Fig 2: ELOB is enriched at genomic loci bound by H3K27M and enriched in active histone marks.(A and B) Heatmap (A) and average profile plots (B) of ELOB binding sites (3,076 peaks) identified by CUT&RUN (C&R), demonstrating co-enrichment in H3K4me3, H3K4me1, H3K27ac, and H3K27M oncohistones and ELOB-bound regions. H3K27M CUT&RUN data was also generated in additional control SU-DIPGXIII cells where a single copy of H3.1 was knocked out by CRISPR-targeting (HIST1H3B, H3.1KO) to confirm H3K27M antibody specificity. (C) Pie chart showing overlap between ELOB binding sites and various genomic elements, revealing frequent ELOB binding at promoters, intergenic regions, and introns. (D) Overrepresented gene categories from Gene Ontology (GO) analysis of the genes near ELOB binding sites (+/− 200 kb) revealing ELOB binding at genes associated with transcription and chromatin regulation, mitosis, and DNA damage pathways. (E and F) Representative genome browser snapshots showing co-localization of H3K27M (grey boxes), H3K4me1, H3K4me3, and H3K27ac at ELOB peaks (blue boxes) located near the ELL (E) and ARNT (F) gene loci. (G) Overrepresented transcription factor motifs in the ELOB peak set, including factors involved in glial and brain development like SOX9, SOX10, and EN1 from HOMER motif enrichment analysis.

Fig 3: ELOBKO alters Pol2 transcriptional activity at bivalent chromatin regions to control developmental and metabolic genes.(A) Heatmap showing Z-scores from RNA-seq analysis of control and ELOBKO SU-DIPGXIII and BT245 cells. See also table S4. (B) Quantification of PRO-seq reads at ELOBKO-upregulated genes in SU-DIPGXIII cells demonstrating a corresponding increase in Pol2 activity at both the gene promoters (top panel) and gene bodies (bottom panel) due to ELOB loss. (C and D) Average PRO-seq profiles and CUT&RUN signal centered on transcription start sites (TSS) with increased (C) or decreased (D) PRO-seq reads in ELOBKO cells. See also table S3. (E to G) Genome browse snapshots revealing decreased PRO-seq signal at the H3K27me3-enriched ASCL2, KISS1R, and ARID3A gene loci with regional ELOB and H3K27me3 enrichment (E and F) and decreased PRO-seq signal at the NFATC2 gene, which is bound by both ELOB and H3K27M (G). Differential PRO-seq peaks and ELOB-binding regions are marked by green and blue boxes. (H) Heatmap of ChromHMM results integrating ChIP-seq, CUT&RUN, RNA-seq, and PRO-seq data revealing dynamic Pol2 activity (PRO-seq-UP and -DN), enhancer signatures, and bivalent H3K27me3/H3K4me3 chromatin at ELOB-regulated targets. (I and J) GSEA (I) and GO analysis (J) of ELOBKO RNA-seq data showing downregulation of genes associated with cell division, mitochondria, and metabolism, and upregulation of genes associated with neuronal function and excitatory tissues. GSEA results represent normalized enrichment scores (NES) for non-redundant gene sets (FDR<0.01). (K) Representative images of NESTIN staining (left panel) and quantification of the %NESTIN(+) (right panel) stem/progenitor-like cells in control and ELOBKO SU-DIPGXIIIP* xenografts. Scale bar is 800 microns. (L) Genome browser snapshots showing ELOB binding at histone genes. (M) RT-PCR quantification of H1–4 and H2BC21 transcript expression normalized to ACTB from control versus ELOBKO SU-DIPGXIII cells.

Fig 4: ELOB promotes H3K27M-mutant DMG tumor growth.(A to C) Representative images (A) and quantification (B and C) from tumor bioluminescent imaging of NSG immunocompromised mice bearing pontine xenografts of control and ELOBKO SU-DIPGXIIIP*+Zsgreen/luciferase cells. In panels (B and C), two different sgRNA sequences were used to target ELOB (KO1 and KO2), and statistical significance was determined using an unpaired Student’s t-test. (D to F) Representative fluorescence microscopy images of ELOB-stained SU-DIPGXIIIP* xenograft brain sections (D) and quantification demonstrating increasing proportions of ELOB(+) cells over time (E) and a greater percentage of ELOB(+) cells collected at days 22–23 versus days 18–20 (F). In panel (D), the top scale bar is 800 microns, and the bottom scale bar is 100 microns. (F and G) Representative images (F) and quantification (G) of the %MKI67 (+) proliferating cells in control and ELOBKO SU-DIPGXIIIP* xenograft tumors. (H) Kaplan-Meier survival curve comparing the survival of mice bearing control versus ELOBKO SU-DIPGXIIIP* xenograft tumors (*p<0.0295, log-rank test).

Fig 5: The Pol2 SIII complex is a genetic dependency in H3K27M-mutant diffuse midline glioma (DMG).(A) Protein-protein interaction network illustrating selected genetic dependencies from a chromatin-focused CRISPR dropout screen in two H3K27M-mutant glioma cell lines (SU-DIPGXIII and SU-DIPGVI) highlighting roles for regulators of Pol2 elongation, histone acetylation, and RNA splicing in DMG growth and survival. Node width is scaled to the average STARS score, with larger sizes correlating with stronger genetic dependencies. (B) Schematic of RNA polymerase II (Pol2) SIII transcriptional regulatory complex of ELOA, ELOB, and ELOC. (C and D) Immunoblot analysis revealing reduced ELOB protein in SU-DIPGXIIIP*+Cas9 (C) and HSJD-DIPG007+Cas9 DMG (D) cells transduced with two independent ELOB sgRNAs (ELOBKO1 and KO2). (E and F) Normalized cell counts of control and ELOBKO SU-DIPGXIIIP* (E) and HSJD-DIPG007 (F) cells at various time points following selection. (G to I) Normalized cell counts in Cas9 (+) H3K27M-mutant SU-DIPGXIII and SU-DIPGVI cells used in the original screen and H3K27M mutant BT869 cells, expressing either control or ELOB sgRNAs. (J to L) Normalized cell counts of Cas9 (+) H3K27M-mutant DMG cell lines following CRISPR-based targeting of ELOA, ELOB, or ELOC. (M) BrdU pulse labeling and cell cycle analysis showing the average percent of live SU-DIPGXIII+Cas9 cells transduced with control, ELOA, ELOB, or ELOC sgRNAs in different cell cycle phases, suggesting reduced BrdU(+) S-phase cells and increased sub-G1 cells following ELOB loss. Error bars show S.E.M. with *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001 as determined by one-way ANOVA or Student’s t-test.