Does MtDNA Nucleoid Organization Impact Aging?

Abstract

Somatic cells in tissue culture package several copies of mitochondrial DNA (mtDNA) in aggregates known as nucleoids that appear to be remarkably stable. The clustering of multiple mtDNA genomes in a single nucleoid complex may promote the progressive age-related accumulation of deletion and point mutations in mtDNA in many somatic tissues, particularly in post-mitotic cells. In contrast, oocytes appear to have the ability to select against deleterious mutations in mtDNA, at least in mice. This fundamental difference suggests that oocytes may be better able to detect and remove defective mtDNA genomes than somatic cells, possibly due in part to the simpler organization of the mtDNA in smaller nucleoids. These observations suggest the hypothesis that a complex nucleoid structure containing several mtDNA molecules may impair the ability of the cell to select against deleterious mtDNA mutations, thereby contributing to age-related mitochondrial dysfunction.

Introduction

The oxidative theory of aging has enjoyed a phoenix-like resilience in the years since its initial formulation by Harmon in 1956 (Harman, 1956) and its restatement implicating mitochondria as the mainspring of a biological clock (Harman, 1972). Successive waves of criticism have remodeled the landscape without eroding the underlying concept that progressive mitochondrial dysfunction is a hallmark of the aging process. Still, a comprehensive understanding of exactly how mitochondrial function degrades with age has been elusive. For many years, the most popular notion has maintained that aging results from progressive oxidative damage emanating from the mitochondrial electron transport chain. One variant of the mitochondrial oxidative stress model suggests that oxidative damage to mtDNA may lead to the formation of aberrant respiratory complexes that are increasingly likely to generate even more reactive oxygen and nitrogen species (RONS) leading, through a vicious cycle, to progressively greater mtDNA damage (Bandy and Davison, 1990). An alternative model maintains that mtDNA damage leads to decreased oxygen consumption and a corresponding reduction in further oxidative damage to mitochondrial lipids, resulting in slower turnover of mitochondria bearing mutant or damaged mtDNA genomes and, ultimately, to a progressive accumulation of defective mtDNA genomes (De Grey, 1997). This model has been referred to as the “survival of the slowest” (SOS) model. These theories have stimulated additional research, which, in many cases, has failed to support the a priori models.

The mitochondrial oxidative stress theory now faces a stringent challenge posed by several major experimental findings, as follows:

First, two laboratories have independently generated genetically engineered mouse strains that express error-prone versions of mitochondrial DNA polymerase γ (Kujoth et al., 2005; Trifunovic et al., 2004). These mice accumulate a high burden of mtDNA mutations and show several phenotypes consistent with premature aging, but do not exhibit increased oxidative stress. These models permit the conclusion that mtDNA mutations may decrease longevity independent of any effects on oxidative stress. The spectrum of mutations observed in mtDNA does not show a signature consistent with oxidative stress as a major factor in mutagenesis. Unrepaired 8-oxo-deoxyguanosine would be expected to frequently base pair with deoxyadenosine during replication, leading to G to T transversion mutations (Pinz et al., 1995). The relative lack of such mutations suggests that other mechanisms, such as the inherent error frequency of pol γ are more important sources of mtDNA mutations than oxidative stress (Zheng et al., 2006).

Second, a series of transgenic and knockout mice have been generated in an effort to reduce the mitochondrial production of RONS or to over-express enzymes expected to alleviate oxidative stress. With one exception, the mouse expressing mitochondrially-targeted catalase (Schriner et al., 2005), these mice have failed to show any significant increase in longevity (reviewed in (Jang and Remmen, 2009; Lapointe and Hekimi, ; Pérez et al., 2009; Van Remmen and Jones, 2009)).

Third, a comparison of two rodent species that differ considerably in their maximum life expectancy has found that the long-lived naked mole rat generates as much or more RONS as the shorter-lived Mus musculus (Perez et al., 2009).

Finally, an extensive series of studies in many laboratories (reviewed in (Chan, 2006)) has revealed that mitochondria in most cell types engage in rapid cycles of fission and fusion that would tend to homogenize mitochondrial lipid contents so that oxidized lipids in the mitochondrial membranes would not be expected to remain stably associated with damaged mtDNA molecules. This argues against at least one version of the SOS model.

Taken as a whole, these observations have led to considerable confusion in the field. It appears that mitochondrial oxidative stress is at most only one of several variables that contribute to aging and that mtDNA mutations may be a more important factor in age-related mitochondrial dysfunction (Wallace, 2005). We know that mtDNA mutations are commonly found in the cytochrome oxidase-deficient post-mitotic cells that accumulate with age. This phenomenon is well-established for neurons in the substantia nigra (Bender et al., 2006; Kraytsberg et al., 2006) and in skeletal muscle (Herbst et al., 2007; McKiernan et al., 2009), but the extent to which such deletions contribute to normal human aging is still disputed. The mechanism underlying the accumulation of these mtDNA mutations with age has not been established firmly.

In this article, I will present the argument that the packaging of multiple mtDNA molecules in large aggregates known as nucleoids may promote the age-related accumulation of mtDNA mutations.

The fundamental organization of mtDNA nucleoids

The discovery of mtDNA in the 1960’s strongly supported the hypothesis that mitochondria were derived from an endosymbiotic bacterium. As a bacterial chromosome is typically confined to a distinct nucleoid domain associated with the inner membrane, Nass applied the term “nucleoids” to the mtDNA nucleoprotein complexes (Nass, 1969). These mtDNA-protein structures can be visualized using PicoGreen and other DNA stains (Bereiter-Hahn and Vöth, 1996) or by immunofluorescence with anti-DNA antibodies (Iborra et al., 2004). Visualization of nucleoids is facilitated by their typical content of 5-7 mtDNA molecules packed in structures with a diameter estimated at 70 nm (Iborra et al., 2004; Legros et al., 2004). Since a single mtDNA molecule has a contour length of 5 μm, this packaging requires a considerable compaction of the DNA, which is probably largely accomplished by the DNA wrapping ability of the HMG-box protein TFAM (Antoshechkin and Bogenhagen, 1995; Kaufman et al., 2007). The protein composition of nucleoids has been studied for complexes affinity-purified using antibodies directed against major mtDNA binding proteins, mtSSB and TFAM, and for complexes cross-linked using formaldehyde (Bogenhagen et al., 2008; Wang and Bogenhagen, 2006). These studies detected most of the proteins known to be involved in mtDNA maintenance, such as DNA pol γ, the T7-like helicase, mitochondrial RNA polymerase, TFBM1, TFBM2, Terf1, and mitochondrial topoisomerase I. In addition several less well-studied proteins with helicase motifs were identified in nucleoids, such as the suv3-like helicase, DEAD box protein 28 and a novel helicase known as DHX30 which was first identified as a mitochondrial protein based on its detection in nucleoids (Wang and Bogenhagen, 2006).

While nucleoids are clearly foci for mtDNA replication and transcription, they may play an even broader role in mitochondrial biogenesis. Proteomic analysis of nucleoids also revealed a number of factors involved in protein folding and quality control including Hsp60, Hsp70, Hsp40, prohibitins 1and 2, and the complex IV assembly factor LRPPRC ((Mootha et al., 2003) also known as LRP130) along with the ClpX and Lon proteases. In independent studies, Lon has been reported to have DNA binding activity (Liu et al., 2004). Experiments using an immunoaffinity method to search for proteins interacting with a mitochondrial translation factor found nucleoid proteins co-purifying with mitochondrial ribosomes (Rorbach et al., 2008). These results suggested that nucleoids might have a layered structure with an inner core enriched in proteins involved in nucleic acid synthesis and processing and an outer zone involved in the assembly of nascent polypeptides into respiratory complexes (Bogenhagen et al., 2008). This layered structure may help explain why somatic cells tend to assemble multiple mtDNA genomes in single clusters in the first place. Large nucleoids may permit the products of several mtDNA genomes to be pooled to support assembly of respiratory complexes, thus serving as a component of a putative “unit mitochondrion” as envisioned by Capaldi and coworkers (Capaldi et al., 2002).

A nucleoid containing several mtDNA genomes and such a wide array of proteins would have a molecular mass of over 100 megadaltons, making it the largest structure in a mitochondrion, dwarfing even the massive respiratory supercomplexes (Acín-Pérez et al., 2008; Schafer et al., 2007) and pyruvate dehydrogenase (Zhou et al., 2001). Nucleoids move freely within cells, but only as passengers within the dynamically motile organelles that they sustain. They appear to be very stable and largely immobile within mitochondria, and do not readily fuse or exchange mtDNA genomes (Gilkerson et al., 2008). Other aspects of the dynamic behavior of nucleoids are not well understood. As cells grow and divide, there must be some mechanism for the fragmentation and distribution of nucleoids. This has been studied in more detail in lower eukaryotes (Kuroiwa et al., 1994), but even in these instances, the molecular mechanisms involved are poorly understood.

What is the impact of nucleoid organization on mtDNA quality control and aging?

It has long been recognized that the asexual nature of mtDNA inheritance may promote the accumulation of mtDNA mutations according to Muller’s ratchet (Muller, 1932). The organization of mtDNA in nucleoids may greatly influence the phenotypic expression of mtDNA defects. Typically, a single newly-acquired mutation in one mtDNA copy within a nucleoid is not detrimental since it can be complemented by the remaining wild-type genomes. Such complementation may even permit a mitochondrion to function relatively well when a nucleoid retains a single wild-type mtDNA molecule alongside several mutant copies. This correlates well with the general observation that mtDNA mutation rates must reach or exceed about 80% before a heteroplasmic cell shows signs of mitochondrial dysfunction (DiMauro and Schon, 2003; Taylor and Turnbull, 2005). Of course, genetic complementation between mtDNA molecules is not strictly limited to molecules co-localized in a single nucleoid, but it is reasonable to expect that complementation is less effective if mtDNA gene products must diffuse over long distances. Indeed, in many post-mitotic cells containing mitochondria in confined locations, such as neurons and cardiomyocytes, genetic complementation between nucleoids separated by large distances may be relatively ineffective.

Intra-nucleoid mtDNA genetic complementation may be an example of “antagonistic pleiotropy” as expressed by Williams (Williams, 1957), in other words, a two-edged sword. As noted above, it helps protect against the deleterious effects of relatively rare mutations. However, this same local complementation makes it difficult to selectively cull defective mtDNA genomes in somatic cells, resulting in more rapid accumulation of mutations. By analogy, we can consider that mitochondrial biogenesis is a group project in which the contributions of a few functional genomes can mask the poor performance of their ineffective collaborators. Although the gradual acquisition of mtDNA mutations with age in somatic tissues appears to be an inescapable consequence of Muller’s ratchet, new results indicate that nature has evolved a mechanism to select against deleterious mutations in the germline. How is this apparent rejuvenation reconciled with nucleoid structure?

The critical importance of the variability in germline transmission of mutant mtDNAs in human diseases has been evident since the early days of mitochondrial medicine when it was documented that the siblings and mother of a proband with a mitochondrial disorder could all exhibit different degrees of heteroplasmy (Shoffner et al., 1990). Remarkably, individual oocytes from a mother heteroplasmic for a mtDNA disease mutation can contain essentially homoplasmic mutant or wild-type mtDNA (Blok et al., 1997). Earlier work attributed such rapid generational shifts in the level of mtDNA heteroplasmy in cattle to a stage in early oocyte development characterized by a very low mtDNA copy number, which Hauswirth and Laipis termed the mtDNA bottleneck (Hauswirth and Laipis, 1985; Laipis et al., 1988). This bottleneck in mtDNA population size is estimated at 200 genomes in primordial germ cells (PGCs) in mice (Cree et al., 2008; Jenuth et al., 1996). Another recent publication reported a somewhat larger number of mtDNA molecules in PGCs and suggested that these might be organized into approximately 200 segregating units, presumably nucleoids (Cao et al., 2007).

PGCs, which are first detectable in mice as clusters of as few as eight cells during gastrulation (Ginsburg et al., 1990; McLaren, 2003), may be derived from a small segment of ooplasm, raising the possibility that there may be a highly non-random selection of oocyte mtDNAs to populate the next generation of germ cells. Non-random selection of mtDNA molecules for replication may be another significant source of mtDNA genetic variability (Wai et al., 2008). During embryonic development in the human female, the sparse population of PGCs expands to roughly 7 million primary oocytes each containing about 8000 mtDNA molecules in small, spherical mitochondria (Jansen and deBoer, 1998; Jansen and Burton, 2004). The final stages of oocyte maturation involve a further expansion of mtDNA copy numbers to about 200,000 per egg. Thus, as shown in Figure 1, embryogenesis entails a massive PCR-like amplification of the small pool of mtDNAs in the initial population of PGCs, providing an opportunity for considerable variation in the degree of heteroplasmy.

Fig. 1. The life cycle of mtDNA inheritance in the female germline.

MtDNA in somatic cells is not inherited and may be organized differently than in oocytes.

While there is broad agreement that oogenesis is accompanied by rapid shifts in heteroplasmy, until recently this variance was thought to be random, incapable of filtering or selectively removing deleterious mutations (Chinnery et al., 2000; Shoubridge, 2000; Van Blerkom, 2004). This view must be modified in light of two new genetic studies documenting purifying selection against deleterious mutations (Fan et al., 2008) or non-synonymous codon changes (Stewart et al., 2008b) in the germline. These papers imply that mtDNA is subject to some sort of quality control during oogenesis, as discussed in a recent review (Stewart et al., 2008a). While the precise mechanism for this purifying selection remains uncertain, it must be remembered that it is not highly effective since mutations, even substantial deletions in an engineered mouse strain (Inoue et al., 2000), can be transmitted through the germline.

Purifying selection against mtDNA mutations could function at two levels, as discussed by Stewart et al. (Stewart et al., 2008a). First, we know that the vast majority of oocytes are lost through atresia, raising the possibility that there may be selective destruction of oocytes containing an excessive proportion of defective mitochondria (Krakauer and Mira, 1999; Tilly, 2001). In addition to selection at the cellular level, there may be selection against individual mitochondria harboring defective mtDNAs, possibly based on the inability of these mitochondria to develop a sufficient membrane potential (Δψ). Under some conditions, mitochondria with a relatively low membrane potential have been shown to be more likely to be removed by autophagy (Twig et al., 2008). Additional research will be necessary to define the mechanism of purifying selection against mutant mtDNA genomes and to determine the extent to which mtDNA quality control may contribute to oocyte atresia.

An effective mtDNA quality control process must be able to link defective genomes with stable phenotypes reflecting mitochondrial dysfunction. This is difficult to accomplish if mutant mtDNAs are genetically complemented by wild-type genomes within nucleoids or even if high rates of mitochondrial fusion and fission permit functional complementation in heteroplasmic cells. Thus, if purifying selection functions in the female germline but not in most somatic cells, this must reflect some distinctive features of mitochondrial biology in oocytes. There are at least two unusual aspects of mtDNA genome organization in oocytes that may enhance quality control. First, measurements of the total number of mitochondria and of mtDNA in oocytes indicate that, on average, each mitochondrion may harbor only a single mtDNA genome (Piko and Matsumoto, 1976). This is a statistic that has been quoted frequently in reviews (Cummins, 2002; Jansen and Burton, 2004; Van Blerkom, 2004), but it is important to remember that this has not been demonstrated formally, it is only an average value. The possibility that some organelles have several mtDNA genomes while others have none has not been eliminated. Nevertheless, it is intriguing to consider that oocytes may indeed have accomplished a “mitochondrial meiosis” (Hauswirth and Laipis, 1985) that permits quality control at the single-genome level. Second, oocyte mitochondria are generally characterized as small, rather spherical organelles (Jansen and deBoer, 1998; Piko and Matsumoto, 1976), suggesting that there is not a large interconnected mitochondrial reticulum to permit facile exchange of mitochondrial gene products in these cells. We suggest that the balance between fusion and fission may be tipped in favor of fission in oocytes. Fission and fusion cycles are necessary during later embryonic development (Chen et al., 2003), but may be rather quiescent in oocytes.

The concept that purifying selection against deleterious mtDNA mutations may operate more efficiently in oocytes containing smaller nucleoids is in accord with the “disposable soma” theory of aging (Kirkwood and Holliday, 1979). In this view, organization of mtDNA in larger nucleoids is favored in somatic cells that need to build large energy farms of respiratory complexes, but this organization makes it more difficult to eliminate mtDNA mutations, since somatic mutations are not inherited. Additional studies are required to further define these mechanisms and to explore nucleoid structure and the status of mitochondrial fission and fusion accomplished in stem cells, as suggested recently (Rajasimha et al., 2008). Such investigations may ultimately suggest a means to intervene to improve selection of healthy mtDNA genomes in oocytes as well as in somatic cells.