7.5% B2 O3 , 6.5% ZnO, 5.5% TiO2 , 3% Al2 O3 and 0.5% Sb2 O3 . This formulation is expressed in terms of atomic percentages in Table 1. Squares (12 × 12 mm) provided a constant circumference for Wilhelmy-plate, contact-angle goniometry (infra). Circles (15 mm diameter) fit snugly in 24-well plates for cell culture. Before silanization, glass samples were cleaned and etched while mounted in polytetrafluoroethylene racks, as previously described [15]. Silanes, silanization and annealing (3,3,3-Trifluoropropyl) trimethoxysilane (F3) and 3-(2-aminoethylamino) propyltrimethoxysilane (N2) were used as received from PCR (Gainesville, FL, USA). Organic solvents were from J.T. Baker (Phillipsburg, NJ, USA) and were ‘analyzedreagent’ grade or better. In all cases, silanization mixtures had a total silane concen-

Acetone-rinsed glass

Detergent-washed glass

Alkaline-etched glass

Silanized with 0/3 F3 (100% N2)

Silanized with 3/3 F3 (100% F3)

O Si Na B K Al Zn Ti Sb S N F C

Glassmaker’s nominal formulation*

Table 1. ESCA characterization of borosilicate glass surfaces after cleaning, etching and two different organosilanizations

61.79 22.60 4.04** 4.50 2.66** 1.23 1.67 1.44 0.07 0.00 0.00 0.00 0.00

53.2 22.8 4.5 3.0 2.4 1.5 1.5 0.6 0.0 0.0 0.0 0.0 10.5

49.4 21.8 3.0 3.4 1.3 2.0 0.6 0.6 0.0 0.8 1.4 0.0 15.8

57.0 24.4 2.6 2.5 1.5 1.6 1.0 0.6 0.0 0.0 0.0 0.0 8.9

48.8 24.3 2.3 2.8 0.6 2.1 0.6 0.4 0.0 0.0 2.5 0.0 15.6

49.4 23.9 1.9 1.9 1.0 1.9 0.6 0.6 0.0 0.0 0.4 8.1 10.4

Values are given as atomic percentages. ∗ Theoretical values, ∗∗ approximate values.


407

tration of 0.1 M in 50 : 50 ddH2 O /isopropanol, apparent pH 4.75. Extensive experimentation showed that departures from this solvent condition led to instability of mixtures of N2 and F3, with rapid polymerization or precipitation of the monomers, particularly those richest in the F3 monomer. Solutions were used immediately after adjustment of pH. Silanization proceeded 2 h at approx. 20◦ C, with stirring, followed by extensive rinsing in 50 : 50 ddH2 O /isopropanol, apparent pH 4.75, followed by a final rinse in 70% (v/ v) ethanol in ddH2 O to disinfect samples in preparation for cell culture. Annealing by heat is necessary to give stable deposits of siloxanes on glass surfaces, but can destroy amines if carried to excess [6, 58, 59]. To assess the effects of annealing regimens on amine reactivity, we conjugated fluorescein-5isothiocyanate (FITC, Aldrich, Milwaukee, WI, USA) to surfaces, using 1 mg/ ml FITC in N,N-dimethylformamide (DMF), with 0.712 µl/ ml triethylamine as a base catalyst (Fig. 2). Thiourea conjugation proceeded for 2 h at approx. 20◦ C, unstirred. After a 30-min rinse in DMF, samples were rinsed thrice in ddH2 O and dried in vacuo at approx. 20◦ C, about 0.5 mmHg. Bound FITC residues were then stripped in 0.1 M NaOH for 2 h, unstirred, at approx. 20◦ C. Fluorescence of the acid-neutralized stripping solution was determined in a Perkin-Elmer LS5B luminescence spectrometer, using a xenon source (excitation 495 nm, emission 519 nm; Perkin-Elmer, Norwalk, CT, USA).

Figure 2. Reactivity of aminosiloxane (N2) residues with fluorescein-5-isothiocyanate.

408


Electron spectroscopy for chemical analysis (ESCA) Analyses were performed with an X-Probe ESCA instrument (Surface Science Instruments, Mountain View, CA, USA). This instrument permits analysis of the outermost approx. 50 Å of a sample in an elliptical area whose short axis is aobut 600 µm (D. Leach-Scampavia, University Washington, Seattle, WA, USA, personal communication). An aluminum Kα1,2 monochromatized X-ray source was used to stimulate photoemission. Energy of the emitted electrons was measured in a hemispherical energy analyzer at pass energies ranging from 25 to 150 eV. Spectral data were collected with the analyzer at 55◦ with respect to the surface normal of the sample. SSI data-analysis software was used to calculate elemental compositions from the peak areas and to peak-fit the high-resolution spectra. An electron flood gun set at 5 eV was used to minimize surface charging. Binding energy was referenced by setting the CHx peak maximum in the C1s spectrum to 285.0 eV. Typical pressures in the analysis chamber during spectral acquisition were 10−9 torr. Contact-angle goniometry Advancing (θA ) and receding contact angles (θR ) with ddH2 O (pH approx. 7.4), were determined by the Wilhelmy-plate method [60], using a Cahn DCA 312 Contact Angle Analyzer (Cahn Instruments, Cerritos, CA, USA). Purity of ddH2 O was monitored by daily measurements of the surface tension of water (γLV ddH2 O), using freshly-flamed cover glasses, which were used as soon as they cooled to room temperature. Crosshead velocity was 10 mm /min. Three independent samples were assayed, and θA and θR were measured during the first three immersion-emergence cycles. Data presented below are the means of these nine values. Throughout this report, error bars show ±1 standard deviation. Atomic-force microscopy (AFM) Annealed, silanized surfaces were imaged in air in tapping mode with the NanoScope Dimension 3100 Scanning Probe Microscope (Digital Instruments, Santa Barbara, CA, USA). Root-mean-square (RMS) roughness, the standard deviation of height measurements (z), was calculated over a square sample area 3 µm on a side (512 × 512 pixels). Heights, widths and lengths of randomly selected surface features were measured in the same area. Images were recorded with a false-color scheme showing a 5-nm full z-axis in all cases. Statistical analysis Pair-wise comparisons of data were evaluated post hoc with the Tukey– Kramer honestly significantly different (HSD) test [61], using version 3.0 of the JMP software (SAS Institute, Cary, NC, USA). Differences were considered significant at P < 0.05.


409


Surface composition by electron spectroscopy for chemical analysis Silicon and oxygen dominate the surface of borosilicate cover glasses cleaned with acetone (Table 1). Theoretical values for composition are based on formulation data from the glassmaker. Presence of carbon is the most striking departure of the observed composition from theoretical values. This optical glass does not contain carbonates; carbon is probably introduced by physisorption of hydrocarbons and other organic contaminants onto the surface from the atmosphere. This is perhaps inevitable on such high-energy surfaces as clean metals and glasses [35, 62– 64]. Though the surface can be cleaned by flaming, as was done with the standards used in the Wilhelmy-plate experiments (supra), it rapidly fouls upon standing in the laboratory air (in approx. 20 min). Carbon is further enriched in the alkalinedetergent-washed sample (Table 1), possibly by deposition of surfactants from the Micro cleaning solution [15]. Detergent residue would also account for the appearance of small amounts of sulfur and nitrogen at this stage (Table 1). Nitrogen, sulfur and nearly half the carbon are subsequently removed by etching in 0.5 M NaOH on the day of silanization (Table 1). Consistent with published reports that describe etching of glass surfaces at extremes of pH, the two alkaline treatments (Micro detergent, followed by the 0.5 M NaOH etchant) depleted the surface of most minor elements (Na, B, K, Zn and Ti) and the surface was concomitantly enriched with silicon and oxygen [62, 63]. Antimony (Sb) was not detected by ESCA (Table 1), probably because it is present at less than 0.1 atomic percent and because of strong interference from the 2s and 2p lines of silicon. Silanization with pure, 0.1-M solutions of the diaminosilane (N2) gave 2.5 atomic percent nitrogen, while the pure, 0.1-M trifluoropropyl silane (F3) gave 8.1 atomic percent fluorine (Table 1). The trace amount of nitrogen present in the surface silanized with a pure solution of F3 might represent residual soap. Normalized data on atomic ratios of nitrogen and fluorine show good agreement between observed surface compositions and those predicted from stoichiometry of the reactive silane monomers (Fig. 3). A slight excess of nitrogen was observed in every case except the pure aminosiloxane deposits. Reactivity of amine groups with fluorescein-5-isothiocyanate Organosiloxane deposits condensed on glass are unstable in water, a property that is often overlooked by biologists who culture cells on silanized surfaces. Hydrolysis of organosiloxanes can be greatly slowed if surfaces are heat-annealed prior to use [6, 36, 64]. An interactive series of short studies on annealing, contact-angle goniometry and reactivity of FITC toward surface amine groups (Fig. 2) demonstrated that an air-oven cure of 2 h at 100◦ C offered an acceptable compromise between annealing and loss of amine reactivity, producing silanized surfaces resistant to hydrolysis, while at the same time preserving approx. 85% of the reactive amines of the uncured surface (data not shown). Gradual loss of

410


amine reactivity upon heating aminosiloxanes in air might be due to reactions with CO2 , conversion of amines to amides, or the formation of internal Zwitterions when amines ‘bite back’ with free silanol anions or other anions on the surface (e.g. borates, metal oxides) [65, 66]. Figure 4 shows a monotonic decrease of binding and subsequent alkaline hydrolysis of FITC residues with increasing mole fraction

Figure 3. Normalized ESCA data on atomic percentages of nitrogen (N) and fluorine (F) in mixed aminofluorosiloxane surfaces, expressed as the ratio F/(F + N).

Figure 4. Decrease in binding and subsequent lysis of fluorescein-5-isothiocyanate residues in silanized surfaces with increasing mole fractions of the trifluoropropyl residues (r 2 = 0.977).


411

of silane monomer F3 in surfaces of mixed composition. These surfaces had been annealed at 100◦ C for 2 h prior to FITC conjugation. Contact angles All pure and mixed surfaces of N2 and F3 showed advancing contact angles, θA , greater than 80◦ and receding contact angles, θR , greater than 55◦ (Fig. 5). The relatively high contact angles observed on many aminosilanized surfaces have been ascribed to the aforementioned formation of internal Zwitterions and to concealment of high-energy head groups among alkane and siloxane groups [67]. Heat-annealed surfaces were stable during repeated wetting, showing only a slight decrease in θA and θR as the polysiloxanes hydrated during the first three immersion-emergence cycles. Surprisingly, both θA and θR reached their maxima at a mole fraction of 2/3 F3 (Fig. 5) and not at 100% F3, as expected. Surface roughness, hydrogen F2 CF· · ·HNH ), or lateral heterogeneity in surface chemistry bonding (e.g. might account for the peak in θA and θR at intermediate mixtures of N2 and F3. Nanometer-scale roughness, as studied with the AFM, did not vary significantly among the glass and silanized surfaces (P > 0.05, infra). In view of the insular

Figure 5. Advancing and receding water-contact angles (θA and θR , respectively) for silanized surfaces. Data enclosed in dashed ellipses do not differ from one another (Tukey– Kramer HSD test, P 0.05).

412


Figure 6. Contact-angle hysteresis (θH ) for silanized surfaces. Data enclosed in dashed ellipses do not differ from one another (Tukey– Kramer HSD test, P 0.05).

nature of the silanized surfaces (infra), patchiness in surface chemistry [51] seems to be the most likely explanation for the peak in θA and θR at 2/3 silane F3. Chemical heterogeneity of surfaces is often invoked to explain contact-angle hysteresis, θH , defined as θA minus θR [53]. Though we had expected θH to reach its maximum in mixed surfaces of N2 and F3, θH was in fact greatest in surfaces of pure N2 (26.6◦ ) and declined gradually with increasing mole fraction of F3, reaching a minimum of 17.9◦ in surfaces of pure F3 (Fig. 6). Declining θH with increasing fluorine content was significant at P < 0.05 (Tukey– Kramer HSD test; Fig. 6). This might have been due to greater hydration of the diamino functional groups on surfaces richer in N2 residues. Atomic-force microscopy (AFM) As expected, solvent-cleaned and alkaline-etched borosilicate glass surfaces were extremely smooth. Four randomly selected spots were evaluated on both materials and every area sampled had a root-mean-square (RMS) roughness of less than 0.5 nm (Figs 7 and 8). This is consistent with previous reports of roughness values 0.5 nm for the smoothest known surfaces of glass, muscovite mica, silica and silicon [67– 70]. On all seven silanized surfaces studied by AFM, the RMS roughness averaged more than 0.5 nm (Figs 7 and 8). Four of the seven silanized surfaces had average RMS roughness values greater than 0.9 nm (viz., the pure diaminosilanized surface, 0/3 F3, along with three mixed surfaces, 1/2 F3, 2/3 F3 and 5/6 F3; Fig. 8). These same four surfaces exhibited the most striking decoration with ellipsoid, elevated features in the false-color portraits (Fig. 7). Such features, which Brunner et al. [71] called ‘submonolayer islands’, probably represent domains of polymerized silanes on a background of flat, unmodified glass. This is consistent with the strong reporting by minor constituents of the underlying borosilicate glass in all samples evaluated by ESCA (e.g. Na, B, K, Al, Zn and Ti; Table 1). Lateral


413

Figure 7. False-color portraits of glass, etched-glass and silanized-glass surfaces under atomic-force microscopy. Each image represents a square sample area of 3 µm × 3 µm. This figure is published in colour on http://www.ingenta.com.

Figure 8. Root-mean-square (RMS) roughness of silanized and glass surfaces. Note the logarithmic scaling. Average RMS roughness values for all materials are statistically indistinguishable (P 0.05, Tukey– Kramer HSD test).

414


variability in surface roughness was most pronounced in the 0/3 F3 and 1/2 F3 surfaces (Fig. 8). One randomly selected area on the 1/2 F3 surface had an RMS roughness in excess of 7 nm (Fig. 8). Roughness and its variability (Figs 7 and 8) showed no clear relationship with contact angle and its hysteresis (Figs 5 and 6). Section analysis was performed to profile individual, randomly selected surface asperities on all materials (data not shown). Ellipsoid islands on silanized surfaces generally had lengths of 140 to 240 nm and aspect ratios (greatest length over narrowest width) of 1.3 to 1.9. Heights generally ranged from 1.5 to 5 nm. Previous investigators have noted asperities of similar size emerging from flat backgrounds of silanized surfaces. Numerous studies suggest that the size and shape of features created during silanization of flat surfaces, along with numerous other physicochemical characteristics of such surfaces, are influenced by experimental variables, including cleaning, etching and hydration prior to surface modification, as well as silanization conditions and any subsequent annealing by heat [7, 37, 67– 75]. In the most heavily studied silanized surface, residues of octadecyltrichlorosilane or OTS, with their long (C18 ), pendant alkyl chains, are known to form orderly structures, including densely packed, self-assembled monolayers, under some conditions. Independent observers have measured the height of condensed phases of OTS and similar C18 silane residues as 2.2 to 2.5 nm or so, which corresponds well to the theoretical extended-chain length of the octadecyl residue [7, 69, 71– 77]. In contrast to these orderly OTS systems, we speculate that the short, chemically disparate, trialkoxy silane monomers used in the present study (viz., N2, F3) form more random or dispersed networks, without discernible nanostructure [67]. Falsecolor images such as those shown in Fig. 7 might give the reader an exaggerated impression of surface roughness. Please note that while each image covers an area of 3 µm × 3 µm, the vertical scale is in nanometers. Though slightly rougher than clean glass, all culture surfaces evaluated in this study were quite smooth on the scale of mammalian cells.

CONCLUSIONS

Competitive chemisorption creates mixed surfaces in a rational manner Table 2 summarizes the results of this study. As we had hoped, we were able to control the surface chemistry by adjusting the reaction mixture. ESCA and FITC binding confirmed this (Figs 3 and 4). Topography of silanized surfaces was insular and complex, though their root-mean-square roughness did not differ significantly from that of smooth glass (Figs 7 and 8). Surfaces richest in diamine residues were the most hydrophilic (θA 90◦ , Fig. 5). We conclude that chemisorption from the N2-F3 organosilane system provides a simple and inexpensive, one-pot method for tailoring the chemistry of glass culture surfaces. Cell-culture data are presented in the accompanying paper [1].


415

Table 2. Summary of physical studies of etched glass and silanized surfaces Atomic ratio (F/F + N) by ESCA (%) Reactive amines by FITC binding Advancing water-contact angle θA (◦ ) Receding water-contact angle, θR (◦ ) Contact-angle hysteresis, θH (◦ ) Roughness by AFM, root-mean-square (nm) Morphology at nanometer scale

Etched glass 0 nd

Proteins, Cells and Materials: Festschrift in Honor of the 65th Birthday of Dr. John L. Brash

Harmonic analysis, signal processing, and complexity: Festschrift in honor of the 60th birthday of C.A. Berenstein

Asymptotic and computational analysis: conference in honor of Frank W.J. Olver's 65th birthday

Additive number theory: Festschrift in honor of 60th birthday of M.B.Nathanson

Revisiting the Foundations of Relativistic Physics: Festschrift in Honor of John Stachel

Revisiting the Foundations of Relativistic Physics: Festschrift in Honor of John Stachel

Logic, Language and Computation: Festschrift in Honor of Satoru Takasu

Focus on Language and Ethnicity: Essays in Honor of Joshua A. Fishman (Focusschrift in honor of Joshua A. Fishman on the occasion of his 65th birthday)

The Life and Death of Benjamin Brash

The Grothendieck Festschrift, Volume I: A Collection of Articles Written in Honor of the 60th Birthday of Alexander Grothendieck

Harmonic Analysis, Signal Processing, and Complexity: Festschrift in Honor of the 60th Birthday of Carlos A. Berenstein

Topological Methods in Modern Mathematics: A Symposium in Honor of John Milnor's Sixtieth Birthday

Design Methodologies for Secure Embedded Systems: Festschrift in Honor of Prof. Dr.-Ing. Sorin A. Huss

Topological Methods in Modern Mathematics: A Symposium in Honor of John Milnor's Sixtieth Birthday

Buddha Nature: A Festschrift in Honor of Minoru Kiyota

The Unity of Mathematics: In Honor of the Ninetieth Birthday of I.M. Gelfand (Progress in Mathematics)

Buddha Nature: A Festschrift in Honor of Minoru Kiyota

Complex, Contact and Symmetric Manifolds: In Honor of L. Vanhecke

Materials for Fuel Cells

Semigroups and Formal Languages: Proceedings of the International Conference, in Honour of the 65th Birthday of Donald B. Mcalister

The Secrets of Proteins

The Breadth of Symplectic and Poisson Geometry: Festschrift in Honor of Alan Weinstein

The Breadth of Symplectic and Poisson Geometry: Festschrift in Honor of Alan Weinstein

The breadth of symplectic and Poisson geometry: Festschrift in honor of Alan Weinstein

The Unity of Mathematics: In Honor of the Ninetieth Birthday of I.M. Gelfand

Geometry, Mechanics, and Dynamics: Volume in Honor of the 60th Birthday of J. E. Marsden

Geometry, Mechanics, and Dynamics: Volume in Honor of the 60th Birthday of J. E. Marsden

The physics of proteins

Nonlinear Wave Dynamics: Selected Papers of the Symposium Held in Honor of Philip L-F Liu's 60th Birthday

Advances in Statistics: Proceedings of the Conference in Honor of Professor Zhidong Bai on His 65th Birthday, National University of Singapore, 20 July 2008

In the Name of Honor

Proteins, Cells and Materials: Festschrift in Honor of the 65th Birthday of Dr. John L. Brash

Harmonic analysis, signal processing, and complexity: Festschrift in honor of the 60th birthday of C.A. Berenstein

Asymptotic and computational analysis: conference in honor of Frank W.J. Olver's 65th birthday

Additive number theory: Festschrift in honor of 60th birthday of M.B.Nathanson

Revisiting the Foundations of Relativistic Physics: Festschrift in Honor of John Stachel

Revisiting the Foundations of Relativistic Physics: Festschrift in Honor of John Stachel

Logic, Language and Computation: Festschrift in Honor of Satoru Takasu

Focus on Language and Ethnicity: Essays in Honor of Joshua A. Fishman (Focusschrift in honor of Joshua A. Fishman on the occasion of his 65th birthday)

The Life and Death of Benjamin Brash

The Grothendieck Festschrift, Volume I: A Collection of Articles Written in Honor of the 60th Birthday of Alexander Grothendieck

Harmonic Analysis, Signal Processing, and Complexity: Festschrift in Honor of the 60th Birthday of Carlos A. Berenstein

Topological Methods in Modern Mathematics: A Symposium in Honor of John Milnor's Sixtieth Birthday

Design Methodologies for Secure Embedded Systems: Festschrift in Honor of Prof. Dr.-Ing. Sorin A. Huss

Topological Methods in Modern Mathematics: A Symposium in Honor of John Milnor's Sixtieth Birthday

Buddha Nature: A Festschrift in Honor of Minoru Kiyota

The Unity of Mathematics: In Honor of the Ninetieth Birthday of I.M. Gelfand (Progress in Mathematics)

Buddha Nature: A Festschrift in Honor of Minoru Kiyota

Complex, Contact and Symmetric Manifolds: In Honor of L. Vanhecke

Materials for Fuel Cells

Semigroups and Formal Languages: Proceedings of the International Conference, in Honour of the 65th Birthday of Donald B. Mcalister

The Secrets of Proteins

The Breadth of Symplectic and Poisson Geometry: Festschrift in Honor of Alan Weinstein

The Breadth of Symplectic and Poisson Geometry: Festschrift in Honor of Alan Weinstein

The breadth of symplectic and Poisson geometry: Festschrift in honor of Alan Weinstein

The Unity of Mathematics: In Honor of the Ninetieth Birthday of I.M. Gelfand

Geometry, Mechanics, and Dynamics: Volume in Honor of the 60th Birthday of J. E. Marsden

Geometry, Mechanics, and Dynamics: Volume in Honor of the 60th Birthday of J. E. Marsden

The physics of proteins

Nonlinear Wave Dynamics: Selected Papers of the Symposium Held in Honor of Philip L-F Liu's 60th Birthday

Advances in Statistics: Proceedings of the Conference in Honor of Professor Zhidong Bai on His 65th Birthday, National University of Singapore, 20 July 2008

In the Name of Honor

Recommend Documents