Restricted HIV-1 escape with miR-based shRNAs targeting mutliple conserved genomic sequences

Abstract

Lentiviral delivery of monospecific short hairpin RNAs (shRNA) can stably suppress HIV-1 infection in cell lines and primary cells. However, given the requirement of sequence homology for RNAi-mediated gene silencing, viral heterogeneity is a major hurdle for its therapeutic use against HIV-1 as even a single nucleotide substitution, especially in the central target region (regions 8-12) can completely abolish silencing. Therefore, simultaneously targeting multiple highly conserved viral sequences is thought to be the efficient way of safeguarding against viral propensity for escape from RNAi. As RNAi is known to induce the selection of escape variants with nucleotide changes in the target sequence, we screened conserved sequences in naturally occurring HIV isolates, since these are likely the ones with minimal loss in viral fitness. Of the 875 naturally occurring HIV isolates listed in the Los Alamos database, we selected two highly conserved 19 nucleotide sequences in the vif and tat genes with minimal/wobble nucleotide variations at positions 9, 10 and 11 of the target sequence to retain efficient siRNA binding even upon mutation. We tested efficacy of silencing the selected targets on HIV replication and analyzed whether combinatorial targeting of Vif and Tat would prevent viral escape. For this, we coexpressed these sequences as precursor microRNA (miRNA) stem-loop structures using a vector expressing pol II driven miR-155 miRNA backbone that allows efficient polycistronic expression of multiple miRNAs. We have confirmed siRNA expression by northern blots and successful inhibition of HIV by cotransfecting 293T cells with NL4-3, a molecular clone of HIV with the miR construct. Furthermore, no escape variants were detected upon culture. To test the effect of positional mutations in the target sequence on loss of silencing, we generated artificial mRNA targets expressing the target sequences in fusion with luciferase mRNA as an indicator. The mRNA targets were based on sequences from naturally occurring HIV variants reported in Los Alamos Database. Naturally variants with changes at the 9th position, either G9T in Tat or G9A in Vif were no longer responsive to silencing, but no natural variant has been reported with changes at both these positions. A variant of Vif, A11G was still amenable to inhibition, due to wobble base pairing of the variant residue with the siRNA strand as confirmed by the loss of inhibition observed with A11C that abolishes wobble binding. Interestingly, no mutations at position 10 were detected in any isolate. Taken together, our results suggest that bispecific miRNA targeting the highly conserved tat and vif sequences can completely inhibit all known isolates of HIV and prevent generation of escape variants since no natural isolate exists with a propensity to harbor all the possible mutations.