The two-component serine protease of flaviviruses such as Dengue virus (DENV) and West Nile virus (WNV) are attractive targets for inhibitor/therapeutic design. Peptide aldehyde inhibitors that bind to the covalently tethered two-component WNV protease (WNVpro) with 50% inhibitory concentration (IC(50)) at sub-micromolar concentrations, bind the equivalent DENV-2 protease (DEN2pro) with IC(50) of micromolar concentrations at best. Conversely, the protease inhibitor aprotinin binds DEN2pro ∼1000-fold more tightly than WNVpro. To investigate the residues that are crucial for binding specificity differences, a binding-site network of hydrogen bonds was transplanted from WNVpro onto DEN2pro. The transplantations were a combination of single, double and triple mutations involving S79D, S83N and S85Q. The mutant DENV proteases, except those involving S85Q, proved to be more efficient enzymes, as measured by their kinetic parameters. The binding affinities of the mutants to peptide inhibitors however showed only marginal improvement. Protein structure modeling suggests that the negatively charged residue cluster, Glu89-Glu92, of the NS2B cofactor may play an important role in determining substrate/inhibitor-binding specificity. These same residues may also explain why aprotinin binds more tightly to DEN2pro than WNVpro. Our results suggest that structure-based inhibitor design experiments need to explicitly consider/include this C-terminal region whose negative charge is conserved across the four DENV serotypes and also among the flavivirus family of proteases.