The SWI/SNF chromatin remodelling complex is required for maintenance of lineage specific enhancers

Abstract

Genes encoding subunits of SWI/SNF (BAF) chromatin remodelling complexes are collectively altered in over 20% of human malignancies, but the mechanisms by which these complexes alter chromatin to modulate transcription and cell fate are poorly understood. Utilizing mouse embryonic fibroblast and cancer cell line models, here we show via ChIP-seq and biochemical assays that SWI/SNF complexes are preferentially targeted to distal lineage specific enhancers and interact with p300 to modulate histone H3 lysine 27 acetylation. We identify a greater requirement for SWI/SNF at typical enhancers than at most super-enhancers and at enhancers in untranscribed regions than in transcribed regions. Our data further demonstrate that SWI/SNF-dependent distal enhancers are essential for controlling expression of genes linked to developmental processes. Our findings thus establish SWI/SNF complexes as regulators of the enhancer landscape and provide insight into the roles of SWI/SNF in cellular fate control.

Figure 1. Deletion of SWI/SNF subunits results in H3K27ac loss and enhancer deactivation.

(a) Western blots of selected factors, histones and histone modifications in control MEFs (-CRE) and MEFs with CRE-inactivated SWI/SNF (Arid1a, Smarca4, Smarcb1) or Polycomb (Ezh2) subunits. (b). H3K27ac signal at promoters and enhancers in Smarcb1-deficient or Smarca4-deficient versus wild-type MEFs. (c). Representative screenshot showing localization of subunits SMARCC1 and SMARCA4 and histone marks H3K4me3, H3K4me1 and H3K27Ac in wild-type, Smarcb1-deficient and Smarca4-deficient MEFs showing lost enhancers. (d) Log-fold changes of histone modifications H3K4me1 versus H3K27ac at promoters and enhancers in Smarcb1-deficient or Smarca4-deficient relative to wild-type MEFs.

Figure 2. SWI/SNF binds most cis-regulatory elements in wild-type but typical enhancers are most sensitive to loss of SWI/SNF subunits.

(a) Histogram of average SMARCA4 and SMARCC1 (SWI/SNF) ChIP-seq enrichment over input at promoters and enhancers in wild-type MEFs. (b) Boxplots for SWI/SNF ChIP-seq enrichment over input at promoters and enhancers in wild-type and Smarcb1-deficient MEFs. The boxes indicate first, second and third quartiles, and whiskers show 1.5 × interquartile range below and above the first and third quartile, respectively. Two sided t-test P values are shown. (c). Fold changes of H3K27ac versus SWI/SNF ChIP-seq signal at promoters and enhancers in Smarcb1-deficient relative to wild-type MEFs. (d) Representative screenshot depicting clusters of enhancers with different sensitivities to Smarcb1 or Smarca4 loss (red highlights high sensitivity, blue highlights low sensitivity). (e). Boxplots for SWI/SNF and H3K27ac signal fold changes upon Smarcb1 loss at promoters, enhancers, and enhancers classified based on super-enhancers (SE) and transcription (PolII enrichment). The boxes indicate first, second and third quartiles, and whiskers show 1.5 × interquartile range below and above the first and third quartile respectively. Two sided t-test P values are shown.

Figure 3. The SWI/SNF complex physically interacts with enzymes catalysing histone acetylation at enhancers.

(a) Western blots of selected factors, SMARCB1, enhancer-specific binding factors, histone H3, and different modifications in SMARCB1-deficient rhabdoid cell lines, G401 and BT16. Parental lines (P) and 0, 5 or 10 days after Doxycycline (Dox) induced SMARCB1 re-expression. (b). Immunoprecipitation (IP) of the SWI/SNF complex subunit SMARCC1 in wild-type and Smarca4- or Smarcb1-deficient MEFs after CRE-inactivation. Immunoblotted for P300, SMARCA4, SMARCB1, SMARCC1 and EZH2. IgG pulldown is shown as control. (c) Immunoprecipitation (IP) of the SWI/SNF complex subunits SMARCC1, SMARCB1 or SMARCA4 from the nuclear extracts of G401 cell line with or without Doxycycline (Dox) induced SMARCB1 re-expression. Immunoblotted for P300, SMARCA4, SMARCB1, and EZH2. IgG is shown as control. (d) In vitro HAT assay with Acetyl-CoA as donor group and detection of histone acetylation by immunoblot using H3K27ac, H3K9ac, and H3 antibodies. After pulling down the SWI/SNF complex with SMARCC1 antibody, acetylation activity was measured in the presence or absence of SMARCB1 expression. Ponceau staining was shown as loading control for substrate.

Figure 4. SWI/SNF subunit deletion results in downregulation of specific developmental targets.

(a). Number of TF motif matches within lost enhancers relative to an expectation based on numbers in stable enhancers. The strongest enrichment/depletion is seen for motifs similar to AP-1, ETS and CTCF motifs, which are highlighted. (b). Correlation of nearby gene expression (5–100 kb) changes with H3K27ac signal changes at enhancers upon Smarcb1 or Smarca4 deletion. The boxes indicate first, second, and third quartiles, and whiskers show 1.5 × interquartile range below and above the first and third quartile, respectively. Two sided t-test P values are shown. (c) Selected GO terms enriched in downregulated genes. Hypergeometric test P values are shown after Benjamini–Hochberg correction for multiple-hypothesis testing.