Fig. 6: Inhibiting H3K4me3 accumulation under hypoxia blocks TSS switching and decreases cellular fitness.

a, Representative immunoblots of H3K4me3 from T47D cells pre-treated with 25 µM OICR-9429 or DMSO for 48 h and treated with hypoxia or normoxia, with the addition of 25 µM OICR-9429 or DMSO for an additional 48 h. H3 and α-tubulin were used as loading controls, and HIF1α as a positive control for hypoxia (n = 3 independent experiments). b, Densitometry of H3K4me3 normalized to α-tubulin and H3 loading controls. Bars indicate mean ± s.d. P = 0.01; two-sided t-test (n = 3 independent experiments). c, Kernel density estimation P value and FDR distributions for differential TSS usage in T47D cells between co-treatments of hypoxia and OICR-9429 or DMSO (n = 5,016). Dotted grey line indicates an FDR threshold of 0.15. d, Bar plot indicating the proportions of protein-coding transcripts with significantly altered TSS usage between T47D cells co-treated with hypoxia and OICR-9429 versus DMSO (FDR < 0.15). e, TSS-switching events under hypoxia with lower FDRs are more likely to be reversed by OICR-9429 treatment. The FDR range for TSS switching under hypoxia was divided into ventiles and the proportion of opposite changes in 5′UTR isoforms between hypoxia + OICR-9429 versus hypoxia + DMSO comparisons was determined. Dashed line marks the proportion of changes expected to be the same by chance (22.9%, estimated by Monte Carlo simulation). Red line represents smoothed linear regression, with 95% CI shaded. f, Bar plot of categories of 5′UTR sequences enriched in transcripts with significant TSS switching after co-treatment with hypoxia and OICR-9429, identified by change-point analysis (n = 5,016). g, Bar plot of categories of 5′UTR sequences enriched for transcripts with significant TSS switching between hypoxia and normoxia. The percentage reversed with OICR-9429 is indicated. In total, 786 (31.1%) of change-point-defined 5′UTR categorical changes under hypoxia were reversed by OICR-9429. h, Quantification of 5′UTR isoforms of NM_004618 (TOP3A mRNA) in hypoxia and normoxia-treated (top) and hypoxia + DMSO or OICR-9429-treated T47D cells (bottom). Change-point analysis (dotted orange lines) identified enriched and depleted 5′UTR segments in both comparisons (right). OICR-9429 restores expression of shorter 5′UTR isoforms that were lost under hypoxia. i, Venn diagram showing the overlap of transcripts where hypoxia-induced TSS switching was recapitulated by C48 treatment and reversed by OICR-9429 treatment in T47D cells. j, Quantification of 5′UTR isoforms of NM_001191028 (TNPO3 mRNA) in hypoxia and normoxia-treated (top), C48 and DMSO-treated (middle) and hypoxia-treated T47D cells co-treated with either DMSO or OICR-9429 (bottom). Change-point analysis (dotted orange lines) identified enriched and depleted 5′UTR segments in all three comparisons (right). C48 treatment recapitulates the enrichment in expression of shorter 5′UTR isoforms that occurs under hypoxia, whereas OICR-9429 reverses this effect. k, Trypan blue exclusion assays to quantify viable cells under hypoxia. T47D cells treated the same as in a were counted at the end point. Displayed is the mean of the ratio of cell count in hypoxia versus normoxia in OICR-9429 or DMSO-treated cells ± s.d. P = 0.012; two-sided t-test (n = 6 independent experiments).

Source Data