Human Molecular Genetics, 2000, Vol. 9, No. 19 2821-2835
© 2000 Oxford University Press
Human mtDNA sublimons resemble rearranged mitochondrial genomes found in pathological states
1Institute of Medical Technology and 2Department of Forensic Medicine, University of Tampere and Tampere University Hospital, Tampere, Finland, 3Department of Neurology, University of Oulu, Oulu, Finland, 4Department of Paediatrics, University of Oxford, Oxford, UK, 5MRC Dunn Human Nutrition Unit, Cambridge, UK and 6Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow G12 8QQ, UK
Received 29 August 2000; Revised and Accepted 25 September 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Sublimons, originally identified in plant mitochondria, are defined as rearranged mtDNA molecules present at very low levels. We have analysed the primary structures of sublimons found in human cells and tissues and estimated their abundance. Each tissue of a given individual contains a wide range of different sublimons and the most abundant species differ between tissues in a substantially systematic manner. Sublimons are undetectable in
0 cells, indicating that they are bona fide derivatives of mtDNA. They are most prominent in post-mitotic tissue subject to oxidative stress. Rearrangement break-points, often defined by short direct repeats, are scattered, but hotspot regions are clearly identifiable, notably near the end of the D-loop. The region between the replication origins is therefore frequently eliminated. One other hotspot region is located adjacent to a known site of protein binding, suggesting that recombination may be facilitated by protein–protein interactions. For a given primary rearrangement, both deleted and partially duplicated species can be detected. Although each sublimon is typically present at a low level, at most a few copies per cell, sublimon abundance in a given tissue can vary over three orders of magnitude between healthy individuals. Collectively, therefore, they can represent a non-negligible fraction of total mtDNA. Their structures are very similar to those of the rearranged molecules found in pathological states, such as adPEO and MNGIE; therefore, we propose that, as in plants, human mtDNA sublimons represent a pool of variant molecules that can become amplified under pathological conditions, thus contributing to cellular dysfunction. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Rearranged mitochondrial DNA (mtDNA) molecules, including both heteroplasmic deletions and partial duplications, are found in a great variety of human pathological states, notably in syndromic disorders affecting heart and skeletal muscle, the central nervous system and other post-mitotic tissues (1–3). These disorders can be divided into two categories, based on inheritance patterns and the types of rearranged molecule present, although their clinical features overlap. The first category shows sporadic occurrence or occasionally maternal inheritance, with a single primary rearrangement found in the affected tissue(s). This is presumed to have expanded clonally during early development. The second category shows autosomal inheritance, with multiple rearranged molecules found in different tissues, although individual cells or muscle fibre segments may contain just one or a few clonally expanded, rearranged mtDNA species (4).
The precise rearrangements found in the two categories of disorder show some similarities and also some differences. Clonally expanded ‘single’ rearrangements are often present in different forms that are theoretically interconvertible by homologous recombination (5,6), as illustrated in Figure 1. Rearranged molecules with a given pair of primary break-points can be present within a single individual simultaneously as deletions, partial duplications, deletion dimers and other multimers. Such break-points frequently occur at short, directly repeated sequences of up to 13 bp, suggesting that illegitimate recombination is the mechanism of primary rearrangement, although replication slippage has also been proposed. The deletion usually lies within the region between the replication origins for the two strands, although the exact break-points are scattered. Different deletions can be associated with similar clinical phenotypes and vice versa. Rearranged molecules are invariably a substantial fraction of total mtDNA in affected tissues, such as brain and skeletal muscle, in patients with Kearns–Sayre syndrome or sporadic ocular myopathy (PEO).
|
Multiple deletions, as found in such disorders as autosomal dominant PEO (adPEO) (7,8), mitochondrial neurogastrointestinal encephalomyopathy (MNGIE) (9,10) and inflammatory muscle disease, such as sporadic (11) and hereditary (12) inclusion body myositis (IBM), are also usually located between the replication origins, but break-points are more clustered. In particular, one end of the deletion very commonly lies within a short region near to the end of the D-loop. Direct repeats at break-points are rarer, or sometimes very short, and short direct and inverted repeats are often found near rather than at the exact break-points. Although each individual deleted species is present typically at a low level, deleted molecules as a whole again represent a substantial fraction of mtDNA (10–50% or more) in affected, post-mitotic tissues (10,13).
Recently, we as well as others have detected rearranged mtDNAs even in control subjects, using long PCR methods (14–16). Since such molecules cannot be detected by Southern blot, at least in young, healthy subjects, it has been assumed that they are present only at very low levels. At high template concentration, however, long PCR preferentially amplifies these molecules and exaggerates their apparent abundance. Reconstruction experiments that we performed showed that deleted molecules present at one or just a few copies per cell are easily detected by long PCR (17). Blot hybridization indicated that these molecules generally lack the region between the replication origins, just like the deleted mtDNAs found much more abundantly in cases of mitochondrial disease. Some reports, using PCR, have suggested that such deleted mtDNAs accumulate significantly during the course of ageing (18–22), as well as in some cases of dilated cardiomyopathy (23).
In an analogy with plant mitochondria (24,25), we have tentatively termed these low abundance rearranged mtDNAs ‘sublimons’ (14). The term was originally applied to low abundance DNA molecules found in the mitochondria of healthy (fertile) maize plants, whose structures were similar to those of rearranged mtDNAs found in certain male sterile lines (24). However, some very basic questions remain regarding the structure, abundance and tissue distribution of these molecules. In this paper, we present a detailed structural characterization of the sublimons present in heart muscle and compare the sublimon profiles of different individuals and tissues for clues as to the mechanisms that might generate them, in order to assess their relationship with pathological deletions.
The results confirm that heart sublimons have very similar structures to those of pathological multiple deletions. The same primary rearrangement can be represented as deleted or partially duplicated mtDNA, as well as deletion multimers. Break-points, frequently involving short direct repeats, are clustered near the end of the D-loop and in at least one other region of protein binding and the most abundant sublimon classes show break-point heterogeneity.
Remarkably, the spectrum of sublimons varies between tissues, but very similar profiles of sublimons are present in a given tissue from different individuals. This tissue specificity applies even at the microscopic level, with different and characteristic patterns of break-point heterogeneity in each tissue. Based on a semi-quantitative multiplex PCR assay and counting a set of species exhibiting break-point heterogeneity within a limited region as a single sublimon, the more abundant sublimons may be present at up to 190 copies per cell, although in other individuals their representation can be much less than 1 copy per cell or even below the detection limit. We did not find any evidence of a simple relationship between age and sublimon abundance.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Sublimons are tissue specific
As previously (14,17), we used long PCR with adjacent, control region primers, to amplify full-length mtDNA, as well as any rearranged molecules containing deletions or partial duplications that were present in each DNA sample analysed. Note that the method, based on outwardly orientated primers, does not distinguish between deleted and partially duplicated molecules, which give identical products for a given primary rearrangement that still contains the priming sites. The pattern of products representing rearranged mtDNAs was different in the various tissues of a given individual (Fig. 2a). Products were in general shorter than 6 kb, although some longer products were also generated. A very prominent product of 3.75 kb was evident, for example, in both heart and skeletal muscle, but the same product was either much less abundant or not evident at all in other tissues. The other major products from heart and skeletal muscle were not identical, each tissue showing a set of products that were either unique or much less prominent in other tissues. The template concentration artefact, in which small circular molecules are preferentially amplified at high template DNA concentrations, makes it hard to judge accurately the relative amounts of sublimons in different tissues. However, based crudely on the amount of template dilution needed to give a substantial yield of the 16.6 kb wild-type mtDNA band, sublimons were markedly more abundant in post-mitotic tissues proposed to be subject to oxidative stress (heart, skeletal muscle, brain, as well as liver), but less abundant in other tissues such as skin, pancreas and blood. In contrast, very similar patterns of products were generated from template DNAs of a given tissue from different individuals: heart DNAs typically generated a ‘heart-specific’ set of products (Fig. 2c), although with some exceptions (Fig. 2b), whereas skeletal muscle or kidney DNAs gave more muscleor kidney-specific patterns (Fig. 2b). Repeated deproteinization of DNA samples by prolonged proteinase K digestion and phenol extraction made no difference to the pattern or abundance of products generated, indicating that they are not the result of residual protein contamination of the templates.
|
HindIII digest, combined in (a) with 
X174 BsuRI digest]. (a) Sublimon products obtained using template DNAs extracted from different tissues of a single individual (s2, 80-year-old male) post mortem, as follows: skeletal muscle (13 ng), heart (5.2 ng), brain (frontal cortex, 14 ng), liver (56 ng), kidney (13 ng), skin (41 ng), pancreas (12 ng), lung (57 ng) and blood leukocytes (38 ng). The 16.6 kb linear mtDNA product and the prominent 3.75 kb sublimon product obtained from skeletal muscle, heart, kidney and liver, are indicated by arrows. (b) Sublimon products obtained from four tissues of each of three individuals post mortem. Lanes 1, 2 and 3 used template DNAs from individuals s1 (22-year-old male), s2 and s3 (75-year-old female), respectively, as follows. Skeletal muscle: s1, 125 ng; s2, 13 ng; s3, 72 ng. Heart: s1, 24 ng; s2, 5.2 ng; s3, 2.0 ng. Kidney: s1, 26 ng; s2, 13 ng; s3, 18 ng. Blood: s1, 90 ng; s2, 38 ng; s3, 28 ng. Major products as in (a) are indicated by the arrows. (c) Sublimon products from DNAs extracted from heart muscle of five male alcoholic cardiomyopathy patients (lanes 6–10, ages 47, 65, 46, 46 and 48 years, respectively), plus five male controls (lanes 1–5, ages 65, 58, 52, 62 and 48 years, respectively). Approximately 10–100 ng of template DNA were used in each case. Major products as in (a) are indicated by the arrows. (d) Sublimon products from increasing amounts of template DNAs extracted from brain (0.02, 0.12, 0.58, 2.9, 14, 72 ng) and heart (0.21, 1.0, 5.2, 26, 130, 652 ng) of subject s2 post mortem, and from 143B osteosarcoma-derived Sublimon-derived products were also amplified from cultured cell line DNAs (data not shown). However, such products could not be generated from either of two different
0 cell templates (Fig. 2d), indicating that they are derived from bona fide mtDNA and are not the products of amplification from nuclear pseudogenes of mtDNA. Pre-digestion of the template with BamHI, which cuts human mtDNA once, at nucleotide position 14 258 in the ND6 gene, abolished synthesis of the 16.6 kb product, but almost all of the sublimon products were synthesized just as efficiently as from uncut DNA, indicating that the region in which the BamHI site lies is deleted from almost all of the sublimons (data not shown). This applied to all tissues tested.
Sublimon break-points are concentrated in hotspot regions
To determine the primary structure of a representative set of sublimons, we cloned long PCR products from control heart DNA of a single individual, initially at random and subsequently by pre-selecting gel-purified products in the >4 kb size class, to counteract the inevitable bias towards cloning shorter products. Shorter cloned products were sequenced in their entirety, longer products as far as the break-points that defined the corresponding rearrangement. Most of the products characterized in the initial survey were equivalent to very large deletions, removing the previously assigned origin of light-strand synthesis OL. The combination of long PCR and plasmid cloning discriminates against longer sublimons. However, some heart sublimons detected by long PCR were clearly in the size range >7 kb, so would retain OL if their deletions began in or near the D-loop. To amplify sublimon-derived products whose deletions did not include OL we used, as previously (14), a different pair of primers: one located in the D-loop, the other in the COXI gene. Nested PCR using these primers on gel-purified products of the first PCR reaction in the 7–15 kb size range gave an identical set of final products (data not shown). The findings from the full set of cloned products are summarized in Table 1 and Figure 3. Only one clone of 65 represented a mispriming artefact: the remainder were correctly primed products containing an internal mtDNA rearrangement.
|
|
The rearrangement break-points were concentrated in specific hotspot regions, the clearest example being a common break-point region (CBR) towards the end of the D-loop, around nucleotide position 16 070, which was involved in >80% (52 of 64) of the clones, although the exact break-point was heterogeneous (Fig. 3b). A less prevalent break-point region around nucleotide position 3260 was found in eight clones, in six of which the second break-point was within the CBR. The other two had a break-point at nucleotide position 16 034, which was also found in one other clone. The highly prominent long PCR product of 3.75 kb (Fig. 2) represents heart sublimons bounded on one side by the nucleotide position 3260 break-point region and on the other by the CBR. A second prevalent rearrangement, represented as seven clones, was a 7.4 kb deletion, bounded on one side by the CBR and on the other, at nucleotide positions 8637–8648, by an exact 12 bp repeat of a sequence found at the CBR. All clones analysed represented molecules in which a large portion of the genome adjacent to the D-loop was deleted.
Several other features emerged from sequence analysis. In general, aside from these gross rearrangements, sublimon sequences matched those of bona fide mtDNA, apart from a small number of polymorphisms and probable PCR-induced mutations. The widespread scatter of low level substitutions and microdeletions characteristic of nuclear pseudogenes of mtDNA were absent. Less than half of all sublimons (28 of 64) showed evidence of significant direct repeats (>1 bp) at the break points (Fig. 3c), but the presence of short direct or inverted repeats very close to (but not exactly at) the rearrangement sites was commonly noted. Nine clones had direct repeats of 10 bp or more across the junctions, although seven of these represented the 7.4 kb deletion. One clone had a 1 bp insertion at the break-point. Note that because the primers used for PCR lie far from the rearrangement break-points, these sequence features are very unlikely to be PCR-induced artefacts and must instead represent the intrinsic structures of the sublimons themselves.
Microscale analysis reveals tissue specificity not individual specificity
The sequence analysis described above indicated microscale break-point heterogeneity even for ‘classes’ of sublimons that resolve as a single PCR product on agarose gels, e.g. the prevalent 3.75 kb species. To investigate this further, conventional PCR using one fluorescent and either of two non-fluorescent primers was used to amplify across the break-points of two of the major sublimon classes detected by sequencing: the prevalent (3.75 kb) sublimons, as well as a sublimon class of 2.83 kb, had one break-point also in the CBR, and the other within the 16S rRNA gene (represented as clones 3, 13 and 17). These primer pairs were tested first on myocardial DNA to confirm that they gave the predicted products (Fig. 4).
|
The products showed a similar heterogeneity as was evident from sequence analysis, although the patterns of heterogeneity were different for the two sublimon classes. The prevalent sublimon class manifested as 27 distinct products, differing in size by increments of one nucleotide, although these may represent many more than 27 different molecular species, since heterogeneity at both break-points was evident from sequence analysis. The second sublimon analysed, with a break-point in the 16S rRNA gene, manifested as four discrete products (shoulders on each of these peaks can probably be discounted, since they are attributable to the standard PCR artefact of incomplete addition of non-templated terminal A, by Taq DNA polymerase). Neither primer pair amplified any product from
0 cell DNA, in a multiplex reaction in which a single copy nuclear gene control was efficiently synthesized. The products yielded by these primers are therefore specific to mtDNA. Fluorescent PCR was used to analyse further the structure and representation of the prevalent 3.75 kb sublimon class in different tissues of the same individual, post mortem (Fig. 4a). This revealed a subtle but clear difference in the pattern of products derived from the various tissues. Skeletal muscle gave the least heterogeneous picture, with only two major products representing sublimons of 3762 and 3785 bp, plus a third product corresponding with a 3770 bp species. The patterns of products were highly reproducible for a given DNA sample. Even more remarkably, very similar profiles were produced using template DNAs from a given tissue of different individuals (Fig. 5). This was particularly striking in the case of heart muscle, where the product profiles from virtually all individuals, resolved at the nucleotide level, were almost exactly superimposable. In skeletal muscle samples, a product corresponding with a prevalent sublimon of 3762 bp was found in all individuals surveyed. The amount of heterogeneity was always less than in the heart, but seemed more variable between individuals. From some individuals, no product could be amplified from skeletal muscle DNA at all, even though a single copy nuclear amplicon was efficiently synthesized in all cases, as an internal control. From kidney DNA (as well as brain; data not shown) a heterogeneous set of products was amplified from all individuals, although its exact profile was somewhat variable. Sperm DNA gave single species representing sublimons of 3754, 3762 and 3784 bp, but the relative levels of these three species varied considerably between individuals. The same primer pair consistently gave no product at all with template DNAs from lung, pancreas or blood of multiple individuals, even though control amplicons were efficiently synthesized from these templates. Very minute amounts of product were evident in liver or skin of some individuals.
|
Sublimon abundance varies between <0.1 and >100 copies per cell
In order to estimate the abundance of sublimons of the prevalent (3.75 kb) class, we carried out multiplex PCR using primers across the deletion junction, combined with primers for a single copy nuclear gene giving a similarly sized PCR product (172 bp) as an internal standard. The products, labelled by two different fluorescent labels, were analysed by capillary electrophoresis at increasing cycle times to ensure that the reactions had not yet reached saturation. Data from any reaction points where saturation was inferred were discarded from analysis. Ratios were computed of the amount of nuclear and sublimon product at each sampling point. To guard against artefacts arising from impurities in different template preparations, we diluted template DNAs 5- and 25-fold to check that the ratios obtained were not altered. The method is illustrated in Figure 6. The results shown in Table 2 are for a set of six control hearts and for various different tissues of other individuals, mainly collected post mortem.
|
|
Three general points may be extracted from these data. Firstly, sublimon abundance varied between tissues and between individuals over a range of at least three orders of magnitude, from 95 single copy equivalents, i.e. 190 copies per cell down to <0.1 copies per cell (but still detectable). Secondly, in any individual, a clear and similar pattern of relative abundances in different tissues was evident, regardless of the absolute levels present: heart > muscle > brain
kidney > skin liver, with no detectable sublimons of this class detectable in the pancreas, in the lungs or in the blood. Thirdly, sublimon abundance showed no simple relationship with age. Thus, the 22- and 77-year-old males surveyed both had low levels of sublimons in all tissues, whereas a second male of 80 years of age had very high levels, especially in heart and skeletal muscle. One other skeletal muscle sample from an aged person (82-year-old female, s10), had very high levels of the prevalent sublimon class (190 copies per cell). In addition, we were unable to detect the prevalent 3.75 kb sublimon class in muscle, heart, brain or liver DNA samples from paediatric subjects. The number of sublimons of the prevalent (3.75 kb) class can clearly reach levels representing at least several per cent of total mtDNA. Note, however, that this represents a set of at least 20 and potentially hundreds of different molecular species, because of microheterogeneity at both of the break-points. Each individual sublimon is therefore present at most as a few copies per cell and usually much less.
Sublimons represent partially duplicated and deleted mtDNA
The above analyses do not distinguish between the various molecular forms carrying the same primary rearrangement (Fig. 1). To investigate this, we designed PCR primers capable of amplifying across the potentially duplicated region (Fig. 7). Primers X and Y, lying just within the deleted region, generated long PCR products of 3.8, 7.6 and 
11.5 kb using heart DNA templates in which the prevalent sublimons were easily detectable by PCR.
|
The larger products are exactly as predicted for partial duplications and triplications and cannot be generated from deleted molecules. The junctional primer Z, extending by just 6 nucleotides across the break-point of sublimons with the exact sequence of clones 19 and 22, gave the predicted, duplication-specific product of 3.8 kb, when used in combination with primer X on sublimon-containing template DNA. Together with primer W, lying just within the deleted region, primer Z gave the predicted sublimon-specific 3.6 kb product. The identity of these products was confirmed by direct sequencing. Neither can be generated from wild-type mtDNA. A myocardial DNA template in which significant levels of sublimons could not be detected did not give these products.
These findings indicate that a proportion of the prevalent sublimons exist as partial duplications and probably even as triplications, although the presence, in addition, of deletion monomers and multimers cannot be excluded. Primer pair XZ (but not WZ) also gave a faint band in the region of 7.5 kb, that may derive from deletion dimers. In order to demonstrate more clearly the presence of deletion monomers and multimers, we used a Southern blotting approach. This was technically demanding, since the background smear from the many thousands of wild-type mtDNA molecules per cell completely obliterated any signal from the low percentage of molecules with the anticipated structures, even when using probes specific for the non-deleted region. We therefore probed such blots using a junctional oligonucleotide. However, to obtain significant signal, we found it necessary to reduce the stringency of hybridization to the point where hybridization to undeleted mtDNA was still unavoidable. Nevertheless, under these conditions we did reproducibly detect low molecular weight bands uniquely in those myocardial DNA samples previously shown by PCR to have a high sublimon content (Fig. 8). These were not easily detected in undigested DNA, where the majority of mtDNA apparently migrated as higher order, probably catenated structures. After digestion with a restriction enzyme that cuts mtDNA only once, in the deleted region, bands were detected that correspond in mobility with supercoiled circles of the sizes expected for deletion monomers (3.75 kb), deletion dimers (7.5 kb) and trimers (11.5 kb). None of these species was detected by a probe for the ND4/ND5 region of mtDNA. A BamHI-linearized duplication in the 20 kb range would migrate close to the 16 kb linear band, hence could not easily be detected by this test.
|
Under similar hybridization conditions, digests predicted to cut once in the non-deleted region, for example DraI (Fig. 8d) or AflII (data not shown), gave only the 3.75 kb linear monomer band of the sublimon, as well as the D-loop containing fragment from wild-type mtDNA.
The bands proposed to represent the circular deleted mtDNAs were undetectable in all other tissues, including even muscle, where the predominant sublimon of 3762 bp is 2 bp shorter than that represented by the oligonucleotide used and has a junctional sequence that is probably too poorly matched to cross-hybridize efficiently. DNAs from 0 cells that were probed with the oligonucleotide were completely blank.
In summary, PCR and Southern blot analysis revealed that myocardial sublimons with a precisely defined junctional sequence can exist in a variety of molecular species, with properties consistent with the various interconvertible forms illustrated in Figure 1.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
In this study we set out to analyse the structure and representation of rearranged mtDNA molecules in healthy human tissues and to determine the relationship between such molecules and those found in association with pathological states. To describe them we have adopted the term sublimon, previously used to describe rearranged mtDNA molecules present at low abundance in mitochondria of healthy maize plants (24), with structures akin to those represented much more abundantly in strains exhibiting the mitochondrial cytopathy of cytoplasmic male sterility. Our analysis reveals features of human mtDNA sublimons that are very similar to those characteristic of pathological rearrangements, plus others that seem to be more specific.
Tissue specificity of sublimons
The spectrum of sublimons that appear to be present is strikingly tissue specific, at both the gross- and fine-structure levels. In general, post-mitotic tissues such as muscle or brain contain higher levels of sublimons than perennial tissues such as blood or skin and each tissue has a characteristic set of abundant sublimons. This could reflect different strengths and types of phenotypic selection operating in different tissues, as already implied by the observation of stereotypic shifts in heteroplasmy in different mouse tissues (26). Alternatively, it could indicate tissue-specific differences in the mode or machinery of mtDNA maintenance.
How far do sublimons resemble pathological rearrangements?
Within any given tissue, the spectrum of different sublimons is considerable. However, except for the most abundant species, all are present at low levels, typically of the order of 1–2 copies per cell or less. Even sublimons of the prevalent class, considered collectively, are rarely present at >50 copies per cell and usually much less. Putting together the data of Figure 2 and Table 2, we estimate that at most a few dozen different sublimons are ever represented at 1 copy per cell or more, with many others represented at significantly lower levels. The total number of sublimon molecules per cell is therefore typically of the order of 100 or less, in other words representing only a small percentage of the mtDNA population. In contrast, rearranged mtDNA molecules found in pathological states typically represent much larger fractions of total mtDNA, for example 30–70% for single clonal deletions or partial duplications and at least 10%, commonly much more, for multiple deletions, with some individual rearranged molecules representing up to 10% of total mtDNA in a given tissue.
Structurally, sublimons show a number of features in common with multiple deletions, and many deletions previously reported in diseased individuals (5,8,11,27) correspond precisely or almost precisely with those that we found in controls. The prevalent 3.75 kb sublimon class was also previously found as a partial duplication/triplication at high abundance in the muscle of an individual without overt disease (28). As was previously observed for multiple mtDNA deletions in multiple families with adPEO, MNGIE and IBM, the vast majority of sublimons show break-points in the region of nucleotide position 16 070, near the end of the D-loop (7,8,11,12,29). This CBR shows heterogeneity at the nucleotide level, but the other rearrangement break-points are more scattered, as is also the case in adPEO deletions. This leads to under-representation of the region adjacent to the end of the D-loop in the rearranged molecules (i.e. the region between the replication origins), as seen also in pathological deletions. A striking difference is that sublimons, at least in some tissues, for example heart, kidney or frontal cortex, mainly represent very large deletions lacking OL, whereas pathological deletions generally retain OL. However, because our analysis is based on PCR and plasmid cloning, techniques that both favour short products, it has an intrinsic bias in favour of very large deletions that may not reflect the true spectrum of sublimons present in vivo. This may account, for example, for the intriguing failure to find the common deletion, although it may not in any case be as common in heart as in skeletal muscle.
In sublimons as in pathological rearrangements (5,6), a single primary rearrangement can be associated with multiple molecular species that are theoretically interconvertible by homologous recombination (Fig. 1). The largest sublimon ‘deletions’ that remove OL are clearly found also as partial duplications.
What mechanism generates sublimons?
The location of the common break-point region near the end of the D-loop suggests that sublimons may be generated by a recombinational mechanism: the free 3' end of the paused H-strand could serve as an inappropriate primer for onward DNA synthesis, following strand invasion elsewhere in the genome, a mechanism already proposed in the case of pathological deletions (30). Even without readthrough of the origin, resolution of the resulting structure could generate either a deletion or a partial duplication. The second most commonly observed break-point hotspot region, at least in heart and skeletal muscle, is in the vicinity of the binding site for the transcription termination factor mTERF (31,32), within the gene for tRNA-leuUUR. The termination region of the D-loop is also known to be bound by proteins (33–35). Illegitimate recombination events that generate sublimons might therefore be facilitated by protein–protein interactions, juxtaposing distant regions of the genome.
The tissue specificity of sublimons, even at the microscopic scale, could therefore arise from systematic tissue differences in the properties or representation of mtDNA-binding proteins, some of which may also induce replication pausing (36), to generate specific 3' ends. In vivo footprinting studies have revealed a similar microscale variation in the effects of protein binding on the state of mtDNA, for example in response to thyroid hormone (37), or the different stages of early development in sea urchins (38). Because different sublimons are present in different tissues, in some cases at <1 copy per cell, they cannot be inherited and must instead be generated de novo somatically, perhaps on a continuous basis.
Why do sublimons remain at low levels?
Previous studies have shown that high levels of deleted molecules are highly deleterious to oxidative function (39,40), whereas partial duplications have only very modest effects on phenotype (40,41). However, despite the replicative advantage of partially duplicated molecules carrying supernumerary copies of the replication origin (41,42), their accumulation should be balanced by homologous recombination, resolving them to wild-type mtDNA and deleted molecules that are then counterselected phenotypically (41). Deleted species lacking OL may be lost even without phenotypic selection, as a result of inefficient replication. Although the idea that the light-strand origin OL is strictly necessary for mtDNA replication may not apply universally (43), its presence is nonetheless likely to confer a replicative advantage. Pathological duplications can indeed resolve in cell culture to wild-type and deleted mtDNAs, but the latter typically remain at a lower level (42) and in some cases are undetectable.
We therefore suggest that sublimons and wild-type mtDNA probably co-exist in an equilibrium condition in which sublimons are continuously lost and regenerated, but never accumulate to significant levels except under highly abnormal conditions. The exact position of this equilibrium may vary, however, even between healthy individuals, since sublimon abundance appears to vary over a considerable range even in individuals with no known history of systemic disease.
Could sublimon amplification be a pathological mechanism?
Given that at least some sublimons resemble pathological rearrangements, for example the multiple deletions seen in such conditions as adPEO, we need to re-evaluate thinking about the mechanism underlying the emergence of rearranged mtDNAs in pathological states. Instead of invoking a rare, spontaneous aberration in DNA replication or repair, it may be more appropriate to regard pathological rearrangements as illegitimately amplified sublimons. This suggests three new classes of mechanism by which rearranged mtDNAs could accumulate to physiologically damaging levels: (i) via an enhancement of the process(es) by which sublimons are continuously generated; (ii) via inhibition of the process(es) by which they are usually lost; and (iii) by a (possibly transient) alteration in their selective value, whether at the level of phenotype or replicative advantage.
If replication pausing, strand invasion and protein binding are involved in the generation of sublimons, any change or abnormality in the rate or processivity of DNA replication, the level and activity of mtDNA-binding proteins or the efficiency of lagging strand initiation could shift the equilibrium in favour of sublimon accumulation. This is strongly suggested by the observation that mutations in either of two genes whose products are involved in nucleotide metabolism, namely thymidine phosphorylase (44) and one isoform of the mitochondrial adenine nucleotide translocator (45), have recently been shown to cause multiple deletion disorders (MNGIE and one type of adPEO, respectively). Subtle differences in the spectrum of rearranged mtDNAs in different disorders (10) may give clues as to the precise mechanisms by which sublimons become amplified in pathological states. Since recovery from mtDNA depletion appears to involve a different mode of mtDNA replication from that operating in proliferating cells maintaining a stable mtDNA copy numbr (43), transient or periodic copy number aberrations may be sufficient to promote sublimon accumulation. It has also been proposed that deleted mtDNAs could affect the lysosomal turnover of mitochondria, by virtue of effects on the rate of oxygen radical generation (46); if this is the case, pathological sublimon amplification might be a consequence of abnormalities in this process, rather than anything specifically affecting mtDNA metabolism.
Sublimons and ageing
Our attention was first drawn to sublimons because of the use, by others, of long PCR to detect deleted mtDNAs that were purported to accumulate in the course of somatic ageing and in association with acquired myocardial diseases. We demonstrated earlier that the sensitivity of the technique is so great that it can detect rearranged molecules present at only one or a few copies per cell, even against a background of thousands of molecules per cell of ‘wild-type’ mtDNA (17), an assertion supported by the analysis presented here. This raises considerable doubts as to the significance of reports linking deleted mtDNAs to ageing.
However, sublimon levels clearly vary between individuals and it cannot be excluded that they might accumulate under some conditions, even in an essentially healthy person, and could contribute to age-related degeneration. Our findings indicate that there is no simple relationship between age and sublimon abundance. A careful and rigorously controlled study is now required to evaluate whether sublimon amplification does occur at the expense of normal mtDNA in at least some individuals or tissues and whether this has any phenotypic consequences.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Patients and tissue samples
Frozen samples of heart muscle were obtained from medico-legal autopsies performed in 1991–1992. These comprise five cases of alcoholic cardiomyopathy (see legend to Fig. 2), originally collected for other studies, plus controls excluded from having any heart disease or any known history of alcohol abuse, as previously (14). Paediatric biopsy samples (11 skeletal muscle, 4 heart, 2 liver, 1 brain from children from 14 weeks gestation to 14 years) were collected in Tampere University Hospital (Finland) and the John Radcliffe Hospital (Oxford, UK) from individuals without diagnosed mitochondrial or other relevant systemic disease. Muscle biopsies from six adult controls (ages 18–82) were taken at the Neurology clinic in University of Oulu (Finland) or in Oxford. Post-mortem tissue samples (time from death to autopsy 2–3 days) were taken from each of six other individuals (s1–s6) during the routine autopsy protocol (for the selected tissues see Table 2) and the DNA was extracted immediately after sampling. Control sperm samples from anonymous males known to have fathered children were kindly provided by Dr L. Wichmann (Department of Anatomy, University of Tampere, Finland).
Cell lines
143B osteosarcoma-derived and A549 lung carcinoma-derived 0 cells (lines 206 and B2, respectively) (47,48) were cultured in Dulbecco’s modified Eagle’s medium/10% fetal bovine serum, supplemented with pyruvate and uridine (47).
DNA extraction
DNA was prepared by standard methods as previously (14), essentially by proteinase K digestion and phenol–chloroform extraction, from tissue samples and cell lines.
Oligonucleotide primers
Custom-designed primers for PCR and DNA sequencing were purchased from DNA Technology (Aarhus, Denmark) or Genset (Paris, France). The adjacent primer pair OK1H/OK2L, located in the mtDNA control region and primer FR31H from within the COXI gene, were as described previously (14). Additional PCR primers used in the study, all given as 5'
3' and, where appropriate, followed by the corresponding co-ordinates in the human mtDNA sequence (49), are as follows: mt3150 (5'-TACTTCACAAAGCGCCTTC-3', 3150–3168), mt2204 (5'-TTCAAGCTCAACACCCACTA-3', 2204–2223), mt16153FAM (5'-CAGGTGGTCAAGTATTTATGG-3', 16153–16133, incorporating a 5' FAM fluorescent label), W (5'-CCGGTCTGAACTCAGATCAC-3', 3080–3060), X (5'-GTTGGCCATGGGTATGTTGT-3', 3321–3302), Y (5'-TCCACCATTAGCACCCAAAG-3', 15 976–15 995) and Z (5'-GGTAATCGCATAAAACCATCA-3', 3247–3261 and 16 072–16 077). PCR primers for cytochrome P450 2E1 were: DraIR (5'-TCCCAAAGTGCCAGGATT-3') and DraIFROX (5'-ATCATGGCTCATTGTAGCTTC-3', incorporating a 5' ROX fluorescent label). Oligonucleotide Junc-1 (5'-GGTAATCGCATAAAACCATCAACAACCGCT-3', nucleotide positions 3247–3261 and 16 072–16 086 of human mtDNA) was used as a hybridization probe. In addition, a series of sequencing primers corresponding with appropriate portions of the human mtDNA sequence were designed for analysis of cloned sublimons.
Long PCR
Reactions were carried out using primer pair OK1H/OK2L and analysed by agarose gel electrophoresis exactly as described previously (14). The amounts of template DNA were as indicated in the legend to Figure 2. For cloning, scaled-up long PCR reactions used 26 ng of a control heart DNA template (individual BT91, 49-year-old male) that had given diverse and abundant sublimon products. For cloning the largest sublimons (i.e. those with the shortest deletions) PCR was carried out using ‘sub-genomic’ primer pair FR31H/OK2L (14). Identical products were obtained using nested PCR, on gel-purified products of the first PCR reaction, migrating in the 7–15 kb size range.
Molecular cloning and sequencing
Long PCR products were cloned into pCR-XL-TOPO vector (Invitrogen BV, Groningen, The Netherlands) by the manufacturer’s recommended protocol. Sixty-five clones (42 obtained by PCR using ‘genome length’ primers, plus 23 derived using ‘sub-genomic’ primers) were selected, sized by miniprep analysis and sequenced using BigDye terminator chemistry (Applied Biosystems, Foster City, CA), with a combination of vector-specific and mtDNA-specific primers. Sequencing products were analysed by capillary electrophoresis on an Applied Biosystems 310 Genetic Analyzer, using the manufacturer’s software.
Fluorescent PCR
Fluorescent PCR to characterize specific sublimon subclasses used fluorescent primer mt16153FAM, plus either of primers mt3150 or mt2204. For semi-quantitative analysis, multiplex reactions included also primers DraIR and DraIFROX as an internal single copy gene standard. After a manual hot start for 3 min at 95°C, a PCR cycle consisting of 30 s at 95°C, 20 s at 59°C and 20 s at 72°C was repeated 22–30 times and the fluorescent products were analysed by capillary electrophoresis using GeneScan software on an Applied Biosystems 310 Genetic Analyzer. An additional 7 min final extension step to maximize the 3' A-overhang addition efficiency did not affect the relative or absolute amounts of the fluorescent products in any sample tested. Reaction products resolved at the nucleotide level were quantified as peak areas in the electrophoretogram and ratios computed of the amount of sublimon product versus the single copy standard, at increasing cycle number. Data from points where saturation had not yet occurred were pooled, to extrapolate a mean copy number per cell, i.e. twice the number of single copy equivalents, for all sublimons of the prevalent 3.75 kb class detected by the primers, considered collectively.
Southern blotting
Aliquots (5 or 10 µg) of genomic DNA, with or without BamHI digestion, were fractionated by 0.7% agarose gel electrophoresis in the absence of ethidium bromide. Southern blotting, hybridization and washing were as described previously (14). The sublimon junctional oligonucleotide Junc-1 was end-labelled with [
-32P]ATP (6000 Ci/mmol; Amersham Pharmacia Biotech, Little Chalfont, UK) using T4 polynucleotide kinase (MBI Fermentas, Vilnius, Lithuania) in the supplied reaction buffer A, according to the manufacturer’s instructions, before addition to the hybridization reaction. Hybridization was carried out in a standard buffer (14) at 50°C for 16 h after a pre-hybridization step of 1 h at 50 °C without the probe. The blot was washed for 10 min in 5x SSC, 0.1% SDS, 0.025 M sodium phosphate pH 6.8 and for 5 min in 1x SSC, 0.1% SDS, 0.025 M sodium phosphate pH 6.8, both at 50°C. The radioactive signal was detected by autoradiography. Stripping and reprobing with a probe for the ND4/ND5 region of human mtDNA were as described previously (14).
|
ACKNOWLEDGEMENTS |
|---|
We thank Mino Cantatore, Vito Pesce, Pierre Rustin, Jyrki Kaukonen and Anu Suomalainen for many useful discussions. This work was supported by grants from the Finnish Academy, Muscular Dystrophy Group, British Diabetic Association, Tampere University Hospital Medical Research Fund, Yrjö Jahnsson Foundation, Finnish Foundation of Alcohol Research and the Pirkanmaa Region Fund of the Finnish Cultural Foundation.
|
FOOTNOTES |
|---|
+ To whom correspondence should be addressed at: Institute of Medical Technology, 33014 University of Tampere, Finland. Tel: +358 3 215 7731; Fax: +358 3 215 7731; Email: howy.jacobs@uta.fi
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
1 Grossman, L.I. and Shoubridge, E.A. (1996) Mitochondrial genetics and human disease. Bioessays, 18, 983–991.[ISI][Medline]
2 DiMauro, S., Bonilla, E., Davidson, M., Hirano, M. and Schon, E.A. (1998) Mitochondria in neuromuscular disorders. Biochim. Biophys. Acta, 1366, 199–210.[Medline]
3 Zeviani, M., Tiranti, V. and Piantadosi, C. (1998) Mitochondrial disorders. Medicine, 77, 59–72.[Medline]
4 Rotig, A., Bourgeron, T., Chretien, D., Rustin, P. and Munnich, A. (1995) Spectrum of mitochondrial DNA rearrangements in the Pearson marrow-pancreas syndrome.Hum. Mol. Genet., 4, 1327–1330.
5 Moslemi, A.R., Melberg, A., Holme, E. and Oldfors, A. (1996) Clonal expansion of mitochondrial DNA with multiple deletions in autosomal dominant progressive external ophthalmoplegia. Ann. Neurol., 40, 707–713.[ISI][Medline]
6 Poulton, J., Deadman, M.E., Bindoff, L., Morten, K., Land, J. and Brown, G. (1993) Families of mtDNA re-arrangements can be detected in patients with mtDNA deletions: duplications may be a transient intermediate form. Hum. Mol. Genet., 2, 23–30.
7 Zeviani, M., Servidei, S., Gellera, C., Bertini, E., DiMauro, S. and DiDonato, S. (1989) An autosomal dominant disorder with multiple deletions of mitochondrial DNA starting at the D-loop region. Nature, 339, 309–311.[Medline]
8 Moslemi, A.R., Melberg, A., Holme, E. and Oldfors, A. (1999) Autosomal dominant progressive external ophthalmoplegia: distribution of multiple mitochondrial DNA deletions. Neurology, 53, 79–84.
9 Papadimitriou, A., Comi, G.P., Hadjigeorgiou, G.M., Bordoni, A., Sciacco, M., Napoli, L., Prelle, A., Moggio, M., Fagiolari, G., Bresolin, N. et al. (1998) Partial depletion and multiple deletions of muscle mtDNA in familial MNGIE syndrome. Neurology, 51, 1086–1092.
10 Carrozzo, R., Hirano, M., Fromenty, B., Casali, C., Santorelli, F.M., Bonilla, E., DiMauro, S., Schon, E.A. and Miranda, A.F. (1998) Multiple mtDNA deletions features in autosomal dominant and recessive diseases suggest distinct pathogeneses. Neurology, 50, 99–106.
11 Moslemi, A.R., Lindberg, C. and Oldfors, A. (1997) Analysis of multiple mitochondrial DNA deletions in inclusion body myositis. Hum. Mutat., 10, 381–386.[ISI][Medline]
12 Jansson, M., Darin, N., Kyllerman, M., Martinsson, T., Wahlstrom, J. and Oldfors, A. (2000) Multiple mitochondrial DNA deletions in hereditary inclusion body myopathy. Acta Neuropathol., 100, 23–28.[Medline]
13 Santorelli, F.M., De Joanna, G., Casali, C., Tessa, A., Siciliano, G., Amabile, G.A., Pierelli, F., Vilarinho, L. and Santoro, L. (2000) Multiple mtDNA deletions: clinical and molecular correlations. J. Inherit. Metab. Dis., 23, 155–161.[ISI][Medline]
14 Kajander, O.A., Kunnas, T.A., Perola, M., Lehtinen, S.K., Karhunen, P.J. and Jacobs, H.T. (1999) Long-extension PCR to detect deleted mitochondrial DNA molecules is compromised by technical artefacts. Biochem. Biophys. Res. Commun., 254, 507–514.[ISI][Medline]
15 Ruppert, V. and Maisch, B. (2000) Mitochondrial DNA deletions in cardiomyopathies. Herz, 25, 161–167.[ISI][Medline]
16 Tengan, C.H. and Moraes, C.T. (1996) Detection and analysis of mitochondrial DNA deletions by whole genome PCR. Biochem. Mol. Med., 58, 130–134.[ISI][Medline]
17 Kajander, O.A., Poulton, J., Spelbrink, J.N., Rovio, A., Karhunen, P.J. and Jacobs, H.T. (1999) The dangers of extended PCR in the clinic. Nature Med., 5, 965–966.[ISI][Medline]
18 Ozawa, T. (1999) Mitochondrial genome mutation in cell death and aging. J. Bioenerg. Biomembr., 31, 377–390.[ISI][Medline]
19 Kowald, A. (1999) The mitochondrial theory of aging: do damaged mitochondria accumulate by delayed degradation? Exp. Gerontol., 34, 605–612.[ISI][Medline]
20 Kovalenko, S.A., Kopsidas, G., Kelso, J., Rosenfeldt, F. and Linnane, A.W. (1998) Tissue-specific distribution of multiple mitochondrial DNA rearrangements during human aging. Ann. N. Y. Acad. Sci., 854, 171–181.
21 Melov, S., Schneider, J.A., Coskun, P.E., Bennett, D.A. and Wallace, D.C. (1999) Mitochondrial DNA rearrangements in aging human brain and in situ PCR of mtDNA. Neurobiol. Aging, 20, 565–571.[ISI][Medline]
22 Gadaleta, M.N., Cormio, A., Pesce, V., Lezza, A.M. and Cantatore, P. (1998) Aging and mitochondria. Biochimie, 80, 863–870.[Medline]
23 Li, Y.Y., Hengstenberg, C. and Maisch, B. (1995) Whole mitochondrial genome amplification reveals basal level multiple deletions in mtDNA of patients with dilated cardiomyopathy. Biochem. Biophys. Res. Commun., 210, 211–218.[ISI][Medline]
24 Small, I., Suffolk, R. and Leaver, C.J. (1989) Evolution of plant mitochondrial genomes via substoichiometric intermediates. Cell, 58, 69–76.[ISI][Medline]
25 Yesodi, V., Izhar, S., Gidoni, D., Tabib, Y. and Firon, N. (1995) Involvement of two different urf-s related mitochondrial sequences in the molecular evolution of the CMS-specific S-Pcf locus in petunia. Mol. Gen. Genet., 248, 540–546.[ISI][Medline]
26 Jenuth, J.P., Peterson, A.C. and Shoubridge, E.A. (1997) Tissue-specific selection for different mtDNA genotypes in heteroplasmic mice. Nature Genet., 16, 93–95.[ISI][Medline]
27 Kleinle, S., Wiesmann, U., Superti-Furga, A., Krahenbuhl, S., Boltshauser, E., Reichen, J. and Liechti-Gallati, S. (1997) Detection and characterization of mitochondrial DNA rearrangements in Pearson and Kearns–Sayre syndromes by long PCR. Hum. Genet., 100, 643–650.[ISI][Medline]
28 Tengan, C.H. and Moraes, C.T. (1998) Duplication and triplication with staggered breakpoints in human mitochondrial DNA. Biochim. Biophys. Acta, 1406, 73–80.[Medline]
29 Zeviani, M., Bresolin, N., Gellera, C., Bordoni, A., Pannacci, M., Amati, P., Moggio, M., Servidei, S., Scarlato, G. and DiDonato, S. (1990) Nucleus-driven multiple large-scale deletions of the human mitochondrial genome: a new autosomal dominant disease. Am. J. Hum. Genet., 47, 904–914.[ISI][Medline]
30 Zeviani, M. (1992) Nucleus-driven mutations of human mitochondrial DNA. J. Inherit. Metab. Dis., 15, 456–471.[ISI][Medline]
31 Kruse, B., Narasimhan, N. and Attardi, G. (1989) Termination of transcription in human mitochondria: identification and purification of a DNA binding protein factor that promotes termination. Cell, 58, 391–397.[ISI][Medline]
32 Fernandez-Silva, P., Martinez-Azorin, F., Micol, V. and Attardi, G. (1997) The human mitochondrial transcription termination factor (mTERF) is a multizipper protein but binds to DNA as a monomer, with evidence pointing to intramolecular leucine zipper interactions. EMBO J., 16, 1066–1079.[ISI][Medline]
33 Madsen, C.S., Ghivizzani, S.C. and Hauswirth, W.W. (1993) Protein binding to a single termination-associated sequence in the mitochondrial DNA D-loop region. Mol. Cell. Biol., 13, 2162–2171.
34 Kumar, S., Suzuki, H., Onoue, S., Suzuki, S., Hattori, N. and Ozawa, T. (1995) Rat mitochondrial mtDNA-binding proteins to inter-specifically conserved sequences in the displacement loop region of vertebrate mtDNAs. Biochem. Mol. Biol. Int., 36, 973–981.[ISI][Medline]
35 Roberti, M., Musicco, C., Polosa, P.L., Milella, F., Gadaleta, M.N. and Cantatore, P. (1998) Multiple protein-binding sites in the TAS-region of human and rat mitochondrial DNA. Biochem. Biophys. Res. Commun., 243, 36–40.[ISI][Medline]
36 Qureshi, S.A. and Jacobs, H.T. (1993) Two distinct, sequence-specific DNA-binding proteins interact independently with the major replication pause region of sea urchin mtDNA. Nucleic Acids Res., 25, 2801–2808.
37 Enriquez, J.A, . Fernandez-Silva, P., Garrido-Perez, N., Lopez-Perez, M.J., Perez-Martos, A. and Montoya, J. (1999) Direct regulation of mitochondrial RNA synthesis by thyroid hormone. Mol. Cell. Biol., 19, 657–670.
38 Roberti, M., Polosa, P.L., Musicco, C., Milella, F., Qureshi, S.A., Gadaleta, M.N., Jacobs, H.T. and Cantatore P. (1999) In vivo mitochondrial DNA-protein interactions in sea urchin eggs and embryos. Curr. Genet., 34, 449–458.[ISI][Medline]
39 Spelbrink, J.N., Zwart, R., van Galen, M.J.M. and van den Bogert, C. (1997) Preferential amplification and phenotypic selection in a population of deleted and wild-type mitochondrial DNA in cultured cells. Curr. Genet., 32, 115–124.[ISI][Medline]
40 Tang, Y., Schon, E.A., Wilichowski, E., Vazquez-Memije, M.E., Davidson, E. and King, M.P. (2000) Rearrangements of human mitochondrial DNA (mtDNA): new insights into the regulation of mtDNA copy number and gene expression. Mol. Biol. Cell., 11, 1471–1485.
41 Holt, I.J., Dunbar, D.R. and Jacobs, H.T. (1997) Behaviour of a population of partially duplicated mitochondrial DNA molecules in cell culture: segregation, maintenance and recombination dependent upon nuclear background. Hum. Mol. Genet., 6, 1251–1260.
42 Tang, Y., Manfredi, G., Hirano, M. and Schon, E.A. (2000) Maintenance of human rearranged mitochondrial DNAs in long-term cultured transmitochondrial cell lines. Mol. Biol. Cell, 11, 2349–2358.
43 Holt, I.J., Lorimer, H.E. and Jacobs, H.T. (2000) Coupled leading- and lagging-strand synthesis of mammalian mitochondrial DNA. Cell, 100, 515–524.[ISI][Medline]
44 Nishino, I., Spinazzola, A. and Hirano, M. (1999) Thymidine phosphorylase gene mutations in MNGIE, a human mitochondrial disorder. Science, 283, 689–692.
45 Kaukonen, J., Juselius, J.K., Tiranti, V., Kyttala, A., Zeviani, M., Comi, G.P., Keranen, S., Peltonen, L. and Suomalainen, A. (2000) Role of adenine nucleotide translocator 1 in mtDNA maintenance. Science, 289, 782–785.
46 de Grey, A.D. (1997) A proposed refinement of the mitochondrial free radical theory of aging. Bioessays, 19, 161–166.[ISI][Medline]
47 King, M.P. and Attardi, G. (1989) Human cells lacking mtDNA: repopulation with exogenous mitochondria by complementation. Science, 246, 500–503.
48 Bodnar, A.G., Cooper, J.M., Holt, I.J., Leonard, J.V. and Schapira, A.H.V. (1993) Nuclear complementation restores mtDNA levels in cultured cells from a patient with mtDNA depletion. Am. J. Hum. Genet., 53, 663–669.[ISI][Medline]
49 Anderson, S., Bankier, A.T., Barrell, B.G., de Bruijn, M.H., Coulson, A.R., Drouin, J., Eperon, I.C., Nierlich, D.P., Roe, B.A., Sanger, F. et al. (1981) Sequence and organization of the human mitochondrial genome. Nature, 290, 457–465.[Medline]
