Viscoelastic Properties of Fibrinogen Adsorbed to the Surface of

Biomaterials Used in Blood-Contacting Medical Devices. Norbert Weber ... The State UniVersity of New Jersey, and REVA Medical Inc., San Diego, Califor...
3 downloads 0 Views 174KB Size
3298

Langmuir 2007, 23, 3298-3304

Viscoelastic Properties of Fibrinogen Adsorbed to the Surface of Biomaterials Used in Blood-Contacting Medical Devices Norbert Weber,† Aaron Pesnell,† Durgadas Bolikal,† Joan Zeltinger,‡ and Joachim Kohn*,† Department of Chemistry and Chemical Biology and New Jersey Center for Biomaterials, Rutgers, The State UniVersity of New Jersey, and REVA Medical Inc., San Diego, California ReceiVed February 21, 2006. In Final Form: October 24, 2006 The hemocompatibility of polymeric vascular implants is in part dependent on the propensity of fibrinogen to adsorb to the implant surface. Fibrinogen surface adsorption was measured in real time using a quartz crystal microbalance with dissipation monitoring (QCM-D). Six new, biodegradable tyrosine-derived polycarbonates were used as test surfaces. Stainless steel, poly(L-lactic acid), poly(D,L-lactide-co-glycolide), and poly(ethylene terephthalate) surfaces served as controls and provided a comparison of the test surfaces with those of commonly used biomaterials. Our study addressed the question regarding to which extent systematic variations in polymer structure can be used to optimize X-ray visibility and provide tunable degradation rates while generating protein-repellant surface properties that minimize fibrinogen adsorption. QCM-D revealed surface-dependent changes in fibrinogen layer thickness (2 to 37 nm), adsorbed wet mass (0.2 to 4.3 µg/cm2), and viscosity (0.001 to 0.005 kg/ms). While we did not find an overall correlation between surface air-water contact angle measurements and fibrinogen adsorption (R2 ) 0.08), our data demonstrate that gradually increasing the poly(ethylene glycol) content within a subgroup of polymers having the same polymer backbone will lead to decreased fibrinogen adsorption. Within this subgroup of polymers, there was a strong correlation between decreasing air-water contact angles and decreasing fibrinogen adsorption (R2 ) 0.95). We conclude that it is possible to minimize fibrinogen adsorption to tyrosine-derived polycarbonates while optimizing X-ray visibility and degradation rates. Some of the tyrosine-derived polycarbonates were identified as useful materials for the design of blood-contacting implants on the basis of their substantially lower levels of fibrinogen adsorption relative to the commonly used controls.

Introduction Hemocompatibility evaluation of synthetic biomaterials used in blood-contacting applications is critical in order to prevent undesirable clinical outcomes. Upon implantation in the body, plasma proteins in the circulating blood ubiquitously adsorb on the device surface. Thromboembolic complications result when this proteinacious layer mediates the adhesion and activation of platelets triggering a clotting reaction. It is widely accepted that the plasma protein fibrinogen (Fg) plays a predominant role in mediating the adhesion of platelets. For instance, Pitt et al.1 reported that the adsorption of Fg on a biomaterial surface leads to a reduction in hemocompatibility. In a separate study, Stanford and co-workers2 found that surfaceadsorbed Fg increased the binding and activation of platelets potentially leading to thrombotic events. More recently, Tsai et al.3 demonstrated that platelet adhesion to artificial surfaces preadsorbed with blood plasma is mediated mostly by surfacebound Fg. Finally, surface-adsorbed Fg reportedly initiates an acute inflammatory response on implanted surfaces.4 In the above-mentioned studies, a number of experimental methods have been advocated for the quantitative measurement of protein adsorption at solid-liquid interfaces. Techniques such * Corresponding author. New Jersey Center for Biomaterials, 145 Bevier Road, Piscataway, NJ 08854. Email: kohn@biology.rutgers.edu. Phone: 732-445-3888. Fax: 732 445 5006. † Rutgers, The State University of New Jersey. ‡ REVA Medical Inc. (1) Pitt, W. G.; Park, K.; Cooper, S. L. J. Colloid Interface Sci. 1986, 111, 343-362. (2) Stanford, M. F.; Munoz, P. C.; Vroman, L. Ann. N.Y. Acad. Sci. 1983, 416, 504-512. (3) Tsai, W. B.; Grunkemeier, J. M.; McFarland, C. D.; Horbett, T. A. J. Biomed. Mater. Res. 2002, 60 (3), 348-359. (4) Tang, L.; Eaton, J. W. J. Exp. Med. 1993, 178 (6), 2147-2156. (5) Davies, J. Nanobiology 1994, 3, 5-16.

as protein radiolabeling and immunofluorescence assays rely on chemical modification of the protein or the binding of a specific fluorescent antibody, respectively. By contrast, surface plasmon resonance (SPR),5 optical waveguide lightmode spectroscopy (OWLS),6,7 and quartz crystal microbalance with dissipation monitoring (QCM-D)8 offer the advantage of providing data in real time without the necessity of chemical labeling. While all three aforementioned techniques possess sensitivity in the ng/cm2 range, QCM-D provides unique information on the viscoelastic properties (density, viscosity, shear).9 Moreover, QCM-D measures quantitative values of the adsorbed “wet” mass and layer thickness.10 This information can provide a qualitative fingerprint of the conformational state (rigid/nonrigid) of the adsorbed layer.11 In a recent example, Roach et al.12 reported that albumin adsorbed on methyl- and hydroxyl-terminated model surfaces in a single-step process, whereas Fg adsorption proceeded through a more complex multistage process. On the basis of the ability of the QCM-D technique to measure real-time formation, dynamics, and kinetics of adsorbing Fg layers at the materialliquid interface, we selected the QCM-D technique for this study. In a previous study, we reported on the formation of viscoelastic protein layers on polymeric surfaces relevant to platelet adhe(6) Ho¨o¨k, F.; Voros, J.; Rodahl, M.; Kurrat, R.; Boni, P.; Ramsden, J. J.; Textor, M.; Spencer, N. D.; Tengvall, P.; Gold, J.; Kasemo, B. Colloids Surf., B 2002, 24 (2), 155-170. (7) Voros, J.; Ramsden, J. J.; Csucs, G.; Szendro, I.; De Paul, S. M.; Textor, M.; Spencer, N. D. Biomaterials 2002, 23 (17), 3699-3710. (8) Rodahl, M.; Kasemo, B. ReV. Sci. Instrum. 1996, 67 (9), 3238-3241. (9) Voinova, M. V.; Rodahl, M.; Jonson, M.; Kasemo, B. Phys. Scr. 1999, 59 (5), 391-396. (10) Ho¨o¨k, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Anal. Chem. 2001, 73 (24), 5796-5804. (11) Ho¨o¨k, F.; Rodahl, M.; Brzezinski, P.; Kasemo, B. Langmuir 1998, 14 (4), 729-734. (12) Roach, P.; Farrar, D.; Perry, C. C. J. Am. Chem. Soc. 2005, 127 (22), 8168-8173.

10.1021/la060500r CCC: $37.00 © 2007 American Chemical Society Published on Web 02/10/2007

Viscoelastic Properties of Adsorbed Fibrinogen

Figure 1. Resorbable polymer stent made of tyrosine-derived polycarbonates using a slide-and-lock design. Photograph provided by Reva Medical, Inc.

sion.13 The adsorption of Fg and the subsequent binding of the soluble platelet receptor GPIIb-IIIa on different polymeric surfaces was quantified in real time by QCM-D. A good correlation was found between in vitro platelet adhesion and the quantitative binding efficiency of GPIIb-IIIa to polymer-adsorbed Fg. To validate the specificity of GPIIb-IIIa binding, a synthetic RGDcontaining peptide was shown to inhibit the glycoprotein binding to polymer-adsorbed Fg. Recently, the development of a polymer stent was reported, based on a slide-and-lock design using degradable polymers selected from the library of tyrosine-derived polycarbonates (Figure 1).14 This development prompted us to explore the effect of three structural variations of tyrosine-derived polycarbonates on Fg adsorption (incorporation of poly(ethylene glycol), carboxylic acid groups, and iodine; Figure 2). To allow the comparison of our results to those obtained by others, we also included a range of clinically used biomaterials in our study. The incorporation of poly(ethylene glycol) (PEG) was investigated, since it is known to improve biomaterial hemocompatibility by substantially reducing protein adsorption and cell adhesion.15 As a second structural variation, copolymers containing desaminotyrosyl tyrosine (DT) were explored. The incorporation of DT units into the polymer backbone not only provides a source for negatively charged carboxylic acid groups on the polymer surface, but also affects the polymer degradation rate. Previous studies have shown that the molar ratio of DT units within the polymer backbone is closely correlated with the rate of polymer degradation.14,16 However, the effect of negatively charged DT units on Fg adsorption had not been explored. Finally, we investigated the effect of iodine atoms on Fg adsorption. Since iodine scatters X-rays with high efficiency, we iodinated the aromatic ring of desaminotyrosine to produce polymers useful in the design of medical implants that can be visualized by X-ray fluoroscopy.17,18 A number of other iodinated polymer composi(13) Weber, N.; Wendel, H. P.; Kohn, J. J. Biomed. Mater. Res., Part A 2005, 72A (4), 420-427. (14) Zeltinger, J.; Schmid, E.; Brandon, D.; Bolikal, D.; Pesnell, A.; Kohn, J. Advances in the development of coronary stents. Biomaterials Forum, Official Newsletter of the Society for Biomaterials, 2004, 26(1), 2004; pp 8-9. (15) Harris, J. M. Polym. Prepr. (Am. Chem. Soc., DiV. Polym. Chem.) 1997, 213, 21. (16) James, K.; Levene, H.; Parsons, J. R.; Kohn, J. Biomaterials 1999, 20 (23-24), 2203-2212. (17) Kohn, J.; Bolikal, D.; Pendharkar, S. M. Radio-opaque Polymeric Biomaterials. U.S. Patent 6,852,308, 2005. (18) Pendharkar, S. M.; James, K.; Kohn, J. Society for Biomaterials: Transactions, 24th Annual Meeting, San Diego, 1998; Vol. 21, p 386.

Langmuir, Vol. 23, No. 6, 2007 3299

Figure 2. General chemical structures of tyrosine-derived polymers and possible variation in their I2DTE, I2DT, and poly(ethylene glycol) (PEG) content. DTE stands for desaminotyrosyltyrosine ethyl ester. In I2DTE and I2DT, the desaminotyrosyl ring is iodinated. The molar fraction of PEG (Mw 2000 g/mol) units in the copolymer was varied between 0 and 15 mol %. The following six polymers were synthesized and abbreviated as follows: poly(DTE carbonate) (PDTEC), poly(I2DTE carbonate) (PIDTEC), poly(I2DTE-co2.5%PEG2000 carbonate) (PIDTE-co-2.5%PEGC), poly(I2DTE-co3.4%PEG2000 carbonate) (PIDTE-co-3.4%PEGC), poly(I2DTE-co15%PEG2000 carbonate) (PIDTE-co-15%PEGC), poly(I2DTE-co10%I2DT-co-2.5%PEG2000 carbonate) (PIDTE-co-10%DT-co2.5%PEGC).

tions have recently been described,19,20 but the effect of surface iodine on Fg adsorption had not yet been investigated. Materials and Methods Reagents. Tetrahydrofuran (THF) and methylene chloride were purchased from EMD Chemicals (Gibbstown, NJ) and Fisher Scientific (Tustin, CA), respectively. Hexafluoroisopropanol (HFIP) was obtained from DuPont (Wilmington, DE). Phosphate-buffered saline (PBS) and Tween 20 were purchased from Sigma (St. Louis, MO). Fg from human plasma was from Calbiochem (La Jolla, CA). Hydrogen peroxide and ammonium hydroxide were from Fisher Scientific and EMD Chemicals (Gibbstown, NJ), respectively. Test Materials. Poly(DTE carbonate) and PEGylated copolymers (Figure 2) were synthesized according to published procedures.17,21 Iodinated polycarbonates were synthesized using modified methods reported by Pendharkar et al.18 The mole fraction of PEG units in the copolymer was varied between 0 and 15 mol % PEG, and the molecular weight of the PEG blocks was 2000 g/mol. Poly(D,Llactide-co-glycolide) (PLGA, Resomer 506) and poly(L-lactic acid) (PLLA, Resomer L-206) were purchased from Boehringer Ingelheim (Ridgefield, CT). Poly(ethylene terephthalate) (PET) was obtained from Sigma (St. Louis, MO). Stainless steel-coated quartz crystals (QSX 304) and gold-coated crystals (QSX 301) were from Q-Sense (Go¨teborg, Sweden). Polymer Characterization. Routine polymer characterization included proton nuclear magnetic resonance (1H NMR) spectrometry, measurement of molecular weight by gel permeation chromatography (GPC), dry glass transition temperature by differential scanning calorimetry (DSC), decomposition temperature by thermogravimetric analysis (TGA), and air-water contact angle by goniometry as described before.22,23 (19) Mottu, F.; Rufenacht, D. A.; Laurent, A.; Doelker, E. Biomaterials 2002, 23 (1), 121-131. (20) Jones, D. S.; Djokic, J.; McCoy, C. P.; Gorman, S. P. Biomaterials 2002, 23 (23), 4449-4458. (21) Yu, C.; Kohn, J. Biomaterials 1999, 20, 253-264. (22) Brocchini, S.; James, K.; Tangpasuthadol, V.; Kohn, J. J. Biomed. Mater. Res. 1998, 42 (1), 66-75. (23) Ertel, S. I.; Kohn, J. J. Biomed. Mater. Res. 1994, 28 (8), 919-930.

3300 Langmuir, Vol. 23, No. 6, 2007

Weber et al.

Figure 3. Real-time QCM-D data of the normalized fifth overtone for Fg adsorption to PLLA (A), stainless steel (B), and poly(I2DTEco-15%PEG2000 carbonate) (C) surfaces. (1) Injection of Fg solution; (2) rinsing steps with protein-free buffer. The frequency shifts, ∆f, (plot a) in conjunction with dissipation shifts, ∆D, (plot b) are indicative of the formation of a viscous Fg layer at the solution-polymer interface. Quartz Crystal Microbalance with Dissipation Monitoring (QCM-D). In a QCM-D experiment, a quartz crystal is driven by applying an alternating electric field at its resonance frequency of 5 MHz or at one of its first three overtones: 15, 25, or 35 MHz. An increase or decrease in mass bound to the crystal (∆m) caused by adsorption or desorption, respectively, is measured in real time by following changes in frequency (∆f) and dissipation (∆D). Adsorbed mass of thin and nondissipative layers can be calculated using the Sauerbrey24 relationship where ∆f is linearly related to the adsorbed mass ∆m ) -C∆f/n where C is a constant based on the physical properties of the quartz crystal (in this case, C ) 17.7 ng/(cm2 Hz) at f ) 5 MHz) and n ()1, 3, ...) is the overtone number. The Sauerbrey equation is not applicable for hydrated and highly viscous surface-adsorbed protein layers that result in high dissipation shifts. In order to calculate the adsorbed mass and viscoelastic properties of such layers, theoretical modeling of the QCM-D response is necessary by considering a “Voigt model”.9 The Voigt model has been described and applied previously by Weber and co-workers.13 In summary, the adsorbed layer is represented by four unknown parameters: effective density (F), shear viscosity (η), shear elastic modulus (µ), and thickness (δ). Since the viscous layers give rise to different penetration depths of the harmonic acoustic frequencies, the three overtones can be simultaneously fitted using the Voigt model to calculate values of F, η, µ and δ. Substrate Preparation. Gold-coated quartz crystals (5 MHz, QSX 301; Q-Sense AB, Go¨eteborg, Sweden) were spin-coated using a spin coater (Headway Research, Inc., TX) with 1% (w/v) polymer solutions in methylene chloride. However, poly(I2DTE-co-10%I2DT-co-2.5%PEG2000 carbonate) and PET had to be dissolved in THF and HPIP, respectively. Stainless steel-coated crystals (5 MHz, QSX 304) were used as received. To permit crystal reuse, crystals were incubated (15 min) in the respective solvent followed by extensively rinsing with solvent. Next, the crystals were treated with a mixture of H2O2 (30%), NH4OH, and deionized water in a 1:1:5 ratio (15 min, 80 °C). Thereafter, crystals were rinsed with deionized water, dried under nitrogen, and exposed to UV and ozone for 10 min (UVO Cleaner, Jelight Company, Irvine, CA) followed by rinsing with pure ethanol. Protein Adsorption Studies. Human Fg adsorption to polymercoated crystals was performed using the QCM-D model D300 (QSense AB, Go¨eteborg, Sweden) at 37 °C. Fg was diluted in PBS (3 mg/mL). Next, frequency and dissipation shifts induced by adsorbed Fg were monitored in real time at 3, 5, and 7 times the crystal’s natural frequency of 5 MHz. All experiments were performed in PBS under nonflow conditions. Protein-free PBS was used for all rinsing steps to remove nonbound Fg from the polymer-coated sensor surface after the adsorption process. Following the conclusion (24) Sauerbrey, G. Z. Phys. 1959, 155, 206-222.

of each experiment, the QCM-D chamber and tubing system were filled with 1% (v/v) Tween 20, extensively rinsed with deionized water, and dried under nitrogen. To start a typical experiment (see Figure 3), a 5 mL polypropylene pipet tip was connected to the inlet tube of the QCM-D chamber. The liquid (1.5 mL solution is necessary for the adsorption step) was exchanged by opening the valve, allowing the liquid in the QCM-D chamber to exchange via gravitational flow. After the instrument showed a constant baseline for the frequency and for the dissipation, the Fg solution was exposed to the polymer-coated crystal until a binding saturation was reached (∼60-70 min), followed by rinsing steps with protein-free PBS. Values of the normalized fifth overtone (25 MHz) for frequency shifts (∆f) and dissipation shifts (∆D) were then plotted for each type of test surface. Modeling of the QCM-D Response. Three overtones (third, fifth, seventh) were used to model the viscoelastic properties (viscosity, elasticity, thickness) using the Voigt model (Q-Tools, Q-Sense AB, Go¨eteborg, Sweden). The layer density was fixed to 1200 kg/m3. Parameters fitted were (i) layer viscosity between 0.001 and 0.01 kg/ms, (ii) layer shear between 105 and 1012 Pa, and (iii) layer thickness between 10-10 and 10-7 m. Statistics. The Student’s paired t-test was used for the statistical analysis of the data. Probability (P) values of 3.5 mol % of PEG exhibited smaller ∆f values than the stainless steel surface. Finally, PET, known to provoke inflammatory and thrombotic responses had the greatest ∆f value (-176 Hz) of all ten test surfaces evaluated in this study. Modeling QCM-D Data. While QCM-D can be used to monitor the real-time organization and formation of protein layers with nanogram sensitivity on test substrates, interpretation of ∆f and ∆D values using the Sauerbrey equation24 to calculate the adsorbed mass is only valid for thin and nondissipatiVe layers.11 (29) Menz, B.; Knerr, R.; Gopferich, A.; Steinem, C. Biomaterials 2005, 26 (20), 4237-4243.

3302 Langmuir, Vol. 23, No. 6, 2007

Weber et al.

Figure 8. Time-resolved effect of various PEG2000 concentrations in the polymer structure on the adsorbing Fg layer thickness. (a) poly(I2DTE carbonate); (b) poly(I2DTE-co-2.5%PEG2000 carbonate); (c) poly(I2DTE-co-3.4%PEG2000 carbonate); (d) poly(I2DTE-co15%PEG2000 carbonate). QCM-D data were modeled according to the Voigt model.9 Experiments were performed as illustrated in Figure 3. Normalized data for the fifth overtone are shown.

Figure 6. (a) Surface-adsorbed mass and (b) thickness calculated according to the Sauerbrey equation24 and Voigt model.9 Experiments were performed as illustrated in Figure 3. Normalized data for the fifth overtone were used. Results are expressed as mean ( SD (n ) 3).

Figure 7. Viscosity of polymer-adsorbed Fg layers calculated according to the Voigt model.9 Experiments were performed as illustrated in Figure 3. Normalized data for the fifth overtone are shown. Results are expressed as mean ( SD (n ) 3).

Hence, a Voigt model9 must be applied to calculate the adsorbed mass and viscoelastic properties of soft or thick adsorbed layers. In this study, we found that surface-adsorbed Fg layers are viscous and dissipative, as demonstrated by high ∆D values (up to 13 × 10-6) (Figure 4b). Using the Sauerbrey equation and Voigt model, calculated values for the adsorbed mass and layer thicknesses are summarized for all the test surfaces in Figure 6. This comparison of the two models reveals that the Sauerbrey equation underestimates the adsorbed mass and the thickness of dissipative Fg layers by 25-30%. This is discussed in greater detail at the end of this section. Taken together, the Voigt thickness of the Fg layers on all ten test surfaces ranged from 2 to 37 nm (Figure 6b), the adsorbed Voigt wet mass from 0.2 to 4.3 µg/cm2 (Figure 6a), and the viscosity from 0.001 to 0.005 kg/ms (Figure 7). For example, the average thickness of the Fg layer on poly(I2DTE carbonate) (a highly protein adhesive polymer) was 28 nm, while for poly(I2DTE-co-15%PEG carbonate) (a protein repelling polymer), the Fg layer thickness was reduced to 2 nm (Figure 6b).

Addition of desaminotyrosyl tyrosine (DT) resulted in a noticeable reduction in the ∆f value. However, after modeling, the reduction in ∆f did not translate into a significant impact on the surface-adsorbed “wet mass” and thickness as demonstrated by comparison of values for poly(I2DTE-co-2.5%PEG2000 carbonate) versus poly(I2DTE-co-10%I2DT-co-2.5%PEG2000 carbonate) (Figure 6). This result demonstrated the importance of modeling the raw data before quantitative conclusions are derived from the observed ∆f values. Addition of iodine (I2) showed a slight decrease in absorbed Fg mass and thickness compared to the non-iodinated poly(DTE carbonate), which was not significant (p > 0.05; Figure 6). Addition of increasing amounts of PEG to poly(I2DTE carbonate) showed a significant and continuous decrease in absorbed Fg mass (Figure 6a), Fg layer thickness (Figure 6b), and Fg layer viscosity (Figure 7). In Figure 8, the calculated Voigt layer thickness of Fg adsorptiondesorption are reported for all PEG-containing tyrosine-derived polycarbonates. The higher the PEG concentration in the polymer structure, the thinner the Fg layer is and the less the adsorbed Fg mass (Figures 6 and 8). In particular, the polymer with the highest PEG content (15 mol %) showed essentially no Fg adsorption. Of interest are the observed polymer-dependent Fg adsorption Sauerbrey mass values in Figure 6a that range between 0.2 and 3.1 µg/cm2 for the test surfaces. The highest adsorption values are significantly greater than the values reported by Ho¨o¨k and co-workers6 using QCM-D (1.18 µg/cm2; “Sauerbrey wet mass”). This difference is most likely related to the fact that Ho¨o¨k et al.6 used a different test substrate (titanium oxide) as well as a much lower Fg solution concentration (80 µg/mL) in their study. In this study, we used a 37.5-fold higher Fg concentration (3 mg/ mL), which is the reported Fg concentration in human blood plasma.30 We noticed that the precision of the Voigt modeling technique was limited with respect to shear moduli, leading to large standard deviations (data not shown). Other investigators using the same instrument and software31 have described the limited precision of the model, especially for shear moduli. However, the other modeling parameters (adsorbed mass, thickness, and viscosity) displayed reproducible results (Figures 6-8). As the QCM-D technique is relatively new, it is important to compare the QCM-D values to those obtained by more established (30) Grunkemeier, J. M.; Tsai, W. B.; McFarland, C. D.; Horbett, T. A. Biomaterials 2000, 21 (22), 2243-2252. (31) Munro, J. C.; Frank, C. W. Macromolecules 2004, 37, 925-938.

Viscoelastic Properties of Adsorbed Fibrinogen

Figure 9. (a) Air-water contact angle of 10 test surfaces. The polymers marked (1) through (4) are a series of poly(I2DTE carbonate)s with increasing PEG content of 0, 2.5, 3.4, and 15 mol %. (b) Relationship (R2 ) 0.95) between adsorbed Fg mass and air-water contact angle of four selected polymers having identical backbone structures.

techniques. This validation was performed by Hook and coworkers6 who compared the values obtained for the mass of adsorbed Fg on titanium oxide surfaces using ellipsometry (ELM), optical waveguide light mode spectroscopy (OLWS), and QCMD. These authors report surface-adsorbed Fg mass values of 0.39 µg/cm2 measured by ELM, 0.45 µg/cm2 measured by OWLS, and 1.18 µg/cm2 measured by QCM-D using the Sauerbrey equation (which underestimates the mass by about 25% to 30%). This discrepancy can be readily explained by the fact that ELM and OWLS are based on optical principles and detect the adsorbed dry mass, while QCM-D detects hydrodynamically coupled water in addition to the protein mass, i.e., the “wet mass”. Air-Water Contact Angle and Adsorbed Mass. The average air-water contact angles for all test substrates ranged from about 55° to 78° (Figure 9A). Stainless steel, PET, PLGA, and PLLA were all significantly more hydrophilic than poly(DTE carbonate) and poly(I2DTE carbonate) (p < 0.01). The hydrophobicity increased significantly (p < 0.01) with the addition of iodine into poly(DTE carbonate), which can be explained by the known hydrophobic character of iodine.32 About 3.4 mol % of PEG2000 must be incorporated into the polymer structure in order to compensate for the increased hydrophobicity caused by iodination (PIDTEC compared to PDTEC in Figure 9). As expected, the addition of PEG to iodinated polymers decreased the hydrophobicity of the polymers.33 No overall correlation was found between air-water contact angle values and levels of Fg adsorption (R2 ) 0.08; not shown). However, gradually increasing (32) Wang, R.; Fu, Y.; Lai, L. J. Chem. Inf. Comput. Sci. 1997, 37, 615-621. (33) Chung, C. W.; Kim, H. W.; Kim, Y. B.; Rhee, Y. H. Int. J. Biol. Macromol. 2003, 32 (1-2), 17-22. (34) Horbett, T. A. CardioVasc. Pathol. 1993, (2), 137-148. (35) Cacciafesta, P.; Humphris, A. D. L.; Jandt, K. D.; Miles, M. J. Langmuir 2000, 16, 8167-8175. (36) Voros, J. Biophys. J. 2004, 87 (1), 553-561.

Langmuir, Vol. 23, No. 6, 2007 3303

the PEG content within a subgroup of polymers having the same polymer backbone resulted in decreased Fg adsorption that correlated with decreased air-water contact angles (R2 ) 0.95; Figure 9b). Possible Orientation of Surface-Adsorbed Fg. It is widely accepted that proteins adsorb to artificial surfaces in monolayers.34 Cacciafesta et al.35 summarized Fg adsorption data reported by different authors using various substrates measured by atomic force microscopy (AFM). The molecular length of surfaceadsorbed Fg molecules ranged between 46 and 66 nm, and the height varied between 1.4 and 3.4 nm. Further, Fg has a rodlike structure of width approximately 10 nm,6,28 which may lie planar at the polymer interface.28 Our results show surface-dependent Fg layer thicknesses ranging between 1.3 and 35.5 nm (Voigt thickness). Interestingly, nine out of ten test surfaces showed thicknesses greater than 10 nm (Figure 6b), which might indicate an end-on adsorption of Fg molecules to these surfaces (assuming a monolayer). In this case, the adsorbed Fg mass should be close to the calculated dry mass for Fg monolayer coverage (closepacked end-on monolayer), which has been reported as 1.05 µg/cm2.6 We found a maximum Fg adsorption on our test polymers around 3 µg/cm2 Voigt mass (Figure 6a), which corresponds to about 1 µg/cm2 dry mass in the case of Fg adsorption.6 Thus, under the assumption of monolayer formation on surfaces,34 we conclude an end-on adsorption of Fg to surfaces with thicknesses greater than 10 nm. On the basis of the available QCM-D data, however, no conclusion regarding the end-on or side-on Fg adsorption can be made for thicknesses smaller than 10 nm. Detailed experiments using AFM might give more insights into the material-dependent orientation of adsorbed Fg molecules. However, this is beyond the scope of the present study. Error Analysis. All modeled data in this study have been obtained using a fixed and assumed Fg layer density of 1200 kg/m3. However, this value might vary between 1000 kg/m3 (water) and 1350 kg/m3 (protein) depending on protein coverage. To test the error introduced by this assumption, the modeling of the experimental data was made with different (fixed) densities corresponding to lower (1050-1150 kg/m3) or higher (12501350 kg/m3) protein coverage than the 1200 kg/m3 that was chosen on the basis of a saturated protein coverage slightly above 50%.6,36 This resulted in less than 15% variations in Voigt thickness/adsorbed mass and less than 10% changes in viscosity values (data not shown). This means, in turn, that, although the protein layer density may vary by more than 50% on different surfaces, dependent on the protein coverage, the modeled values will be accurate within 10-15%seven in the case of significant differences in saturated frequency change between the different test polymers. Thus, possible variations of the adsorbed Fg layer density (slightly higher or lower than 1200 kg/m3) will not affect our general conclusions regarding the material-dependent differences in adsorbed mass, thickness, viscosity, and the possible orientation of surface-adsorbed fibrinogen molecules on the test surfaces.

Conclusions The QDM-D technique provided a powerful tool for the analysis of protein adsorption phenomena on polymeric surfaces. Detailed real-time adsorption studies showed material-dependent changes in adsorbed Fg mass, thickness, and viscoelastic properties. Our data show that the polymer structure of poly(DTE carbonate) can be modified by iodination and incorporation of PEG, resulting in biodegradable polymers which provided X-ray visibility while effectively decreasing human Fg adsorption to levels below those

3304 Langmuir, Vol. 23, No. 6, 2007

found on clinically used materials such as stainless steel, PLLA, PLGA, and PET. On the basis of these results in part, such modified tyrosine-derived polycarbonates were selected for the development of a fully resorbable polymer stent. Acknowledgment. This research has been supported by NIH Grant HL60416; RESBIO, an Integrated Technology Resource for Polymeric Biomaterials, a program funded by NIH Grant

Weber et al.

EB001046; the New Jersey Center for Biomaterials, and a SBIR Grant (1 R43 HLO75925-01) for Radio-opaque Resorbable Stents. The authors thank Prof. Fredrik Ho¨o¨k (Lund University, Lund Sweden) and Mr. Patrik Bjo¨o¨rn (Q-Sense AB, Go¨teborg, Sweden). We also thank Dr. Paul F. Holmes for his assistance in completing this manuscript. LA060500R