Genome-scale DNA methylation mapping of clinical samples at single-nucleotide resolution

Abstract

Bisulfite sequencing measures absolute levels of DNA methylation at single-nucleotide resolution, providing a robust platform for molecular diagnostics. We optimized bisulfite sequencing for genome-scale analysis of clinical samples: here we outline how restriction digestion targets bisulfite sequencing to hotspots of epigenetic regulation and describe a statistical method for assessing significance of altered DNA methylation patterns. Thirty nanograms of DNA was sufficient for genome-scale analysis and our protocol worked well on formalin-fixed, paraffin-embedded samples.

Figure 1. Optimizing bisulfite sequencing for genome-scale profiling of human disease samples

(a) Typical RRBS coverage of gene promoters (2-kilobase regions centered on RefSeq-annotated transcription start sites), CpG islands (annotated with a stringent version of the Gardiner-Garden criteria, requiring a minimum length of 700 basepairs) and putative regulatory elements (mapped by DNase hypersensitivity). All data are for colon tumor, run no. 1 in Table 1. The values refer to the number of individual CpG measurements in the region that pass quality control. Correlation between (b) RRBS performed on 30 ng and 100 ng of input DNA derived from mouse ES cells and (c) DNA methylation measurements for colon tumor, obtained by RRBS and Infinium, at 1,027 high confidence CpGs (RRBS: sequencing coverage ≥ 20, Infinium: detection p-value < 0.05). (d) Distribution of DNA methylation in the promoter regions of SOX17 (hypermethylated in the colon tumor) and HNF4A (hypomethylated). Unmethylated reads are shown in red, partially methylated reads in grey and methylated reads in blue.