A proteomics study identifying interactors of the FSHD2 gene product SMCHD1 reveals RUVBL1-dependent DUX4 repression

Abstract

Structural Maintenance of Chromosomes Hinge Domain Containing 1 (SMCHD1) is a chromatin repressor, which is mutated in > 95% of Facioscapulohumeral dystrophy (FSHD) type 2 cases. In FSHD2, SMCHD1 mutations ultimately result in the presence of the cleavage stage transcription factor DUX4 in muscle cells due to a failure in epigenetic repression of the D4Z4 macrosatellite repeat on chromosome 4q, which contains the DUX4 locus. While binding of SMCHD1 to D4Z4 and its necessity to maintain a repressive D4Z4 chromatin structure in somatic cells are well documented, it is unclear how SMCHD1 is recruited to D4Z4, and how it exerts its repressive properties on chromatin. Here, we employ a quantitative proteomics approach to identify and characterize novel SMCHD1 interacting proteins, and assess their functionality in D4Z4 repression. We identify 28 robust SMCHD1 nuclear interactors, of which 12 are present in D4Z4 chromatin of myocytes. We demonstrate that loss of one of these SMCHD1 interacting proteins, RuvB-like 1 (RUVBL1), further derepresses DUX4 in FSHD myocytes. We also confirm the interaction of SMCHD1 with EZH inhibitory protein (EZHIP), a protein which prevents global H3K27me3 deposition by the Polycomb repressive complex PRC2, providing novel insights into the potential function of SMCHD1 in the repression of DUX4 in the early stages of embryogenesis. The SMCHD1 interactome outlined herein can thus provide further direction into research on the potential function of SMCHD1 at genomic loci where SMCHD1 is known to act, such as D4Z4 repeats, the inactive X chromosome, autosomal gene clusters, imprinted loci and telomeres.

Figure 1

SILAC-MS identifies novel binding partners of SMCHD1. (A): Experimental set-up of the SILAC-MS screen in U2OS cells stably expressing GFP or GFP-SMCHD1. (B): Visual representation of potential SMCHD1 protein interaction partners. Median Log2 transformed Heavy GFP-SMCHD1 SILAC ratios compared to the light GFP-only controls (FW) of 4 replicate sets are plotted on the X-axis, against their corresponding Light GFP-SMCHD1 SILAC ratios compared to the heavy GFP-only controls (RV). Colour coding denotes the number of samples out of 8 replicates in which a protein’s H/L was > 1.5. Green: 8–7/8; Blue: 6/8; Orange: 5/8; Red: 4/8; grey: < 4/8. The identity of nuclear proteins with a median H/L ratio > 1.5 identified in 7 or 8 replicates are indicated. The dotted lines indicate the H/L ratio threshold of 1.5 (Log2 transformed: 0.585). The correlation of the median FW and RV (R-square) was calculated to be 0.7515 (red line). (C): Table representation of the 9 proteins identified in the SILAC-MS analysis with highest confidence. Individual H/L ratios of each replicate samples and peptide counts are listed, as well as the median H/L ratio used in (B). (D): Venn diagram showing overlap between SMCHD1 interaction partners identified in this study (all nuclear proteins with median H/L ratio > 1.5), compared to D4Z4 chromatin components identified by Campbell et al. The 12 overlapping proteins identified are listed on the right and coloured as outlined in (B). Asterisk indicates detection in > 5 replicates of GFP-SMCHD1 SILAC-MS.

Figure 2

Validation Co-IPs confirm the interaction of SMCHD1 with RUVBL1 and EZHIP. (A): Western blot after HA-IP in various cell types. 3xHA-SMCHD1 was ectopically expressed in HEK293T, U2OS and HeLa cells as indicated. Cells were lysed, and HA-tagged proteins were precipitated using HA-agarose beads. Probing the membrane for endogenous RUVBL1 shows Co-IP with SMCHD1 in all cells analysed. (B): Western blot after immunoprecipitation of endogenous RUVBL1 in HEK293T, U2OS and HeLa cells. Upon purification of RUVBL1, endogenous SMCHD1 can be detected in the IP fraction for each cell type. The asterisk denotes the presence of IgG heavy chains present in the sample. (C): Western blot after immunoprecipitation of endogenous RUVBL1 in control myoblasts. Upon purification of RUVBL1, endogenous SMCHD1 can be detected in the IP fraction. (D): HA-IP in HEK293T cells after ectopic expression of 3xFLAG-SMCHD1 and HA-EZHIP. Immunoreactivity for FLAG-tagged SMCHD1 is only detected in the IP fraction in the presence of HA-EZHIP. (E): FLAG-IP in HEK293T cells after ectopic expression of 3xFLAG-SMCHD1 and HA-EZHIP. Immunoreactivity for HA-tagged EZHIP is only detected in the IP fraction in the presence of 3xFLAG-SMCHD1. (F): Western blot analysis after GFP-IP in HEK293T cells shows co-immunoprecipitation of RAD21 with GFP-SMCHD1.

Figure 3

Decreased RUVBL1 levels in FSHD myotubes leads to increased DUX4 expression. (A): Representative western blot showing reduced RUVBL1 protein levels in FSHD myotube samples treated with independent shRNAs targeting RUVBL1 (shRUVBL1-1 and shRUVBL1-2) compared to scrambled control shRNA (shSCR). Tubulin was used as loading control. Percentages indicate levels of RUVBL1 protein normalized to tubulin. (B): RT-qPCR analysis of FSHD myotube samples corresponding to western blot in (A). Expression of DUX4 and DUX4 target gene ZSCAN4 is significantly increased upon RUVBL1 depletion (shRUVBL1-1: N = 8, shRUVBL1-2: N = 6). FSHD1: Triangles, FSHD2: Circles. (C): Confocal immunofluorescent microscopy analysis of FSHD derived myotubes depleted for RUVBL1 or SMCHD1. Top panels show nuclei (DAPI—Blue), DUX4 (Green) and Myosin (Red). Bottom panels show DUX4 only. Scalebar: 25 µm. (D): Violin plot of high content analysis for DUX4 positive nuclei of slides shown in (C). (N = 3, total nuclei counted: shSCR: 55,642; shRUVBL1: 29,721; shSMCHD1-S4: 26,423). (E): Box and whisker plot of myogenic fusion index determined in slides from C by high content analysis. (N = 3). (F): ChIP-qPCR analysis of SMCHD1 occupancy at two D4Z4 loci and negative control locus GAPDH, relative to input. Knockdown of RUVBL1 does not lead to loss of SMCHD1 at D4Z4 in FSHD myotubes. Dotted line indicates enrichment for IgG control. (N = 2). Error bars: SEM. (*P value < 0.05, **P value < 0.01, ***P value < 0.001, ****P value < 0.0001, NS: Not-Significant—Kruskal–Wallis One-Way ANOVA).