Mitochondrial DNA damage patterns and aging: revising the evidences for humans and mice.

A significant body of work, accumulated over the years, strongly suggests that damage in mitochondrial DNA (mtDNA) contributes to aging in humans. Contradictory results, however, are reported in the literature, with some studies failing to provide support to this hypothesis. With the purpose of further understanding the aging process, several models, among which mouse models, have been frequently used. Although important affinities are recognized between humans and mice, differences on what concerns physiological properties, disease pathogenesis as well as life-history exist between the two; the extent to which such differences limit the translation, from mice to humans, of insights on the association between mtDNA damage and aging remains to be established. In this paper we revise the studies that analyze the association between patterns of mtDNA damage and aging, investigating putative alterations in mtDNA copy number as well as accumulation of deletions and of point mutations. Reports from the literature do not allow the establishment of a clear association between mtDNA copy number and age, either in humans or in mice. Further analysis, using a wide spectrum of tissues and a high number of individuals would be necessary to elucidate this pattern. Likewise humans, mice demonstrated a clear pattern of age-dependent and tissue-specific accumulation of mtDNA deletions. Deletions increase with age, and the highest amount of deletions has been observed in brain tissues both in humans and mice. On the other hand, mtDNA point mutations accumulation has been clearly associated with age in humans, but not in mice. Although further studies, using the same methodologies and targeting a larger number of samples would be mandatory to draw definitive conclusions, the revision of the available data raises concerns on the ability of mouse models to mimic the mtDNA damage patterns of humans, a fact with implications not only for the study of the aging process, but also for investigations of other processes in which mtDNA dysfunction is a hallmark, such as neurodegeneration.

Aging and Disease • Volume 4, Number 6, December 2013 338 lipids, sugars and proteins, thus inducing a pattern of damage that would lead directly to a measurable deficiency in cellular oxidative phosphorylation activity. A variant of the Free radical theory of aging is the Mitochondrial theory [7][8][9][10]. Mitochondria contribute to the majority of ROS generation as a product of electron transfer during oxidative phosphorylation. These organelles contain a second cellular genome, which spans 15.6-kbp in humans, and encodes 13 essential proteins of the respiratory chain, a set of mitochondrial tRNAs and the small and large subunit of the mitochondrial rRNAs [11] ( Figure 1A). MtDNA is located in close proximity to the oxidative phosphorylation machinery, presumably being subjected to more oxidative damage than nuclear DNA. A combination of elevated oxidative damage, possible deficiencies in mismatch and nucleotide excision repair, an excess of direct repeats, and the asymmetrical replication of mtDNA, leaving a considerable portion of the H-strand displaced for an extended period of time, may result in accelerated mutation accumulation [12,13]. This vulnerability of mtDNA leads to the suggestion that it should accumulate mutations progressively during life, and produce cells with a decreased oxidative capacity. As originally formulated, the Free radical and the Mitochondrial theories of aging correspond to stochastic theories that do not take into account the evolutionary process; even if they accommodate well the observations in humans, they fail to explain the variation observed in an evolutionary context. Currently, evolutionary theories of aging are the most accepted ones and basically two models for how aging can evolve have been proposed, the Mutation Accumulation theory [14] and the Antagonistic Pleiotropy theory [15], these last comprising the Disposable Soma theory [16][17][18]. These two evolutionary theories can integrate the observations reported for mtDNA, since in both the accumulation of mutations is predicted. In fact, the Disposable Soma theory closes the gap between stochastic and evolutionary theories of aging, by suggesting that aging results from progressive accumulation of molecular and cellular damage, as a direct consequence of evolved limitations in the genetic settings of maintenance and repair functions [16][17][18].
The investigation of the association between mtDNA damage and age has benefited from the use of animal models, which are invaluable tools for in-depth study of aging in vivo. Mouse models have been frequently used in the understanding of the process of mitochondrial Aging and Disease • Volume 4, Number 6, December 2013 339 alterations associated with age [19]. Murine in general and the mouse in particular share physiological and genomic similarities with humans; the mouse nuclear genome sequence contains about 30000 genes, with 99% having direct counterparts in humans [19]. Similarly to other mammals the mouse mitochondrial genome ( Figure 1B) also exhibits an extensive similarity with its corresponding in humans [20]. However, important differences on what concerns physiological properties, disease pathogenesis as well as life-history exist between mice and humans ( [21]; for a detailed revision see [22]). The extent to which such differences limit the translation from mice to humans of results on the association between mtDNA damage and aging remains to be established. In this paper we revise the studies that analyze the association between patterns of mtDNA damage and aging, and that investigate putative alterations in the number of mtDNA copies, the accumulation of deletions, as well as the accumulation of point mutations. Results reported for humans are compared with those disclosed for mice, and the impact of the discrepancies observed are discussed in terms of the potential informativity of mouse models on what concerns the aging process in humans.

Copy number
Humans Studies performed to date in humans are not consensual, providing distinct insights concerning the relation between aging and mtDNA copy number. All possible Aging and Disease • Volume 4, Number 6, December 2013 340 outcomes are represented in the literature, with studies reporting a clear tendency for the decrease in mtDNA content with age [23][24][25], others reporting an increase of mtDNA amount with age [26][27][28] and even some cases in which no significant change in the absolute mtDNA copy number was observed across lifetime [29][30][31][32] (Table 1A). A possible explanation for the lack of consensus between studies could be the differential methodology used. Realtime PCR (RT-qPCR) is the most widely used technique to detect and quantify mtDNA copy number, which in comparison to alternative methodologies (such as southern blot), should produce the most reliable results (Table 1A). The methodology used, however, cannot be the sole reason for discrepancies observed since RT-qPCR has been used both in studies reporting a decrease of mtDNA copy with age and the absence of significant correlations between these two variables [23][24][25]31]. In the study of Miller et al. [31], for example, these authors used RT-qPCR to quantify mtDNA content and still failed to detect an association between mtDNA copy number and age. The region analyzed for estimation of mtDNA copy number could also affect the interpretation of results, especially if it corresponds to segments susceptible to deletions. However, even when comparing studies using the same regions, such as NADH dehydrogenase subunit 1 (ND1) (which falls outside the "common deletion" region, the 4977 deletion), results are not consensual [23,24,27,28,31] (Table 1A). Differences observed between studies could also be due to the collection, processing and storage of samples. However, all reported studies, with the exception of Cree et al. [23], used fresh tissue, directly frozen and without additional treatment that could interfere with the determination of mtDNA content, a fact that makes discrepancies difficult to justify on the grounds of sample manipulation. The hypothesis that dissimilar results for patterns of mtDNA copy number and age might reflect the behavior of different tissues could also be considered. However, skeletal muscle, which is represented in the majority of studies, does not provide consistent results on what concerns the association between copy number and age [24-26, 28, 29, 31, 32], therefore confirming that the reason for discrepancies cannot be solely attributed to a tissue-specific behavior (Table 1A).

Mice
The analysis of the mtDNA content in mice showed results similar to those reported for humans; the association between mtDNA copy number and age is not clear, and controversial results are present in the literature (Table 1B). Comparing studies, which used the same methodology (RT-qPCR), opposite results are also reported for mice: decrease of mtDNA copy number with age [33] and increase with age [34,35] (Table 1B). To control for a possible effect of the tissue being analyzed, Masuyama et al. [35] reported results for several tissues from mice (brain, heart, lung, kidney, liver, spleen, skeletal muscle and bone), at five different ages (2,4,8,21.7 and 65.2 weeks); their results demonstrated that mtDNA content of the liver decreased with age, whereas it increased in the remaining analyzed tissues. Similar results were reported by Dai et al. [34] who, analyzing cardiac muscle from 170 mice, also reported an increase in mtDNA copy number with age. More controversial results are presented for brain tissues, a fact that could be due to a differential pattern of mtDNA content in the different regions of the brain [33,[35][36][37].  (Table 2). Since only one method of estimation of the amount of deletions (% dmtDNA/total mtDNA) is used both for humans and mice, a compilation of such studies was performed based on this method for further comparison ( Table 2). The comparison was limited to the "common deletion": the 4977-bp deletion in humans and its analogue, the 3867-bp deletion in mice (Table 2, Figure  1). Studies analyzing multiple deletions were also considered in comparisons (Table 2).

Humans
Studies analyzed demonstrated a clear pattern of mtDNA deletions accumulation with age (Table 2A). Although most studies have shown that the overall percentage of mtDNA with deletion is relatively low [27,[38][39][40][41], others demonstrated a high percentage of deleted mtDNA (up to 90% of the total mtDNA [42,43]) (Table 2A). Accumulation of age-dependent mtDNA deletions was found to be tissue-specific and more pronounced in tissues with greater energetic demands, e.g. muscle [42,44] and brain [43,[45][46][47]. The proportion of mtDNA with the 4977-bp deletion was reported to be increased with age in all revised studies, with the exception of Chen et al. [48].
Similarly to the general trend identified for deletions, the accumulation of the mtDNA 4977-bp deletion is tissuedependent, occurring at much higher levels in tissues with Aging and Disease • Volume 4, Number 6, December 2013 341 high oxygen consumption of aged individuals [27,49] than in all other tissues studied, such as liver [27,38,50] (Table 2A). However, Zhang et al. [39] and Lee et al. [27] found a relatively low frequency of the mtDNA 4977-bp deletion in muscle samples of aged individuals. Among brain tissues, the highest amounts of mtDNA 4977-bp deletion were detected in cortex, putamen and substantia nigra (SN), corresponding to the brain areas characterized by a high dopamine metabolism [43,49,51] (Table 2A).
From the areas analyzed, the cerebellum corresponds to the one for which studies report the lowest amount of deletions, or even the absence of deletions [49]. Beside this, it was observed that high levels of deletions were not exclusive for tissues with high oxygen consumption, because a relatively high amount of the mtDNA 4977-bp deletion was detected in hair samples of aged individuals [40]. The timing of appearance of mtDNA deletions also varies among tissues. The earliest mtDNA deletions were reported for humans at 12-48 weeks (skin fibroblasts) [38]. In other tissues the first detection of deletions was recorded at 0.3 years (cardiac muscle) [32], 0.5 years (skeletal muscle) [52], 10 years (brain tissues) [52], 31 years (liver) [53], 0-15 years (buccal swabs) [54] and 60 years (testis) [27] (Table 2A) (Table 2B). The highest percentage of the 3867-bp deletion was observed in mouse brain tissues, as well as in blood [33,56 ]. Two studies [45,55], however, demonstrated low levels of deletions in brain tissues, which can be due to the particular regions of the brain and to differences in the methodology used (as reported above for copy number). Other tissues, such as kidney, also demonstrated low levels of 3867-bp deletion [55]. A relatively high level of deletions was observed in mouse liver [57]. The earliest deletions were reported for blood and brain tissues (8 weeks) [33].

Humans
A clear association between mtDNA point mutations accumulation and age has been reported in almost all studies [58][59][60][61][62][63][64][65][66][67] (Table 3A). In addition to the evidence corroborating the accumulation of point mutations with age, results for a wide variability of tissues are available; these indicate the presence of a differential pattern of mtDNA point mutations accumulation between tissues (Table 3A). Wang et al. [64] analyzed 8 different tissues from 40 individuals and reported that the m.189A>G and m.408T>A mutations have a tissue-specific frequency, presenting their highest values in muscle. On the other hand, the most frequent fibroblast-specific mutation (m.414T>G) has been reported in skin, but not in muscle. Theves et al. [67] analyzed the age-related point mutation m.189A>G in buccal cells as well as in skeletal muscle, concluding that the accumulation of this mutation was higher in the latter tissue (20-50% in the muscle of older individuals; 12.6% maximum in buccal cells).

Mice
Reports on mtDNA point mutations accumulation with age are scarce in mice. Notwithstanding, in almost all reported studies no correlation is observed between mtDNA point mutations and age. In fact, with the exception of results published by Khaidakov et al. [68], no point mutations have been reported in mice [34, 69-71; Ramos et al., unpublished observations] (Table 3B). Khaidakov et al. [68] sequenced the D-loop region from the liver of 8 mice and reported that the presence of point mutations was exclusive of aged mice. These results are contrary to the remaining studies, including those by Song et al. [69], who analyzed the D-loop in 6 different tissues (including liver) and reported the absence of mutations in all analyzed tissues. The lack of mtDNA point mutations in the analyzed strain (C57BL/6) seems to be in accordance with other strains, as reported by Goios et al. [71] and Dai et al. [34]. Moreover, results from Goios et al. [71] include the whole mitochondrial genome, indicating that the absence of point mutations should be a general trait of the mouse (Table 3B).

Discussion
Evidences accumulated over the years strongly suggest that damage in mtDNA contribute to aging in humans. Contradictory data, however, are reported in the literature, with some studies failing to provide support to this hypothesis. Despite the fact that mice have differences in physiological properties, disease pathogenesis as well as life history, when compared to humans, they share genomic similarities and have been extensively used as models of human aging, namely on what concerns the investigation of the association of aging with mtDNA alterations. But are mouse models really relevant to human aging, from the perspective of the investigation of mtDNA damage accumulation? In this review we have analyzed the available studies on the association between mtDNA damage (copy number alterations, accumulation of deletions and of point mutations) and aging, for humans and mice. The analysis performed evidenced the existence of similar patterns for some of the damage indicators, such as deletions, as well as some important differences between humans and mice, namely on what concerns the accumulation of point mutations.
Discrepancies obtained in the studies revised prevent the establishment of a clear association between mtDNA copy number and age. Further studies, using a wide spectrum of tissues and a higher number of individuals would be necessary to elucidate this pattern.
Likewise humans, mice demonstrated a clear pattern of age-dependent and tissue-specific accumulation of mtDNA deletions. Deletions increase with age, and the highest amount of deletions was observed in brain tissues, both in humans and mice. Nevertheless, important tissuespecific differences were observed. In brain, for instance, SN demonstrated one of the highest levels of deletions in humans, whereas in mice this tissue presents one of the lowest levels of deletions. In fact, the study of Guo et al. [45] evidenced this pattern, since SN in aged humans presented 10-fold amount of deletions, as compared to aged mice. Another example of differences among tissues was liver, which had one of the lowest levels of deletions in humans and an opposite pattern in mice. The differences in SN and liver between humans and mice could be due to the existence of differential mechanisms of ageing in these particular tissues. For example, dopaminergic neurons from SN of humans and mice have Aging and Disease • Volume 4, Number 6, December 2013 346 differential patterns of accumulation of neuromelanin, which is thought to induce oxidative stress in mitochondria, with much lower level of neuromelanin accumulation in mice compared to humans (discussed in details in [45]). Differences in liver can also be explained by different ways of master regulatory proteins functioning in human and mouse liver cells, with very small number of genes having identical regulation in the liver of both species. This is supported by the fact that extensive variation between human and mouse hepatocytes have been described, on what concerns the binding sites for highly conserved tissue-specific transcription factors [72].
MtDNA point mutations accumulation has been clearly associated with age in humans, but not in mice. The discrepancies between humans and mice could be at least partially explained by the different lifespans. Mouse has a shorter lifespan, which could be insufficient to accumulate a significant amount of mutations [22]. Moreover, a differential pattern of mitochondrial mutation rate has been reported between humans and mice. In humans, the mutational rate for mitochondrial control region has been estimated empirically as 0.1675 mutations/site/Myr (for a revision of different mutation rates estimated see [73]). By contrast, control region in mouse strain C57B1 has a mutational rate estimated for as 0.056 mutations/site/Myr [71], which is about 3-fold lower than in humans. It should be noted that inferences on the association between mtDNA damage and age in both humans and mice are particularly limited by the fact that the regions analyzed vary between studies, and that few positions are targeted in some of reports. A comprehensive study of the entire molecule, on what concerns the accumulation of point mutations, covering coding and non-coding regions would be crucial in the establishment of definitive conclusions on what concerns the pattern of point mutations accumulation in humans and mice.
The revision of the literature performed in this paper revealed a differential pattern of accumulation of mtDNA somatic mutations between mice and humans, but a similar pattern concerning mtDNA deletions and copy number. This behavior could be explained by their differential process of generation. According to a recent study that made an ultra-deep sequencing of mouse mtDNA [74] and the latest review of mtDNA mutations and free radicals in disease and ageing [75], most somatic mutations are due to errors in mtDNA replication. Regarding mtDNA deletions generation, it seems that the oxidative stress and direct DNA damage are the likely instigators of their formation, being the mtDNA repair the predominant pathway involved in the formation of deletions [76].
Although further studies, using the same methodologies and targeting a larger number of samples would be mandatory to draw definitive conclusions, the revision of the available studies raises concerns on the ability of mouse models to mimic the mtDNA damage patterns of humans, a fact with implications not only for the study of the aging process, but also for investigations of other processes in which mtDNA dysfunction is a hallmark, such as neurodegeneration.