Novel Rationally Developed Substrates for Factor IXa Demonstrate Active Site Differentiation between Factor IXa and Factor IXa-Padua

Background: The goal of achieving therapeutically relevant levels of FIX activity in hemophilia B (HB) has been enhanced with the use of the hyperactive factor IX variant R338L (FIX-Padua). FIX-Padua has between 4 to 9-fold higher specific activity than wild-type (wt-) FIX, although the exact mechanism of enhanced activity is still unknown. In particular, characterizing the active site differences between wt-FIXa and FIXa^R338L has been significantly limited by the lack of adequate small peptidyl substrates for FIXa. FIXa demonstrates very weak amidolytic activity with respect to existing substrates (Basavaraj et al., 2020), particularly in the absence of ethylene glycol or other alcohols. These conditions can potentially create non-physiological and detrimental effects on both the FIXa active site and intrinsic Xase complex formation, stability, and function. Given that FIX^R338L is the transgene in two FDA-approved HB gene therapy products, understanding the mechanism of its enhanced activity is critical. To this end, we rationally designed novel, efficient FIXa peptidyl substrates that do not require alcohols. These peptides take advantage of the crystallographic observation that P' side interactions (Johnson et al., 2010) are important for FIXa substrate recognition. We then employed these peptides to study the active site function of FIXa^R338L.

Methods: We synthesized a series of peptides incorporating various residues from natural FIXa substrates on both the P and P’ sides of the scissile bond. Typical serine protease peptidyl substrates contain only P side residues, with a fluorophore on the carboxy-terminal side that is liberated after hydrolysis to enable readout. Since our peptides have residues on either side of the scissile bond, we incorporated either MCA/Lys(DNP) or ACC/Lys(DNP) FRET pairs on the amino and carboxy-terminals, respectively, with MCA or ACC serving as the fluorophore and Lys(DNP) functioning as the quencher. Following cleavage by the enzyme, spatial separation from the quencher occurs, rendering the fluorophore available for excitation. Multiple peptide sequences were characterized to evaluate FIXa’s ability to hydrolyze them. We then performed detailed kinetic characterization of the most promising sequences and evaluated them for protease selectivity. Finally, we performed kinetic studies with FIXa^R338L to determine if there are measurable active site differences between the wild-type form and the variant.

Results: We identified a 9-amino acid candidate peptide that had up to 100-fold enhanced k_cat and higher affinity for the enzyme compared to existing FIXa peptidyl substrates. This peptide incorporated 3 residues on the P side and 6 residues on the P’ side of the scissile bond. Importantly, this enhanced activity was seen without the need for ethylene glycol, and addition of ethylene glycol had no significant effect on amidolytic activity. The FRET pair employing ACC as the fluorophore showed dramatically greater fluorescence and enhanced solubility compared to MCA. To ensure that this was indeed an active site probe, we confirmed that active site blocked FIXa (FIXa-DEGR) did not hydrolyze the substrate. Further, we demonstrated that the active site serine protease inhibitor benzamidine inhibited FIXa competitively with respect to our novel substrate (K_i = 2 mM), consistent with the kinetics of other active site probes. Surprisingly, FIXa^R338L was considerably less efficient at hydrolyzing the substrate compared to wt-FIXa while still exhibiting similar substrate affinity (K_M = 35.9 – 39.4 uM) but 2 to 6-fold slower turnover (72.5 vs 8.9 RFU/s), which could not be rescued by imparting FVIIIa cofactor function with either FVIIIa, emicizumab, or substrate binding. Finally, we redemonstrated that previous observations with both wt-FIXa and FIXa^R338L are inhibited by antithrombin identically.

Conclusions: A novel active site probe for FIXa demonstrates that the active site of FIXa^R338L is significantly different from that of wt-FIXa. This does not mean that FIXa^R338L is less functional, but rather implies that understanding the different activity of FIXa^R338L will require understanding how a remote mutation affects the enzyme’s active site.