P(AAm-co-EG) Interpenetrating Polymer Networks Grafted to Oxide ...

Langmuir 1997, 13, 5175-5183

5175

P(AAm-co-EG) Interpenetrating Polymer Networks Grafted to Oxide Surfaces: Surface Characterization, Protein Adsorption, and Cell Detachment Studies J. P. Bearinger,†,‡ D. G. Castner,§ S. L. Golledge,§ A. Rezania,†,‡ S. Hubchak,| and K. E. Healy*,†,‡ Division of Biological Materials, Northwestern University Dental School, Chicago, Illinois 60611, Department of Biomedical Engineering, Northwestern University, Evanston, Illinois 60208, Department of Pathology, Northwestern University Medical School, Chicago, Illinois 60611, and NESAC/Bio, Department of Chemical Engineering and Center for Bioengineering, University of Washington, Box 351750, Seattle, Washington 98195 Received February 3, 1997. In Final Form: June 27, 1997X We investigated the composition, properties, and utility of a novel copolymer of P(AAm-co-EG) designed to be an adaptable, durable, and biocompatible surface treatment of metallic, polymeric, and ceramic materials. Solution deposition and photoinitiation reactions were employed to graft a silane layer and then two sequential polymer layers (a discontinuous two stage polymerization) onto oxide surfaces. Different solvents, polymer concentrations, and cross-linker concentrations in the top polymer layer were compared. Contact angle measurements, spectroscopic ellipsometry, and X-ray photoelectron spectroscopy were used to characterize layer wettability, thickness, and chemistry, respectively. A sandwich type network formed between acrylamide and poly(ethylene glycol) when acetone was used as the solvent for both layers. In contrast, an interpenetrating polymer network between acrylamide and poly(ethylene glycol) formed when acetone and methanol were used as the solvents for polymerization of the acrylamide and poly(ethylene glycol) layers, respectively. Interpenetrating polymer network configured samples were tested for protein adsorption and strength of cell attachment. Protein adsorption experiments in 15% fetal bovine serum indicated that significant amounts of protein do not adsorb to the surface of the thin polymer films (∼20 nm). Cell detachment experiments indicated that cells contacting copolymer-modified surfaces were removed by lower shear stresses than cells contacting clean and amine-terminated, (N-(2-aminoethyl)-3-aminopropyl)trimethoxysilane modified surfaces.

Introduction In the fields of biotechnology and biomaterials products such as sensors, diagnostic tests, commodity polymers, and biomedical devices are often modified so that materials exhibit interfaces with improved performance compared to the original bulk materials. Polyacrylamide P(AAm) and poly(ethylene glycol) (PEG) are hydrophilic polymers that have demonstrated low protein, cell, and bacterial binding characteristics1-5 and have been identified as suitable candidates for coatings to minimize interactions in biological environments. In this work, we incorporated P(AAm) and PEG into interpenetrating polymer networks (IPNs) grafted to oxide surfaces (e.g., SiO2, quartz). Interpenetrating polymer networks are combinations of two or more cross-linked polymers in network form that are synthesized in juxtaposition.6 For scientific and patent literature, six main categories of IPNs have been identified: sequential IPNs, simultaneous IPNs, latex IPNs, gradient IPNs, thermoplastic IPNs, and semi IPNs. Compared to individual networks, IPNs typically demonstrate improved strength and mechanical properties * To whom correspondence should be addressed. † Northwestern University Dental School. ‡ Northwestern University. § University of Washington. | Northwestern University Medical School. X Abstract published in Advance ACS Abstracts, September 1, 1997. (1) Tazuke, S.; Kimura, H. Makromol. Chem. 1978, 179, 2603. (2) Drumheller, P. D.; Hubbell, J. A. J. Biomedical Mat. Res. 1995, 29, 207. (3) Desai, N. P.; Hubbell, J. A. Biomaterials 1991, 12, 144. (4) Dunkirk, S. G. J. Biomater. Appl. 1991, 6, 131. (5) Gristina, A. G. Science 1987, 237, 1588. (6) Klempner, D.; Sperling, L. H.; Utracki, L. A. Interpenetrating Polymer Networks; American Chemical Society: New York, 1994; Vol. IV, pp 1-38, 77-123, 638.

S0743-7463(97)00101-7 CCC: $14.00

due to a synergistic effect induced by forced compatibility of the components.7 (In polymer science and engineering, compatibility is often used to characterize desirable or beneficial properties upon mixing of the polymers.8) IPNs can be formulated into gradient IPNs that can act as ion exchange resins9 or can permit drug delivery of medication at predetermined rates.10 There are many other potential applications for IPNs in the prevention of protein fouling and loss in the biotechnology and pharmaceutical industries. Alternatively, IPNs can be used as high-performance specialty polymers in advanced microelectronic devices11 and provide useful materials for sound and vibration damping that may be used in automotive, aircraft, and submarine technology.12 Whereas P(AAm) has primarily been attached to substrates by photoinduced graft polymerization,1,13,14 PEG has been applied to surfaces using a larger variety of methodologies. PEG based molecules have been chemisorbed onto gold for the prevention of protein adsorption,15 centrifugally cast from a polymer reaction mixture into vascular conduits, and γ irradiated onto (7) Xie, H. Q.; Zhang, C. X.; Guo, J. S. Castor Oil PolyurethaneAcrylic or Vinyl Polymer Interpenetrating Polymer Networks Cured at Room Temperature; Xie, H. Q., Zhang, C. X., Guo, J. S., Eds.; American Chemical Society: Washington, DC, 1994; pp 557-570. (8) Paul, D. R.; Newman, S. In Polymer Blends; Academic Press: New York, 1978; Vol. 1, pp 17-18. (9) Barrett, J. H.; Clemens, D. H. U.S. Patent 3,966,489, 1976. (10) Mueller, K. F.; Heiber, S. J. J. Appl. Polym. Sci. 1982, 27, 4043. (11) Ree, M.; Yoon, D. Y. Rodlike Cross-Linked, Flexible Polyimide Semi-interpenetrating Polymer Network Composites; American Chemical Society: Washington, DC, 1994; pp 557-570. (12) Yamamoto, K.; Takahashi, A. Sound and Vibration damping with Polymers; American Chemical Society: Washington, DC, 1990. (13) Smithson, R. L. W.; Evans, D. F.; Monfils, J. D.; Guire, P. E. Colloids Surf., B 1993, 349. (14) Yao, Z. P.; Ranby, B. J. Appl. Polym. Sci. 1990, 40, 1647. (15) Pale-Grosdemange, C.; Simon, E. S.; Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1991, 113, 12.

© 1997 American Chemical Society

5176 Langmuir, Vol. 13, No. 19, 1997

silane-treated glass for the prevention of platelet adhesion.16,17 Common problems with these methodologies include difficulty in achieving a consistent molecular structure and difficulty in achieving strong adhesion to substrates when used as a coating. Photoinitiated grafting of PEG, although not the predominant method of grafting, is dependable and consistent. It has been used to prevent blood component deposition when grafted onto synthetic polymer surfaces and to prevent postoperative adhesions when grafted directly to tissue, as exemplified in the work of Drumheller and Hubbell2 and Sawhney.18 It has also been used to microfabricate polymer surfaces for directed tissue formation.19,20 We have developed a novel copolymer that minimizes protein adsorption and cell attachment. We report here on synthesis, characterization, and testing. In the development of this grafted copolymer, several critical factors which determined the resulting properties of the IPN were considered, including the solvent system, solubility of monomer/polymer system, and concentration of polymer and cross-linker. Other important factors, such as method of polymerization, type of initiator, and order of synthesis, had been previously defined.20 We show that sequential photoinitiated grafting of the P(AAm) and PEG led to uniform coverage and efficient cross-linking within an interpenetrating polymer network. Materials and Methods Surface Modification. Polished silicon wafers (n type, 〈100〉, International Wafer service, Portola Valley, CA) were used for all studies involving contact angle measurements, spectroscopic ellipsometry (SE), and X-ray photoelectron spectroscopy (XPS). Quartz disks (Quartz Scientific Inc., Fairport Harbor, OH) were used for all radial flow apparatus (RFA) studies. Substrates were sequentially cleaned ultrasonically for 10 min in ASTM grade I water (18 MΩ cm resistance) (further referred to as ultrapure water), acetone, and hexane, and then subjected to a Piranha etch (90% (v/v) sulfuric acid, 10% (v/v) hydrogen peroxide) for 15 min and thoroughly rinsed in ultrapure water. (All chemicals were purchased from Aldrich (Milwaukee, WI, USA) unless otherwise noted, and used as received.) Surfaces that were further modified with silanes and polymers underwent additional cleaning by an oxygen plasma (March Plasmod, Concord, CA) set at 0.5 Torr and 150 W power for 5 min. Interpenetrating networks (IPNs) of P(AAm) and PEG were grafted to the silicon or quartz substrates via an unsaturated organosilane, allyltrichlorosilane (ATC). ATC was reacted with either silicon or quartz substrates by soaking samples in a 1.25% (v/v) ATC-anhydrous toluene solution for 5 min; the solution was prepared under a nitrogen atmosphere in a glovebox and then transferred to a laminar flow fume hood for the deposition. Samples were then rinsed three times for 1 min each in toluene and baked in an oven at 120 °C for 5 min. A two-step polymerization scheme was developed to graft the P(AAm-co-EG) to the substrate surface. First the P(AAm) was coupled to the ATC-modified surface, followed by formation of the IPN with PEG. Figure 1 schematically shows the synthesis protocol. Taking monomer concentration, cross-linker concentration, photoinitiator, and photoinitiation time into account, the P(AAm) layer was optimized by constraining the cross-linker level below a “critical point” determined by Campbell.21 Acrylamide (AAm, Polysciences, Warrington, PA) was grafted to ATC (16) Chaikof, E. L.; Merrill, E. W.; Coleman, J. E.; Ramberg, K.; Connolly, R. J.; Callow, A. D. AIChE J. 1990, 36, 994. (17) Amiji, M.; Park, K. J. Colloid Interface Sci. 1993, 155, 251. (18) Sawhney, A. S.; Pathak, C. P.; van Rensburg, J. J.; Dunn, R. C.; Hubbell, J. A. J. Biomed. Mater. Res. 1994, 28, 831. (19) Matsuda, T.; Sugawara, T. J. Biomed. Mater. Res. 1995, 29, 749. (20) Bearinger, J. P.; Rezania, A.; Chen, J.; Healy, K. E. Minimization of Protein Adsorption and Cell Adhesion to a Grafted Copolymer of P(AAm-co-EG); Fifth World Biomaterials Congress; University of Toronto Press: Toronto, Canada, 1996; Vol. 1, p 308. (21) Campbell, W. P.; Wrigley, C. W.; Margolis, J. Anal. Biochem. 1983, 129, 31.

Bearinger et al.

Figure 1. Schematic of synthesis protocol: (a) ATC silane grafted to oxide surface; (b) AAm and BIS grafted to ATCmodified surface; (c) IPN formed upon grafting PEG and BIS to P(AAm) modified surface. by the following protocol: 0.1485 mg/mL AAm, 0.0015 mg/mL N,N-methylenebis(acrylamide) (BIS, Polysciences, Warrington, PA), and 0.03324 mg/mL mL dl-camphorquinone (CQ, Polysciences, Warrington, PA) were dissolved in acetone. These values correspond to a 15% T, 1% C system; % T represents grams of monomer plus cross-linker/100 mL of solvent, and % C represents grams of cross-linker/grams of monomer plus crosslinker.21 Nitrogen was bubbled through the mixture for 15 min to remove residual oxygen. ATC-treated silicon or quartz samples were placed in glass dishes containing the AAm solution. After 5 min of adsorption time, the samples underwent photoinitiated polymerization to gelation for 1.5 min using a Heraeus Amersil Z7 indium light source that has a predominant emission peak at approximately 470 nm. Substrates were removed from the petri dishes and then rinsed and sonicated for 30 min in ultrapure water to remove unreacted materials. When optimizing the poly(ethylene glycol) network structure, the solubility of BIS was examined in conjunction with a system incorporating poly(ethylene glycol) monomethyl ether monomethacrylate (MW 1000) (PEG, Polysciences, Warrington, PA) and CQ. The goal was to swell the acrylamide network to allow the penetration of the PEG monomer and BIS cross-linker. The effects of using acetone versus methanol as the solvent for the photoinitiation were compared and are reported in the contact angle, ellipsometry, and XPS depth profile results. When acetone was the designated solvent, the following material concentrations were used: 0.1000 mg/mL PEG, 1.25 × 10-4 mg/mL BIS, and 0.03324 mg/mL CQ. These values correspond to a 10% T, 1% C system. When methanol was the designated solvent, the material concentrations were 0.0200 mg/mL PEG, 0.0300 mg/mL BIS, and 0.03324 mg/mL CQ. These values correspond to a 5% T, 60% C system. For both solvent systems, nitrogen was bubbled through the mixture for 15 min and the mixture was adsorbed onto substrates that had been modified with ATC and P(AAm) and placed in glass dishes. Photoinitiation took place under the same conditions as stated for AAm for 5 min. In contrast to the acrylamide polymerization, PEG polymerization did not result in gelation of the solvent/polymer mixture. However, each of the mixtures did become more viscous. Substrates were removed from the dishes and then rinsed and sonicated in ultrapure water for 3 h to remove unreacted materials. Materials were left in water an additional 10 h in order for the films to reach a consistent thickness, as determined by ellipsometry. The PEG layer used to perform protein adsorption and adhesion tests was grafted to the P(AAm) layer using methanol as the solvent. Surface Characterization. Surfaces were characterized by contact angle measurements, spectroscopic ellipsometry (SE), and X-ray photoelectron spectroscopy (XPS). Contact angle measurements were made using a customized micrometer microscope fitted with a goniometer eyepiece (Gaertner, Chicago, IL). Quasistatic advancing (θadv) and receding (θrec) angles were measured as previously described.22 Spectroscopic ellipsometry was used to determine the thicknesses of the oxide layer, immobilized organosilanes, P(AAm), P(AAm-co-EG), and any protein layers adsorbed to silicon wafer samples. All film measurements were taken on samples that had been placed in a desiccator overnight, and data were collected in air at room temperature (∼25 °C) and relative humidity (∼3050%). A MOSS ES4G Sopra ellipsometer, equipped with a highpressure xenon arc lamp (75 W) was used and controlled by ELLISPEC, a software program supplied with the ellipsometer. (22) Healy, K. E.; Thomas, C. H.; Rezania, A.; Kim, J. E.; McKeown, P. J.; Lom, B.; Hockberger, P. E. Biomaterials 1996, 17, 195.

Interpenetrating Polymer Networks Wavelengths from 250 to 900 nm were used in the measurements necessary for determination of thickness. Prior to each measurement, the signal was optimized by focusing the X, Y, and Z settings of the light beam. An offset calibration served to determine the starting analyzer angle, Ψ, and polarizer angle, ∆, with respect to the plane of incidence. The angle of incidence was initially input in the computer at 75°, and then the actual angle was calculated using a silicon wafer standard with an oxide 109.8 nm thick. The actual angle was usually within 0.5° of the set value. Regression analysis was used to determine the thickness of the native oxide layer using a library of n(λ) and k(λ) data for the bulk silicon and silicon oxide layer. To calculate the thickness of multilayer samples, the Cauchy dispersion model was utilized.23 The thicknesses of the oxide and silane layers were assumed to remain constant as the polymer layers were synthesized on top of them. A bulk thickness measurement fit on n(λ) (real part of the index of refraction as a function of wavelength) and k(λ) (extinction coefficient; imaginary part of the index of refraction as a function of wavelength) was always taken for P(AAm) and P(AAm-co-EG) samples, and previously determined layers were subtracted from the bulk thickness measurement to determine the thickness of only the newly synthesized outer layer. The results were then fitted as multilayer polymer structures on all the layers to confirm the consistency of the copolymer network. At least two spots were examined per wafer to test the precision of the measurement protocol. XPS analyses were conducted on a Surface Science Instrument (SSI) X-probe spectrometer with a monochromatic Al KR1,2 X-ray source (1486.6 eV) to stimulate photoemission. Emitted electron energies were measured with a hemispherical energy analyzer at pass energies ranging from 25 to 150 eV. Binding energy was referenced by setting the CHx peak maximum in the C 1s spectrum to 285.0 eV. Survey spectra were collected at a takeoff angle of 55° and high resolution C 1s spectra were collected at takeoff angles of 0°, 55°, and 80°. (Takeoff angle (Φ) is defined as the angle between the surface normal and the axis of the analyzer lens system. The solid acceptance angle of the analyzer lens was decreased to 12° × 30° by placing an aperture over the analyzer lens to improve the depth resolution at each takeoff angle.24) Survey spectra were collected over a binding energy range of 0-1000 eV and high-resolution C 1s spectra were collected over a 275-295 eV range. Analyzer resolution was on the order of 1.5 eV for survey scans and 0.25 eV for high-resolution scans. SSI data analysis software was used to calculate elemental compositions from peak areas and to peak-fit the high resolution spectra. Angle-dependent XPS measurements were performed to determine and compare depth profiles of the P(AAm) and PEG layers polymerized in acetone and methanol. A regularization algorithm developed by Tyler (1989)25 was used to fit the angledependent study data from the near surface region of the samples (∼15 nm) and determine a compositional depth profile (CDP). The CDP generated in this manner is consistent with, but not necessarily unique to, the experimental angle-dependent XPS data. The mean free paths used in determining the CDPs were calculated from the equations given by Seah and Dench.26 These calculations predict the sampling depth (three times the mean free path) should decrease from about 9 to 1.5 nm as the takeoff angle increases from 0° to 80°. Protein Adsorption Experiments. Protein adsorption experiments on clean and modified silicon wafers in serum were performed to determine the extent of protein deposition on surfaces at two time intervals. An aminosilane, (N-(2-aminoethyl)-3-aminopropyl)trimethoxysilane (EDS, Huls America, Piscataway, NJ) was bound to oxygen plasma treated substrates by a previously described protocol.22 EDS surfaces served as positive controls in protein adsorption and cell adhesion experiments. Clean, EDS, P(AAm), and P(AAm-co-EG) samples were immersed in 70% ethanol for 10 min, rinsed three times in (23) Born, M.; Wolf, E. In Principles of Optics; Pergamon: New York, 1975; p 95. (24) Tyler, B. J.; Castner, D. G.; Ratner, B. D. J. Vac. Sci. Technol. 1989, A7, 1646. (25) Tyler, B. J.; Castner, D. G.; Ratner, B. D. Surf. Interface Anal. 1988, 14, 443. (26) Seah, M. P.; Dench, W. A. Interface Anal. 1979, 1, 2.

Langmuir, Vol. 13, No. 19, 1997 5177 phosphate-buffered saline (PBS), and then soaked in Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco, Grand Island, NY) containing 15% fetal bovine serum (FBS) for either 2 or 24 h. All samples were rinsed in PBS and then ultrapure water after the allotted times to remove unbound proteins and/or salt residue. Thicknesses of materials and any adsorbed layers were then analyzed using spectroscopic ellipsometry and compared to samples not exposed to the protein solution. Endothelial Cell Culture. Human umbilical vein endothelial cells (HUVEC) were isolated by collagenase digestion from freshly delivered umbilical cords obtained after cesarean births using established methods.27 The growth medium consisted of RPMI 1640 (Gibco, Grand Island, NY) supplemented with 20% iron fortified bovine calf serum (BCS, Hyclone Laboratories, Logan, UT), 200 µg/mL endothelial growth supplement (ECGS, Collaborative Research, Bedford, MA), 100 U/mL penicillin, 100 µg/mL streptomycin, 50 µg/mL gentamicin, 2.5 µg/mL amphotericin B, 2 mM glutamine (GIBCO-BRL), and 5 U/mL sodium heparin (Fischer Scientific, Itasca, IL). The cells were confirmed as endothelial by positive Von Willebrand factor staining and negative cytokeratin and alpha smooth muscle actin staining. Cultures were found free of mycoplasmal contamination according to the method of Chen.28 Cells were maintained in 75 cm2 flasks with HUVEC medium and were fed every other day. All cells were used between passages 3 and 6, which corresponds to approximately 20 population doublings. Radial Flow Apparatus. A radial flow apparatus was used to perform an in vitro cell detachment assay and thereby determine the adhesive nature of the modified substrates. Details of the flow apparatus and methods of analysis are given elsewhere.29 Briefly, cells were detached from culture flasks upon exposure to 0.5 mM ethylene glycol-bis[β-aminoethyl ether]-N,N,N′,N′-tetraacetic acid (EGTA) for 5 min and then rinsed with PBS in an attempt to remove 90-95% of the cells. Cell suspensions were then centrifuged for 5 min and cells were resuspended in HUVEC medium. Substrates were sterilized in 70% ethanol for 10 min and rinsed thoroughly in PBS prior to plating of cells. Cells were then plated at an approximate density of 10.25 × 104 cells mL-1 (∼8.34 × 104 cells/cm2, ∼2.64 × 106 cells/surface) on either clean, EDS, P(AAm), or P(AAm-co-EG) substrates and incubated for 20 min. After the incubation period, the quartz disk plated with cells was removed from the cell suspension and carefully inverted onto the flow chamber which was placed on an inverted light microscope. The chamber was connected to a reservoir containing Minimum Essential Media (R-MEM, Gibco, Grand Island, NY), without serum. Temperature was maintained at 37 °C and constant flow rate was achieved by maintaining a constant hydrostatic pressure difference between the R-MEM reservoir and the flow chamber. Three rates were achieved by adjusting the height difference between the reservoir and the microscope stage. Flow rate was calculated by measuring the volume of fluid collected per time. Initial cell counts and cell counts after each flow rate were recorded at varying radial positions from the center of the disk using light microscopy. The shear stress at the wall was determined using the following equation based on the continuity and Navier Stokes equations29

τ)

3FQ2 3Qµ 4πh2r 140π2hr3

(1)

where Q is the volumetric flow rate, 2h is the gap height between plates (0.022 cm), r is the radial distance, F is the density of the flowing media (∼1 g/cm3), and µ is the fluid viscosity at 37 °C (0.85 cP). Only laminar flow regimes were used to calculate τ. Flow rates ranging from 0.2 to 2.0 mL/s at radial distances of 0.4-2.0 cm from the origin of flow were used in order to maintain the local channel Reynolds number under 500.

Results Surface Characterization. The results of the contact angle measurements are reported in Table 1. Clean oxide (27) Schnaper, H. W.; Grant, D. S.; Stetler, W. G.; Fridman, R.; D’Orazi, G.; Murphy, A.; Bird, R. E.; Hoyhtya, M.; Fuerst, T. R.; Quigley, J.; French, D.; Kleinman, H. K. J. Cell Physiol. 1993, 156, 235. (28) Chen, T. R. Exp. Cell Res. 1977, 104, 255. (29) Rezania, A.; Thomas, C. H.; Healy, K. E. Ann. Biomed. Eng. 1997, 25, 190.

5178 Langmuir, Vol. 13, No. 19, 1997

Bearinger et al.

Table 1. Advancing (θadv) and Receding (θrec) Contact Angles ( Standard Deviation of Modified Surfaces surface Si/SiO2 EDS ATC P(AAm) PEGa PEGb a

contact angle (deg) θadv θrec