Pre-mRNA Splicing and Gene Density are Significant Determinants of Lentiviral Integration

Abstract

Nuclear speckles harbor proteins important for transcription and pre-mRNA splicing. HIV-1 preintegration complexes localize to speckles in a capsid (CA)-CPSF6 dependent manner and preferentially target speckle-associated domains (SPADs) for integration. Because the CPSF6-CA interaction is also crucial for integration into gene-dense regions and spliced genes, we correlated the extents of SPAD targeting with gene-dense regions and spliced genes. Our bioinformatic analysis mapped SPADs to gene-dense regions. While the average gene density of the human genome is 9 genes/Mb, nearly 98% of SPADs mapped to regions with gene density > 9 genes/Mb. Our analysis revealed that human chr17 and 19 contained the highest number of genes, intron-containing genes, SPADs, pol II sites, and H3K36me3 sites per Mb. Analysis of HIV-1 integration sites revealed enrichment on chr17 and 19 in wildtype but not in CPSF6 knockout cells. Sites of LEDGF/p75 occupancy and lamina-associated domains were not enriched on human chr17 and 19. Non-primate lentiviruses EIAV, FIV, BIV only marginally targeted SPADs for integration in human cells. As expected, these viruses did not preferentially target gene-dense human chr17 and 19. We next asked whether these lentiviruses targeted gene-dense regions in their natural hosts. As control, the primate lentivirus SIV preferentially targeted gene-dense chr17 and 19 in human cells and gene-dense chr16 and 19 in rhesus cells. Further, the ovine lentivirus MVV did not preferentially target human chr17 and 19 but targeted gene-dense chr11 in the sheep genome. Plausibly, species-specific host factors help guide primate versus non-primate lentiviruses to target gene-dense chromosomal regions for integration. We are currently determining non-primate lentiviral integration sites in additional native host cells. We next asked whether MVV targets spliced genes in sheep. We determined a high correlation (R2=0.6) between integration per kb and the number of sheep introns. By contrast, this correlation was low (R2=0.2) for MVV in human cells due to the lack of SPAD targeting. SPAD-associated human genes contained 0.4 introns per kb, whereas SPAD non-associated genes contained 0.1 introns per kb (P<2E-16), suggesting that SPAD- associated genes are highly spliced. These data suggest a conserved role for pre-mRNA splicing in both primate and non-primate lentiviral replication and indicates that the viruses utilize species-specific factors for trafficking to their preferred integration sites. Acknowledgment: Harvard University Center for AIDS Research (HU CFAR NIH/NIAID fund 5P30AI060354-17).