Genome wide DNA methylation profiling of normoxic and hypoxic non-small-cell lung cancer samples for 5mC and 5hmC. The Illumina Infinium 450k Human DNA methylation Beadchip v1.2 was used to obtain DNA methylation and hydroxymethylation profiles across 485,512 CpGs from DNA extracted from fresh-frozen tumor samples. Samples included 12 hypoxic and 12 normoxic tumor samples, with hypoxia determined according to the hypoxia metagene score (Buffa et al, Br J Cancer 2010). To profile hydroxymethylation, 5hmC was glycosylated and 5mC was oxidised as described by Yu and colleagues (Nat Protoc 2012), and hydroxymethylation and methylation were differentially profiled according to the Nazor and colleagues (Genomics 2014). Hypermethylation of tumor suppressor gene (TSG) promoters confers growth advantages to cancer cells, but how these changes arise is poorly understood. Here, we report that tumor hypoxia reduces the activity of oxygen-dependent TET enzymes, which catalyze DNA de-methylation through 5-methylcytosine oxidation. This occurs independently of hypoxia-associated alterations in TET gene expression, basal metabolism, HIF activity or nuclear reactive oxygen species, but directly depends on oxygen shortage. Hypoxia-induced loss of TET activity increases hypermethylation at gene promoters in vitro, while also in patients, gene promoters are markedly more methylated in hypoxic than normoxic tumors. Affected genes are frequently involved in DNA repair, cell cycle regulation, angiogenesis and metastasis, indicating cellular selection of hypermethylation events. Overall, up to 50% of the tumor-associated hypermethylation is ascribable to hypoxia across various cancer types. Accordingly, spontaneous murine breast tumors become hypermethylated when rendered hypoxic through vessel pruning, whereas vessel normalisation rescues this effect. Tumor hypoxia thus acts as a novel regulator underlying DNA methylation.
Overall design
12 normoxic and 12 hypoxic non-small-cell lung cancer samples were compared for their 5hmC and 5mC distribution
Less...
More...