Abstract
While most research suggests mitochondrial DNA (mtDNA) harbors low or no methylation, a few studies claim to report evidence of high-level methylation in the mtDNA. The reasons behind these contradictory results are likely to be methodological but remain largely unexplored. Here, we critically reanalyzed a recent study by Patil et al. (2019) reporting extensive methylation in human mtDNA in a non-CpG context. Our analyses refute the original findings and show that these do not reflect the biology of the tested samples, but rather stem from a combination of methodological and technical pitfalls. The authors employ an oversimplified model that defines as methylated all reference positions with methylation proportions above an arbitrary cutoff of 9%. This substantially exacerbates the overestimation of methylated cytosines due to the selective degradation of unmethylated cytosine-rich regions. Additional limitations are the small sample sizes and lack of sample-specific controls for bisulfite conversion efficiency. In conclusion, using the same dataset employed in the original study by Patil et al., we find no evidence supporting the existence of extensive non-CpG methylation in the human mtDNA.