Mitochondrial Metabolism and RNA Methylation in DLBCL

D-2-hydroxyglutarate dehydrogenase (D2HGDH) is a mitochondrial enzyme that catalyzes the oxidation of D-2-HG into alpha-ketoglutarate (α-KG). Homozygous loss of the D2HGDH gene leads to accumulation of D-2-HG and causes type I D-2-hydroxyglutaric aciduria (D-2-HGA), an autosomal recessive neurometabolic disease. Conversely, type II D-2-HGA is caused by de novo gain-of-function mutations in IDH2, which via its neomorphic activity also leads to accumulation of D-2-HG. The link between D2HGDH and IDH2 is also present in cancer. Gliomas and acute myeloid leukemia (AML) frequently display IDH1/2 mutations, which result in competitive inhibition of α-KG-dependent dioxygenases (e.g., DNA hydroxylases and histone demethylases) by D-2-HG, and epigenetic remodeling. We recently reported (Cancer Cell . 2017 31:619-620) that in IDH-mutant tumors, this epigenetic deregulation extends to RNA as D-2-HG also competitively inhibits the α-KG dependent RNA demethylase FTO, resulting in elevation of N6-Methyladenosine (m6A) RNA marks. Contrary to the gain-of-function IDH1/2 mutations of AML, we discovered somatic heterozygous loss-of-function mutations of D2HGDH in diffuse large B cell lymphomas (DLBCL) (Nat Commun . 2015 6:7768), which remarkably led to a DNA/histone hypermethylation phenotype similar to that of IDH-mutant tumors.

However, we still do not know if loss of D2HGDH deregulates RNA methylation in DLBCL and if a haplo-insufficiency model explains the broad epigenetic remodeling found in D2HGDH-mutant DLBCL. Further, the role of the related mitochondrial enzyme, L2HGDH, which converts the enantiomer L-2-HG into α-KG, on RNA/DNA methylation has not been investigated. Addressing these knowledge gaps will delineate the extent of the metabolism/epigenetic interplay in DLBCL, in particular RNA methylation, which is an emerging pathogenic event in cancer.

First, we utilized CRISP-Cas9 technology to create models of D2HGDH and/or L2HGDH knockout (KO) in HEK-293T cells. In all instances, at least two independent RNA guides were used per gene, and a minimum of three unique clones examined in downstream assays. Then, we edited the genome of DHL6, a D2HGDH-mutant (1262G/AG, pArg421His) DLBCL cell line, and KO or stably expressed D2HGDH in 5 additional DLBCL cell lines. Quantification of m6A RNA, and 5hmC or 5mC DNA, were completed as we described (Cancer Cell . 2017; Nat Commun . 2015). Histone methylation was defined by western blotting.

Global m6A level was significantly elevated (~40% higher) in HEK-293T cells with D2HGDH or L2HGDH KO when compared to isogenic controls expressing empty lentiCRISPR-v2 (p<0.01, Student's t-test, three biological replicates). Double D2/L2HGDH KO further elevated m6A levels (~80% higher than control cells) suggesting that the accumulation of both D-2/L-2-HG metabolites maximized the competitive inhibition of the α-KG dependent FTO. We also confirmed that single L2HGDH KO significantly decreases 5hmC and increases 5mC DNA marks, plausibly in association with competitive inhibition of α-KG-dependent TET-hydroxylases (p<0.01, Student's t-test, data obtained in triplicate). In DLBCL, we found that D2HGDH KO significantly elevated m6A levels (35% to 70%) in the D2HGDH-WT Ly1 or Ly8 cell lines (p<0.05, Student's t-test, three biological replicates). Further, stable expression of WT-D2HGDH in the D2HGDH-mutant DLBCL cell lines DHL8, WSU-NHL and Ly7 led to a consistent decrease in m6A levels (p<0.05, Student's t-test). Lastly, correcting the heterozygous (R41H) D2HGDH mutation in DHL6 broadly remodeled its epigenome. We found a significant decrease in m6A RNA and 5mC DNA marks, as well as H3K4me3 methylation, in cells expressing the corrected D2HGDH-WT gene than in their isogenic counterparts harboring the R421H mutation (p<0.05, Student's t-test, three biological replicates).

These data establish possible oncogenic consequences of D2HGDH and L2HGDH loss. In particular, we show that: a) D2HGDH integrity regulates RNA methylation in DLBCL; b) deficiency of the related L2HGDH enzyme also influences RNA/DNA methylation and, c) CRISPR-based genome editing firmly establishes that D2HGDH haplo-insufficiency has pathogenetic consequences in DLBCL. We conclude that the imbalance between 2-HG and α-KG rewires the epigenome, and suggest that modulation of the cellular levels of these metabolites may represent a novel epigenetic therapeutic strategy.