Furilisin: a variant of subtilisin BPN' engineered for cleaving tribasic substrates.

The serine protease, subtilisin BPN', was engineered to cleave proteins after tribasic sequences in a manner that resembles the substrate specificity of furin, one of the mammalian subtilisin homologs that processes prohormones. As a starting point we used a double mutant of subtilisin BPN' (N62D/ G166D) that showed substantial preference for cleaving after sequences having consecutive dibasic residues (namely, at the P1 and P2 substrate positions) [Ballinger et al. (1995) Biochemistry 34, 13312-13319]. Additional specificity for basic residues was engineered at the P4 position by introducing subtilisin-to-furin substitutions at three hydrophobic residues that composed the S4 subsite (Y104, I107, and L126). Initial attempts to incorporate a Y104D or I107E mutation or the Y104D/I107E double mutation into the dibasic specific enzyme failed to generate the processed enzyme. The problem was traced to the inability of the mutant prosubtilisins to process themselves and fold correctly. Replacing the natural processing site sequence (AHAY) with a good furin substrate sequence (RHKR) resulted in expression of the triple subtilisin mutant (N62D/Y104D/G166D) we call "furilisin". Furilisin hydrolyzes synthetic tribasic substrates (succinyl-RAKR-pNA or succinyl-KAKR-pNA) with high catalytic efficiency (kcat/K(m) > 3 x 10(5) M-1 s-1) and discriminates in favor of Arg versus Ala at the P4 position by a factor of 360. The overall specificity change versus the wild-type enzyme was dramatic. For example, succinyl-RAKR-pNA was cleaved approximately 60000 times faster than succinyl-AAPF-pNA, a good substrate for wild-type subtilisin. Similarly, furilisin was inhibited (K1* = 29 nM) by a variant of the turkey ovomucoid third domain inhibitor that contained an engineered furin substrate site (RCKR decreases) [Lu et al. (1993) J. Biol. Chem. 268, 14583-14585] and not by one having a good wild-type subtilisin substrate sequence (ACTL decreases). Interestingly, the extreme changes in substrate specificity resulted from substantial synergy between the engineered subsites. These studies provide a basic example of how to manipulate substrate specificity in a modular fashion, thereby creating an engineered-enzyme that may be useful as a protein processing tool.