Methylation on CpG residues is one of the most important epigenetic modifications of nuclear DNA, where it regulates gene expression. Methylation of mitochondrial DNA (mtDNA) has been studied using whole genome bisulfite sequencing (WGBS), but recent evidence has uncovered major technical issues, which introduce a potential bias during methylation quantification. In this study, we first validate the technical concerns with WGBS using publicly available datasets. Then we develop and assess the accuracy of a protocol for variant-specific methylation identification using long-read based technology Oxford Nanopore Sequencing. Our approach circumvents mtDNA-specific confounders, while enriching for native full-length molecules over nuclear DNA. Variant calling analysis against Illumina deep re-sequencing showed that all expected mtDNA variants can be reliably identified. By using simulated data sets, we were able to determine that the mtDNA methylation levels identified were likely false positives introduced by the technique. This observation was consistent across the multiple human primary and cancer cell lines and human tissues analysed in this study.

We therefore conclude that CpG methylation is not an epigenetic modification occurring in human mtDNA, thus resolving previous controversies. Additionally, we developed a reliable protocol to study epigenetic modifications of mtDNA at single-molecule and single-base resolution, with potential applications beyond CpG methylation