Factor V (FV) links procoagulant amplification to anticoagulant feedback, but how FV limits tissue factor-initiated coagulation are not fully defined. We hypothesized that procofactor FV downregulates factor X (FX) activation by tissue factor:factor VIIa (TF:FVIIa), independently of tissue factor pathway inhibitor α (TFPIα), and that this mechanism is especially important in hemophilia. Thrombin generation was measured in FV/FVIII‑immunodepleted plasma and synthetic plasma while titrating FV, with TFPIα removed, blocked, or re-added. TF:FVIIa activation of FX was measured on phosphatidylserine‑containing or phosphatidylserine‑free liposomes with antibodies against the FV light chain and C2 domain. FV and TFPIα levels modulated thrombin generation in plasma from people with hemophilia A. In TF‑initiated coagulation lacking TFPIα, thrombin generation peaked at 2 nM FV and decreased as FV increased; at 20 nM FV (normal concentration), peak thrombin and thrombin generation rate were reduced by up to 50-80%, with larger effects at low FVIII. In purified TF:FVIIa assays, FV reduced FX activation by ~80% at physiologic concentration and inhibited FX activation on TF-expressing fibroblasts. Increasing PS content enhanced FX activation and increased the FV-sensitive component, while blocking the FV light chain or C2 domain partially relieved inhibition. In hemophilia A plasma, higher FV was associated with longer lag time and time-to-peak, and lower peak thrombin independent of TFPIα. Thus, FV is an endogenous anticoagulant that inhibits TF-initiated coagulation by limiting FX activation by TF:FVIIa through a membrane-dependent mechanism. This mechanism refines models of coagulation initiation and may help explain how FV variation contributes to bleeding and thrombosis.

Factor VIIIa (FVIIIa) mimetic bispecific antibodies, emicizumab and mim8, mimic one of the key functions of FVIIIa in the intrinsic tenase complex by approximating the substrate factor X (FX) and protease factor IXa (FIXa). However, unlike FVIIIa, these mimetics do not bind to procoagulant membranes, which likely contributes to their significantly lower cofactor activity on a molar basis when compared to that of FVIIIa. Although single chain antibodies derived from the FIXa binding arm of emicizumab or mim8 retain some ability to enhance FIXa function, the rate of FX activation is greatly augmented in the bispecific format in which the antibodies engage both FIXa and FX. We have continued exploration of how the addition of membrane binding features to monospecific FIXa binding derivatives of emicizumab or mim8 could surpass the cofactor mimetic activities of the bispecific antibodies. Similar to one-arm antibodies derived from the FIXa binding arm of emicizumab that can only bind FIXa and impart a modest cofactor mimetic activity, our emicizumab FIXa arm derived single chain antibody fragment (scFv), EVH9VL, facilitated FIXa mediated FX activation at a rate several times lower than that of the bispecific emicizumab. The monospecific antibody fragment EVH9VL served as a foundation to introduce membrane binding properties by fusing it with the factor V C2-domain. A repeating Gly-Ser linker separated the antibody fragment from the C2-domain, allowing the variable regions to potentially orient correctly on the membrane surface for optimal engagement with FIXa. The resulting construct, EVH9VLC2 (GR8), demonstrated a substantial amplification in cofactor activity compared to EVH9VL. GR8 bound to synthetic phospholipid vesicles containing phosphatidylcholine (PC) and phosphatidylserine (PS) (PC:PS, 75:25 %) with an estimated affinity constant (Kd) of ~ 0.5 µM and strongly colocalized with FIXa at the membrane. In kinetic assays with purified proteins, GR8 showed approximately 22-fold faster FX activation as compared to EVH9VL. Remarkably, despite lacking FX binding ability, GR8 showed a cofactor mimetic activity about ten-fold higher than emicizumab in activated partial thromboplastin time (APTT) assays using FVIII-deficient hemophilia A (HA) plasma. This potent enhancement of FX activation by GR8 was also evident in thrombin generation assays (TGA). At 100 nM concentration, GR8 normalized thrombin generation in HA plasma supplemented with 4 μM PC:PS, activated by low tissue factor (0.1 pM) or factor XIa (0.1 nM). The C2 domain-mediated increase in cofactor activity of GR8 was effectively reversed by an antibody fragment (E9scFv) that impairs C2-domain binding to the membrane, providing a built-in reversal mechanism for GR8's cofactor activity if necessary. Further modifications, including changing the C-terminus 6-His tag to HPC4 or removing the tag entirely, increased cofactor mimetic activity by an additional 3-fold, suggesting that the histidine charge cluster at the C2-domain's C-terminus may hinder optimal membrane interaction. Substituting the FV C2-domain in GR8 with lactadherin's C2-domain maintained similar cofactor activity, indicating that membrane dependent enhancement of cofactor activity is not limited to FV C2-domain. Similarly, mim8 FIXa binding arm-derived scFv fused with the FV C2-domain also showed several-fold increased cofactor activity compared to its unmodified counterpart. These engineered FIXa binding variants were able to overcome FVIII inhibitors in HA plasma. To prolong half-life and introduce inherent latency until activation, human albumin was fused to GR8 at the N-terminus, linked with a FIX activation peptide. This fusion aimed to exploit cellular recycling via neonatal Fc receptor interaction, potentially maintaining a latent state until triggered by a procoagulant stimulus. The albumin-fused GR8 retained cofactor mimetic activity similar to GR8 and was cleaved by human FXIa in vitro. The engineered scFv with a factor V C2 domain fusion, demonstrates that simply adding a membrane binding feature to the FIXa arm of emicizumab surpasses the mimetic activity achieved by a bispecific antibody that exploits both bridging and allostery for effective cofactor function. This advancement underscores the potential of single-chain, monospecific, membrane-anchored FVIIIa mimetics to significantly enhance FIXa's catalytic activity.

Potent and selective inhibition of the structurally homologous proteases of coagulation poses challenges for drug development. Hematophagous organisms frequently accomplish this by fashioning peptide inhibitors combining exosite and active site binding motifs. Inspired by this biological strategy, we create several EXACT inhibitors targeting thrombin and factor Xa de novo by linking EXosite-binding aptamers with small molecule ACTive site inhibitors. The aptamer component within the EXACT inhibitor (1) synergizes with and enhances the potency of small-molecule active site inhibitors by many hundred-fold (2) can redirect an active site inhibitor's selectivity towards a different protease, and (3) enable efficient reversal of inhibition by an antidote that disrupts bivalent binding. One EXACT inhibitor, HD22-7A-DAB, demonstrates extraordinary anticoagulation activity, exhibiting great potential as a potent, rapid onset anticoagulant to support cardiovascular surgeries. Using this generalizable molecular engineering strategy, selective, potent, and rapidly reversible EXACT inhibitors can be created against many enzymes through simple oligonucleotide conjugation for numerous research and therapeutic applications. Inspired by the biologics of hematophagous organisms such as leeches, the authors in this work design and create inhibtors of thrombin and factor Xa by linking exosite-binding aptamers with small molecule active site inhibtors. They coin these inhibitors EXACT inhibitors.

Background : Coagulation cascade cofactors are essential proteins that control homeostasis and are appealing targets for anticoagulants. However, targeting such proteins has been challenging due to their lack of an active site. Aims : Development of FV/FVa and FVIII-inhibiting aptamers as potent and reversible anticoagulants that can be used to aid (or replace) heparin for therapeutic anticoagulation. Reveal new mechanisms of cofactor inhibition. Methods : 2 ' F modified RNA Aptamers were identified via in vitro systematic evolution of ligand by exponential enrichment (SELEX) and post-SELEX truncation. Aptamer affinity was characterized by nitrocellulose filter binding assays and surface plasmon resonance. Aptamer bioactivity was analyzed using clotting assays and biochemical assays. Aptamers ' binding mechanism was determined using fluorescent anisotropy, fluorescence resonance energy transfer, dynamic light scattering, and molecular modeling. Results : An aptamer termed T18.3 binds to both human FV/FVa with K D ∼10 nM and prolongs aPTT in normal human plasma by 3.5-fold at a concentration of 500 nM. The aptamer inhibits membrane docking of FVa by binding to FV/FVa light chain, thereby reduces prothrombinase assembly and thrombin generation. Notably, the aptamer showed similar anticoagulant potency in normal, FV Leiden, and COVID-19 patient plasma. The aptamer also synergistically inhibits clotting with enoxaparin and can be rapidly reversed by protamine in in vitro assays. Another aptamer termed F8-3.1 binds to human FVIII with a K D of 0.67 nM and prolongs aPTT in normal human plasma by 1.9-fold at a concentration of 500 nM. F8-3.1 also has full cross-reactivity to canine FVIII (K D = 0.68 nM) and significantly prolongs the aPTT in canine and porcine plasma. Conclusions : The work describes the generation of aptamers targeting FV/FVa and FVIII that can achieve clinically relevant anticoagulant activity. These results not only demonstrate the feasibility of using cofactor-binding aptamers as therapeutic anticoagulants but also reveal a novel mechanism of aptamer-mediated protein inhibition by interrupting protein-membrane interactions.

Emicizumab is a bispecific antibody that augments Factor X (FX) activation by factor IXa (FIXa) in the absence of factor VIIIa (FVIIIa) by binding and approximating the substrate FX and protease FIXa. Clinical use of this FVIIIa mimetic significantly reduces bleeding episodes in Hemophilia A (HA) patients. Membrane dependent assembly of FVIIIa-FIXa-FX is an absolute requirement for the intrinsic tenase activity. In contrast, emicizumab lacks any membrane binding ability itself and exhibits significant FXa generation even in the absence of membranes. This major distinction likely allows FVIIIa to function at sub-nanomolar concentrations to greatly accelerate the activation of FX while only much lower enhancements of rate are achieved at several 100-fold higher concentrations of emicizumab. We sought to test whether the addition of a membrane binding feature in the monospecific FIXa binding fragment of emicizumab could alone approximate the cofactor activity obtained with the bispecific antibody. We first generated a single chain antibody fragment (scFv) of the FIXa binding arm of emicizumab by joining the variable heavy chain (V H) with the variable light chain (V L) to produce V H9V L. In line with previous observations that FIXa one arm antibodies of emicizumab that can only bind FIXa and impart a modest cofactor mimetic activity, V H9V L, facilitated FIXa-mediated FX activation at a much lower rate in comparison to bispecific emicizumab. The scFv, V H9V L, was used as a template to introduce membrane binding properties. The Factor V C2-domain, which binds membrane, was fused at the C-terminus of V H9V L via a repeating Gly 4Ser flexible linker of ~ 5.7 nm to allow the V H and V L domains of the scFv to potentially orient correctly at the membrane surface. The resulting construct V H9V LC2 showed a substantial amplification in cofactor mimetic activity as compared to V H9V L. In vitro kinetic analyses with purified proteins and synthetic phospholipid vesicles containing phosphatidylcholine (PC) and phosphatidylserine (PS) (PC:PS, 75:25 %) showed approximately 22-fold faster activation of FX by V H9V LC2 as compared to V H9V L. Remarkably, despite its inability to bind FX, cofactor mimetic activity of V H9V LC2 was about two-fold higher than that of emicizumab. Potent enhancement of FX activation by V H9V LC2 observed with purified proteins was also reflected in thrombin generation assays (TGA). V H9V LC2 added at 100 nM brought thrombin generation into the reference range in congenital FVIII-deficient HA plasma supplemented with 4 µM PC:PS and triggered with either low tissue factor (0.1 pM), or factor XIa (0.1 nM). The role of C2 domain-mediated increase in cofactor activity of V H9V LC2 was further assessed by using an antibody fragment (E 9scFv) that inhibits the binding of factor V to membranes. Increasing concentrations of E 9scFv caused a dose-dependent reduction in the rate of FXa formation catalyzed by the complex of V H9V LC2: FIXa (20nM:25nM). Inhibition saturated at ~ 85% at 100 nM E 9scFv, indicating a substantial contribution of membrane binding to the cofactor mimetic function of V H9V LC2. To evaluate whether V H9V LC2 can restore clotting in HA plasma with FVIII inhibitors, normal pooled plasma was supplemented with a FVIII neutralizing antibody to mimic FVIII inhibitor plasma. Addition of V H9V LC2 to this plasma restored clotting time to the levels shown by normal pooled plasma, suggesting that V H9V LC2 can bypass FVIII inhibitor activity. The engineered scFv with a factor V C2 domain fusion, V H9V LC2, demonstrates that mere addition of a membrane binding feature to the FIXa arm of the emicizumab surpasses the mimetic activity achieved by a bispecific antibody that exploits both bridging and allostery to be an effective cofactor. Such a construct also avoids the burden of productive assembly from a combination of three separate polypeptide chains, as in a functional bispecific antibody. Replicating the membrane binding feature into bispecific FVIIIa mimetic antibodies could further amplify function and more closely resemble FVIIIa in both activity and membrane dependent regulation. Our results provide surprising insights into how a single chain, monospecific, membrane-anchored FVIIIa mimetic enhances the catalytic activity of FIXa.

Coagulation cofactors profoundly regulate hemostasis and are appealing targets for anticoagulants. However, targeting such proteins has been challenging because they lack an active site. To address this, we isolate an RNA aptamer termed T18.3 that binds to both factor V (FV) and FVa with nanomolar affinity and demonstrates clinically relevant anticoagulant activity in both plasma and whole blood. The aptamer also shows synergy with low molecular weight heparin and delivers potent anticoagulation in plasma collected from patients with coronavirus disease 2019 (COVID-19). Moreover, the aptamer's anticoagulant activity can be rapidly and efficiently reversed using protamine sulfate, which potentially allows fine-tuning of aptamer's activity post-administration. We further show that the aptamer achieves its anticoagulant activity by abrogating FV/FVa interactions with phospholipid membranes. Our success in generating an anticoagulant aptamer targeting FV/Va demonstrates the feasibility of using cofactor-binding aptamers as therapeutic protein inhibitors and reveals an unconventional working mechanism of an aptamer by interrupting protein-membrane interactions.

Unfractionated heparin (UFH), the standard anticoagulant for cardiopulmonary bypass (CPB) surgery, carries a risk of post-operative bleeding and is potentially harmful in patients with heparin-induced thrombocytopenia-associated antibodies. To improve the activity of an alternative anticoagulant, the RNA aptamer 11F7t, we solved X-ray crystal structures of the aptamer bound to factor Xa (FXa). The finding that 11F7t did not bind the catalytic site suggested that it could complement small-molecule FXa inhibitors. We demonstrate that combinations of 11F7t and catalytic-site FXa inhibitors enhance anticoagulation in purified reaction mixtures and plasma. Aptamer-drug combinations prevented clot formation as effectively as UFH in human blood circulated in an extracorporeal oxygenator circuit that mimicked CPB, while avoiding side effects of UFH. An antidote could promptly neutralize the anticoagulant effects of both FXa inhibitors. Our results suggest that drugs and aptamers with shared targets can be combined to exert more specific and potent effects than either agent alone.

Factor V, the inactive precursor to factor Va, has a domain organization of A1-A2-B-A3-C1-C2. Factor Va is formed by the proteolytic excision of the central B domain, which resolves the molecule into a heterodimer (A1-A2/A3-C1-C2). Removal of the B domain enables the cofactor to engage factor Xa on phosphatidylserine-containing membranes, assemble prothrombinase and greatly enhance the rate of thrombin formation. Recent studies have shown a key role for a basic region (BR), which lies approximately in the center of the B domain, in enforcing procofactor properties in human factor V (hFV). Exogenously added recombinant BR fragments can bind with high affinity to a cofactor-like variant of human hFV (hFVDT), in which a large central portion of the B domain has been deleted, interfere with Xa binding and restore procofactor-like properties. Biochemical evidence suggests that BR binding results from its interaction with an acidic region (AR2) at the C terminus of the B domain and likely also an acidic sequence (AR1) at the C terminus of the A2 domain. Our recent crystal structure of hFVDT provided the first structural evidence that AR1 and AR2, ~800 residues apart in the primary structure of hFV, are positioned adjacent to each other and could plausibly form an extended surface for high affinity BR binding to reconstitute a tripartite procofactor-regulatory region (AR1/BR/AR2). However, the lack of BR in hFVDT precluded independent structural verification of this possibility. In a computational approach, we created a molecular model for the 58 residue BR peptide. The top scoring three-dimensional models of the 58 residue BR peptide showed a helix-loop arrangement, contrary to the general belief that the B domain lacks structured regions. The best scoring BR peptide model was used for ab initio docking studies using the crystal structure of hFVDT to predict possible binding sites using PIPER and ClusPro. The most highly represented and statistically probable solutions showed the BR peptide in intimate contact with juxtaposed surfaces provided by AR1 and AR2. Interestingly, the docked BR peptide contacted regions in AR1 and on the A2 domain implicated in FXa binding in the structure of Pseudonaja textilis FV bound to snake venom factor X. Computational predictions were tested using hydrogen-deuterium exchange detected by protein fragmentation and mass spectroscopy (HDX). Proteolytic fragmentation of hFVDT and fragment detection by LC-MS was optimized to cover >95% of its 1514 residues with an average redundancy of 4.27 peptides/residue. Only 4 or 5 segments of ~10-15 residue length were not covered. Addition of the BR peptide had minor effects on amide proton exchange over the bulk of the molecule. However, BR peptide binding was accompanied by reductions in amide proton exchange rates of ~7-30-fold in immediately adjacent regions of hFVDT corresponding to sequences within A2 (626-634), AR1 (658-695), AR2 (872-881) and A3 (983-995). BR peptide binding to hFVDT is accompanied by perturbations in these spatially adjacent regions covering a small fraction of the surface area at approximately the 3 o’clock position with the molecule in the standard orientation. The marked agreement between the HDX findings and the computational docking studies supports our proposal that the BR engages an extended surface contributed by AR1 and AR2 to form a tripartite procofactor-regulatory region. The interaction of BR with AR1 and a small region in A2, both implicated in binding Xa, potentially explains how the BR might restrict Xa binding to the procofactor. Destabilization of BR binding by proteolysis at the C terminus of AR2 is envisioned to result in cofactor formation by releasing the BR and revealing sites responsible for binding Xa. Our findings provide a structural explanation for the long standing puzzle of factor V activation and pave the way for further definition of mechanistic details of procofactor and cofactor function. They have implications for how interactions with TFPIα through the basic region at its C-terminus might regulate FV(a). They also reveal previously unanticipated strategies to modulate functions of hFV and hFVa for therapeutic gain.

Coagulation factor V (FV) circulates as an inactive procofactor with a domain organization of A1-A2-B-A3-C1-C2. Factor Va (FVa), the active cofactor, is produced in steps essential for rapid thrombin formation, by the proteolytic excision of the B domain which resolves the molecule into a heterodimer (A1-A2/A3-C1-C2). Removal of the B domain imbues the resulting FVa with the ability to bind factor Xa (Xa) on a membrane surface to assemble prothrombinase and greatly enhance the rate of thrombin formation. A recombinant variant of human factor V (HFVDT) with a shortened B domain exhibits constitutive cofactor activity. Cofactor activity even without proteolysis arises from its lack of a conserved basic region (BR) located in the large B domain of FV. Exogenously added BR peptide binds tightly to HFVDT in a Ca2+-dependent fashion and restores procofactor-like properties. The BR is proposed to restrict Xa binding and cofactor function by interacting with an acidic region at the C terminus of the B domain (AR2) and likely also an acidic sequence at the C terminus of the A2 domain (AR1). These two sequences are ~800 residues apart in FV with no structural information to explain how AR1 and AR2 might cooperate to engage the BR in the central portion of the B domain to autoinhibit FV. The available structures of an inactivated form of bovine factor Va (BFVai), of a FV ortholog from Pseudonaja textilis (FVPtex) and a lower resolution structure of B domainless human factor VIII (HFVIII) shed no light on this problem. We obtained diffraction quality crystals of HFVDT complexed with a single chain antibody (scFvE10) directed to FVa. Crystals were not obtained in the absence of scFvE10. The crystals diffracted to a resolution of 2.8 Å and the structure was solved by molecular replacement. The refined structure shows high similarity to BFVai, FVPtex and HFVIII. Insufficient electron density precluded the placement of scFvE10 in the modeled structure. The three homologous A domains in HFVDT adopt a typical cupredoxin-fold with the A domains arranged in a pseudo-three-fold axis of symmetry. The two C domains are cylindrical and oriented side-by-side to form the base of the A domain rosette. These features are equivalent to those seen in structures of BFVai, HFVIII and FVPtex. Two bound calcium ions are evident, one in A1 and the other in the A3 domain. The most important feature newly revealed in the structure of HFVDT is the close spatial proximity of AR1 and AR2 at the outer edge of the A domain rosette at the 3 o'clock position in the standard orientation. These acidic regions form adjacently positioned surfaces in spite of being bisected by the long primary sequence of the intervening B domain. Our observations provide the first structural evidence that the two distinct acidic regions come together in space to provide an extended surface. This provides a plausible explanation for how the BR in the middle of the B domain may bind to both AR1 and AR2 to restrict cofactor function in FV. The need for this extended but bipartite acidic surface, to which the BR may bind, also provides a plausible explanation for how proteolytic cleavage at position 1545 at the C terminus of AR2 destabilizes BR binding and results in cofactor formation. The Ca2+-stabilized loop in the A3 domain abuts the bisegmental acidic cluster potentially explaining why BR binding to HFVDT is strongly dependent on Ca2+. In the structure of FVPtex bound to snake venom factor X, AR1 extends away from the body of the cofactor to make intimate contacts with factor X. If this is mirrored in human prothrombinase, then our findings provide a structure-based model to phrase the long-standing procofactor activation paradox. BR binding to the AR1/AR2 extended surface ties up surfaces necessary for Xa binding and restricts cofactor activity. Proteolytic processing of the B domain and probably most importantly following AR2 destabilizes the BR/AR1/AR2 complex to free up surfaces including AR1 necessary to support Xa binding. This model reconciles the biochemical evidence with structural findings to provide new insights into the role played by the BR/AR1/AR2 complex in restricting Xa binding and cofactor function in FV. It provides a platform to further explore mechanistic details of FV and FVa function and for the development of novel strategies to modulate their functions to regulate thrombin formation for therapeutic gain.

The prothrombinase complex assembles through high affinity interactions between factor Xa and factor Va on the surface of membranes exposing phosphatidylserine. Assembly of prothrombinase allows for the highly accelerated formation of thrombin localized at the injury site. Because the rate of thrombin formation by factor Xa alone is profoundly increased upon its incorporation into prothrombinase, disrupting the proteinase-cofactor interaction could represent an effective therapeutic strategy for the treatment of thrombosis. However, the extended interface between Xa and Va, likely involving shallow surfaces of the interacting proteins, limits the available strategies for the development of ligands that interfere with prothrombinase assembly with high affinity. The RNA aptamer, 11F7t, was selected in an unbiased SELEX screen using human factor Xa as a target and identified as a potent anticoagulant. Detailed functional characterization revealed that 11F7t binds both X and Xa with high affinity (~1 nM) and competes for the interaction between Xa and Va without occluding active site function of the proteinase. Competititive inhibition of the assembly of prothrombinase likely accounts for a major fraction of its anticoagulant effects. However, additional inhibitory effects of 11F7t on coagulation also probably arise from its ability to partially inhibit X activation by the intrinsic Xase and the inhibition of factor VIII activation by Xa. Here we report the crystal structure of 11F7t complexed with desGla-Xa containing the catalytic Ser195 replaced with Ala at 2.5 A resolution and solved by molecular replacement. In the Xa model, the first 30 residues expected in the light chain, 5 residues in the autolysis loop and 7 residues at the C terminus were not defined by electron density. In 11F7t, 25 of 38 nucleotides could be unambiguously modeled. The tertiary fold of 11F7t presents an extended molecular surface for interactions with desGla-Xa and buries 1384 A2 of solvent accessible surface area in the interacting species. The RNA aptamer exclusively binds the proteinase domain with contacts over a broad surface area that includes residues Leu59-Lys62, Val87-Arg93, Phe101, Lys236, Trp237 and Arg240 but not the catalytic site. Some of these residues have been implicated in heparin binding and have also been proposed to participate in Va binding based on modest changes in function upon mutagenesis. Even though 11F7t does not make any contacts in the 165 helix previously proposed to be a key region for the interaction with Va, aptamer binding is sufficient to disrupt the high affinity cofactor-proteinase interaction. The structure also reveals that most of the aptamer–protein contact involves nucleotides that constitute a base-paired stem (formed between G3-C6 and 31G-34C) and loop (C7-G11) of 11F7t with these interactions mostly specified by the nucleotide bases rather than the phosphate backbone which might be expected for a protein surface already implicated in binding a polyanion such as heparin or an acidic peptide from the C terminus of the A2 domain of Va. The regions of desGla-Xa occluded by the aptamer point to an extended surface that plays an important role in mediating interactions with factor Va within prothrombinase and suggests commonalities between this binding mode and the way that factor X might bind to factor VIIIa as a substrate within intrinsic Xase or the way that factor Xa might engage factor VIII as an enzyme. Disclosures No relevant conflicts of interest to declare.

Factor Va binds to membranes exposing phosphatidylserine (PS) with nanomolar affinity. This high affinity interaction plays a vital role in the assembly of membrane-bound prothrombinase thereby supporting robust thrombin formation at the site of vascular damage. Although the importance of the C1 and C2 domains of FVa in membrane binding has long been recognized, mechanistic details are incompletely understood. Unlike human prothrombinase, Pseudonaja textilis (common brown snake, P. tex) has evolved a membrane-independent form of prothrombinase composed of a FVa-like protein (VPtex) tightly bound to a FXa-like protein in solution. Our structural advances with FVPtex provide new tools to address major unresolved questions related to membrane binding by FVa. The high resolution (1.95Å) structure of VPtex resembles previously published structures of inactivated bovine FVa and human FVIII of lower resolution with all the structural features considered critical for membrane binding by FVa. However, VPtex bound to PS-containing membranes with very poor affinity (Kd > 1 µM). Substitution of 9 residues in the C1 and C2 domains of VPtex with residues present in the hemostatic form of FV present in the plasma of the snake yielded a derivative (VPtexC1C2) that bound to PS-containing membranes with nanomolar affinity equivalent to hFV. Interestingly, the newly acquired function of membrane binding in VPtexC1C2 does not affect its ability to function in solution, suggesting that membrane binding and solution-phase function are controlled independently. Variants containing substitutions in the individual C domains (VPtexC1 and VPtexC2) exhibited intermediate affinities (Kd=100 nM and Kd=50 nM) for binding to PS-containing membranes. However, the binding energy contributions from the individual C-domains did not additively explain the affinity of VPtexC1C2 for membranes. The large connection energy (-8.7 kcal/mole) implies substantial energetic expenditure, possibly through a conformational rearrangement, upon membrane binding. This correlates well with higher thermal factors observed in the C1 and C2 domains of structures of VPtexC1C2 and VPtexC2 as compared to VPtex. It is also supported by rapid kinetic studies illustrating equivalence in the bimolecular association rate constants for human Va and VPtex variants regardless of their membrane affinity. Thus, high affinity membrane binding results from large decreases in the dissociation rate constant expected from a conformational change that allows the protein to adopt a new stable membrane-bound configuration. A second explanation for the lack of an obvious correlation between x-ray structures of VPtexC1C2, VPtexC2 and VPtex and their affinity for membranes lies in the possibility that their solution-phase conformations differ. We explored this using small angle x-ray scattering (SAXS) of VPtex, VPtexC2 and VPtexC1C2 in solution. The low resolution SAXS envelope for VPtex could be accounted for by minor shifts in the individual domains, particularly in C1 and C2, as seen in the crystal structure. However, the SAXS envelopes for membrane binding variants (VPtexC2 and VPtexC1C2) showed major shape changes in the C-domains. Rigid body modeling revealed an increasingly extended end-on arrangement, rather than a side-by-side configuration, of the C1 and C2 domains seen in the x-ray structure and in the SAXS envelope for FVPtex. The C2-domain was found to extend away from the base of the C1 domain along the long axis of the molecule correlating major structural differences in these VPtex variants with increasing affinity for membranes. Accordingly, the spatial disposition of the C1 and C2 domains in VPtexC2 appears intermediate to their arrangement in VPtex and VPtexC1C2.These findings contrast to the arrangement seen in the crystal structures of all factor V forms, where the C-domains are arranged side-by-side, probably due to limitations imposed by crystal packing. Our SAXS studies provide clear evidence of an unforeseen framework of C-domains associated with the ability of factor V forms to bind to membranes with high affinity. The findings reveal new mechanistic insights into the structural correlates of the membrane binding function of factor V.

Abstract Abstract 375 Poisonous snakes frequently harbor activators of mammalian coagulation as part of the toxin repertoire in their venom. The venom of Pseudonaja textilis (Ptex, common brown snake) contains an efficient activator of human prothrombin comprised of a Xa-like protein tightly bound to a Va-like protein. The constituents of this complex exhibit high sequence homology to the corresponding activated coagulation factors in mammalian blood. Factors Xa and Va are produced in blood upon proteolytic activation of their precursors, complex with each other in membrane-dependent reactions to form prothrombinase and catalyze thrombin formation at the site of vascular damage. In contrast, the venom proteins are constitutively active, form a complex in solution and can efficiently catalyze prothrombin activation in the absence of membranes. These properties likely drive the disseminated and consumptive coagulopathy associated with evenomation by P. textilis. The Va-like component (Ptex-Va) of P. textilis venom is a single chain glycoprotein of 1430 residues with 53% identity to human factor V (hV) and a common A1-A2-B-A3-C1-C2 domain organization. The B-domain of Ptex-Va is significantly shorter than its counterpart in hV (46 vs. 836 residues). We now report a high resolution x-ray structure of recombinant Ptex-Va collected at 1.9 Å resolution and solved by molecular replacement. The resulting structure closely mimics those seen at lower resolution for inactivated bovine factor Va lacking the A2 domain and full length B-domainless human factor VIII (hVIII). Each A domain is formed by two cupredoxin-like β barrels with the A domains arranged in a pseudo-three-fold axis of symmetry. The two C domains are roughly cylindrical and oriented side-by-side to form a pedestal for the A1-A2-A3 rosette. The A3 domain makes extensive contacts with the C1 and A2 domains. The structure also reveals a disulfide bond unique to Ptex-Va, linking Cys642 in the A2 domain with Cys1002 in the A3. These features likely account for the high stability of the molecule even after proteolytic processing of the B domain and/or cleavage between the A1 and A2 domains. Although the C2 domain is significantly more disordered than the other domains, both C1 and C2 each contain protruding loops at their base with hydrophobic residues pointing outward. These structural features replicate those found in inactivated bovine Va and hVIII considered critical for membrane binding by the hemostatic cofactors. Surprisingly, despite the presence of these structures, light scattering measurements revealed negligible binding of Ptex-Va to synthetic membranes composed of phosphatidylcholine and phosphatidylserine with an estimated 103-fold weaker affinity than that of hV. We reasoned that this unexpected property of the venom protein, not conducive to regulated coagulation, was unlikely to be replicated in factor V from the plasma of the snake. Sequence alignment of Ptex-Va with factor V from snake plasma revealed 11 differences in the 328 residues of the C1 and C2 domains. Of these, 9 were located on the distal region of these domains, occupying a band approximately 7 Å thick across the molecule. Mutagenesis of Ptex-Va to introduce these 9 substitutions followed by its expression and purification yielded a derivative that bound to membranes with high affinity in a manner equivalent to hV. This striking gain in function sheds new and unexpected light on the structural determinants of high affinity membrane binding in Ptex-Va and by extension, its homologues hV and hVIII. Our high resolution structure of this hV-like species with a series of unusual properties provides a unique platform to address major but unresolved questions related to the structural correlates of hV function. It also reveals the basis for molecular mimicry whereby a cofactor essential for regulated blood coagulation has served as a scaffold for the evolution of a potent toxin by simultaneous loss in the ability to bind membranes and a gain in the ability to bind its proteinase with high affinity in a membrane-independent fashion. Disclosures: No relevant conflicts of interest to declare.