Engineered Factor Xa Variants Retain Procoagulant Activity Independent of Direct Factor Xa-Inhibitors

The venom of the Australian Elapid snake Pseudonaja textilis contains a prothrombin-activating complex that consists of factor Xa (FXa) and factor Va (FVa) homologs which are evolutionary adapted to derail the hemostatic system of its prey, leading to runaway coagulation. These adaptations include functional resistance to inactivation by the main inhibitors of coagulation, antithrombin and activated protein C. Further studies revealed that venom FXa, unlike other FXa species, is also resistant to inhibition by direct oral FXa-inhibitors such as rivaroxaban and apixaban (K_i >1000 nM for venom FXa vs. 1 nM for human FXa).  Accordingly, venom FXa is able to support thrombin generation (TG) in FX-depleted plasma spiked with pharmacological concentrations (0.4-2 μM) of these FXa-inhibitors. While human FXa-initiated TG resulted in a 8-fold prolonged lag time and a 70% reduced thrombin peak, those parameters were within normal range in venom FXa-triggered TG.

Venom FX homologs produced by Elapid snakes comprise a heterogeneous insertion between His91-Tyr99 (chymotrypsin numbering) in the serine protease domain. A recent crystal structure of one of these homologs shows that this insertion is in close proximity of the active site pocket. In contrast, P. textilis liver-derived plasma FX, which, when activated, is fully inhibited by the FXa-inhibitors (K_i10 nM), lacks this structural feature. We investigated whether the His91-Tyr99 insertion is at the basis of the reduced sensitivity of venom FXa towards FXa-inhibitors. To do so, we constructed and stably expressed human-snake FX chimeras (FX-A, -B, -C) that incorporate His91-Tyr99 insertions from three venom FXa homologs. The chimeric FX variants were purified by successive ion-exchange and hydrophobic interaction chromatography steps, and FXa was generated following RVV-X-activation and size-exclusion chromatography.

Evaluation of the kinetic parameters of prothrombin conversion in the presence of saturating amounts of FVa and anionic phospholipids revealed that the chimeric FXa variants exhibit an up to ~4-fold enhanced affinity for prothrombin (K_m 0.11-0.29 μM) as compared to recombinant human FXa (rhFXa; K_m 0.41 μM). The rate of prothrombin activation was 3-10-fold reduced (k_cat 118-370 min^-1 vs. 1243 min^-1 for rhFXa), which may be indicative of a modified active site conformation. Consistent with this, the rate of chimeric FXa inhibition by antithrombin was impaired (k_app 0.12-0.95 x 10³ M^-1s^-1 vs. 4.07 x 10³ M^-1s^-1 for rhFXa). Furthermore, the variant that was most poorly inhibited by antithrombin (variant A) also exhibited the lowest catalytic rate of prothrombin conversion and vice versa (variant C). Conversely, apixaban or edoxaban inhibition of the FXa variants assembled into prothrombinase led to the highest K_i for chimeric variant C (2.3 or 0.3 µM), followed by variants B (1.4 or 0.2 µM), and A (0.2 or 0.006 µM) compared to rhFXa (0.004 or 0.0005 µM). Evaluation of the inhibition of uncomplexed FXa variants employing peptidyl substrate conversion revealed a similar decrease in sensitivity to the FXa-inhibitors (≤500-fold). These data suggest that insertion of the snake venom His91-Tyr99 regions indeed results in impaired engagement of the FXa active site pocket.

We next assessed whether chimeric variant C, which is most resistant to inhibition by the direct FXa-inhibitors, is able to restore thrombin generation in a plasma system in the presence of apixaban or edoxaban. While rhFXa-triggered (5 nM) thrombin formation in FX-depleted plasma was inhibited by 2 μM apixaban, initiation with FXa-C (5 nM) resulted in normal thrombin generation parameters (peak thrombin 98%). In addition, the zymogen form of variant C also supported tissue factor-initiated (2 pM) thrombin generation in FX-depleted plasma with inhibitor concentrations up to 6 μM (apixaban) or 2 μM (edoxaban). Under these conditions, little if any thrombin was formed with rhFX present (peak thrombin 5%). We obtained similar results when performing these experiments in normal pooled plasma.

Taken together, these results show that chimeric FX is able to restore hemostasis in plasma inhibited by the direct FXa-inhibitors, both in the zymogen as well as protease state. As such, these variants have the potential to serve as rescue therapeutic agents to overcome the effect of FXa-inhibitors in case of potential life-threatening bleeding events or emergency surgical interventions.