GENE SILENCING. Epigenetic silencing by the HUSH complex mediates position-effect variegation in human cells

Abstract

Forward genetic screens in Drosophila melanogaster for modifiers of position-effect variegation have revealed the basis of much of our understanding of heterochromatin. We took an analogous approach to identify genes required for epigenetic repression in human cells. A nonlethal forward genetic screen in near-haploid KBM7 cells identified the HUSH (human silencing hub) complex, comprising three poorly characterized proteins, TASOR, MPP8, and periphilin; this complex is absent from Drosophila but is conserved from fish to humans. Loss of HUSH components resulted in decreased H3K9me3 both at endogenous genomic loci and at retroviruses integrated into heterochromatin. Our results suggest that the HUSH complex is recruited to genomic loci rich in H3K9me3, where subsequent recruitment of the methyltransferase SETDB1 is required for further H3K9me3 deposition to maintain transcriptional silencing.

Fig. 1. A haploid genetic screen identifies a requirement for the HUSH complex for epigenetic repression in human cells

(A) Schematic view of the GFP reporter construct (see (14) for further details). (B) Transduction of KBM7 cells with the GFP reporter results in a majority GFP^bright population plus a repressed GFP^dim population. (C) A haploid genetic screen to identify genes required for repression of the GFP reporter. (D) Bubble plot illustrating the hits from the screen. Bubble size is proportional to the number of independent inactivating gene-trap integrations identified (shown in brackets). (E and F) Validation of the screen hits in HeLa cells, using an independent lentiviral reporter (E) and a non-viral reporter containing the phosphoglycerate kinase 1 (PGK) promoter driving GFP delivered by transfection (F). Histograms were gated on GFP⁺ cells. (G) TASOR, MPP8 and periphilin form a complex. Endogenous TASOR, MPP8 and periphilin were immunoprecipitated from KBM7 cells, and the indicated co-immunoprecipitating proteins identified by immunoblot. We were unable to blot for periphilin as the antibody does not recognise its epitope following NP-40 lysis.

Fig. 2. The HUSH complex functions through H3K9me3 via the targeted recruitment of SETDB1

(A) GFP^dim integrations are marked by H3K9me3 but not H3K27me3. H3K9me3 and H3K27me3 levels across the reporter were assessed by ChIP-qPCR in sorted GFP^dim and GFP^bright populations. (B and C) Knockdown of HUSH components results in a loss of H3K9me3 across a GFP^dim reporter as assessed by ChIP-qPCR (B), concomitant with an increase in GFP transcript levels (C). (D) Subcellular fractionation showing that HUSH subunits are found in the chromatin fraction. GAPDH, Rb and HP1α were used to validate the fractionation. (E) V5-tagged HUSH subunits bind to a repressed GFP reporter as assessed by ChIP-qPCR. (F) Co-immunoprecipitation of TASOR and MPP8 with SETDB1. (G) Knockdown of HUSH components results in impaired recruitment of SETDB1 to a GFP^dim reporter, as assessed by ChIP-qPCR.

Fig. 3. The HUSH complex functions at genomic loci rich in H3K9me3

(A) HUSH represses the majority of GFP^dim reporter integrations. Populations of GFP^dim and GFP^bright cells were isolated by FACS and subjected to knockdown of TASOR. (B to D) The HUSH complex acts at genomic loci marked by high levels of H3K9me3. Reporter integration sites were mapped among the GFP^dim and GFP^bright populations (B). Correlating integration sites with ENCODE ChIP-seq peaks for K562 cells showed that GFP^dim integration sites were most enriched in proximity to H3K9me3 (C). Examples of the top “dim genes” and “bright genes” are shown in (D). (E and F) HUSH-mediated repression of a GFP reporter construct targeted to ZNF594. (G) Re-establishment of reporter repression upon reconstitution of a HUSH triple knockout clone.

Fig. 4. The HUSH complex maintains H3K9me3 at endogenous loci and represses viruses integrated into heterochromatin

(A to C) Loss of HUSH results in decreased H3K9me3 at endogenous genomic loci. Global levels of H3K9me3 were measured by ChIP-seq in wild-type HeLa cells and cells lacking HUSH subunits and SETDB1. The CRISPR/Cas9-mediated disruption of TASOR, MPP8 and periphilin resulted in decreased H3K9me3 at shared loci (A). Knockout of SETDB1 resulted in decreased H3K9me3 at over 99% of these loci, with the majority (91%) showing a >3-fold reduction (B). Four example loci are shown in (C). (D) Loss of HUSH resulted in increased expression of the four example genes from (C), as measured by qRT-PCR. (E to H) Proviral repression by the HUSH complex. Schematic view of the HIV-1 LTR-Tat-IRES-GFP virus (E). Infection of Jurkat cells resulted in a range of GFP expression levels (F); the proviruses in the GFP^dim cells were repressed by the HUSH complex (G) through H3K9me3 (H). TNF-α activates transcription from the HIV-1 LTR through the NF-κB pathway (18).