MitoRCA-seq reveals unbalanced cytocine to thymine transition in Polg mutant mice

Mitochondrial DNA (mtDNA) encodes important genetic information in addition to the nuclear genome. Unlike the diploid nuclear genome, each mammalian cell contains thousands of copies of mtDNAs. The heteroplasmic nature of mtDNA mutations make their analysis more challenging1,2. Mutations in mtDNAs, especially the high-frequency point mutations, have been linked to more than 100 human disorders3,4, including cardiovascular disease, diabetes, deafness and Parkinson’s disease. Emerging studies have begun to characterize the significance of low-frequency mtDNA mutations in tumor progression and aging2,3,5,6,7. Moreover, recent studies indicated that low-level mtDNA heteroplasmy can affect the progeny’s phenotypes through unbalanced transmission between generations7,8. Thus, there is a need for a reliable method to detect low-abundance point mutations in the mitochondrial genome. Recent advances in high-throughput sequencing technology had led to several strategies for global identification of mtDNA mutations9. In these methods, mitochondrial DNA fragments are enriched by direct isolation of mitochondrion1, multiplex Polymerase Chain Reaction (PCR)2,10,11,12 or capture-based approach with mtDNA-specific oligonucleotides2,11,13,14. However, each approach has its own limitations. Isolation of mitochondria is labor-intensive and not practical for large number of samples or circumstances with very limited material, for which mitochondria cannot be isolated (e.g. paraffin-embedded tissues). One major challenge for PCR and oligonucleotide-capture based methods is the contamination of Numts, the DNA sequences in the nucleus with high homology to mtDNA15,16. There are hundreds of Numts in the mouse and human genomes, overlapping with 84.5% and 99.99% of the mtDNA sequences in the respective organism (see Supplementary Methods for details). The presence of Numts obscures the identification of real point mutations in mtDNA especially those occurring at a low frequency11,14,17. The mice expressing a proofreading-deficient version of the mitochondrial DNA polymerase γ (POLG) provide an excellent model to study low-frequency mtDNA mutations and the related phenotypes (e.g., decreased body weight, kyphosis, reduced hair density and induction of apoptosis)5,18,19,20. By mutating a critical aspartate residue in the POLG exonuclease domain to alanine (D257A), the mtDNA polymerase γ showed reduced proofreading activity during mtDNA replication5,20. Sanger sequencing results showed that mice containing the homozygous Polg mutated allele (denote Polg mutant hereafter) have a higher load (~3–5 times) of somatic mtDNA point mutations for Cytb (Cytochorme b) gene and the control region (or D-loop region) in brain, liver and heart5. To obtain a full spectrum of somatic mutations for mitochondrial genome, Ameur et al. sequenced the whole mtDNA region from Polg mutant mice by SOLiD platform1. They found homozygous Polg mutant mice have a median point mutation load of 12 × 10−4 per site, while the wild-type mice have mutational load of 1.3–1.8 × 10−4. Mutational load can be separated into the number of mutation sites and the frequency at individual sites. When 0.5% was used as the minimum mutation frequency, Ameur et al. found homozygous Polg mutant mice have more mutation sites (~5.3–6.6 times) compared with wild-type mice at 30–40 weeks of age1. Although Trifunovic et al. had examined 2-month old Polg mutant and wild-type mice for mutations in the Cytb gene and control region5, it remains unclear whether there is a global change in mitochondrial somatic mutation spectrum before the premature ageing phenotypes becoming apparent at ~25 weeks. If so, how this might contribute to the development of premature ageing phenotypes at the molecular level? In this study, we developed mitoRCA-seq, a simple and robust mitochondrial DNA sequencing procedure, to compare the mutational spectrum of 6-week old wild-type and Polg mutant mice (early stage before premature phenotypes appear). MitoRCA-seq employs rolling circle amplification (RCA), which has two advantages: (i) It amplifies full-length mitochondrial DNA and thus minimize Numt contamination; (ii) mtDNA can be enriched from as little as 1 ng of total DNA, which makes it suitable for broad applications. The reliability of mitoRCA-seq is demonstrated by profiling (i) different amount of total DNA (1 ng, 5 ng and 50 ng) or (ii) technical replicates starting from different RCA reactions. Strikingly, we found that mutational load is increased in the Polg mutant mice, by expanding the number of mutation sites and elevating mutation frequency at the existing sites, at the age of 6 weeks. The time point is much earlier than the 25-week old when the premature ageing phenotypes start to appear5. Further analyses revealed that cytocine (C) to thymine (T) transitions are overrepresented among all point mutations, and specific to Polg mutant mice in a context dependent manner. We further demonstrated that C → T transitions often lead to an increase in the hydrophobicity of the underlying amino acids. These findings suggest that mutational events are ahead of the cellular changes and accumulation of mtDNA mutations might ultimately contribute to premature ageing phenotype.