Clear cell renal cell carcinomas (ccRCCs) harbor frequent mutations in epigenetic modifiers including SETD2, the H3K36me3 writer. We profiled DNA methylation (5mC) across the genome in cell line-based models of SETD2 inactivation and SETD2 mutant primary tumors because 5mC has been linked to H3K36me3 and is therapeutically targetable. SETD2 depleted cell line models (long-term and acute) exhibited a DNA hypermethylation phenotype coinciding with ectopic gains in H3K36me3 centered across intergenic regions adjacent to low expressing genes, which became upregulated upon dysregulation of the epigenome. Poised enhancers of developmental genes were prominent hypermethylation targets. SETD2 mutant primary ccRCCs, papillary renal cell carcinomas, and lung adenocarcinomas all demonstrated a DNA hypermethylation phenotype that segregated tumors by SETD2 genotype and advanced grade. These findings collectively demonstrate that SETD2 mutations drive tumorigenesis by coordinated disruption of the epigenome and transcriptome,and they have important implications for future therapeutic strategies targeting chromatin regulator mutant tumors.

To address the need to study frozen clinical specimens using next-generation RNA, DNA, chromatin immunoprecipitation (ChIP) sequencing and protein analyses, we developed a biobank work flow to prospectively collect biospecimens from patients with renal cell carcinoma (RCC). We describe our standard operating procedures and work flow to annotate pathologic results and clinical outcomes. We report quality control outcomes and nucleic acid yields of our RCC submissions (N=16) to The Cancer Genome Atlas (TCGA) project, as well as newer discovery platforms, by describing mass spectrometry analysis of albumin oxidation in plasma and 6 ChIP sequencing libraries generated from nephrectomy specimens after histone H3 lysine 36 trimethylation (H3K36me3) immunoprecipitation. From June 1, 2010, through January 1, 2013, we enrolled 328 patients with RCC. Our mean (SD) TCGA RNA integrity numbers (RINs) were 8.1 (0.8) for papillary RCC, with a 12.5% overall rate of sample disqualification for RIN <7. Banked plasma had significantly less albumin oxidation (by mass spectrometry analysis) than plasma kept at 25°C (P<.001). For ChIP sequencing, the FastQC score for average read quality was at least 30 for 91% to 95% of paired-end reads. In parallel, we analyzed frozen tissue by RNA sequencing; after genome alignment, only 0.2% to 0.4% of total reads failed the default quality check steps of Bowtie2, which was comparable to the disqualification ratio (0.1%) of the 786-O RCC cell line that was prepared under optimal RNA isolation conditions. The overall correlation coefficients for gene expression between Mayo Clinic vs TCGA tissues ranged from 0.75 to 0.82. These data support the generation of high-quality nucleic acids for genomic analyses from banked RCC. Importantly, the protocol does not interfere with routine clinical care. Collections over defined time points during disease treatment further enhance collaborative efforts to integrate genomic information with outcomes.