Introduction: The serine protease factor Xa (FXa) and its cofactor factor Va (FVa) assemble into prothrombinase, the complex that catalyzes thrombin generation. Tissue factor pathway inhibitor alpha (TFPIα), a trivalent Kunitz-type serine protease inhibitor, inhibits the initiation of thrombin generation in part by blocking the initial prothrombinase complex. The TFPIα C-terminus contains a conserved basic region (252-LIKTKRKRKK-261; BR) which binds to a regulatory acidic region (AR) present in FXa-activated and platelet-released forms of FVa, the alternatively spliced FV-short, and genetically altered forms of FV found in the East Texas bleeding disorder and FV Amsterdam. The AR is removed by thrombin. Thus, inhibition only occurs during the initiation phase of blood coagulation. Similar to TFPIα, peptides which mimic the BR inhibit the activity of prothrombinase assembled with forms of FVa that retain the AR. BR peptides have been shown to block FXa from binding FVa and prevent prothrombinase assembly. However, this inhibition requires 2-3 orders of magnitude higher concentrations than are required for inhibition by TFPIα. A possible explanation for this difference is that TFPIα can both block prothrombinase assembly, through its BR, and block prothrombinase activity through its second Kunitz domain (K2). To further define the inhibitory mechanism, we investigated potential interactions between TFPIα and prothrombinase outside of the charged-based BR/AR binding, including direct inhibition of FXa by K2 and binding of the FVa heavy chain by conserved uncharged residues within the BR (L252, I253, T255).

Methods and Results: Direct inhibition of FXa was investigated using TFPIα with an altered K2 (TFPI-R107A) incapable of binding FXa, as well as a protein containing only the third Kunitz domain and C-terminus of TFPIα (K3C). TFPI-R107A inhibited purified prothrombinase 17-fold weaker than TFPIα (IC50 = 30.6nM vs. 1.8nM), and K3C was 34-fold weaker (IC50 = 61.4nM). In contrast to TFPIα, neither TFPI-R107A nor K3C inhibited FXa-initiated thrombin generation in platelet-rich plasma (PRP) at concentrations up to 10nM. Thus, direct binding of FXa by K2 is required for efficient inhibition of prothrombinase in PRP. TFPI-AAKA, in which L252, I253, and T255 were substituted with alanine, was used to assess the functional significance of the basic region uncharged residues. TFPI-AAKA was an intermediate inhibitor of prothrombinase activity, relative to TFPI-R107A or K3C, both in purified prothrombinase assays (IC50 = 10.4nM) and in PRP, revealing that these residues are necessary for optimal inhibition. The function of the uncharged residues was further probed using the peptide LIKTKRKRKK, which inhibited purified prothrombinase (IC50 = 1.0μM) and thrombin generation in PRP at 1μM. Substitution of the uncharged residues with alanine (AAKAKRKRKK) resulted in near complete loss of inhibitory activity in either assay (~20% prothrombinase inhibition with 225μM peptide). Studies with single alanine substitution peptides revealed that all three uncharged residues contribute to the inhibitory activity. In contrast to their functional significance, LIKTKRKRKK and AAKAKRKRKK bound the FVa AR equivalently (Kd= 6.0nM and 5.9nM, respectively). Therefore, the basic residues are responsible for AR binding, while a second step, mediated by L252-T255, is necessary for inhibitory activity. FXa binds FVa near heavy chain residue R506, BR peptides compete with FXa for FVa binding, and TFPIα is a weaker inhibitor of FVa containing the R506Q substitution. Based on these observations, we investigated the possibility that the BR uncharged residues bind near R506. Consistent with this hypothesis, TFPI-AAKA was an equivalent inhibitor of prothrombinase containing FVa or FVa-R506Q, while both TFPIα and TFPI-R107A were weaker inhibitors of FVa-R506Q prothrombinase.

Conclusions: These results collectively demonstrate that inhibition of prothrombinase by physiological low-nanomolar concentrations of TFPIα requires at least three interactions: (1) high-affinity binding of the basic residues of the TFPIα BR to the FVa AR; (2) low-affinity binding of residues L252-T255 to the FVa heavy chain, near R506, likely blocking FXa binding; and (3) high-affinity binding of K2 to the FXa active site. Loss of any of these interactions results in a substantial reduction in inhibitory activity.

Disclosures

Mast: Novo Nordisk: Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.