Genome-wide mutagenesis screen identifies genes required to activate a HIF response.
To identify genes required to activate HIF target loci, we used a dynamic HIF-mCherry reporter17, 18, which reversibly accumulates in hypoxia through a fused oxygen dependent degradation domain (ODD), and requires endogenous HIF binding to a consensus triplicate HRE for activation (Figure 1a-c). The mutagenesis screen was performed by transducing Cas9-expressing HeLa reporter cells with a genome-wide sgRNA library, incubating the cells in 1% oxygen for 24 hr, and undertaking iterative-flow activated cell sorting (FACS) for those cells that failed to activate the reporter (Figure 1d, e). sgRNAs enriched in the sorted population that failed to activate the reporter were identified by high throughput deep sequencing in comparison to sgRNAs from cells that had only been exposed to 21% oxygen (Figure 1d, e). HIF1b (ARNT) and HIF-2a (EPAS1) were highly enriched for sgRNA in the mutagenesis screen, validating our approach (Figure 1e). We also identified several genes implicated in histone modifications that were significantly enriched in hypoxic mutagenised cells compared to control cells: SET1B, UBE2A (also known as Rad6) and PPP4C (Figure 1e). UBE2A is a ubiquitin E2 conjugating enzyme involved in global H2B ubiquitination by the RNF20 and RNF40 E3 ligases 19, 20. We proceeded to validate UBE2A by making a mixed CRISPR knockout (KO) population of HeLa reporter cells, which showed decreased activation of the HIF reporter (Supplementary Figure 1a). Depletion of RNF20 and RNF40 also decreased expression of the HIF reporter (Supplementary Figure 1b), consistent with their reported recruitment of UBE2A for ubiquitination19, 20. A role for histone acetylation in the hypoxic response was supported by the identification of PPP4C, a phosphatase that controls histone-deacetylation activity (HDAC3)21, which we also validated with CRISPR mixed KO populations of PPP4C (Supplementary Figure 1c).
SET1B, a mammalian H3K4 methyltransferase22, was of particular interest as it was highly enriched in the screen, and suggested that histone methylation is involved in the activation of HIF target genes. We confirmed that sgRNA depletion of SET1B decreased activation of the HIF reporter in hypoxia with mixed knockout populations, multiple sgRNAs and in isolated clones (Figure 1f, Supplementary Figure 1d-f). In all cases SET1B loss decreased HIF reporter fluorescent levels in hypoxia (Figure 1f, Supplementary Figure 1d, e), without altering HIF reporter levels in 21% oxygen (Supplementary Figure 1g). Although SET1B depletion did not prevent activation of the HIF reporter to the same extent as HIF1b loss, time course analyses showed a marked delay in reporter fluorescence with SET1B depletion in 1% oxygen (Figure 1g), consistent with the involvement of SET1B in activation of the HIF pathway.
The HIF transcriptional response is impaired by depletion of the SET1B methyltransferase.
We next examined whether SET1B depletion altered the activation of the endogenous HIF response, and if this was applicable to other cell types. Using the well-validated HIF-1a target, Carbonic Anhydrase 9 (CA9), we confirmed that CA9 levels decreased in HIF HeLa reporter cells, consistent with a decrease in transcriptional activation of the HIF response (Figure 2a, b). Similar findings were observed in other cancer cell lines (A549 lung adenocarcinoma and MCF7 breast cancer cells), and skin fibroblasts following SET1B sgRNA depletion in hypoxia (Figure 2c-f), with a similar temporal and partial decrease in CA9 levels to those observed with the HIF reporter (Supplementary Figure 2a). We verified that decreased activation of the reporter was not due to changes in HIF-1a levels, as SET1B loss did not alter HIF-1a mRNA expression or protein abundance in several cancer cell lines or immortalised skin fibroblasts (Supplementary Figure 2b-e). HIF-2a and HIF1b protein levels were also unchanged in SET1B deficient cells (Supplementary Figure 2c-e). Furthermore, reconstituting SET1B following its siRNA-mediated depletion increased CA9 levels in HeLa and A549 cells (Figure 2g, Supplementary Figure 2f, g), confirming the specific involvement of SET1B in the transcriptional activation of HIF targets.
Hypoxia activates the HIF response by decreasing prolyl hydroxylation of the HIF-a subunits, preventing their VHL dependent ubiquitination and proteasome-mediated degradation. We therefore asked whether SET1B loss in the context of PHD inhibition altered activation of HIF targets independently of hypoxia. Treating HIF reporter cells with the PHD inhibitor dimethyloxalylglycine (DMOG) stabilised HIF-1a and activated the HIF reporter, but we observed decreased HIF reporter fluorescence levels and endogenous CA9 in SET1B depleted cells treated with DMOG (Supplementary Figure 3a, b). We also observed that proteasome inhibition (MG132) did not rescue CA9 levels following SET1B depletion in hypoxia (Supplementary Figure 3e), consistent with SET1B enhancing transcriptional activation of HIF targets rather than altering the stability of HIF-1a or HIF targets.
Six H3K4 methyltransferases have been identified in humans: SET1A, SET1B and 4 mixed lineage leukaemia (MLL) genes, which together comprise the SET/COMPASS family22. SET1A and SET1B are closely related, each forming a large complex that shares many subunits with other COMPASS members, aside from CFP1 and WDR8222 (Figure 2h). Therefore, we tested if depletion of SET1A, CFP1 or WDR82 altered activation of the HIF reporter following DMOG treatment. In contrast to SET1B, SET1A depletion had no effect on HIF reporter activation (Figure 2i, j). However, depletion of CFP1 and WDR82, which associated with SET1B, decreased activation of the reporter similarly to SET1B loss (Figure 2i, k, l). Thus, the SET1B complex, but not SET1A, alters activation of the HIF reporter.
SET1B selectively drives mRNA expression of HIF target genes.
Decreased activation of the HIF-fluorescent reporter and the reduction in CA9 levels in hypoxia suggested that SET1B modulated the activation of HIF target genes. Therefore, we determined the global effect of SET1B on gene expression in hypoxia using RNA-seq. Mixed populations of SET1B knockout, HIF-1b knockout or wildtype HeLa cells were exposed to 12 hr 21% or 1% oxygen, and analysed in duplicate by RNA-seq (Figure 3a). SET1B depletion had a minimal effect on global gene transcription in 21% oxygen, and these changes were less than observed with HIF1b depletion alone (Figure 3b, c). In hypoxia, 1033 genes were significantly induced (Supplementary Figure 4a, b) , of which 764 (74%) were decreased following HIF-1b loss (Figure 3d, Supplementary Figure 4a, c). SET1B loss significantly decreased the expression of 20% of these hypoxia dependent genes (Supplementary Figure 4a, d), favouring those that were highly upregulated in the hypoxia control HeLa cells (CA9, VEGF and PHD3) (Figure 3d, e). Quantitative PCR confirmed these findings in several cell types, with reduced expression of CA9, PHD3 and VEGF but not a control gene, BAP1, or the HIF-1a target GLUT1 (Figure 3f, Supplementary Figure 4e-g). Moreover, reconstituting SET1B following siRNA mediated depletion of SET1B restored HIF activation of CA9 and VEGF (Supplementary Figure 4h, i).
Differential activation of the HIF-a isoforms did not appear to account for the subset of genes altered by SET1B loss. Analyses of available HIF-1a and HIF-2a binding sites from MCF7 or HepG2 cells showed no clear bias towards one isoform (Supplementary Figure 5a-d), although there was incomplete overlap with our RNA-seq analysis. To test HIF-a selectivity further, we generated isoform specific fluorescent reporter lines by clonal knockout of HIF-1a, HIF-2a or both together (Supplementary Figure 5e, f). HeLa HIF-1a KO (HIF-2 reporter) or HIF-2a KO (HIF-1 reporter) HRE-GFPODD cells, depleted of SET1B, both showed decreased GFP levels following PHD inhibition (DMOG), with combined HIF-1a/HIF-2a KO reporter cells showing no response to DMOG treatment (Supplementary Figure 5f, g). Thus HIF-a isoform specificity could not account for the involvement of SET1B in activation of a subset of hypoxia inducible genes.
The finding that GLUT1 was not altered by SET1B loss suggested that some cellular processes typically activated by HIFs may be differentially regulated. Gene ontology analysis supported this, as SET1B loss did not alter the transcriptional activity of all HIF target genes equally, but preferentially altered expression of genes involved in angiogenesis (Figure 3h, Supplementary Figure 5h), with little effect on those involved in pathways such as glycolysis (Supplementary Figure 5h). These findings were further supported by a marked decrease in the secretion of VEGF, a key angiogenic factor, in SET1B and HIF1b depleted A549 or HeLa cells (Figure 3i). In contrast, SET1B loss had a minimal effect on the HIF-dependent shift to glycolysis using bioenergetic profiling (Supplementary Figure 6a-d). HeLa cells treated with DMOG or selective PHD inhibitors (Roxadustat and Daprodustat) showed a shift to glycolysis, with a decrease in oxygen consumption relative to extracellular acidification (Supplementary Figure 6a, b). HIF1b deficient HeLa cells prevented this HIF-dependent glycolytic shift, as expected. However, SET1B deficient HeLa cells behaved the same as the wildtype cells following PHD inhibition, consistent with HIF-mediated activation of glycolysis remaining intact. Similar findings were observed in A549 cells following selective PHD inhibition, confirming that SET1B selectively regulates a subset of HIF target genes, with a strong effect on angiogenic genes and a minimal effect on glycolysis.
SET1B loss decreases cell survival in hypoxia and delays tumour establishment.
To understand the functional requirement for SET1B in the HIF response, we first examined the effect of SET1B loss on cell growth in hypoxia. HeLa or A549 mixed SET1B knockout populations were incubated in 21% or 1% oxygen and total cell numbers were measured at daily intervals. HIF1b-depleted HeLa or A549 cells were used as a control for inactivation of the HIF response. HeLa or A549 cells grew at similar rates at 21% O2 irrespective of SET1B or HIF1b depletion (Supplementary Figure 7a, b), but we observed a decrease in total cell number of SET1B or HIF1b deficient cells after 24-48 hr of hypoxia (Figure 4a, b). These findings were consistent with the ability of the HIF response to favour cell survival in hypoxia, as demonstrated in certain cancers23, 24. In these experiments, we also observed that hypoxia altered the levels of SET1B itself, with a marked transcriptional downregulation and decrease in protein levels after 24 hr in both HeLa and A549 cells (Figure 4c-e), suggesting that reducing SET1B expression may reduce the downstream response to HIF in prolonged hypoxia through a negative feedback loop. Importantly, the hypoxic-specific growth defect in the SET1B deficient cells was distinct from SET1A loss, which reduced cell numbers even when incubated in 21% oxygen (Supplementary Figure 7c), indicating a more global involvement in transcriptional regulation for SET1A. This was also evidenced by decreased mRNA of both HIF target genes and control genes following SET1A depletion (Supplementary Figure 7d).
We next examined whether the decrease in cell numbers in the SET1B depleted cells was due to decreased proliferation or increased cell death. Cell tracing with CFSE staining showed that proliferation rates were not altered in HIF1b or SET1B-deficient cells irrespective of oxygen availability (Supplementary Figure 7e); however, there was an increase in PARP1 cleavage and caspase 3 activity in both the HIF1b and SET1B-depleted cells (Supplementary Figure 7f, g), consistent with increased apoptosis during hypoxia.
The contribution of HIF to cell survival in hypoxia does not only relate to growth or apoptosis, but involves activation of several pathways to improve oxygenation of tissues25. Microenvironmental activation of HIF can increase vascularisation and growth, and this angiogenic effect of HIF is evident in tumour xenograft models26-28. Given the preponderance of genes influenced by SET1B involved in angiogenesis, we explored whether SET1B loss altered tumour establishment and growth using cervical (HeLa) and lung adenocarcinoma (A549) xenografts. HeLa cells or A549 cells, depleted of SET1B or HIF1b, were injected into the backs of nude mice, and the time to tumour establishment (200mm3 volume) or to reach a tumour volume of 1000mm3 was measured. In the HeLa xenografts, HIF1b or SET1B depletion resulted in a delay in tumour establishment, but not in the subsequent growth of the tumour (Figure 4g, Supplementary Figure 7h). A549 SET1B-deficient tumours also showed a marked delay in the initial tumour growth, which was strikingly more than observed with HIF1b loss (Figure 4h, Supplementary Figure 7i). Further analysis of the tumour xenografts confirmed that SET1B deficient tumours had increased hypoxic regions compared to the control A549 tumours, similarly to HIF1b loss (Figure 4i, j), consistent with decreased angiogenesis. Moreover, using CD31 as a marker of blood vessels within the tumour sections, we found that SET1B-deficient tumours had decreased blood vessel density (Figure 4i, k). Together, these findings indicate that SET1B loss alters HIF activation of angiogenic pathways, contributing to delayed tumour establishment in xenograft models.
SET1B accumulates on chromatin in hypoxia and is selectively recruited by HIF-1a.
To understand how SET1B was mechanistically involved in activating HIF target genes, we first examined the interaction of SET1B with the HIF heterodimer. Endogenous SET1B associated with both HIF-a isoforms and HIF1b (Figure 5a, Supplementary Figure 8a-c), however the association between HIF-1a and SET1B was more prominent. The formation of the mature heterodimeric HIF complex was not required, as SET1B still associated with HIF-1a in HIF1b null cells, indicating that the interaction was not mediated through HIF1b (Figure 5a). Interestingly, the interaction still occurred when HIF-1a was prolyl-hydroxylated (Supplementary Figure 8d, e) and was independent of HIF chromatin binding, as treatment with DNAse did not prevent the association of HIF-1a with SET1B (Figure 5b). Other components of the SET1B complex were included in this interaction, as we observed association of HIF-1a with CFP1 (Supplementary Figure 8f). To map the interaction domain of HIF-1a with the SET1B complex, we generated HIF-1a truncation mutants in HEK293T cells (Figure 5c). Immunoprecipitated SET1B associated with HA-tagged HIF-1a constructs containing the PAS domains, but did not require the presence of the ODD or the c-terminal transactivating domain. However, a truncation mutant lacking the second PAS-B domain, a region involved in heterodimerisation with HIF-b, prevented the association with the SET1B complex (Figure 5d, e). Thus, recruitment of SET1B is mediated through the PAS domains on HIF-1a, and mainly through PAS-B.
SET1B has been observed to localise to the cytosol and nucleus in 21% oxygen29, suggesting that its differential localisation may affect its function. We therefore determined if hypoxia altered SET1B localisation, and if HIF-1a recruited SET1B to chromatin. Subcellular fractionation studies in HeLa and A549 cells showed that SET1B was evenly distributed between the cytosol, nucleoplasm and chromatin fractions in 21% oxygen, as previously reported29, but following incubation in hypoxia, increased levels of SET1B were found in the chromatin fraction (Supplementary Figure 8g, h). Time course analyses in HeLa or A549 cells showed an increase in the SET1B chromatin fraction after 4 hr hypoxia, with nearly all SET1B present on chromatin after 24 hr (Figure 5f, g Supplementary Figure 8i, j). This translocation of SET1B closely coincided with the movement of HIF-1a to the nucleus (Figure 5f, g, Supplementary Figure 9i, j). Furthermore, stabilisation of HIF-1a with DMOG also resulted in accumulation of SET1B on chromatin (Figure 5h), indicating that HIF stabilisation, irrespective of hypoxia, resulted in the translocation of SET1B.
To further understand the recruitment of SET1B to HREs we used ChIP-PCR at validated HIF target genes (CA9, PHD3 and VEGF). In all cases ChIP-PCR for HIF-1a confirmed that SET1B was only recruited to HIF target sites following binding of the HIF complex (Figure 6a, b), and that SET1B or the CFP1 component of the SET1B complex did not bind selected HIF-1 target loci in HIF1b null cells (Figure 6c, d). Thus, SET1B associates first with HIF-1a and is then recruited to selected HIF target genes in hypoxia.
Accumulation of H3K4me3 at HIF target genes in hypoxia is HIF dependent and mediated by SET1B.
We next determined whether SET1B altered H3K4me3 in hypoxia, and if this occurred at HIF target gene loci. Hela or A549 cells exposed to hypoxia showed a transient increase in total levels of H3K4me3 (Supplementary Figure 9a, b), as previously reported, and attributed to hypoxic inhibition of lysine demethylase activity13, 14. ChIP-PCR for H3K4me3 at selected HIF targets also confirmed that this mark increased in hypoxia at promoter regions (Figure 7a). However, H3K4me3 deposition was clearly HIF dependent, as we observed a substantial decrease in H3K4me3 levels at CA9, PHD3 and VEGF promoter regions in HIF1b null cells (Figure 7b). This indicated that HIF binding was associated with active deposition of H3K4me3, rather than hypoxic inhibition of demethylases. To determine if SET1B was responsible for H3K4me3 at HIF targets, we depleted SET1B in HeLa cells and used ChIP-PCR to measure H3K4me3 at HIF loci promoter regions (Figure 7b). SET1B depletion markedly decreased H3K4me3 at these HIF target loci in hypoxia, without altering H3K4me3 levels at the promoter of the non-HIF target BAP1 (Figure 7b).
To further explore the association of SET1B with H3K4 methylation in hypoxia, we used ChIP-seq to determine locus-specific changes in H3K4me3 levels across the genome in control, HIF1b or SET1B deficient HeLa cells incubated in 21% or 1% oxygen for 6 hr. In each dataset, the overall distribution of H3K4me3 was similar (Figure 7c) being predominantly located at the promoter regions of active genes as previously described30. We then focused on those sites at which H3K4me3 was induced by hypoxia (false discovery rate - fdr < 0.00001) in control cells. The observed hypoxia-associated increase in H3K4me3 was reduced in both HIF1b and SET1B deficient cells (Figure 7d). The effect of SET1B depletion was even more marked when H3K4me3 signal was examined at the subset of hypoxia-dependent sites that were regulated by HIF (i.e. also downregulated by HIF1b depletion in hypoxia, fdr < 0.00001) (Figure 7e). (Figure 7e). Analysis of individual HIF target genes that showed a transcriptional dependence on SET1B clearly demonstrated decreased deposition of H3K4me3 at promoter regions following SET1B or HIF1b loss (Figure 7f). The specificity for HIF target genes involved in angiogenesis was again apparent (Supplementary Figure 9d), with little change in H3K4me3 in several key glycolytic genes (Supplementary Figure 9e) or a control gene, BAP1, following SET1B loss. (Supplementary Figure 9f).
Lastly, we explored the relationship between H3K4me3 and acetylation as several studies have highlighted crosstalk between promoter methylation and acetylation in the control of gene transcription31-33. We therefore measured the levels of H3K27ac at HIF target gene promoter regions by ChIP-PCR, with or without SET1B depletion (Figure 7g). H3K27ac increased at selected HIF target genes in hypoxia, consistent with active gene transcription34 (Figure 7g). However, either HIF1b or SET1B depletion prevented H3K27ac deposition at the promoter regions (Figure 7g), providing further evidence for the recruitment of SET1B in promoting active gene transcription in hypoxia. Together, these findings demonstrate that histone methylation of a subset of HIF target loci is mediated by SET1B, and that HIFs direct this epigenetic regulation through recruitment of the SET1B complex, shedding insights into selective transcriptional regulation of the HIF response (Figure 8).