|Home | About | Journals | Submit | Contact Us | Français|
mtDNA deletions are pathogenic mutations that remove substantial portions of the mitochondrial genome. mtDNA deletions accumulate with age and have been implicated in various degenerative diseases. There are multiple mtDNA per cell and mtDNA mutations become toxic only if they accumulate to substantial intracellular levels, i.e. exceed so-called “phenotypic threshold”. This is usually achieved via clonal expansion of a single initial mutated molecule. Intracellular mitochondrial genomes are analogous to a population of individuals in that mitochondria are born by division and die by degradation. Clonal expansion within cells is thus analogous to genetic drift within populations and is driven by a combination of random processes and selection. mtDNA mutations occurring early in development are expected to end up spread across tissues, while mutations of late origin are expected to be localized, i.e. limited a single post-mitotic cell or progeny of a single stem cell. We have explored the extent and timing of clonality of mtDNA deletion in human muscle using single-molecule PCR. We analyzed deletions from two nearby locations within the same tissue sample. Altogether we analyzed over 130 mutant molecules, but almost every deletion type detected was represented by several identical mutant molecules, so that altogether there were only 21 different kinds of deletions, implying that essentially all deletions were clonal. At the same time the sets of deletions in the two locations were completely different. This observation implies that all of the clonal expansions spanned very small areas and therefore that the corresponding mutations were likely events of older age. More studies are necessary to further validate these findings in muscle and to explore the other tissues.
Accumulation of mtDNA mutations in aging tissues is considered as one of the possible causes of degenerative changes associated with old age and age-related disease, though more work is necessary to evaluate this hypothesis (see (Khrapko and Vijg, 2009) for a recent review). Each cell contains multiple mitochondria, each of which contain a few copies of mitochondrial genome. Any newly mutated copy of mtDNA is therefore surrounded by a great majority (hundreds to thousands) of wild-type copies and will not be harmful for the cell unless mutations accumulate in quantities exceeding so called phenotypic threshold, i.e. about 60–95% of the total mtDNA copies per cell, depending on the type of mutation and the cell type. This may not be true for still hypothetical “dominant-negative mtDNA mutations that are postulated to kill the cell even when present in very low numbers (Dubec et al., 2008), though in this paper we will concentrate on “conventional” mtDNA mutations. Of note, mutations that are counted towards the phenotypic threshold cannot be merely random detrimental mutations. In fact, mutations comprising the threshold need to be of the “same type”, or more precisely, should knock down the same function of the mitochondrial genome. Indeed, if a mutation in one copy of the mitochondrial genome deactivates one mtDNA-encoded polypeptide, another – a tRNA, still another mutation – an rRNA or another polypeptide, then other mitochondrial genomes within the same cell are able to mutually compensate these various deficiencies by exchanging non-mutated (or mutated in a different site) products with mitochondria that happened to contract a mutated mtDNA. This exchange has been demonstrated experimentally (Nakada et al., 2001) and is probably achieved via mitochondrial fusion/fission. In other words, the most deadly scenario is when all mtDNA mutations in a cell are of the same kind. Perhaps surprisingly, this scenario is the one realized in aging cells. It has been demonstrated for a wide range of tissues that mtDNA mutations have a tendency to clonally expand in individual cells, that is, all mutations in a cell are the same, but different between different cells, which implies that mutations in a cell are progeny of one initial mutation that are generated by clonal expansion. Clonal expansion of mtDNA mutations is thus an important and potentially harmful phenomenon, which allows these mutations to realize their patholological potential.
Despite apparent importance of clonal expansion of mtDNA mutations, this process remains insufficiently studied. Mechanisms of clonal expansion remain a matter of debate, but perhaps any expansion is governed by a combination of random genetic drift and selection (Coller et al., 2002). Clonal expansions of mtDNA deletions were demonstrated in muscle more than a decade ago in pioneering work of Muller-Hocker (Muller-Hocker et al., 1993) and Aiken (Schwarze et al., 1995). Since then, the existence of clonal expansions was confirmed in a variety of tissues, both for mtDNA deletions and point mutations. It is still unclear, however, what proportion of mutations present in tissue are involved in clonal expansion. In addition to clonally expanded mutations, there exist individual mutations, which cumulatively may potentially constitute a very significant portion of total mutational burden of the tissue. It is not clear what proportion of mutations belong to each of these two classes (clonal vs. individual). This issue is difficult to resolve because measuring of non-expanded mutations is complicated by artifacts. It is usually easy to prove that a clonally expanded mutation is not an artifacts (because artificial mutations such as PCR errors, do not expand clonally, and thus are easily recognized by their distribution). In contrast non-clonally expanded mutations are indistinguishable from artifacts, which by some accounts may constitute large proportions of reported mutant fractions (Vermulst et al., 2007). Furthermore, clonally expanded mutations are usually measured by methods (such as cell-by-cell analysis) that differ from those used for non-expanded mutations (which are usually measured in tissue homogenates). We were able to determine the fraction of clonally expanded mtDNA deletions in muscle by using a special approach, the single molecule PCR (Kraytsberg and Khrapko, 2005) in tissue homogenates. In our experiments we were able to address the question by measuring clonally expanded and non-expanded deletions simultaneously.
The other unsolved question pertains to the time when mutations arise. Clonal expansion may take a long time to proceed. This implies that mutation event that seeded a clonal expansion might have happened long before expansion is complete, potentially in early development or even in the maturing oocyte. Intense replication of mtDNA during rapid growth of the embryo and/or during oocyte maturation could potentially provide many opportunities for mutation. Those early mutations would have an advantage as they have more time to expand than mutations occurring later. This is particularly relevant to expansions of mtDNA deletions that seem to be usually driven in most tissues by selective pressure rather than random drift (Sato et al., 2007), which is less certain for point mutations. The idea of early origin of mtDNA mutations is corroborated by observations (evidence reviewed in (Khrapko et al., 2004)) that mtDNA mutations tend to cluster in individuals, i.e. a certain mutation may be repeatedly detected in one person’s tissues, while being completely absent in another individual, though an opposite pattern was observed in rats (Khaidakov et al., 2005). We have argued (Khrapko et al., 2004) that the timing of the seed mutation event of a clonal expansion could be estimated by studying the spatial distribution of expanded mutations within/among tissues. An expanded mutation that was seeded before gastrulation is expected to be distributed across the whole body, while mutations arising later during organogenesis should be limited to smaller areas, such as a particular tissue. Mutations arising upon completion of organogenesis should be limited to single cells or muscle fibers where mutations arise, or to the progeny of a stem cell that acquired such a mutation.
Examples of intracellular clonal expansions are deletions in neurons (Kraytsberg et al., 2006), point mutations in colonic crypts (Taylor et al., 2003), bronchial epithelium turnover units (Coller et al., 2005), and expansions of deletions and/or point mutations in muscle fibers (Bua et al., 2006) (Herbst et al., 2007) (Fayet et al., 2002) (Del Bo et al., 2003) and cardiomyocytes and buccal cells (Nekhaeva et al., 2002). The question is whether such clonal expansions were seeded locally (i.e. within the respective cells/fibers or stem cells and thus occurred relatively late in development, or can be traced back to earlier stages and thus are shared between many cells/fibers, or perhaps can even be found across whole tissue or even the whole body. Intriguingly, it was indeed reported that within the same individual, independent fibers (Fayet et al., 2002) (Del Bo et al., 2003), and colonic crypts (Greaves et al., 2006) indeed occasionally share identical point mutations or deletions as if these lesions could be traced back to common precursors of these fibers/cells. In the case of the crypts, it was proposed that common mutations are the result of late in life crypt fission (Greaves et al., 2006) rather than because of sharing an early common precursor. In the case of muscle, the origin of recurrent mutations is not known. As we know now (Bua et al., 2006), clonal expansions sometimes span very long distances (>2mm) within muscle fibers. There is a possibility therefore that different parts of the same fiber with the same expansion might have been collected from sections taken at different levels as independent fibers. Alternatively, fibers sharing the same expanded mutation might have arisen from the same precursor, i.e. resident satellite cell (then the fibers sharing the same mutation are expected to be confined within small area, perhaps a few millimeters across (Jockusch and Voigt, 2003). If the mutation originated earlier in the development, it is expected to be scattered much wider. We addressed this issue in muscle by analyzing the sets of deletions in two proximal but separate locations of the same muscle sample.
A sample of normal skeletal muscle from an 82 year old individual has been obtained through a tissue network. The sample was snap frozen in liquid nitrogen and stored at −80C. The 1 gram tissue sample was scraped by a razor blade from two faces, so that the locations of the sampling were positioned about 0.5 cm apart. The two samples were about 10 mg each. The scraped muscle powder was transferred without thawing into 0.4 ml of lysis buffer (10mM EDTA, 0.5% SDS, 0.1 mg/ml proteinase K) and incubated for 1hr with mild agitation at 45C. The lysate was then diluted as necessary with 10mM Tris pH8.5 1mM. The lysate and dilutions were stored at −80C.
To identify mtDNA deletions, we used single molecular PCR (smPCR), an efficient, error resistant method of mutational analysis (Kraytsberg and Khrapko, 2005). DNA was digested with XhoI, which cuts the mitochondrial genome at nucleotides 14,956 and SacI (cuts at 41 and 9,648), diluted to the concentration of about 0.5 mtDNA deletion copy per PCR plate well and amplified in two stages using nested primers. (outer primer pair: 2999F30/16260R32, inner primer pair: 3204F34/16201R32, where the first number is the position of the 5’-nucleotide, the letter denotes orientation, Forward or Reverse with respect to the conventional sequence orientation of the mtDNA, and the last number is the length of the primer). Of note, single molecule PCR may impose some bias in favor of shorter PCR products (Kraytsberg et al., 2008) because longer templates have a higher probability of sustaining DNA damage that prevents amplification. One may thus suspect that higher copy numbers of come deletions are in fact the result of such a bias. This bias, however is unlikely to skew the results of this study. Indeed, if the bias was significant, we would see high prevalence of molecules with large deletions (shorter PCR products) among clonally expanded ones. There is no such correlation in the data (see table 1).
The relative position of primers and restriction cuts is such that only deleted mtDNA with breakpoints between nucleotide positions 3,204 and 9,648 (5’ breakpoint) and 14,956 and 16,201 (3’ breakpoint) were amplified. This arrangement is particularly convenient for deletion identification as in most cases breakpoint can be identified by a single sequencing reaction of the PCR product using 16201R32 as a sequencing primer because sequence covers essentially all region where 3’ breakpoint may potentially lie. PCR reactions containing single deleted PCR products were identified by gel electrophoresis and PCR products were sequenced using the primer 16201R32 as the sequencing primer. In case no breakpoint was discovered, the next 5’ primer 15783R34 was used to sequence the remaining part of the 3’ breakpoint region.
We have characterized over 80 mtDNA molecules containing large DNA deletions from two locations of a muscle sample from an 82-year old individual. Our initial intent was to characterize the diversity of mtDNA deletion breakpoints in muscle tissue. We soon discovered however, that our approach was very inefficient: once just a few deletion molecules were characterized, sequencing of more molecules essentially did not bring any new types of breakpoints as deletions started to repeat themselves as if in that tissue sample there was a very limited set of types of deletions and every type what represented by a large number of identical molecules. Indeed, almost all types of deletions were isolated multiple times (Table 1), only 7 molecules out of 132 isolated appeared as singlets, i.e. carried deletion types that were not present on any other molecule. This was unexpected, because in principle, there exist an enormous number of various possible deleted mtDNA molecules (MITOMAP), and we hoped that sampling into a pool that wide would bring a new deletion type with each molecule sequenced.
Repeated isolation of the same type of mutation from a given tissue sample implies two possibilities. First, this may be a mutational hotspot, that is, mutational rates at this particular site are high so that mutations may repeatedly and independently arise at this site so that the multiple mutant molecules that we detect represent independent mutational events. Alternatively, all molecules carrying a particular mutation type may represent a clonal expansion, i.e. be the product of the same initial mutational event, i.e. initial mutant mitochondrial genome might have proliferated to create the multiple copies.
To distinguish between these two alternatives, consider that hotspots should be defined by the DNA sequence, and/or specific biochemical processes in the cell and/or the peculiarities of the cellular environment. An important corollary is that mutational hotspots should be relatively universal, that is, they should be the same from cell to cell, at least as far as the same cell type is considered. For example, there is a known hotspot of mtDNA deletions called the “common deletion”, that is associated with a 13-bp direct repeat in mtDNA sequence. Another hotspot for deletions is associated with the site around nucleotide position 16,070. It is still not known what mechanisms are responsible for the generation of deletions at these and other potential hotspots, although some plausible hypothesis have been suggested, e.g. via repair of DNA damage (Krishnan et al., 2008). Exact mechanisms are not important for this discussion, however. Whatever the mechanism, the mutational rates at hotspots should depend on superposition of the above mentioned factors which by their nature are universal, or at most cell type-specific. Different type of damage or the prevalence of different biochemical processes (e.g. repair vs. replication) in different cell types may conceivably result in different sets of hotspots, but the hotspots should be the same in cells of the same type. In other words, by repeatedly sampling the same tissue one would expect to see the same set hotspot mutations over and over again.
Our results clearly show, however, that the sets of mutations are completely different between two nearby tissue locations. This implies that the repeated occurrence of particular deletion types within each location is not due to repeated independent generation of deletions at hotspots, but rather to clonal expansions of a few seed mutations. Having concluded that repeatedly isolated mutations represent clonal expansions, we can now estimate the extent of clonal expansion in muscle of this individual.
The question we are interested in is how important is clonal expansion for generation mtDNA deletions? We have proposed some time ago (Khrapko et al., 2003) that clonal expansion rather than de novo mutation generation may be a major force behind age-dependent accumulation of mtDNA mutations with age, at least for some types of mutations and some tissue types. This hypothesis seems to be correct, at least for mtDNA deletions in muscle of the particular individual described in this report. Indeed, a randomly sampled mtDNA molecule with a deletion most likely (odds approximately 125 to 7, see Table 1) will be a part of a clonal expansion. The other way to look at these data is to consider that starting with 21 different (and supposedly independent) mutational events clonal expansion resulted in 132 mutant genomes, that is, clonal expansion is responsible roughly for the generation of 85% of mutant genomes in this tissue sample. It should be noted that these estimates of the proportion of mutants involved in clonal expansions may underestimates. Indeed, some of the deletions that we count as single unexpanded mutations may in fact be part of a small clonal expansion that we might have failed to recognize because of the small total number of mutant molecules that was analyzed. Had we sampled more molecules, then we might have detected more deletions of the kinds that so far are represented by a single molecule only and thus are classified as non-expanded mutations. In other words, our estimates of the significance of clonal expansion are lower estimates.
Comparison of the set of mutations in two locations of the same muscle sample (separated by the distance of about 10 mm) revealed that the sets are completely different with no mutations in common. This implies that the distribution of mutant DNA molecules belonging to every clonal expansion is local. As discussed in the Introduction, this implies that mutations are of relatively late origin, i.e. they have arisen within the cells that contain them, or at most in the satellite precursor cells.
This is a result based on a single tissue sample and a limited data. Our results should not be considered more than a demonstration that mtDNA deletions may be highly clonal and of late origin. More studies are necessary to validate our results for a wider cohort of individuals. It is quite possible that our conclusions will not always hold or will be just not true for other tissues like brain. The approach that we used is universally applicable and will help to clarify these issues in the future. Of particular particular interest, in our view, are individuals with increased rate of mtDNA deletions. The mechanisms responsible for excess of deletions in these cases may be different (Spelbrink et al., 2001) (Hudson et al., 2008) and thus the timing of origin of deletion may be different too. It would be interesting to explore whether there are non-local early origin clonal expansions of mtDNA deletions in such individuals.
This work was supported in part by NIH grants NS058988 and AG19787 to KK.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.