Abstract
After viral entry and reverse transcription, HIV-1 proviruses that fail to integrate are epigenetically silenced, but the underlying mechanism has remained unclear. Using a genome-wide CRISPR/Cas9 knockout screen, we identified the host SMC5/6 complex as essential for this epigenetic silencing. We show that SMC5/6 binds to and then SUMOylates unintegrated chromatinized HIV-1 DNA. Inhibition of SUMOylation, either by point mutagenesis of the SMC5/6 component NSMCE2—a SUMO E3 ligase—or using the SUMOylation inhibitor TAK-981, prevents epigenetic silencing, enables transcription from unintegrated HIV-1 DNA and rescues the replication of integrase-deficient HIV-1. Finally, we show that blocking SMC5/6 complex expression, or inhibiting its SUMOylation activity, suppresses the establishment of latent HIV-1 infections in both CD4+ T cell lines and primary human T cells. Collectively, our data show that the SMC5/6 complex plays a direct role in mediating the establishment of HIV-1 latency by epigenetically silencing integration-competent HIV-1 proviruses before integration.
Main
Integration of proviral DNA into the host cell genome is a defining feature of the retroviral life cycle that is essential for proviral transcription and replication1,2. Integrase (IN) inhibitors potently inhibit HIV-1 replication3. In the absence of functional IN, unintegrated HIV-1 proviruses accumulate repressive epigenetic marks, including trimethylation of lysine 9 on histone H3 (H3K9me3), and are depleted of activating marks, such as H3 acetylation (H3Ac)4,5. While the epigenetic silencing of transcription from unintegrated HIV-1 DNA probably represents a host defence against foreign DNA, the underlying mechanisms and cellular factors that mediate this effect remain incompletely defined6,7.
In murine leukaemia virus (MLV), a genomic screen identified components of the human silencing hub (HUSH) complex, as well as the DNA-binding protein NP220, as critical for unintegrated MLV DNA silencing8. However, subsequent work9,10 failed to detect any role for the HUSH complex or NP220 in silencing unintegrated HIV-1. More recently, a screen of 1,217 human genes found to be downregulated by the HIV-1 Vpr protein identified a component of the structural maintenance of chromosome (SMC) 5/6 complex, SMC5/6 complex localization factor 2 (SLF2), as critical for unintegrated HIV-1 DNA silencing. This screen also showed that six other components of the SMC5/6 complex, including SMC5 and 6 as well as the four SMC5/6 associated proteins non-structural maintenance of chromosomes element 1 through 4 (NSMCE1–4), but not the SMC5/6 associated factor SLF1, were also critical for the epigenetic silencing of unintegrated HIV-1 DNA9. Of note, the SMC5/6 complex was previously shown to be degraded by the hepatitis B virus (HBV) non-structural protein HBX and, in the absence of HBX, episomal HBV DNA is also epigenetically silenced11,12. Thus, the SMC5/6 complex not only participates in chromosomal replication, recombination and repair13 but can also silence invasive viral DNA. Here we sought to determine whether the SMC5/6 complex mediates the establishment of latent HIV-1 infections.
Results
A genomic screen for factors that silence HIV-1 proviruses
To identify factors that transcriptionally silence unintegrated HIV-1 DNA, we performed a genome-wide CRISPR/Cas9 knockout screen14 in the human CD4+ T cell line CEM-SS. We transduced a CEM-SS subclone that expresses Streptococcus pyogenes Cas9 with a lentiviral library expressing ~80,000 single guide RNAs (sgRNAs) targeting 19,114 human genes15. Seven days later, we infected these cells with IN− NL-GFPΔEnv10, an HIV-1 derivative harbouring a deletion in env, the inactivating D64V mutation16 in IN, and the green fluorescent protein (GFP) open reading frame in place of nef. This virus retains intact copies of the other six HIV-1 genes, including vpr. At 48 h post infection (hpi), GFP+ cells were collected by fluorescence activated cell sorting (FACS), the sgRNAs recovered by (polymerase chain reaction) PCR then cloned into the same lentiviral vector. After three rounds of selection for GFP+ cells, the sgRNAs were sequenced and analysed for enrichment compared to the starting sgRNA library. As shown in the volcano plot in Fig. 1a, we identified 9 genes that were enriched >16-fold and had a P value <0.0005. These included 3 cell surface receptors (LY9, OR52N2 and SSTR2) and 1 motor protein (MYO1B) which were not further analysed. This analysis also recovered 4 of the 8 known components of the SMC5/6 complex, namely SMC5, SMC6, SLF1 and NSMCE3, as well as the DNA repair protein SWI5 (Fig. 1a)….
