Ultra-deep whole genome bisulfite sequencing reveals a single methylation hotspot in human brain mitochondrial DNA

Epigenetics. 2022 Aug;17(8):906-921. doi: 10.1080/15592294.2022.2045754. Epub 2022 Mar 7.

Abstract

While DNA methylation is established as a major regulator of gene expression in the nucleus, the existence of mitochondrial DNA (mtDNA) methylation remains controversial. Here, we characterized the mtDNA methylation landscape in the prefrontal cortex of neurological healthy individuals (n=26) and patients with Parkinson's disease (n=27), using a combination of whole-genome bisulphite sequencing (WGBS) and bisulphite-independent methods. Accurate mtDNA mapping from WGBS data required alignment to an mtDNA reference only, to avoid misalignment to nuclear mitochondrial pseudogenes. Once correctly aligned, WGBS data provided ultra-deep mtDNA coverage (16,723 ± 7,711) and revealed overall very low levels of cytosine methylation. The highest methylation levels (5.49 ± 0.97%) were found on CpG position m.545, located in the heavy-strand promoter 1 region. The m.545 methylation was validated using a combination of methylation-sensitive DNA digestion and quantitative PCR analysis. We detected no association between mtDNA methylation profile and Parkinson's disease. Interestingly, m.545 methylation correlated with the levels of mtDNA transcripts, suggesting a putative role in regulating mtDNA gene expression. In addition, we propose a robust framework for methylation analysis of mtDNA from WGBS data, which is less prone to false-positive findings due to misalignment of nuclear mitochondrial pseudogene sequences.

Keywords: NUMTs; epigenetics; mitochondria; mtDNA; parkinson’s disease.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Brain
  • CpG Islands
  • DNA Methylation
  • DNA, Mitochondrial* / genetics
  • Humans
  • Parkinson Disease* / genetics
  • Sequence Analysis, DNA / methods
  • Sulfites

Substances

  • DNA, Mitochondrial
  • Sulfites
  • hydrogen sulfite

Grants and funding

This work was supported by grants from The Research Council of Norway (Norges Forskningsråd 288164, ES633272) and Bergen Research Foundation (Bergens Forskningsstiftelse BFS2017REK05).