Engineered factor Xa variants retain procoagulant activity independent of direct factor Xa inhibitors

Figures

• 

  Fig. 1. The S4 subsite of factor Xa coordinates apixaban binding and inhibition. a The minimal interatomic (MI) distances from the side chains of FXa residues to apixaban were calculated every nanosecond during a 750 ns MD simulation of apixaban-bound human FXa. Box plots (with whiskers from minimum to maximum) display the distribution of the observed MI distances for each of the 15 FXa residues with the shortest side chain to apixaban MI distance (sorted by average). The surface representation (inset) depicts the FXa-bound apixaban (blue) configuration throughout the MD simulation. Color coding (corresponding to box plots): S4 subsite residues Tyr99 (orange), Phe174 (cyan), and Trp215 (yellow), catalytic residue Ser195 (red), and others (gray). b The rate of peptidyl substrate conversion by 5 nM of RVV-X-activated wild-type (gray circles, WT), Tyr99Ala (orange squares, Y99A), Phe174Ala (cyan squares, F174A), or double mutant (black squares, Y99A/F174A) FX was determined with saturating amounts of the cofactor FV-810 (30 nM) and anionic phospholipid vesicles (PCPS, 50 µM) in the absence (V 0) or presence (V i) of increasing apixaban concentrations (0.001–100 µM). The lines were drawn…

• 

  Fig. 2. Functional characterization of the effect of the direct factor Xa inhibitors on factor Xa paralogs. a The rate of peptidyl substrate conversion by purified recombinant P. textilis venom FXa (10 nM; squares) or plasma-derived human FXa (2 nM; circles) was determined in the absence (V 0) or presence (V i) of increasing concentrations (0.01–100 µM) of apixaban (APX; open symbols) or rivaroxaban (RVX; closed symbols). b The rate of peptidyl substrate conversion by RVV-X activated P. textilis venom FX (closed squares), isoform FX (semi-closed diamonds), or liver FX (open diamonds) was determined in the absence (V 0) or presence (V i) of increasing apixaban concentrations (0.01–100 µM). a, b The lines were drawn following nonlinear regression analysis of the data sets, and the fitted parameters for IC50 are shown in the inset. The data are the means of two independent experiments. *For these experiments, FXa inhibition was inefficient, precluding an accurate assessment of the IC50 values

• 

  Fig. 3. Molecular dynamics (MD) simulations of apixaban bound to factor Xa variants. Root-mean-square deviations (RMSDs) of atomic positions of the apixaban ligand during independent 750 ns MD simulations of apixaban binding to either P. textilis isoform FXa (Iso FXa, five independent MD simulations (a–e)) or to human FXa (four independent MD simulations (f–i)) are presented as block averages over 10 ns intervals. The corresponding single standard deviation interval is indicated (gray density). For each independent simulation, the molecular configurations at 750 ns are depicted, in which apixaban (green), the 99-loop (magenta), the isoform FXa extended 99-loop region PQKAYKFDL (yellow), and the catalytic triad residues (red) are highlighted. Molecular configurations at 250 and 500 ns of each simulation are compared with those at 750 ns in Supplementary Fig. 4

• 

  Fig. 4. Snake–human factor X(a) variants. a An alignment of region His91-Asp102 in human FX (WT-FX) and FX variants A (FX-A), B (FX-B), and C (FX-C) is shown. The sequences inserted between His91 and Tyr99 originate from P. textilis venom FX (FX-A), from T. carinatus venom FX (FX-B), or from P. textilis isoform FX (FX-C). b Proteins (3 µg per lane) were subjected to SDS-PAGE under reducing conditions and visualized by staining with Coomassie Brilliant Blue. Lane 1: plasma-derived FXa; lane 2: recombinant wild-type FXa; lane 3: FXa variant A; lane 4: FXa variant B; lane 5: FXa variant C. The protein bands corresponding to the heavy chain derived from α or β FXa (α/β), the FXa light chain (LC), and the apparent molecular weights (kDa) of the standards are indicated. While autoproteolytic excision of the C-terminal portion of FXa-α (residues 436–447) yields the β form of FXa, both isoforms are functionally similar with respect to prothrombinase assembly, prothrombin activation, antithrombin recognition, and peptidyl substrate conversion62. The purified products of wt-FXa and FXa variants B and C migrate predominantly as FXa-β; FXa variant A migrates as a 50/50 mixture of α and β FXa. The…

• 

  Fig. 5. Zymogen factor X-C corrects thrombin generation in plasma spiked with apixaban. a, b Thrombin generation (TG) was measured for 30 min at 37 °C in FX-depleted plasma supplemented with 1 U ml−1 FX-C (15 μg ml−1) in the absence (open symbols) or presence of 2 or 6 μM apixaban (a, APX, blue symbols) or 0.6 or 2 μM edoxaban (b, EDX, orange symbols) with 2 pM tissue factor (TF) and 20 μM PCPS. Thrombin generation was initiated with CaCl2 and a thrombin fluorogenic substrate as detailed in ‘Methods’. The TG profiles of FX-depleted plasma supplemented with 1 U ml−1 (7 μg ml−1) wt-FX in the absence (light gray area under the curve) or presence (dark gray area under the curve) of 2 µM apixaban (a) or 0.6 µM edoxaban (b) are shown. c, d Thrombin generation was measured for 30 min at 37 °C in normal pooled plasma in the presence of 2 pM (c) or 6 pM (d) TF, supplemented with increasing concentrations FX-C (5–40 µg ml−1), 2 μM apixaban, and 20 μM PCPS. The TG profiles of normal pooled plasma in the absence (light gray area under the curve) or presence (dark gray area under the curve) of 2 µM apixaban are shown. All curves are representatives of three to six independent experiments