The KDM5 Family of Histone Lysine Demethylases Are Targets of R-2-Hydroxyglutarate in IDH Mutant Tumors

Gain-of-function mutations in isocitrate dehydrogenase enzymes IDH1 and IDH2 occur in ∼10% of acute myeloid leukemias (AML) and >80% of gliomas. The mutant enzymes convert 2-oxoglutarate (2OG) to the oncometabolite R-2-hydroxyglutarate (R-2HG). R-2HG promotes cellular transformation by modulating the activities of 2OG-dependent dioxygenases (2OGDDs). The only functionally validated direct target of R-2HG is TET2, a 2OGDD myeloid tumor suppressor that catalyzes the conversion of 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC).

Interestingly, in clonal myeloid disorders the patterns of IDH and TET2 mutations are vastly different. TET2 mutations occur at similar frequencies in clonal hematopoiesis of unknown significance (CHIP), lower- and higher-grade myeloproliferative (MPN) and myelodysplastic (MDS) disorders, and primary and secondary AML. IDH mutations, on the other hand, are associated with higher-grade and blast-phase MPN and MDS and with de novo AML and are rare in CHIP and low-grade MDS. This suggests that mutant IDH promotes a more aggressive disease phenotype and that R-2HG has additional targets other than TET2 that contribute to its leukemogenic activity.

To ask if the in vitro transforming activity of R-2HG directly correlates with TET2 inhibition, we treated TF-1 cells, a cytokine-dependent human AML cell line, with a dose range of cell-permeable esterified R-2HG. We found that R-2HG induces cytokine independence at concentrations that have no effect on 5hmC levels. To identify other 2OGDD myeloid tumor suppressors that could be contributing to R-2HG-mediated transformation, we performed a positive-selection CRISPR-Cas9 screen under cytokine-poor conditions in TF-1 cells. We identified three H3K4 histone lysine demethylases, KDM5A, KDM5C and KDM5D, as genes whose sgRNAs were enriched upon cytokine withdrawal. Triple knockout of KDM5A, KDM5C and KDM5D (TKO) in TF-1 cells induces robust cytokine independence. Likewise, treatment of TF-1 cells with KDM5c70, a specific inhibitor of KDM5 enzymes, strongly induces TF-1 cytokine independence. Of note, KDM5 inhibition has no effect on TET2 expression or 5hmC levels. We further found that R-2HG is a more potent inhibitor of KDM5A, KDM5C and KDM5D than of TET2. We then assessed the effect of mutant IDH1 expression, TKO, R-2HG treatment and KDM5c70 treatment on H3K4 trimethylation by ChIP-seq and found that each of these perturbations results in a significant enrichment in H3K4me3 peaks relative to controls.

TET enzymes are not recurrently mutated in glioma and although there is a strong correlation between mutant IDH status and the CpG island methylator phenotype (CIMP), direct inhibition of TET2 by R-2HG has not been reproducibly demonstrated in glioma. To ask if TET2 activity is suppressed in IDH mutant glioma, we quantified 5hmC levels in a panel of primary IDH wild-type and IDH mutant glioma and AML samples by mass spectrometry. We found that, unlike in AML, in glioma there is no correlation between IDH1 mutation status and loss of 5hmC. We likewise saw no correlation between 5hmC levels and either IDH mutation status or intracellular R-2HG levels in patient derived xenograft (PDX) models of glioma. Given the lack of evidence that TET enzymes are tumor suppressor targets of R-2HG in IDH mutant glioma, we asked if mutant IDH positivity is associated with increased levels of H3K4 methylation in glioma. We performed ChIP-seq on a panel of IDH wild-type and IDH mutant glioma PDX lines and found H3K4me peaks to be highly enriched in the IDH mutant lines when compared to IDH wild-type lines. Trimethyl-H3K4 levels were likewise increased in isogenic normal human astrocyte (NHA) cells ectopically expressing mutant IDH1.

Collectively, these data suggest that R-2HG inhibits KDM5 histone lysine demethylases to promote mutant IDH-mediated transformation in AML and glioma. These studies identify a novel direct target of R-2HG in IDH mutant tumors and provide a functional link between IDH mutations and dysregulated histone lysine methylation in cancer.