Conservative mutations in the C2 domains of factor VIII and factor V alter phospholipid binding and cofactor activity

Abstract

Factor VIII and factor V share structural homology and bind to phospholipid membranes via tandem, lectin-like C domains. Their respective C2 domains bind via 2 pairs of hydrophobic amino acids and an amphipathic cluster. In contrast, the factor V-like, homologous subunit (Pt-FV) of a prothrombin activator from Pseudonaja textilis venom is reported to function without membrane binding. We hypothesized that the distinct membrane-interactive amino acids of these proteins contribute to the differing membrane-dependent properties. We prepared mutants in which the C2 domain hydrophobic amino acid pairs were changed to the homologous residues of the other protein and a factor V mutant with 5 amino acids changed to those from Pt-FV (FV(MTTS/Y)). Factor VIII mutants were active on additional membrane sites and had altered apparent affinities for factor X. Some factor V mutants, including FV(MTTS/Y), had increased membrane interaction and apparent membrane-independent activity that was the result of phospholipid retained during purification. Phospholipid-free FV(MTTS/Y) showed increased activity, particularly a 10-fold increase in activity on membranes lacking phosphatidylserine. The reduced phosphatidylserine requirement correlated to increased activity on resting and stimulated platelets. We hypothesize that altered membrane binding contributes to toxicity of Pt-FV.

Figure 1

Crystal structures of the FVIII C2 domain (PDB entry 1D7P) and FV C2 domain (PDB entry 1CZT). The membrane-interactive hydrophobic amino acids for spikes 1 and 3 of FVIII (left) and FV (right) are shown with side chains in dark gray, whereas the rest of the structures show the backbone only in light gray. Residue numbers relate to the full-length proteins.

Figure 2

Relative activities, membrane interaction, and affinities for FVIII mutants. (A) Membrane interaction for FVIII_WT (■) or FVIII_LS (▾) was evaluated by mixing FIXa, FX, and various concentrations of sonicated vesicles. The quantity of FXa formed was measured with chromogenic substrate S-2765. The membrane reactivity of FVIII_LS was less than FVIII_WT. The vesicles had a composition of PS:PE:PC 4:20:76; concentration of FVIII or mutants was 0.2nM; FIXa, 4nM; and FX, 130nM. Results represent the mean ± SEM for at least 3 experiments, each performed in duplicate and fitted to a single-site binding model (smooth curves). (B) Apparent K_M for the FVIIIa-FIXa complex was determined as in panel A, except that the phospholipid concentration was constant (100μM) and the FX concentration varied. (A-B) Fitted results were normalized to the B_max value for each curve to optimize comparison of the relative affinities. (C) The specific activities, apparent phospholipid interactivity, and apparent FX affinities for each mutant are displayed with reference to control experiments performed with wild-type FVIII (as in panels A and B for FVIII_LS). The displayed values represent the mean ± SEM. Each fit was performed on a minimum of 3 separate experiments, each performed in duplicate. (D) Apparent VWF affinity was determined by incubating FVIII_WT and mutants with various concentrations of VWF for 60 minutes before addition to a microtiter well coated with a monoclonal antibody that competes for the VWF-binding epitope. Bound FVIII was detected with a second antibody. The fraction of VWF-bound FVIII was calculated as the difference between the maximum signal in the absence of VWF and the signal at each concentration of VWF. Smooth curves represent best-fit curves corresponding to K_D values of 0.7 ± 0.1nM, 2.2 ± 0.7nM, and 4.8 ± 0.8nM for FVIII_WT (■), FVIII_WW (▴), and FVIII_LS (▾), respectively. Displayed results are the mean ± SEM of at least 2 experiments, each performed in duplicate.

Figure 3

Relative activities, membrane interaction, and affinities for FV mutants. (A) Membrane reactivity for FV_WT (■) or FV_LL (▾) was determined by mixing with FXa, prothrombin, and various concentrations of sonicated vesicles. The quantity of thrombin formed was measured with chromogenic substrate S-2238. The apparent affinity of FV_LL was higher than FV_WT. The vesicles had a composition of PS:PE:PC 4:20:76; concentrations of FV and mutants were 0.2nM; FXa, 2.5pM; and prothrombin, 2μM. Displayed represent the mean ± SEM, fitted to a single-site binding model (smooth curves). (B) Apparent FXa affinity was determined as in panel A, except that the phospholipid concentration was held constant (50μM) and the FXa concentration was varied. (C) Apparent K_M for the FVa-FXa complex was determined as in panel A, except that the phospholipid concentration was constant and the prothrombin concentration varied. (A-C) Fitted results were normalized to the B_max value for each curve to optimize comparison of the relative affinities. (D) The specific activities, phospholipid vesicle reactivity, apparent FXa affinities, and apparent prothrombin affinities (K_M) for each mutant are displayed with reference to control experiments performed with wild-type FV (as in panels A-C for FV_LL). The displayed values represent the mean ± SEM. Each fit was performed on a minimum of 3 separate experiments, each performed in duplicate.

Figure 4

Apparent membrane-independent activity of FV mutants resulting from copurification of phospholipid. (A) The potential presence of phospholipid was evaluated by testing the capacity of the factor FV_WT and mutants to support the FXase complex. FV preparations were mixed with FVIIIa, FIXa, and FX in the absence of phospholipid vesicles. FV mutants supported FXase activity, whereas FV_WT supported little or no activity. Data represent the mean ± SEM for 2 experiments for FV_LL and FV_WT and one experiment for the other mutants each performed in duplicate. (B) Lactadherin inhibition of FV_MF-supported FXase activity was evaluated using increasing concentrations of lactadherin. Activity was inhibited with half-maximal inhibition at approximately 4nM. Data represent the mean ± SEM for a single experiment performed in duplicate. (C) Prothrombinase activity supported by CHAPS-washed FV_WT and FV mutants in the absence of phospholipid vesicles was evaluated in the presence of various concentrations of lactadherin. Lactadherin inhibited activity supported by FV_MF (▵) with half-maximal inhibition at approximately 16nM lactadherin. Half-maximal inhibition of activity supported by FV_LL (▿) occurred at approximately 1nM, whereas half-maximal inhibition of FV_MF/LL (♢) occurred at an intermediate concentration. Data represent the mean ± SEM for 2 experiments. (D) Inhibition of prothrombinase activity by phospholipase A₂ was evaluated for FV_LL (▿) in the absence of phospholipid vesicles and for FV_WT (□) in the presence of phospholipid vesicles (inset). Phospholipase A₂ inhibited prothrombinase activity under both conditions. Data represent the mean ± SEM for 3 experiments performed in duplicate.

Figure 5

P textilis–motivated FV mutant. (Left) Sequence alignment of the C2 domains from human FVIII (H-FVIII-C2), human FV (H-FV-C2), bovine FV (B-FV-C2), and the FV-homologous subunit of pseutarin C from P textilis venom (Pt-FV-C2). Black bars represent the 4 lipid binding spikes with the hydrophobic residues marked with triangles. Amino acids that were mutated to make FV_MTTS/Y are starred. (Right) Locations of the mutated amino acids on the C2 domain of human FV are shown with original side chains in dark gray with the rest of the backbone in light gray. Residue numbers relate to the full-length human FV. Numbering references follow HGVS standard, using Met of respective propeptides as 1.

Figure 6

Membrane-independent activity, apparent prothrombin and FXa affinities, and phospholipid binding affinity of FV_MTTS/Y. (A) Various concentrations of FV_WT or FV_MTTS/Y with or without a 1% CHAPS wash during purification were mixed with FXa (0.4nM) and prothrombin (1μM) in the absence of phospholipid vesicles. Washing with 1% CHAPS during purification abolished all apparent lipid-independent activity. (B) Apparent K_M for the prothrombinase complex with mutant or WT CHAPS-washed FV was determined with saturating concentrations of vesicles of composition 4:20:76 (PS:PE:PC). FV_MTTS/Y showed the same apparent affinity for prothrombin as FV_WT. Values represent mean ± SEM for at least 2 experiments, each performed in duplicate. (C) Direct affinity for phospholipid vesicles was evaluated for FV_WT or FV_MTTS/Y by immobilizing FV to mAb CBC-MOR101 covalently linked to Superose beads. Beads were incubated overnight with either FV_WT or FV_MTTS/Y and then mixed with various concentrations of fluorescein-labeled vesicles of composition 4:5:20:71 (PS:PE-CF:PE:PC). After 30 minutes, the quantity of vesicles bound to FV_WT or FV_MTTS/Y was measured by flow cytometry. The vesicle dissociation constants were 4.8 ± 1.1μM for FV_MTTS/Y and 1.7 ± 0.4μM for FV_WT. Data represent the mean ± SEM for at least 4 experiments.

Figure 7

Decreased PS dependence of FV_MTTS/Y activity on vesicles and platelets. (A) Activity on vesicles with high PS content was determined with 5pM CHAPS-washed FV_WT or FV_MTTS/Y and various concentrations of vesicle of composition 15:20:65 (PS:PE:PC). (B) Activity on vesicles with limiting PS was evaluated with vesicles of composition 2:20:78 (PS:PE:PC) using 10pM FV. (C) Activity on vesicles lacking PS was evaluated on vesicles of composition 20:80 (PE:PC) using 50pM FV. FV_MTTS/Y showed increasingly higher relative prothrombinase activity with lower vesicle PS content. Concentrations of FXa and prothrombin were 1nM and 1μM, respectively. Data represent the mean ± SEM for 3 experiments, each performed in duplicate. (D) Prothrombinase activity of FV_MTTS/Y on unstimulated platelets or platelets incubated for 10 minutes with 10μM thrombin receptor agonist protein was measured in the presence of FXa and prothrombin. Platelets were suspended at 4 × 10⁷/mL, and concentrations of FV and mutant were 20pM; FXa, 25pM; and prothrombin 1μM. Data represent the mean ± SD for 2 experiments, each performed in duplicate. Probabilities were calculated using the Student t test.