Functional Reconstitution of Thrombomodulin within a Substrate

Functional Reconstitution of Thrombomodulin within a. Substrate-Supported Membrane-Mimetic Polymer Film. June Feng, Po-Yuan Tseng, Keith M. Faucher, ...
0 downloads 0 Views 264KB Size
Langmuir 2002, 18, 9907-9913

9907

Functional Reconstitution of Thrombomodulin within a Substrate-Supported Membrane-Mimetic Polymer Film June Feng, Po-Yuan Tseng, Keith M. Faucher, Janine M. Orban, Xue-Long Sun, and Elliot L. Chaikof* Laboratory for Biomolecular Materials Research, Departments of Surgery and Biomedical Engineering, Emory University School of Medicine, Atlanta, Georgia 30322, and School of Chemical Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332 Received May 9, 2002. In Final Form: August 22, 2002 A stable, protein C activating membrane-mimetic film was produced on a polyelectrolyte multilayer (PEM) by in-situ photopolymerization of a phospholipid assembly containing thrombomodulin (TM). The monoacrylated phospholipid monomer was initially synthesized and prepared as unilamellar vesicles with varying molar concentrations of TM. Notably, in mixed-lipid systems, Km values for protein C activation increased in direct proportion to the mole fraction of polymerizable lipid, which was likely due to reduced membrane mobility after photopolymerization. Membrane-mimetic films were also constructed on planar substrates with predictable surface concentrations of catalytically active TM. Significantly, at a TM surface concentration of 170 fmol/cm2, the rate of protein C activation was comparable to that measured for a variety of confluent endothelial cell monolayers. Serial measurements of contact angles and protein C activation confirmed short-term film stability under a variety of in vitro conditions. Moreover, 125I labeling of TM demonstrated little change in TM surface concentration over periods of up to 28 days. Significantly, polymeric lipid membranes functionalized with thrombomodulin efficiently inhibited thrombin generation. We believe that the design of membrane-mimetic films that have the capacity to activate the protein C pathway will provide a useful strategy for generating “actively” antithrombogenic surfaces.

Introduction The control of thrombus formation on molecularly engineered surfaces will be an important step in the development of a small diameter arterial prosthesis critical to the fields of cardiac, plastic, and vascular surgery, as well as to the successful implantation of artificial organs and metabolic support systems. It has been postulated that a clinically durable vascular prosthesis may be achievable by identifying and incorporating, at the bloodmaterial interface, physiologically relevant antithrombogenic mechanisms that are normally operative under a range of hemodynamic conditions. Characteristically, the inhibition of blood coagulation is primarily achieved by two complementary mechanismssserine proteinase inhibitors, such as antithrombin III (ATIII), also known as “serpins” and the protein C pathway.1 Despite the presence of serpin binding sites on heparan sulfates and the well-characterized anticoagulant properties of these glycosaminoglycans, the physiological significance of the anticoagulant/antithrombotic functions attributed to heparan sulfates at the endothelial cell surface have not been conclusively established. For example, high-affinity ATIII binding sites have not been localized to heparan sulfates that are in direct contact with blood.2 Moreover, Lollar et al.3 could not confirm the alleged catalytic effect of heparan sulfate on the ATIII-thrombin reaction in recirculating rabbit Langendorff heart preparations. In contrast, there is growing evidence that thrombomodulin (TM), as a critical regulator of the protein C pathway, represents * To whom correspondence should be addressed at 1639 Pierce Drive 5105 WMB, Emory University, Atlanta, GA 30322. Phone: 404 727-8413. Fax: 404 727-3660. E-mail: [email protected]. (1) Bourin, M. C.; Lindahl, U. Biochem. J. 1993, 289, 313. (2) de Agostini, A. I.; Watkins, S. C.; Slayter, H. S.; Youssoufian, H.; Rosenberg, R. D. J. Cell. Biol. 1990, 111, 1293. (3) Lollar, P.; MacIntosh, S. C.; Owen, W. G. J. Biol. Chem. 1984, 259, 4335.

the major anticoagulant mechanism that is operative in both normal and injured blood vessels under physiologic conditions in vivo.4 TM is a 60 kD type I transmembrane protein that provides high-affinity binding sites for thrombin at the luminal surface of the vascular endothelium.5-9 In forming a 1:1 molar complex with thrombin, TM not only inactivates thrombin by an ATIII-mediated mechanism but also markedly enhances thrombin’s ability to activate protein Csa potent inhibitor of coagulation factors Va and VIIIa.10-12 Thus, TM switches off all known procoagulant/ proinflammatory functions of thrombin and instead channels the catalytic power of the enzyme into complex anticoagulant/antiinflammatory activities. While TM is a critical mediator of protein C activation, the lipid bilayer in which it resides serves as an essential “cofactor”, locally concentrating and coordinating the appropriate alignment of reacting cofactors and substrates. In concert with TM, the membrane accelerates protein C activation and subsequently optimizes activated protein C (APC) anticoagulant activity. Given this molecular framework, we believe that a substrate-supported membrane-mimetic assembly that contains TM, as an activator of the endogenous protein C anticoagulant (4) Kalafatis, M.; Egan, J. O.; van’t Veer, C.; Cawthern, K. M.; Mann, K. G. Crit. Rev. Eukaryotic Gene Expression 1997, 7, 241. (5) Esmon, N. L.; Owen, W. G.; Esmon, C. T. J. Biol. Chem. 1982, 257, 859. (6) Esmon, C. T.; Gu, J. M.; Xu, J.; Qu, D.; Stearns-Kurosawa, D. J.; Kurosawa, S. Haematologica 1999, 84, 363. (7) Esmon, C. T.; Ding, W.; Yasuhiro, K.; Gu, J. M.; Ferrell, G.; Regan, L. M.; Stearns-Kurosawa, D. J.; Kurosawa, S.; Mather, T.; Laszik, Z.; Esmon, N. L. Thromb. Haemostasis 1997, 78, 70. (8) Owen W. G.; Esmon, C. T. J. Biol. Chem. 1981, 256, 5532. (9) Esmon, C. T.; Owen W. G. Proc. Natl. Acad. Sci. (U.S.A.) 1981, 78, 2249. (10) Esmon, C. T. J. Biol. Chem. 1989, 264, 4743. (11) Esmon, C. T. Biochim. Biophys. Acta 2000, 1477, 349. (12) Esmon, C. T. FASEB J. 1995, 9, 946.

10.1021/la025931y CCC: $22.00 © 2002 American Chemical Society Published on Web 11/07/2002

9908

Langmuir, Vol. 18, No. 25, 2002

Feng et al.

Figure 1. Schematic representation of a polymeric phospholipid monolayer supported on an alkylated polyelectrolyte multilayer.

pathway, provides a rational design strategy for generating an actively antithrombogenic blood contacting interface. Noncovalently associated planar lipid assemblies, in and of themselves, are insufficiently robust for medical implant application.13-15 As a consequence, we believe that polymerization of the lipid provides at least one route to the generation of stabilized, chemically heterogeneous and biologically active surfaces, which closely mimic the structure of cell membranes. In this report, we describe the reconstitution of thrombomodulin into polymerizable lipid assemblies composed of 1-palmitoyl-2-[12-(acryloyloxy)dodecanoyl]-sn-glycero-3-phosphorylcholine (monoacrylate-PC)16-19 (Figure 1). The catalytic efficiency of TM within these membranes and the effect of the polymerization process on TM activity are reported herein. Moreover, the capacity to generate systems that maintain persistent catalytic activity under a variety of short-term in vitro conditions is demonstrated. Significantly, the ability to produce membrane-mimetic surfaces that are capable of activating protein C at rates that are comparable to or exceed those of native endothelial cells has been achieved and the abrogation of an induced thrombin generation response confirmed. Experimental Methods Materials. Rabbit lung thrombomodulin (in 0.1% lubrol PX), human factors Va, Xa, and prothrombin were obtained from Haematologic Technologies, Inc. Human protein C and human thrombin were obtained from Calbiochem. Human antithrombin III, activated protein C, and Spectrozyme PCa substrate (H-DLys(γ-carbobenzoxy)-Pro-Arg-pNA‚2AcOH) were obtained from (13) Winger, T. M.; Chaikof, E. L. Langmuir 1998, 14, 4148. (14) Winger, T. M.; Chaikof, E. L.; Ludovice, P. J. Langmuir 1998, 14, 5255. (15) Winger, T. M.; Ludovice, P. J.; Chaikof, E. L. Langmuir 1999, 15, 3866. (16) Orban, J. M.; Faucher, K. M.; Dluhy, R. A.; Chaikof, E. L. Macromolecules 2000, 33, 4205. (17) Chon, J. H.; Marra, K. G.; Chaikof, E. L. J. Biomater. Sci., Polym. Ed. 1999, 10, 95. (18) Sells, T. D.; O’Brien, D. F. Macromolecules 1994, 27, 226. (19) Sun, X. L.; Liu, H.; Orban, J. M.; Sun, L.; Chaikof, E. L. Bioconjugate Chem. 2001, 12, 673.

American Diagnostica Inc. S-2238 was purchased from Chromogenix. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) was obtained from Avanti Polar Lipids, Inc., and used as received. Eosin Y (EY, 5% in water), triethanolamine (TEA), and 1-vinyl2-pyrrolidinone (VP) were obtained from Aldrich. Nucleopore polycarbonate filters, circular glass coverslips (15 mm diameter, 0.17 mm thickness), and Contrad 70 detergent were obtained from Fisher. Kelco alginate (alg) was obtained from ISP alginates. Poly(L-lysine) (PLL, MW ∼400 kD) and all buffer salts were obtained from Sigma. Monoacrylate-PC (1-palmitoyl-2-[12-(acryloyloxy)dodecanoyl]-sn-glycero-3-phosphocholine) was synthesized as described previously.20 The synthesis of the terpolymer that consists of 3:6:1 3-acryloyl-3-oxapropyl-3-(N,N-dioctadecylcarbamoyl)propionate):2-hydroxyethyl acrylate:sodium styrene sulfonate (AOD3:HEA6:SS1) has been described elsewhere.21 Cell Culture. Three types of endothelial cells were examined. Bovine aortic endothelial cells (BAECs, passage 4, Cell Systems) were cultured in medium 199 (Gibco) 10% fetal bovine serum (FBS) (Hyclone), amino acid solution (Gibco), and basal medium Eagle vitamin solution (Gibco). Human dermal microvascular endothelial cells (HDMECs, passage 2, Emory Skin Diseases Research Center) were cultured in MCDB131 media (Mediatech) containing 20% human serum (Irvine Scientific), EGF, P/S/F (penicillin/streptomycin/ fungizone, Gibco), L-glutamine (Mediatech), hydrocortisone, and cAMP (Sigma). Human umbilical vein endothelial cells (HUVECs, passage 3, Emory Skin Diseases Research Center) were cultured in medium 199 (Gibco) with Eagle’s salt, 20% FBS, L-glutamine, endothelial mitogen (Biomedical Technologies, Inc.), heparin (Sigma), and P/S/F. The protein C activation rate was measured on postconfluent BAECs (passage 7), HDMECs (passage 3), and HUVECs (passage 6) plated onto gelatin-coated coverslips. The activation mixture, as described below (Analysis of TM Catalytic Activity), was added onto cell-coated substrates for 1 h activation, quenched by ATIII, and assayed by Spectrozyme PCa substrate for 30 min. Instrumentation. Irradiation was performed using a DynaLume quartz halogen illuminator equipped with a heat shield (Scientific Instruments). Light intensity was measured using a radiometer model IL 1400A equipped with a SL021 photodetector (20) Marra, K. G.; Winger, T. M.; Hanson, S. R.; Chaikof, E. L. Macromolecules 1997, 30, 6483. (21) Liu, H.; Faucher, K. M.; Sun, X.-L.; Feng, J.; Johnson, T. L.; Orban, J. M.; Apkarian, R. P.; Dluhy, R. A.; Chaikof, E. L. Langmuir 2002, 18, 1332.

Functional Reconstitution of Thrombomodulin and FQI filter (International Light). Contact angle measurements were performed using a Rame´-Hart NRL contact angle goniometer model 100-00-115 with 0.45 µm filtered water as the wetting solvent. Measurements are reported as average advancing/receding degree ( standard deviation of 10 data points. All spectrophotometric measurements were executed on a Cary 50 Bio UV-visible spectrophotometer (Varian) equipped with a temperature-regulated cell compartment. Sucrose gradient centrifugation was conducted in a Beckman L7 ultracentrifuge. Layer-by-Layer Assembly of an Alkylated Alginate/Polyl-lysine (Alg/PLL) Multilayer on a Glass Surface. Coverslip substrates were cleaned by 30 min of sonication in 10% Contrad 70 detergent solution followed by extensive washing and sonication in deionized water. The substrate was then immersed in a solution of 0.10% PLL in 20 mM PBS for 30 s and then washed with water. The procedure was repeated using a solution of 0.15% Alg in 20 mM PBS. This process was repeated until a total of 11 layers (six PLL and five alginate layers) were coated, with the top layer comprised of PLL. Samples were then immersed in a 0.1 mM (in [SS]) terpolymer (AOD3:HEA6:SS1) solution in 1% (v/v) DMSO/PBS buffer for 1 min. After extensive rinsing with water, samples were air-dried. Advancing contact angles of >100° were characteristic of these substrates. Comprehensive surface analysis of these films with or without a supported lipid assembly, including contact angle goniometry, ellipsometry, external reflectance infrared spectroscopy, and high-resolution scanning electron microscopy, has been detailed elsewhere.21 Incorporation of TM into Lipid Vesicles. Large unilamellar vesicles (LUV) of 12 mM lipid solution in 20 mM sodium phosphate buffer, pH 7.4, were prepared by four successive freeze/ thaw/vortex cycles using liquid N2 and a 45 °C water bath. Thrombomodulin was then added to obtain the desired thrombomodulin/lipid molar ratio, which ranged in these investigations from 1:8000 to 1:200 000. The lipid/thrombomodulin solution was gently vortexed for 1 h at room temperature before it was extruded 21 times, each through two back-to-back 2000 nm and then 600 nm polycarbonate filters. Sucrose density centrifugation was carried out to assess the extent to which TM was incorporated within the lipid membrane after vesicle formation. Lipid vesicles containing reconstituted thrombomodulin were layered on top of a 6 mL 5-30% discontinuous sucrose gradient prepared in TBS and centrifuged for 16 h at 130 000g. Aliquots of 0.6 mL were transferred into microcentrifuge tubes, and TM activity in each fraction was determined. Formation of a TM-Containing Membrane-Mimetic Thin Film. Extruded lipid vesicles were diluted to 1.2 mM with 20 mM sodium phosphate buffer, pH 7.4, and a final salt concentration of 150 mM NaCl was achieved using 750 mM NaCl in water. This solution was then purged with argon for 20 min. A terpolymer-coated glass coverslip was then added to a scintillation vial containing 1.2 mL of the vesicle solution. The coverslip was completely immersed in vesicle solution and faced upward. The vial was quickly sealed and maintained overnight at 40 °C. Photopolymerization of acrylic-PC was carried out as described previously22 with some modifications. A stock solution of coinitiators was prepared as 10 mM Eosin Y, 225 mM TEA, and 37 mM VP in water and was stored in an opaque amber bottle. In a glovebag purged with argon, the desired amount of initiator stock solution was added to the vial containing the substrate fused with vesicles so that a 12:1 ratio of [acrylic-PC monomer]: [Eosin Y] was achieved. The sample was irradiated for 30 min under ambient conditions from above at a distance of approximately 6 cm (light intensity ∼ 40 mW/cm2). Following the photopolymerization period, the sample was removed from the polymerization media and washed extensively with water. Analysis of TM Catalytic Activity. Activity of TM was accessed via the activation of human protein C by human thrombin-rabbit lung thrombomodulin complex, as described in the literature with some modification.23 Activation was performed at 37 °C in 20 mM Tris-HCl buffer, pH 7.5, containing (22) Orban, J.; Faucher K.; Dluhy, R. A.; Chaikof, E. L. Macromolecules 2000, 33, 4205. (23) Esmon, N. L.; Debault, L. E.; Esmon, C. T. J. Biol. Chem. 1983, 258, 5548.

Langmuir, Vol. 18, No. 25, 2002 9909 100 mM NaCl, 0.1% BSA, and 5 mM Ca2+. A typical activation mixture contains 0.5 nM of TM as a component of LUVs, 5 nM of thrombin, and 800 nM of protein C. Incubation periods were adjusted such that less than 10% of protein C was converted to activated protein C. Activation was terminated by the addition of antithrombin III (300 µg/mL final concentration). Activated protein C concentration was determined using the Spectrozyme PCa substrate (absorbance 405 nm). Km and kcat values were calculated assuming Michaelis-Menton reaction kinetics. Thrombomodulin activity on glass coverslips coated with a membrane-mimetic thin film was determined in similar fashion in a shaking water bath at 37 °C. For stability tests, samples were incubated in PBS at 4 °C or human plasma at 37 °C for specified time periods prior to measuring TM activity. Fresh samples were used for all stability tests, and a total of six samples were assayed to obtain each data point. Determination of TM Surface Density and Stability within a Polymeric Lipid Membrane. Rabbit lung thrombomodulin at a concentration of 1 mg/mL was radiolabeled with 125I using Iodobeads (Pierce Chemical Co., Rockford, IL) according to the manufacture’s protocol. The concentration of radiolabeled TM was determined by a modification of the Bradford Protein Assay (Bio-Rad) using a TM standard curve. Specific activity of the protein solution averaged 1.8 × 107 cpm/µg (1330 cpm/fmol). Large unilamellar vesicles of 12 mM lipid solution in 20 mM sodium phosphate buffer, pH 7.4, were prepared by four successive freeze/thaw/vortex cycles using liquid N2 and a 45 °C water bath. 125I-TM and unlabeled TM were mixed to form 1:16 molar ratio solution, and TM was then added to obtain the desired thrombomodulin/lipid molar ratio. The lipid/thrombomodulin solution was gently vortexed for 1 h at room temperature before it was extruded 21 times, each through two back-to-back 2000 nm and then 600 nm polycarbonate filters. Planar photopolymerized lipid/TM assemblies were produced as described above and washed extensively with water prior to measuring surface radioactivity in a γ counter. TM surface density (fmol/cm2) was calculated as {cpm/(specific activity × sample area)} and specific activity corrected for decay. For stability analysis, samples were stored in PBS at 37 °C and radioactivity periodically measured after extensive washing in water over a 1-month period. Test samples were generated in quadruplicate. Defining the Capacity of TM-Containing MembraneMimetic Thin Films To Inhibit de Novo Thrombin Generation. TM was incorporated into acrylate-PC vesicles fused and photopolymerized onto an alkylated substrate, as described above. A defined “prothrombin activation” mixture containing 0.1 mM 25% phosphatidylserine/75% phosphatidylcholine lipid vesicles, 250 pM human factor Xa, 2.5 pM human factor Va (limiting reagent), 0.1 µM human protein C, and 5 mM Ca2+ was prepared in Tris buffer saline (50 mM Tris, 175 mM NaCl, 0.5 mg/mL BSA, pH 7.9). TM-containing membrane-mimetic thin films were added to the activation mixture and the samples incubated at 37 °C in a shaking water bath (100 rpm). Thrombin production was initiated by the addition of 200 nM (final concentration) of human prothrombin. Serial aliquots were obtained, quenched with 20 mM EDTA in Tris buffer for 1-2 min, and thrombin concentration measured using the thrombinspecific chromogenic substrate, S-2238 (absorbance 405 nm). For comparative purposes, this analysis was repeated with free TM. Briefly, a lipid mix of 0.1 mM of 25% phosphatidylserine/ 75% phosphatidylcholine in Tris-buffered saline was sonicated for 4 min. Defined concentrations of TM (60 vs 100 nM) were then added, and the mixture was gently vortexed for 1 min. Vortexing was then repeated four times, after which the other components of the “prothrombin activation” mixture were added, as described above, and incubated for 15 min at 37 °C. Thrombin production was initiated by the addition of prothrombin, and serial measurements of thrombin generation were performed, as detailed previously.

Results and Discussion Since the mid-1980s investigators have noted that the phosphorylcholine headgroup appears to limit the induc-

9910

Langmuir, Vol. 18, No. 25, 2002

tion of blood clot formation on synthetic surfaces.24-26 Although not well understood, it has been speculated that this biological property may be related to the large amount of water bound to this zwitterionic headgroup or, conceivably, the selective adsorption to phosphorylcholine of specific plasma proteins that inhibit the blood clotting process.27 While our group has also observed limited thrombus formation and neointimal hyperplasia on phospholipid functionalized surfaces using short-term in vivo assays,20,28 we believe that the inherent strength of a membrane-mimetic-based approach is the capacity to incorporate within these systems a variety of physiologically significant processes relevant to the control of blood coagulation. As a critical regulator of the protein C pathway, TM represents the major anticoagulant mechanism that is operative under physiologic conditions in vivo and thereby provides an appropriate target for functionalizing blood contacting membrane-mimetic thin films. TM is Efficiently Reconstituted into Lipid Assemblies. TM was initially reconstituted into unilamellar phospholipid vesicles using an extrusion method. To minimize the loss of TM activity, the protein was added to the phospholipid solution prior to extrusion but after freeze-thaw-vortex cycles. Use of a sucrose gradient confirmed that over 95% of the TM activity was associated with the lipid vesicles, located at the top of the gradient. In contrast, only a small amount of residual free TM was present in the middle to the bottom portion of the gradient. This high level of reconstitution made further separation of free TM unnecessary. While dialysis-based approaches have been described as an alternative method for the incorporation of transmembrane proteins into lipid vesicles, it is noteworthy that the described extrusion/reconstitution method is significantly less time-consuming (3 h; cf. 36 h).29 Overall, TM was successfully reconstituted into a variety of mixed phospholipid vesicles ranging from 100% monoacrylate-PC to 100% POPC. Photopolymerization of a Lipid Assembly Influences the Rates of TM-Mediated Protein C Activation. Vesicles were exposed to visible light for 30 min in the presence of eosin Y/triethanolamine after TM was reconstituted into vesicles of varying POPC:monoacrylatePC molar ratio. All vesicle systems, regardless of lipid composition exhibited similar rates of protein C activation prior to polymerization. However, following photopolymerization a modest reduction in the protein C activation rate was noted (Figure 2A). We believe that this effect may be attributed to two factors. In part, a small reduction of TM activity was observed due to direct TM inactivation by free radicals generated during the photoinitiation process. In addition, the catalytic efficiency of TM was also diminished as a consequence of reduced lipid membrane mobility. Specifically, the decrease in the rate of protein C activation was greater when TM was reconstituted into vesicles composed of increasing concentration of polymerizable monoacrylate-PC lipids, despite similar concentration of free radical initiator (Figure 2B). TM-Mediated Protein C Activation is Reduced in Polymeric Lipid Assemblies Largely Due to an (24) Ishihara, K.; Tsuji, T.; Kurosaki, T.; Nakabayashi, N. J. Biomed. Mater. Res. 1994, 28, 225. (25) Hayward, J. A.; Chapman, D. Biomaterials 1984, 5, 135. (26) Hall, B.; Bird, R. R.; Kojima, M.; Chapman, D. Biomaterials 1989, 10, 219. (27) Chapman, D. Langmuir 1993, 9, 39. (28) Chen, C.; Ofenloch, J. C.; Yianni, Y. P.; Hanson, S. R.; Lumsden, A. B. J. Surg. Res. 1998, 77, 119. (29) Galvin, J. B.; Kurosawa, S.; Moore, K.; Esmon, C. T.; Esmon, N. L. J. Biol. Chem. 1987, 262, 2199.

Feng et al.

Figure 2. (A) Effect of the photoinitiation process on the rate of protein C activation (µM/min). TM was reconstituted at a concentration of 10 nM into either POPC or monoacrylate-PC vesicles, and initiator stock solution added at a 12:1 molar ratio of [PC]:[Eosin Y]. (B) Influence of lipid polymerization on the rate of protein C activation (µM/min). TM was reconstituted at a concentration of 10 nM into mixed POPC/monoacrylate-PC vesicles, and the lipid/initiator mix irradiated with visible light for 30 min. The data are representative of experiments performed in duplicate. Table 1. Catalytic Efficiency of Thrombomodulin as a Function of the Local Lipid Microenvironment monomer % mono% (M) or acrylate-PC POPC polymer (P) 0 100 80 50 100 80 50 a

100 0 20 50 0 20 50

M M M M P P P

Km (µM)

kcat (min-1)

kcat/Km (µM-1 min-1)

1.3 ( 0.1 1.1 ( 0.1 1.4 ( 0.1 1.0 ( 0.1 3.0 ( 0.2 3.0 ( 0.6 2.1 ( 0.4 4.2 ( 0.3

29 ( 1 30 ( 1 27 ( 1 28 ( 1 23 ( 1 18 ( 2 22 ( 1 39 ( 1

22.3 ( 2.5 27.3 ( 3.4 19.3 ( 2.1 28.0 ( 3.8 7.7 ( 0.8 6.0 ( 1.9 10.5 ( 2.6 9.3 ( 0.9

a Thrombomodulin (in 0.1% lubrol PX, Haematologic Technologies, Inc.) without added phospholipid.

Increase in Km. Kinetic parameters, kcat and Km, were obtained by nonlinear regression analysis of the rates of activated protein C production as a function of protein C concentration (Table 1). As anticipated, TM reconstituted into lipid vesicles exhibited Km values 3- to 4-fold lower than that of free TM. Although Km values were initially similar regardless of lipid composition, after photopolymerization Km values increased in direct proportion to the mole fraction of monoacrylate-PC. This was likely indicative of reduced molecular mobility as a consequence of lipid polymerization. Indeed, others have reported that by limiting the movement of TM and/or protein C within

Functional Reconstitution of Thrombomodulin

Figure 3. (A) Relationship between the rate of surface mediated protein C activation (nM/min/cm2) and the concentration of TM in the fusion mixture. (B) Rate of surface-mediated protein C activation (nM/(min/cm2)) for a membrane-mimetic thin film ([TM] ) 170 fmol/cm2) and confluent endothelial cell monolayers (HDMEC, human dermal microvascular endothelial cells; HUVEC, human umbilical vein endothelial cells; BAEC, bovine aortic endothelial cells).

a membrane, typically by using lipid constituents that are associated with a high melting transition temperature, the rate of protein C activation is reduced in a commensurate fashion.30,31 Robust Membrane-Mimetic Films can be Constructed on Planar Substrates with Predictable Surface Concentrations of Catalytically Active TM. A schematic of the membrane-mimetic construct is illustrated in Figure 1, and a detailed description of the surface physiochemical properties of this system have been published elsewhere.21 The molar ratio of TM:phospholipid was systematically varied from 1:8000 to 1:200 000 with an observed linear relationship between the concentration of TM in the lipid mixture and the rate of activated protein C formation (Figure 3A). Significantly, at a TM surface concentration of 170 fmol/cm2, the rate of protein C activation was comparable to that measured for a variety of confluent endothelial cell monolayers (Figure 3B). Contact angle goniometry was used as an indirect measure of film stability. After the polymerization of TM containing lipid films, advancing and receding contact angles were 61 ( 2 and 43 ( 2°, respectively, comparable to other membrane-mimetic systems developed in our laboratory.22 After 10 days of storage in PBS at 4 °C, (30) Mann, K. G.; Jenny, R. J.; Krishnaswamy, S. Annu. Rev. Biochem. 1988, 57, 915. (31) Smirnov, M. D.; Ford, D. A.; Esmon, C. T.; Esmon, N. L. Biochemistry 1999, 38, 3591.

Langmuir, Vol. 18, No. 25, 2002 9911

Figure 4. (A) Serial measurements (mean ( SD) of the rate of protein C activation for a TM-containing lipid film incubated in PBS at 4 °C. (B) Serial measurements (mean ( SD) of the rate of protein C activation for a TM-containing lipid film incubated in plasma at 37 °C. All values are normalized with respect to the initial rate protein C activation in PBS.

contact angles were unchanged (65 ( 7°/49 ( 7°). Moreover, TM activity was unchanged over a 5-day incubation period in PBS at 4 °C (Figure 4A). Incubation of films in plasma at 37 °C revealed a small decrease in TM activity (∼20%) within 30 min. Nonetheless, continued incubation for an additional 3 days was not associated with any further loss of TM activity (Figure 4B). Surface associated TM activity was ∼20% higher when films were generated from mixed lipid (monoacrylate-PC:POPC) vesicles, as compared to films produced solely from monoacrylate-PC (data not shown). However, partial loss of the lipid film, as assessed by contact angle measurements, was observed as early as 1 day after incubation in PBS at 4 °C. 125 I-protein labeling was used to determine the absolute surface density of TM on planar-supported membranes as a function of its molar concentration in the vesicle solution (Figure 5). A linear relationship was observed over a wide range of TM concentrations. A small decrease in surface concentration was noted within the first 3 days on incubation at 37 °C in PBS with relatively little change observed over a period extending up to 28 days. We speculate that the small decrease TM concentration soon after the initiation of PBS incubation may be related to the loss of partially adsorbed lipid vesicles. Notably, samples were removed from the bathing media on multiple occasions during the incuabtion period, which emphasizes the relative stability and robustness of this system. TM-Containing Vesicles and Polymeric Thin Films Inhibit de Novo Thrombin Generation. Thrombin was produced in a predictable manner upon the addition of prothrombin to a factor Va containing solution, and as anticipated, thrombin generation inhibited in the presence of both TM containing polymeric thin films and a TM/ lipid mixture (Figure 6). The rate of thrombin production

9912

Langmuir, Vol. 18, No. 25, 2002

Feng et al.

Figure 5. TM surface density and stability within a polymeric lipid membrane. TM was radiolabeled with 125I and reconstituted into lipid vesicles, and surface concentration was determined after the formation of a polymeric lipid assembly. The stability of this system was assessed in PBS at 37 °C over a 28-day period. Films were transferred to fresh PBS at indicated time points and radioactivity measured in a γ counter.

declined in both systems with complete inhibition achieved within 30-60 min. The observed lag in eliminating thrombin generation was greater for TM localized to planar films. However, it bears emphasis that the total amount of TM bound to membrane-mimetic surfaces ranged from 460 fmol (170 fmol/cm2) to 3600 fmol (1700 fmol/cm2), which is 20-200 times less than the total amount of TM present in the lipid vesicle system (60-100 nM). We suspect that the extent of the lag period is largely dependent on both the total amount of accessible TM, as well as mass transfer limitations. Covalent immobilization of TM onto polymeric surfaces has been previously described by several investigators. For example, Kishida et al.32-35 conjugated TM to both aminated and carboxylated surfaces, including poly(vinylamine) and poly(acrylic acid) surface-grafted poly-

ethylene and a surface-hydrolyzed poly(ether urethaneurea). Similarly, Vasilets et al.36 reported the immobilization of TM onto poly(acrylic acid) surface-grafted PTFE. In all cases, the conjugation scheme utilized a carbodiimide-based coupling reaction in which TM was coupled to the substrate via freely available amino or carboxyl (32) Kishida, A.; Ueno, Y.; Maruyama, I.; Akashi, M. Biomaterials 1994, 15, 1170. (33) Kishida, A.; Ueno, Y.; Maruyama, I.; Akashi, M. ASAIO J. 1994, 40, M840. (34) Kishida, A.; Ueno, Y.; Fukudome, N.; Yashima, E.; Maruyama, I.; Akashi, M. Biomaterials 1994, 15, 848. (35) Kishida, A.; Akatsuka, Y.; Yanagi, M.; Aikou, T.; Maruyama, I.; Akashi, M. ASAIO J. 1995, 41, M369. (36) Vasilets, V. N.; Hermel, G.; Konig, U.; Werner, C.; Muller, M.; Simon, F.; Grundke, K.; Ikada, Y.; Jacobasch, H. J. Biomaterials 1997, 18, 1139.

Functional Reconstitution of Thrombomodulin

Langmuir, Vol. 18, No. 25, 2002 9913

protein C activation were enhanced, and this activity appeared to be directly proportional to TM surface density, as determined by a ninhydrin assay. However, the ability to control TM surface concentration was substrate dependent with reported TM densities ranging between 0.15 and 0.45 µg/cm2 (∼2000-6000 fmol/cm2). Furthermore, TM bioactivity was significantly reduced after surface coupling. We believe that compromised TM activity may have been attributable to both the protein immobilization procedure, which coupled any accessible functional group to the test surface, and the absence of an associated lipid membrane. In contrast, the studies reported herein demonstrate that TM can be predictably incorporated into a stable, membrane-mimetic thin film over a wide range of surface concentrations by a process of lipid/protein selfassembly and in situ photopolymerization. While some reduction in the catalytic efficiency of TM was observed, due to reduced membrane mobility and direct photoinactivation, the capacity to activate protein C was largely retained. In the process, the capacity of surface-bound TM to completely eliminate thrombin generation was confirmed.

Figure 6. (A) Rate of thrombin generation (nM/min) as a function of time in the presence of TM/lipid vesicles. (B) Rate of thrombin generation (nM/min) as a function of time in the presence of TM-containing planar membrane-mimetic film.

functionalities on the protein surface. As anticipated, in vitro studies demonstrated that both clotting time and

Acknowledgment. This work was supported by grants from the NIH, the Juvenile Diabetes Research Foundation International, and the Molecular Design Institute at the Georgia Institute of Technology. The authors wish to acknowledge the Emory University Mass Spectrometry Center provided by grants from the NIH and NSF and helpful discussions with Professor Charles Esmon of the Howard Hughes Institution at the Oklahoma Medical Research Foundation. LA025931Y