Fig 2: Knockout of ARID1A downregulates the mevalonate pathway. Top scoring canonical pathways associated with ARID1A knockout in OVCA429 are shown (IPA, p < 0.005 by Fisher's exact test right-tailed). Black bars: negative log p values for each canonical pathway. The ratio of annotated proteins that significantly changed in level in ARID1A knockout to total identified proteins for a given canonical pathway is shown above each bar. Red arrows: canonical pathways that include the mevalonate pathway and are also significantly enriched among proteins that changed upon ARID1A induction. Refer to Fig. 4.

Fig 3: ARID1A inactivation causes defective telomere cohesion. a, b Representative images of prometaphase chromosome spreads (a) and quantification of distance between distal ends of sister chromatids (b) enriched by mitotic shake-off from parental and ARID1A knockout RMG1 cells. c–e Representative images of chromosome spreads (c) and quantification of distance between distal ends of sister chromatids (d) enriched by colcemid treatment from parental and ARID1A knockout RMG1 cells, and ARID1A mutated TOV21G cells. And quantification of distance between distal ends of sister chromatids enriched by colcemid treatment from the indicated clear cell ovarian cancer cell lines or primary cultures highlighted in red (e). f, g Representative images of telomere fluorescent in situ hybridization (f) and quantification of mitotic telomere signal loss (g) in parental and ARID1A knockout RMG1 cells. h Quantification of mitotic telomere signal loss in the indicated clear cell ovarian cancer cell lines. i, j Representative images of telomere fluorescent in situ hybridization (i) and quantification of mitotic telomere signal loss (j) in cells isolated from normal mouse ovary and Arid1a−/−/Pik3caH1047R genetic clear cell ovarian tumors respectively. n = 3 independent experiments unless otherwise stated. Data represent mean ± s.e.m. P values were calculated using a two-tailed t test except in 1e and 1 h by multilevel mixed-effects models

Fig 4: The expression levels of ARID1A were significantly increased in α-OG treated condition. (A) Quantification result from RT-PCR analysis to measure the gene expression levels of six candidate genes (ARID1A, HDAC4, PKD1, β-CATENIN, SMARCA2, and PBRM1) in the presence or absence of α-OG. *P < 0.05 (two-sided Student’s t-test). (B) Representative western blot show the levels of candidates (ARID1A, HDAC4, PKD1, β-CATENIN, SMARCA2, and PBRM1) and β-actin. (C) DNMT and (D) TET activity were measured in response to α-OG treatment. *P < 0.05 (two-sided Student’s t-test). (E) Representative Immunohistochemical stain showed ARID1A expression in bladder tissues obtained from IC patients (IC) and healthy donors (C). (F) A diagram showing the potential mechanism that redundant α-OG in urine may regulate bladder epithelial cells and contribute to suppression of cell proliferation.

Fig 5: Reciprocal regulation between BAF and MLL4 on active enhancers during adipogenesis.Mll3‒/‒Mll4f/f or Smarcb1f/f brown preadipocytes were infected with adenoviruses expressing GFP or Cre, followed by adipogenesis assays. Cells were collected before (D-3) and during (D2) adipogenesis for ChIP-Seq, ATAC-Seq and Western blot analyses. a, b Decreased SMARCA4 and SS18 binding and chromatin accessibility on adipogenic enhancers in Mll4 KO (Cre) cells during adipogenesis. Adipogenic enhancers were defined as active enhancers bound by either C/EBPβ, C/EBPα, or PPARγ at D2 of adipogenesis. Published ChIP-Seq data sets for C/EBPβ, C/EBPα, and PPARγ were used (GSE74189)18. Heat maps around SMARCA4+ SS18+ adipogenic enhancers (a) and average profiles around SMARCA4+ SS18+ MLL4+ adipogenic enhancers (b) are shown. Adipogenic enhancers shown in the heat maps were ranked by the binding intensity of SMARCA4 at the center in D2 control (GFP) cells. c–f Deletion of Smarcb1 reduces binding of BAF subunits (SMARCA4, SS18, and ARID1A) as well as chromatin opening, and impairs MLL4 binding and activation of adipogenic enhancers. Adipogenic enhancers at D2 of adipogenesis were re-defined using ChIP-Seq data sets obtained in Smarcb1f/f cells. c Heat maps for ChIP-Seq of BAF subunits (SMARCA4, SS18, and ARID1A) and ATAC-Seq around the center of SMARCA4+ SS18+ MLL4+ adipogenic enhancers. d Heat maps for ChIP-Seq of MLL4, CBP, H3K4me1 and H3K27ac on adipogenic enhancers. Adipogenic enhancers shown in the heat maps were ranked by the intensity of SMARCA4 at the center in D2 control (GFP) cells. e Fold changes of intensities between control (GFP) and Smarcb1 KO (Cre) cells on adipogenic enhancers (n = 3530) are shown in box plots. The box represents the first and third quartiles with the horizontal line showing the median, and whiskers indicating minimum and maximum in all box plots. Outliers were not included. Statistical significance levels are as follows (Wilcoxon signed rank test, one-sided): SMARCA4 (p = 0); SS18 (p = 0); ARID1A (p = 6.3E-101); ATAC (p = 0); MLL4 (p = 4.5E-281); CBP (p = 0); H3K4me1 (p = 0); H3K27ac (p = 0). Exact p-values cannot be computed for SMARCA4, SS18, ATAC, CBP, H3K4me1 and H3K27ac due to ties in the ranks. f Western blot analyses of H3K4me1 and H3K27ac in control and Smarcb1 KO cells during adipogenesis. Total histone H3 was used as a loading control. n = 3 biological replicates.