Judd M. Aiken, Department of Animal Health and Biomedical Sciences, University of Wisconsin-Madison, Madison, WI 53706, USA. Tel. (608) 262 7362; fax: (608) 262 7420; e-mail: firstname.lastname@example.org
The hypothesis that mitochondrial DNA damage accumulates and contributes to aging was proposed decades ago. Only recently have technological advancements, which facilitate microanalysis of single cells or portions of cells, revealed that mtDNA deletion mutations and, perhaps, single nucleotide mutations accumulate to physiologically relevant levels in the tissues of various species with age. Although a link between single nucleotide mutations and physiological consequences in aging tissue has not been established, the accumulation of deletion mutations in skeletal muscle fibres has been associated with sarcopenia. Different, and apparently random, deletion mutations are specific to individual fibres. However, the mtDNA deletion mutation within a phenotypically abnormal region of a fibre is the same, suggesting a selection, amplification and clonal expansion of the initial deletion mutation. mtDNA deletion mutations within a muscle fibre are associated with specific electron transport system abnormalities, muscle fibre atrophy and fibre breakage. These data point to a causal relationship between mitochondrial DNA mutations and the age-related loss of muscle mass.
Mammalian mitochondrial DNA (mtDNA) is a double-stranded, circular genome that encodes 13 polypeptides of the electron transport system (ETS), two ribosomal RNAs and 22 transfer RNAs. The major non-coding region of the mitochondrial genome is the displacement loop (control region), which contains regulatory elements for the replication and transcription of the mitochondrial genome. The mtDNA is intertwined, in the matrix of the inner membrane of the mitochondria, in close proximity to the ETS (Clayton, 1982). The ETS is dependent on the mtDNA for many of its subunits but also produces free radicals which can react and damage nearby molecules. The mtDNA is thought to be particularly susceptible to oxidative damage resulting from its proximity to the free radical source as well as its relative lack of protective proteins. The DNA repair systems are different to those in the nucleus. The direct restitution of carbon-centred free radicals in the DNA backbone by sulfhydryls, described by Henle & Linn (1997), have not been reported in the mitochondrion. Base excision repair, nucleotide repair and recombinational activity have all been identified in the mitochondrion (Thyagaragjan et al., 1996; LeDoux et al., 1999). Since only 5.5% of the mtDNA is non-coding (the 880 bp control region), damage is likely to involve genes coding for the polypeptides, rRNAs and/or tRNAs, resulting in mutational dysfunction.
High levels of mtDNA mutations cause disease
Mitochondrial myopathies are diseases caused by both single nucleotide and deletion mutations. Although many of these diseases result from direct mutation events in the mitochondrial genome, for others, the mitochondrial mutations have been linked to mutations in the nuclear genome (Hirano & DiMauro, 2001). These mitochondrial diseases provide important theoretical and experimental foundations for current mtDNA and aging studies. Many phenotypes associated with mtDNA mutations were first observed in mitochondrial myopathies. These myopathies present with clinical symptoms that include: psychomotor retardation, dementia, ataxia, stroke-like episodes and deafness. At the biochemical level, mitochondrial myopathies affect the ETS in post-mitotic tissues such as nerves and muscles (Harding & Holt, 1989). Muscle biopsies from myopathy patients often display specific fibres that stain in a distinct manner with Gomorri trichrome stain, termed ‘ragged-red’ fibres (RRF). Genetically, these diseases are caused by mtDNA mutations.
Mitochondrial myopathies have two distinct genotypes: mtDNA point mutations and mtDNA deletion mutations. Point mutations in the mitochondrial DNA genome result in diseases such as Leber's hereditary optic neuropathy (LHON), myoclonic epilepsy with ragged red fibre (MERRF) and mitochondrial encephalomyopathy, lactic acidosis and stroke-like episodes (MELAS). Mitochondrial myopathies caused by point mutations, unlike most caused by deletion mutations, are mostly inherited (Shoffner & Wallace, 1994). Most pathogenic point mutations of mtDNA are located in tRNA or rRNA genes, globally affecting mitochondrial protein synthesis. Less common are point mutations in protein-coding genes that contribute to specific defects in the ETS.
mtDNA deletion mutations cause Kearns–Sayre syndrome (KSS) and progressive external opthalmeplegia (PEO). Many of the deletion mutation myopathies contain the same large-scale mtDNA deletion. This common deletion results in the loss of 4977 bp of mtDNA and is referred to as ‘mtDNA4977’. mtDNA4977 occurs between positions 8470 and 13477 on the human mitochondrial genome resulting in the loss of subunits of ETS complexes I & IV as well as many tRNAs.
The mutational threshold has been determined for several of the mitochondrial myopathies. In general, low levels at heteroplasmy (i.e. a mixture of wildtype and mutated mitochondrial genomes in the same cell or fibre) do not result in disease. With many of the myopathies, it has been demonstrated that the levels of wildtype genome above 10% are sufficient to maintain mitochondrial function. For example, Attardi's group has shown, in cell lines, that the threshold for a specific MELAS mutation is 90%. Below this level, respiratory rate is completely normal; above this level, the respiratory rate decreased rapidly (Chomyn et al., 1992; Bentlage & Attardi, 1996). The threshold appears to be different for different mitochondrial mutations and for different tissues.
Myopathies represent one end of the spectrum with respect to the impact of mitochondrial DNA mutations. In these diseases, mtDNA mutations have accumulated to sufficiently high levels in cells and tissues to have profound physiological impacts. The cellular and biochemical phenotypes associated with these types of mutations are now being used to locate and define mitochondrial mutations that accrue with age. Since myopathies are primarily associated with post-mitotic tissues (DiMauro & Schon, 2001), much of the focus on age-related accumulation of mtDNA mutations has focused on these tissues.
mtDNA point mutations: do they accumulate with age?
The search for single nucleotide changes is technically challenging since the point mutations can, potentially, occur anywhere on the 16-kb genome. Initial efforts emphasized single nucleotide changes associated with mitochondrial myopathies and the point mutations characteristic of MERRF and MELAS were identified in normally aging humans (Muenscher et al., 1993; Zhang et al., 1993; Liu et al., 1997). Although these studies produced variable results, point mutations are present at very low levels (0.04–2.2%) compared with levels in myopathy patients, in tissue homogenates from various tissues of aging individuals (Muenscher et al., 1993; Liu et al., 1997).
When single cells were analysed, point mutations were observed to accumulate to high levels in an age-dependent and tissue-specific manner (Clark et al., 1999; Taylor et al., 2001; Nekhaeva et al., 2002). Several specific point mutations occur at high levels (up to 50%) in skin fibroblasts from individuals older than 65 years (Michikawa et al., 1999). Gerhard et al. (2002) did not, however, detect any single nucleotide mutations in similarly cultured skin fibroblasts. The point mutations, A189G and T408A, were identified in skeletal muscle. The fibroblast-specific T414G was absent in muscle and heart in several studies (Calloway et al., 2000; Wang et al., 2001; Nekhaeva et al., 2002) yet another study found T414G in muscle but not in brain (Murdock et al., 2000). A more recent study has linked reduced COX activity to an increasing burden of multiple mtDNA point mutations, rather than a single specific point mutation, present in the genes coding for COX (Lin et al., 2002). The mechanism by which point mutations accrue is not clear. A replicative advantage of mitochondria containing MELAS point mutations was demonstrated in cell line studies (Yoneda et al., 1992).
Although point mutations appear to increase with age, the lack of overt phenotypes has hampered attempts to relate these mutations to aging phenomenon. Cellular impacts of age-associated point mutations in mtDNA have not been reported. The point mutations in the control region of mtDNA may have little impact on cellular physiology, so cells could accommodate and accumulate them to high levels. If single cells can accumulate high levels of point mutations and still function within normal parameters without expressing abnormal phenotypes, point mutations may be an age-associated phenomenon but not a causal link to aging.
Age-associated mtDNA deletion mutations
Initial attempts to determine whether mtDNA deletion mutations accumulated with age were performed using Southern blot hybridization analyses for detection. These experiments were unsuccessful due to the low abundance of these deletion mutations in the tissue homogenates. The advent and application of the polymerase chain reaction (PCR), combined with focusing on the common mtDNA4977 deletion, facilitated identification of this deletion product in many different tissues in normally aged individuals (Cortopassi & Arnheim, 1990; Linnane et al., 1990). The highest levels of this age-associated deletion mutation are observed in post-mitotic tissues with high-energy requirements such as heart, skeletal muscle and brain. It soon became clear that deletion mutations could occur at many sites in the mitochondrial genome (Melov et al., 1995). Multiple age-associated deletion products have been found in a variety of tissues from mice, rats, rhesus monkeys and humans (Cortopassi & Arnheim, 1990; Linnane et al., 1990; Cortopassi et al., 1992; Lee et al., 1993, 1994; Chung et al., 1994; Tanhauser & Laipis, 1995; Van Tuyle et al., 1996). One characteristic of all mtDNA deletion mutations is that the deleted region is large, ranging in size from ∼4 kb to > 9 kb. The majority of the deletion mutations occur within the major arc of the mitochondrial genome and generally result in the loss of subunits of ETS complexes I, IV and V as well as numerous tRNA genes. In humans, rhesus monkeys and, to a lesser extent, rodents, some deletion mutations, the ‘common deletions’, were found to occur more frequently. Sequence analysis of the mtDNA deletion breakpoints indicate that direct repeat sequences are present at the breakpoints of the ‘common deletions’ but not at the breakpoints of ‘unique deletions’ (Chung et al., 1996).
Although these initial studies demonstrated an age-associated accumulation of small mitochondrial genomes, the requisite use of PCR led to two important considerations. First, the use of PCR raised concerns regarding the nature of the target sequence. Although the PCR reaction amplified small products that were expected if deletion mutations were present, other interpretations are possible. For example, the binding of one oligonucleotide primer to a second (unexpected) site on the genome (i.e. mispriming) could produce a smaller than expected amplification product. Primer shift experiments (amplification reactions using primers internal to the initial primer sets), were employed by a number of investigators to address this possibility. Although these methods confirmed many amplification products as originating from deletion mutations, the potential remained for misinterpretating some amplification products, especially for whole genome amplifications in which multiple amplification products were observed. Small amplification products can also result from amplification of duplications of the mtDNA genome. Deletion events are distinguished from possible amplification events by sequence analysis. A second consequence of using a highly sensitive assay (PCR) for detection was the issue of mtDNA deletion abundance. The requirement for PCR suggested these abnormalities are not abundant when compared to total mtDNA present in the tissue homogenates. Quantitative PCR analyses were performed by a number of laboratories. These technically challenging studies were, in hindsight, probably not necessary since the inability to detect these mutations by Southern blot would indicate an abundance of less than 0.1% of the total mtDNA present in the tissue homogenate being analysed. These findings raised concerns about the relevance of low abundance mutations that had been established in the mitochondrial myopathy field to be at a level of > 50% to exhibit a phenotype.
This low calculated abundance of mtDNA deletion mutations in tissue homogenates was, however, misleading. In situ hybridization studies with specific mtDNA probes in a variety of aged tissues including skeletal muscle, extra-ocular muscle and heart muscle demonstrated that individual cells contain high levels of mtDNA deletion mutations, while others appear to be genotypically normal (Müller-Höcker et al., 1993). An analysis of discrete muscle fibre bundles from the skeletal muscle of aged rhesus monkeys further confirmed the mosaic distribution of these abnormalities. As the number of fibres analysed decreased (from tens of thousands to 10), the calculated abundance of specific mtDNA deletion mutations increased (from 0.02% to 13.2%) (Schwarze et al., 1995). Together, these results suggest that single unique mtDNA deletion mutations occur within discrete regions of individual fibres and accumulate to levels that are physiologically relevant.
One interesting and hotly debated question is how these deletion mutations accumulate clonally to high abundance. Initially it was argued that the small genomes had a replicative advantage over wildtype genomes (Wallace, 1992). Others have argued that mitochondrial replication times are sufficient to accommodate the replication of genomes of diverse sizes. Mathematical modelling suggests that clonal accumulation on a random basis is theoretically possible (Elson et al., 2001). More recently, Diaz et al. (2002) have demonstrated, in cell lines containing heteroplasmic mtDNA mutations, that smaller mtDNA genomes (i.e. deletion containing genomes) have a repopulating advantage over wildtype genomes, particularly when copy number controls are relaxed. A different hypothesis (de Grey, 1997) assumes that mitochondria containing deletion mutations are not generating free radical molecules and, thus, mitochondrial proteins and membranes are not being damaged. As a result, mitochondria containing deletion mutations are not removed from the population and, thus, increase in abundance. The presence of high levels of oxidative damage in muscle fibre regions containing high levels of mtDNA deletion mutations (Wanagat et al., 2001) would appear to contradict this ‘survival of the slowest’ hypothesis.
mtDNA deletion mutations are physiologically relevant
All mitochondrial-encoded polypeptides are components of the ETS–OXPHOS system. Therefore, mutations of the mitochondrial genome are most likely to affect these critical energy transduction pathways. Histological examination of skeletal muscle from myopathy patients exhibited the ‘ragged-red’ phenotype, characterized by the activities of two enzymes, succinate dehydrogenase (SDH) and cytochrome C oxidase (COX), of the ETS. RRF stain hyperreactive for SDH (SDH++) and, in the majority of cases, negative for COX (COX−) activity. COX has subunits partially encoded by the mitochondrial genome, while SDH is entirely nuclear encoded. These enzymatic abnormalities also accumulate with age (Fig. 1).
The link between genotype and phenotype of mtDNA deletion mutations was initially demonstrated by in situ hybridization studies using mtDNA probes from differing regions of the mitochondrial genome. Skeletal muscle samples from aged humans (Müller-Höcker et al., 1993) and rhesus monkeys (Lee et al., 1998; Lopez et al., 2000) were hybridized with mitochondrial nucleic acid probes from various defined regions of the mitochondrial genome. The presence of mtDNA deletion mutations, indicated by the lack of hybridization to some of the mitochondrial genome probes, was observed only in the ETS abnormal fibres. The genomic location of the deletion mutation could be localized through the selective use of probes to numerous regions of the mitochondrial genome. The lack of signal with some of the probes, in addition to indicating the region of the genome absent, also implies an absence of wildtype genome.
One consequence of the in situ hybridization and muscle fibre bundle studies was the realization that tissue homogenate studies were not appropriate for quantification of either age-associated mtDNA deletion mutations or the associated ETS abnormal phenotype. The availability of microdissection techniques facilitated the direct analysis of sections of single cells from histological sections. In rat skeletal muscle, all ETS abnormal fibres contained mtDNA deletion mutations with all the deletions being large (4–9 kb deleted) and located within the major arc region of the genome (Cao et al., 2001). In skeletal muscle of both rats and monkeys, the same deletion product was observed in different segments of the same ETS abnormal region (Fig. 2). As noted in the in situ hybridization studies, deletion mutations were not detected in phenotypically normal adjacent fibres or in phenotypically normal regions within the same fibre (Cao et al., 2001; Wanagat et al., 2001).
Our initial efforts to estimate the abundance of mtDNA deletion mutations and associated ETS abnormalities involved the analysis of two cross-sections of tissue from the midbelly of specific muscles. During the course of these studies, it became apparent that these deletion mutations and associated ETS abnormalities were not distributed throughout the entire individual fibres but rather were localized to specific regions or segments of muscle fibres. The segmental distribution of these abnormalities considerably underestimated the number of affected fibres. We therefore modified our approach and began sectioning and analysing 1–2 mm of muscle tissue at 10-µm intervals. This approach allowed us to estimate more accurately the abundance of fibres containing ETS abnormalities. One consequence of these longitudinal studies of aged muscle fibres was the identification of intrafibre atrophy present and unique to ETS abnormal regions. As the ETS abnormal fibres were followed along their length (typically for 1000 µm), many of the ETS abnormal fibres displayed a gradual decline in cross-sectional area, showing intrafibre atrophy within the ETS abnormal region of the same fibre. Some fibres continued to atrophy until they were no longer observable, suggesting that they were broken. In some instances, the fibres reappeared several sections later. Furthermore, an apparent decrease in nuclear number within the ETS abnormal regions was observed, suggesting the activation of cellular apoptotic pathways resulting in fibre death (Wanagat et al., 2001). These segmental mitochondrial abnormalities clearly demonstrate clonal expansion of mtDNA deletion mutations to a high level in a subset of cells and that mutations play a causal role in the loss of muscle mass or fibre. This physiological decline, referred to as sarcopenia, is a clinically recognized aging process (Lexell et al., 1988). Assuming that ETS abnormalities occur once per fibre, Wanagat et al. (2001) estimated that 15% of the fibres contain ETS abnormal regions in the rectus femoris muscle of 38-month-old hybrid rats, but only 2.8% in 5-month-old rats. For rhesus monkeys, up to 60% of the fibres were estimated to display ETS abnormalities in the vastus lateralis muscle of 34-year-olds compared with 4% in 11-year-olds (Lopez et al., 2000).
The link between mitochondrial genomic alterations and fibre atrophy and breakage suggests a molecular basis for fibre loss that occurs in skeletal muscle with age. One test of this hypothesis was to examine specific muscle from aged animals and compare mitochondrial enzymatic abnormalities in muscles that exhibit pronounced fibre loss with those muscles that maintain fibre number with age. Four muscles were studied, vastus lateralis, rectus femoris, soleus and adductor longus. Vastus lateralis and rectus femoris are muscles that exhibit substantial fibre loss with age while soleus and adductor longus are less prone to age-associated fibre loss. Vastus lateralis and rectus femoris, which exhibited the largest muscle mass loss and fibre loss with age, also exhibited the greatest number of ETS abnormalities. In contrast, the adductor longus muscle, which exhibited no muscle mass loss and little change in fibre number with age, did not generate detectable levels of ETS abnormalities. In the soleus muscle, which exhibited little fibre loss, a single ETS abnormal fibre per 1000 µm was found in 36-month-old rats and no ETS abnormal fibres were observed in 5- or 18-month-old rats (Bua et al., 2002). This is strong evidence that the levels of ETS abnormalities are associated with the degree of sarcopenia and fibre number declines exhibited in various muscle types.
Conclusions and perspectives
Addressing questions of whether age-related mitochondrial DNA mutations accumulate to levels of physiological relevance and whether they are associated with age-related phenotypes has required a significant change in experimental design, namely, a shift from the analysis of tissue homogenates to cell-by-cell screening. Mitochondrial DNA deletion mutations have been linked to fibre loss with age. Deletion mutations accumulate with age, clonally expanding within segments of skeletal muscle fibres to high levels. These mtDNA abnormalities are associated with, and appear to cause, muscle fibre atrophy and fibre loss. The impact of deletion mutations on other tissues known to accumulate these aberrant molecules, such as cardiac and nervous tissue, has not been firmly established. Single nucleotide mutations, assumed by many investigators to have potentially the largest impact on aging, have, to date, had less experimental support. One reason is the technical challenge of identifying single nucleotide mutations in a 16-kb genome, especially if one is searching for the accumulation of random mutations within a cell. The lack of an easily detectable phenotype associated with point mutations has also hindered ability to identify and characterize the physiological role of point mutations. The recent discovery of the clonal accumulation to high abundance of single nucleotide mutations suggests that these too may be a factor in aging. Identifying cellular impacts of these point mutations is clearly an important future direction.
We would like to thank all the members of the Aiken lab, especially Susan McKiernan, for their helpful discussions and their critiques of this manuscript. Our mitochondrial DNA abnormality research is funded by the National Institutes on Health.