Evidence Suggesting Absence of Mitochondrial DNA Methylation

Abstract

Methylation of nuclear genes encoding mitochondrial proteins participates in the regulation of mitochondria function. The existence of cytosine methylation in the mitochondrial genome is debated. To investigate whether mitochondrial DNA (mtDNA) is methylated, we used both targeted- and whole mitochondrial genome bisulfite sequencing in cell lines and muscle tissue from mouse and human origin. While unconverted cytosines were detected in some portion of the mitochondrial genome, their abundance was inversely associated to the sequencing depth, indicating that sequencing analysis can bias the estimation of mtDNA methylation levels. In intact mtDNA, few cytosines remained 100% unconverted. However, removal of supercoiled structures of mtDNA with the restriction enzyme BamHI prior to bisulfite sequencing decreased cytosine unconversion rate to <1.5% at all the investigated regions: D-loop, tRNA-F+12S, 16S, ND5 and CYTB, suggesting that mtDNA supercoiled structure blocks the access to bisulfite conversion. Here, we identified an artifact of mtDNA bisulfite sequencing that can lead to an overestimation of mtDNA methylation levels. Our study supports that cytosine methylation is virtually absent in mtDNA.

Keywords: DNA methylation; bisulfite sequencing; epigenetics; mitochondria; whole genome bisulfite sequencing.

FIGURE 1

Inverse relationship between methylation of mtDNA and sequencing depth. The number of reads and methylation percentage is shown in the mitochondrial genome from position 1000–3000 on the sense strand. In red, human HEK293 cells; in green, human primary muscle cells; in blue, mouse skeletal muscle. We analyzed all cytosines in each of the samples corresponding to 6,000–12,000 cytosines.

FIGURE 2

Correlation analysis between sequencing depth and cytosine unconversion rate. WMGBS shows an inverse relationship between mtDNA cytosine unconversion rate (Methylation %) and sequence depth (Number of reads) in both human and mouse cells. Log log scales are shown in the upper right corner. The correlation coefficient is R = –0.907, P < 2.2E-16 and obtained by pooling all samples.

FIGURE 3

BamHI digestion prior to bisulfite sequencing decreases cytosine unconvertion rate. Targeted bisulfite sequencing was used to compare undigested and digested DNA methylation levels at five different regions of the mtDNA from human muscle cells and SKOV3 cells (N = 3). (A) Drawing displays the mtDNA regions investigated by targeted bisulfite sequencing. (B) Percentage methylation for undigested and digested mtDNA. Full circle represents cytosines in CpG context whereas open circle is cytosines in non-CpG context. Results are presented with a min-max interval and a sign test was used to test for significant methylation differences. D-loop (6–298) P = 2.02E-41 (Lonza) and P = 1.40E-24 (SKOV3); D-loop (279–458): P = 5.00E-12 (Lonza) and P = 8.20E-08 (SKOV3); tRNA-F+12S: P = 9.22E-15 (Lonza) and P = 1.29E-9 (SKOV3); 16S: P = 6.50E-05; ND5: P = 1.70E-05 (Lonza) and P = 9.97E-13 (SKOV3); CYTB: P = 2.96E-09 (Lonza), and P = 6.68E-13 (SKOV3). D-loop (6–298) includes origin of replication and tRNA-F+12S includes heavy strand promoter 2. P_H1: heavy-strand promoter 1; P_H2: Heavy-strand promote 2; P_L: Light-strand promoter; OH: Origin of replication from heavy-strand; O_L: Origin of replication from light-strand. ND: not determined.