The active HIV-1 protease is a homodimer made up of monomers containing 99 amino acid residues. All drugs directed against the protease have targeted the substrate-binding site of the dimer. Although at least six drugs have been developed that binding tightly to the active site, one unfortunate consequence of targeting a single drug binding site on the protease has been the emergence of viral strains which carry multi-drug resistant mutations on their protease constructs. For this reason, there is considerable interest in identifying other protease sites that are suitable drug targets. Although the protease monomer is completely inactive, it is folded and therefore contains numerous potential alternative drug-binding sites, making it an attractive anti-HIV target. In order to identify drug-binding sites of the folded monomer, we have been made a series HIV protease constructs with the goal of obtaining a stable folded monomer in solution which would be suitable for studies of drug interactions using NMR spectroscopy. We have shown that N- and C-terminal deletion mutations, as well as point mutations R87K, D29N and T26A all shift the monomer-dimer equilibrium in favor of the folded monomer. However, all these constructs are prone to aggregation. In order to alleviate this undesirable property we have designed proteins in which the N- and C-terminal regions are linked by intra-monomer disulfide bonds. In particular, cysteine residues were introduced at positions 2 and at either 97 or 98 of the protease amino acid sequence. A procedure for the efficient preparation an intra-chain linked monomer was developed, and we have shown that the Q2C/L97C monomer construct exhibits a fold similar to that observed for the monomer subunit in the active dimer. It is anticipated that monomeric proteases of this kind will aid in the discovery of novel inhibitors, aimed at binding to the monomer at the dimerization interface. This extends the target area beyond that of current inhibitors, all of which bind across the active-site formed by both monomers in the active dimer. We anticipate that these inhibitors may address problems related to multidrug-resistance invariably observed with current HIV-1 protease drugs. Although hundreds of crystal structures of active HIV-protease bound to potent inhibitors are available, structures of are not available of the protein bound to substrate because of rapid catalysis. The binding of the fully active HIV-1 protease to peptide substrate IV was studied in solution using NMR spectroscopy. Unfortunately we were not able to follow the binding of the substrate to the active protease, because many of the protease amide signals were missing from the HSQC spectrum of the protease, recorded after adding the substrate. This result was ascribed to severe broadening of the protease signals due to chemical exchange of the protease with various products of the catalytic reaction. In contrast, an inactive protease structural homologue, PRD25N, binds to the substrate in the slow exchange limit, and yields high quality HSQC spectra over a wide range of substrate concentrations. Following sequential assignment of free and substrate bound PRD25N, substrate titration HSQC experiments and 15N-edited NOESY experiments were recorded at 20?C, pH 5.8. Analysis of these spectra yielded an equilibrium dissociation constant, KD, of 0.27?0.05 mM for the PRD25N/substrate complex, and upper limits of the association and dissociation rate constants of 2x104 M-1s-1 and 10 s-1, respectively. This association rate constant is not in the diffusion limit, which indicates that the association is controlled by a rare event, such as full opening of the protease flaps. In addition to these results on binding kinetics, information about internal protein dynamics of the substrate-bound PRD25N was obtained from the analysis of 15N relaxation experiments. In the case of PRD25N bound to the peptide substrate, a larger number of flap residues were found to be flexible on the millesecond-microsecond timescale than has been observed for the protease bound to inhibitors which are significantly smaller than the peptide substrate. Because the protease D25N construct is stable as a free homodimer in solution, we were able to carry out an extensive series of transverse relaxation dispersion measurements. This data enabled us to compare the slow (millesecond-microsecond) dynamics of the free and inhibitor bound proteins. The results obtained clearly confirmed the presence of such motions in the N- and C-terminal regions of the molecule, which contain the auto-catalytic loop and which form the main dimer interface. The flexibility observed in this region of the molecule allows one to rationalize how the N- and C-terminal strands of the protease, which form a highly ordered beta sheet in crystal structures, can separate and be available for cleavage. Cleavage at these the N- and C- terminal sites of the protease is required for maturation of the Gal-Pol polyprotein, which yields functional HIV proteins, including the protease itself. If cleavage is blocked the HIV cannot mature and reproduce.
Showing the most recent 10 out of 14 publications
