Genome-wide analysis of differential methylation and gene expression in a testicular cancer cell line. Homo sapiens

Aberrant methylation has been postulated to play an important role in tumorigenesis. We report the use of methylated DNA immunoprecipitation (MeDIP) and whole-genome tiling arrays to investigate methylation changes in testicular germ cell tumor (TGCT) cells. Coupled to expression profiling changes, we found that only 22-26% of differentially methylated genes were also expressed differentially. This phenomenon was independent of the presence of CpG islands in the promoter. Differential methylation and expression of some of these genes were confirmed in testicular tumor tissue. A substantial number of differentially methylated regions in the human genome were not linked to annotated gene loci. Subsequent analysis indicated several microRNAs and small nucleolar RNAs were regulated by these differentially methylated regions. Our results demonstrate the power of the combination of MeDIP-chip analysis and expression profiling for discovery in cancer cells of epigenetically regulated genes and non-coding RNAs in cancer cells. Overall design: MeDIP: Genomic DNA was extracted from the cultured TGCT cell line Ntera2 and a normal testis cell line (CRL-7002). Methylated DNA immunoprecipitation (MeDIP) was performed as previously described (Weber, et al, 2005). Methylation-enriched DNA was amplified and hybridized to Human Tiling Array 2.0R chips (Affymetrix). Triplicate sets of hybridization were performed from 3 independent MeDIP experiments for each cell line. The raw data were analyzed by Tiling Analysis Software (TAS). Arrays from each group (cancer vs normal) were quantile-normalized and differential methylation between groups of tumor and normal was compared by choosing the “two-sample comparison analysis” option in TAS. A two-sided test was performed to evaluate both hypermethylation and hypomethylation. Transfrags (or DMRs) were generated by Interval Analysis with a p-value cutoff at 20 (p < 0.01) and a bandwidth of 275. Transfrags generated by p-value cutoff with a positive signal difference were defined as hypermethylation while those of negative difference were defined as hypomethylation. Genomic bisulfite sequencing was performed to confirm the sensitivity of the observed DMRs. Expression profiling: Total RNA was extracted from Ntera2 (cancer) and CRL-7002 (normal) cells. Triplicate sets of hybridization were performed for each cell line. Raw data were normalized and analyzed in Partek Genomics Suite Software. Differential gene expression was evaluated using one-way ANOVA. The supplementary bed files contain 22,452 hypermethylated (GSE15220_Chip_ALL_HYPER.bed) and 12,756 hypomethylated (GSE15220_Chip_ALL_HYPO.bed) regions reported in the paper.