Clonal expansion of early to mid-life mitochondrial DNA point mutations drives mitochondrial dysfunction during human ageing

Abstract

Age-related decline in the integrity of mitochondria is an important contributor to the human ageing process. In a number of ageing stem cell populations, this decline in mitochondrial function is due to clonal expansion of individual mitochondrial DNA (mtDNA) point mutations within single cells. However the dynamics of this process and when these mtDNA mutations occur initially are poorly understood. Using human colorectal epithelium as an exemplar tissue with a well-defined stem cell population, we analysed samples from 207 healthy participants aged 17-78 years using a combination of techniques (Random Mutation Capture, Next Generation Sequencing and mitochondrial enzyme histochemistry), and show that: 1) non-pathogenic mtDNA mutations are present from early embryogenesis or may be transmitted through the germline, whereas pathogenic mtDNA mutations are detected in the somatic cells, providing evidence for purifying selection in humans, 2) pathogenic mtDNA mutations are present from early adulthood (<20 years of age), at both low levels and as clonal expansions, 3) low level mtDNA mutation frequency does not change significantly with age, suggesting that mtDNA mutation rate does not increase significantly with age, and 4) clonally expanded mtDNA mutations increase dramatically with age. These data confirm that clonal expansion of mtDNA mutations, some of which are generated very early in life, is the major driving force behind the mitochondrial dysfunction associated with ageing of the human colorectal epithelium.

Figure 1. Analysis of mitochondrial DNA point mutation frequency with age by Random Mutation Capture (RMC).

(A) Schematic diagram describing the RMC methodology. (i) Schematic diagram of the structure of the human colorectal crypt. (ii) Schematic diagram showing mtDNA isolation. Colonoscopic biopsies are homogenized and the mitochondrial fraction isolated by differential centrifugation. MtDNA is then prepared by phenol/chloroform extraction and quantified using real-time PCR (standard curve method). (iii) MtDNA is digested for 10 hours with Taq1α. PCR is then carried out over the restriction site. Only molecules with mutations in the restriction site are able to successfully amplify. (iv) Agarose gel showing PCR products from a typical RMC run. Each reaction contained ∼10000 target base pairs. 488 base pair bands show amplified mutated molecules (wells 4,6,13,16 and 20). The wild-type control well (Wt) shows complete digestion of wild-type DNA following PCR. (v) Example electropherograms showing mutations (asterisks) within the restriction site (highlighted in blue). (B) Frequency of all RMC detected mtDNA mutations in human colorectal mucosa (n = 207). There was no correlation between mtDNA mutation frequency and age (Pearson correlation = 0.127 (P = 0.07)). (C) Data from (B) presented on a log 10 scale to show the spread of the data. Note that the zero values cannot be displayed in this way therefore n = 175. (D) Frequency of all mtDNA mutations detected in human colonic mucosa, grouped by decade. Subjects were grouped as follows, 17–26 (n = 12), 27–36 (n = 19), 37–46 (n = 58), 47–56 (n = 51), 57–66 (n = 43), 67–77 (n = 23). There were no significant differences between any of the groups (P = 0.343, One Way ANOVA).

Figure 2. Schematic diagram showing the detection limits of each of the techniques utilised in this study.

Figure 3. Mitochondrial DNA mutations in human colorectal epithelium of subjects below 26 years of age (n = 8) and over 70 years of age (n = 8) measured by Next Generation Sequencing (NGS).

(A) MtDNA mutation frequency. There was a significantly higher mutation frequency in the subjects >70 years (p = 0.0361, unpaired t-test). (B) Types of mutations detected by NGS frequency. There was no significant difference in the types of mutations detected in subjects <26 and >70 years of age (p = 1.00, Fisher's exact test).

Figure 4. Exclusion of mitochondrial DNA (mtDNA) mutations occurring in the germline or in early embryogenesis.

(A) Somatic mtDNA frequency (mtDNA mutations present in colon only) measured by NGS. There was a significantly higher mutation frequency in the subjects >70 years (P = 0.0351, unpaired t-test). (B) Percentage of synonymous/polymorphic variants and non-synonymous mtDNA mutations which were of germline or early embryological origin compared to those which were somatic in adults <26 years of age and over 70 years of age. There was a significantly lower frequency of non-synonymous mtDNA mutations in the somatic mtDNA mutation groups compared to the germline or early embryological mtDNA mutation group (p = 0.041 Fisher's, exact test).

Figure 5. Quantification of mitochondrial dysfunction in human colonic crypts by cytochrome c oxidase (COX)/succinate dehydrogenase (SDH) histochemistry.

(A) COX/SDH sequential histochemistry on a transverse section of human colorectal mucosa. The crypts which stain brown have functional COX and SDH activity; those which are blue have lost COX activity but retain SDH activity. (i) shows an image from a subject aged 25 in whom no COX deficient crypts were detected. (ii) shows an image from a subject aged 72 in whom 28% of crypts were COX deficient (B) COX activity was measured in colorectal biopsies from 207 subjects. A crypt was defined as deficient if more than 50% of cells had lost COX activity. There was a significant correlation between age and the percentage of COX deficient crypts. (Pearson correlation = 0.603, p<0.0001). (C) Correlation analysis between somatic mtDNA mutation frequency determined by NGS and percentage COX deficient crypts showed a significant correlation (Pearson correlation = 0.511, p = 0.043). (D) Correlation analysis between germline mtDNA mutation frequency determined by NGS and percentage COX deficient crypts showed no significant correlation (Pearson correlation = 0.369, p = 0.176). (E) Correlation analysis between the frequency of random mtDNA mutations by RMC and percentage COX deficient crypts showed no significant correlation (Pearson correlation = 0.007, p = 0.918). (F) Correlation analysis between the frequency of random mtDNA mutations by RMC and those by NGS in paired samples showed no significant correlation (Pearson correlation = 0.145, p = 0.381).