Fig 2: Integrating the direct and indirect roles of BEND3 on chromatin and transcription. (A) Summary of the H3K27me3 CUT&RUN results in WT and Bend3 KO. P-value: hypergeometric tests. (B) Comparison of H3K27me3 loss and differential gene expression in Bend3 KO cells. (C) Comparison of H3K27me3 gain and differential gene expression in Bend3 KO cells. (D) Percentage of statistically significant hypermethylated tiles in the indicated genomic compartments, when comparing Bend3 KO to WT cells. The red dashed line shows the value in intergenic regions, for comparison. (E) Model for the direct and indirect effects of BEND3 on chromatin and gene expression.

Fig 3: Cells lacking BEND3 have increased DNA methylation and decreased DNA hydroxymethylation. (A) Evaluation by a restriction enzyme-based method (LUMA) of the global level of DNA methylation in the indicated cellular contexts. Bend3 KO cells are hypermethylated, and the expression of a BEND3 rescue construct brings DNA methylation back to WT levels. P-value: ANOVA followed by Tukey's post-hoc tests. **P< 0.01; ***P< 0.001; ****P< 0.0001. (B) Quantitation of 5-mC abundance in the cells by LC–MS/MS. P-value: ANOVA followed by Dunnett's post-hoc tests. *P< 0.05; ****P< 0.0001. (C) Distribution of methylation levels on 250-bp tiles containing 5 or more CpGs (n = 2.1 million). The three individual Bend3 KO clones are shown on the left, together with WT cells. (D) Differential methylation analysis on 250-bp tiles containing five or more CpGs (n = 2.1 million). (E) DNA methylation values for the indicated genomic elements in WT cells grown in 2i (pink) or serum (orange), and Bend3 KO cells grown in serum (blue) (F, G) Percentage of statistically significant hypermethylated tiles in the indicated genomic compartments, when comparing Bend3 KO to WT cells. The red dashed line shows the value in intergenic regions, for comparison. See Methods for details. (H) Venn diagram showing the overlap between genes that are downregulated in Bend3 KO and genes whose promoter gains DNA methylation in Bend3 KO and compared to WT. P-value: hypergeometric tests. (I) Quantitation of 5-hmC abundance in the cells by LC–MS/MS. p-value: ANOVA followed by Dunnett's post-hoc tests. **P< 0.01; ****P< 0.0001. (J) GSEA analysis on the RNA-seq data (Bend3 KO vs. WT). (K) RNA-seq data for the genes of ‘Oxoglutarate metabolic process’ GO term. Red: higher expression, blue: lower expression. The genes with promoters directly bound by BEND3 are indicated with arrows.

Fig 4: BEND3 represses Dazl expression. (A) Procedure for validation of Bend3 as a bona fide hit. (B) FACS analysis on serum-grown cells. (C) RT-qPCR on serum-grown cells. (D) Western blotting on serum-grown cells. (E) FACS analysis on serum-grown Bend3 mutant cells, rescued with empty vector or with a Bend3-expressing vector. (F) Survival assay on rescued cells. Only one of the three clones is shown, for clarity. (G) Western blotting shows expression of the V5-tagged BEND3 protein in rescue cells, and concomitant repression of DAZL protein expression. The three independent clones are shown.

Fig 5: BEND3 regulates the transcription of bivalent promoters. (A) Principal Component Analysis on RNA-seq data from the indicated cells (three biological replicates for each condition). (B) Volcano plot showing genes that are downregulated (blue) or upregulated (red) in the Bend3 KO cells. (C) Venn diagrams showing the overlap between genes that are differentially expressed (DE, either up or down-regulated), upregulated, or downregulated, between Bend3 KO and WT, and genes that have a bivalent promoter (35). P-value: hypergeometric tests. (D) Proportion of genes with the indicated promoter marks within up- or down-regulated genes. (E) RT-qPCR following reintroduction of V5-BEND3 in Bend3 KO cells. Genes that were downregulated upon KO (left panel) regain expression upon rescue, while genes that were upregulated in the KO become repressed upon rescue (right panel). P-value: Student's t-test, corrected by two-stage Benjamini, Krieger, & Yekutieli FDR procedure. *P < 0.05; **P < 0.01.