Background: HIV-1 RNA is spliced to produce more than 40 different transcripts. Most previous research on HIV-1 splicing used model single exon systems and PCR gel quantification. Our straightforward assay quantifies HIV-1 RNA splicing patterns in the context of viral infection using Illumina paired-end deep sequencing and PrimerID. This system can be used to assess changes in splicing profiles by mutations that affect splice regulatory elements and to assess the differential effect of APOBEC3G-mediated mutagenesis on the 4 kb partially spliced mRNA population compared to the 1.8 kb mRNA population.
Methods: PrimerID is a random sequence tag in the cDNA primer, which is incorporated into a cDNA and all PCR products subsequently made from that tagged cDNA. Each cDNA gets a unique PrimerID and after sequencing, the PCR products can be sorted by their PrimerIDs. Each occurrence of a PrimerID is counted exactly once, and skewing in the PCR steps is filtered out. Additionally, a consensus sequence can be made from multiple reads with the same PrimerID, making it possible to filter out PCR and sequencing errors. PrimerID-tagged reverse primers specific for either the 1.8kb or 4kb splice classes and a common forward primer were used to tag the 40+ HIV-1 spliced mRNAs. Following amplification and Illumina paired-end deep sequencing, mRNAs were quantified according to splice type and collapsed into consensus sequences. Mutations in a stem near A1 and in the ESE(GAR) sequence located between A5 and D4 were correlated with splicing outcomes.
Results: Mutations to a highly conserved stem-loop containing splice acceptor A1 caused a 10-fold reduction in A1 usage, resulting in reduced vif transcript levels. The ESE(GAR) splice control element between A5 and D4 was analyzed for APOBEC3G mutations. Reduced vif transcripts led to increased C=>T mutations, which were unevenly skewed towards the 1.8kb splice class. T=>C mutations, though less frequent, skewed towards the 4kb splice class. As this panel of mutations in the ESE(GAR) region illustrates, C=>T and T=>C transitions differentially affect upstream (A5) splicing and/or downstream (D4-A7) splicing.
Conclusions: Combining PrimerID with paired-end deep sequencing allows mutations in a splice control element to be screened and correlated to real-time changes in HIV-1 splicing. We identified mutations near A1 and between A5 and D4 that affect utilization of the proximal splice sites.