PRC2-independent actions of H3.3K27M in embryonic stem cell differentiation

Abstract

The histone H3 variant, H3.3, is localized at specific regions in the genome, especially promoters and active enhancers, and has been shown to play important roles in development. A lysine to methionine substitution in position 27 (H3.3K27M) is a main cause of Diffuse Intrinsic Pontine Glioma (specifically Diffuse Midline Glioma, K27M-mutant), a lethal type of pediatric cancer. H3.3K27M has a dominant-negative effect by inhibiting the Polycomb Repressor Complex 2 (PRC2) activity. Here, we studied the immediate, genome-wide, consequences of the H3.3K27M mutation independent of PRC2 activity. We developed Doxycycline (Dox)-inducible mouse embryonic stem cells (ESCs) carrying a single extra copy of WT-H3.3, H3.3K27M and H3.3K27L, all fused to HA. We performed RNA-Seq and ChIP-Seq at different times following Dox induction in undifferentiated and differentiated ESCs. We find increased binding of H3.3 around transcription start sites in cells expressing both H3.3K27M and H3.3K27L compared with WT, but not in cells treated with PRC2 inhibitors. Differentiated cells carrying either H3.3K27M or H3.3K27L retain expression of ESC-active genes, in expense of expression of genes related to neuronal differentiation. Taken together, our data suggest that a modifiable H3.3K27 is required for proper histone incorporation and cellular maturation, independent of PRC2 activity.

Figure 1.

H3.3K27M/L mutations have little effect on H3.3 genome-wide distribution. (A) Heatmaps of H3K27ac around TSSs (±10 kb), at 72 h after Dox, in ESCs expressing WT H3.3 (blue), H3.3K27M (red) or H3.3K27L (orange). Genes are sorted by H3.3 binding in WT. (B) Heatmaps of H3.3-HA around TSSs (±10 kb), at 72 h after Dox, in ESCs expressing WT H3.3 (blue), H3.3K27M (red) or H3.3K27L (orange). Genes are sorted by H3.3 binding in WT. (C) H3.3 incorporation profile around TSSs 72 h after Dox. ESCs expressing H3.3K27M (red) or H3.3K27L (orange) show increased H3.3 binding in TSSs, not observed in ESCs treated with the PRC2 inhibitor GSK343 (purple), compared with ESCs expressing WT H3.3 (blue). (D) Genomic distribution of mapped reads of H3.3 72 h after Dox.

Figure 2.

Increased incorporation of both H3.3K27M and H3.3K27L around TSSs and enhancers. (A) Experimental layout. Dox was added at the indicated time points. (B–D) H3.3 binding around TSSs, shaded within two standard errors at the indicated time points. ESCs expressing H3.3K27M (red) or H3.3K27L (orange) show increased H3.3 binding around TSSs, not observed in ESCs treated with the PRC2 inhibitor GSK343 (purple), compared with ESCs expressing WT H3.3 (blue). (E–G) H3.3 binding around enhancers, shaded within two standard errors at the indicated time points. ESCs expressing H3.3K27M (red) or H3.3K27L (orange) show increased H3.3 binding around enhancers, not observed in ESCs treated with the PRC2 inhibitor GSK343 (purple), compared with ESCs expressing WT H3.3 (blue). (H) Correlation of H3.3 binding across all time points (4, 8, 24 and 72 h after Dox) in two independent experiments. (I) Correlation of H3.3 binding (at 4, 8, 24 and 72 h after dox induction) with WT-1st (ESCs expressing WT-H3.3, first experiment).

Figure 3.

H3.3K27M/L mutations lead to increased correlation between H3.3 binding and gene expression. (A) Gene expression correlations between the different lines and WT. (B) Correlation between H3.3 incorporation and gene expression. (C) Mean gene expression of all genes (left bars), genes enriched with H3.3K27M (middle) or H3.3K27L (right), in ESCs expressing WT-H3.3 (blue), H3.3K27M (red) and H3.3K27L (orange). (D, E). Mean incorporation of WT-H3.3 (blue), H3.3K27M (red) and H3.3K27L (orange) across the different time-points around silent genes (D) and highly expressed genes (E).

Figure 4.

Expression of H3.3K27/L in differentiated ESCs support the retention of ESC-expressed genes. (A) Experimental layout. (B) Correlation of gene expression between ESCs and RA treated cells. ESCs expressing both H3.3K27M (red) and H3.3K27L (orange) mutants exhibit higher correlation with the undifferentiated state compared with ESCs expressing WT-H3.3 (blue) (**P < 10⁻³; ***P < 10⁻⁴). (C) Same as B in ESC-derived OPCs (*P < 0.05). (D) Log gene expression values in RA-induced ESCs, comparing ESCs expressing WT-H3.3 (Y-axis) with ESCs-expressing H3.3K27M (X-axis). Blue and red denote genes higher in RA treated ESCs expressing H3.3-WT and RA treated ESCs expressing H3.3K27M, respectively. (E) Quantification of gene expression differences in ESCs in 2i condition, of gene groups defined by activity in RA (D) (***P < 10⁻⁵⁰). (F) Most enriched GO terms for genes higher in ESCs expressing H3.3K27M versus ESCs expressing WT-H3.3, in RA-induced ESCs. (G) Same as (F) for genes higher in ESCs expressing WT-H3.3 versus ESCs expressing H3.3K27M, in RA-induced ESCs. (H, I) Same as (D, E) for ESCs expressing H3.3K27L (I: ***P < 10⁻⁵⁰). (J) Same as (E) in ESCs and OPCs (***P < 10⁻⁵).

Figure 5.

Mutant cells differentiated to neural progenitors retain the expression of genes which are highly expressed in the pluripotent state. (A) Correlation of H3.3 incorporation around TSSs between ESCs and RA-induced cells in ESCs expressing WT-H3.3 (blue), ESCs expressing H3.3K27M (red) and ESCs expressing H3.3K27L (orange) (***P < 10⁻⁵). (B) Correlation of H3.3 incorporation around TSSs between ESCs expressing WT-H3.3 and ESCs expressing H3.3K27M (left); ESCs expressing WT-H3.3 and ESCs expressing H3.3K27L (middle), and ESCs expressing H3.3K27M and ESCs expressing H3.3K27L (right) (**P < 10⁻³), in RA-induced ESCs. (C) H3.3 meta-gene enrichment plots in WT ESCs around TSSs of genes expressed higher in the RA-treated H3.3K27M cells compared to WT-H3.3 cells (red) and vice versa (blue). (D) H3.3 meta-gene enrichment plots in WT ESCs around TSSs of genes expressed higher in the RA-treated H3.3K27L cells compared to WT-H3.3 cells (yellow) and vice versa (blue). (E) H3.3 meta-gene enrichment plots in H3.3K27M-expressing ESCs around TSSs of genes expressed higher in the RA-treated H3.3K27M cells compared to WT-H3.3 cells (red) and vice versa (blue). (F) H3.3 meta-gene enrichment plots in H3.3K27L-expressing ESCs around TSSs of genes expressed higher in the RA-treated H3.3K27L cells compared to WT-H3.3 cells (yellow) and vice versa (blue). (G) H3.3 meta-gene enrichment plots in human embryo-derived cells (NS19) expressing WT-H3.3 from Brien et al (45) around TSSs of genes higher in human embryo-derived cells expressing WT-H3.3 compared with human embryo-derived cells expressing H3.3K27M (blue) and vice versa (red). (H) Same as (G) in human embryo-derived cells expressing H3.3K27M. (I) Ratio of High-H3.3-at-8-h-genes expressed at 72 h in ESCs expressing WT-H3.3 (blue), ESCs expressing H3.3K27M (red) and ESCs expressing H3.3K27L (orange). Top: in undifferentiated ESCs (**P < 0.005); Bottom: in RA-induced ESCs (***P < 10⁻⁵).