A catalog of homoplasmic and heteroplasmic mitochondrial DNA variants in humans

Abstract

Robust characterization of mitochondrial variation provides an opportunity to map regions under high constraint, and identify essential functional domains. We sequenced the mitochondrial genomes of 196,554 unrelated individuals, and identified 15,035 unique variants. We found that 47% of the mitochondrial genome was invariant across the population, and generated a map of constrained mitochondrial regions. We find that the longest intervals in the mitochondrial genome without any variant were in the two rRNA genes (26 of 40 intervals >10bp long). We also showed that the 13 protein-coding genes in the mitochondrial genome did not tolerate loss-of-function variants. The only frameshift or nonsense variant observed at homoplasmic levels was a nonsense at the start codon of MT-ND1 , which may be rescued by the methionine at amino acid position 3. Lastly, we applied these data to variants reported to be pathogenic for Leber's Hereditary Optic Neuropathy (LHON). We found that 42% of variants (19 of the 45) reported to be pathogenic have a frequency above the maximum credible population allele frequency for an LHON-causing variant, including the primary LHON mutation m.14484T>C, which suggests that m.14484T>C cannot be causing LHON by itself. This result showed that allele frequency information across a large unselected population is important to assess the pathogenicity of variants in the context of rare mitochondrial disorders. We made HelixMTdb -the list of variants and their allele frequency in 196,554 unrelated individuals --publicly available.

Respiratory complex and tissue lineage drive mutational patterns in the tumor mitochondrial genome

“…In total, 11.9% of tumors across all cancers (95% CI: 11.0-12.9%) harbored a truncating mtDNA variant absent in the patient's matched normal sample. In contrast, only 0.15% of normal blood samples exhibited a truncating variant (95% CI: 0.13-0.17%) based on a recent analysis of ~200,000 mtDNA genomes 14 (Fig. 1i).…”

“…Nuclear somatic mutations for TCGA samples were obtained from the MC3 MAF, subset for the samples for which mtDNA whole-exome sequencing BAMs were available. Finally, mtDNA mutation calls for 195,983 normal samples were obtained from the HelixMTdb cohort of sequenced saliva samples from healthy individuals 14 .…”

A novel phylogenetic analysis and machine learning predict pathogenicity of human mtDNA variants

“…Along with the mutations tabulated in MITOMAP, we also included mtDNA changes found in the recently established HelixMT database (HelixMTdb). This database was generated during a genetic survey, performed agnostic to phenotype, of nearly 200,000 independent human samples (Bolze et al 2019). Positive training sets for our SVM consisted of protein or tRNA mutations listed as confirmed pathogenic alterations within the MITOMAP database.…”

“…MITOMAP data used in this study (Lott et al 2013) were downloaded on October 1, 2019. HelixMTdb data used in this study (Bolze et al 2019) were downloaded on October 15, 2019.…”

Risk of mitochondrial deletions is affected by the global secondary structure of the human mitochondrial genome

“…) . Recent database(Bolze et al 2019) containing the maximal up to date(196,554) number of complete mitochondrial human genomes confirms rareness of the disrupted common repeat: the most common disrupted variant is m.8473T>C (1%: 2118 out of 196,554) and the second most common variant is m.8472C>T (0.4%: 833 out of 196,554). It is interesting to mention that both frequent synonymous variants m.8473T>C and m.8479A>G are associated with increased longevity.…”