Glycophase glass revisited: protein adsorption and cell growth on glass surfaces bearing immobilized glycerol monosaccharides

Biomaterials 23 (2002) 3347–3357

Glycophase glass revisited: protein adsorption and cell growth on glass surfaces bearing immobilized glycerol monosaccharides James R. Baina,*,1, Allan S. Hoffmanb a

Department of Pharmacology and Cancer Biology, Duke University Medical Center, DUMC 3813, Durham, NC 27710, USA b Department of Bioengineering, Box 352255, University of Washington, Seattle, WA 98195, USA Received 16 May 2001; accepted 22 January 2002

Abstract g-Glycerylpropylsilyl or ‘‘glycophase’’ glass has been promoted as a non-fouling surface, resistant to protein adsorption and cell attachment, on which one can immobilize oligopeptide ligands, and thus create cell-type-specific culture surfaces. The present study confirmed that the glycerol-rich glycophase surface is a useful support for peptide immobilization. But glycophase glass was observed to adsorb more albumin than glass. At pH 7.4, desorption studies revealed that albumin bound more tightly to glycophase glass than to glass. Moreover, the growth rates, morphologies, and functions of transgenic bG I/17 insulinoma cell cultures were equivalent on the two surfaces. Glycophase glass is neither protein- nor cell-repellant. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Non-fouling surfaces; Surface modification; Silanes; Peptide immobilization; Protein adsorption; Cell and tissue culture; Insulinoma cells

1. Introduction Biomaterials scientists have long sought to create surfaces that would resist fouling by adsorption of proteins and attachment of living cells [1–7]. Such protein- and cell-repellant surfaces would be useful in numerous applications, including medical implants, immunoassays, bioreactors, materials for pharmaceutical packaging, and surfaces exposed to sewage and seawater. Glycophase glass has been promoted as a non-fouling surface on which one can create cell-type-specific culture surfaces via covalent immobilization of oligopeptide cell-adhesion ligands [8–18]. Glycophase glass, prepared by hydrolysis of an oxirane silane residue (Figs. 1 and 2), bears pendant glycerol groups on its surface. Data from the present study confirm observations that this surface is a useful support for ligand immobilization [19–33], but do not support the contention that

*Corresponding author. Department of Pharmacology and Cancer Biology, Duke University Medical Center, DUMC 3813, Durham, NC 27710, USA. Tel.: +1-919-684-5224. E-mail address: [email protected] (J.R. Bain). 1 Preliminary reports of this work were published as abstracts [34,40].

glycophase glass resists protein adsorption and cell attachment [8–18,33].

2. Materials and methods 2.1. Selection, cleaning, and silanization of borosilicate glass Borosilicate cover glasses (12 mm circles, thickness B200 mm) were purchased from Carolina Biological Supply (Burlington, NC). Raw glass sheets of the borosilicate glass D263 were poured by Schott-Deutsche Spezialglas (Mainz, Germany), and cut into the final shape and sold under the trade name ‘‘Assistents’’ by Glaswarenfabrik Karl Hecht GmbH and Company . Germany). According to the manu(Sondheim/Rhon, facturer, the bulk formulation, in weight percent, is 65% SiO2, 12% (Na2O+K2O), 7.5% B2O3, 6.5% ZnO, 5.5% TiO2, 3% Al2O3, and 0.5% Sb2O3. Surface-analysis data are consistent with a glass of this identity (data not shown). Cleaning, etching, and silanization were conducted on cover glasses mounted in custom racks machined from polytetrafluoroethylene (PTFE). Two such racks (holding 32 cover slips each) were placed side-by-side in a

0142-9612/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 2 ) 0 0 0 3 5 - 2

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

3348

γ-Glycidoxypropyl trimethoxysilane (H3CO)3Si(CH2)3 O

H C C H2

CH2 O

Si OH

CH3OH O H Si OSi(CH2)3 O CH2 C C O H2 O

γ-Glycidoxypropyl silanized glass Fig. 1. Organosilanization of etched borosilicate glass with an epoxy silane. A monolayer is shown for simplicity.

γ-Glycidoxypropyl silanized glass O H Si OSi(CH2)3 O CH2 C C O H2 O 1 mM HCl 1 hour 90 °C

O Si OSi(CH2)3 O CH2CHCH2 O H O OH Glycerylpropylsilyl glass or "glycophase glass" Fig. 2. Acidic hydrolysis of the epoxy group to a diol.

Wheaton histological staining dish with a matching glass lid (Wheaton 400 USP Type I borosilicate glass; Wheaton Industries, Millville, NJ). Volumes of reagents for glass treatment were 150 ml unless otherwise noted. Aqueous solutions were based on deionized, then distilled water (ddH2O), resistance B18 MO/cm. Organic solvents were from J.T. Baker, Inc. (Phillipsburg, NJ), and were ‘‘analyzed-reagent’’ grade or better. Other reagents were from Sigma-Aldrich (Saint Louis, MO), unless noted otherwise. After two dip rinses in fresh acetone, racked cover glasses were ultrasonically cleaned 10 min in 2% Micros All-Purpose Liquid Cleaner (International Products

Corporation, Burlington, NJ) in ddH2O, pH 9.6, at 335 W, 47 kHz (Bransonics model B3210DTH, Branson Ultrasonics Corporation, Danbury, CT). Samples were then rinsed in running ddH2O, followed by three dip rinses in ddH2O, an acetone dip, 5 min ultrasonication in methylene chloride, then dried at 901C in air until just dry, and stored in vacuo at B0.5 mm Hg in polystyrene Petri dishes. Clean borosilicate glass coverslips were etched with sodium hydroxide (pH 13.7) for 30 min at 201C to expose free silanol anions, then silanized with a 1% v/v solution of (g-glycidoxypropyl) trimethoxysilane (Huls . America, Inc., Piscataway, NJ) in water, pH 5.3, for 2 h at 901C (Fig. 1). Epoxy groups were then hydrolyzed to the diol by immersion in 1 mm hydrochloric acid for 1 h at 901C (Fig. 2), creating glycophase glass. Reaction vessels were etched with 0.5 n NaOH between uses. Electron spectroscopy for chemical analysis (X-Probe ESCAs instrument, Surface Science Instruments, Mountain View, CA) revealed that this silanization procedure increased the surface content of both carbon and oxygen. Incremental changes in carbon and oxygen had an atomic ratio (C/O) of 2.1, close to the theoretical C/O ratio of 2.0 for the Si–O-linked glyceryl groups shown in Fig. 2. A brief note on nomenclature is in order. The term ‘‘glycophase glass,’’ alone and with several suffixes, has been used to denote a variety of different surfaces, including several ion-exchange resins. To complicate matters further, the word ‘‘Glycophase’’ and some of its variants were formerly registered trademarks of the Pierce Chemical Company (Rockford, IL) and Corning, Inc. (Corning, NY). As used here, ‘‘glycophase glass’’ refers to glycerylpropylsilyl glass, shown schematically in Fig. 2. In early 2002, a search of trademark records at the United States Patent and Trademark Office (Washington, DC) uncovered no active trademarks based on the term ‘‘glycophase’’. 2.2. Ligand design, iodination, and covalent immobilization Next, an attempt was made to create culture surfaces attractive to cells bearing the integrin, a4b1. Useful cells that express a4b1 include certain hæmatopoietic progenitors [34], among many others. The heptapeptide ligand, Gly–Ile–Asp–Ala–Pro–Ser–Tyr or GIDAPSY, was selected for solid-phase immobilization. The GIDAPSY molecule was designed around the fibronectin sequence, IDAPS, a ligand of the integrin, a4b1. An amino-acid sequence bound by the a4b1 cell-surface receptor is variously given as IDAPS or IDAP [35–39]. All immobilizations were carried out via the amino (or N) terminus. N-terminal glycine was added as a steric swivel, and C-terminal tyrosine was added to enable radio-iodination [8]. The Howard Hughes Medical

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

Institute, University of Washington, synthesized and purified GIDAPSY, and provided it in lyophilized form. In preparation for radio-iodination reactions, three Iodo-Beadss (Pierce, Rockford, IL) were rinsed in phosphate-buffered saline (PBS; catalog number P3813, Sigma, Saint Louis, MO; 0.138 m NaCl and 0.0027 KCl, pHB7.4) to remove any oxidant that was dislodged from the surface during shipping and handling. The three non-porous beads were then incubated for 5 min at B201C with a mixture of 10 ml Na125I in water (B100 mCi/ml; Amersham Life Science, Inc., Arlington Heights, IL) and 490 ml PBS. Oligopeptide solution was then added to the reaction mixture (50 ml at 1.0 mg GIDAPSY/ml PBS). Iodination proceeded 20 min at B201C, and was terminated by removal of the beads with forceps. For separation of peptide from free iodine species, the reaction mixture was immediately loaded onto a disposable C18 column (Waters SepPakPluss, catalog number 20515, Waters Chromatography Division, Milford, ME). Just prior to loading, the C18 columns were rinsed with 0.8 ml of a mixture comprising: 80% v/v methanol; 19.8% v/v deionized, distilled water (ddH2O); and 0.2% v/v trifluoroacetic acid (TFA); they were then equilibrated with PBS. Once loaded, columns were eluted by sequential reversal of the mobile phase, beginning with 99% water: 1% TFA (99W1TFA), and followed by stepwise 10% increases in methanol, to a maximum of 20% 99W1TFA: 80% methanol. Most free iodine eluted in the 0%-methanol loading solution, while approximately 90% of the peptide eluted in the 40–60%-methanol fractions. The latter three fractions were pooled, lyophilized, and stored frozen at
201C for later use. Glycophase glass discs were loaded into PTFE racks. For peptide immobilization, glycophase glass was activated for 15 min with 2.5% w/v 1,10 -carbonyldiimidazole (CDIZ) in acetone. Activated glass was rinsed twice in fresh acetone to remove free CDIZ. Many coupling experiments were performed. In a typical experiment, coupling proceeded unstirred for 20 h at 201C, with the oligopeptide dissolved in 0.2 m NaHCO3 buffer, pH adjusted to 9.0. The covalent linkage that forms between the peptide and the surface is an N-alkylcarbamate or urethane bond. After ligand immobilization, racked glass discs were rinsed five times in large volumes of fresh ddH2O. Racks were blotted on Kimwipess (Kimberly Clark Corporation, Roswell, GA) after the first three rinses. Gamma counting (GammaTracs 1185, TM Analytic, Elk Grove Village, IL) consistently showed that radiation levels in the rinse baths fell to background after the third rinse. In some experiments, the effects of a subsequent blocking reaction were evaluated. The blocking solution, 0.8 m ethanolamine, pH 9.0, in 0.2 m NaHCO3, was applied for 2 h, followed by three aqueous rinses.

3349

2.3. Why study protein adsorption and cell growth on glycophase glass? In the present series of investigations, suspicions that glycophase glass might not be a useful non-fouling surface were first aroused by the observation that the a4b1-positive human melanoma cell line, A-375 (American Type Culture Collection, Manassas, VA), attached, spread, and grew equally well on tissue-culture polystyrene, glycophase glass, etched glass, and GIDAPSYliganded glycophase glass. These A-375 cultures were performed under serum-containing medium [34,40]. This observation raised questions about claims that the short glyceryl groups on the surface of glycophase glass are able to repel protein adsorption and cell attachment [8–18,33]. These questions stimulated a new study of protein adsorption and cell growth on glycophase glass. 2.4. Protein selection, iodination, adsorption, and desorption Bovine serum albumin (BSA; Sigma catalog number A7638) was chosen as a model protein, since it is the most abundant protein in most cell-culture systems, and the only extracellular protein in many cell-attachment assays and cell-sorting procedures [34,40]. A sub-sample of BSA was radiolabeled with iodine-125 using IodoBeads (supra), and then separated from free iodine species by two passes on disposable gel-filtration columns (Sephadexs G-25 m, Amersham Pharmacia Biotech AB, Uppsala, Sweden). Non-labeled albumin was spiked with radiolabeled albumin to a specific activity of two million cpm/mg [40]. Adsorption studies were conducted at 0.1 mg BSA/ml in citrated, phosphate-buffered saline, with azide and iodide, or CPBSzI buffer (0.01 m sodium citrate, 0.01 m Na2HPO4, 0.11 m NaCl, 0.01 m NaI, 0.02% w/v NaN3, pH adjusted to 7.4). This phosphate-saline buffer included citrate for consistency with concomitant studies of fibrinogen adsorption in which citrate served as an anticoagulant (data not shown). Iodide was added to competitively reduce binding by any free iodine-125 species, and sodium azide (NaN3) was employed as an antimicrobial agent. Samples subjected to protein adsorption-desorption studies included low-density polyethylene (LDPE, Penn Fibre and Specialty Company, Inc., Fort Washington, PA), etched borosilicate glass, glycophase glass, CDIZ-activated-then-ethanolamine-blocked glycophase glass, and glycophase glass liganded with 10 pmol/cm2 GIDAPSY [40]. Prior to adsorption experiments, samples were hydrated overnight in CPBSzI at 41C. Adsorption proceeded 2 h at 371C, and was terminated by dilution-displacement rinsing with 25 volumes (100 ml) CPBSzI delivered hydraulically at a flow rate of 100 ml/min [40]. During

3350

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

and after gamma counting of the protein adsorbate (supra), desorption in citrate-phosphate buffer, along a pH series 2.2–7.4, proceeded for 48 h at B201C, followed by dip rinsing and recounting [40]. 2.5. Culture of bG I/17 transgenic insulinoma cells on glycophase glass

wells of untreated polystyrene (Falcon 1147), the negative control. Cultures proceeded seven days in a humidified incubator set to 95% air, 5% CO2, 371C, with daily changes of media. Conditioned media were frozen for later assays of metabolites. 2.6. Basal insulin secretion at seven days

To evaluate whether glycophase glass is indeed a cellrepellant surface [8–18], the insulinoma cell line, bG I/17, was cultured on glycophase glass and on two controls, etched borosilicate glass and untreated polystyrene, as follows. The transgenic rat insulinoma cell line, bG I/17 [41], was a gift from Dr. Christopher B. Newgard (Duke University Medical Center, Durham, NC). Cells were used at the 53rd passage after transfection with a single copy of the human proinsulin gene, whose expression is driven by the cytomegalovirus promoter/enhancer element [41]. Immediately prior to use in this study, bG I/17 stock was shown to be free of contamination with mycoplasmata by the Hoechst bisbenzamide 33258 fluorochrome. Near-confluent cultures of bG I/17 on tissue-culture polystyrene were rinsed twice with Hanks’ balanced-salt solution without CaCl2, MgCl2, MgSO4, or Phenol Red (HBSS; Gibco BRL Products, Grand Island, NY), and then trypsinized (0.05% porcine trypsin, 0.53 mm tetrasodium salt of ethylene diamine tetraacetic acid, in HBSS; GibcoBRL). Cells were suspended and centrifuged in the culture fluid, Medium 199 (with Earle’s salts, 100 mg/l l-glutamine, 2.2 g/l NaHCO3, and 25 mm N-2-hydroxyethylpiperazine-N0 -2ethane sulfonic acid (HEPES) buffer; Gibco BRL), plus 4.8% v/v fetal bovine serum (HyClone Laboratories, Inc., Logan, UT), and 7.54 mg/l gentamicin sulfate (Gibco BRL), enriched to a final concentration of 188 mg/dl (10.4 mm) d-glucose using 10% w/v glucose (Sigma). Supernatant was aspirated; cells were resuspended in culture medium, and counted (Coulter Counters, model ZBI, Coulter Electronics, Inc., Hialeah, FL). A final cell suspension was prepared in culture media such that 1 ml of media was added to each dry sample well, with a plating density of 80,000 cells/cm2, approximately one fifth of confluence. Etched glass and glycophase glass discs were mounted for cell culture in the bottom of 24-well, surfacedoxidized, tissue-culture plates (Falcon Multiwells 3047, Becton Dickinson Labware, Franklin Lakes, NJ), and were held in place by fluoroelastomeric o-rings made of poly (vinylidene fluoride-co-hexafluoropropylene), size 014 (Viton As, Kontes Glass Company, Vineland, NJ). Prior to use, o-rings were ultrasonically cleaned 10 min in neat ethanol, rinsed thrice in ddH2O, ultrasonically cleaned 10 min in 2% Micro (supra), rinsed four times in ddH2O, dried at 901C, and autoclaved. In addition to wells containing glass circles, o-rings were placed in

After a week of culture, insulinoma cells were washed twice for 15 min at 371C in Roswell Park Memorial Institute (RPMI) Medium 1640 (300 mg/l l-glutamine, and 0 mm d-glucose; GibcoBRL) containing 0.1% w/v radioimmunoassay-grade bovine-serum albumin (RIABSA; Sigma), pH adjusted to 7.4. Basal levels of insulin secretion were then established by a 1-h 371C incubation in RPMI 1640, with 0.1% RIA-BSA, 20 mm HEPES, 100 mm 7-chloro-3-methyl-2H-1,2,4-benzothiadiazine 1,1-dioxide or diazoxide (Sigma), pH 7.4. Diazoxide, or DZ, is a strong inhibitor of regulated insulin secretion [42]. Under the influence of DZ, secretion is reduced to basal levels, and occurs only by the constitutive pathway of secretion. DZ is poorly soluble in water at neutral pH, so it was necessary to first create a 1000  stock solution by suspending 23 mg DZ in 600 ml dimethyl sulfoxide, adding 5 ml 10 n NaOH, agitating until the DZ dissolved, then diluting to 1 ml with ethanol. 2.7. Maximal stimulation of insulin secretion Cultures were then washed for 15 min with RPMI 1640 containing 0.1% RIA-BSA, then stimulated for one hour with a ‘‘Swiss cocktail’’ of secretagogues to assess their maximum potential for insulin secretion. Formulation of the cocktail used in the present study was devised by workers at BetaGene, Inc. (Dallas, TX), modified from a mixture originally developed by Dr. Philippe Halban, Laboratoires Jeantet, Centre Medical Universitaire, University of Geneva, Switzerland. In the present study, the cocktail comprised 0.1 mm 3-isobutyl-1-methylxanthine or IBMX, 12.1 mm l-arginine, 5 mm (90 mg/dl) d-glucose, 10 mm l-leucine, 10 mm l-glutamine, 0.11 mm carbamylcholine chloride or carbachol (an acetylcholine analogue), 20 mm HEPES buffer, and 1 mg/ml RIA-BSA (all ‘‘Swiss cocktail’’ reagents were from Sigma), dissolved in RPMI 1640, pH adjusted to 7.4. Finally, samples were fixed in 10% neutral, buffered formalin, stained with Harris’s hematoxylin and eosin, inverted, and mounted on microscope slides. 2.8. Assays of glucose consumption and lactate production Media conditioned by the bG I/17 cultures were frozen in polypropylene microcentrifuge tubes, thawed, and assayed for glucose consumption and lactate

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

2

1.E+02

1.E+01

1.E+00

01 1.E+

00 1.E+

1 1.E-0

1.E-0

2

1.E-01 3

Insulin levels were assessed in the conditioned media from the seventh day of routine culture, the basal incubation under diazoxide, and the Swiss-cocktail stimulation, using the radioimmunoassay (RIA) kit from DiaSorin, Inc. (Stillwater, MN), calibrated for detection of human insulin. Calibration history is traceable to World Health Organization insulin standard 66/304. Porcine-insulin standards are used to create a calibration curve, using a curved spline fit. Data are then converted from (mUnits, mU) mU/ml to ng/ml using a conversion factor of 25 mU/ng.

1.E+03

1.E-0

2.9. Assay of human insulin

1.E+04

Immobilized oligopeptide (picomoles/cm )

production on a Kodak Ektachems DT60 II dry-slide analyzer (Ortho-Clinical Diagnostics, Inc., Johnson and Johnson, Rochester, NY). Unconditioned media were assayed daily to assure consistency in the initial glucose concentration (B188 mg/dl) and the presence of minimal initial lactate introduced by the fetal-calf serum (1.0 mm). Aliquots of 10 ml conditioned media were spotted onto dry slides. Reagents in each slide consisted of 0.7 Units of the oxidases of glucose and lactate, respectively, with a chromogenic system based on 7 Units peroxidase, 150 mg 1,7-dihydroxynaphthalene, and 250 mg 4-aminoantipyridine hydrochloride.

3351

Oligopeptide concentration (mg/mL) Fig. 3. N-terminal immobilization of the heptapeptide ligand, GIDAPSY, on carbonyldiimidazole-activated glycophase glass (r2 ¼ 0:993).

2.10. Statistical analysis Pairwise comparisons were evaluated post hoc with the Tukey–Kramer ‘‘honestly significantly different’’ (HSD) test [43], using version 3.0 of the JMPs software (SAS Institute, Inc., Cary, NC). Differences were considered significant at po0:05:

became cloudy, and the oligopeptide began to precipitate during the 20 h incubation. Glycophase glass liganded with 10 pmol GIDAPSY/ cm2 was selected for protein-adsorption studies. Previous workers reported that at this surface density, another fibronectin-inspired oligopeptide, GRGDY, provided a useful cell-culture surface [12].

3. Results 3.1. Solid-phase immobilization of the heptapeptide, GIDAPSY Empirical experimentation with reaction conditions led to a coupling procedure that yielded defined surface concentrations of the fibronectin-inspired ligand spanning three logarithmic decades (Fig. 3). In the example shown in Fig. 3, the coupling reaction was allowed to continue for 20 h at 201C, in stagnant 0.2 m NaHCO3 buffer, pH adjusted to 9.0. In this case, variability in the outcome was most pronounced at low peptide concentrations, due to radiation levels near background, and at high concentrations, where the peptide approached the limit of its solubility in the linkage buffer (Fig. 3). Under similar conditions, when peptide concentration was raised above 2.0 mg/ml in an attempt to create even higher surface densities of the ligand, the solution

3.2. Adsorption and desorption of the model protein, bovine serum albumin Contrary to contentions by two groups of prior investigators that this surface is protein-repellant [8–18,33], plain glycophase glass adsorbed more protein than the glass control (Fig. 4) [34,40]. GIDAPSY-linked glycophase glass adsorbed half the BSA of glass, and a third as much as glycophase glass (Fig. 4). The positive control, low-density polyethylene, adsorbed more than six times the BSA of all glass-based surfaces (216.3711.1 ng BSA/cm2 adsorbed on LDPE, n ¼ 27). Desorption in the pH series (Fig. 5, n ¼ 3 at each point) revealed that albumin was most tightly bound to all surfaces near its isoelectric point (pIB4.9). Above BSA’s pI, all glass-based surfaces showed decreasing protein retention with increasing pH. This included

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

3352

surfaces bearing the immobilized oligopeptide, with its two carboxylate groups (aspartate, C-terminal tyrosine). Borosilicate glass, etched at pH 13.7 to expose silanol anions prior to use, showed the strongest effect of pH on albumin desorption (Fig. 5). Tight binding of albumin to LDPE was relatively insensitive to pH (Fig. 5). Not unexpectedly, in all cases, at pH 7.4, 1% w/v sodium dodecyl sulfate in buffer was more effective at stripping albumin from surfaces than buffer alone (data not shown).

Albumin adsorbed (ng/cm2)

1000

100

3.3. Cell culture on glass and glycophase glass

10

Growth of the transgenic insulinoma line, bG I/17, was equivalent on glass and glycophase glass in the week of culture, as evidenced by data on glucose consumption and lactate production (Figs. 6 and 7). Cultures grown on glass and glycophase glass exhibited similar insulin secretion under the disparate conditions of diazoxide inhibition and secretagogue stimulation (Fig. 8). When comparing cultures on glass and glycophase glass, neither the authors of the present work nor any of their coworkers in the lab could distinguish the morphology of the living cultures under phase-contrast microscopy. Nor could observers detect any morphologic differences when examining the fixed, stained cultures by conventional bright-field microscopy. Photomicrographs documenting morphology of cultures

ne le et ly

po y sit en

-d

PS G

Lo

ID

w

A

ha Et

hy

e/ ol pm 0 ,1

Y

la no

ly G

2 cm

ke oc bl ein

m

co

Et

ph

ch

as

ed

eg

gl

la

as

s

ss

d

1

Fig. 4. Albumin adsorption on model surfaces. n ¼ 27 in all cases, except CDIZ-activated, ethanolamine-blocked glycophase glass, where n ¼ 3: Adsorption on etched glass, GIDAPSY-liganded glycophase glass, and LDPE differed significantly from adsorption on glycophase glass (Tukey–Kramer HSD test, po0:01).

100 90

Percent of albumin retained

80

Low-density polyethylene

70 60

GIDAPSY, 10 pmole/cm2

50 40

Glycophase glass

30 20 Etched glass 10 0 2.0

3.0

4.0

5.0

6.0

7.0

8.0

pH of desorption buffer Fig. 5. Desorption of albumin from four surfaces.

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

3353

Glucose consumption

8

CH2OH O OH 7

OH

Nanomoles glucose consumed/well/day

OH

OH

6

Glass 5

Glycophase glass Untreated polystyrene

4

3

2

1

0 1

2

3

4

5

6

7

Days in culture Fig. 6. Glucose consumption by transgenic bG I/17 insulinoma cells cultured on glycophase glass and two control materials. At no time point did cultures on glass differ from those on glycophase glass (p > 0:05).

grown on glycophase glass were qualitatively indistinguishable from those of cultures grown on glass.

4. Discussion 4.1. Carbonyldiimidazole-activated glycophase glass is a useful support for immobilization of oligopeptide ligands This investigation confirms that glycophase glass is a useful support for solid-phase immobilization of organic molecules [19–33]. By chance, the synthetic pathway described in the present study yielded much higher surface densities of fibronectin-inspired oligopeptides than those achieved by previous workers who used glycophase glass,

probably because the reactive intermediates formed by their sulfonyl-chloride methods [8–18] are less stable in aqueous linkage buffers than the imidazolyl-carbamate intermediate employed here. By working near the limits of solubility of the GIDAPSY oligopeptide ligand in the linkage buffer, the present study led to ligand densities that were more than a hundred-fold greater than the highest densities of sundry ligands achieved by those prior investigators (compare Fig. 3 to their Fig. 1 [9]). However, this result came at a cost. Their sulfonylchloride method creates a given surface concentration of peptide at much lower input concentrations of peptide. Thus, in choosing between the two methods for immobilizing ligands on glycophase glass, one might select the sulfonyl-chloride chemistry for economy of peptide, or the carbonyldiimidazole chemistry to achieve high ligand densities.

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

3354

14

Secretion during the seventh day of routineculture Lactate production

Basal secretion under diazoxideinhibition

OH 12

H3C

C

Maximal secretion under Swiss-cocktail stimulation *

COOH

H

40

Insulin secreted (ng/well/hour)

Lactate produced (mM), volume 1 mL

10 Glass Glycophase glass 8 Untreated polystyrene

6

30

20

10

4 0 Glass

2

Glycophase glass

Untreated polystyrene

Fig. 8. Insulin secretion by transgenic bG I/17 insulinoma cells cultured on glycophase glass and two control materials. *Differs significantly from glass (0:05 > p > 0:01).

0 1

2

3

4

5

6

7

Days in culture Fig. 7. Lactate production by transgenic bG I/17 insulinoma cells cultured on glycophase glass and two control materials. At no time point did cultures on glass differ from those on glycophase glass (p > 0:05).

4.2. Glycophase glass is not protein-repellant Under the conditions of this study, glycophase glass adsorbed more protein than glass itself (Fig. 4). Moreover, at physiologic pH, albumin bound more tightly to glycophase glass than to glass (Fig. 5). Clearly, glycophase glass cannot be considered a non-fouling surface. As early as 1989, Bruin et al. [22] reported that chromatographic surfaces with immobilized poly (ethylene glycol) or PEG provided better non-fouling surfaces than glycophase glass. In their study, glycophase glass exhibited strong interactions with lysozyme, trypsin, and chymotrypsinogen at pH>5 (their Fig. 5b). Similarly, Xiao and Truskey [32] observed more fibronectin adsorption to glycophase glass than to glass under two of the three adsorption conditions they evaluated (their Table 1). In the current study, relatively low binding of albumin to GIDAPSY-liganded glycophase glass at pH 7.4

(Fig. 4, note logarithmic scaling) might have been due in part to charge–charge repulsion between the net negative charges on albumin (pIB4.9) and the N-terminal-immobilized GIDAPSY groups (aspartate, carboxy terminus). Binding of bovine serum albumin to glass was strongest near BSA’s isoelectric point (pIB4.9; Fig. 5), a phenomenon observed by others since at least the 1950s [44]. Glycophase glass and GIDAPSY-liganded glycophase glass also showed decreasing BSA retention with increasing pH above pHB5 (Fig. 5). Many proteins exhibit adsorption maxima near their isoelectric points, particularly when adsorbed from unitary solutions onto uncharged or weakly charged surfaces [45,46]. The positive control, low-density polyethylene, adsorbed much more albumin than any glass-based surface (Fig. 4). Binding of BSA to LDPE was relatively tight and pH-insensitive (Fig. 5), presumably due to hydrophobic interactions between the protein adsorbate and the surface of this polyolefin [40]. 4.3. Glycophase glass is not cell-repellant Neither morphologic appearance nor metabolic data revealed any substantive differences between cell growth and function on glycophase glass and glass (Figs. 6–8).

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

For most anchorage-dependent vertebrate cells that can be propagated in vitro, glass is a permissive culture surface. This observation dates from the earliest days of vertebrate cell culture [47–49]. In serum-containing media, bG I/17 insulinoma cells, which can be exquisitely sensitive to surface chemistry in other organosilane systems [50], grow and function equally well on glass and surface-oxidized, tissue-culture polystyrene (TCPS, data not shown). In recent decades, TCPS has come to supplant glass as a routine, inexpensive, and permissive culture surface in laboratories worldwide. In view of all this, glycophase glass should be considered a permissive culture surface, perhaps on par with glass and TCPS, and not a cell-repellant surface, as proposed by previous workers [8–18]. In the present study, one datum did suggest a subtle difference in cell function on glass versus glycophase glass. Hourly insulin secretion during the seventh and final day of routine culture (Fig. 8) was significantly higher on glycophase glass. This apparent difference was not confirmed under the more rigorous metabolic conditions of inhibition by diazoxide and stimulation with the ‘‘Swiss cocktail’’ of secretagogues (Fig. 8). Such drug-influenced secretory changes serve as standards by which the performance of many endocrine cells is evaluated [51–58]. Therefore, the authors attach no great biologic importance to the apparent differences in insulin secretion during the seventh day of routine culture. Independent studies support the notion that cells can attach, spread, and grow on glycophase glass. Subsequent to publication of the present results in abstract form [34,40], Cle! mence et al. [24] suggested, ‘‘unmodified glycophase glass is an adhesive substrate for NG 108-15 [hybrid: mouse neuroblastoma  rat glioblastoma] cells, both in serum-containing medium and in serum-free medium’’, and Xiao and Truskey [32] observed that bovine aortic endothelial cells adhered well to fibronectin adsorbed onto either glass or glycophase glass. Drumheller and coworkers [59,60] found that 100% of the human foreskin fibroblasts they plated onto glycophase glass had adhered to the surface within the first day of culture. Glycophase glass, created by chemisorption of an organosilane to glass, bears O-linked glyceryl monosaccharides on its surface (Fig. 2). Several workers have fabricated analogous surfaces by chemisorption of mercaptoglycerol to gold. The S-linked mercaptoglyceryl groups immobilized by this method bind immunoglobulins and activate the complement cascade of serine proteases in vitro, and provoke inflammation in vivo [61,62]. Taken together, data from silane-on-glass and thiol-on-gold systems suggest that surfaces rich in short diol residues offer little promise as anti-fouling surfaces.

3355

5. Conclusions The authors do not doubt that earlier workers observed differences in the attachment and spreading of cells in the initial minutes and hours after plating cells onto glycophase glass and sundry surfaces derived from it. It is tempting to speculate that most such differences would disappear in a day or two in cultures grown under serum-containing media. In view of the data presented here, and considering the very short-term nature of their studies, claims by previous investigators that glycophase glass resists protein adsorption and cell attachment [8–18] must be interpreted narrowly. While it might indeed be possible to create cell-typespecific culture surfaces by immobilizing ligands on nonfouling surfaces, earlier claims of success appear to have been premature. Useful chromatographic packings of glycerylpropylsilyl silica were commercially available at the time this manuscript was written (early 2002; e.g., UltraPureDiol3s, SiliCycle Chemical Division, Que! bec City, Que! bec, Canada), and investigators continue to create useful, diol-rich surfaces on inorganic supports through the use of g-glycidoxypropyl silanes. Persistent claims that the immobile glycerol groups created in this way can repel protein adsorption and cell attachment [63,64] should be interpreted with care. Those in search of non-fouling surfaces might find more productive hunting in the great diversity of poly(ethylene glycols) and polysaccharides.

Acknowledgements This work was supported by the US National Institutes of Health (NIH # 1-T32-GM-08437-01 and 5-T32-HL-07403-15), Gore Hybrid Technologies, Inc., Flagstaff, AZ, and BetaGene, Inc., Dallas, TX. Deb Ballard, Bobbie D. Becker, Mark E. Blaylock, Nina Burns, Dr. Sam Clark, Katie Gooby, Dr. Thomas A. Horbett, and Beth Roche provided technical assistance.

References [1] Goessl A, Bowen-Pope DF, Hoffman AS. Control of shape and size of vascular smooth muscle cells in vitro by plasma lithography. J Biomed Mater Res 2001;57:15–24. [2] Shen M, Pan YV, Wagner MS, Hauch KD, Castner DG, Ratner BD, Horbett TA. Inhibition of monocyte adhesion and fibrinogen adsorption on glow discharge plasma deposited tetraethylene glycol dimethyl ether. J Biomater Sci Polym Ed 2001;12:961–78. [3] Novick SJ, Dordick JS. Protein-containing hydrophobic coatings and films. Biomaterials 2002;23:441–8. [4] Brady Jr RF. A fracture mechanical analysis of fouling release from nontoxic antifouling coatings. Prog Org Coat 2001;43: 188–92.

3356

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357

[5] Wang P, Tan KL, Kang ET, Neoh KG. Plasma-induced immobilization of poly(ethylene glycol) onto poly(vinylidene fluoride) microporous membrane. J Membr Sci 2002;195:103–14. [6] Timmons RB, Wu Y. Method for depositing a coating comprising pulsed plasma polymerization of a macrocycle. US Patent Appl No. 20020004104, 2002. [7] Watanabe T, Hayakawa M, Machida M. Antifouling member and antifouling coating composition. US Patent No. 6337129, 2002. [8] Hubbell JA, Massia SP, Desai NP, Drumheller PD. Endothelial cell-selective materials for tissue engineering in the vascular graft via a new receptor. Bio/Technol 1991;9:568–72. [9] Massia SP, Hubbell JA. Covalent surface immobilization of Arg– Gly–Asp- and Tyr–Ile–Gly–Ser–Arg-containing peptides to obtain well-defined cell-adhesive substrates. Anal Biochem 1990;187:292–301. [10] Massia SP, Hubbell JA. Evaluation of vascular endothelial cell interactions with surfaces containing covalently grafted adhesion peptides. Trans Soc Biomater 1990;13:240. [11] Massia SP, Hubbell JA. Human endothelial cell interactions with surface-coupled adhesion peptides on a nonadhesive glass substrate and two polymeric biomaterials. J Biomed Mater Res 1991;25:223–42. [12] Massia SP, Hubbell JA. An RGD spacing of 440 nm is sufficient for integrin aVb3-mediated fibroblast spreading and 140 nm for focal contact and stress fiber formation. J Cell Biol 1991;114:1089–100. [13] Massia SP, Hubbell JA. Biomaterials which selectively support the attachment of vascular endothelial cells via a new adhesion receptor. Trans Soc Biomater 1991;14:238. [14] Massia SP, Hubbell JA. Vascular endothelial cell adhesion and spreading promoted by the peptide REDV of the IIICS region of plasma fibronectin is mediated by integrin a4b1. J Biol Chem 1992;267:14019–26. [15] Massia SP, Hubbell JA. Covalently immobilized laminin peptide Tyr–Ile–Gly–Ser–Arg (YIGSR) supports cell spreading and colocalization of the 67-kilodalton laminin receptor with a-actinin and vinculin. J Biol Chem 1993;268:8053–9. [16] Hubbell JA, Massia SP, Desai NP. Surfaces having desirable cell adhesive effects. US Patent No. 5330911, 1994. [17] Hossiany SFA, Desai NP, Hubbell JA. Avoidance of photoactivation in the epifluorescence video microscopic observation of thrombosis. J Biomed Mater Res 1992;26:1535–42. [18] Drumheller PD, Massia SP, Hubbell JA. Endothelial cell-specific biomaterials obtained by immobilized Arg–Glu–Asp [sic] peptide analogues. Proceedings of the Fourth World Congress on Biomaterials, Berlin, Germany, 1992. p. 216. [19] Oliver R, Royer GP. Selective adsorbent for use in affinity chromatography. US Patent No. 4061591, 1977. [20] Bandi ZL, Moslen MT, Reynolds ES. Ion-exchange chromatography of phosphatides and glycolipids on quaternary triethylammonium glycophase covalently bound to controlled pore glass. J Chromatogr 1982;249:93–101. [21] Borrego F, Tari M, Manjon A, Iborra JL. Properties of pectinesterase immobilized on glycophase-coated controlled-pore glass. Appl Biochem Biotech 1989;22:129–40. [22] Bruin GJM, Huisden R, Kraak JC, Poppe H. Performance of carbohydrate-modified fused-silica capillaries for the separation of proteins by zone electrophoresis. J Chromatogr 1989;480:339–49. [23] Burdick BA, Esders TW, Schaeffer JR, Lynn S. Polymeric thiols as enzyme activators of serum creatine phosphokinase. Appl Biochem Biotech 1987;16:145–56. [24] Cl!emence J-F, Ranieri JP, Aebischer P, Sigrist H. Photoimmobilization of a bioactive laminin fragment and pattern-guided selective neuronal cell attachment. Bioconjugate Chem 1995;6:4111–7.

[25] Ernst-Cabrera K, Wilchek M. Coupling of ligands to primary hydroxyl-containing silica for high-performance affinity chromatography: optimization of conditions. J Chromatogr 1987;397:187–96. [26] Nilsson K, Mosbach K. Immobilization of ligands with organic sulfonyl chlorides. Methods Enzymol 1984;104:56–69. [27] Regnier FE, Chang S-H. Method for rapid analysis of hemoglobin variants by high speed liquid chromatography. US Patent No. 4108603, 1978. [28] Schaeffer JR, Snoke RE, Harris HW. Composition for chromatography. US Patent No. 4111838, 1978. [29] Schacht J. Purification of polyphosphoinositides by chromatography on immobilized neomycin. J Lipid Res 1978;19:1063–7. [30] Stark M-B, Holmberg K. Covalent immobilization of lipase in organic solvents. Biotechnol Bioeng 1989;34:942–50. [31] Stuart JK, Hlady V. Feasibility of measuring antigen-antibody interaction forces using a scanning force microscope. Colloids Surf B: Biointerf 1999;15:37–55. [32] Xiao Y, Truskey GA. Effect of receptor-ligand affinity on the strength of endothelial cell adhesion. Biophys J 1996;71:2869–84. [33] Regnier FE. Bonded carbohydrate stationary phases for chromatography. US Patent No. 3983299, 1976. [34] Bain JR, Hoffman AS, Horbett TA. Protein adsorption on glycerylpropylsilyl glass, with and without an immobilized ligand for a4b1+ hematopoietic cells. J Cell Biochem, Suppl 1994;18C:282. [35] Mould AP, Humphries MJ. Identification of a novel recognition sequence for the integrin a4b1 in the COOH-terminal heparinbinding domain of fibronectin. Eur Mol Biol Org J 1991;10: 4089–95. [36] Sanchez-Aparicio P, Dominguez-Jimenez C, Garcia-Pardo A. Activation of the a4b1 integrin through the b1 subunit induces recognition of the RGDS sequence in fibronectin. J Cell Biol 1994;126:271–9. [37] Clements JM, Newham P, Shepherd M, Gilbert R, Dudgeon TJ, Needham LA, Edwards RM, Berry L, Brass A, Humphries MJ. Identification of a key integrin-binding sequence in VCAM-1 homologous to the LDV active site in fibronectin. J Cell Sci 1994;107:2127–35. [38] Humphries MJ, Sheridan J, Mould AP, Newham P. Mechanisms of VCAM-1 and fibronectin binding to integrin a4b1: implications for integrin function and rational drug design. Ciba Found Symp 1995;189:177–99. [39] Sharma A, Askari JA, Humphries MJ, Jones EY, Stuart DI. Crystal structure of a heparin- and integrin-binding segment of human fibronectin. Eur Mol Biol Org J 1999;18:1468–79. [40] Bain JR, Hoffman AS, Horbett TA. Protein adsorption on plain and peptide-linked glycerylpropylsilyl glass. Trans Soc Biomater 1994;17:336 [Nota bene: Data in Fig. 2 of this abstract contain a minor calculation error. Corrected data appear in the present report]. [41] Clark SA, Quaade C, Constandy H, Hansen P, Halban P, Ferber S, Newgard CB, Normington K. Novel insulinoma cell lines produced by iterative engineering of GLUT2, glucokinase, and human insulin expression. Diabetes 1997;46:958–67. [42] Petit P, Loubati"eres-Mariani MM. Potassium channels of the insulin-secreting b cell. Fundam Clin Pharmacol 1992;6:123–34. [43] Sokal RR, Rolf FJ. Biometry, 2nd ed. San Francisco: Freeman, 1981. [44] Bull HB. Adsorption of bovine serum albumin on glass. Biochim Biophys Acta 1956;19:464–71. [45] Brash JL. Role of plasma protein adsorption in the response of blood to foreign surfaces. In: Sharma CP, Szycher M, editors. Blood compatible materials and devices: perspectives towards the 21st Century. Basel: Technomic Publishing Co., Inc., 1991. p. 3–24.

J.R. Bain, A.S. Hoffman / Biomaterials 23 (2002) 3347–3357 [46] Horbett TA. Adsorption of proteins and peptides at interfaces. In: Ahern TJ, Manning MC, editors. Stability of protein pharmaceuticals, part A: Chemical and physical pathways of protein degradation. New York: Plenum Press, 1992. p. 195–214. [47] Carrel A, Burrows MT. An addition to the technique of the cultivation of tissues in vitro. J Exp Med 1911;14:244–6. [48] Harrison RG. On the stereotropism of embryonic cells. Science 1911;34:279–81. [49] Weiss P. Experiments on cell and axon orientation in vitro: the role of colloidal exudates in tissue organization. J Exp Zool 1935;100:353–86. [50] Bain JR, Hoffman AS, Horbett TA. Growth and function of transgenic insulinoma cells, bG I/17, cultured on thin films of mixed aminofluorosiloxanes. 37th International Symposium on Macromolecules, July 12–17, 1998, Gold Coast, Queensland, Australia, World Polymer Congress, International Union Pure and Applied Chemistry, preprints. p. 789. [51] Hohmeier HE, Mulder H, Chen G, Henkel-Rieger R, Prentki M, Newgard CB. Isolation of INS-1-derived cell lines with robust ATP-sensitive K+ channel-dependent and -independent glucosestimulated insulin secretion. Diabetes 2000;49:424–30. [52] Irminger JC, Vollenweider FM, Neerman-Arbez M, Halban PA. Human proinsulin conversion in the regulated and the constitutive pathways of transfected AtT20 cells. J Biol Chem 1994;269:1756–62. [53] Desai TA. Microfabricated biocapsules for the immunoisolation of pancreatic islets of Langerhans, PhD dissertation. University of California, San Francisco and Berkeley, 1998. xii + 256 pp. [54] Persaud SJ. Pancreatic beta-cell lines: their roles in beta-cell research and diabetes therapy. Adv Mol Cell Biol 1999;29:21–46. [55] Lucas-Clerc C, Massart C, Campion JP, Launois B, Nicol M. Long-term culture of human pancreatic islets in an extracellular

[56]

[57]

[58] [59]

[60]

[61]

[62]

[63]

[64]

3357

matrix: morphological and metabolic effects. Mol Cell Endocrinol 1993;94:9–20. Bonner-Weir S, Taneja M, Weir GC, Tatarkiewicz K, Song K-H, Sharma A, O0 Neil J. In vitro cultivation of human islets from expanded ductal tissue. Proc Natl Acad Sci USA. 2000;97: 7999–8004. Tiedge M, Elsner M, McClenaghan NH, Hedrich H-J, Grube D, Klempnauer J, Lenzen S. Engineering of a glucose-responsive surrogate cell for insulin replacement therapy of experimental insulin-dependent diabetes. Hum Gene Ther 2000;11:403–14. Thul!e PM, Liu J, Phillips LS. Glucose regulated production of human insulin in rat hepatocytes. Gene Ther 2000;7:205–14. Drumheller PD. Polymer networks of poly (ethylene glycol) as biospecific cell adhesive substrates, PhD dissertation. University of Texas at Austin, 1994. xiii+146 pp. Drumheller PD, Elbert DL, Hubbell JA. Multifunctional poly (ethylene glycol) semi-interpenetrating polymer networks as highly selective adhesive substrates for bioadhesive peptide grafting. Biotechnol Bioeng 1994;43:772–80. Tengvall P, Askendal A, Lundstr.om I. Complement activation by 3-mercapto-1,2-propanediol immobilized on gold surfaces. Biomaterials 1996;17:1001–7. Tang L, Liu L, Elwing HB. Complement activation and inflammation triggered by model biomaterial surfaces. J Biomed Mater Res 1998;41:333–40. Sch.afer C, Lubda D. Alkyl diol silica: restricted access precolumn packings for fast liquid chromatography-integrated sample preparation of biological fluids. J Chromatogr A 2002;909:73–8. Ligler FS, Bhatia S, Shriver-Lake LC, Georger J, Calvert J, Dulcey C. Method for surface modification to create regions resistant to adsorption of biomolecules. US Patent No. 5965305, 1999.