Aging-Dependent Large Accumulation of Point Mutations in the Human mtDNA Control Region for Replication

See allHide authors and affiliations

Science  22 Oct 1999:
Vol. 286, Issue 5440, pp. 774-779
DOI: 10.1126/science.286.5440.774


Progressive damage to mitochondrial DNA (mtDNA) during life is thought to contribute to aging processes. However, this idea has been difficult to reconcile with the small fraction of mtDNA so far found to be altered. Here, examination of mtDNA revealed high copy point mutations at specific positions in the control region for replication of human fibroblast mtDNA from normal old, but not young, individuals. Furthermore, in longitudinal studies, one or more mutations appeared in an individual only at an advanced age. Some mutations appeared in more than one individual. Most strikingly, a T414G transversion was found, in a generally high proportion (up to 50 percent) of mtDNA molecules, in 8 of 14 individuals above 65 years of age (57 percent) but was absent in 13 younger individuals.

One postulated cause of aging is the accumulation of mutations in mtDNA (1). This notion is supported by the observation of an aging-related accumulation in human mtDNA of oxidative and alkylation derivatives of nucleotides (2), of small deletions and insertions (2), and of large deletions (3), although their low frequency (<1 to 2%) has raised questions about their functional significance. Furthermore, it has not been clear whether there is an accumulation of aging-dependent point mutations in human mtDNA (4), due in part to the lack of a reliable method for detecting heteroplasmic mutations (that is, mutations that occur together with wild-type mtDNA) and to the search having been largely limited to the protein- and RNA-coding regions of mtDNA.

The main control region of mtDNA [the D-loop and adjacent transcription promoters (DLP) (5)] is the most variable portion of the human mitochondrial genome (6) and may contain heteroplasmic point mutations (7). We therefore analyzed the mtDNA control region with a sensitive method for detection of aging-related heteroplasmic mutations. This approach, which allows the identification of as few as 2 to 4% point mutations while excluding false positives from nuclear mtDNA pseudogenes (8), is based on the use of mtDNA highly purified by cell fractionation and enzyme digestion and of a sensitive technique—denaturant gradient gel electrophoresis (DGGE)—that can identify single-nucleotide mismatches in artificially produced heteroduplexes (9), in combination with cloning, second-round DGGE, and sequencing. The mtDNA-less cell line ρ0 206 (10) was used as a negative control. For DGGE analysis, the main mtDNA control region was subdivided into seven segments, 160 to 440 base pairs in length (DLP1 to DLP7), each with a single uniform melting domain (Fig. 1A) (11). The two segments DLP4 and DLP6, which correspond to one of the hypervariable portions of the main control region (12), were used. DLP4 contains the primary origin of heavy (H)–strand mtDNA synthesis (OH1), while DLP6 contains the two evolutionarily conserved sequence blocks CSB2 and CSB3 (13) and the promoter and start site for light (L)–strand transcription and H-strand replication RNA primer synthesis (14). The block CSB2 carries the homopolymeric tract (HT) D310, a sequence of 12 to 18 Cs interrupted by a T at position 310 (15), which exhibits length variation among individuals (12, 16) and heteroplasmic variation within an individual (7,17).

Figure 1

(A) Map positions of segments into which the main control region of human mtDNA was subdivided for DGGE analysis. OH1: primary origin of heavy (H)-strand synthesis. H1, H2, L: start sites for transcription of rDNA, whole H-strand and light (L)-strand, respectively (5). CSB1, CSB2, and CSB3, conserved sequence blocks 1, 2, and 3 (13). Pro, Phe: tRNAPro and tRNAPhegenes. (B) Effect of different purification steps of mtDNA on PCR amplification with mtDNA- and nuclear gene–specific primers. (a) Untreated total ρ0206 cell DNA. (b) Total ρ0206 cell DNA digested with Bgl II, Dra III, RNase A, and Exonuclease III. (c) Untreated DNA extracted from micrococcal nuclease–treated mitochondrial fraction of ρ0206 cells. (d) Same DNA as in (c), after digestion with Bgl II, Dra III, and Exonuclease III. (e) DNA extracted from MN-treated mitochondrial fraction of fibroblasts from 19-year-old individual, and digested as in (d). The PCR products were run in a 1.5% agarose gel. ATP: α-subunit of ATP synthase. (C) Effects of different enzymatic treatments on total ρ0206 cell DNA (19). R.E.: Bgl II + Dra III restriction enzymes; Exo III: Exonuclease III. M: Hind III–digested λ DNA marker. The variously treated DNA samples were run in a 0.6% agarose gel.

To evaluate the role of nuclear mtDNA pseudogenes in yielding polymerase chain reaction (PCR) products with mtDNA-specific primers, total DNA (18) from ρ0206 cells was tested with primers specific for the mtDNA DLP4 and DLP6 segments, for two mitochondrial tRNA genes [tRNAPhe, tRNALeu(UUR)], for the region containing the origin of mtDNA L-strand synthesis (OL), and, as a control, for three nuclear genes [coding for the 28S rRNA, the p53, and the adenosine triphosphatase (ATPase) α-subunit (19)]. PCR products were obtained with primers for the three nuclear genes, with primers for the two mitochondrial tRNA genes, for OL, and, in small amounts, for DLP4 (Fig. 1B, panel a). These results indicate the presence of nuclear pseudogenes corresponding to these mtDNA segments in ρ0206 cells. Using as a template total DNA from ρ0206 cells in which nuclear DNA and nucleo-cytosolic RNA had been nearly completely digested with Bgl II, Dra III, ribonuclease A (RNase A), and exonuclease III (Fig. 1C) (20), products were still obtained with the primers for the 28S rRNA, p53 and ATPase α-subunit genes, as well as for the two mitochondrial tRNA genes and OL, but no products with the primers for DLP4 and DLP6 (Fig. 1B, panel b). In further tests, DNA extracted from the micrococcal nuclease-treated mitochondrial fraction of ρ0206 cells (18) yielded products only with primers for the tRNAPhe, p53 and 28S rRNA genes (Fig. 1B, panel c). After digestion with Bgl II, Dra III, and Exo III, the same DNA yielded only products of 28S rRNA genes, presumably reflecting the very large number of copies of these genes (Fig. 1B, panel d). MtDNA purified in the same way from fibroblasts of a 19-year-old individual also yielded, aside from the expected PCR amplification products of the five mtDNA segments tested, a small amount of 28S rRNA gene products (Fig. 1B, panel e). This purification method was used for all the analyses below.

The PCR products of the DLP4 and DLP6 segments of untreated total cell DNA and highly purified mtDNA from fibroblast cultures of 18 randomly chosen, genetically unrelated normal individuals between 20-week fetal (20 wf) and 101 years in age and of nine normal individuals twice-sampled, with a 9- to 19-year interval, for the Gerontology Research Center (National Institutes of Health) longitudinal study (LS), were subjected to first-DGGE analysis (21, 22). Figure 2A, panel a, shows the DLP4 profiles of highly purified mtDNA from representative fibroblast cultures, including the two LS5 samples; Fig. 2A, panels b and c, show the DLP6 profiles of untreated total cell DNA (b) or highly purified mtDNA (c) from the same fibroblast samples; and Fig. 2A, panel d, shows the DLP6 profiles of purified mtDNA from fibroblasts of the other eight LS individuals. The DLP4, and especially the DLP6 PCR products from old individuals (>65 y), showed a tendency to exhibit, besides the homoduplex band and a band corresponding to uncrosslinked molecules (UX), one or more slower migrating bands, indicative of sequence variants, which resulted from heteroduplex formation in the final melting and annealing steps (9). Especially significant is the fact that, in five of the nine LS sample pairs (LS1, LS4, LS6, LS7, and LS9), sequence variants of DLP6 appeared only, or were more diverse (in LS6), in the sample taken at a more advanced age.

Figure 2

(A) Ethidium bromide stained gels illustrating the first-DGGE analysis of the PCR products of the DLP4 (a) and DLP6 (b through d) segments of the mtDNA main control region from total-cell DNA (b) and highly purified mtDNA (a, c, d) of fibroblasts from differently aged individuals and of ρ0206 cells. The first lanes show an uncrosslinked sample from 10 y (a, b, and c) and from LS1-52 y (d). UX: uncrosslinked molecules. (B) Scheme of second-DGGE analysis of cloned mtDNA fragments. Shown are the patterns of homoduplexes and/or heteroduplexes expected after hybridization of the PCR products of plasmid DNAs with a randomly chosen cloned fragment (23). (C) Second-DGGE profiles of heteroduplexes between a wild-type cloned fragment and representative sets of DLP6 cloned fragments from highly purified fibroblast mtDNA of 25 y (a), 100(2) y (b) and 100(3) y (c) and of DLP4 cloned fragments from 78(1) y (d). The first lane in each panel shows an uncrosslinked sample from the wild-type reference clone. In panel (c), clones selected as representative of the different types of mtDNA sequence were analyzed in the same gel. In this panel, the random mutations, the number of C residues upstream (u) of T at position 310 in the D310 homopolymeric tract (HT' u. C's) and the nucleotide at position 414 in the various clones are indicated. (D) Scheme of a portion of the mtDNA main control region showing the positions of the specific mutations identified in the present work. The positions of binding of the mtTFA transcription factor (the densely hatched rectangle indicates a position of high affinity binding) and the site of the promoter for L-strand transcription (LSP) are shown. The homopolymer tract D310 is located within CSB 2. OH1, primary initiation site, and OH2, secondary initiation site of mtDNA H-strand synthesis. 383i, T insertion after position 383.

After the first-DGGE assays, the PCR-amplified DLP6 fragments from nearly all purified mtDNA samples and all DLP4 fragments exhibiting abnormal bands were further analyzed. For this purpose, these fragments were cloned in Escherichia coli, and a large number of plasmids (in general, 42 to 48) were isolated from each source, and their DNAs were subjected to PCR amplification and a second-DGGE step, after hybridization with one of them randomly chosen as a reference clone (23). The possible patterns of homoduplexes or heteroduplexes, or both, expected in this hybridization step are illustrated schematically in Fig. 2B (23). All presumptive mutant clones and several of the presumptive wild-type clones identified by the second-DGGE analysis were then sequenced, together with one or more reference clones (24). Furthermore, the PCR-amplified DLP4 fragments of all individuals below 48 years in age were also directly sequenced.

Figure 2C shows second-DGGE profiles produced by heteroduplexes between wild-type mtDNA and representative sets of cloned DLP6 fragments from individuals of different ages: 25 y (panel a), 100(2) y (panel b) and 100(3) y (panel c), and of cloned DLP4 fragments from 78(1) y (panel d). These profiles illustrate the main patterns of heteroplasmic point-mutations that we detected.

Three main types of mutations were revealed by the sequence analysis (25): (i) specific point mutations, present in general in a high proportion of mtDNA molecules; (ii) length and sequence variations of the HT D310 and of a (CA)n repeat; and (iii) random point mutations. Two experiments aimed at estimating the background mtDNA sequence variation associated with the PCR and cloning steps [plasmid controls (26)] revealed a level of heteroplasmy of 0 and 2% (25).

We found an aging-related accumulation of high copy specific point mutations, almost always base substitutions, in the DLP6 region (Fig. 3). This was in contrast with the behavior of the HT D310 or (CA)n length variation, or of random single-base substitutions or deletions that were not restricted to old individuals (Fig. 3) (25). In fact, 170 of a total of 802 DLP6 plasmid clones derived from 10 of the 14 individuals older than 65 years carried a specific point mutation outside of HT D310. Between 5 and 50% of the clones from each individual contained mutations. In contrast, no such mutation was present in DLP6 in a total of 581 plasmid clones from 13 younger individuals (Fig. 3). Some mutations occurred in more than one individual. The most conspicuous example was a T414G transversion in DLP6 clones from fibroblasts of eight genetically unrelated individuals older than 65 y (Fig. 3) (25). No T414G transversion was found in the plasmid control. The mutations T414G, T285C, A368G, and T insertion after 383 were not previously reported (27).

Figure 3

Diagram summarizing the age distribution and frequency of the various types of heteroplasmic mutations detected in the present work. The upper four panels show the DLP6 mutations, the lowest panel, the DLP4 mutations. The dashes below each abscissae axis indicate the individuals analyzed, with the age in years (fw = 20-week fetal). In the uppermost panel, the two samples of each LS pair are indicated by a distinct symbol (*, +, #, ∼).

Strong support for the age-dependency of the accumulation of specific point mutations in DLP6 was provided by the analysis of samples taken twice from the same individuals who were between 15 and 19 years apart in age. In three such studies, a specific mutation (A368G in LS1 and T414G in LS4 and LS7) was absent (LS1, LS4) or present at a marginal level (T414G in LS7) in the earlier sample, but was abundant (23 to 50%) in the later sample. In another study (LS6), the T414G transversion was found in a significant proportion of mtDNA in both the earlier and later samples, although considerably decreased in the latter (28). Two other LS studies (LS3 and LS8) did not reveal any specific mutation in either the earlier or the later sample.

The presence of the A414G mutation described above was confirmed by other methods (Fig. 4). Although not precisely quantifiable, the T414G mutation could be detected when particularly abundant (≥20%), by direct DNA sequencing of the PCR product of DLP6 fragment (estimated proportion 30 to 40% in Fig. 4B), and in all cases—except in one with minimal amount of mutation—by allele-specific termination of primer extension (Fig. 4, C and D) (29). The frequencies of the mutation detected by the primer extension method agree with those determined by DGGE-cloning-sequencing, the tendency toward somewhat lower values presumably reflecting a change in secondary structure of the template caused by the AT to CG base-pair mutation at position 414 (Fig. 4D).

Figure 4

Detection of T414G (A414C) mutation by DNA sequencing and allele-specific termination of primer extension. [(A) and (B)] Sequences of DLP6 fragments amplified from highly purified fibroblast mtDNAs of LS7-70 y and LS7-86(2) y, estimated to contain, respectively, (A) 3% and (B) 23% mutation by the DGGE-cloning-sequencing method (Fig. 3). [(C) and (D)] Primer extension data of a representative set of DLP6 fragments (C), and comparison of frequencies of mutation determined by the primer extension and the DGGE-cloning-sequencing methods (D).

The analysis of the DLP4 clones (Fig. 3) (25) revealed a T152C transition in two individuals, that is, in 23 and 31%, respectively, of the DLP4 clones from 48 y and 100(2) y. The latter individual also exhibited, in the same 31% of its DLP4 clones, a T146C transition and a T195C transition, while 48 y exhibited in all its clones the T195C transition, presumably an inherited polymorphism. The T → C transitions at positions 146, 152, and 195 were previously described as polymorphisms (27). None of the high-frequency base substitutions detected in the DLP4 fragments from the individuals ≥48 years old analyzed here was present in the DLP4 fragments of nine individuals younger than 48 years, as judged from the first-DGGE patterns and from the results of direct DNA sequencing of the PCR-amplified DLP4 fragments (30).

The age distribution and the results of the longitudinal studies indicate that the specific base substitutions are not inherited. A role of nuclear mtDNA pseudogenes in this phenomenon is excluded by several lines of evidence: (i) the failure to obtain any PCR products with primers specific for the DLP6 and DLP4 mtDNA segments from total cell DNA of mtDNA-less ρ0206 cells, digested with Bgl II, Dra III, RNase A, and Exo III; (ii) the identity of the DLP6 first-DGGE patterns obtained from total cell DNA and from purified mtDNA of individual fibroblast samples exhibiting such specific base substitutions (Fig. 2A, panels b and c); (iii) the absence of any of the specific base substitutions in the DLP6 and DLP4 mtDNA fragments from individuals younger than 48 years; and (iv) the results of three longitudinal studies.

The specific point mutations identified here may occur during aging due to oxygen radical–induced mtDNA damage, to mtDNA polymerase errors, or to a phenomenon akin to the bacterial SOS response, that is, activation of genes involved in error-prone DNA repair, recently described also in yeast (31). The mutation may become amplified, because of an intracellular replicative advantage of the mtDNA molecules carrying them, in fibroblasts from old individuals, as previously shown for pathogenic point mutations (32) and deletions (33). The cells with amplified mtDNA may then take over the whole population due to their clonal growth advantage. A similar mechanism has recently been proposed for the homoplasmic or near-homoplasmic somatic mutations in human colorectal tumors (34).

The T414G transversion occurs in the middle of the promoter for mtDNA H-strand replication primer synthesis and L-strand transcription (LSP), at a position immediately adjacent to a segment with high affinity for the mtTFA transcription factor (Fig. 2D) (35). Also the other seven specific mutations observed in old individuals occur at positions critical for mtDNA replication, in particular, either in the coding sequence of the RNA primer for H-strand synthesis, within mtTFA binding segments (36), or very close to the OH1 primary site or to the OH2 secondary site of DNA synthesis initiation (Fig. 2D) (35). These mutations occur in DNA sequences that either unwind and bend because of mtTFA binding at the same or at an adjacent position (35, 36), or form persistent RNA-DNA hybrids giving rise to an R loop with a tRNA-like cloverleaf structure at one or more origins of H-strand synthesis (37). These conformational changes would likely expose single-stranded DNA stretches, which may be more susceptible to oxygen radical damage. Although preliminary results have failed to reveal any relationship between any of these mutations and mtDNA content in the fibroblasts, replicative advantage of one subset of mtDNA molecules need not be accompanied by changes in mtDNA content (32).

  • * To whom correspondence should be addressed. E-mail: attardig{at}


View Abstract

Stay Connected to Science

Navigate This Article