The dengue virus (DEN) protease is a heterodimer of nonstructural protein 2B (NS2B) and NS3. The amino-terminal third of NS3 is a trypsin-like serine protease. Previously, we performed a mutagenesis study of some of the residues in NS3 of DEN type 2 (DEN2) predicted to be involved in substrate binding, in order to characterize their requirement for protease activity and to identify mutants with partial cleavage defects. In all, 46 mutations were analyzed for their effect on intramolecular cleavage at the NS2B-NS3 junction and intermolecular cleavage at the NS4B-NS5 junction in vitro. Subsequently, the crystal structure of the DEN2 NS3 protein was reported by Padmanabhan and Murthy, and the observed effects of most of our mutations were consistent with their expected perturbations of the protease structure. We are now collaborating with these investigators to express, purify, and crystallize various mutant NS3 proteins, to see if the structures are indeed perturbed as predicted. Work has also been done to introduce some of these NS3 mutations back into the DEN2 genome, in an effort to confirm the observed effect on protease function and possibly to attenuate the virus. Thirteen mutants that had wild-type or nearly wild-type activity in vitro were analyzed by reverse engineering them into a full-length DEN2 cDNA clone and testing mutant RNA transcripts for infectivity: 5 were non-infectious, 7 were recovered as virus, and in 1 case the recovered virus was a pseudorevertant. The phenotypes of these viruses are currently being investigated in growth curves and by biochemical analyses of protease processing in infected cells. Among this set of 13 mutations, the 5 that were lethal had less activity in vitro, suggesting that DEN is not able to tolerate much reduction in protease activity. In an unrelated set of experiments, we have observed that transcripts made from DEN2 cDNA clones with short deletions at the 3' end are infectious. Constructs missing up to 7 nt are viable, while deletion of 8 or more nt is lethal. In all but one case, the 3' end sequences of the recovered viruses are wild-type, suggesting that the virus has some mechanism to repair incorrect 3' end structures. Virus recovered from the 7 nt deletion has a 3' end sequence that differs from wild-type at 1.5 positions, and this mutant virus appears to grow more slowly than wild-type DEN2. Work is in progress to further investigate the phenotype of this mutant, and to attempt to understand the mechanism involoved in the 3' end repair.
