@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Medicine, Faculty of"@en, "Biochemistry and Molecular Biology, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Krisinger, Michael J."@en ; dcterms:issued "2011-02-03T21:58:36Z"@en, "2007"@en ; vivo:relatedDegree "Doctor of Philosophy - PhD"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """Haemostasis is a highly regulated, fundamental, physiological process featuring numerous peripheral membrane proteins. Of these, the membrane and calcium binding properties of the vitamin K-dependent proteins are dependent on a common N-terminal γ-carboxyglutamic acid (Gla)-containing domain. Previous work on Gla proteins has provided a wealth of affinity and kinetic membrane binding information. These studies have employed a number of biophysical techniques using artificial phosphatidylserine-containing model membranes. However, many aspects of the membrane binding interaction, in terms of mechanism and modulation by protein cofactor remain obscure. This thesis examines two methods for studying the membrane binding properties of human plasma derived Gla proteins with emphasis on prothrombin. In Chapter 3 differential centrifugation combined with immunoaffinity detection was used to quantify the effect the cofactor Factor Va had on the membrane binding affinity of prothrombin for membrane. Factor Va bound to anionic phospholipid membrane undoubtedly enhanced the membrane binding affinity of prothrombin relative to prothrombin binding in the absence of the cofactor. Thus, these results indicate that Factor Va can recruit prothrombin or prethrombin 1, a Gla-domain less fragment of prothrombin, to the membrane surface, plausibly contributing to its cofactor function. In Chapters 4 and 5, surface plasmon resonance (SPR) was used to evaluate the Ca²⁺-specific binding properties of a number of Gla proteins to immobilized membranes. Membrane affinity, molar binding preference and kinetics controlling complex formation and complex breakdown varied widely between Gla proteins. The comparative results obtained by SPR indicate that the majority of homologous Gla proteins bind membranes with a complex mechanism which may involve membrane induced protein dimers. Unlike prothrombin, the binding profiles for fragment 1 and fragment 1.2 fitted closely to a one-site binding model. Apparent biphasic association and biphasic dissociation phases were observed for prothrombin and commonly amongst the other Gla proteins at a wide range of protein concentrations including physiological concentrations. For prothrombin, dimerization appears to be specific to the protease domain as neither fragment 1 nor fragment 1.2 displays such binding complexities. It is possible that dimerization increases the half-life of membrane-bound Gla proteins thereby promoting their participation in complex assembly and function."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/31087?expand=metadata"@en ; skos:note "Membrane Binding Properties of Prothrombin and other Gamma-Carboxyglutamic Acid-Containing Coagulation Proteins by Michael J. Krisinger B.Sc., University of Calgary, 2001 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Biochemistry and Molecular Biology) THE UNIVERSITY OF BRITISH COLUMBIA January 2007 © Michael J. Krisinger, 2007 Abstract Haemostasis is a highly regulated, fundamental, physiological process featuring numerous peripheral membrane proteins. Of these, the membrane and calcium binding properties of the vitamin K-dependent proteins are dependent on a common N-terminal y-carboxyglutamic acid (Gla)-containing domain. Previous work on Gla proteins has provided a wealth of affinity and kinetic membrane binding information. These studies have employed a number of biophysical techniques using artificial phosphatidylserine-containing model membranes. However, many aspects of the membrane binding interaction, in terms of mechanism and modulation by protein cofactor remain obscure. This thesis examines two methods for studying the membrane binding properties of human plasma derived Gla proteins with emphasis on prothrombin. In Chapter 3 differential centrifugation combined with immunoaffinity detection was used to quantify the effect the cofactor Factor Va had on the membrane binding affinity of prothrombin for membrane. Factor Va bound to anionic phospholipid membrane undoubtedly enhanced the membrane binding affinity of prothrombin relative to prothrombin binding in the absence of the cofactor. Thus, these results indicate that Factor Va can recruit prothrombin or prethrombin 1, a Gla-domain less fragment of prothrombin, to the membrane surface, plausibly contributing to its cofactor function. In Chapters 4 and 5, surface plasmon resonance (SPR) was used to evaluate the Ca2+-specific binding properties of a number of Gla proteins to immobilized membranes. Membrane affinity, molar binding preference and kinetics controlling complex formation and complex breakdown varied widely between Gla proteins. The comparative results obtained by SPR indicate that the majority of homologous Gla proteins bind membranes with a complex mechanism which may involve membrane induced protein dimers. Unlike prothrombin, the binding profiles for fragment 1 and fragment 1.2 fitted closely to a one-site binding model. Apparent biphasic association and biphasic dissociation phases were observed for prothrombin and commonly amongst the other Gla proteins at a wide range of protein concentrations including physiological concentrations. For prothrombin, dimerization appears to be specific to the protease domain as neither fragment 1 nor fragment 1.2 displays such binding complexities. It is possible that dimerization increases the half-life of membrane-bound Gla proteins thereby promoting their participation in complex assembly and function. Table of Contents Abstract ii Table of Contents iv List of Tables vii List of Figures viii List of Abbreviations x Acknowledgements xiii Chapter 1. Introduction 1 1.1 Blood coagulation 1 1.2 Gla proteins ....3 1.2.1 Prothrombin 3 1.2.2 Other Gla proteins 5 1.3 Procoagulant membranes 8 1.4 Gla protein membrane interaction 11 1.5 Techniques to study protein-membrane interactions 15 1.5.1 Equlibrium binding 15 1.5.2 Kinetic analysis 17 1.6 Surface plasmon resonance 18 1.6.1 Overview 18 1.6.2 Optical Configuration and Detection Principles 20 1.7 Objectives and Overview 21 Chapter 2. Methods 24 2.1 Materials, proteins and miscellaneous reagents 24 2.2 Liposome preparation 25 2.3 SDS-PAGE electrophoresis, western blotting and densitometry 26 2.4 Differential centrifugation (FVa mediated binding) 28 2.5 Thrombin generation assay 29 2.6 Prothrombin enzymatic digestion 30 2.7 Amino-terminal sequence analysis 30 2.8 Surface plasmon resonance: 31 2.8.1 Membrane immobilization 31 2.8.2 Protein binding experiments 31 2.8.3 Data analysis 32 Chapter 3. Effect of Factor Va on Prothrombin-Membrane Interaction 34 3.1 Rationale 34 3.2 LV characterization by prothrombinase activity 35 3.3 Factor Va mediated prothrombin binding: immunoaffinity quantification 37 3.4 Determination of -50 kDa species 46 3.5 Discussion 48 Chapter 4. Mechanism of Prothrombin-Membrane Interaction 50 4.1 Rationale 50 4.2 Membrane immobilization and stability 51 4.3 Validation of interactive membrane surface prior to detailed kinetic analysis 54 4.3.1. BSA binding .. 54 4.3.2. Mass transport 56 4.3.3. Specificity of membrane interaction 57 4.4 Fl, F1.2 and prothrombin membrane binding: Affinity and qualitative kinetic analysis.. 60 4.5 DOPS dependence of prothrombin binding 63 4.5.1 Non-cooperative binding 67 4.6 On the binding mechanism of prothrombin 69 4.6.1 Prothrombin membrane binding involves a linked reaction mechanism 69 4.6.2 Stability of membrane bound species is dependent on prothrombin concentration.... 70 4.6.3 Analysis of cross-linked prothrombin dimer membrane binding 71 4.6.4 Homogeneous Analyte 74 4.7 Estimation of kinetic parameters 75 4.8 Inhibition of prothrombin membrane binding by des-Gla prothrombin 78 4.9 Discussion 81 4.9.1 The prothrombin-membrane interaction is biphasic 81 4.9.2 Fast prothrombin-membrane phases are mediated by 82 fragment 1 82 4.9.3 Dimerization model for slow prothrombin-membrane binding phases 82 4.9.4 Linked mechanism for prothrombin-membrane interaction 83 4.9.5 Previous studies concur with membrane induced dimerization 85 4.9.6 Apparent discrepancies from previous studies about the secondary kinetic event 86 4.9.7 Other models to explain prothrombin membrane binding 87 4.9.8 Significance of prothrombin dimerization 94 Chapter 5. Comparison of Coagulation Gla Protein-Membrane Interactions 96 5.1 Rationale 96 5.2 Results 99 5.2.1 Purity analysis by SDS-PAGE 99 5.2.2 Gla protein binding profile to 25% DOPS-containing membrane 101 5.3 Discussion 110 5.3.1 Binding comparison of Gla proteins 110 5.3.2 Zymogen versus activated enzyme 113 5.3.3 Membrane binding is influenced by regions outside of Gla 115 5.3.4 Vitamin K-dependent protein dimerization 115 Chapter 6. Summary and General Discussion 118 Chapter 7. Future Directions 122 Bibliography 126 List of Tables Table 1 Comparison of vitamin K-dependent Gla proteins involved in blood coagulation 6 Table 2 Binding affinity of prothrombin, fragment 1, fragment 1.2 and xl-prothrombin to DOPS-containing membrane 65 Table 3 Parameters obtained from global fitting 77 Table 4 Binding models for peripheral membrane proteins 89 Table 5 Prothrombin membrane induced dimerization: Is there another model or process to explain the second kinetic event involved in membrane binding? 91 Table 6 Comparison of membrane binding properties of Gla proteins assessed by SPR 107 Table 7 Dissociation constants determined for Protein C I l l List of Figures Figure 1 Domains and fragments of prothrombin 3 Figure 2 Immobilization strategy and detection principle of SPR 19 Figure 3 Prothrombinase activity of LUV and LV phospholipid surfaces as a function of total phospholipid concentration 36 Figure 4 Optimization of differential centrifugation experiments: Effect of large vesicle concentration 39 Figure 5 Factor Va-dependent prothrombin binding to BPS/EPC (25%/75%) large vesicles.... 41 Figure 6 Prothrombin concentration standard curve 42 Figure 7 Prothrombin membrane binding: dependence on large vesicle and Factor Va 44 Figure 8 Factor Va mediated prothrombin binding to BPS/EPC (25/75) large vesicles assessed by differential centrifugation coupled with immunoaffinity detection 45 Figure 9 Cleavage analysis of prothrombin by thrombin 47 Figure 10 Liposome immobilization using the Biacore Sensor Chip LI 52 Figure 11 Baseline stability of immobilized membrane and reference surfaces over time 54 Figure 12 BSA binding to immobilized DOPS-containing membrane 55 Figure 13 Prothrombin membrane interaction under variable flow rate 57 Figure 14 Protein and DOPS dependent binding specificity 59 Figure 15 Fragment 1, fragment 1.2 and prothrombin binding profiles to 25% DOPS-containing membrane 61 Figure 16 Prothrombin binding isotherms at different percentage DOPS membrane concentration 64 Figure 17 Cooperativity analysis for prothrombin interaction with DOPS-containing membrane. 68 Figure 18 Membrane dissociation of prothrombin is dependent on duration of association phase. 70 Figure 19 Initial extent of membrane dissociation is dependent on prothrombin concentration. 71 Figure 20 xl-prothrombin dimer binding profile to 25% DOPS-containing membrane 73 Figure 21 Characterization of proteins by SDS-polyacrylamide gel electrophoresis 75 Figure 22 Determination of kinetic parameters of prothrombin and derivatives binding to 25% DOPS containing membrane using SPR 76 Figure 23 Des-Gla prothrombin inhibition of prothrombin membrane interaction 80 Figure 24 Model for prothrombin membrane binding based on a linked reaction mechanism involving membrane induced dimerization of the Protease domain 84 Figure 25 Amino acid sequence alignment of the Gla domain of human Gla proteins involved in blood coagulation 97 Figure 26 Structural comparison of three homologous Gla domains 98 Figure 27 Characterization of proteins by SDS-polyacrylamide gel electrophoresis 101 Figure 28 Gla protein binding profiles to 25% DOPS-containing membrane 105 List of Abbreviations AP activation peptide APC activated protein C a-PT mouse monoclonal antibody against kringle 2 of prothrombin BCA bicinchoninic acid BPS bovine brain phosphatidyl-serine BSA bovine serum albumin DOPS 1,2-dioleoyl-.s'tt-glycero-3-[phospho-L-serine] EDTA ethylenediaminetetraacetic acid EGF epidermal growth factor EPC egg L-lecithin-(phosphatidylcholine) F1 fragment 1 F1.2 fragment 1.2 FYa Factor Va FRET fluorescence resonance energy transfer Gla y-carboxyglutamic acid HBS hepes buffered saline HBC HBS plus BSA plus calcium HEPES 4-(2-Hydroxyethyl)piperazine-l -ethanesulfonic acid HRP-2°Ab horseradish peroxidase conjugated goat anti-mouse IgG Kd dissociation constant rsapp ^d apparent dissociation constant LUV large unilamellar vesicle LV large vesicle MLV multilamellar vesicle Mr molecular weight PC phosphatidylcholine PEG polyethylene glycol POPC 1 -palmitoyl-2-oleoyl-.w-glycero-3-phosphocholine Prel pre thrombin 1 PS phosphatidylserine QCM-D quartz crystal microbalance with dissipation R m a x maximum response RU response unit S-2238 //-D-phenylalanyl-L-pipecolyl-L-arginine-P-nitroanilide SDS-PAGE sodium dodecyl sulphate - polyacrylamide gel electrophoresis SHBG sex hormone binding globulin SPR surface plasmon resonance SUV small unilamellar vesicle TIR total internal reflection TIR-FM TIR-fluorescence microscopy TIR-FPR TIR-fluorescence photobleaching recovery TBS tris buffered saline TBS-T TBS plus Tween-20 TRIS tris (Hydroxymethyl) aminomethane TSR thrombin sensitive region xl-prothrombin cross-linked prothrombin dimer ZPI protein Z dependent protease inhibitor Acknowledgements I would like to thank a number of people that have helped me out during my time here at UBC. I greatly appreciated the scientific discussions I had during my directed studies course with Drs. Dana Devine and Ed Pryzdial. They also provided great support on my project as committee members. A special thanks to Ed for his help and guidance that helped make my project productive. And also, for allowing me to take part at countless of his lab meeting - an area where I was able to soak up a lot of knowledge. I like to thank the numerous and enlightening conversations I have had with members of the MacGillivray Lab, past and present, and for their support and camaraderie. Special thanks to Iain, Cedric, Les, Dave, Mike P., Jeff, Mark, and Ian T. that were especially frequent at the Friday beer sessions in the MacGillivray lab. And the good times I had with you guys - you made this place a lot of fun. I like to also thank my family, friends and Anne that have been there for me at times when I needed support. Lastly, my supervisor, Ross MacGillivray, always understanding and patient. Thanks for having me in your lab and sending me to several conferences - the learning experiences I got from them are irreplaceable and everlasting. I wish you all the best. Chapter 1. Introduction 1.1 Blood coagulation Blood coagulation is a highly regulated, fundamental, physiological process required to stop blood from leaking out of the vasculature. Subsequent to vessel injury, platelets aggregate and activate setting off a series of membrane bound clotting reactions that result in a primary hemostatic plug and preventing further blood loss. The end result of this process is the production of an insoluble fibrin mesh or clot, which maintains the integrity of the circulatory system. At sites of vascular damage, initiation of coagulation occurs when prothrombotic sub-endothelial tissue elements, usually hidden from circulation, are exposed and platelets accumulate. Membranes containing anionic phospholipids promote the assembly of coagulation factors on their surface; this in turn results in a dramatic acceleration of the proteolytic reactions that result in thrombin formation. The initiating pathway of blood coagulation is known as the tissue factor pathway or historically as the extrinsic pathway. The tissue factor pathway is initiated when the integral membrane protein tissue factor (extrinsic to blood) comes in contact with blood after vascular injury (i). Tissue factor interacts with the zymogen Factor VII, or with its activated form, Factor Vila, which is continuously present at low levels in the circulation. The tissue factor-Factor Vila complex converts the zymogens Factor IX, Factor X, and Factor VII itself into active serine protease enzymes. Although assembly and catalytic activity of the tissue factor-Factor Vila complex is effective in the absence of anionic phospholipids, its activity is enhanced by phosphatidylserine (2). However, surface exposure of phosphatidylserine on procoagulant membranes such as activated platelets is essential in promoting membrane assembly and efficient catalytic activity of two subsequent enzyme complexes, namely tenase and prothrombinase. The tenase complex forms when the active protease Factor IXa binds to a high affinity site created by the interaction of the cofactor Factor Villa with a phosphatidylserine-containing membrane surface and calcium ions (3). Tenase rapidly converts the zymogen Factor X into its active protease Factor Xa in the presence of calcium ions. Likewise, Factor Xa binds to its high affinity site created by the interaction of the cofactor Factor Va with a phosphatidylserine-containing membrane surface and calcium ions (4). The prothrombinase complex rapidly converts prothrombin into the multifunctional protease thrombin. Through its ability to promote further platelet aggregation and fibrin formation, thrombin ensures efficient hemostatic plug formation at the site of injury. Thrombin also activates Factor XIII to Factor XHIa, which stabilizes the fibrin mesh via its transglutaminase activity. Phosphatidylserine is equally important in cutting back the thrombin formation by the anticoagulant protein C pathway (5). After being activated by the thrombin-thrombomodulin complex on endothelial cells, protein C and its cofactor protein S degrade Factors Villa and Va when bound to a phosphatidylserine-containing membrane surface. This leads to disassembly of the tenase and prothrombinase complexes and in conjunction with a number of serine protease inhibitors, down-regulates coagulation. 1.2 Gla proteins 1.2.1 Prothrombin Human prothrombin is comprised of a single glycopolypeptide chain of 579 amino acids that circulates in blood at an average concentration of 1.4 jjM (6, 7). Post-translational modifications include vitamin K-dependent y-carboxylation of 10 glutamic acid residues (<§) and glycosylation at three sites (9). Prothrombin consists of four domains. Starting from the N-terminus, these include the y-carboxyglutamic acid rich (Gla) domain, two Kringle domains and a C-terminal Protease domain which is homologous to the chymotrypsin/trypsin family (Figure 1). Figure 1 Domains and fragments of prothrombin. Prothrombin consists of four domains. From the N-terminus there is a negatively charged y-carboxyglutamic acid (Gla) domain which mediates prothrombin binding to procoagulant phospholipid surfaces. Following this domain are two kringle domains, which are thought to be involved in protein-protein interactions, and lastly the inactive Protease domain which is converted to the thrombin upon prothrombin activation. Disulfide bonds and cleavage sites are also indicated. Nomenclatures of prothrombin fragments used throughout this study are also indicated. Protease domain prothrombin des-gla prothrombin prethrombin 1 prethrombin 2 fragment 1.2 fragment 1 A number of different fragments of prothrombin can be produced by enzymatic cleavages, which have been used successfully in studies relating function to the various domains. Thrombin auto-catalytically cleaves prothrombin at Arg155 and Arg284 forming fragment 1/prethrombin 1 and fragment 1.2 + 13 residues/prethrombin 2 des-13 cleavage product pairs, respectively (10). Similarly, Factor Xa cleaves at two sites: Arg271 and Arg320 forming fragment 1.2/prethrombin 2 and disulfide linked 2-chained meizothrombin, respectively. Finally, in a non-physiological reaction, chymotrypsin cleaves at Tyr44 producing Gla and des-Gla prothrombin. In vivo, prothrombin is activated by prothrombinase, an enzyme complex composed of the membrane-bound protease Factor Xa, its membrane-bound cofactor Factor Va, an anionic phospholipid membrane and calcium ions. Prothrombinase catalyzes the cleavage of the zymogen prothrombin to the multifunctional serine protease thrombin. The combined effect of these interactions on prothrombin activation results in a 280,000-fold increase in catalytic efficiency compared to Factor Xa alone (11, 12). Activation of prothrombin involves cleavage at Arg271 and Arg320 (Figure 1). Accordingly, two cleavage pathways exist; however, the activation mechanism in vivo proceeds via an ordered sequential reaction, with meizothrombin as the sole intermediate (13). If released during activation, meizothrombin has anticoagulant function, as it has an active site and enzymatic activity towards protein C (14, 15). Thrombin generation is mainly concerned with procoagulant functions including fibrinogen cleavage, activation of Factors V, VIII & XIII and platelet aggregation (16). Prothrombin dimerization has been reported when bound to anionic phospholipids (17) or at very high concentration (14-200 jjM) in solution (18, 19). However, only the membrane-dependent dimers can form at or below the physiological concentration of prothrombin and are likely facilitated by intermolecular protease domain interactions (17). Many prothrombin interactive sites have been identified for the prothrombin-Factor Xa and prothrombin-Factor Va binary complexes. Each of the prothrombin domains has been implicated in Factor Va interactions. Kringle 1 has been implicated in Factor Va binding (20). Kringle 2 has also been shown to mediate an interaction between prothrombin and Factor Va by data derived from prothrombin deletion mutants (21) and a peptide inhibition study (22). One study also suggests that the Gla domain has an interactive site with the cofactor (23). Finally the Protease domain has been implicated in mediating prothrombin-Factor Va interaction through the anionic binding proexosite I (24, 25) and residues 473-487 (26). 1.2.2 Other Gla proteins Members of the vitamin K-dependent Gla proteins involved in blood coagulation include the procoagulant factors; prothrombin, Factor VII, Factor IX and Factor X and co-regulators of coagulation; protein C, protein S and protein Z. These plasma proteins involved in blood coagulation require vitamin K for normal biosynthesis. Vitamin K is required for carboxylation of specific glutamic acid residues to y-carboxyglutamic acid. Some properties of these proteins are shown in Table 1. Table 1 Comparison of vitamin K-dependent Gla proteins involved in blood coagulation®. post-translational modification plasma concentration (nM) molecular weight KDa carbohydrate p-hydroxylation Y -c a r b o x y | a t i on domain structure % # asp/asn residues # gla residues (N-term to C-term) prothrombin 1400 72 8 0 1 G l a - k r i n g l e - k r i n g l e - p r o t e a s e Factor VII 10 50 13 8 9 Gla-EGF-EGF-protease Factor IX 70-90 55 17 1 12 G la-EGF-EG F-AP-protease Factor X 170 59 15 1 11 Gla-EGF-EGF-AP-protease protein C 65-80 62 23 1 9 Gla-EGF-EGF-AP-protease protein S 145 (free) 360 (total) 69 7 3 10 Gla-TSR-EGF-EGF-EGF-EGF-SHBG like protein Z 45 62 ? ? ? Gla-EGF-EGF-pseudoprotease mature plasma protein\" zymogen/cof actor Enzyme(cofactor) catalysis General Function prothrombin one chain zymogen FXa(FVa) to form thrombin numerous; converts fibrinogen to fibrin (pro/anti-coagulant) Factor VII one chain zymoqen FXa, FIXa, thrombin or FXIIa to form FVIIa initiation phase of coaqulation (procoaqulant) Factor IX one chain zymoqen FXIa or FVIIa(TF) to form FIXa propaqation phase of coaqulation (procoaqulant) Factor X two chain zymoqen FIXa(FVIIIa) or FVIIa(rF) to form FXa propaqation phase of coagulation (procoagulant) protein C two chain zymoqen thrombin(tfirombomocMn) to form APC APC cleaves Villa and Va (anticoagulant) protein S one chain cofactor - cofactor for APC (anticoagulant) - binds Factor Va and inhibits prothrombinase - binds and inhibits Factor Xa - 60% circulates in plasma in complex with C4b bindinq protein (complement) protein Z one chain cofactor - binds thrombin and directs to endothelial membrane - cofactor for ZPI involved in inhibition of FXa (coagulant property unknown) (a) references: www.haemtech.com, (7), (9), (27), (31), (34), (47), (133) supplementary information Gla - region containing y-carboxyglutamic acid residues EGF - region containing sequences homologous to human epidermal growth factor AP - activation peptide TSR - thrombin sensitive region Pseudoprotease - region which replaces the protease domain in vitamin K-dependent serine proteases SHGB - region which replaces the protease domain APC - activated protein C ZPI - protein Z dependent Protease Inhibitor, a serpin (b) two chained proteins are disulfide linked as Besides having a homologous Gla domain (discussed in section 1.4), the vitamin K dependent proteins possess a similar domain organization. With the exception of prothrombin, the Gla domain is followed by regions homologous to epidermal growth factor (EGF domains) for all other Gla proteins. This 50-60 amino acid, disulfide-bonded, often calcium bound domain is thought to be responsible for mediating the many protein-protein interactions involved with these proteins (27). Like the kringle domains of prothrombin, EGF domains likely provide spacers that allow the proper distance and positioning of other parts of the molecule for optimal biological function (28). Thus all Gla proteins are believed to have an elongated structure. For example, prothrombin, having an overall length of 120 A , when bound to membrane has its FXa susceptible peptides bonds placed near the active site of prothrombinase (29). Finally, a large, carboxy-terminal domain makes up the other half mass of the Gla proteins. For the zymogens prothrombin, Factor VII, Factor X, Factor IX and protein C, this is the inactive serine protease domain. Upon activation via proteolysis, a new N-terminus is created, which inserts into a preformed hydrophobic pocket by an internal salt bridge believed to trigger the conformational change from inactive to a functional active site with a correctly shaped substrate binding region (30). In protein Z, which is not a serine protease, the C-terminal region is a pseudoprotease domain since it lacks two of the three characteristic catalytic triad residues, namely Ser and His (31, 32). Like protein Z, protein S is also a cofactor lacking enzymatic activity. Its C-terminal domain is homologous to sex hormone binding globulin (SHBG) (33, 34). Finally, other Gla proteins not involved in clotting have been described including Gas 6 (growth factor (35)), osteocalcin & matrix Gla protein (bone development (36)), and a family of four transmembrane Gla proteins (putatively involved in signal transduction (37, 38)). Lastly a number of Gla containing proteins have been described from invertebrates (39). 1.3 Procoagulant membranes Membranes containing phosphatidylserine and other anionic phospholipids play an essential role in blood coagulation. In vivo, suitable membranes for the interaction of vitamin K-dependent Gla proteins are provided by activated platelets or other cells that have exposed anionic phospholipids to their outer leaflet (40, 41). In resting platelets, an asymmetric bilayer is maintained with the majority of phosphatidylserine and phosphatidylethanolamine hidden from plasma and confined to the membrane's inner cytoplasmic leaflet (see (42) and references within). The regulation of membrane lipid sidedness is controlled by a number of specific membrane proteins, referred to as lipid transporters (reviewed in (43)). Thus in a resting platelet, the bulk composition of the outer leaflet is phosphatidylcholine which is inefficient to procoagulant complex assembly (44). Exposure of phosphatidylserine in platelets serves as a second messenger linking platelet activation to fibrin clotting (45). By binding to proteases and respective cofactors, the membrane facilitates the assembly of macromolecular complexes which efficiently convert circulating zymogens to active enzymes (46). Involving both K m and kcat, the precise enzymatic role of the membrane is incompletely understood but has been attributed to conformation, orientation and concentration effects resulting in co-localization of the enzyme, cofactor and substrate (47). Binding of the coagulation enzymes together with their nonenzymatic cofactors and substrates to phosphatidylserine-containing membranes has been shown to cause conformational changes that serve to induce productive enzyme-substrate interactions. For example, the scissile bonds of prothrombin and meizothrombin appear to be properly aligned with the Factor Xa active site due to membrane-induced conformational changes in both substrates and enzyme (29, 48, 49). Moreover, the high local membrane-bound protein concentrations lead to a substantial decrease in apparent K m of the substrates (e.g Factor X and prothrombin) from far above to far below their respective plasma concentrations (46). Efficient membrane binding of Gla proteins and cofactors as well as catalysis by the enzyme complexes (e.g. tenase and prothrombinase) requires anionic phospholipids with phosphatidylserine being most effective (50-52). In addition, binding of Factor Va and Factor Xa to naturally occurring phosphatidyl-L-serine is stereospecific and occurs with lower affinity to membranes containing phosphatidyl-D-serine (53), indicating specific proteinaceous sites interact with the serine headgroup moiety. Furthermore, soluble phosphatidylserine molecules have been shown to support prothrombinase assembly and support catalytic activity comparable to membranes (54, 55). These recent findings question the classical view the membrane surface plays as often thought to have \"a surface dimensionality reduction\". Collapse of membrane phospholipid asymmetry not only exposes phosphatidylserine but also leads to an increase of phosphatidylethanolamine in the outer membrane leaflet, which has been shown to also have a procoagulant effect (42). Finally, it should be noted that specific protein receptors in addition to phospholipid binding sites on the surface of activated platelets have also been shown to be important to promote the assembly of the various coagulation enzyme-cofactor-substrate complexes (56). Synthetic membranes used to study blood coagulation usually contain the phospholipids phosphatidylcholine, phosphatidylserine and/or more recently phosphatidylethanolamine and can be incorporated into diverse structures including small/large unilamellar vesicles, monolayers and immobilized bilayers. Historically, a composition of 20-25% phosphatidylserine and 75-80% phosphatidylcholine was chosen as a standard to study the enzymatic clotting reactions as this ratio produced optimal activity and most closely matched clotting times observed on platelet membranes (11). Although the activated platelet outer leaflet has been shown to contain only 4-10% phosphatidylserine (57), high experimental phosphatidylserine concentrations continue to be used to study binding and enzymatic function of clotting factors (58-63). Binding affinity and activation rates are greatly dependent on the molar fraction of phosphatidylserine and at sub-optimal phosphatidylserine concentrations, phosphatidylethanolamine also has an enhancing effect. Another variable which effects membrane protein binding is the structure of the membrane. Surface curvature and accessibility to the hydrocarbon region of the membrane varies between the different membrane architectures and consequently has an impact on the binding dependent on hydrophobic interactions. For instance, prothrombin binding affinity to membranes of identical composition was dependent on vesicle size with 0.11, 0.23, 0.86 jiM for liposome diameter of 27.5, 119, and 328 nm (64). A similar binding preference was observed with Factor V (65). 1.4 Gla protein membrane interaction The Gla domain has been solely implicated in the membrane contact process of vitamin K-dependent proteins. Indeed, proteolytic fragments of prothrombin including fragment 1 (Fl; Gla and first Kringle residues 1-155) and fragment 1.2 (F1.2; Gla and both Kringles, residues 1- 273) can interact with membranes, whereas prethrombin 1 (Pre 1; second Kringle and Protease domain, residues 156-579) cannot interact with membranes (66). Prothrombin deletion studies, using a crude gel filtration membrane binding assay, have indicated that neither kringle domain contributes to phospholipid binding (67). Like prothrombin, anionic phospholipid membrane recognition by all of the homologous Gla proteins has been solely attributed to the Gla domain (67-69). A prerequisite for prothrombin membrane binding is the saturation of seven Ca2+ sites in the Gla domain that changes its tertiary structure from an unfolded and non-functional conformation to a tightly folded domain facilitating membrane binding (70, 71). This calcium binding requires Gla residues. The Gla domain within the various vitamin K-dependent proteins, comprised of ~ 45 amino acids, contains between 9 to 12 Gla residues which mediate this calcium interaction. In protein C and prothrombin, a detailed analysis of the function of each of these Gla residues has been evaluated (68, 72). Of the Gla residues, nine are strictly conserved throughout the Gla proteins. From crystal structures of the Gla domain of prothrombin and Factor Vila, the placement of these calcium ions in relation to their Gla ligands is nearly identical in the two proteins (28). The conformational transition induced by the cooperative binding of calcium ions turns the N-terminal part of the Gla domain inside out exposing the hydrophobic co-loop to solvent and burying the majority of the Gla residues. Of the 7 calcium ions, only 2 are accessible to solvent and may play a role in membrane binding. The other calcium ions are buried and are integral to maintaining the membrane binding conformation (70, 73). Calcium binding to the Gla domain occurs at free calcium concentrations (half maximal binding: 0.5-0.7 mM) only slightly below those in blood (free Ca2+ = 1.2 mM) (27). Thus, it has been speculated that calcium binding may serve as a regulatory function in blood coagulation (28). Another conserved feature among the Gla proteins is a hexapeptide disulfide loop which contributes to the overall Ca2+-dependent Gla conformation (63). However, it was shown not to be essential to phospholipid binding (63). Thus, only part of the Gla structure appears to be involved in binding to membrane. Other notable conserved features of the Gla domain are a solvent inaccessible N-terminus and an aromatic cluster at the C-terminal end (70). Changes at these conserved sites result in the destabilization or alteration in the three-dimensional fold of the Gla domain and subsequent loss of calcium and phospholipid binding (68, 74). Despite the conserved structural features and sequence similarity, there is a wide range in affinity of the Gla proteins for biological membranes. Although numerous studies focusing on the various Gla domains have been conducted, no single site or single type of interaction appears to be exclusively responsible for the affinity of the protein-membrane interaction. A hydrophobic interaction mediated by the small conformation-dependent ©-loop, was suggested by fluorescence quenching of a doxyl spin label within the bilayer (60) and from the derived crystal structure (75). However, this was considered insignificant to prothrombin-membrane affinity when studies were compared using small unilamellar vesicles (SUV) and large unilamellar vesicles (LUV) despite differences in headgroup packing and accessibility to the hydrophobic membrane core (64). Adding to the uncertainty, another study showed that an aberrant ©-loop in a bovine prothrombin molecule lacking the first four N-terminal residues caused a modest 5-fold decrease in membrane affinity (74), indicating that other regions are also involved or exposed aberrantly in the mutant. Similarly, site directed mutagenesis of various residues in the co-loop had only small impact on membrane affinity. For example, replacement of Phe4 of protein C with Gin had no detected impact on membrane affinity (76). In addition to a hydrophobic contribution, an electrostatic interaction has been postulated, but is proposed to contribute only minimally to binding in a Ca2+ saturated system (77, 78). Specific phosphatidylserine contact points have been identified. Using a disulfide constrained co-loop peptide sequence of Factor IX, Lys5, Leu6, Phe9 and VallO have been directly implicated in phosphatidylserine binding (58). These findings have been confirmed by structural studies using human prothrombin Gla domain in complex with calcium ions and lysophosphatidylserine (75). The binding isotherms obtained by dynamic light scattering suggest the existence of 3-4 phosphatidylserine binding sites per prothrombin molecule (50). Others have speculated additional phosphatidylserine sites on the protein surface (79, 80). Finally calcium ions in addition to the ones mentioned are bound during membrane contact as was shown for prothrombin (81) suggesting that the Gla domain crystal structures solved thus far in the absence of membrane are insufficient to explain membrane binding. Membrane binding is undoubtedly complex and likely requires multiple synergistic interaction types between Gla protein and anionic phospholipid membrane. These multi-facetted interactions complicate the study of Gla domain membrane binding and inherently cause the interaction to deviate from a simple binding model as observed by several other laboratories (82-87). Despite this progress, the mechanism of the Gla protein-membrane binding interaction remains unclear. In addition, it is surprising that the Gla proteins bind membrane with a great range in affinity. For instance, bovine protein C and bovine factor X have only five major differences in the amino terminal 34 residues, although the binding affinity for membrane differs by approximately 400-fold, with K'' tpp > 15,000 and ~ 40 nM, respectively (79). The reason for this difference is still not obvious as mutagenesis studies have not been successful at completely deciphering this discrepancy (76, 88). Although some non-conserved residues of the Gla domain are related to affinity (e.g. (89)), there appear to be additional factors that control membrane affinity. The causal nature of this complexity remains obscure. Prothrombin is known to have a number of different conformations with unknown function. A cis to trans isomerization of Pro22 of bovine prothrombin has been shown by fluorescence due to Ca2+ binding (90) but its impact on membrane binding remains controversial. Fluorescence studies suggest the trans conformer binds membrane while the Ca2+ bound crystal structure data argues that the cis conformer binds membrane. Furthermore, molecular dynamics simulations have shown that the cis form is less likely to occur; however, it is permitted by the structure (91) suggesting that during membrane binding a cis-trans equilibrium may result. Bovine prothrombin has also been shown to have membrane induced conformational changes (29). In solution, prothrombin is an elongated molecule with an overall length of 120 A as determined by fluorescence resonance energy transfer (FRET) measurements (66). In the presence of calcium and phosphatidylserine containing membranes, prothrombin was found to undergo a shape change to a more tightly folded, compressed, bent molecule with an overall length of 94 A (29). This compression shape change is not observed with meizothrombin, although both have the same shape in solution (66). Membrane-induced changes in prothrombin secondary structure as well as alterations in interdomain interactions have also been implicated from thermal denaturation studies (92). These studies together suggest that fragment 2 may act as a hinge region upon membrane binding. If and how these conformations of prothrombin affect membrane binding is not known. 1.5 Techniques to study protein-membrane interactions A number of techniques have been used to study protein-membrane interactions. Separation assays generally provide only equilibrium binding data whereas direct assays provide additional information on the kinetics of an interaction. 1.5.1 Equlibrium binding Equilibrium separation assays commonly used to study the membrane affinity of peripheral membrane proteins include differential centrifugation and chromatographic gel filtration. These techniques offer the advantage of being inexpensive and require no labelling but, are labour intensive and often have problems associated with the separation of free protein and membrane-bound protein. Centrifugation assays can give reliable Kd measurements provided that the vesicle or phospholipid-coated bead centrifugation efficiency is high (95%) which can be verified by using a radiolabeled lipid. In additional, complexes must be relatively stable (small &0ff) to allow accurate determination of bound and free protein. Gel filtration chromatography has been used on the basis of size exclusion to separate the large vesicle with associated protein from the smaller free protein (67). However this technique is less sensitive, requires large amounts of protein and is not amenable to interactions governed by fast dissociation. Other less frequendy used equilibrium binding techniques have also been used to determine peripheral protein-membrane dissociation constants. Requiring fluorescently labelled protein and liposome adsorbed latex beads, flow cytometry can be used to derive protein ratios of membrane bound to free. A biophysical technique called quartz crystal microbalance has been used with prothrombin and other peripheral proteins to study binding to immobilized bilayers by monitoring dissipation and resonance frequency changes at the membrane surface (87). In addition to frequency measurements at various bulk protein concentrations used to derive a dissociation constant, dissipation changes related to frictional (viscous) losses at the bilayer give insight to rigidification or intra-molecular protein interaction. Finally, ellipsometry, another biophysical technique requiring adsorbed mono or bi-layers, has also been used for phospholipid affinity measurements for a number of Gla proteins. 1.5.2 Kinetic analysis Stopped flow light scattering and FRET measurements at right angle are two conventional, frequently used kinetic analysis methods that enable the direct measurement of formation or breakdown of a peripheral protein - membrane complex. Fluorometric assays based on either intrinsic tryptophan residues in the protein or extrinsically introduced fluorescent probes on either protein and/or lipid head group, measure changes in the microenvironment of these reporter groups which are representative of complex formation. Membrane association can be followed by mixing SUVs with peripheral protein or membrane dissociation can be followed by mixing an equilibrated protein-SUV with excess unlabelled SUVs. The highest sensitivity for binding is obtained with the introduction of both a donor and an acceptor fluorescent probe attached to either of the interacting components. Binding is then observed by following the change in FRET between donor and acceptor group which is proportional to complex formation. Both light scattering and fluorescence measurements require small liposomes and dilute protein concentrations (e.g. < 3 |J.M prothrombin) to maintain the vesicles in solution. Stopped flow measurements allow the accurate determination of fast kinetic events in the ms range as observed for many Gla proteins. Equilibrium dissociation constants at surfaces coated with monolayers have been determined from total internal reflection (TIR) fluorescence microscopy (TIR-FM) measurements of the fluorescence as a function of the concentration of fluorescently labelled protein in solution (e.g. prothrombin (93)). TIR has been combined with fluorescence photobleaching recovery (TIR-FPR) to also provide kinetic binding constants (84, 85). 1.6 Surface plasmon resonance 1.6.1 Overview Biacore instruments employ surface plasmon resonance (SPR) to monitor the formation and breakdown of complexes in real time. This optical biosensor technique provides detailed information on the binding mechanisms and rate constants associated with macromolecular interactions. In a SPR experiment, one of the molecules is immobilized on a surface (referred to as immobilized ligand) and the other is flowed past the surface in solution (referred to as analyte). In the study of peripheral protein - membrane interactions, the membrane is directly coupled to the biosensor surface. This biosensor, known as an LI chip in Biacore terminology has been sequentially derivatized. First, a gold thiolate bond covalently links to a C-16 alkane containing a terminal hydroxyl group. Second, a carboxymethyl dextran (a carboxymethyl glucose polymer) is attached via two additional reactions. The carboxy groups are then derivatized with lipophilic alkyl chains capable of capturing liposomes by interacting with the acyl chain moieties of phospholipids (Figure 2A). These captured bilayers are chemically and physically stable (94) and can be used to probe for peripheral protein interactions. When an analyte molecule interacts with the immobilized membrane ligand a surface mass change is detected and a signal is generated in real time. Thus, the progress of an interaction experiment is followed directly in a sensorgram, a data plot of response versus time. There are several advantages in using SPR to study interactions including lack of labelling, low sample consumption and high sensitivity. A dextran matrix thin gold coated glass surface lipophilic surface plasmon resonance In thin metal film light source exponentially decaying evanescent field bulk solution with analyte o detector e resonance intensity dip at resonance angle = f(n sur() Figure 2 Immobilization strategy and detection principle of SPR. A. LI chip showing capture of liposomes. Adaptedfrom www.biacore.com B. Schematic diagram illustrating a surface plasmon resonance biosensor. TIR: total internal reflection; nsurf: refractive index of the bulk solution in the vicinity of the sensor surface. Adapted from P. Schuck (95). 1.6.2 Optical Configuration and Detection Principles Two basic requirements for SPR are a source of polarized light and a metal (e.g. gold) coated glass prism as illustrated in Figure 2B (96). When polarized light is shone at a glass prism, at an appropriate angle, the incident light is totally reflected, a principle known as total internal reflection (TIR). At this critical angle, all photons are reflected and none pass through the prism. If the prism is coated directly on top of the TIR surface with a thin (50 nm) layer of gold, then the set-up is known as the Kretchmann Configuration (95). In this configuration, TIR of light can be used to excite surface plasmons in the metal surface. At a specific angle of incident light, 0 excitation occurs. The energy of the incident photons are absorbed and converted to surface plasmons traveling along the metal surface. This resonance results in an electromagnetic field composed of evanescent waves that decay exponentially with increasing distance perpendicular to the sensor surface. The resonance causes an energy loss in the reflected light, which is visible as a sharp minimum in the angle-dependent reflectance, an experimentally recorded quantity. The resonance angle (0) strongly depends on the refractive index of the sample within the evanescent field above the sensor surface (nsurf). Thus, adsorption or desorption of analytes at the sensor surface change the refractive index and produce a shift in resonance angle that can be precisely measured. The shift in resonance angle has been shown to be directly proportional to the mass that is bound to the sensor surface (97). The Biacore Instrument consists of a number of integrated components that allow data collection under conditions of continuous flow (www.biacore.com). The sensor chip surface forms one wall of the flow cell (dimensions: 1 = 2.4 mm, w = 0.5 mm, h = 0.05 mm). The Biacore 3000 instrument has four flow cells in parallel which allow experiments and controls to be performed with the same analyte sample simultaneously. Samples and reagents are delivered to the surface through a microfluidic system that ensures control over sample delivery times. Valves in the microfluidic system can switch between buffer and sample with high precision. This ensures that the sample is delivered as a defined liquid segment with minimum mixing between sample and buffer. A photo-detector precisely detects an array of reflected light angles dependent on surface plasmon resonance. The shift in angle required for resonance, called the SPR-angle response is the quantitative measurement of analyte binding to the surface. SPR-angle response is then converted to the characteristic Response Unit (1 RU = 0.0001°) reported in the raw data sensorgrams (98) which is equivalent to 1 pg/mm2 of bound protein (96). 1.7 Objectives and Overview Although a wealth of information has been gathered, several questions remain about the overall mechanism of phospholipid binding of Gla proteins. In this thesis, prothrombin was chosen as the central molecule of study as it services a pivotal role in coagulation. The impact of Factor Va on the prothrombin-membrane interaction is addressed in Chapter 2. Although a number of studies have suggested an interaction between Factor Va and prothrombin, none of them have been conducted as a direct binding assay in the presence of membrane. As a potential mechanism leading to prothrombinase assembly and substrate binding, the cofactor may directly recruit prothrombin on the membrane surface. Thus, it was my goal to address the question: Does Factor Va enhance the affinity of prothrombin for membrane? Semi-quantitative binding data obtained by differential centrifugation experiments implicate the cofactor in aiding the membrane affinity of prothrombin and also its non-Gla containing fragment prethrombin 1. As outlined in previous introductory sections, the detailed kinetic events governing Gla protein binding have mainly been characterized using light scattering and FRET, and the results of these studies make up the bulk of known biophysical data describing the membrane interaction. In Chapter 3, the previously unexploited technique of surface plasmon resonance is developed to further probe the membrane interaction kinetics and affinity of prothrombin and derivatives of prothrombin. The main objective of this chapter is to determine if the Gla domain - containing derivatives of prothrombin (fragment 1 and fragment 1.2) behave similarly in membrane binding to prothrombin and, if not, to determine the causative nature of the observed differences. Results of these studies suggest that prothrombin membrane binding deviates from the simple 1:1 membrane binding shown for fragment 1.2, but rather follow complex binding likely explained by a slow dimerization reaction on the membrane surface. As all Gla proteins involved in blood coagulation contain the highly conserved Gla domain (assessed by either sequence or available structure), it is surprising that the membrane binding properties of these proteins vary widely in the literature. One problem has been in technique and membrane variability preventing valid comparisons between studies. To alleviate this discrepancy and on the basis of the results obtained with prothrombin and its fragments, a thorough membrane binding comparison of kinetic and affinity parameters is made of the other Gla proteins, including zymogens, cofactors and activated enzyme forms. In doing so, I wished to address the question whether all Gla proteins have conserved membrane binding properties? As detailed in Chapter 5, results of these studies indicate that the Gla proteins have adopted a wide range of membrane binding properties that appear not to be solely caused by Gla domain differences but also due to differences in regions removed from the membrane contact site. The significance of these findings and future directions are also discussed. Chapter 2. Methods 2.1 Materials, proteins and miscellaneous reagents The synthetic lipids l,2-dioleoyl-sn-glycero-3-[phospho-L-serine] (DOPS) and 1-palmitoyl-2-oleoyl-.sn-glycero-3-phosphocholine (POPC) were purchased from Northern Lipids, Inc. (Vancouver, BC) and Avanti Polar Lipids, Inc. (Alabaster, AL), respectively. The natural lipids L-lecithin-(phosphatidylcholine) from egg (EPC) and phosphatidyl-serine from bovine brain (BPS) were purchased from Sigma-Aldrich (Oakville, ON). Polycarbonate filters (100 nm diameter) were purchased from Corning (Acton, MA). The peptidyl substrate, //-D-phenylalanyl-L-pipecolyl-L-arginine-P-nitroanilide (S-2238) was purchased from Chromgenix (West Chester, OH). Pre-stained protein marker was obtained from New England BioLabs (Pickering, ON). Human coagulation proteins (prothrombin, Pre 1, Fl, F1.2, a-thrombin, Factor Va, Factor X, Factor Xa, Factor Xa-DEGR, Factor IX, Factor XIa, Protein C, Activated Protein C, Protein S and Protein Z) were purchased from Haematological Technologies (Essex Junction, VT). Prothrombin-Gla domainless was bought from Enzyme Research Laboratories (South Bend, IN). Bovine serum albumin was purchased from Sigma-Aldrich. Cross-linked prothrombin dimer (xl-prothrombin) was graciously provided by Dr. Peter Anderson (University of Ottawa) (77). Mouse monoclonal antibody against kringle 2 of prothrombin (a-PT) was bought from Haematological Technologies. Chemiluminescent reagents were purchased from Pierce (Rockford, IL) including horseradish peroxidase conjugated goat anti-mouse IgG (HRP-2°Ab) and luminal - peroxide solution. All coagulation proteins were judged to be greater than 98% pure from an overloaded Coomassie blue-stained sodium dodecyl sulphate (SDS)-PAGE gel. Protein concentrations were determined by using the bicinchoninic acid assay, BCA (Pierce, Milwaukee, WI) and/or by absorbance using the following extinction coefficients (E280 1%, 1cm) and molecular weights (Mr) given by the supplier; prothrombin: 13.8, 72,000; prothrombin des-Gla: 14.5, 67,000; Prel: 17.8, 49,900; Fl: 11.9, 21,700; F1.2: 10.8, 34,600, a-thrombin: 18.3, 36,700; Factor Va: 17.4, 168,000; Factor X: 11.6, 58,900; Factor Xa and Xa-DEGR: 11.6, 46,000; Factor IX: 13.2, 55,000; Factor IXa: 14.0, 45,000; protein C: 14.5, 62,000; activated protein C 14.5, 56,200; protein S 9.5, 69,000; and protein Z: 12.0, 62,000. For ease of comparison, the xl-prothrombin concentrations are given in terms of prothrombin monomer. All surface plasmon resonance reagents, including LI sensor chips and HBS running buffer were purchased from Biacore (Biacore Inc., Piscataway, NJ). HEPES buffered saline used for liposome immobilization experiments (HBS) consisted of 10 mM HEPES, pH 7.4 and 150 mM NaCl. Running buffer used for protein interaction experiments (HBC) consisted of 10 mM HEPES, pH 7.4, 150 mM NaCl, 0.1% bovine serum albumin (BSA) and 5 mM CaCb. All buffers were filtered through a 0.22 jim filter and degassed before use. All stock solutions were briefly centrifuged before use to remove any potential insoluble material. All other chemicals were from Sigma Chemical Co. and were of the highest grade available. 2.2 Liposome preparation Liposomes were prepared as described previously (99) with minor revisions. Briefly, lyophilized lipids were dissolved in chloroform to approximately 30-100 mM and concentrations were established by inorganic phosphate determination as described previously (100). The appropriate molar ratios of phospholipid in chloroform was dried, first under a stream of argon and then under vacuum for at least 3 hours. The resultant residue of lipids was resolubilized in HEPES-buffered saline (HBS) [10 mM HEPES, pH 7.4 and 150 mM NaCl] or HBS containing 15% sucrose (for sucrose loaded vesicles) to a final concentration of approximately 10 mM. The resultant multilamellar vesicle (MLV) suspension was subjected to a rapid freeze-thaw technique five times by cycling in liquid nitrogen and warm water. Large unilamellar vesicles (LUV) were generated by extrusion of MLVs under pressure through two stacked Nucleopore polycarbonate filters with a 100 nm pore size (10 passes) using an extrusion device obtained from Northern Lipids (Vancouver, BC). To prepare sucrose loaded large vesicles (LV), MLVs were extruded through an 800 nm pore size filter. Non-encapsulated sucrose was replaced by HBS by adding equal volume HBS followed by a 5 min, 14,000g centrifugation. Sucrose loaded LV in the pellet were resuspended in HBS. Due to the large pore extrusion size LV remained multilamellar. Final liposome concentrations were determined by inorganic phosphate determination. Membrane composition is stated as the mole percentage of DOPS (or BPS) with the remainder of the phospholipid being POPC (or EPC). The homogeneity and mean diameter of LVs and LUVs was determined by dynamic light scattering using a NICOMP 370 particle sizer (Nicomp Particle Sizing Inc., Santa Barbara, CA) and found to be > 1000 nm and 110 +/- 25 nm, respectively. 2.3 SDS-PAGE electrophoresis, western blotting and densitometry For differential centrifugation experiments, membrane bound prothrombin concentrations were determined by immunoaffinity detection. Standard concentration samples along side experimental samples were denatured in Laemmli sample buffer under reduced conditions for 10 min at 60 °C and subjected to SDS-PAGE on a 12% minigel. After electrophoresis, the separating gel and nitrocellulose paper were briefly soaked in transfer buffer (25 mM Tris.HCI/192 mM glycine/10% methanol, pH 8.3) and assembled in a transfer chamber (Bio-Rad). Proteins were electroblotted onto the nitrocellulose in transfer buffer by application of 0.4 A for 2 hr. Nitrocellulose blots were soaked in SuperBlock (Pierce, Rockford, EL) containing 0.05% Tween-20 for 1 h and then at 4 °C overnight in SuperBlock containing 4 |ig (saturating amounts) of a-PT per mL, washed in Tris Buffered Saline containing Tween-20 (TBS-T: 150 mM NaCl, 50 mM Tris.HCl, 0.05 % Tween-20, pH 7.4), and incubated for 1 hr in 40 ng HRP-2°Ab per mL SuperBlock. The blots were washed in TBS-T and for detection incubated with a HRP chemiluminescent substrate - peroxide solution (Pierce) for 5 min. The image was directly captured on a Syngene GeneGenius Imager (Cambridge, U.K.) and the band density determined prior to pixel saturation using the integrated densitometry software. Experimental bands were quantified using a prothrombin concentration standard curve. Experiments were performed and analysed three times. The data are plotted as the mean +/- the standard deviation of the three results. Prior to SPR analysis, analyte samples were examined for homogeneity and total protein content by SDS-polyacrylamide gel electrophoresis. Samples were denatured under reducing or non-reducing conditions for 10 min at 95 °C and electrophoresed (120V for 45 min) in a pre-cast 4-20% gradient polyacrylamide SDS gel (Promega, Nepean, ON). Protein bands were visualized following staining with EZBlue™ Coomassie Brilliant Blue G-250 (Sigma) and destaining in water. The destained gel was subsequently photographed over visible light. Products of prothrombin proteolysis were examined for cleavage products by SDS-PAGE (7.5 |ig/lane) or western blot analysis (10 ng/lane). Samples were denatured with SDS under reducing or non-reducing conditions for 10 min at 95 °C and electrophoresed in a 12% polyacrylamide SDS gel. Protein bands were visualized following staining with Coomassie Blue R-250 (Sigma-Aldrich) and destaining in 40% methanol/5% acetic acid. Alternatively for western analysis, procedures were the same as the immunoaffinity detection experiments (differential centrifugation) above with the exception that images were captured on x-ray film (HyperFilm ECL, Amersham: Uppsala, Sweden). 2.4 Differential centrifugation (FVa mediated binding) In a 10 nL reaction volume, prothrombin (15 - 400 nM) was incubated in the presence and absence Factor Va (350 or 738 nM) with sucrose loaded LV (2400 |xM, 25% BPS) in HBS (20 mM Hepes, 150 mM NaCl, pH 7.4), 2 mM Ca2+ and 0.1% PEG 8000 at 22 °C for 25 min. The equilibrated sample was then carefully layered on top of a 200 |iL 15% sucrose sedimentation cushion without mixing the two aqueous solutions. The LV with associated proteins were separated from unbound protein and pelleted by centrifugation at 19,000g for 10 min and quantified by immunoaffinity detection (see Section 2.3). Experiments were also carried out in the absence of LV to assess background signal and using a control protein (bacterial protein MT 1704) in place of Factor Va to assess non-specific binding. 2.5 Thrombin generation assay Different membrane preparations were tested for their ability to enhance thrombin generation by prothrombinase. Membrane preparations were incubated in a flat bottom microtiter plate with a 5:1 or 3:1 molar ratio of Factor Va to Factor Xa (at concentrations indicated in the figure legends) along with excess prothrombin (1.4 jiM final). Proteins were mixed at 22°C in HBS pH 7.4 supplemented with 0.01% PEG 8000 to minimize protein losses by adsorption. Prothrombinase assembly and subsequent prothrombin activation were initiated by the addition of Ca2+ (2 mM final). Continuous shaking during the reaction maintained the membrane preparation in suspension. The reaction was stopped after an indicated time by the addition of HBS pH 7.4 containing 0.01% PEG 8000 and 90 mM (final) EDTA. An aliquot of the assay solution was diluted in HBS containing 0.01% PEG 8000 and mixed with S-2238 to a final concentration of 200 fiM at 22°C. The amount of thrombin that formed (progress curve) was measured spectrophotometrie ally by the absorbance change at 405 nm in a Spectramax kinetic microplate reader (Molecular Devices). From the change in absorbance at 405 nm the amount of thrombin was calculated from a calibration curve made with a purified thrombin standard. The calibration curve was determined with the assay conditions described above. A mean value of the initial thrombin formation rate (or Vmax) was determined from three separate replicates. Experimental data were fit by non-linear regression analysis using Graphpad Prism 4.0 software (Graphpad). 2.6 Prothrombin enzymatic digestion Prothrombin (52.5 |ig = 5.4 jiM) was incubated for 5 min in HBS containing 0.01% PEG 8000 and 2 mM Ca2+. At time 0 s, thrombin (3.5 |ig = 675 nM) was added to start the proteolysis reaction a total reaction volume of 140 |iL. At indicated timed intervals 20 |iL aliquots (seven in total) of the reaction mixture were removed and added to 95 °C sample loading buffer containing 5% P-mercaptoethanol and analysed by SDS-PAGE analysis (see Section 2.3). For western blot analysis, the same conditions were used except prothrombin (70 ng = 6.94 nM) and thrombin (5 ng = 965 pM) concentrations were changed as indicated. 2.7 Amino-terminal sequence analysis Protein species in the prothrombin sample (10 ng total protein) from the commercial supplier were separated by SDS-PAGE on 12% reduced gels, transfered to Immobilon-P membranes (Millipore) in 25 mM Tris-base, 192 mM glycine, (pH 8.3) and 20% methanol, and stained with diluted Coomassie Brilliant Blue R-250. The 50 kDa protein band was cut from the blots and subjected to protein sequencing using Edman chemistry and a Perkin-Elmer ABI automated sequencer model 476A by S.C. Perry of the Nucleic Acid-Protein Service Unit (NAPS, University of British Columbia). 2.8 Surface plasmon resonance: 2.8.1 Membrane immobilization All SPR experiments were conducted on a Biacore 3000 (Uppsala, Sweden). Prior to lipid immobilization, the lipophilic LI sensor chip was washed with 50 mM octyl glucoside (1 min at 20 ^iL/min). LUV (500 |aM) composed of either synthetic DOPS/POPC (or natural BPS/EPC) lipids were injected for 17 min at a 3 ^L/min flow rate in HBS running buffer. LUV were immobilized to a reading of 5000-8500 RU depending on DOPS concentration, with higher DOPS concentrations resulting in lower immobilization values. The use of new chips was crucial for reproducibility of membrane immobilization values as used chips bound significantly less phospholipid membrane. Weakly adhering LUV were removed with five consecutive 10 mM EDTA pH 8.0 injections (2 min at 20 |iL/min) resulting in a stable membrane surface as was indicated by an insignificant loss in SPR signal for the following 12 hours (data not shown). Prior to protein binding, running buffer was changed to HBC and flow cells were equilibrated until the baseline stabilized to less than 0.05 RU/min. An excess concentration of Ca2+ (5 mM) was included to avoid limiting prothrombin membrane association (101) and BSA (0.1%) was included to block any non-specific protein-lipid and protein-protein interactions (102). 2.8.2 Protein binding experiments SPR experiments were performed at 24 °C with a flow rate of 20 |jL/min. Association times were typically between 4 and 25 min and dissociation was monitored as indicated on figure legends. Controls for bulk refractive index changes between sample and HBC running buffer, instrument drift and non-specific binding were performed in parallel with an underivatized flow cell. A 100% POPC membrane surface was not used as a reference surface since very low (< 2% binding compared to 25% DOPS), yet significant binding was detected. The underivatized flow cell showed no appreciable change in SPR response upon an injection of prothrombin and was used as a reference control. The immobilized membrane surface was then regenerated by removing membrane bound protein with a 10 mM EDTA pH 8.0 injection, which returned the baseline to the value prior to introducing protein. The binding of various peripheral proteins (prothrombin, Fl, F1.2, xl-prothrombin and other Gla proteins) was examined by multiple injections of the analyte diluted in HBC running buffer over a concentration series typically spanning a 10-fold range above and below the apparent equilibrium dissociation constant (K%pp) of the interaction. All SPR experiments were carried out with new LI sensor chips, duplicate analyte concentration standards and replicate analyses. 2.8.3 Data analysis BIAevaluation v4.1 was used to make reference corrections and to display sensorgram data (response units (RU) vs. time). All data was corrected for non-specific binding by subtracting the value from the underivatized reference flow cell. A second reference correction was made by subtracting buffer injections thereby eliminating any possible systematic artefacts observed between the reaction and reference surfaces. The K a p p was determined by measuring the near steady state response units (Req) at several ligand concentrations. Graphpad Prism 4.0 was used to fit binding data globally to a one-site binding hyperbola according to the relationship R e q = RmaxC/( Kf p +C), where Rmax is the response signal at saturation, C corresponds to the injected analyte concentration (Co), and Kj''p is the equilibrium dissociation constant. The dissociation constant is an approximation (hence Kf p) since the binding model is known to be more complex than the fitted one-site binding model. Sensorgram data were globally fit to various models made available through the BIAevaluation v4.1. They included a simple one-site (Langmur) model (A + B AB), a heterogeneous analyte-competition reaction model (A1 + B <-»• A1B; A2 + B A2B), a two-state reaction (conformation change) model (A + B <-> AB <-> A*B), and a bivalent analyte (first step: A + B <->• AB; second step: AB + B AB2) model. Together with a surface dimerization model (first step A + B <-+ AB; second step AB + AB <-> AAB) provided by Biacore, these models were used to determine which reaction mechanism dominated an interaction. Chapter 3. Effect of Factor Va on Prothrombin-Membrane Interaction 3.1 Rationale Previous binding studies between prothrombin and Factor Va have been based on experimental systems in solution lacking an anionic phospholipid membrane. Sedimentation equilibrium analysis has shown that prothrombin and Factor Va form a weak 1:1 complex in solution with a Kf p = 10 (J.M (103). Although informative, this result seems insignificant to complex formation in vivo in light of the prothrombin (1.4 jjM; (104)) and Factor V (21 nM; (105)) concentrations in plasma. Anionic phospholipids (especially phosphatidylserine) present in the outer leaflets of activated platelets and at sites where complex assembly occurs, have been shown to be crucial in the forming of a fully active enzyme complex since they determine the proper conformation and orientation of prothrombin, Factor Va and Factor Xa (54). Also, a tubular flow enzymatic study has shown that prothrombin can directly contribute to the assembly of the Factor Va-Factor Xa complex on phosphatidylserine-containing membranes (106). There is also ample indirect evidence, from prothrombinase kinetic studies using recombinant prethrombin 1 mutants (25), a proexosite 1 directed inhibitor (24), Factor Va heavy chain directed inhibitors (107), and a fast dissociating form of Factor Va (108) that indicate a possible direct interaction between prothrombin and Factor Va on the membrane. It was my goal to elucidate the putative binding interaction between prothrombin and Factor Va in the presence of an anionic phospholipid membrane in a simplified Factor Xa-free 3-component system. It was anticipated that multiple regions on the prothrombin surface contribute to the overall binding energy required to bind to the complementary exosites found on Factor Va. By dissecting these binary protein interactions, I hoped to gain insight to the binding interaction of prothrombin to prothrombinase. Differential centrifugation of large vesicles and associated proteins from unbound proteins was used as a technique to assess the effect that membrane bound Factor Va had on the prothrombin membrane interaction. A similar method had been employed successfully to measure cell receptor binding of specific yet low-affinity plasma proteins (109). Since Factor Va forms a tight complex with membrane with a IQ = 2.7 nM (110), this technique was anticipated to be useful. 3.2 LV characterization by prothrombinase activity Prothrombin membrane binding was evaluated using sucrose loaded LV. To assess any deleterious effects the sucrose may have on these liposomes, they were first characterized in their ability to support prothrombinase activity. As expected, prothrombinase activity was supported by LUV composed of 25% BPS and 75% EPC, giving a saturation curve with half-maximal activity at -0.95 pM phospholipid and a maximum velocity or kc&t of 80 +/- 5 nM of thrombin min\"1 (nM of Factor Xa)\"1 (Figure 3). 0.0 2.5 5.0 7.5 10.0 12.5 [phospholipid] (nM) 15.0 B c \"c I 2000 pM) were required to observe prothrombin in the bound fraction (Figure 4). 66 KDa — No LV (Background) B \\ v HHHHHHHIH I Mmm 66 KDa — 1500 |JM LV / * 66 KDa — J A \" # # A ' A o* ©• •BflSBHSnHHI Sfil 18 pixel counts (or > 1700 pg prothrombin). Over these protein concentrations, pixel count is no longer proportional to concentration. Western blot results show an enhanced prothrombin membrane binding affinity compared to the experimental background (Figure 7). Prothrombin membrane binding and background levels were quantified by performing prothrombin band densitometry and shown graphically in Figure 8A. Results indicate an enhanced detection of prothrombin with increasing reaction prothrombin concentrations regardless of inclusion of LV and/or Factor Va in reactions. Real prothrombin membrane binding was clearly obscured by background levels. Although with high error, reference corrected membrane binding data show an increase in prothrombin binding in the presence of Factor Va compared to without Factor Va (Figure 8B). Prothrombin binding data in the presence of Factor Va was loosely fitted to a 2-site binding model (r2 = 0.86). Data was further corrected for self-mediated binding illustrating prothrombin binding specifically to protein site(s) on membrane bound Factor Va (Figure 8C). This data was fitted to a 1-site binding model (r2 = 0.76) with an K« pp ~ 0.05 jjM. 66 KDa 66 KDa 66 KDa 66 KDa 66 KDa— 0.014 0.070 0.140 0.210 0.350 Figure 7 Prothrombin membrane binding: dependence on large vesicle and Factor Va. Representative western blots (triplicate) of SDS-PAGE gel probed with saturating amounts of monoclonal anti-human prothrombin directed against fragment 2 subjected to centrifugation assay (see Materials & Methods). Prothrombin at indicated concentration (fiM; right) was equilibrated for 25 min either in the absence ( - / -) or presence of 2400 ^M LV (- / LV) or presence of both 185 nM Factor Va and 2400 jxM LV (FVa / LV) in 20 mM HEPES, pH 7.4, 150 mM NaCl, 2 mM CaCl2, 0.1% PEG 8000. Prothrombin was detected in the pellet after differential centrifugation. Location of molecular weight marker indicated for size reference. 0.1 0.2 0.3 [prothrombin] (nM) 0.1 0.2 0.3 [prothrombin] (p.M) [prothrombin] (|iM) Figure 8 Factor Va mediated prothrombin binding to BPS/EPC (25/75) large vesicles assessed by differential centrifugation coupled with immunoaffinity detection. Prothrombin assayed by differential centrifugation in the absence of LV (A), presence of 2000 pM LV (•) and presence of both 2000 pM LV and 185 nM Factor Va (T) was subsequentiy detected by immunoaffinity detection (see Materials & Methods). A. Detected prothrombin data was derived from densitometry performed on experimental samples and concentration standards from western blots. Each data point was an average of 3 repeats and error bar represents the standard deviation. B. Background reference corrected prothrombin binding data. Binding data in the presence of LV and FVa were fitted to a 2-site binding model. C. Prothrombin binding to FVa-bound LV. Data corrected for self-mediated binding and fitted to a 1 -site binding model. (Kd ~ 0.05 pM). 3.4 Determination of ~so kDa species It was of interest to determine the identity of the -50 kDa degradation species that appeared in differential centrifugation experiments. Initial predictions were that this band was prethrombin 1 from its size determined from SDS-PAGE gels and reactivity to anti-prothrombin antibody confirming the presence of kringle 2 epitope. This hypothesis was confirmed by cleavage analysis of prothrombin by thrombin and amino-terminal sequence analysis. The time course of product generation during proteolytic digestion of prothrombin by thrombin was followed by SDS-PAGE and visualization of protein bands by staining with Coomassie Blue (Figure 9A) or western blotting (Figure 9B). Considerable amounts of prethrombin 1 were formed in the early phase of the incubation. Its concentration decreased after prolonged incubation with formation of products with lower molecular weights, namely fragment 2 (not visible on the gel) and prethrombin 2. The appearance of these bands is indicative for thrombin-catalyzed cleavages in prethrombin 1. The prethrombin 1 fragment electrophoresed at an identical location as the -50 kDa band in the commercial prothrombin sample. Amino-terminal sequence analysis of the -50 kDa band present in prothrombin commercial sample yielded the following sequence: Ser-Glu-Gly-Ser-Ser-Val-Asn-Leu. Comparison with the amino acid sequence of human prothrombin {111) indicates that this product results from thrombin-catalyzed cleavages at Argl55-Serl56 in prothrombin giving rise to prethrombin 1 (residues 156-581) and fragment 1 (residues 1-155). 10 60 120 240 ^ — — prothrombin • M * — prethrombin 1 X fr^* » * M&m 31— 21.5— ''ff ej^' ^ ^ ^ ^ ^^^^^ ^^^^^^ thrombin / prethrombin 2 — fragment 1.2 i J (0 a -200 -800 0 0.74 83 0 1000 2000 3000 4000 5000 6000 Time (s) Figure 11 Baseline stability of immobilized membrane and reference surfaces over time. Baseline stability response is shown over extended time of immobilized membrane and reference surfaces to 20 (iL/min running buffer (150 mM NaCl, 10 mM Hepes pH 7.4). Surfaces characterized were 25% BPS (black trace), 25% DOPS (gray trace) and reference or blank (light gray trace). Baseline stability is indicated next to trace in RU change per min. 4.3 Validation of interactive membrane surface prior to detailed kinetic analysis 4.3.1. BSA binding Prior to protein binding experiments, HEPES-buffered saline (HBS) running buffer was changed to HBC (HBS including bovine serum albumin (BSA) and Ca2+). The BSA bound to 0% DOPS (100% POPC) liposomes resulting in a SPR signal increase of 750 RU. Liposomes with DOPS (analyzed from 1 to 25% DOPS) allowed additional albumin adhesion to the membrane surface with a 1000 RU signal increase at 25% DOPS (Figure 12). o (0 c o a. <0 3 a> cc cc < ffl 1000-9 0 0 -800 -700 DOPS Figure 12 BSA binding to immobilized DOPS-containing membrane. After membrane immobilization running buffer was changed from 150 mM NaCl, 10 mM HEPES pH 7.4, to 150 mM NaCl, 10 mM HEPES pH 7.4, 5 mM CaCl2 and 0.1% BSA. Stabilized response changes due to BSA were measured and indicated as a function of DOPS membrane concentration. Error bar represent standard deviation from 4 separate experiments. BSA adsorption to membrane has been described previously and was shown to increase membrane permeability (120). BSA is included in the running buffer as it is an important additive to block any non-specific protein-protein, protein-lipid and protein-underivatized LI chip interactions. If unblocked, these unoccupied hydrophobic sites could give false binding of the protein of interest. After BSA addition to buffer and response stabilization (~ 30 min), no significant change in SPR signal (<15 RU) was observed during 12 hours upon HBC running buffer conditions, which vastly exceeded the time required to complete the protein binding experiments. The 0.1% concentration of BSA used in experiments is below the physiological concentration of albumin (3.5 - 5.5%) (121) and was not expected to interfere with subsequent protein binding analysis. 4>3»2* Mass transport The existence of mass transport (the rate of transport of solution analyte to the sensor surface) was tested by injecting prothrombin over the membrane surface at variable flow rate. This was performed to ensure that the sensorgram data collected was actually kinetic and not influenced by a mass transport parameter. Inspection of the variable flow rate sensorgrams show no difference between the 20 jxL/min and the 62 pL/min data and thus no significant mass transport term appears to be present in the data (Figure 13). Thus, mass transport of the protein in solution to the surface is not rate limiting and can be ignored. As a consequence, the data collected were not diffusion-limited, but rather reaction-limited. Subsequent SPR experiments were performed with a flow rate of 20 (xL/min. Time (s) Figure 13 Prothrombin membrane interaction under variable flow rate. Prothrombin at 0.1 pM was injected (4 min) over a stable 25% DOPS containing immobilized membrane surface and dissociation phase was followed for approximately 7 min. Flow rate was varied: 20 nL/min (dark gray trace) and 62 fiL/min (black trace). Experimental conditions: temperature = 25 °C, running buffer =150 mM NaCl, 10 mM HEPES pH 7.4, 5 mM CaCl2 and 0.1% BSA. All data was reference subtracted using a blank flow cell followed by a double reference correction using a blank running buffer injection. 4.3.3. Specificity of membrane interaction To further validate the application of this method for studying prothrombin-membrane interactions, I determined that binding was highly dependent on the DOPS concentration. The SPR signal increase during association of an approximately physiological concentration of prothrombin to the 25% DOPS membrane was much greater than the 4% DOPS membrane (Figure 14A). The 0% DOPS (100% POPC) membrane surface had very low (< 2% binding compared to 25% DOPS) but significant and specific prothrombin binding. The specificity of the prothrombin membrane binding interaction was shown by the lack of detectable binding for the Gla-domainless prothrombin molecules: prethrombin 1 and des-Gla prothrombin (Figure 14B). In addition, the interaction of prothrombin was completely reversible and dependent on Ca2+, as any remaining prothrombin was completely removed from the membrane surface with EDTA which restored the original baseline (Figure 14A). The phospholipid specificity requirement of the Gla-domain and Ca2+ dependence of the prothrombin-phospholipid interaction verifies that the binding of prothrombin to the phospholipid-Ll biosensor chip resembled the well-characterized interaction of prothrombin with phospholipid liposomes in solution. A => cc d> (ft c o Q, (ft V cc 2000 -1500 -1000 -500 -200 400 600 800 Time (s) 1000 B 3 cc a) (0 c o Q. (0 0) CC 200 400 600 Time (s) 800 1000 Figure 14 Protein and DOPS dependent binding specificity. A. Prothrombin at 1 pM was injected (4 min) over stable 0%, 4%, 10%, 14%, 19% and 25% DOPS (light gray to black traces at increasing DOPS concentration) containing immobilized membrane surface. Dissociation phase was followed for 7 min before a 10 mM EDTA pH 8.0 injection (2 min) removing any remaining membrane adhering protein. Experimental conditions: flow rate = 20 pL/min, temperature = 25 °C, running buffer = 150 mM NaCl, 10 mM HEPES pH 7.4, 5 mM CaCl2 and 0.1% BSA. All data was reference subtracted using a blank flow cell followed by a double reference correction using a blank running buffer injection. B. Des-Gla prothrombin (black trace) and prethrombin 1 (gray trace) at 25 pM were injected over the 25% DOPS surface under the same conditions. 4.4 Fi, F1.2 and prothrombin membrane binding: Affinity and qualitative kinetic analysis Prothrombin and Gla-domain-containing fragments of prothrombin (Fl and Fl .2) were analyzed at various concentrations for their interaction to a 25% DOPS membrane surface. From the 340 s time-point indicating the amount bound within the plateau region of the association phase (Figure 15A-C), the apparent equilibrium dissociation constants ( K f 9 ) for Fl, F1.2 and prothrombin were calculated to be 2.20 +/- 0.23, 2.48 +/- 0.26 and 0.82 +/- 0.10 |iM, respectively, from data fitted to a simple rectangular hyperbola representing a single type of binding (Figure 15A-C insets). 0 200 400 600 800 1000 Time (s) 0 200 400 600 800 1000 Time (») Time (s) Figure 15 Fragment 1, fragment 1.2 and prothrombin binding profiles to 25% DOPS-containing membrane. LUV composed of 25% DOPS were stably immobilized to a Biacore LI sensor chip as described under \"Materials & Methods\". A titration series (indicated concentrations in )j.M) of Fl (A), Fl .2 (B) and prothrombin (C) was analysed for membrane binding. Data collection for association phase (protein injection) occurred between 100 - 340 s (4 min) and dissociation phase from 340 - 775 s (~7 min). Experimental conditions: flow rate = 20 (iL/min, temperature = 25 °C, running buffer = 150 mM NaCl, 10 mM HEPES pH 7.4, 5 mM CaCb and 0.1% BSA. Each cycle of protein data collection was followed by a brief 10 mM EDTA, pH 8.0 injection to remove any remaining protein and returning the response signal to lipid baseline. All data was reference subtracted using a blank flow cell followed by a second reference subtraction using a blank running buffer injection. Above sensorgrams (Response vs Time) were obtained as a titration series overlay using BIAevaluation v4.1. Responses obtained at association phase end were used to generate a binding isotherm fitted to a one-site binding hyperbola (see figure insets) using GraphPad Prism v4. Binding isotherms were used to obtain an K f and R m a x . I next used the kinetic data provided by SPR to relate the differences in binding profile between Fl, F1.2 and prothrombin to on and off rate parameters. The kinetic binding profiles observed for Fl and F1.2 differed substantially from prothrombin in both the association and dissociation phases. Fl and F1.2 membrane binding are dictated by a single extremely fast on and off rates as observed by the rapid response change that is completed within 3-5 s of the start and end of protein injection. Nearly complete dissociation is observed for Fl and F1.2 in the 7 min dissociation phase as responses return to membrane baseline. Binding isotherms shown in insets of Figure 15A (Fl) and Figure 15B (F1.2) were well represented by a mathematical model consistent with a single type of binding (one-site model; r2 = 0.99). Remarkably different than that of Fl or F1.2, prothrombin association occurred by two independent processes characterized by an initial extremely fast on rate (that was similar to the Fl and F1.2 rates) and a secondary slow on rate. This biphasic nature was absent in the Fl and F1.2 binding data. The secondary on rate was influenced by prothrombin concentration, as higher prothrombin concentrations resulted in an increase in this rate. The more complex mechanism of the prothrombin association phase was also reflected in dissociation phase data characterized by an initial fast off rate and a secondary very slow off rate. The initial fast off rate, complete within 3-5 s after protein injection, appears similar to the fast off rate observed with Fl and F1.2. 4.5 DOPS dependence of prothrombin binding Prothrombin binding affinity was dependent on the membrane DOPS concentration as shown by binding isotherms (Figure 16) and is evident from K f and R m a x determined from such plots (Table 2). 2.5 5.0 7.5 10.0 12.5 600 450 300 • 150-DC 0 300-CC o!o 2.5 5.0 7.5 10.0 12.5 i.O 2.5 5.0 7.5 10.0 12.5 2.5 5.0 7.5 10.0 12.5 [prothrombin] (pM) Figure 16 Prothrombin binding isotherms at different percentage DOPS membrane concentration. Data was collected as described in Figure 15 and Materials & Methods. Responses obtained at association phase end were used to generate a binding isotherm fitted to a one-site binding hyperbola at indicated DOPS percentage using GraphPad Prism v4. Binding isotherms were used to obtain an Kf and R m a x reported in Table 2. T a b l e 2 B i n d i n g aff inity o f p r o t h r o m b i n , f r a g m e n t 1, f r a g m e n t 1 .2 a n d x l - p r o t h r o m b i n to D O P S - c o n t a i n i n g m e m b r a n e 3 outer leaflet DOPS maximum protein protein % DOPS immobilized (molecules)/mm R m a x ( R U ) r2 bound (molecules)/mm n prothrombin 4 8.94 x 1010 20.0 +/- 7.0 1036 0 . 9 9 3 9 0.87 x 1010 10.3 prothrombin 7 1.98 x 1011 9.1 +/- 1.8 1689 0 . 9 9 8 4 1.42 x 1010 13.9 prothrombin 10 2.33 x 1 0 u 5.0 +/-0.5 1995 0 . 9 9 8 7 1.67 x l O 1 0 14.0 prothrombin 14 3.36 x 1011 3.2 +/-0.2 2758 0 . 9 9 9 5 2.31 x 1010 14.5 prothrombin 19 4.14 x 1011 1.66+/-0.14 2856 0 . 9 9 9 5 2.39 x 1010 17.3 prothrombin 25 4.45 x 1011 0.82+/-0.10 3309 0 . 9 9 9 5 2.77 x 1010 16.1 F1.2 25 6.28 x 1011 2.48 +/- 0.26 879 0.9911 1.52 x 1010 41.3 F.l 25 6.28 x 1011 2.20 +/- 0.23 1122 0.9911 3.22 x 1010 19.5 xl-prothrombin 25 6.06 x 10\" 0.28 +/- 0.07 3107 0 . 9 9 9 8 2.6 x 1010 23.3 (a) explanatory information The number (n) of DOPS molecules bound per protein molecule can be estimated if the following assumptions are made: 1) all outer leaflet phospholipids are accessible, 2) '/2 of the DOPS molecules are present on the outer leaflet and participate in protein binding, 3) 1000 RU correspond to 0.92 ng immobilized lipid/mm2 chip (Cooper etal. (1998) Biochim Biophys Acta 1373, 101-11), 4) the value of n is determined at R lrax, a value derived from a direct fit of a rectangular hyperbola using a 1:1 binding model and 5) 1000 RU protein binding response is equivalent to 1 ng protein/mm2 (www.biacore.com). The trend observed in prothrombin binding affinity due to DOPS concentration was similar to previous studies using bovine brain PS and SUV characterized by 90° light scattering (50). Dissociation constants determined at the lower DOPS concentration (e.g. 4% and 7%) have higher associated uncertainty and must be considered with caution as they were obtained with prothrombin concentration below or approaching the determined Kf' concentration. Prothrombin binding affinities were not determined for DOPS concentration below 4% as vast amounts of protein would have been required. Prothrombin binding profiles for various DOPS membrane concentrations showed the same binding profile characterized by two apparent on rates and two apparent off rates as described for 25% DOPS (Figure 15C). It is evident that the biphasic binding mechanism described for prothrombin is intrinsic to the protein and is independent of DOPS concentration. The phospholipid binding site on prothrombin and its derivates can be approximated in terms of the number of DOPS molecules that define a binding site. Although the exact structure of the immobilized membrane is not known, the number (n) of DOPS molecules bound per prothrombin molecule can be estimated if the following assumptions are made a) all outer leaflet phospholipids are accessible, b) half of the DOPS molecules are present on the outer leaflet and are accessible to participate in protein binding, c) 1000 RU correspond to 0.92 ng immobilized lipid/mm2 chip (122), d) the value of n is determined at R m a x > a value derived from a direct fit of a rectangular hyperbola using a 1:1 model and e) 1000 RU protein binding response is equivalent to 1 ng protein/mm2 (www.biacore.com). The estimated value of n at different DOPS membrane concentrations is consistently between 10-17 DOPS molecules bound per prothrombin molecule and appears to be independent of DOPS concentration. This is consistent with the current view of peripheral membrane protein binding, in which a given number of DOPS molecules are required to comprise a protein binding site that is stable on the membrane for long enough time to be detected by SPR. Not necessarily are all DOPS molecules involved in direct contact with the protein 3-D structure but are required to make up a \"binding site\". This value of n agrees with a previous report using natural phospholipid SUV and bovine prothrombin studied by light scattering which determined n = 10-20 (50). 4.5.1 Non-cooperative binding There was no evidence for positive or negative cooperativity since Scatchard plots were linear (Figure 17 A) and a Hill plot of binding data had a Hill coefficient (h) of ~1 (Figure 17B) regardless of DOPS concentration analyzed. Data with cooperative binding have curved Scatchard plots and a Hill plot with a slope of either >1 (positive cooperativity) or <1 (negative cooperativity). Both plots indicate that prothrombin membrane binding is a non-cooperative process. Thus complex binding process observed for prothrombin to the DOPS-containing membrane must be explained by another process. 0.00 0.00 0.01 0.02 0.03 0.04 0.05 0.06 r B ® O) o -1.0 -0.5 0.0 0.5 1.0 1.5 log [free prothrombin] Figure 17 Cooperativity analysis for prothrombin interaction with DOPS-containing membrane. Experimental data from SPR experiments were fitted to a Scatchard plot (A) and to a Hill plot (B). Scatchard analysis: r is expressed as bound prothrombin per total outer leaflet DOPS. 25% DOPS (•), 19% DOPS (•), 7% DOPS (•) are linear. Hill analysis: O is equal to Y/(l-Y), where Y is the fraction of DOPS binding sites occupied and defined as r/n. At 25% DOPS; n = 0.060 prothrombin sites/DOPS; (•), 10% DOPS; n = 0.077 prothrombin sites/DOPS; ( • ) , 4% DOPS; n = 0.072 prothrombin sites/DOPS; (T). The Hill coefficient (h) is the slope at log 9 = 0 and determined to be 0.956 +/- 0.039 at 25% DOPS (•), 0.995 +/- 0.024 at 10% DOPS (A) and 1.001 +/- 0.059 at 4% DOPS ( • ) . 4.6 On the binding mechanism of prothrombin 4.6.1 Prothrombin membrane binding involves a linked reaction mechanism To investigate the complexity of prothrombin-membrane binding further, prothrombin was injected at a constant concentration (5 |oM) over an immobilized 25% DOPS membrane surface for variable periods of time ranging from 4 to 25 min. After the injection, dissociation of bound prothrombin was followed by flowing only buffer over the chip. The dissociation rate was much slower after long association times (Figure 18). Clearly in the case of prothrombin binding, the curves were not identical as the extent of initial dissociation decreased with increasing association time. The change in the stability of the prothrombin-membrane complex over time indicates that the binding is controlled by an obligate sequential mechanism or linked reaction involving an intermediate. In case of independent reactions the complex remains stable over time and the dissociation curves are identical for short and long analyte injections (associations). This was clearly not observed verifying that the reaction is indeed linked. o & -200 d) DC - 3 0 0 -0 200 4 0 0 Time (s) 600 800 Figure 18 Membrane dissociation of prothrombin is dependent on duration of association phase. Each cycle of protein data collection was obtained as described in Figure 15 with the following exceptions: A fixed concentration of prothrombin (5 nM) was used for each cycle and allowed to associate either for 4, 8, 12, 20 or 25 min. Data was transformed (using BIAevaluation v4.1) to 0 s and 0 RU at the end of association phase resulting in the above displayed dissociation phase overlay. Only dissociation phase is shown. Time of association phase in minutes is indicated on individual traces. 4.6.2 Stability of membrane bound species is dependent on prothrombin concentration An interesting observation regarding the dissociation data is that at concentrations between 0.05 and 1.5 (iM, prothrombin dissociated 80%, whereas above 1.5 |iM, a trend is observed in which increasing prothrombin concentrations result in a decrease in percent dissociation from the membrane (Figure 19). After 30s of dissociation, the percentage of prothrombin released was; 2.5 fiM, 76%; 5 jiM, 69%; 7.5 fiM, 64%; 10 pM, 61%; 20 fiM, 49%. These results indicate that a more stably bound form of prothrombin becomes increasingly prevalent at higher prothrombin concentrations. This was not observed with either Fl or F1.2. The formed prothrombin-membrane complex appears to be stabilized at higher prothrombin concentrations, plausibly via an intermolecular protein interaction not present in F1.2. Figure 19 Initial extent of membrane dissociation is dependent on prothrombin concentration. Each cycle of protein data collection was obtained as described in Figure 15. The amount of prothrombin dissociated 30 s into the dissociation phase was quantified (% of R m a x ) and plotted against prothrombin concentration. 4.6.3 Analysis of cross-linked prothrombin dimer membrane binding To test the hypothesis that the slow phase may be due to prothrombin dimerization, a purified, cross-linked prothrombin dimer (xl-prothrombin) was used to determine if the complex mechanism seen for prothrombin-membrane binding could be explained by membrane-induced dimerization. xl-prothrombin was graciously prepared by Dr. Peter Anderson (University of Ottawa) using an activated polyglutamate cross-linking agent (10 residues in length) in the presence of unilamellar vesicles and 5 mM CaCl2 and purified by gel filtration as described previously (17). Prothrombin could only be cross-linked in this reaction in the presence of Ca2 + and PS -containing liposomes (17). Purified xl-prothrombin was analyzed at several concentrations for its interaction with a 25% DOPS surface. From the overlaid dose-response binding curves (Figure 20A), the K ' f was calculated to be 0.28 +/- 0.07 pM from data fitted to a simple rectangular hyperbola (Figure 20B). The interaction of xl-prothrombin to membrane was Ca2+ dependent and completely reversible by EDTA, like the other prothrombin derivatives. From inspection of dissociation phase data, it is apparent that cross linking two prothrombins had a profound effect on dissociation kinetics compared to Fl .2 or prothrombin. The off rate was very slow and a resembled the second off rate observed with prothrombin. A fast off rate was not seen as was observed for both F1.2 and prothrombin. 0 200 400 600 800 1000 Time (s) B ^ 2500 0.2 0.4 0.6 [xl-prothrombin] (pM) Figure 20 xl-prothrombin dimer binding profile to 25% DOPS-containing membrane. LUV (100 nm extruded pore size, composition: 25% DOPS 75% POPC) were stably immobilized to a Biacore LI sensor chip. A. A titration series (indicated concentrations in nM) of xl-prothrombin was analysed for membrane binding. Data collection for association phase (protein injection) occurred between 100 - 340 s (4 min) and dissociation phase from 340 - 775 s (~7 min). Experimental conditions: flow rate = 20 H.L/min, temperature = 25 °C, running buffer = 150 mM NaCl, 10 mM HEPES pH 7.4, 5 mM CaCl2 and 0.1% BSA. Each cycle of protein data collection was followed by a brief 10 mM EDTA, pH 8.0 injection to remove any remaining protein and returning the response signal to lipid baseline. All data was reference subtracted using a blank flow cell followed by a double reference correction using a blank running buffer injection. Above sensorgrams (Response vs Time) were obtained as a titration series overlay using BIAevaluation v4.1. Response obtained at association phase end were used to generate a binding isotherm fitted to a one-site binding hyperbola (B) using GraphPad Prism v4. Binding isotherms were used to obtain an Kf p and R m a x . 4*6.4 Homogeneous Analyte To further refine my experimental system, I investigated the homogeneity of our analyzed analyte molecules by SDS-PAGE. For heterogeneous protein mixtures, a phenomenon of sequential binding has been described, in which a relatively abundant protein with low affinity binds the membrane surface and is replaced in time by a much less abundant protein with higher affinity (123). A protein mixture could explain the complex membrane binding seen with prothrombin. However, this was not the case as all proteins used were judged to be greater than 98% pure from an overloaded Coomassie blue stained gel (Figure 21) and thus any type heterogeneous analyte model can likely be ruled out. Heterogeneity in terms of prothrombin solution multimers (i.e. monomer, dimer, etc.) can also be ruled out as the concentration used in experiments (0.05-20 pM) are predominantly below the concentration where prothrombin is known to start dimerizing in solution (14 pM, (18, 19)) and 10 to 100-fold below the Kd of a dimerization reaction in solution. In addition, Sere et. al.(62) showed that no multimeric forms of purified prothrombin exist when analyzed by native PAGE in the absence of SDS. Taken together, these findings rule out that the complex binding results for the prothrombin-membrane interaction are caused by the existence of prothrombin multimers in solution, prothrombin degradation species, or other contaminating molecules. xl-prothrombin prothrombin fragment 1.2 fragment 1 1 2 3 4 5 Figure 21 Characterization of proteins by SDS-polyacrylamide gel electrophoresis. Proteins (5 pg) were applied to a gradient SDS gel (from 4 to 20% polyacylamide) under reducing conditions. Lane indicated are 2, Fl; 3, F1.2; 4, prothrombin; and 5, xl-prothrombin. Lane 1 contains the molecular weight markers with indicated protein sizes. Coomassie Brilliant Blue was used for staining. 4.7 Estimation of kinetic parameters Kinetic data were analysed by global fitting of multiple concentrations of the various models made available by the BIAevaluation software. Fl and F1.2 data gave mediocre fits using a simple 1:1 model as indicated by obvious deviation from the experimental data curves (Figures 22A and B) and high chi squared (x2) values (Table 3). 1:1 model 800 800 C xl-prothrombin bivalent model 2000 -800 D prothrombin 1:1 model 800 Figure 22 Determination of kinetic parameters of prothrombin and derivatives binding to 25 % DOPS containing membrane using SPR. Fitting of kinetic data from protein titration experiments using BIAevaluation v4.1. A representative data curve of Fl (A, 1.5 pM, global fitting), F1.2 (B, 1.5 pM, global fitting), xl-prothrombin (C, 0.55 pM, global fitting) and prothrombin (D, 1.5 pM, single concentration fitting) is shown interacting with a 25% DOPS membrane surface. Continuous and dotted lines correspond to experimental and fitted data, respectively. Indicated models gave best fits from experimental data. Table 3 Parameters obtained from global fitting. F1a F1.2a xl-prothrombinb prothrombin all datac (2nd dissociation phase)\" fronO^S-1) 36,600 42,000 8,460 nd — • Kringle Gla X cross- l inker • monomer-membrane complex dimer-membrane complex total bound: monomer + dimer complexes fa»t mL ^^ JC-ODf ^OQQQoodc«, weakly bound (monomer-membrane complexes) A tightly bound (dimer-membrane complex) D. 5-1000 ae V g 500 a 8 « a. 1000 2000 Time (s) 3000 4000 Figure 24 Model for prothrombin membrane binding based on a linked reaction mechanism involving membrane induced dimerization of the Protease domain. A. Fl, F1.2, B. xl-prothrombin and C. prothrombin putative binding reaction schemes on immobilized DOPS-containing membranes. D. Prothrombin (5 pM) was injected for either 4 min (gray trace) or 25 min (black trace) over a 25% DOPS membrane immobilized surface followed by a 10 min dissociation phase. Experimental conditions and data analysis were as described in figure 2. At selected time points, traces marked at \"x\" were analysed using the linked reaction mechanism for predicted contributions of monomer and dimer to total membrane bound prothrombin (black square-double gray triangle). Numerical calculations are based on the availability of 100 prothrombin membrane binding sites. Symbols used for weakly membrane bound monomelic prothrombin and tightly membrane bound dimeric prothrombin are black square and double gray triangle, respectively. According to the proposed linked kinetic model, the prothrombin-membrane association would be followed by a similar desorption process. Initial membrane dissociation is dominated by monomelic prothrombin molecules controlled by a fast dissociation rate. Remaining membrane bound dimeric prothrombin has a very slow dissociation phase that may either directly dissociate from the membrane as a dimer (not shown) or slowly separate into membrane bound monomers that dissociate rapidly from the membrane individually (Figure 24C). The off rates derived for the second dissociation phase of prothrombin and the single dissociation phase of xl-prothrombin are slow (4 and 6 x 10\"4 s\"1, respectively), whereas the dissociation rates for Fl or F1.2 (-0.2 s\"1) are comparatively much faster. These observations argue that prothrombin dimerizes on the membrane to account for the slow dissociation not observed for Fl or F1.2. Consistent with the model a prolonged association time of prothrombin with membrane would allow a more stable form of prothrombin (i.e. dimers) to assemble on the membrane, which I observed and illustrate schematically (Figure 24D). 4.9.5 Previous studies concur with membrane induced dimerization Numerous studies have been conducted to help understand the mechanism by which prothrombin interacts with PS-containing membranes. While several of these have concluded that single exponential association and dissociation can occur (64, 82), others have predicted a biphasic mechanism using a variety of experimental approaches: A) Biphasic association was observed by two independent laboratories (83, 86) following prothrombin binding to spread phospholipid monolayers using ellipsometry. The secondary kinetic event was not discussed (83) or attributed to non-specific adsorption (86). B) Deviations from a simple binding model were obtained from prothrombin (1 fiM) association (<2 min) binding data assessed by TTR-FPR fluorescence recovery curves (85). In their discussion, Pearce et al. suggested the formation of dimers on the membrane surface as a possible explanation. C) Very similar dissociation data as I report here has been noted using QCM-D. Richter et al. showed only 80% prothrombin dissociated from supported lipid bilayers after 1 hr. (87); complete dissociation was shown with EDTA. Their measurements of dissipation deviated from a one-site model and this deviation was attributed to membrane induced rigidification of prothrombin, which they hypothesized was due to lateral interactions on the membrane surface. Although these examples from the literature support our biphasic model, there are also examples in the literature where a simpler mechanism was described as discussed in the following section. 4.9.6 Apparent discrepancies from previous studies about the secondary kinetic event Binding data from stopped flow intrinsic fluorescence, stopped flow FRET and light scattering experiments have concluded the prothrombin-membrane reaction proceeds by a simple model lacking a secondary kinetic step (64, 82) (also see introduction). Data fit well to one-site model under the conditions that were analyzed and gave good estimations of the initial association and dissociation kinetic rates. Observations of the secondary kinetic step were not made for several reasons. Firstly, the prothrombin concentrations (less than 2.8 pM (82) and 1 |iM (64)) used were below membrane binding site saturation. As a consequence the secondary kinetic step (a concentration dependent, protein - protein interaction dependent event) was minimal and thus not noted. At higher concentrations (12.5 fiM (82)) the authors acknowledged that association became complex, but the cause of this complexity was not addressed. Secondly, time of analysis of association (500 ms(S2) and < 1 s (64)) and dissociation (< 5 s (64)) were not long enough at concentrations used to observe a secondary kinetic step. Thirdly, dissociation experiments (Figure 4a in Wei et al.) confirmed an absence of complete prothrombin release after 5 sec of data analysis indicating dissociation had not gone to completion. The prothrombin concentration (0.5 jjM (64)) used in this experiment gave a percent dissociation (-80% completion) that is in agreement with our results. Taken together these studies gave an accurate account of the fast kinetic event that is completely in line with our observations of the fast phase of prothrombin membrane binding. By evaluating on and off processes simultaneously, the dimerization model presented here is consistent with all of the prior reports. 4.9.7 Other models to explain prothrombin membrane binding SPR experiments suggest that prothrombin membrane induced dimerization is likely the dominating process responsible for the second kinetic event apparent in the biphasic prothrombin binding data. Much criticism has been directed for attempting to define a mechanistic model describing the experimental data for prothrombin membrane binding. In order to address these criticisms, other models that can explain peripheral membrane binding reactions are introduced and outlined in Table 4. In these reactions peripheral membrane protein, P is in equilibrium with membrane, M. Rate designations (fast or slow) and reaction descriptions are included that theoretically may produce data similar to the experimentally described prothrombin-membrane binding data. The reaction can be linked and thus involve an association time dependent event or not linked, as the case for a simple bimolecular interaction of a 1:1 (Langmuir) model. Reactions are presented in a way that could potentially result in the observed experimental prothrombin data. Thus, rates, binding types and strengths were chosen for each model to accommodate prothrombin-like data. Table 4 Binding models for peripheral membrane proteins. Model Name Reaction Linked Reaction (Association Time Dependent Event) 1:1 (Langmuir) model P + M v. v MP No description: P binds with a single proteinaceous site to a single site on the membrane. 1:2 heterogeneous immobilized P + Ma v PMa No ligand model P + Mb ^ PMb description: P can bind one of two independent membrane sites; Ma or Mb. The two sites may represent different PS clusterings 2:1 heterogeneous analyte model P0 + M f a s t s PaM yes linked slow P b + M s PbM description: Pa is a fast, weak binder and Pb is a slow, tight binder. Pa and Pb can bind to a competing site on the membrane. Pa can be displaced by Pb over time. fast slow 1:1 conformational change model P + M v v MP v s MP* yes description: Two conformational states are present; P and P*. Only P can bind the membrane and then undergoes a slow conformational change producing a tightly bound species P*. The conformational change is much slower than the binding event. fast slow 1:1 multivalent analyte model P + M ^ MP ^ MP* yes description: P can bind membrane with 2 proteinaceous sites. The binding of the 1 -site, producing MP, facilitates the binding of the second proteinaceous site, producing a tightly bound molecule MP*. The second binding interaction is much slower than the initial binding event. fast slow 2:1 membrane induced P + M s MP + MP N MPP yes dimerization model description: P is a fast, weak binder. P can slowly dimerize with another membrane bound P forming a membrane bound dimer; MPP. P is peripheral membrane protein M is membrane For example, a 1:2 heterogeneous immobilized ligand model can be characterized by two independent binding sites (Ma and Mb). Each membrane site may have a kinetically distinctive binding rate constant designated simply as fast and slow, making each site unique. Since each membrane site is independent the reactions at each site would not be linked, ruling out such a reaction to describe the prothrombin data. In addition it has been determined that mixed PS/PC membranes are homogeneous and do not contain acidic lipid domains or clusters prior to and after prothrombin binding (112) and thus only a single type of binding site, M is available. Thus any theory of membrane heterogeneity causing complex binding has to disagree with these previous studies and our linked reaction experiment. Finally, F1.2 is a useful control to further rule out such a model to explain prothrombin membrane binding mechanism, as F1.2, having an identical membrane contact site, showed simple binding kinetics. Other models and their descriptions in Table 4 include a 2:1 heterogeneous analyte model, 1:1 conformational change model, 1:1 multivalent analyte model and a 2:1 membrane induced dimerization model. The description of these models is not discussed in further detail here. Refuting evidence against these models and other processes to explain the second kinetic event is outlined in Table 5. Table 5 Prothrombin membrane induced dimerization: Is there another model or process to explain the second kinetic event involved in membrane binding? Proposed model or process Refuting evidence 1) non-specific adsorption to chip surface • F1, F1.2, prethrombin 1 and des-gla prothrombin do not show second on/off rate 2) multiple membrane bound species A) heterogeneity in analyte • SDS-PAGE analysis for proteolytic fragmentation (2:1 heterogeneous analyte model) • F1.2 would have the same minor heterogeneity such as carboxylation and glycosylation as prothrombin B) multi-step binding process via protein sub-sites • F1 and F1.2 do not show although Gla domain is the only membrane e.g. PS pocket(s), oj-loop, electrostatic interactions contact site and is identical to prothrombin (1:1 multivalent analyte model) C) conformers • None, No evidence for different membrane bound conformers either. (1:1 conformational change model) - Would be a Protease dependent Gla conformational change induced by membrane binding - Can not be ruled out 3) two dimensional surface causing binding anomalies • F1 & F1.2 do not show. Mass Transfer experiments. e.g. crowding, rebinding, electrostatic or • prothrombin at low concentration has second kinetic event orientational effects • Surface is more representative of platelet membrane immobilized in a clot under flow compared to SUV in solution Most of the models can be eliminated as an appropriate model by using F1.2 and other fragments of prothrombin as comparative controls. For example non-specific adsorption to the LI sensor chip (e.g. to carboxymethyl dextran, unblocked lipophilic anchors, or Au surface) could cause a slow association process and potentially result in a stable interaction. However, since none of the other fragments of prothrombin show this type of interaction process and these fragments together cover the entire primary amino acid sequence of prothrombin such a process seems unlikely. Another model that should be addressed is the heterogeneous analyte model. Although SDS-PAGE analysis shows greater than 98% purity in the prothrombin preparations used in the analysis, prothrombin is well known to have multiple heterogeneous variants of similar molecular weight not distinguishable on SDS-PAGE. Variants in gamma-carboxylation (10 sites) and N-linked glycosylation (3 sites) have been described for plasma purified prothrombin (9, 125). Again, the possibility of two variants with different membrane affinity; one being a fast, weak binder and the other a slow, tight binder could exist. Over time a linked reaction would ensue resulting in the replacement of weak binders by tighter binders giving this slow kinetic phase. However, this situation seems unlikely in light of the fact that F1.2 is a product of plasma purified prothrombin and thus contains the same heterogeneity of Gla residues and 2 of 3 carbohydrate sites (the 3 r d site being on the protease domain). Another concern that should be addressed here on the topic of heterogeneous analyte model is the issue of a prothrombin dimer contaminating the monomer in solution. A heterogeneous monomer/dimer mixture would undoubtedly complicate the analysis of prothrombin kinetic data and would represent a form of the heterogeneous analyte model. As was apparent from xl-prothrombin data, a natural solution dimer will have different membrane kinetics compared to monomer. It has been shown that bovine prothrombin can dimerize in solution starting at a concentration of 14 pM as assessed by cross-linking experiments (19) and sedimentation measurements (18). The concentration used here for prothrombin kinetic experiments were for the most part well below this solution dimerization concentration. In addition, Sere et. al. (62) showed that no multimeric forms of purified prothrombin exist when analyzed by native PAGE in the absence of SDS. Taken together, these findings rule out that the above described results for prothrombin-membrane binding characterized by a linked reaction mechanism are caused by prothrombin degradation products, variants, solution phase dimers or other contaminating molecules. The binding of prothrombin at different sub-sites on the membrane contact site as described by the 1:1 multivalent analyte model could potentially also result in a biphasic binding mechanism. Specific interactions between prothrombin and membrane have been under extensive investigation and considerable knowledge is now available. The presence of multiple interactive sites mediated by PS pockets, hydrophobic and electrostatic interactions (for details see introduction) is not disputed however these interactions likely occur too fast and on a time scale indistinguishable by a SPR experiment. If multivalent interactions were responsible for the biphasic prothrombin-membrane interaction, then one would expect similar slow secondary kinetics to be seen with F1.2, which was not the case. One process in the prothrombin-membrane interaction that cannot be ruled out at this time is the existence of more than one membrane-bound prothrombin conformer with varying membrane affinities. A membrane induced conformational change does occur for bovine prothrombin (29, 124, 126)', however, multiple membrane-bound conformers have not been described. If present, these membrane-bound conformers would depend on the presence of the Protease domain as F 1.2 lacking the Protease domain does not show the characteristic prothrombin biphasic kinetic data. At this time I can only speculate on these membrane bound conformers. 4-9-8 Significance of prothrombin dimerization The K'* p (0.82 +/- 0.10 pM) determined using SPR was similar to Kd values reported previously for the interaction of prothrombin with membranes composed of 25% PS using other techniques, including measurements with 90° light scattering and fluorescence (0.11 - 0.86 pM dependent on liposome diameter (64, 68)) and gel filtration chromatography (1.1 pM) (67). Also consistent with my observations, prothrombin membrane binding was determined to be 2-3 fold higher in affinity compared to Fl and F1.2 as was shown previously by other studies (67, 85, 127). Neither Kringle domain has been shown to affect membrane binding, as prothrombin Kringle domain-deleted mutants have identical membrane affinity compared to native prothrombin (67). Other studies have also suggested that portions of prothrombin that are C-terminal to Fl also contribute to membrane binding. Scanning calorimetry of prothrombin suggested a difference in denaturation patterns of the C-terminal region of prothrombin when it was associated with membranes compared to free in solution (92, 126). These results indicate a role for the Protease domain in overall membrane binding. Although Fl, F1.2 and prothrombin make direct membrane contact through their identical Gla domains, only prothrombin forms a bivalent, high avidity membrane interaction via our postulated dimerization binding mechanism. The dimerization site in the Protease domain is proposed to be near proexosite I of membrane bound prothrombin, as the molecules can be cross-linked at proexosite I under similar experimental conditions (17). To support the prothrombin membrane-induced dimerization hypothesis, we purified a previously reported stable prothrombin dimer (xl-prothrombin) and demonstrated reduced K' cipp and slow kinetic properties compared to Fl and F1.2 binding to membranes. Thus, protease domain-dependent dimerization may account for previously reported discrepancies between prothrombin and Fl or Fl .2 binding to membranes. At this time, I can only speculate on the physiological role of prothrombin dimerization on procoagulant membrane in vivo. Membrane-induced dimerization will increase the half-life of prothrombin on the membrane. This may increase its availability to bind to the prothrombinase complex thereby increasing the subsequent production of thrombin. Chapter 5. Comparison of Coagulation Gla Protein-Membrane Interactions 5.1 Rationale To further understand how the Gla domain and regions removed from the Gla domain facilitate membrane binding, other Gla proteins were analyzed for their membrane binding properties. With the binding profile differences observed between F1.2 and prothrombin, it was anticipated that the other Gla proteins would also categorize themselves into one of two types of binders: simple 1:1 or multiphasic. Although many studies have characterized membrane binding of individual Gla proteins, few studies have attempted to characterize them in a single study making comparisons difficult and unclear. Thus, many binding property comparisons that have been made between Gla proteins, were performed with a variety of parameters that made them difficult to compare. These include the use of a wide spectrum of techniques (see introduction), protein origin (usually human versus bovine), membrane type (small unilamellar vesicles, large unilamellar vesicles, immobilized monolayer versus bilayer) and membrane composition (percentage and types of phospholipids). Consequently at this time, it is difficult to compare the membrane binding properties of Gla proteins. In terms of binding strength, the seven vitamin K-dependent clotting proteins vary tremendously in their affinity for PS-containing membranes; with protein S the strongest binder (mid nM range) and protein C/Factor VII the weakest binders (low-mid (iM range). An explanation for this 100-1000-fold difference in binding affinity is still under scrutiny. A most obvious reasoning for this difference would be attributed to differences in the protein regions involved in the membrane contact site. A sequence comparison of the Gla domains (Figure 25), however, shows a high sequence identity (55-68% amino acid identity for any two sequences) with 16 of 44 residues of the domain being strictly conserved. propeptidase chymotrypsin cleavage w _ | 0 0 p cleavage I prothrombin #: 1 10 30 40 prothrombin ANTFLXXVRKGNLXRXCVXXTCSYXXAFXALXSSTATDVFWAKY protein C ANSFLXXVRHSSLXRXCIXXICDFXXAKXIFQNVDDTLAFWSKH FIX NSQKLXXFVQGNLXRXCMXXKCSFXXARXVFXNTXRTTXFWKQY FX ANS F LXXM KKGH LX RX CMXXTC S YXXARXVFXD SD KTNX FWNKY FVII ANAFLXXLRPGSLXRXCKXXQCS FXXARXIFKDAXRTKLFWISY protein Z AQSYLLXXLFXGNLXKXCYXXICVYXXARXVFXNXVVTDXFWRRY protein S ANSLLXXTKQQNLXRXCIXXLCNKXXARXVFXNDPXTDYFYPKY * * * * * . * * * * * * * * * . . * * . Figure 25 Amino acid sequence alignment of the Gla domain of human Gla proteins involved in blood coagulation. Gla sequence is shown defined between propeptidase and chymotrypsin cleavage sites. Strictly conserved residues are indicated with an asterisk (*). Highly conserved residues are indicated with a colon (:). Conserved residues are indicated with a period (.). Positions in the sequence at which gamma-carboxylation of glutamic acid residues is either known to occur or may occur is indicated by X. The position of the disulfide loop within the Gla domain is also indicated. Amino acids are colored according to their chemical properties: non-polar amino acids (G,A,V,L,I,P,W,F,M) are red, polar (S,T,Y,C,Q,N) green, basic (K,R,H) purple, and acidic (D,X) amino acids are blue and black, respectively. The numbering at the top refers to the prothrombin sequence. Prothrombin sequence 1 -44 was used as the query sequence to search the protein database using ClustalW multiple sequence alignment program (128). Furthermore, a structural comparison of the Gla domains in the presence of saturating calcium made possible from x-ray crystallography data shows that the protein backbone orientation as well as position of seven calcium ions required for membrane binding are strictly maintained in these homologous proteins (Figure 26). Figure 26 Structural comparison of three homologous Gla domains. Bovine prothrombin (2PF2, (70)) white trace (residues 1-45 and associated seven Ca2+ (white ball)) was used as reference structure for fitting bovine Factor X (1IOD, (129)) in red trace (residues 1-44 and associated seven Ca2+ (red ball)) and human Factor IX (1CFI, (73)) in orange trace (residues 2-45). Factor IX associated Ca2 + are not shown for clarity. Hydrophobic residues (Phe4, Leu5 and Val8 from left to right) of the W-loop known to interact with the membrane interior are also indicated on the prothrombin structure. Gla and other residues are not shown for clarity. Crystal coordinates were obtained from the Protein Data Bank using indicated accession numbers. Deep view/Swiss-PdbViewer 3.7 was used to make illustration. Many of the strictly conserved residues undoubtedly are required for domain structure and highly conserved residues are required for the membrane interaction involving hydrophobic (e.g. co-loop as illustrated in Figure 26), Ca2+ bridging, electrostatic and specific PS headgroup interactions as detailed earlier (see introduction). Again a most obvious reason for the affinity difference would be in these minor differences in sequence at the least conserved amino acid positions. However, numerous studies have tried to modulate membrane binding of individual Gla proteins by altering individual or several, non-strictly conserved residues (altering these has profound devastating effects on membrane binding presumably by destabilizing the Gla domain) of one Gla proteins to another. These studies have been unfulfilling as they have failed to drastically alter the affinity of the mutant protein. Thus it seems unlikely that several or (as often proposed) a single residue in the Gla domain involved in membrane contact cause this astounding 100-1000-fold difference in binding affinity. It seems apparent that sites removed from the membrane contact site may play a role in modulating the membrane binding process of the Gla proteins. So the question remains in the field: what really is the basis for this difference in membrane affinity among the Gla proteins? It was hoped that in light of the binding profile differences observed with prothrombin and F1.2, SPR data could shed light on this question. 5.2 Results 5.2.1 Purity analysis by SDS-PAGE In order to assess the binding properties of all coagulation Gla proteins (except Factor VII - due to cost restrictions), proteins were first analyzed for purity by SDS-PAGE (Figure 27). From a heavily overloaded (in order to observe any heterogeneous protein) gradient gel electrophoresed under non-reducing conditions to maintain the integrity of the disulfide bonded two-chained proteins, it is apparent that the majority of proteins migrate as a single band. These single band proteins (> 98%) included prothrombin, protein Z, protein S, Factor IX and Factor X. Factor Xa predominantly ran as a doublet equally represented by Factor Xaa (upper band) and Factor Xap (lower band) as a result of a known autocatalytic cleavage of a C-terminal fragment of its heavy chain (130). Factor IXa predominantly ran as an expected single band (131) (-90%); however, two smaller bands presumed to be degradation products were also apparent. Protein C ran as a known doublet (132) equally represented by protein C a (upper band) and protein Cp (lower band) due to proteolytic cleavage of a susceptible C-terminal peptide (Mr = 3000) of the heavy chain. No functional distinctions between a - and P-protein C have been observed. Activated protein C also had this heavy chain doublet in addition to a minor degradation fragment of low abundance (< -10%). I concluded that the proteins were of expected homogeneity and were suitable as analytes for SPR characterization. Figure 27 Characterization of proteins by SDS-polyacrylamide gel electrophoresis. Proteins (7.5 (ig), as indicated in lanes were applied to a gradient SDS gel (from 4 to 20% polyacylamide) under non-reducing conditions as described in Materials & Methods. Coomassie Brilliant Blue was used for staining. 5.2.2 Gla protein binding profile to 25% DOPS-containing membrane Gla proteins of human origin were analyzed over a wide concentration series for their binding properties to an identical 25% DOPS membrane immobilized to a biosensor surface. This allowed a direct comparison of their binding properties in a non-biased fashion. As expected, all Gla proteins bound to the 25% DOPS membrane surface in a concentration dependent manner (Figure 28A1-I1). Response (RU) Ni w S § § 8 o © o o Response (RU) Response(RU) Response(RU) D1 Factor IXa [Factor IXa](uM) 200 400 600 800 1000 1200 1400 1600 Time (s) E18oo,PrPtqin,C 0.5 1.0 [protein C] (uM) 1000 1500 2000 2500 Time (s) F18oo activated protejn C 0.5 1.0 [APC] (nM) 1000 1500 Time (s) 2000 2500 G 1 p r o t e i n S => cc 0) tn c o o_ tn a> cc 4000 -3000 • 2000 1000 -G2 4500 -5.0 0.1 0.2 0.3 0.4 0.5 0.6 [protein S] (uM) Time (s) H ^ o o P r o t e i n Z => tr o tn c o a. FXa > FIXa > APC ~ FIX > FX ~ prothrombin > protein Z > protein C with determined K f values indicated in Table 6. Another intriguing finding is a comparison of the total membrane binding capacities for the individual Gla proteins (Table 6). Factor Xa has the highest surface density for all the Gla proteins ( R m a x = 6530 ry RU, 6.53 ng =141.9 fmol bound per mm chip) analyzed. Relative to Factor Xa, the binding capacity from highest to lowest is FXa > FIXa > FX > protein S > prothrombin > protein Z > protein C > FIX ~ APC with maximum binding percentage indicated in Table 6. Although these proteins have such high sequence similarity and conserved structure, the preferred membrane analyte varies by almost a factor of 10 when Factor Xa and Factor IX/activated protein C are compared. 5.3 Discussion 5.3.1 Binding comparison of Gla proteins Using surface plasmon resonance, I have investigated the detailed kinetic interaction between Gla proteins and a phosphatidylserine containing membrane surface. Most studies focusing on the Gla domain in relationship to membrane binding have evaluated single Gla proteins and a series of mutants residing in the N-terminal half of the Gla domain (e.g. protein C/APC (133), prothrombin (68), Factor VII (134)), particularly hydrophobic residues thought to insert into the membrane. These experiments primarily focused on a single Gla protein and have provided detailed biochemical data describing the structure and functional relationship between protein and membrane. However, data comparisons that are made between studies from different laboratories have large discrepancies and thus such comparisons are difficult and often invalid. Factors implicated in these data discrepancies between laboratories including technique, protein origin, membrane type, phospholipid composition and other experimental conditions (Ca2+, pH and temperature). For example, with protein C, a wide range of Kj values to membranes have been reported: Table 7 Dissociation constants determined for Protein C techniaue membrane protein source KH reference light scattering SUV (% PS = ?) bovine 0.23 pM (135) light scattering SUV (50% PS) human 1.9 pM (136) light scattering SUV (20% PS) human 7.3 pM (88) SPR immobilized human 3.5 pM (88) membrane (20% PS) light scattering SUV (25% PS) human 1.5 pM (79) light scattering SUV (25% PS) bovine 17 pM (79) light scattering SUV (25% PS) bovine 15 uM (137) Similarly for prothrombin a IQj ranging from 0.11 to 1.1 pM were determined for 25% PS membranes of various types by different laboratories employing various techniques (see introduction for details). To address this discrepancy, I have completed a thorough comparison of binding affinity and kinetic analysis of human Gla proteins using the same membrane surface and same technique. In the current Chapter, phospholipid membrane binding properties were determined using SPR for all plasma derived human Gla proteins; zymogens, cofactors and activated forms with the exception of Factor VII over a wide concentration series using an identical membrane surface. Results from this study are comparable and are summarized in Table 6. To our surprise, there was great variation in all parameters investigated including membrane affinity and binding preference as well as kinetics controlling association and dissociation phases of the membrane interaction. Unexpectedly, none of the Gla proteins matched the binding profile observed previously with prothrombin. With the exception of Factor Xa and protein S, Gla protein - membrane binding was complex and could not be fit to a simple 1:1 model or models made available through BIAevaluation describing a single additional process of binding complexity (heterogeneity, conformers, multivalency - see Table 4). Thus, Gla-membrane binding is complex. A general trend seen throughout the Gla protein - membrane binding interaction is the involvement of apparent biphasic association and biphasic dissociation phases. This process was observed at all protein concentrations tested including physiological concentrations, with the exception of protein Z which showed no detectable binding at physiological concentration. The comparative results obtained by SPR indicate that the homologous Gla proteins involved in different enzymatic reactions of coagulation have diverse membrane binding properties. The concept of Gla interchangeability and similarity in membrane binding is often thought to be a conserved feature of Gla proteins (132). This is put into question with the current results indicating the vast differences in membrane binding properties between these proteins. It is conceivable that the differences in membrane binding observed here may actually facilitate the different roles and functions these protein have in controlling coagulation. 5.3.2 Zymogen versus activated enzyme One would expect the membrane binding properties of specific zymogen and activated enzyme pair (with the exception of prothrombin/thrombin) to be very similar if not identical, as they have an identical Gla sequence responsible for membrane contact. This is, however, clearly not the case. Striking membrane binding property differences are observed between zymogen and respective activated enzyme. This was most dramatic for Factor X and Factor Xa. Affinity increased by over 4-fold for the activated enzyme with simultaneous 2-fold increase in molar capacity for membrane. Similar differences were observed for Factor IX/IXa. Again the activated enzyme had a higher affinity (2-fold) and higher molar capacity for membrane (3-fold) relative to Factor IX. Finally, for the protein C/activated protein C pair, the activated enzyme form showed a 5-fold increase in membrane affinity with similar membrane binding capacity confirming a previous light scattering study using bovine molecules and SUV (135). Furthermore the mechanistic interaction also differed between Factor X and Factor Xa as was evident from the major shape differences of kinetic association and dissociation curves. This was not seen with two other zymogen-activated enzyme pairs studied (Factor IX/IXa and protein C/activated protein C). Clearly, zymogen activation results in membrane binding changes. Since the Gla sequence is unaltered, a sequence independent effect must be the cause in this binding difference. Conformational changes in the Gla domain have not been described for zymogen activation, but if present could help explain such binding differences. Alternatively, regions outside the Gla domain removed from the membrane contact site may be responsible for such binding differences. Possible explanations for this behaviour include the existence of membrane bound conformers and the presence or absence between zymogen and activated enzyme of a surface induced dimerization site. These putative membrane binding modulators deserve more attention and are explained in more detail in sections 5.3.3 and 5.3.4. Interestingly, of the Gla proteins examined, the membrane supported the highest molar binding capacity at saturation for the activated procoagulant proteins; Factor Xa and Factor IXa. The significance of this finding is unclear but implies that the activated procoagulant proteins bind negatively charged membranes favourably in an environment where all proteins are present, such as the competing plasma environment at sites of procoagulant membrane. It makes physiological sense to have the activated procoagulant proteins out compete zymogen proteins for binding sites at activated cell membranes. As enzymatic rates are controlled by the number of activated enzyme complexes present, not substrate concentrations, then it follows that it is more important to have enzymes bind these sites preferentially. Of course, this situation is complicated by the membrane affinity enhancement effect provided by their representative cofactors (Factor Va and Factor Villa, respectively). Similarly, activated protein C has a higher membrane affinity compared to the zymogen, protein C. However, activated protein C was determined to have the lowest molar binding capacity for membrane of all Gla proteins analyzed. Not surprisingly, it relies on a 100-fold enhancement of membrane affinity provided by its cofactor, protein S (135) to carry out its functions localized to membrane surfaces. 5.3.3 Membrane binding is influenced by regions outside of Gla One study has shown that membrane binding can be influenced by regions outside of the Gla domain. The thrombin sensitive region (TSR) located between the Gla domain and the first EGF domain is unique to protein S and was shown to be important to membrane binding (138). In this study, a TSR deleted protein S molecule had a membrane affinity reduced by 15-fold relative to intact protein S although the TSR peptide itself was shown not to bind membranes, indicating that regions removed from the membrane contact site influence membrane binding. Unfortunately, kinetic data for the membrane interaction was not determined for the two proteins. The study proposed a conformational stabilizing effect by TSR on the Gla domain allowing for an optimal interaction with phospholipids. Thus, the protein S molecule appears to have 2 membrane bound conformers depending on the presence or absence of TSR. It is conceivable then that upon zymogen activation (known to cause major Protease domain transitions) or initial membrane binding may result in conformational changes that effect the Gla - membrane interaction. 5.3.4 Vitamin K-dependent protein dimerization The effect of membrane binding of Gla proteins may be complicated by surface induced dimerization reactions. The biphasic kinetic data observed in the current study could be explained by such a reaction. Although no direct evidence exists for such a reaction, Gla protein dimerization on membranes surfaces should not be ignored but rather be considered as a potentially important event occurring on a membrane surface. A number of studies point to the existence of such membrane dimers. It is known that prothrombin dimerizes in solution, although only at high concentration (between 15 - 230 |iM as assessed by cross-linking experiments (19) and sedimentation measurements (18)) which indicates that an interactive site is present between two prothrombin molecules. This provides further support that this intermolecular protein interaction on the membrane is likely to occur at much lower concentrations and as I propose physiological concentrations. Accordingly, Gla protein collisional events facilitating dimer formation are much more probable to occur between membrane bound molecules as opposed to solution molecules, as diffusional and rotational dynamics are restricted on the membrane surface. Additional supporting evidence comes from Gla proteins described to dimerize only in the presence of a membrane component. First, cross-linking studies using prothrombin (17) and Factor VIlai (61) an active site-inhibited form of Factor Vila cross-linked through the active sites (61) at low protein concentrations were only possible in the presence of phospholipid vesicles and calcium. This suggests that the membrane surface facilitates the cross-linking reaction likely by a protein dimerization event prior to cross-linking. Second, cross-linked dimeric factor VIlai is a bivalent membrane binding molecule with two Gla domains. The membrane binding characteristics of this dimeric protein were shown to have an extremely slow off rate compared to monomeric factor Vila, consistent with the xl-prothrombin studied here in Chapter 4, which greatly enhanced the dimer - membrane binding strength. Thus, any transient dimer formation of a Gla protein on membrane would also result in an enhancement of membrane affinity by increasing the membrane resident time of the protein. Third, non-covalently associated Protein S dimers and multimers in solution have been described previously to have greatly increased membrane affinity compared to Protein S monomer (62). Our data are consistent with these finding as the protein S kinetic data was best described by a bivalent analyte model which assumes the presence of solution dimers prior to membrane binding. Fourth, an observation was made with Factor IX crystallized on negatively charged phospholipid layers under near physiological conditions. The Factor IX molecules organized themselves as dimers in the 2-D crystals (139), pointing to yet another example that these Gla proteins have a tendency to dimerize. Finally many heterodimeric Gla protein associations on membrane have been described. Protein S serves as a cofactor to activated protein C by increasing the affinity of activated protein C for membrane (135). Protein S binds to and inhibits Factor Xa (140). These results in light of the SPR data presented here indicate that membrane induced dimerization may be a common mechanism employed by the vitamin K-dependent Gla proteins which contributes to membrane binding. Chapter 6. Summary and General Discussion Blood coagulation depends on the effective assembly of a number of multicomponent enzyme and regulatory complexes held together in the presence of calcium ions by a plethora of intermolecular, reversible interactions including protein-protein and protein-phospholipid. Of these, the phosphatidylserine-containing membrane provided in vivo by a procoagulant surface such as activated platelets, provides both allosteric and concentration effects resulting in efficient protein interactions. Assembly and subsequent activity of these enzyme complexes depends on the binding properties of these individual peripheral membrane proteins. As such, a sensitive and quantitative assay describing membrane affinity and the kinetics governing association and dissociation is an essential tool for understanding the function and regulation of these proteins. Equilibrium differential centrifugation and surface plasmon resonance were employed to assess the membrane interaction of prothrombin and other Gla protein involved in blood coagulation. In Chapter 3, the effect Factor Va had on the affinity of prothrombin for membrane was investigated. Taking advantage of MLV that could be readily centrifuged allowed separation of bound and unbound peripheral protein. Combining this differential centrifugation technique with immunoaffinity quantification allowed the basis of some preliminary affinity data collection. It has to be stressed, however, that the experimental errors associated with this technique were large, preventing the accurate determination of binding constants. Nevertheless, prothrombin-membrane binding results obtained gave a reasonable estimate of the dissociation constant in the presence of Factor Va (-50 nM) when bound to membrane. Factor Va bound to anionic phospholipid membrane undoubtedly enhanced the membrane binding affinity of prothrombin relative to prothrombin binding in the absence of the cofactor. Finally, prethrombin 1, a minor contaminant in the prothrombin preparation, was localized to membrane fractions by Factor Va and showed no association to membrane in the absence of Factor Va. Thus, Factor Va can individually recruit prothrombin or prethrombin 1 to the membrane surface, plausibly contributing to its cofactor function. Surface plasmon resonance was proven to be a useful method for determining the detailed kinetics characterizing the membrane interaction of Gla proteins. As detailed in Chapters 4 and 5, the technique provided further insight into the nature of the Gla protein-phosphatidylserine membrane interactions that had been previously characterized with a number of other membrane binding assays. In Chapter 4, surface plasmon resonance was used to evaluate the Ca2+-specific binding of prothrombin and proteolytic fragments of prothrombin to immobilized membranes composed of mixtures of phosphatidylserine (0-25%) and phosphatidylcholine . Equilibrium binding measurements showed that the apparent dissociation constants were 2-3 fold lower for prothrombin compared to fragment 1 or fragment 1.2, and these dissociation constants all increased with decreasing molar fraction of phosphatidylserine. Equilibrium binding profiles for fragment 1 and fragment 1.2 fitted closely to a one-site binding model. Under the same conditions, however, prothrombin-membrane binding exhibited a biphasic association and dissociation process that deviated strongly from the simple one-site binding model. This binding profile was observed at all prothrombin concentrations tested (0.05 - 20 |xM) but was more pronounced at higher concentrations. Surface plasmon resonance data fitted best to a two step model characterized by an initial fast kinetic step and a secondary slow kinetic step. In this model, the binding of prothrombin consists of an obligate sequential mechanism involving a linked reaction because post-equilibrium dissociation profiles varied with association phase time. A cross-linked dimer of prothrombin formed a high affinity complex with membranes and had a similar off-rate compared to the secondary slow off-rate observed for prothrombin. When combined, these results most likely are described by a model in which prothrombin forms dimers on membranes. Dimerization may also explain the 2-3 fold increase in membrane binding affinity for prothrombin compared to fragment 1 or fragment 1.2. Concurrent with these observations, dimerization appears to be specific to the protease domain of prothrombin as neither fragment 1 nor fragment 1.2 displays such binding complexities. The physiological significance of the observed prothrombin dimerization on procoagulant membranes is unclear at present. It is possible that dimerization increases the half-life of membrane-bound prothrombin. This increased availability may increase the participation of prothrombin in the prothrombinase complex with subsequent thrombin production. The application of SPR to study membrane binding was further expanded to the other blood coagulation Gla proteins in Chapter 5. Membrane affinity (40-fold), molar binding preference (10-fold) and kinetics controlling complex formation and complex breakdown varied widely between Gla proteins. Apparent biphasic association and biphasic dissociation phases were commonly observed amongst the Gla proteins at a wide range of protein concentrations including physiological concentrations. The comparative results obtained by SPR indicate that the majority of homologous Gla proteins bind membranes with a complex mechanism which may involve membrane induced protein dimers (or multimers) and/or multiple membrane bound conformers. Gla proteins appear to have a propensity to dimerize (17-19, 62, 102, 135, 139, 140). Thus, an explanation involving an intermolecular interaction between two Gla proteins on the membrane surface after initial membrane complex formation appears to be the most likely explanation for this binding complexity. Analysis and interpretation of the underlying factors governing the complex membrane kinetic data obtained by SPR is often difficult if not impossible for a given peripheral protein, as was the case for many of the experiments carried out in this thesis. This is often described as major disadvantage of this technique which is currently still in its infancy. However, with further experimentation, these complexities should be overcome and thereby provide significant amount of information. A continuing objective in the field of membrane binding of blood coagulation Gla proteins as well as other peripheral protein membrane is to decipher the mechanistic details of the protein-membrane interaction and other effects the membrane plays on bound proteins. For such a problem to be solved, a very sensitive kinetic techniques such as SPR will be required. Chapter 7. Future Directions Several experimental approaches are possible with the developments from the experimental results in this thesis. Specific molecules or reagents that will allow further investigation of membrane binding properties of Gla proteins include the use of a physiological membrane, recombinant chimeric Gla proteins and prethrombin 1 as a Factor Va specific membrane binding probe and as an inhibitor of prothrombin membrane binding. For reasons of simplicity, Gla protein - membrane binding studies focusing on detailed kinetics have used artificial membranes composed of usually two and sometimes up to three types of phospholipids. Compared to the physiological platelet membrane setting, these artificial membranes clearly oversimplify the membrane composition of activated platelets which contain a number of phospholipid types as well as integral and tightly associated proteins, many of which have unknown function (141). Gla proteins have a tighter affinity to platelets compared to artificial liposome membranes. This may be caused by the presence of specific platelet receptor(s), reviewed in (56). Thus, a binding study involving activated platelet membranes seems imperative. Platelet derived microparticles are prime candidates for a SPR experiment as they are readily immobilized by a similar strategy used here for liposomes. These microparticles are small membrane vesicles released from activated platelets and provide a major catalytic surface for the assembly of blood coagulation enzyme complexes (142, 143). Thus a study using platelet microparticles may give further insight to the role physiological membranes have in Gla protein membrane binding. Vitamin K-dependent protein chimeras have been made for protein C containing the Gla domain of either Factor VII, Factor IX or prothrombin (132,144,145). These studies have indicated an interchangeability of the Gla domain for protein C without dramatically altering its membrane binding properties under study. For example comparison of phospholipid binding of prothrombin, protein C and the protein C containing the Gla domain of prothrombin chimera showed very similar binding for protein C and the chimera, whereas prothrombin was noticeably different (132). This was also reflected in the Kf p obtained in this study; prothrombin (700 nM), protein C (430 nM) and chimera (520 nM). Similarly, membrane affinity appears to be controlled by areas extrinsic to the Gla domain, as protein C and the protein C containing the Gla domain of Factor IX chimera showed very similar binding affinities that differed significantly from Factor IX. Unfortunately, only equilibrium binding data obtained by right angle light scattering were used to characterize membrane binding in these studies, as a detailed kinetic binding analysis was not performed. This emphasizes that the unique binding characteristics observed for individual Gla proteins by SPR (i.e. shape of kinetic profiles and affinity) are perhaps due to unique sites removed from the membrane contact site. Thus an intriguing possibility exists that the characteristic SPR kinetic signature curves encompassing various complex mechanisms observed for each Gla protein is a result of protein regions outside the Gla domain. For example, a chimeric study involving for instance prothrombin containing the Gla domain of various other Gla proteins would be predicted to have membrane binding properties very similar to native/wild type prothrombin. Similarly, it should be feasible to maintain the simpler kinetic profile seen with Factor Xa by swapping its Gla domain with another Gla domain. Of course there are some properties that appear to be intrinsic to the individual Gla domain, as was shown with the slighdy altered phospholipid specificities of protein C and prothrombin (132)\\ however, these differences are expected to be minor compared to the role the remainder of the protein plays to overall membrane binding. Unlike the interaction between Factor Va and Factor Xa, the interaction between Factor Va and prothrombin on membrane is not well understood. In Chapter 3, prethrombin 1 was shown to bind specifically to Factor Va-bound vesicles, as the Gla-less protein was unable to bind vesicles in the absence of the cofactor. It is apparent that when bound to vesicles prothrombin binds to Factor Va tighter (~ 50 nM, this study) then Factor Va in solution (1.9 (iM, (20) and 1.3 (iM, (23)). The use of prethrombin 1 as a specific probe for Factor Va bound to membranes may have several applications. First prethrombin 1 could be used to determine the specific site on the cofactor mediating the interaction by using a number of Factor Va derived peptides to inhibit binding. Furthermore, showing that prethrombin 1 has a higher affinity to Factor Va when bound to vesicles compared to in solution would provide evidence that the membrane allosterically controls the configuration of the Factor Va binding site. Finally, prethrombin 1 was shown in Chapter 4 to clearly inhibit prothrombin membrane binding. The exact mechanism of this inhibition remains unclear and also deserves further study. One possible explanation is that prethrombin 1 interacts with prothrombin in solution, thereby changing the conformation of the membrane contact site on prothrombin. This prethrombin 1 - bound conformation of prothrombin may result in a slower, but tighter binder which may explain the SPR data observed in Chapter 4. It is thus intriguing if one could potentially crosslink prothrombin and prethrombin 1 in solution. If so, a crosslinked prothrombin - prethrombin 1 heterodimer may display simple 1:1 kinetics. Bibliography (1) Nemerson, Y. (1992) The tissue factor pathway of blood coagulation. Semin Hematol. 29, 170-6. (2) Bach, R., Gentry, R., and Nemerson, Y. (1986) Factor VII binding to tissue factor in reconstituted phospholipid vesicles: induction of cooperativity by phosphatidylserine. Biochemistry 25, 4007-4020. (3) Ahmad, S. S„ Rawala-Sheikh, R„ and Walsh, P. N. (1992) Components and assembly of the factor X activating complex. Semin Thromb Hemost. 18, 311-23. (4) Krishnaswamy, S. (1990) Prothrombinase complex assembly. Contributions of protein-protein and protein-membrane interactions toward complex formation. J Biol Chem. 265, 3708-18. (5) 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., and Esmon, N. L. (1997) The protein C pathway: new insights. Thromb Haemost. 78, 70-4. (6) Degen, S. J., and Sun, W. Y. (1998) The biology of prothrombin. Crit Rev Eukaryot Gene Expr. 8, 203-24. (7) Hijikata-Okunomiya, A. (1990) A new method for the determination of prothrombin in human plasma. Thromb Res. 57, 705-15. (8) Zytkovicz, T. H., and Nelsestuen, G. L. (1975) [3H]diborane reduction of vitamin K-dependent calcium-binding proteins. Identification of a unique amino acid. J Biol Chem. 250, 2968-72. (9) Mizuochi, T., Fujii, J., Kisiel, W., and Kobata, A. (1981) Studies on the structures of the carbohydrate moiety of human prothrombin. J Biochem (Tokyo). 90,1023-31. (10) Petrovan, R. J., Govers-Riemslag, J. W., Nowak, G., Hemker, H. C., Tans, G., and Rosing, J. (1998) Autocatalytic peptide bond cleavages in prothrombin and meizothrombin. Biochemistry 37, 1185-91. (11) Nesheim, M. E„ Taswell, J. B., and Mann, K. G. (1979) The contribution of bovine Factor V and Factor Va to the activity of prothrombinase. J Biol Chem. 254, 10952-62. (12) Krishnaswamy, S. (2005) Exosite-driven substrate specificity and function in coagulation. J Thromb Haemost. 3, 54-67. (13) Krishnaswamy, S., Mann, K. G., and Nesheim, M. E. (1986) The prothrombinase-catalyzed activation of prothrombin proceeds through the intermediate meizothrombin in an ordered, sequential reaction. J Biol Chem. 261, 8977-84. (14) Cote, H. C., Bajzar, L., Stevens, W. K, Samis, J. A., Morser, J., MacGillivray, R. T., and Nesheim, M. E. (1997) Functional characterization of recombinant human meizothrombin and Meizothrombin(desFl). Thrombomodulin-dependent activation of protein C and thrombin-activatable fibrinolysis inhibitor (TAFI), platelet aggregation, antithrombin-III inhibition. J Biol Chem. 272, 6194-200. (15) Cote, H. C., Stevens, W. K„ Bajzar, L., Banfield, D. K, Nesheim, M. E., and MacGillivray, R. T. (1994) Characterization of a stable form of human meizothrombin derived from recombinant prothrombin (R155A, R271A, and R284A). J Biol Chem. 269, 11374-80. (16) Walsh, P. N., and Ahmad, S. S. (2002) Proteases in blood clotting. Essays Biochem. 38. (17) Anderson, P. J. (1998) A dimeric form of prothrombin on membrane surfaces. Biochem J. 336, 631-8. (18) Jackson, C. M„ Peng, C. W„ Brenckle, G. M„ Jonas, A., and Stenflo, J. (1979) Multiple modes of association in bovine prothrombin and its proteolysis products. J Biol Chem. 254, 5020-6. (19) Tarvers, R. C„ Noyes, C. M„ Tarvers, J. K„ and Lundblad, R. L. (1986) Mechanism of the calcium-dependent self-association of bovine prothrombin. Use of a covalent cross-linking reagent to study the reaction. J Biol Chem. 261,4855-9. (20) Deguchi, H., Takeya, H„ Gabazza, E. C„ Nishioka, J., and Suzuki, K. (1997) Prothrombin kringle 1 domain interacts with factor Va during the assembly of prothrombinase complex. Biochem J. 321, 729-35. (21) Kotkow, K. J., Deitcher, S. R„ Furie, B„ and Furie, B. C. (1995) The second kringle domain of prothrombin promotes factor Va-mediated prothrombin activation by prothrombinase. J Biol Chem. 270, 4554-7. (22) Kim, B. J., Koo, S. Y„ and Kim, S. S. (2002) A peptide derived from human prothrombin fragment 2 inhibits prothrombinase and angiogenesis. Thromb Res. 106, 81-7. (23) Blostein, M. D„ Rigby, A. C., Jacobs, M„ Furie, B., and Furie, B. C. (2000) The Gla domain of human prothrombin has a binding site for factor Va. J Biol Chem. 275, 38120-6. (24) Anderson, P. J., Nesset, A., Dharmawardana, K. R„ and Bock, P. E. (2000) Role of proexosite I in factor Va-dependent substrate interactions of prothrombin activation. J Biol Chem. 275,16435-42. (25) Chen, L., Yang, L., and Rezaie, A. R. (2003) Proexosite-1 on prothrombin is a factor Va-dependent recognition site for the prothrombinase complex. J Biol Chem. 278, 27564-9. (26) Yegneswaran, S„ Mesters, R. M„ Fernandez, J. A., and Griffin, J. H. (2004) Prothrombin residues 473-487 contribute to factor Va binding in the prothrombinase complex. J Biol Chem. 279, 49019-25. (27) Stenflo, J. (1999) Contributions of Gla and EGF-like domains to the function of vitamin K-dependent coagulation factors. Crit Rev Eukaryot Gene Expr. 9, 59-88 review. (28) Nelsestuen, G. L„ Shah, A. M., and Harvey, S. B. (2000) Vitamin K-dependent proteins. Vitam Horm. 58, 355-89. (29) Chen, Q., and Lentz, B. R. (1997) Fluorescence resonance energy transfer study of shape changes in membrane-bound bovine prothrombin and meizothrombin. Biochemistry 36, 4701-11. (30) Bode, W. (2005) The structure of thrombin, a chameleon-like proteinase. Thromb Haemost. 3, 2379-88. (31) Broze, G. J. J. (2001) Protein Z-dependent regulation of coagulation. Thromb Haemost. 86, 8-13. (32) Broze, G. J. J., and Miletich, J. P. (1984) Human Protein Z. J Clin Invest. 73, 933-8. (33) Edenbrandt, C. M„ Lundwall, A., Wydro, R., and Stenflo, J. (1990) Molecular analysis of the gene for vitamin K dependent protein S and its pseudogene. Cloning and partial gene organization. Biochemistry 29, 7861-8. (34) Rigby, A. C., and Grant, M. A. (2004) Protein S: a conduit between anticoagulation and inflammation. Crit Care Med. 32, S336-41. (35) Munoz, X., Sumoy, L., Ramirez-Lorca, R., Villar, J., de Frutos, P. G., and Sala, N. (2004) Human vitamin K-dependent GAS6: gene structure, allelic variation, and association with stroke. Hum Mutat. 23, 506-12. (36) Price, P. A., Poser, J. W., and Raman, N. (1976) Primary structure of the gamma-carboxyglutamic acid-containing protein from bovine bone. Proc Natl Acad Sci U SA 73, 3374-5. (37) Kulman, J. D., Harris, J. E., Xie, L., and Davie, E. W. (2001) Identification of two novel transmembrane gamma-carboxyglutamic acid proteins expressed broadly in fetal and adult tissues. Proc Natl Acad Sci U S A 98, 1370-5. (38) Khazi, F. R., Chu, K. C., and High, K. A. in The American Society of Hematology 48th Annual Meeting and Exposition December 9-12, 2006, Orlando, Florida. (39) Kulman, J. D., Harris, J. E., Nakazawa, N., Ogasawara, M., Satake, M., and Davie, E. W. (2006) Vitamin K-dependent proteins in Ciona intestinalis, a basal chordate lacking a blood coagulation cascade. Proc Natl Acad Sci USA 103, 15794-9. (40) Ahmad, S. S„ London, F. S„ and Walsh, P. N. (2003) The assembly of the factor X-activating complex on activated human platelets. J Thromb Haemost. 1, 48-59. (41) McGee, M. P., Li, L. C., and Hensler, M. (1992) Functional assembly of intrinsic coagulation proteases on monocytes and platelets. Comparison between cofactor activities induced by thrombin and factor Xa. J Exp Med. 176, 27-35. (42) Falls, L. A., Furie, B., and Furie, B. C. (2000) Role of phosphatidylethanolamine in assembly and function of the factor IXa-factor Villa complex on membrane surfaces. Biochemistry 39, 13216-22. (43) Zwaal, R. F., and Schroit, A. J. (1997) Pathophysiologic implications of membrane phospholipid asymmetry in blood cells. Blood 89, 1121-32. (44) Govers-Riemslag, J. W., Janssen, M. P., Zwaal, R. F., and Rosing, J. (1994) Prothrombin activation on dioleoylphosphatidylcholine membranes. Eur J Biochem. 220, 131-8. (45) Lentz, B. R. (2003) Exposure of platelet membrane phosphatidylserine regulates blood coagulation. Prog Lipid Res. 42, 423-38. (46) Mann, K. G., Nesheim, M. E., Church, W. R., Haley, P., and Krishnaswamy, S. (1990) Surface-dependent reactions of the vitamin K-dependent enzyme complexes. Blood 76, 1-16 review. (47) Furie, B., and Furie, B. C. (1988) The molecular basis of blood coagulation. Cell 53, 505-18. (48) Husten, E. J., Esmon, C. T., and Johnson, A. E. (1987) The active site of blood coagulation factor Xa: its distance from the phospholipid surface and its conformational sensitivity to components of the prothrombinase complex. J Biol Chem. 262, 12953-12961. (49) Armstrong, S. A., Husten, E. J., Esmon, C. T„ and Johnson, A. E. (1990) The active site of membrane-bound meizothrombin. A fluorescence determination of its distance from the phospholipid surface and its conformational sensitivity to calcium and factor Va. J Biol Chem. 265, 6210-8. (50) Cutsforth, G. A., Whitaker, R. N„ Hermans, J., and Lentz, B. R. (1989) A new model to describe extrinsic protein binding to phospholipid membranes of varying composition: application to human coagulation proteins. Biochemistry 28, 7453-61. (51) Nelsestuen, G. L., and Broderius, M. (1977) Interaction of prothrombin and blood-clotting factor X with membranes of varying composition. Biochemistry 16, 4172-7. (52) Bangham, A. D. (1961) A correlation between surface charge and coagulant action of phospholipids. Nature 192,1197-8. (53) Comfurius P, S. E„ Willems GM, Bevers EM, Zwaal RF. (1994) Assembly of the prothrombinase complex on lipid vesicles depends on the stereochemical configuration of the polar headgroup of phosphatidylserine. Biochemistry 33, 10319-24. (54) Majumder, R„ Weinreb, G., Zhai, X., and Lentz, B. R. (2002) Soluble phosphatidylserine triggers assembly in solution of a prothrombin-activating complex in the absence of a membrane surface. J Biol Chem. 277, 29765-73. (55) Majumder, R., Weinreb, G., and Lentz, B. R. (2005) Efficient thrombin generation requires molecular phosphatidylserine, not a membrane surface. Biochemistry 44, 16998-7006.. (56) Walsh, P. N. (2004) Platelet coagulation-protein interactions. Semin Thromb Hemost. 30, 461-71. (57) Bevers, E. M., Comfurius, P., and Zwaal, R. F. (1983) Changes in membrane phospholipid distribution during platelet activation. Biochim Biophys Acta. 736, 57-66. (58) Grant, M. A., Baikeev, R. F., Gilbert, G. E„ and Rigby, A. C. (2004) Lysine 5 and phenylalanine 9 of the factor IX omega-loop interact with phosphatidylserine in a membrane-mimetic environment. Biochemistry 43, 15367-78. (59) McGee, M. P., Teuschler, H., and Liang, J. (1998) Effective electrostatic charge of coagulation factor X in solution and on phospholipid membranes: implications for activation mechanisms and structure-function relationships of the Gla domain. Biochem J. 330, 533-9. (60) Falls, L. A., Furie, B. C., Jacobs, M., Furie, B„ and Rigby, A. C. (2001) The omega-loop region of the human prothrombin gamma-carboxyglutamic acid domain penetrates anionic phospholipid membranes. J Biol Chem. 276, 23895-902. (61) Stone, M. D., Harvey, S. B., Martinez, M. B., Bach, R. R., and Nelsestuen, G. L. (2005) Large enhancement of functional activity of active site-inhibited factor Vila due to protein dimerization: insights into mechanism of assembly/disassembly from tissue factor. Biochemistry 44, 6321-30. (62) Sere, K. M., Janssen, M. P., Willems, G. M., Tans, G., Rosing, J., and Hackeng, T. M. (2001) Purified protein S contains multimeric forms with increased APC-independent anticoagulant activity. Biochemistry 40, 8852-60. (63) Dai, Q., Prorok, M., and Castellino, F. J. (2005) Role of the hexapeptide disulfide loop in the gamma-carboxyglutamic acid domain of protein C in Ca2+-mediated structural and functional properties. Biochemistry 37, 12508-14. (64) Lu, Y., and Nelsestuen, G. L. (1996) Dynamic features of prothrombin interaction with phospholipid vesicles of different size and composition: implications for protein—membrane contact. Biochemistry 35, 8193-200. (65) Abbott, A. J., and Nelsestuen, G. L. (1987) Association of a protein with membrane vesicles at the collisional limit: studies with blood coagulation factor Va light chain also suggest major differences between small and large unilamellar vesicles. Biochemistry 26, 7994-8003. (66) Lim, T. K„ Bloomfield, V. A., and Nelsestuen, G. L. (1977) Structure of the prothrombin- and blood clotting factor X-membrane complexes. Biochemistry 16, 4177-81. (67) Kotkow, K. J., Furie, B., and Furie, B. C. (1993) The interaction of prothrombin with phospholipid membranes is independent of either kringle domain. J Biol Chem. 268, 15633-9. (68) Ratcliffe, J. V., Furie, B., and Furie, B. C. (1993) The importance of specific gamma-carboxyglutamic acid residues in prothrombin. Evaluation by site-specific mutagenesis. J Biol Chem. 268, 24339-45. (69) Pollock, J. S„ Shepard, A. J., Weber, D. J., Olson, D. L„ Klapper, D. G„ Pedersen, L. G., and Hiskey, R. G. (1988) Phospholipid binding properties of bovine prothrombin peptide residues 1-45. J Biol Chem. 263, 14216-23. (70) Soriano-Garcia, M., Padmanabhan, K., de Vos, A. M., and Tulinsky, A. (1992) The Ca2+ ion and membrane binding structure of the Gla domain of Ca-prothrombin fragment 1. Biochemistry 31, 2554-66. (71) Sunnerhagen, M., Forsen, S., Hoffren, A. M., Drakenberg, T., Teleman, O., and Stenflo, J. (1995) Structure of the Ca(2+)-free Gla domain sheds light on membrane binding of blood coagulation proteins. Nat Struct Biol. 2, 504-9. (72) Christiansen, W. T., Tulinsky, A., and Castellino, F. J. (1994) Functions of individual gamma-carboxyglutamic acid (Gla) residues of human protein c. Determination of functionally nonessential Gla residues and correlations with their mode of binding to calcium. Biochemistry 33, 14993-5000. (73) Freedman, S. J., Furie, B. C., Furie, B„ and Baleja, J. D. (1995) Structure of the calcium ion-bound gamma-carboxyglutamic acid-rich domain of factor DC. Biochemistry 34, 12126-37. (74) Weber, D. J., Berkowitz, P., Panek, M. G„ Huh, N. W„ Pedersen, L. G„ and Hiskey, R. G. (1992) Modifications of bovine prothrombin fragment 1 in the presence and absence of Ca(II) ions. Loss of positive cooperativity in Ca(II) ion binding for the modified proteins. J Biol Chem. 267, 4564-9. (75) Huang, M., Rigby, A., Morelli, X., Grant, M., Huang, G., Furie, B., Seaton, B„ and Furie, B. (2003) Structural basis of membrane binding by Gla domains of vitamin K-dependent proteins. Nat Struct Biol. 10, 751-6. (76) Christiansen, W. T., Jalbert, L. R., Robertson, R. M., Jhingan, A., Prorok, M., and Castellino, F. J. (1995) Hydrophobic amino acid residues of human anticoagulation protein C that contribute to its functional binding to phospholipid Vesicles. Biochemistry 34, 10376-82. (77) Resnick, R. M., and Nelsestuen, G. L. (1980) Prothrombin-membrane interaction. Effects of ionic strength, pH, and temperature. Biochemistry 19, 3028-33. (78) McDonald, J. F., Evans, T. C. J., Emeagwali, D. B., Hariharan, M., Allewell, N. M., Pusey, M. L., Shah, A. M., and Nelsestuen, G. L. (1997) Ionic properties of membrane association by vitamin K-dependent proteins: the case for univalency. Biochemistry 36, 15589-98. (79) McDonald, J. F„ Shah, A. M„ Schwalbe, R. A., Kisiel, W„ Dahlback, B., and Nelsestuen, G. L. (1997) Comparison of naturally occurring vitamin K-dependent proteins: correlation of amino acid sequences and membrane binding properties suggests a membrane contact site. Biochemistry 36, 5120-7. (80) Majumder, R„ Wang, J., and Lentz, B. R. (2003) Effects of water soluble phosphotidylserine on bovine factor Xa: functional and structural changes plus dimerization. Biophys J. 84, 1238-51. (81) Evans, T. C. J., and Nelsestuen, G. L. (1994) Calcium and membrane-binding properties of monomeric and multimeric annexin II. Biochemistry 33, 13231-8. (82) Wei, G. J., Bloomfield, V. A., Resnick, R. M„ and Nelsestuen, G. L. (1982) Kinetic and mechanistic analysis of prothrombin-membrane binding by stopped-flow light scattering. Biochemistry 21, 1949-59. (83) Kop, J. M„ Cuypers, P. A., Lindhout, T„ Hemker, H. C., and Hermens, W. T. (1984) The adsorption of prothrombin to phospholipid monolayers quantitated by ellipsometry. J Biol Chem. 259,13993-8. (84) Pearce, K. H., Hiskey, R. G„ and Thompson, N. L. (1992) Surface binding kinetics of prothrombin fragment 1 on planar membranes measured by total internal reflection fluorescence microscopy. Biochemistry 31, 5983-95. (85) Pearce, K. H., Hof, M„ Lentz, B. R., and Thompson, N. L. (1993) Comparison of the membrane binding kinetics of bovine prothrombin and its fragment 1. J Biol Chem. 268, 22984-91. (86) Ellison, E. H., and Castellino, F. J. (1998) Adsorption of vitamin K-dependent blood coagulation proteins to spread phospholipid monolayers as determined from combined measurements of the surface pressure and surface protein concentration. Biochemistry 37, 7997-8003. (87) Richter, R. P., Maury, N., and Brisson, A. R. (2005) On the effect of the solid support on the interleaflet distribution of lipids in supported lipid bilayers. Langmuir 21, 299-304. (88) Sun, Y. H., Shen, L., and Dahlback, B. (2003) Gla domain-mutated human protein C exhibiting enhanced anticoagulant activity and increased phospholipid binding. Blood 101, 2277-84. (89) Shen, L„ Shah, A. M., Dahlback, B„ and Nelsestuen, G. L. (1998) Enhancement of human protein C function by site-directed mutagenesis of the gamma-carboxyglutamic acid domain. J Biol Chem. 273, 31086-91. (90) Evans, T. C. J., and Nelsestuen, G. L. (1996) Importance of cis-proline 22 in the membrane-binding conformation of bovine prothrombin. Biochemistry 35, 8210-5. (91) Perera, L., Darden, T. A., and Pedersen, L. G. (1998) Trans-Cis Isomerization of Proline 22 in Bovine Prothrombin Fragment 1: A Surprising Result of Structural Characterization. Biochemistry 37, 10920-7. (92) Lentz, B. R„ Zhou, C. M„ and Wu, J. R. (1994) Phosphatidylserine-containing membranes alter the thermal stability of prothrombin's catalytic domain: a differential scanning calorimetric study. Biochemistry 33, 5460-8. (93) Tendian, S. W„ Lentz, B. R„ and Thompson, N. L. (1991) Evidence from total internal reflection fluorescence microscopy for calcium-independent binding of prothrombin to negatively charged planar phospholipid membranes. Biochemistry 30, 10991-9. (94) Cooper, M. A., Hansson, A., Lofas, S., and Williams, D. H. (2000) A vesicle capture sensor chip for kinetic analysis of interactions with membrane-bound receptors. Anal Biochem. 277, 196-205. (95) Schuck, P. (1997) Use of surface plasmon resonance to probe the equilibrium and dynamic aspects of interactions between biological macromolecules. Anna Rev Biophys Biomol Struct. 26, 541-66. (96) Hall, D. (2001) Use of optical biosensors for the study of mechanistically concerted surface adsorption processes. Anal. Biochem. 288,109-25. (97) Stenberg, E., Persson, B., Roos, H., and Urbaniczky, C. (1991) Quantitative Determination of surface concentration of protein with surface plasmon resonance using radiolabeled proteins. J Colloid Interface Sci. 143, 513-26. (98) Gizeli, E., and Lowe, C. R. (2002) In Biomolecular Sensors. Taylor and Francis Publishing, New York. (99) Mayer, L. D., Hope, M. J., and Cullis, P. R. (1986) Vesicles of variable sizes produced by a rapid extrusion procedure. Biochim Biophys Acta 858, 161-8. (100) Chen, P. S„ JR. Toribara, T. Y., and Warner, H. (1956) Microdetermination of Phosphorus. Anal. Chem. 28,1756-8. (101) Nelsestuen, G. L., and Lim, T. K. (1977) Equilibria involved in prothrombin- and blood-clotting factor X-membrane binding. Biochemistry 16, 4164-71. (102) Stone, M. D., and Nelsestuen, G. L. (2005) Efficacy of soluble phospholipids in the prothrombinase reaction. Biochemistry 44,4037-41. (103) Luckow, E. A., Lyons, D. A., Ridgeway, T. M., Esmon, C. T., and Laue, T. M. (1989) Interaction of clotting factor V heavy chain with prothrombin and prethrombin 1 and role of activated protein C in regulating this interaction: analysis by analytical ultracentrifugation. Biochemistry 28, 2348-54. (104) Mann, K. G. (1976) Prothrombin. Methods Enzymology 45, 123-56. (105) Tracy PB, E. L., Bowie EJ, Mann KG.. (1982) Radioimmunoassay of factor V in human plasma and platelets. Blood 60, 59-63. (106) Billy, D„ Willems, G. M„ Hemker, H. C., and Lindhout, T. (1995) Prothrombin contributes to the assembly of the factor Va-factor Xa complex at phosphatidylserine-containing phospholipid membranes. J Biol Chem. 270, 26883-9. (107) Beck, D. O., Bukys, M. A., Singh, L. S„ Szabo, K. A., and Kalafatis, M. (2004) The contribution of amino acid region ASP695-TYR698 of factor V to procofactor activation and factor Va function. J Biol Chem. 279, 3084-95. (108) Sorensen, K. W., Nicolaes, G. A., Villoutreix, B. 0., Yamazaki, T„ Tans, G., Rosing, J., and Dahlback, B. (2004) Functional properties of recombinant factor V mutated in a potential calcium-binding site. Biochemistry 43, 5803-10. (109) Levesque, J. P., Hatzfeld, A., and Hatzfeld, J. (1985) A method to measure receptor binding of ligands with low affinity. Application to plasma proteins binding assay with hemopoietic cells. Exp. Cell Res. 156, 558-562. (110) Krishnaswamy, S., and Mann, K. G. (1988) The binding of factor Va to phospholipid vesicles. J Biol Chem. 263, 5714-23. (111) Degen, S. J., and Davie, E. W. (1987) Nucleotide sequence of the gene for human prothrombin. Biochemistry 26, 6165-77. (112) Lentz, B. R. (1995) Are acidic lipid domains induced by extrinsic protein binding to membranes? Mol Membr Biol. 12, 65-7. (113) Rich, R. L„ and Myszka, D. G. (2005) Survey of the year 2004 commercial optical biosensor literature. J Mol Recognit. 18, 431-478. (114) Erb, E. M„ Stenflo, J., and Drakenberg, T. (2002) Interaction of bovine coagulation factor X and its glutamic-acid-containing fragments with phospholipid membranes. A surface plasmon resonance study. Eur J Biochem. 269, 3041-6. (115) Saenko, E., Sarafanov, A., Ananyeva, N., Behre, E., Shima, M., Schwinn, H., and Josic, D. (2001) Comparison of the properties of phospholipid surfaces formed on HPA and LI biosensor chips for the binding of the coagulation factor VIII. J ChromatogrA. 921, 49-56. (116) Saenko, E., Sarafanov, A., Greco, N., Shima, M., Loster, K., Schwinn, H., and Josic, D. (1999) Use of surface plasmon resonance for studies of protein-protein and protein-phospholipid membrane interactions. Application to the binding of factor VIII to von Willebrand factor and to phosphatidylserine-containing membranes. J Chromatogr A. 852, 59-71. (117) Neuenschwander, P. F. (2004) Exosite occupation by heparin enhances the reactivity of blood coagulation factor IXa. Biochemistry 43, 2978-86. (118) Osterlund, M., Persson, E., Carlsson, U., Freskgard, P. O., and Svensson, M. (2005) Sequential coagulation factor Vila domain binding to tissue factor. Biochem Biophys Res Commun. 337, 1276-82. (119) Erb, E. M„ Chen, X., Allen, S., Roberts, C. J., Tendler, S. J., Davies, M. C„ and Forsen, S. (2000) Characterization of the surfaces generated by liposome binding to the modified dextran matrix of a surface plasmon resonance sensor chip. Anal Biochem. 280, 29-35. (120) Dimitrova, M. N., Matsumura, H„ Dimitrova, A., and Neitchev, V. Z. (2000) Interaction of albumins from different species with phospholipid liposomes. Multiple binding sites system. Int J Biol Macromol. 27, 187-94. (121) Cirelli, N., Lebrun, P., Gueuning, C., Delogne-Desnoeck, J., Vanbellinghen, A. M., Graff, G., and Meuris, S. (2001) Physiological concentrations of albumin stimulate chorionic gonadotrophin and placental lactogen release from human term placental explants. Hum Reprod. 16, 441-8. (122) Cooper, M. A., Try, A. C„ Carroll, J., Ellar, D. J., and Williams, D. H. (1998) Surface plasmon resonance analysis at a supported lipid monolayer. Biochim Biophys Acta 1373,101-11. (123) Willems, G. M., Janssen, M. P., Comfurius, P., Galli, M„ Zwaal, R. F„ and Bevers, E. M. (2000) Competition of annexin V and anticardiolipin antibodies for binding to phosphatidylserine containing membranes. Biochemistry 39, 1982-9. (124) Wu, J. R., and Lentz, B. R. (1991) Fourier transform infrared spectroscopic study of Ca2+ and membrane-induced secondary structural changes in bovine prothrombin and prothrombin fragment 1. Biophys J. 60, 70-80. (125) Malholtra, O. P. (1991) Isolation and some properties of 11- and 9-Gla prothrombins from normal plasma. Biochem Cell Biol. 69, 404-8. (126) Lentz, B. R„ Wu, J. R„ Sorrentino, A. M„ and Carleton, J: N. (1991) Membrane binding induces lipid-specific changes in the denaturation profile of bovine prothrombin. A scanning calorimetry study. Biophys J. 60, 942-51. (127) Mayer, L., Nelsestuen, G. L., and Brockman, H. L. (1983) Prothrombin association with phospholipid monolayers. Biochemistry 22, 316-21. (128) Chenna, R., Sugawara, H., Koike, T., Lopez, R., Gibson, T. J., Higgins, D. G., and Thompson, J. D. (2003) Multiple sequence alignment with the Clustal series of programs Nucleic Acids Res 31, 3497-500. (129) Mizuno, H., Fujimoto, Z., Atoda, H., and Morita, T. (2001) Crystal structure of an anticoagulant protein in complex with the Gla domain of factor X. Proc Natl Acad Sci U SA98, 7230-4. (130) Jesty, J., Spencer, A. K., and Y., N. (1974) The Mechanism of Activation of Factor X. J Biol Chem. 249, 5614-5622. (131) Rosenberg, J. S., McKenna, P. W„ and Rosenberg, R. D. (1975) Inhibition of human factor IXa by human antithrombin. J Biol Chem. 250, 8883-8. (132) Smirnov, M. D., Safa, O., Regan, L„ Mather, T., Stearns-Kurosawa, D. J., Kurosawa, S., Rezaie, A. R., Esmon, N. L., and Esmon, C. T. (1998) A chimeric protein C containing the prothrombin Gla domain exhibits increased anticoagulant activity and altered phospholipid specificity. J Biol Chem. 273, 9031-40. (133) Castellino, F. J. (1995) Human protein C and activated protein C : Components of the human anticoagulation system Trends in Cardiovascular Medicine 5, 55-62. (134) Shah, A. M„ Kisiel, W., Foster, D. C„ and Nelsestuen, G. L. (1998) Manipulation of the membrane binding site of vitamin K-dependent proteins: enhanced biological function of human factor VII. Proc Natl Acad Sci U S A 95,4229-34. (135) Walker, F. J. (1981) Regulation of activated protein C by protein S. The role of phospholipid in factor Va inactivation. J Biol Chem. 256,11128-31. (136) Colpitts, T. L., and Castellino, F. J. (1994) Calcium and phospholipid binding properties of synthetic gamma-carboxyglutamic acid-containing peptides with sequence counterparts in human protein C. Biochemistry 33, 3501-8. (137) Nelsestuen, G. L., Kisiel, W., and Di Scipio, R. G. (1978) Interaction of vitamin K dependent proteins with membranes. Biochemistry 17, 2134-8. (138) Borgel, D., Gaussem, P., Garbay, C., Bachelot-Loza, C., Kaabache, T., Liu, W. Q„ Brohard-Bohn, B„ Le Bonniec, B„ Aiach, M„ and Gandrille, S. (2001) Implication of protein S thrombin-sensitive region with membrane binding via conformational changes in the gamma-carboxyglutamic acid-rich domain. Biochem J. 360,499-506. (139) Stoylova, S., Gray, E., Barrowcliffe, T. W., Kemball-Cook, G., and Holzenburg, A. (1998) Structural determination of lipid-bound human blood coagulation factor IX. Biochim Biophys Acta. 1382, 175-8. (140) Heeb, M. J., Rosing, J., Bakker, H. M., Fernandez, J. A., Tans, G., and Griffin, J. H. (1994) Protein S binds to and inhibits factor Xa. Proc Natl Acad Sci USA 91, 2728-32. (141) Garcia, B. A., Smalley, D. M„ Cho, H„ Shabanowitz, J., Ley, K„ and Hunt, D. F. (2005) The platelet microparticle proteome. Journal ofProteome Research 4, 1516-21. (142) Sims, P. J., Faioni, E. M„ Wiedmer, T„ and Shattil, S. J. (1988) Complement proteins C5b-9 cause release of membrane vesicles from the platelet surface that are enriched in the membrane receptor for coagulation factor Va and express prothrombinase activity. J Biol Chem. 263, 18205-12. (143) Gilbert, G. E., Sims, P. J., Wiedmer, T„ Furie, B„ Furie, B. C., and Shattil, S. J. (1991) Platelet-derived microparticles express high affinity receptors for factor v m .J Biol Chem. 266, 17261-8. (144) Christiansen, W. T., and Castellino, F. J. (1994) Properties of recombinant chimeric human protein C and activated protein C containing the gamma-carboxyglutamic acid and trailing helical stack domains of protein C replaced by those of human coagulation factor IX. Biochemistry 33, 5901-11. (145) Geng, J. P., and Castellino, F. J. (1997) Properties of a recombinant chimeric protein in which the gamma-carboxyglutamic acid and helical stack domains of human anticoagulant protein C are replaced by those of human coagulation factor VII. Thromb Haemost. 77, 926-33. "@en ; edm:hasType "Thesis/Dissertation"@en ; edm:isShownAt "10.14288/1.0100368"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Biochemistry and Molecular Biology"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Membrane binding properties of prothrombin and other gamma-carboxyglutamic acid-containing coagulation proteins"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/31087"@en .