CPSF6-Dependent Targeting of Speckle-Associated Domains Distinguishes Primate from Nonprimate Lentiviral Integration

Abstract

Lentiviral DNA integration favors transcriptionally active chromatin. We previously showed that the interaction of human immunodeficiency virus type 1 (HIV-1) capsid with cleavage and polyadenylation specificity factor 6 (CPSF6) localizes viral preintegration complexes (PICs) to nuclear speckles for integration into transcriptionally active speckle-associated domains (SPADs). In the absence of the capsid-CPSF6 interaction, PICs uncharacteristically accumulate at the nuclear periphery and target heterochromatic lamina-associated domains (LADs) for integration. The integrase-binding protein lens epithelium-derived growth factor (LEDGF)/p75 in contrast to CPSF6 predominantly functions to direct HIV-1 integration to interior regions of transcription units. Though CPSF6 and LEDGF/p75 can reportedly interact with the capsid and integrase proteins of both primate and nonprimate lentiviruses, the extents to which these different viruses target SPADs versus LADs, as well as their dependencies on CPSF6 and LEDGF/p75 for integration targeting, are largely unknown. Here, we mapped 5,489,157 primate and nonprimate lentiviral integration sites in HEK293T and Jurkat T cells as well as derivative cells that were knocked out or knocked down for host factor expression. Despite marked preferences of all lentiviruses to target genes for integration, nonprimate lentiviruses only marginally favored SPADs, with corresponding upticks in LAD-proximal integration. While LEDGF/p75 knockout disrupted the intragenic integration profiles of all lentiviruses similarly, CPSF6 depletion specifically counteracted SPAD integration targeting by primate lentiviruses. CPSF6 correspondingly failed to appreciably interact with nonprimate lentiviral capsids. We conclude that primate lentiviral capsid proteins evolved to interact with CPSF6 to optimize PIC localization for integration into transcriptionally active SPADs. IMPORTANCE Integration is the defining step of the retroviral life cycle and underlies the inability to cure HIV/AIDS through the use of intensified antiviral therapy. The reservoir of latent, replication-competent proviruses that forms early during HIV infection reseeds viremia when patients discontinue medication. HIV cure research is accordingly focused on the factors that guide provirus formation and associated chromatin environments that regulate transcriptional reactivation, and studies of orthologous infectious agents such as nonprimate lentiviruses can inform basic principles of HIV biology. HIV-1 utilizes the integrase-binding protein LEDGF/p75 and the capsid interactor CPSF6 to target speckle-associated domains (SPADs) for integration. However, the extent to which these two host proteins regulate integration of other lentiviruses is largely unknown. Here, we mapped millions of retroviral integration sites in cell lines that were depleted for LEDGF/p75 and/or CPSF6. Our results reveal that primate lentiviruses uniquely target SPADs for integration in a CPSF6-dependent manner.

HIV Capsid and Integration Targeting

“…While ectopic expression of CPSF6 isoform 1 in CPSF6 knockout cells in large part restored integration targeting of gene dense regions, expression of CPSF6 mutant F284A, which is defective for CA binding [ 91 , 92 ], did not [ 25 ]. Although Jurkat T cells depleted for CPSF6 via transient transfection of Cas9-guide RNA complexes revealed more modest reductions in integration targeting of gene-dense regions (from ~21 to ~15 genes/Mb; random = 7.9 genes/Mb), the N74D mutant virus in large part lost the preference to target gene-dense regions (8.7 genes/Mb) in Jurkat T cells [ 26 ].…”

“…In silico calculations placed the random chance of SPAD targeting at 2.8%. Remarkably, ~30% to 40% of HIV-1 integrations occurred within SPADs in primary and transformed cell types [ 12 , 26 ]. Specificity of SPAD integration targeting was addressed using the same tools employed for the gene-dense regions targeting studies.…”

“…Specificity of SPAD integration targeting was addressed using the same tools employed for the gene-dense regions targeting studies. For example, N74D mutant virus disfavored SPADs for integration in HEK293T cells [ 12 ] and largely avoided them (3.7% SPAD targeting) in Jurkat T cells [ 26 ]. While ectopic expression of wild-type CPSF6 restored SPAD integration targeting to CPSF6 knockout cells, expression of the F284A mutant did not [ 12 ].…”

“…Reduction of cellular LEDGF/p75 protein levels via mRNA-directed knockdown [ 20 ] or stable knockout of the PSIP1 gene that encodes for LEDGF/p75 [ 21 , 22 , 23 ] significantly reduced the frequency of intragenic HIV-1 integration targeting. Moreover, the genic profiles of the residual proviruses that formed under these conditions uncharacteristically congregated toward the upstream regions of genes [ 24 , 25 , 26 ]. Consistent with these observations, LEDGF/p75 has been shown to interact with mRNA splicing factors [ 24 , 27 ] and can overcome the transcriptional block imposed by nucleosomes in vitro [ 28 ].…”

Factors that mold the nuclear landscape of HIV-1 integration

“…Depletion of CPSF6 or infection with binding-defective viruses such as N74D and A77V CA mutants resulted in PIC and proviral accumulation in the peripheral region of the nucleus ( 17 , 60–62 , 110 , 155 , 156 ). Concurrent genomic DNA analyses revealed significant upticks in LAD-proximal integration site targeting with parallel reductions in integrations into SPADs ( 17 , 18 , 157 ). HIV-1 PICs and CPSF6 were accordingly seen to colocalize with nuclear speckles in a variety of acutely-infected cell types including primary CD4+ T cells and macrophages ( 18 ).…”