The aim of the present study was to reconsider an intriguing observation: a strikingly high frequency of the C150T mutation in mtDNA CR in centenarians . The observation was at odds with the consolidated idea that heteroplasmy is detrimental for attaining longevity, considering that a variety of mtDNA deletions and mutations accumulate with age  and that mitochondrial function declines with age . The novelty of our approach was to compare mtDNA CR heteroplasmy in descendants of centenarians (both offspring and nephews/nieces) and age-matched controls unrelated to centenarians. The analysis of offspring of centenarians is a valuable tool in searching for susceptibility genetic factors in longevity ; however, it has never been used to investigate a putative role of mtDNA somatic variability on longevity.
The first requirement for our study was to set up a fast and reliable method to screen mtDNA heteroplasmy, and DHPLC met our requirement. Till now DHPLC has been applied to detect mutations on the entire mtDNA molecule in samples of rather limited size [12–16]. In our case, we needed to carry out quantitative comparisons of the level of heteroplasmy in a sole mtDNA region, the control region, but in a large population sample (414 subjects in total). The DHPLC protocol we set up provided reliable results (see standard deviations of the reference curve in Fig. 1) and was reasonably sensitive.
A further critical point was to establish that the PCR protocol was specific for the amplification of the 16531 nt-261 nt mtDNA fragment. In fact, it was recently shown that the pseudo-mitochondrial genome can induce errors in heteroplasmy interpretation . The negative results we obtained both by processing rho-zero cells and by a BLAST search excluded that nuclear pseudogenes contaminated mtDNA PCR amplifications.
The most important finding presented here is that the patterns of mtDNA CR heteroplasmy do not differ between centenarians and their descendants, but differ between relatives of centenarians and age-matched controls (Fig. 2). This result ruled out that heteroplasmy was exclusively due to age-related stochastic mutations, however indicated that it was genetically controlled. In agreement, the levels of heteroplasmy were significantly correlated in parent-offspring pairs (Fig. 3). The observation that the correlation was significant in mother-offspring pairs while not in father-offspring pairs (Fig, 4a and 4b) suggested that a possible genetic control on heteroplasmy was due to the mitochondrial genome. However, other clues indicated a different, and probably more complicated, genetic control pattern. First, the lack of association between mtDNA haplotypes and heteroplasmy (Tables 2 and 3) rules out that there are mtDNA molecules more prone than others to somatic mutations. Second, both the tissue specificity of mtDNA CR point mutations  and the concordance of heteroplasmy higher in monozygotic than in dizygotic twins  denote that heteroplasmy is not related to the mtDNA haplotype the offspring inherited from the mother. Thus, although the genetic mechanism modulating the occurrence/accumulation of the mtDNA CR heteroplasmy needs further work to be elucidated, all the data suggest the involvement of nuclear sex specific factors.
It should also be noted that the heteroplasmy revealed by DHPLC does not refer to the sole C150T variability, but to additional possible mutations occurring in the entire 16531 nt-261 nt mtDNA fragment. In fact, the C150T mutation was found to be present in 10 out of 16 subjects (Table 3), while the remaining subjects showed other heteroplasmic mutations . Interestingly, most of the observed heteroplasmic positions were either replication origins (position 146, see Ref. ) or contiguous to replication origins (positions 150 and 152 that flank the 151 replication origin; position 189 which is 2 bp from the 191 replication origin). Since the C150T transition is able to provide alternative replication origins , a similar effect could be hypothesized for the other mutations.
The results reported in Fig. 2, show that mtDNA CR heteroplasmy cannot be accounted for only by to age-related stochastic mutations. What is more, the finding that mtDNA CR heteroplasmy is greater in descendants of centenarians than in age-matched controls suggests a beneficial role of mtDNA heteroplasmy for attaining longevity. In fact, several data show that the offspring of centenarians have a better chance to attain longevity than the general population [20, 21]. How could this apparent paradox be explained from a biological point of view? The well known mitochondrial theory of ageing proposes that age-associated mitochondrial dysfunction is a consequence of age-associated accumulation of somatic mutations in the mtDNA population. However, recent findings suggest that at least some aspects of the above theory require reconsideration . In fact, a key for explaining the paradox that mtDNA heteroplasmy could be beneficial for longevity may be the new emerging concept of mitochondria complementation, which suggests that human cells are protected from mitochondrial dysfunction by complementation of mtDNA products in fused mitochondria . The beneficial effect of complementation may be enhanced by efficient mtDNA replication, as provided by CR mutations which introduce alternative replication sites. In fact, multiple replication origins falling in this DLoop region could play a major role in accelerating mtDNA synthesis to satisfy developmental, physiological, or aging-related demands . However, neither the replicative advantage of some variants nor the mitochondrial complementation can explain, by themselves, the heteroplasmy patterns of Fig. 2. By contrast, it is likely that the interplay among new replication origins, mitochondrial complementation and nuclear factors might provide an advantage for pursuing longevity by counteracting age-related mitochondrial damages. In this frame, the subjects who are genetically predisposed to mtDNA CR heteroplasmy would be clearly favoured in the demographic selection as defined by Perls et al. .