Nanoscale Clustering of RGD Peptides at Surfaces Using Comb

Biomacromolecules 2001, 2, 85-94

85

Nanoscale Clustering of RGD Peptides at Surfaces Using Comb Polymers. 1. Synthesis and Characterization of Comb Thin Films Darrell J. Irvine and Anne M. Mayes* Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

Linda G. Griffith Department of Chemical Engineering & Division of Bioengineering and Environmental Health, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 Received July 13, 2000

Theoretical and experimental studies were conducted to elucidate the structure and properties of amphiphilic comb polymer thin films presenting nanoscale clusters of Arg-Gly-Asp (RGD) peptides for control of cell adhesion on biomaterials. Combs comprised of a poly(methyl methacrylate) backbone and short poly(ethylene oxide) side chains were synthesized, and peptides were tethered to the side chain ends to create nanoscale peptide clusters. In thin films, comb polymers containing g30 wt % six to nine unit PEO side chains completely resisted adhesion of a model fibroblast cell line in the presence of 7.5% serum over 24 h. These same polymers modified with RGD peptides elicited tunable cell adhesion when mixed with unmodified combs in varying proportion. A self-consistent field lattice model of the interface between comb polymer films and water predicts an organization of the top molecular layer of comb polymer with the backbone oriented parallel to the interface in quasi-two-dimensional confinement and hydrophilic side chains extended in a brushlike layer into solution. This picture of a quasi-2D configuration is consistent with the observed surface properties of comb films in water as well as measurements of the RGD cluster density on mixed comb surfaces using fluorescent nanosphere labeling of ligand clusters. Introduction Cell adhesion to extracellular matrix (ECM) in vivo is primarily controlled by cell-surface receptors known as integrins binding to ECM molecules of appropriate type and spatial distribution. However, typical biomaterials when implanted adsorb a nonphysiological layer of protein on their surface, leading to uncontrollable cell-biomaterial interactions and cell responses. The goal of much current biomaterials research is to eliminate these interactions, which are difficult to predict or control, in favor of tailored biochemical communication with cells via surface-localized peptide or protein signals.1-9 This idea has been approached with surface modifications designed to guide the adsorption of specific proteins in specific orientations9 or immobilize cellsignaling peptides or proteins at the surface of materials.1-8,10 To enhance effectiveness, these cell-signaling surfaces often simultaneously resist adsorption of conflicting protein signals from serum, usually by incorporating a highly hydrophilic component in the surface layer such as poly(ethylene oxide) (PEO) or dextran. A variety of methods have been used for implementation of this paradigm and for incorporating PEO at surfaces for protein resistance. A few examples include physisorption,11 chemisorption,12 covalent grafting of polymers (including graft polymerization),13-17 and plasma glow discharge treat-

ment.18 However, the approaches reported to provide resistance to nonspecific cell adhesion coupled with tethered ligand presentation are typically the most inflexible from an application standpoint. For example, several approaches have demonstrated control of cell adhesion by coupling surface modification agents such as self-assembled monolayers to surfaces from solution.2-4,8 Such approaches require chemistry to be carried out on the biomaterial surface, which limits the applicability of the approach to materials with the appropriate reactive sites and further introduces multiple surface treatment steps. Beyond these processing issues, biomaterials scientists have yet to exploit biophysical aspects of cell adhesion unveiled by cell biologists. For example, biochemical signaling from integrins binding to ligand is known to be strongly influenced by the spatial distribution of these receptorss complete signaling only occurs when integrins are clustered in the cell membrane.19-22 Thus the spatial distribution of immobilized ligand presented on a biomaterial surface might have a significant influence on cell responses to the substrate. Promising initial results demonstrating the consequences of ligand clustering on cell function were obtained by Maheshwari et al. in a model tethered-adhesion ligand system.6 By covalently linking RGD peptides to the chain ends of star PEO molecules atop a PEO hydrogel, cell adhesion strength

10.1021/bm005584b CCC: $20.00 © 2001 American Chemical Society Published on Web 12/07/2000

86

Biomacromolecules, Vol. 2, No. 1, 2001

and migration speeds were dramatically altered compared to surfaces presenting unclustered RGD on PEO. Development of practically relevant biomaterials with tunable spatial presentation of ligand could thus offer new capabilities in controlling cell function. The current work was motivated by the need for a simple, inexpensive surface modification route to control both biochemical and biophysical aspects of cell adhesion by nanoscale clustering of ligand at surfaces. In previous work we demonstrated the use of RGD-modified amphiphilic comb copolymers of methyl methacrylate and poly(oxyethylene) methacrylate, P(MMA-r-POEM), as stabilizers for acrylic latexes which could be used as cell-interactive coatings.1 We present here the first part of our studies of RGD-modified and unmodified P(MMA-r-POEM) comb mixtures designed to provide nanoscale ligand clustering on a protein-resistant background. This paper focuses on meeting these goals within the framework of a nondegradable implant coating applied using simple aqueous-based coatings operations. In part two of these studies, we demonstrate that such materials, generally devised, may be of interest for tailoring cell adhesion on 3D bioresorbable cell scaffolds for tissue engineering by an alternative surface modification method, namely, surface segregation. Numerical Modeling Self-consistent mean field (SCF) calculations were used to predict the structure of P(MMA-r-POEM) comb polymer films in equilibrium with water at the film’s free surface. The SCF model used herein is an extension of the original Scheutjens and Fleer lattice SCF model for polymer adsorption.23-25 The method has been treated thoroughly in a recent text25 and will not be described in detail here. Calculations were made for polymer chains and water occupying a hexagonal lattice with concentration profiles varying only in layers perpendicular to the water/film interface. The MMA and EO units of the comb polymer were modeled as hydrophobic segments and hydrophilic segments, designated A and B, respectively. Water was explicitly included in the calculations as a single-segment molecule of a third chemical type, S. All three segment types are assumed to have identical dimensions equal to one lattice site in volume. Calculations were made for combs having 150 backbone segments. Short teeth of four segments each are connected every seven units along the backbone (20 total teeth per comb). This architecture and composition models the C1 comb polymer prepared for experimental studies (described below), providing similar mole and weight fractions of the hydrophilic side chain segments. Interactions between the components of the system were accounted for by the magnitude of the Flory-Huggins segmental interaction parameter χ for each chemical pair. Since the basic repeat units of the comb (methyl methacrylate and ethylene oxide) are miscible,26,27 we set χAB ) 0. To model the interaction of the comb polymer components with water, we set χBS ) 0.4 (modeling poly(ethylene glycol) in water28) and χAS ) 2.5 (water is a nonsolvent for the backbone units). The equilibrium distribu-

Irvine et al.

tion of segments within the lattice is calculated by finding concentration profiles of the components that account for all possible chain configurations, subject to layer potentials that account for the energy of contacts between segments and maintain system incompressibility (the sum of volume fractions of all components within a given lattice layer must be unity). These potentials must be simultaneously selfconsistent with the polymer concentration profiles. Systems were composed of 150 lattice layers, with reflecting boundary conditions at layer 0 and layer 150 to model a semi-infinite interface. The water/polymer interface was initially created by making calculations with an initial guess for the segment potentials, biasing the solvent to localize in layers near the boundary at layer 0 and the polymer at the opposite side of the lattice, with χ parameters set to induce phase separation. The set of nonlinear equations governing the layer potentials and segment distributions is solved by numerical iteration from this initial guess using an unconstrained optimization routine similar to that described by Dennis.29 Equilibrium concentration profiles calculated for each segment type are presented. Experimental Methods Materials. Methyl methacrylate (MMA), poly(ethylene glycol) methacrylate (referred to herein as hydroxy-poly(oxyethylene) methacrylate, HPOEM, Mn ∼ 360 g/mol), poly(ethylene glycol) methyl ether methacrylate (referred to herein as poly(oxyethylene) methacrylate, POEM, Mn ∼ 300, 475, 1100, or 2080 g/mol), tetrahydrofuran (THF), azo(bis)isobutyronitrile (AIBN), 1-methoxyphenol, succinic anhydride, anhydrous dichloroethane, N-methylimidazole, N-hydroxysuccinimide (NHS), dicyclohexylcarbodiimide (DCC), sodium cyanoborohydride, Tween 20 surfactant, and dimethylformamide (DMF) were obtained from Aldrich. Phosphate buffered saline (PBS), pH 7.4, and N-(2-hydroxyethyl)piperazine-N′-(2-ethanesulfonic acid) (HEPES) buffer, pH 7.4, were prepared from prepackaged dry packets from Sigma. Petroleum ether, ethanol, methanol, dichloromethane, anhydrous ethyl ether, ethyl acetate, and deuterated solvents were obtained from VWR Scientific. N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC) was obtained from Pierce Chemical Co. Fluorescent polystyrene nanospheres (yellow-green aldehyde sulfate Fluospheres) with a nominal diameter of 29 nm were obtained from Molecular Probes. Poly(methyl methacrylate) (PMMA, Mn ) 168 000 g/mol, PDI ) 1.07) was purchased from PolySciences. Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP) and GlyArg-Gly-Glu-Ser-Pro (GRGESP) peptides were obtained from Gibco. Gly-Arg-Gly-Asp-Ser-Pro-Lys (GRGDSPK) was obtained from American Peptide Co. All reagents were used as received unless otherwise noted. Comb Polymer Synthesis and Characterization. The chemical structure of comb copolymers prepared for these studies is shown in Figure 1. The combs are random terpolymers of MMA, HPOEM, and POEM. For brevity, comb terpolymers are referred to generically throughout as P(MMA-r-POEM). Combs all contained ∼50:50 weight

Biomacromolecules, Vol. 2, No. 1, 2001 87

Clustering of RGD Peptides

Figure 1. P(MMA-r-POEM) base structure. The combs are random terpolymers of MMA, HPOEM (m ∼ 6), and POEM (n ∼ 9). Table 1. Comb Polymer Physical Data polymer

Mn (g/mol)

Mw (g/mol)

PDI

composition (w/w/w) of MMA/HPOEM/POEM)

C1 C2

25 870 93 900

44 870 192 000

1.73 2.04

62/18/20 66/16/18

ratios of HPOEM and POEM except for two experiments where the total PEO content and side chain length were specifically varied, as noted below. Copolymers comprised of e50 wt % HPOEM/POEM monomers were water insoluble. Combs were synthesized as described previously;1 the following represents a typical synthesis protocol: 21 mL of MMA (0.197 mol), 6.55 g of HPOEM (0.0182 mol), 6.55 g of POEM (0.0138 mol), and 0.239 g of AIBN (0.00146 mol) were added to 500 mL of THF in a 1000-mL roundbottom flask equipped with a condenser. The solution was degassed by bubbling nitrogen for 20 min, followed by refluxing at 70 °C for 18 h. The reaction was terminated by addition of 20 mg of 1-methoxyphenol. The resulting copolymer was purified by two precipitations in 8:1 (v/v) petroleum ether/methanol and dried in vacuo at 25 °C for 24 h. Physical characteristics of the two P(MMA-r-POEM) polymers C1 and C2 used in the majority of these studies are listed in Table 1. Two series of comb polymers were additionally prepared to assess the effects of composition and side chain length on the resistance of comb surfaces to nonspecific cell adhesion. For these experiments, comb polymers with molecular weights comparable to C1 were prepared as described above with no HPOEM macromer. To assess composition effects, combs were prepared containing 20, 30, or 45 wt % POEM. To examine side chain length effects, combs were prepared with 45 wt % total POEM, using POEM monomer having 5, 9, 23, or 45 ethylene oxide repeat units. Comb compositions were determined using a Bruker Avance DPX400 proton NMR operating at 400 MHz. 1H NMR spectra were obtained for 1% copolymer solutions in deuterated chloroform or dimethyl sulfoxide. Figure 2a shows an example of the 1H NMR spectra of comb C1, dominated by the presence of the ether protons at ∼3.6 ppm and the backbone methacrylate protons at 0.5-2 ppm. Strong signals at 2.5 and 3.3 ppm are DMSO and water protons, respectively. Molecular weights of the comb were determined using a Waters Associates gel permeation chromatography-laser light scattering system comprised of a model 510 pump, model 410 differential refractometer, miniDawn laser light scatterer, and two linear Styragel columns connected in series. Filtered copolymer solutions (0.004 g/mL) in THF were eluted at 30 °C. Absolute molecular weights were

Figure 2. NMR spectra of comb polymers: (a) base C1 comb structure; (b) carboxylated C1; (c) NHS-activated C1.

obtained using the comb polymer dn/dc calculated according to30 dn/dc ) (ncomb - nTHF)/Fcomb

(1)

where ncomb is the refractive index of the comb polymer, nTHF is the refractive index of the solvent, and Fcomb is the mass density. The polymer refractive index was measured by ellipsometry (≈1.491), Fcomb was estimated by group contribution methods30 as ≈1.09 g/cm3, and the value for nTHF used was 1.407.31 This gives a value of dn/dc ) 0.077 cm3/ g, intermediate between the values for PEO (0.068) and PMMA (0.089) in THF.32 Yields of subsequent functionalization reactions were calculated from elemental analysis performed on samples by Quantitative Technologies, Inc. Carboxylation of Comb Polymer. Carboxylation of the comb polymer was carried out using a modification of the procedure of Storey and Hickey.33 The following represents a typical synthesis: 8 g of comb polymer (0.00889 mol of -OH) and 4 g of succinic anhydride (0.0889 mol) were added with magnetic stir bar to a hot 500-mL round-bottom

88


Irvine et al. Table 2. RGD-Functionalized Comb Polymer Physical Data

Figure 3. Chemistries used for modification of comb polymers. Products in each step are (1) carboxylated comb, (2) NHS-activated comb, and (3) RGD-comb.

flask removed from a drying oven at T ∼ 150 °C. The reactor was capped with a rubber septum and purged with nitrogen until cool. Anhydrous dichloroethane (200 mL) was cannulated into the flask. The copolymer was observed to quickly dissolve while the succinic anhydride remained suspended in the solvent. The mixture was degassed for 15 min by bubbling nitrogen, then 48 µL of N-methylimidazole (NMIM) was added dropwise to the reactor. Upon addition of NMIM, the solution rapidly became clear. The flask was connected to a condenser and refluxed 15 h at 65 °C. The carboxylated comb polymer was separated from unreacted succinic anhydride and NMIM by concentrating in a rotovaporator, precipitating in petroleum ether, redissolving in THF, and precipitating in 5% aqueous HCl. The polymer was washed 18 h by stirring in 5 vol % aqueous HCl, recovered by filtration, and dried at 60 °C in vacuo. Functionalization was confirmed by elemental analysis (>95 mol %) and by the introduction of a peak due to carboxylate protons at ∼12 ppm as shown in Figure 2b. NHS Activation of Carboxylated Combs. Carboxylic acid groups of the comb polymer were activated using N-hydroxysuccinimide and carbodiimide, following a modification of the procedure of Jo and Mikos.34 The following is a representative synthesis: 4.75 g of carboxylated comb, 0.526 g of N-hydroxysuccinimide, and 45 mL of dichloromethane (DCM) were added to a 100-mL round-bottom flask. DCC (0.942 g) was dissolved in 5 mL of DCM and immediately added dropwise to the stirring comb/NHS/DCM mixture. The mixture was stirred for 12 h at room temperature. The NHS-activated polymer was purified by precipitating in anhydrous ethyl ether three times. NHS-activated comb polymer was stored at -20 °C until used. Introduction of the NHS ester was indicated in NMR by the disappearance of the carboxylate proton signal and introduction of the -CH2CH2- signal of NHS at 2.5-3 ppm. RGD Coupling to Activated Combs. Integrin ligands GRGDSP, GRGDSPK, and the nonsense ligand GRGESP were covalently linked via an amide bond to NHS-activated side chains, as shown in Figure 3. Coupling was achieved by reaction of peptides with NHS-activated comb in DMF/ water mixtures using a modification of the procedure of Jo and Mikos.34 In a typical reaction, 6.2 mg of GRGDSP was dissolved in 2 mL of PBS. A 100-mg portion of NHSactivated comb copolymer (4.5× excess carboxylate groups)

polymer

base material

peptide linked

µg of peptide/ mg of polymer

peptides/ molecule

C1-RGD1 C2-RGD1 C2-RGD2 C2-RGD3 C2-RGD4 C1-RGE1

C1 C2 C2 C2 C2 C1

GRGDSP GRGDSPK GRGDSPK GRGDSPK GRGDSPK GRGESP

50.9 12.4 16.0 26.6 40.0 37.9

2.10 1.69 2.18 3.62 5.44 1.6

was dissolved in 2 mL of anhydrous DMF, and the mixture was added dropwise to the stirring peptide solution at 4 °C. As the reaction progressed, a milky emulsion formed. The reaction was allowed to proceed with stirring at 4 °C for 12 h, after which the solvent was evaporated under low heat (T ) 35-40 °C) in a chemical hood. The recovered polymer was separated from any remaining free peptide and hydrolyzed NHS by dialysis (Pierce Slide-A-Lyzer, 3500 g/mol molecular weight cutoff) against 50/50 (v/v) water/ethanol for 3 days with periodic changes of the medium. Finally, the purified copolymer was recovered by evaporating the dialyzed solution under low heat (T ) 35-40 °C) and subsequently drying the product in vacuo at 25 °C for 24 h. RGD-comb was stored at 25 °C until use. Variations in the total RGD content were obtained by changing the ratio of peptide to polymer during the solution coupling reaction. Solution coupling of GRGESP and GRGDSPK was carried out in a similar manner. A series of combs with varying peptide cluster size (# peptides/comb) was prepared; Table 2 lists the physical data for each polymer, calculated from elemental analysis results. To demonstrate flexibility in ligand coupling to comb polymer surfaces, two alternative coupling approaches were also employed. GRGDSP peptides were coupled to the surface of NHS-activated comb (C1) films cast on tissue culture polystyrene (TCPS) dishes by first dissolving the peptide at 1.0 mg/mL in pH 7.4 PBS followed by incubation of the solution over the surface at 4 °C for 6 h. The surfaces were rinsed twice briefly in PBS and then stored in PBS at 4 °C until use. GRGESP was coupled to NHS-activated C1 surfaces in a similar manner. Controls for the nonspecific adsorption of peptides to comb surfaces were prepared by following the above procedures using unactivated carboxycomb films. As a third strategy, RGD was tethered to comb surfaces through in situ activation and peptide coupling to carboxyl groups at comb polymer film surfaces. For peptide linking, 9 mg of EDC and 15 mg of NHS were dissolved in 4 mL of PBS and immediately applied (1 mL per surface) to films precast on TCPS dishes. Surfaces were left in contact with the solution for 20 min at room temperature, followed by 2× rinsing with 1 mL of PBS. One milliliter of GRGDSP solution (0.5 mg/mL in PBS) was immediately applied to each surface at 4 °C for 6 h. Each surface was subsequently rinsed twice with PBS for 20 min per wash and then stored under PBS at 4 °C until used. Thin Film Preparation. Cell culture substrates were 30 mm glass dishes (Kontes) or tissue culture polystyrene (TCPS) (Falcon 3047 or 3230 plates or 35-mm dishes).


Single-crystal silicon substrates were used for ellipsometric measurements on comb films. Comb polymer thin films were prepared by spin-coating 0.01 g/mL water/ethanol (50/50 vol/ vol) solutions of the comb copolymer on substrates at 1000 rpm and drying in vacuo for 24 h at 25 °C, providing films ∼1500 Å thick. Films for fluorescent labeling of surface peptide clusters were prepared in 96-well opaque PS spectrophotometer plates (Corning) by solvent casting (0.002 g/mL solutions of comb polymer in 50/50 (vol/vol) water/ ethanol) onto surfaces, covering the plates, and drying at room temperature for 5 h followed by drying in vacuo at 25 °C for 24 h, to obtain films ∼5000 Å thick. Blends of RGDmodified combs and unmodified combs were prepared by codissolving the polymers in water/ethanol and casting in a similar manner. Measurements of film thickness and refractive index were made using a Gaertner ellipsometer operating at 633 nm wavelength. Bulk Swelling Measurements. Measurements of water uptake by bulk P(MMA-r-POEM) samples were made to assess the swelling of comb polymers in aqueous environments. Three samples of C2 comb polymer were compression molded at 125 °C into 0.5 mm thick disks 25 mm in diameter. Samples were immersed in 150 mL of filtered deionized water at 37 °C and weighed periodically over 7 days. The measured equilibrium water content, reached between 24 and 48 h, was 16 ( 3.1 wt %. This can be compared with the swelling of pure PMMA samples prepared in a similar manner, which contained 5.3 ( 1.1 wt % water. Peptide Cluster Labeling with Fluorescent Nanospheres. The relationship between RGD surface cluster density and bulk RGD-comb content in comb/RGD-comb blend thin films was determined by covalently linking fluorescent nanospheres bearing surface aldehyde groups to the -amine of surface-accessible peptide lysines. Stock solutions of the nanospheres suspended in pH 7.4 HEPES buffer were prepared (2.0 × 1013 nanospheres/mL) and stored at 4 °C in darkness until used. Coupling of the nanospheres to peptide-presenting surfaces was performed by sonicating the stock solution for 1 h and then applying 625 µL/cm2 spheres and 62.5 µL/cm2 sodium cyanoborohydride (0.033 g/mL in deionized water) to each surface to create permanent alkylamine bonds. Samples were subsequently stored at 4 °C in the dark for 24 h to allow condensation of the nanospheres with the surface peptide. Removal of excess nanospheres and nonspecifically adsorbed particles was accomplished by rinsing surfaces three times with 625 µL/ cm2 Tween 20 solution (0.3 vol % in PBS), 5 min per wash, followed by one wash with deionized water. Measurements of total fluorescence from surfaces after further rinsing treatments showed no further label loss. Determination of surface RGD cluster densities was made by measurement of total fluorescence from labeled surfaces for a range of bulk RGD-comb concentrations in a Molecular Devices SpectraMax Gemini fluorescence plate reader (excitation at 488 nm, emission at 520 nm). Fluorescence measurements were converted to nanosphere surface densities by comparing readings with standard curves prepared from nanospheres at various dilutions in a 96-well plate. Nonspecific adsorption of nanospheres on film defects was accounted for by


measuring the fluorescence of control comb surfaces presenting no ligand and subtracting this baseline signal from that measured on RGD surfaces. As the nanospheres were present in large excess (∼1000×) to the surface peptides for extended time periods during coupling, labeling for surface cluster spacings greater than the nanosphere diameter is expected to be quantitative. Qualitative examination of the ligand distribution on polymer surfaces was made by fluorescence microscopy on labeled films using a Zeiss 35 inverted fluorescence microscope. Cell Culture. All cell culture reagents were obtained from Gibco unless otherwise noted. WTNR6 cells, a cell line derived from NIH 3T3 cells, were obtained from Alan Wells at the University of Pittsburgh. This cell line is prepared from mouse fibroblast 3T3 cells, lacking endogenous epidermal growth factor receptor (EGFR), which have been transfected to express human EGFR.35,36 Fluorescence-activated cell sorting (FACS) analysis has previously demonstrated that these cells express Rvβ3 and R5β1 integrin receptors,37 which are known to bind the RGD amino acid sequence.38 Cells were cultured at 37 °C under 5% CO2 in Modified Eagles’ Medium-R, supplemented with 7.5% fetal bovine serum (FBS), 1% nonessential amino acids (10 mM), 1% sodium pyruvate (100 mM), 1% L-glutamine (200 mM), 1% penicillin (10 000 U/mL), 1% streptomycin (10 mg/mL), and 1% Geneticin antibiotic (35 mg/mL). Cells were grown near confluence in T75 flasks and then suspended with trypsinEDTA solution (10×) and passaged every 3-5 days. Cell passages 16-26 were used for cell attachment experiments. Cell Attachment Assay. Surfaces for cell culture were sterilized by exposure to a UV lamp 30 min prior to use. Fibroblasts to be seeded on samples were grown near confluence in T75 flasks and then suspended using trypsinEDTA. Cell concentrations were determined using a Coulter cell counter and diluted appropriately to seed surfaces. TCPS was used as a control substrate in all cell experiments to monitor for irregularities in media or cell passages. Cell attachment was assayed by culturing cells in contact with surfaces for 24 h followed by counting the fraction which adhered. Cell numbers on polymer surfaces were quantified by direct manual counting of cells by phase contrast microscopy or by using a fluorescent DNA probe assay. Manual cell number determination was performed by counting cells present in five fields at 100× in a phase contrast microscope on three samples. Cells counted in this manner are reported as number of cells/field or percent adhered relative to a TCPS control surface. Alternatively, quantification of total cell number on surfaces was carried out using a DNA-binding assay (CyQuant, Molecular Probes). After culturing cells on samples for the desired time, media were aspirated and surfaces were rinsed with 200 µL of PBS. Samples were then immediately frozen at -70 °C for 18 h. Samples were then thawed at 25 °C for 1 h, and assay solution was prepared with cell lysis buffer (20×) and GR DNA dye (200×) in deionized water. Cells on surfaces were lysed by applying 700 µL of assay solution for 15 min, and fluorescence from the plate(s) was read immediately using a Molecular Devices Spectramax Gemini spectrophotometer with excitation at 480 nm and emission read at 520 nm. Total

90


Irvine et al.

fluorescence was converted to cell number in each well using a standard conversion curve prepared from Coulter counter measurements on cell standards of the same cell passage. Cell number, surface densities, and percent seeded cells adhered are reported as mean values of three to five samples ( standard error. Results and Discussion Resistance of Comb Polymer Films to Nonspecific Cell Adhesion. To control cell adhesion exclusively through RGD-integrin interactions, P(MMA-r-POEM) surfaces completely resistant to nonspecific cell attachment are required. Therefore the effect of comb architecture and composition on the resistance of P(MMA-r-POEM) surfaces to cell adhesion was studied. First, the effect of PEO side chain length was examined. To maintain water insolubility, the total POEM content of combs was fixed at 45 wt %. At a fixed maximum POEM content, the comb could be prepared with a few long side chains or many short ones. However, use of high molecular weight PEO side chains to eliminate protein adsorption may interfere with effective signaling by ligand tethered to PEO chain ends. SCF models predict that lowdensity, high molecular weight grafted polymer brushes have chain ends distributed throughout the grafted layer;39 this might cause ligands to be screened from cell receptors. Short PEO tethers would be preferable to keep ligands spatially well-localized and prevent steric shielding of the peptide by neighboring side chains. In addition, the surface densities obtained for long side chains may provide only “kinetic protein resistance” and, thus, after long times allow adsorption to occur.40 Thus we limited our studies to combs with relatively short side chains (e45 repeat units). Cell resistance of P(MMA-r-POEM) films as a function of side chain length is shown in Figure 4a. Cell adhesion is essentially eliminated (relative to highly adhesive TCPS substrates) for side chains with g9 EO repeat units; because a short tether length is desirable, nine-unit POEM and six-unit HPOEM were used for further studies. For a given PEO side chain length, increasing the molar fraction of macromer units in the comb polymer increases the surface density of the side chains and thus should improve the resistance of surfaces to nonspecific cell attachment. The PEO content of comb polymers necessary for cell resistance for a fixed side chain length of nine EO units was assessed for a series of P(MMA-r-POEM) copolymers containing 20, 30, or 45 wt % POEM. Figure 4b shows that cell attachment to comb films in the presence of serum was effectively eliminated when the weight fraction of POEM units was 30% or greater. On the basis of these results, further studies were carried out using two carboxylated comb polymers of cell-resistant composition (45 wt % nine-unit POEM/six-unit HPOEM), C1 and C2. Carboxylation of the HPOEM units, necessary for tethering ligands by NHS chemistry, did not notably affect the cell resistance of comb films. Example phase contrast micrographs of cell attachment after 24 h to pure PMMA, TCPS, and carboxylated C1 are shown in Figure 5. The comb surfaces are highly cell resistant; rare cells found

Figure 4. Cell adhesion to P(MMA-r-POEM). WTNR6 fibroblasts (20 000 cells/cm2) were seeded on TCPS or films of comb polymers. At 24 h postseeding, cells adhered were counted manually, and data are plotted as fraction of cells adhered relative to TCPS. (a) Cell adhesion to combs bearing different numbers of ethylene oxide units (n) in the side chains. (b) Cell attachment to comb comprised of varied weight fractions of POEM (nine EO unit side chains).

Figure 5. WTNR6 fibroblasts were seeded (20 000 cells/cm2) on carboxylated C1, pure PMMA, and TCPS; shown are phase contrast micrographs (100×) of surfaces 24 h postseeding in serum.

on these surfaces always display a rounded, unspread morphology. Modeling of Comb Surface Structure. To shed light on why such comb polymers, containing only short PEO side chains and up to 70 wt % hydrophobic MMA units, are highly resistant to nonspecific cell adhesion, numerical modeling of the comb/water interface was performed. The predicted equilibrium concentration profiles for a film of C1 comb polymer in contact with water are shown in Figure 6. Taking 5 Å as the layer thickness (equal to the segment size),



Figure 7. Possible chain configurations at the interface of comb polymer films and water. Peptides attached to the comb are denoted by gray spheres, and chains contributing segments to the surface layer are highlighted by heavy lines. (Top) Quasi-confinement of comb at the surface. (Bottom) Case of combs maintaining a 3D coil conformation at the surface.

Figure 6. SCF prediction of comb polymer surface structure in water. (a) Component volume fraction profiles: (2) water; (O) comb side chains; (9) comb backbone. (b) Segment distributions relative to their bulk (φb) values: (O) comb side chains; (2) branch points; (9) other backbone units. Only the interfacial region of the calculated concentration profiles is shown.

SCF predicts that the outermost 10 Å of the polymer surface is composed almost entirely of water and PEO units. Due to the favorable interaction of the comb side chains with water, the backbone of the comb is preferentially localized within two layers adjacent to these outermost layers. Hydrophilic segments and water are depleted from these two layers to accommodate the backbone by extending the side chains into solution. These concentration profiles are consistent with alignment of the comb backbone parallel to the interface, minimizing the backbone/water contacts while maximizing favorable PEO-H2O contacts as depicted in Figure 7. It should be noted that this simple lattice model does not allow explicit accounting of the entropic nature of the hydrophobic effect at the water/polymer interface, which might further enhance the segregation of side chain and backbone units. Given the predicted arrangement of combs in the top layer of the film, a rough estimate of the surface density of 9-mer PEO chains at the water/C1 interface can be obtained by assuming a sharp interface between hydrated side chains anchored to an underlying layer composed only of backbone units. Then the surface density σ (PEO chains/area) is σ ) xPEO/a2

(2)

where xPEO is the mole fraction of POEM macromers in the comb (xPEO ≈ 0.15) and a2 is the area occupied by a backbone segment. Taking a2 ≈ 25 Å2, the surface density of side chains would be ∼0.6 chains/nm2, or ∼13 Å between PEO chains anchored at the surface. The fully extended

length of these very short side chains is ∼18 Å; thus the PEO appears to be providing protein resistance without being strongly stretched in the surface layer (strong stretching occurs when the distance between anchors is much less than the RG of the tethered chains41). On the basis of these simple calculations, the mean distance between the side chains drops from ∼22 Å at 20 wt % nine-unit POEM to ∼17 Å at 30 wt % POEM. Complete cell resistance is thus obtained when the apparent mean distance between side chains at the interface becomes less than the fully extended length of the short PEO teeth. The quasi-2D alignment of combs at the interface would have several important consequences for the surface properties of these films. First, perpendicular to the interface, segregation of the comb segments into a side-chain-rich swollen layer atop a backbone-rich solvophobic layer could explain the observed high cell resistance, even for comb compositions containing 60-70 wt % hydrophobic segments. The PEO teeth form a thin but highly hydrated brush that resists protein adsorption. Such results are consistent with previous studies, which demonstrated that chains with only a few ethylene oxide units can provide excellent protein resistance at sufficiently high grafting densities.12 Second, the surface-layer molecules will be self-segregated, organizing roughly as close-packed “disks”, as depicted in the cross section in Figure 7a. In the bulk, the number of chains interpenetrating a given coil volume of size ∼RG3 scales as N1/2. However, chains confined to two dimensions should be only weakly overlapping each other, failing to interpenetrate substantially even at bulk density.42,43 A third consequence of this 2D organization is an expanded lateral distribution of chains relative to a 3D coil in bulk. The radius of gyration (RG) of a polymer chain in the film interior is44 RG ≈ N1/2a(φbbb)-1/86-1/2

(3)

where N is the number of segments in the chain, φbbb is the bulk volume fraction of backbone segments (accounting for the swelling of chains by water in the bulk), and a is the segment size. For comb C2, taking N ) 600, φbbb ) 0.58 (determined simply by the composition of the comb), and a ) 6.4 Å yield an estimate for RG ≈ 68 Å in bulk.32 Confined

92


Irvine et al.

Figure 8. Cell adhesion after 24 h on RGD-presenting surfaces prepared by different methods. WTNR6 cells (30 000 cells/cm2) were seeded in serum-containing media on RGD surfaces prepared by several routes.

at the interface, the comb backbone takes on larger dimensions. The ratio of the 2D RG to the bulk RG should be RG,2D/RG ≈ C(φbbs)-1/2/(φbbb)-1/8

(4)

where C is a constant (approximately 1.1-1.45 according to simulations42,43,45,46) and φbbs is the volume fraction of backbone segments at the surface (φbbs ) 0.67 from SCF). The radius r of disks of the self-segregated molecules in the surface layer is related to RG,2D by42 r ) 21/2RG,2D

(5)

From the bulk RG estimate and eqs 4 and 5, the diameter of comb “disks” in the surface layer is expected to be ≈320 Å. This molecular arrangement at the interface should greatly influence the presentation of clustered ligand by comb polymer, as depicted in Figure 7. Were the chains organized as three-dimensional coils at the interface (Figure 7b), their small areal contribution at the surface and extensive coil interpenetration would prevent single-comb molecules from presenting ligands effectively; a comb bearing several peptides might only localize a single ligand (on average) accessible to receptors in the top surface layer. Semiconfinement at the interface, on the other hand, makes every ligand on a functionalized molecule in the surface layer potentially accessible. Clustered RGD Surfaces. RGD clustering at comb polymer surfaces was effective for inducing cell adhesion, whether peptide was coupled to comb polymers in solution before film casting, to the surface of chemically preactivated films, or to in situ activated surfaces. Figure 8 compares the number of fibroblasts attached to GRGDSP-bearing surfaces prepared by each of these routes. Cell adhesion to combs prepared by coupling peptide to film surfaces was mediated by covalently tethered RGD and not adsorbed peptide, as indicated by the lack of cell adhesion to nonactivated controls exposed to RGD solution. RGD surface densities obtained by coupling peptide to in situ activated surfaces were ∼30% lower than those obtained by the other two methods (as assessed by fluorescence labeling); however, cells attached and spread on all surfaces. The surface-coupling routes are useful for limiting the amount of precious peptides used, by confining the tethered ligand to the exterior surface of the

film and allowing reuse of the peptide solution if desired. For example, we might consider using the comb polymer as a biomaterial coating with nominal thickness 3 000 Å. A celladhesive coating of C1-RGD1 polymer will use ∼1 × 107 peptides/µm2, most of which are buried in the film. Approximately 1000-fold less RGD tethered by the surfacecoupling routes would provide a similar level of adhesion. This economy is traded for, however, in robustness of the signaling layer (worn surface layers would exhibit no signaling) and extra surface processing steps. RGD-combs having a range of cluster sizes (number of peptides/comb) were prepared by the solution precoupling method for use in further experiments, as listed in Table 2. To control cell adhesion via tethered RGD, it is important that the ligand itself does not promote protein adsorption. To verify this, cell attachment to C1-RGD1 and C1-RGE1 was compared, the latter presenting an inactive analogue of RGD. These results for surfaces prepared by the solution precoupling approach are also shown in Figure 8. Cells adhere and spread on C1-RGD1 but are unable to attach to C1-RGE1 despite culturing in the presence of 7.5% serum. The specificity of interaction between cells and RGD anchored at the surface was further probed by adding 85 µM soluble GRGDSP to the media of cells that had been allowed to attach for 24 h. Cells immediately showed significant rounding and within 4 h completely detached from the surface, indicating that integrins responsive to RGD were the adhesion-promoting cell-surface linkage. Thus cell attachment even on surfaces of very high RGD density (∼32 nm on average between RGD clusters, as discussed below) in the presence of serum is controlled by tethered RGDintegrin interactions. The total surface density of RGD presented on comb polymer films can be controlled by blending RGD-combs with their nonfunctionalized counterparts. Cell-accessible RGD cluster densities at the surface of comb/RGD-comb mixtures were measured by labeling surface ligand clusters with ∼30 nm diameter fluorescent polystyrene nanospheres and measuring the total fluorescence from surfaces. For this analysis, several assumptions were made. First, the nanospheres were assumed to access surface-localized RGD in a manner comparable to cell integrin receptors; to this end, a large label with a diameter comparable to integrin heterodimers was used. Second, it was assumed that each ligand cluster (presented by a single comb molecule) was labeled by only one nanosphere and that each nanosphere labeled only one cluster. This experiment provided both a practical measure of the total cluster density at the surface and an experimental test of the quasi-2D surface conformation implied by the SCF model. If the combs are not strongly confined at the surface but instead mimic the bulk conformation, multiple chains will contribute to the local RGD density at any given surface location due to interpenetration of coils. More specifically, one might expect a monolayer coverage of labels for any mixture of RGD-modified and unmodified combs where more than one RGD-comb is present per ∼N1/2 molecules; i.e., when more than one RGD-comb is present in the volume spanned by a single comb molecule in bulk, ∼RG3 (as in Figure 7b). For the C2 comb polymer, N ≈ 600,



blends. Also shown is a linear best fit to these data, obtained for Fclus ) (9.768 × 1011)φbRGD-comb (number/cm2). This surface density is replotted in Figure 9b as an average cluster spacing on the surface dclus ) (Fclus)-1/2. For the case of comb/ RGD-comb blends where the top molecular layer is confined at the interface, the expected average cluster spacing on the surface is dclus ) dmin/φbRGD-comb-1/2, where dmin is a minimum cluster spacing obtained for films of the pure RGD-comb polymer. The data are well fit using this relationship (χ2 ) 0.792). The best-fit curve plotted in the figure has dmin ) 32 ( 0.98 nm. The labeling data strongly support the hypothesis that the comb polymer is quasiconfined at the water-polymer interface. The measured cluster separation is inconsistent with a 3D coil arrangement of comb molecules at the interface, as surface labeling is not saturated for φbRGD-comb . 0.04. Further, the measured minimum cluster separation is in agreement with the calculated RG of 2D combs at the surface. Also plotted in Figure 9b are data from blends using C2-RGD4, where the number of peptides per comb is 5.4. The same overall cluster distribution is obtained; i.e., the number of peptides per cluster is changed but the separation of clusters at the surface for a given amount of RGD-comb in the film is the same, as expected if changes in functionality of the comb polymer do not significantly change their expression at the surface. On the basis of these results, the average surface density of ligand vs bulk concentration for comb/RGD-comb blends with RGD-combs of different cluster size can be derived, as plotted in Figure 9c. Shown is the estimated range of accessible ligand densities for the four C2-RGD materials prepared. The accessible peptide density can be readily adjusted 2 orders of magnitude higher or lower than RGD densities reported in the literature to support fibroblast attachment and focal contact formation.47 Conclusions

Figure 9. (a) Ligand cluster surface density vs bulk RGD-comb concentration in C2/C2-RGD3 blends. Solid line is a best fit to the data Fclus ) 9.768 × 1011(φbRGD-comb) (number/cm2). (b) RGD cluster spacing from cluster surface density data. Shown are results from two different cluster sizes: ([) 3.6 RGD/comb (C2-RGD3), (9) 5.4 RGD/comb (C2-RGD4). Solid line is the best fit to the data dclus ) dmin/φRGD-comb-1/2, where dmin ) 32 ( 0.98 nm. (c) Total RGD surface densities obtained as a function of bulk RGD-comb concentration in comb/RGD-comb blends from cluster density data. Accessible surface densities derived for four different RGD-comb cluster sizes are shown: 1.7 RGD/comb (C2-RGD1), 2.2 RGD/comb (C2RGD2), 3.6 RGD/comb (C2-RGD3), and 5.4 RGD/comb (C2RGD4).

meaning that φbRGD-comb > 0.04 would saturate surface labeling. Variations in label density would only be expected at extremely low RGD-comb contents in the film. In contrast, confinement of the comb at the interface (Figure 7a) would mean that large fractions of RGD-comb would be necessary to decrease the cluster separation; in this case the cluster separation would scale as (φRGD-comb)-1/2. Figure 9a shows the RGD cluster surface density Fclus measured by nanosphere labeling as a function of the bulk volume fraction of C2-RGD3 (φbRGD-comb) in C2/C2-RGD3

A wide variety of biomedical applications would best be served by materials that guide specific cell responses by combining protein resistance with tethered ligand signaling.1-4,6,7 The comb polymers described here would have definite advantages in practical use outside a laboratory environment, being inexpensive to prepare and compatible with existing industrial coating processes.1,48 The solubility of these amphiphilic molecules in water/alcohol mixtures also eliminates the need for environmentally unsound organic solvents during processing. Clustered ligand presentation by comb polymers offers a powerful tool for engineering cell responses to surfaces. Cluster spacing, number of peptides/cluster, and total ligand density are easily controlled by varying the amount of ligandmodified comb used, comb molecular weight, the degree of ligand functionalization, and the combination of these variables. Use of short side chain tethers allows the immobilized ligand to sample a range of conformations, while maintaining a highly surface-localized chemical signal. Numerical SCF calculations of the structure of comb polymer/water interfaces predict preferential alignment of the comb backbone at the surface, maximizing PEO-water

94


contacts in the top surface layer; our experimental results for comb polymer films are consistent with this picture. Alignment of the comb at the interface apparently (1) segregates the comb teeth into a short hydrophilic brush atop an anchoring backbone layer and (2) arranges the surface layer molecules as weakly interpenetrated disks. This quasi2D conformation leads to highly efficient nanoscale clustering of ligand at the surface: each ligand attached to a given comb molecule in the surface layer can potentially be presented without steric hindrance from subsurface layers, as the surface is molecularly arranged to allow maximum exposure of the PEO side chains. Acknowledgment. We gratefully acknowledge Professors M. F. Rubner and A. Wells for access to the contact angle appartus and for providing the WTNR6 cell line, respectively. This work was supported by Whitaker Foundation Grant No. RG-97-0196, NIH Award Number 1R0GM59870-01, and National Science Foundation Award No. BES-9632714. References and Notes (1) Banerjee, P.; Irvine, D. J.; Mayes, A. M.; Griffith, L. G. J. Biomed. Mater. Res. 2000, 50, 331. (2) Drumheller, P. D.; Elbert, D. L.; Hubbell, J. A. Biotech. Bioeng. 1994, 43, 772. (3) Bearinger, J. P.; Castner, D. G.; Healy, K. E. J. Biomater. Sci., Polym. Ed. 1998, 9, 629. (4) Hubbell, J. A.; Massia, S. P.; Desai, N. P.; Drumheller, P. D. Bio/ Technology 1991, 9, 568. (5) Kuhl, P. R.; Griffith-Cima, L. G. Nat. Med. 1996, 2, 1022. (6) Maheshwari, G.; Brown, G.; Lauffenburger, D. A.; Wells, A.; Griffith, L. G. J. Cell Sci. 2000, 113, 1677. (7) Patel, N.; Padera, R.; Sanders, G. H. W.; et al. FASEB J. 1990, 12, 1447. (8) Tong, Y. W.; Shoichet, M. S. J. Biomater. Sci., Polym. Ed. 1998, 9, 713. (9) Ratner, B. D. Nature 1999, 398, 593. (10) Black, F. E.; Hartshorne, M.; Davies, M. C.; et al. Langmuir 1999, 15, 3157. (11) Lee, J. H.; Kopeckova, P.; Kopecek, J.; Andrade, J. D. Biomaterials 1990, 11, 455. (12) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, 10714. (13) Harris, J. M.; Poly(ethylene glycol) Chemistry: Biotechnical and Biomedical Applications; Plenum Press: New York, 1992. (14) Irvine, D. J.; Mayes, A. M.; Satija, S. K.; Barker, J. G.; Sofia-Allgor, S. J.; Griffith, L. G. J. Biomed. Mater. Res. 1998, 40, 498. (15) Park, K. D.; Kim, W. G.; Jacobs, H.; Okano, T.; Kim, S. W. J. Biomed. Mater. Res. 1992, 26, 739. (16) Tseng, Y. C.; Park, K. J. Biomed. Mater. Res. 1992, 26, 373. (17) Amiji, M.; Park, K. J. Biomater. Sci., Polym. Ed. 1993, 4, 217. (18) Sheu, M. S.; Hoffman, A. S.; Terlingen, J. G. A.; Feijen, J. Clin. Mater. 1993, 13, 41.

Irvine et al. (19) Burridge, K.; Fath, K.; Kelly, T.; Nuckolls, G.; Turner, C. Annu. ReV. Cell Biol. 1988, 4, 487. (20) Burridge, K.; Chrzanowska-Wodnicka, M. Annu. ReV. Cell DeV. Biol. 1996, 12, 463. (21) Shaw, L. M.; Messier, J. M.; Mercurio, A. M. J. Cell Biol. 1990, 110, 2167. (22) Miyamoto, S.; Teramoto, H.; Coso, O. A.; Gurkind, J. S.; Burbelo, P. D.; Akiyama, S. K.; Yamada, K. M. J. Cell Biol. 1995, 131, 791. (23) Scheutjens, J. M. H. M.; Fleer, G. J. J. Phys. Chem. 1979, 83, 1619. (24) Evers, O. A.; Scheutjens, J. M. H. M.; Fleer, G. J. Macromolecules 1990, 23, 5221. (25) Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. M. H. M.; Cosgrove, T.; Vincent, B. Polymers at Interfaces; University Press: Cambridge, U.K., 1993. (26) Schantz, S. Macromolecules 1997, 30, 1419. (27) Utracki, L. A. Polymer Alloys and Blends: Thermodynamics and Rheology; Hanser Publishers: New York, 1990. (28) Barton, A. F. M. CRC Handbook of Polymer-Liquid Interaction Parameters; CRC Press: Boca Raton, FL, 1990. (29) Dennis, J. E., Jr.; Schnabel, R. B. Numerical methods for unconstrained optimization and nonlinear equations; Prentice-Hall: Englewood Cliffs, NJ, 1983. (30) Van Krevelen, D. Properties of Polymers: Their Correlation With Chemical Structure, Their Numerical Estimation and Prediction from Group Contributions; Elsevier: New York, 1990. (31) S.-A. C. Company Product Index, 1999-2000. (32) Brandrup, J.; Immergut, E. H. Polymer Handbook; Wiley: New York, 1989. (33) Storey, R. F.; Hickey, T. P. J. Polym. Sci., Part A: Polym. Chem. 1993, 31, 1825. (34) Jo, S.; Mikos, A. G. ACS Polym. Prepr. 1999, 40, 183. (35) Chen, P.; Gupta, K.; Wells, A. J. Cell Biol. 1994, 214, 547. (36) Pruss, R. M.; Herschman, H. R. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 3918. (37) Maheshwari, G.; Wells, A.; Griffith, L. G.; Lauffenburger, D. A. Biophys. J. 1999, 76, 2814. (38) Ruoslahti, E.; Pierschbacher, M. D. Science 1987, 238, 491. (39) Irvine, D. J.; Mayes, A. M.; Griffith-Cima, L. Macromolecules 1996, 29, 6037. (40) Szleifer, I. Biophys. J. 1997, 72, 595. (41) Alexander, S. J. Phys. (Paris) 1977, 38, 983. (42) Carmesin, I.; Kremer, K. J. Phys. (Paris) 1990, 51, 915. (43) Baumga¨rtner, A. Polymer 1982, 23, 334. (44) Flory, P. J. Principles of Polymer Chemistry; Cornell University Press: Ithaca, NY, 1953. (45) Paul, W.; Binder, K.; Heermann, D. W.; Kremer, K. J. Phys. II 1991, 1, 37. (46) Nelson, P. H.; Hatton, T. A.; Rutledge, G. C. J. Chem. Phys. 1997, 107, 1269. (47) Massia, S. P.; Hubbell, J. A. J. Cell Biol. 1991, 114, 1089. (48) LaPorte, R. J. Hydrophilic Polymer Coatings for Medical DeVices: Structure/Properties, DeVelopment, Manufacture and Applications; Technomic Publishing Co.: Lancaster, PA, 1997.

BM005584B

Nanoscale Clustering of RGD Peptides at Surfaces Using Comb

Recommend Documents