We have studied the dynamics of mitochondrial DNA maintenance and segregation in human cells using serial cybrid transfer of partially duplicated mitochondrial DNA, from a mitochondrial myopathy patient, to two distinct recipient cell types. The results indicate two radically different outcomes dependent upon nuclear background. In one case (lung carcinoma) there is systematic loss of the partial duplication by an implied recombinational mechanism. In another nuclear background (osteosarcoma) the duplicated molecules can survive, having only a marginal effect on mitochondrial respiratory function. Moreover, in the osteosarcoma nuclear background further disturbances of mtDNA maintenance frequently follow from cybrid transfer. These are progressive, catastrophic loss of mtDNA and further rearrangement to generate partially triplicated molecules. The results imply differential expression of nuclear genes regulating mtDNA copy number, replication and recombination in different human cell types.
Many mutations in mitochondrial DNA (mtDNA) are associated with human disease. These mutations can be divided into two classes: point mutations and large-scale rearrangements. Large-scale rearrangements generally occur as partial deletions (1 ) or partial duplications (2 ), and are invariably associated with heteroplasmy, i.e., rearranged and apparently wild-type mtDNA molecules coexist. Metazoan mitochondria, like those of lower eukaryotes, can be depleted entirely of their mtDNA by long-term exposure to a low dose of ethidium bromide (3 ,4 ). Immortal cells lacking endogenous mtDNA (termed [rho]o) can be repopulated with donor mitochondria (containing mtDNA) from normal or patient-derived cells (3 ,5 ). In principle, [rho]o cells provide an ideal system for studying the phenotypic effects of mtDNA mutations in a control nuclear background. The introduction of partially deleted mtDNAs into a HeLa [rho]o cell line demonstrated that respiratory function was impaired at levels of rearranged molecules >65% of total mtDNA. At a level of 78% partially deleted mtDNA there was almost complete failure of mitochondrial translation (6 ). A priori, one might expect partially duplicated (pd) mitochondrial genomes to be benign, as no mitochondrial genes are missing. However, either the imbalance of tRNAs or mitochondrial polypeptides, or a putative fusion gene product have been postulated to perturb mitochondrial function. Moreover, the similarities in clinical features of patients with deleted and duplicated mtDNA (2 ,7 -13 ) strongly suggest that duplications, like deleted mtDNAs, are in some way pathological, or at least that they are inextricably linked to a pathological process.
More recently, temporal analysis of mtDNA mutations in culture has revealed dramatic changes in the proportions of mutant and wild-type mtDNA. Analysis of the 3243 point mutation associated with MELAS (mitochondrial encephalomyopathy, lactic acidosis and stroke-like episodes) in repopulated [rho]o cells showed that the mutant population of mtDNA could increase or decrease dependent on the cellular background (14 ,15 ), while deleted mtDNA in cultured fibroblasts increased in the presence of uridine and decreased when uridine was absent (16 ).
To investigate whether pd mtDNA molecules are maintained, selected or eliminated in a control nuclear background, patient-derived mitochondria were introduced into [rho]o 143B osteosarcoma cells, and the proportion of pd and wild-type mtDNA molecules monitored in individual cybrid clones grown under various conditions. Mitochondria from a cybrid clone homoplasmic for pd mtDNA were then transferred by serial cybridization to two different recipient cell lines in order to investigate the stability of rearranged mtDNA molecules in different cell types. Cybrids were analysed further for respiratory impairment, in order to determine the contribution of respiratory phenotype to the observed segregation phenomena.
The results indicate that rearranged mtDNA molecules containing additional copies of the heavy strand replication origin are at a replicative advantage, that they nevertheless confer a metabolic disadvantage to cells, that they can undergo recombination resulting in elimination or amplification of the duplicated region of the genome, and that their presence in cells can perturb copy number control leading to catastrophic depletion of mtDNA. The balance between these various processes differs between cells of different nuclear/developmental backgrounds, offering a potential explanation for many of the hitherto unexplained phenomena associated with mtDNA disease.
In order to investigate the factors influencing the maintenance of pd mtDNA molecules, cytoplast fusion was used to create cybrid clones containing mixtures of wild-type and pd molecules in a defined nuclear background. The representation of wild-type and rearranged molecules was followed over time, under various conditions. Mitochondria from one derivative that resolved to homoplasmy for pd mtDNA molecules were serially transferred to different recipient cells, in order to study the specific influence of nuclear background on the maintenance of rearranged mtDNA molecules. The results of these two series of experiments revealed a variety of possible outcomes. These included segregation of mitochondrial genotypes, loss or maintenance of heteroplasmy, removal of the duplicated portion of mtDNA, catastrophic depletion of mtDNA and further rearrangement (summarized in Fig. 1 ). The factors deduced to be influencing these outcomes were nuclear genetic background (osteosarcoma verus lung carcinoma) and the presence of uridine in the medium. Some of the observed events are not attributable to any measured variable and may be the result of random occurrences. The experimental procedure giving each of the outcomes is now presented in detail.
Partially deleted mtDNA molecules are generally lost rapidly in myoblasts cultured in growth medium lacking uridine (21 ), presumably through intercellular competition. In the present case, in which the rearranged mtDNA contained a partial duplication rather than a partial deletion,the proportion of pd mtDNA molecules was 25% in the proband's muscle (Fig. 2 , lane 4) and 75% in passage eight myoblasts grown in the absence of uridine (data not shown). Patient myoblasts at passage six were enucleated and fused to [rho]o osteosarcoma cells (cell line 143B.206 used previously in our own and other studies; 3 ,6 ,15 ). At day 61, a clonal cell line with 25% pd mtDNA was split and grown with and without uridine. Screening of the cells at day 224 revealed a decrease in the level of pd mtDNA to 5% in the cells grown in the absence of uridine. Conversely, the proportion of pd mtDNA had increased to 75% in those cells maintained in uridine (Fig. 2 , lanes 5-7). These cells were recloned and four sub-clones were screened after amplification. All four appeared homoplasmic, three for wild-type mtDNA (designated 206.dup1[alpha], -[beta] and -[delta]), the other (206.dup1[gamma]) for pd mtDNA (Fig. 2 , lane 2). Transient ethidium bromide treatment (22 ) of the same cell line (206.dup 1) and recloning in the presence of uridine yielded four sub-clones with ~15, 20, 90 and 90% pd mtDNA. The two clones with 90% pd mtDNA were cultured further, again in the presence of uridine: they segregated to 100% pd mtDNA (206.dup 1A) and 100% wild-type mtDNA (206.dup 1B). These results indicate that a mixture of wild-type and pd mtDNAs may be completely segregated to homoplasmy in either direction in osteosarcoma cells, in the absence of selection.
The mtDNA of sub-clone 206.dup 1A, was rescreened by PstIdigestion 471 days post-fusion and found to contain an additional band (data not shown). The additional fragment migrated more slowly than the 20 kb fragment that included the pd region of mtDNA and represented approximately half of total mtDNA. An additional band was also seen when HpaIand EcoRI digests were examined. Further restriction endonuclease analysis with ApaI, AvaI, EcoRV, SspI and XbaI yielded mtDNA fragments identical to those obtained with DNA from cells with only pd mtDNAs. This suggested that the additional mtDNA formed part of a partially triplicated (pt) molecule, i.e., the duplicated portion of mtDNA found in the patient had been reduplicated in cell line 206.dup 1A. Later experiments described below confirmed this supposition.
The cells containing the presumed pt mtDNAs were named 206.dup 1A* and recloned 471 days after the original fusion of patient-derived myoblasts and osteosarcoma [rho]o cells. DNA from three sub-clones was screened 18 days later (Fig. 3 A). One sub-clone, 206.dup1A1, contained almost exclusively pd mtDNA. In the other two sub-clones (206.dup1A2 and 206.dup1A3) there was a high level of pt mtDNA (98 and 75% respectively) the visible remainder being pd mtDNA (Fig. 3 A, lanes 5 and 6). Frozen stocks of cybrid 206.dup 1A were regrown and the DNA screened in an attempt to determine when the pt molecules were first detectable. Cells were available that dated back to 152 days post-fusion, 50 days post-ethidium bromide treatment: in these cells the partial triplication was detectable but accounted for no more than 2% of mtDNA molecules (Fig. 3 B). Thus, in the mass culture and sub-clones 206.dup1A2 and 206.dup1A3 the pt mtDNA species had largely displaced pd mtDNA, although it had been a minor species a year earlier.
DNA from the 206.dup1A2 cell line, that contained predominantly pt mtDNA molecules, was digested with EcoRI, EcoRV and XbaI (Fig. 4 ).The fragments obtained were again the same as those obtained by cutting pd mtDNA molecules. However, the 4.7 kb XbaI junction fragment was clearly present in larger relative amount compared with the wild-type 4.5 kb fragment (Fig. 4 , lane 8 versus 9). The ratio of the 4.7 and 4.5 kb fragments was estimated by phosphorimager analysis as close to 2 in the 206.dup 1A2 cells. This supports the interpretation thatthe novel species is a pt mtDNA molecule. We conclude that further rearrangement, via an implied recombination mechanism, can occur as a rare event in cells cultured in the absence of selection.The increase in the amount of pt molecules over time is, moreover, consistent with the idea that additional copies of the leading-strand replication origin confer a replicative advantage. The various types of rearranged mtDNA molecule reported here, including pt mtDNA, are shown in diagrammatic form in Figure 5 .
In order to determine whether nuclear background influences the outcome of cybridization,cytoplasts from an osteosarcoma cybrid subclone homoplasmic for pd mtDNA (206.dup1[gamma]) were fused with lung carcinoma A549.B2 [rho]o cells carrying a neomycin resistance marker. Cybrid clones (B2.neo.dup) were selected by growth in the presence of G418 and the absence of uridine, and their mtDNA types were analysed26 days after fusion. Three of five cybrids analysed had no detectable pd mtDNA molecules (Fig. 6 ). The two remaining cybrids had only low amounts of rearranged mtDNA (25 and 20%) which was subsequently lost altogether over the course of 3 months, while the total amount of mtDNA remained stable (Fig. 6 ). Uridine in the growth medium had no discernible effect on the direction or rate of segregation in either of these two cybrids. Given that the donor line was apparently homoplasmic for pd molecules, these results are strongly suggestive of an active recombination mechanism in the lung carcinoma nuclear background, resolving rearranged molecules to wild-type. It may be further presumed, consistent with the findings described in the preceding section, that there is concomitant loss of the reciprocal (deleted mtDNA) product because of its replication incompetence. A second identical fusion yielded similar results as did a fusion between the same donor cytoplasts and A549.B2 [rho]o cells that lacked a neomycin resistance gene (Figs 6 and 7 ). This last experiment indicated that the phenomenon observed had nothing to do with the drug-selection regime per se.
Serial transfer of mitochondria containing 100% pd mtDNA to the original recipient cell type was achieved by cytoplast fusion using sub-clone 206.dup 1[gamma] as donor, and a neomycin-resistant derivative of the osteosarcoma [rho]o cells (143b.206.neo) as recipient. This fusion was carried out three times. The first such fusion yielded two cybrid clones (206.neo.dup i and ii), each initially capable of growing in medium containing G418 but lacking uridine. Southern blotting revealed that, as expected, both cybrids contained predominantly pd mtDNA (>95%). However, wild-type molecules were now detected in both cases, consistent with the operation of a recombination mechanism in this nuclear background also, albeit a much less efficient one than that inferred in lung carcinoma cells. The relative proportions of pd and wild-type mtDNA molecules were stable in both cybrid clones. However, in one of them (206.neo.dup i) the total mtDNA copy number decreased with time (Fig. 8 ). At 65 days post-fusion the amount of mtDNA relative to nuclear DNA, as judged by probing for nuclear 18S rDNA, was 0.52 and 9.2 respectively in the two clones, compared with a ratio of 6.1 in the lung carcinoma parental cell line (A549). Cytochrome c oxidase activity and intact cell oxygen consumption rates were assayed between days 55 and 84 post-fusion; these assays revealed a marked decrease in respiratory function in the cybrid undergoing mtDNA depletion (Fig. 9 ). By 159 days post-fusion the mtDNA:nuclear DNA ratio had decreased to 0.08 in cybrid 206.neo.dup i (Fig. 8 , lane 7), but remained substantially unaltered (8.7) in sister cybrid 206.neo.dup ii (Fig. 8 , lane 5). In 206.neo.dup i cells losing their mtDNA, uridine became an essential supplement for growth at ~160 days post-fusion. From 160 days post-fusion, 206.neo.dup i cells maintained in the absence of uridine stopped growing and by day 225 were dead. A parallel 206.neo.dup i culture, supplemented with uridine, grew at a similar rate to controls during the same period. DNA extracted from these cells at 224 days post-fusion contained no detectable mtDNA relative to nuclear encoded 18S rDNA.
The loss of the pd mtDNA molecules in the lung carcinoma cell background meant that it was only possible to assess the respiratory consequences of pd mtDNA in the osteosarcoma cell background. Measurements of oxygen consumption of intact cells and of permeabilised cells supplemented with respiratory substrates failed to distinguish osteosarcoma cybrids with and without pd mtDNA, nor was there any consistent enzyme deficiency of respiratory complexes I, II/III or IV in cells with >96% pd mtDNA (23 ). In vitro labelling of mitochondrial translation products did not reveal any qualitative or quantitative difference between cybrid clones or sub-clones with 0 and >96% pd mtDNA, nor was a fusion product visible even when employing a gel system capable of resolving a peptide of ~1.5 kDa (24 ) (data not shown). The absence of the predicted fusion peptide (11 ) was not surprising as it would contain only two methionine residues and the complete ND1 polypeptide produces only a weak signal in this assay, thus a much truncated ND1-ND5 chimera (15 amino acids in length) would be highly likely to generate a signal below the detection limit.
Clones with >96% pd mtDNA, however, did show an increased lactate:pyruvate ratio (Fig. 10 ). In humans, exercise-induced high serum lactate levels are a good diagnostic indicator of mitochondrial myopathy and respiratory chain deficiency (25 ). Moreover, measurement of lactate and pyruvate production in cultured fibroblasts can be used to identify individuals with impaired aerobic metabolism (26 ). The ratio oflactate:pyruvate production in cybrids with no pd mtDNA was consistently lower than those with 97, 98 and 100% pd mtDNA (Fig. 10 ). Application of the two-tailed t-test indicated that this difference was highly significant (P < 0.001). These results suggest that pd mtDNA leads to increased dependence on glycolytic ATP production, presumably as a consequence of mildly impaired respiratory chain function. The highest lactate:pyruvate ratio was obtained from the cell line containing predominantly pt mtDNAs (Fig. 10 , lane 11).
The temporal analysis of pd mtDNAs in two types of cell cybrid, as reported here, indicates a variety of possible outcomes that appear to be dependent, in part, on the nuclear background of the recipient cell. Most importantly, in lung carcinoma cybrids, pd mtDNAs were systematically lost, this process being associated with a concomitant reappearance of wild-type mtDNAs which then progressively increased. This strongly suggests that human A549 lung carcinoma cells contain an active recombination system within their mitochondria. Wild-type molecules did reappear, albeit at a low level, in several osteosarcoma cybrid clones that received mtDNA from the donor homoplasmic for pd molecules, although in no case was a subsequent increase in wild-type molecules documented in this nuclear background. This finding implies that the mitochondria of osteosarcoma cells also contain a recombination system, but that it is much less efficient than that of lung carcinoma cells, and thus is only apparent as a transient phenomenon, under conditions of marked respiratory stress such as encountered during serial cybridization of pd mtDNA from a homoplasmic donor. Two additional phenomena were observed in osteosarcoma cybrids containing pd mtDNA. First, a progressive and ultimately catastrophic loss of mtDNA was observed in some neomycin-resistant osteosarcoma cybrid clones containing pd mtDNA. In contrast, there was no disturbance of mtDNA copy number control when wild-type mtDNAs derived from the same patient were introduced into osteosarcoma cells. Second, pt mtDNA molecules appeared independently in two osteosarcoma cybrids. Biochemical analysis demonstrated that this rearrangement enhanced the mild respiratory impairment seen in cells containing pd (as opposed to wild-type) mtDNA molecules. Thus, phenotypically disadvantageous mtDNA variants can be maintained and even positively selected in proliferating cells, presumably by selfish mechanisms.
The observed increase in pd mtDNA molecules in the presence of uridine and decrease in its absence are in agreement with the results of Bourgeron et al. (16 ) for deleted mtDNA. [rho]o cells are uridine auxotrophs, hence uridine might be expected to stabilise deleterious mutant mtDNAs. However, segregation in opposing directions occurred in two sub-clones cultured under seemingly identical conditions, that included supplementation with uridine suggesting that factors other than uridine contribute to the changes seen in the level of rearranged mtDNAs. These segregation phenomena are unexplained, and may reflect further selective stresses on mitochondrial function that operate in the micro-environment of cells.It is possible that variations in the topological state of mtDNA (or of the mitochondria themselves) are involved. How these might respond to differences in culture conditions, or at random, remains to be documented.
Partially triplicated mtDNA molecules presumably arise via some form of recombination event. This, however, leaves unresolved the question of how the newly generated pt species came to be fixed. There are several possible mechanisms of fixation of variants in organelle genomes (27 ). Given that the frequency of pt mtDNAs increased within a population of cells, and that there was no discernible difference in growth rate between cells with wild-type, pd or pt mtDNA under our standard growth conditions (data not shown), intracellular selection is the most plausible mechanism of fixation. Our results are consistent, therefore, with the idea that pt molecules have areplicative advantage over pd mtDNA.
A variant in any genome may be fixed simply because it is associated with an advantageous (driver) mutation, even if it is, of itself, mildly detrimental; the phenomenon of `hitch-hiking' (28 ). The clear candidate for `driver' within the rearranged mtDNA molecules is the heavy strand origin of replication. Wild-type mtDNA molecules contain a single heavy strand origin, the pd and pt molecules contain two and three heavy strand origins respectively. Multiple origins are well recognised as the molecular basis of segregation bias (so-called hypersuppressiveness) in mtDNA of yeast (29 ,30 ). Recently, Kuzminov (31 ) highlighted the similarities between segregation and maintenance of plasmids in Escherichia coli, and mtDNAs of yeast and human cells. His postulated common diffusible inhibitor of replication suggests one way in which a mutant episome containing additional origins of replication (as seen in the pd and pt mtDNA molecules reported here) might establish itself in a cell.
Previous work by others (32 ) showed that pyrimidine dimers in mtDNA were not removed from one murine or two human cell lines, suggesting that mammalian mitochondria lacked a nucleotide excision repair system. However, there is good evidence that mammalian mitochondria contain base excision repair activity (33 -35 ) and other DNA repair pathways (36 ,37 ). More recently, it has been shown, using an in vitro assay, that mammalian mitochondria contain homologous DNA recombination activity (38 ). The systematic loss of pd mtDNAs in lung carcinoma cells, reported here, provides further evidence that human mitochondria contain a recombination system and suggests a role for this system in maintaining the integrity of the human mitochondrial genome.
The maintenance of pd mtDNAs, by contrast, in osteosarcoma cells may indicate either that particular human cell types naturally lack an efficient mitochondrial recombination system or that the system can be inactivated or lost in aneuploid cells in culture. Alternatively, recombination may be inefficient in both cell types with a strong segregation bias against rearranged molecules in the lung carcinoma nuclear background. A third possibility is that an undetectable amount of wild-type molecules in the donor mitochondria could have accounted for the appearance of wild-type mtDNAs in the lung carcinoma cybrids, reflecting an inability to replicate or transmit rearranged mtDNA molecules in this particular nuclear background. However, the persistence of pd molecules in some lung carcinoma cybrid clones indicates their transmission in this background is not completely prohibited, and makes `heteroplasmy at an undetectable level' an unlikely explanation for the reappearance of wild-type moleules upon serial cybridization.
One possible candidate for involvement, based on studies in yeast, is a cruciform junction resolvase: MTG1 (39 ). In the absence of MTG1, [rho]- mtDNA forms aggregates to a greater degree than [rho]+ (wild-type) mtDNA; this substantially decreases the number of segregating units and results in the selection of [rho]+ mtDNAs, and loss of [rho]- mtDNAs.
Interestingly, the pattern of transmission of rearranged mtDNAs in the family studied is consistent with that of a late onset autosomal dominant disorder, i.e., that it is caused by a nuclear gene mutation. The nuclear gene mutation might be an allelic variant of a component of the recombination, segregation or replication machinery that did not function properly in combination with a particular mtDNA haplotype. This would result in a complex inheritance pattern with both maternal and Mendelian contributions.
Mitochondrial DNA depletion is a well recognised cause of disease (40 ) and a progressive decrease in mtDNA copy number has been observed in the cultured fibroblasts of a patient with mitochondrial dysfunction (41 ). Mitochondrial DNA depletion has not hitherto been described in association with a presumed pathological mtDNA mutation. The mechanism of mtDNA depletion in osteosarcoma cybrids that we have observed remains unknown, nor is it clear whether this phenomenon occurs in vivo in association with the rearranged mtDNA. We have only been able to document such depletion following serial cybrid transfer, using donor organelles containing solely pd mtDNA. If mtDNA depletion does occur in vivo in association with this rearrangement it could explain why there was muscle pathology in patient M10 at a measured level of only 25% pd mtDNA, i.e., some muscle cells with a high proportion of pd mtDNA may undergo mtDNA depletion at high frequency and die.
The mild impairment of aerobic metabolism associated with osteosarcoma cybrids containing >96% pd mtDNAs suggests that decreased respiratory capacity may not be the (primary) mechanism by which this mutation causes disease. Alternatively, pd mtDNA may be capable of causing significant respiratory impairment only in particular nuclear genetic or developmental backgrounds; appropriate backgrounds may be most commonly characteristic of islet cells, cochlea and skeletal muscle leading to the phenotypes observed in this pedigree.
Cytochrome c oxidase activity and in vitro mitochondrial translation (data not shown) were not significantly impaired in osteosarcoma cybrids with high levels of pd mtDNA molecules. These results are in striking contrast to those obtained for partially deleted mtDNAs, where the experiments of Hayashi and colleagues (6 ) demonstrated a dramatic decrease in COX activity and mitochondrial translation in HeLa cybrids with 78% partially deleted mtDNAs. The contrasts between the findings of Hayashi et al. (6 ) and those of this study suggest that the respiratory phenotypic consequences of the two types of rearrangement may be systematically different. The apparent paradox of how the two types of rearrangement (deletion and duplication) are associated with rather similarclinical conditions could be reconciled if both types of mutation predisposed particular cells to undergo mtDNA depletion or cell death.
The difficulties we have encountered in detecting a respiratory phenotype in cybrids harbouring a partial mtDNA duplication suggest that a mtDNA mutation may be associated with a disease phenotype without necessarily producing a measurable decrease in respiratory capacity, at least in cybrids of a particular type. Neither is analysis of patient tissue wholly satisfactory as biochemical deficiencies can be masked by co-existing wild-type mtDNA (8 ). It is necessary, therefore, to exercise caution in excluding mtDNA involvement in disease states.
A patient (M10) with a population of pd mtDNA, presented with diabetes mellitus, ophthalmoplegia and proximal muscle weakness (11 ). M10's blood contained 80% pd mtDNA and his muscle 25% pd mtDNA. M10's diabetic mother had ~40% mtDNA in blood carrying the same rearrangement (11 ). More recently, we have learned that M10 suffers high frequency hearing loss and his mother had severe loss of hearing across all frequencies. A sister who is also diabetic, severely deaf and suffers proximal muscle weakness, was screened and found to have 60% pd mtDNA in her blood. Like her brother, there was an absence of ragged red fibres (RRF) in her skeletal muscle. The sister's three children had no detectable pd mtDNA in their blood, either by Southern blotting or PCR (data not shown), and all are currently without symptoms. M10's children also lacked detectable rearranged mtDNAs.
Myoblasts from patient M10, containing a mixture of pd and wild-type sized mtDNA molecules, were grown initially in Dulbecco's Modified Eagle's Medium (DMEM) containing 110 mg/l pyruvate, supplemented with 30% fetal bovine serum (FBS) and chick embryo extract (CEE). After two passages CEE was omitted and the serum reduced first to 20% and subsequently to 10%. The osteosarcoma 143B TK- cells and 143B [rho]o-derived cybrids were grown in 90% DMEM, 10% FBS and bromodeoxyuridine (100 [mu]g/ml). The [rho]o cells derived from the osteosarcoma 143B cell line (143B.206), were, in addition, supplemented with 50 [mu]g/ml uridine. A549 lung carcinoma cells were maintained in 90% DMEM, 10% FBS. The [rho]o cells derived from the lung carcinoma cell line (A549.B2), were, in addition, supplemented with 50 [mu]g/ml uridine. The absence of mtDNA from both these cell lines has been shown previously by Southern blotting and PCR (3 ,17 ). 143B.206.neo and A549.B2.neo cell lines were generated by transfecting pcDNA1neo (Invitrogen) into the 143B.206 cells using Lipofectin (Life Technologies). After each transfection, a stable G418-resistant clonal cell line was selected and maintained by culturing in the presence of 250 [mu]g/ml Geneticin (Life Technologies).
Cytoplasts were generated by inverting 35 mm tissue culture plates, 50-90% confluent, in 95% DMEM, 5% FBS with 10 [mu]g/ml cytochalasin B (Sigma, Poole) and centrifuging at 7000 g for 20 min. The resultant cytoplast lawn was incubated for 3 h at 37oC, with ~8 * 105 [rho]o cells. The addition of 50% (w/v) PEG 1500, 45% DMEM, 5% DMSO-induced cell-cytoplast fusion. After 1 min, the cells were washed twice in 90% DMEM, 10% DMSO and three times in DMEM alone, and incubated overnight in 90% DMEM, 10% FBS without uridine. Putative transformant cells were replated on 90 mm dishes in 90% DMEM, 10% FBS without uridine. Individual colonies were picked ~14 days later using glass rings. Where the nuclear recipient carried a neomycin resistance gene, G418 was included in the growth medium at a concentration of 250 [mu]g/ml, for a period of ~1 month. Osteosarcoma cybrids carrying mtDNA molecules derived from patient M10 were designated 206.dup, even in those cells that did not maintain pd molecules. The equivalent lung carcinoma cybrids were denoted B2.dup. G418-resistant cybrids were designated 206.neo.dup or B2.neo.dup to indicate that their selection was dependent on expression of the neomycin resistance gene.
DNA extraction from cultured cells was as described by Laird et al. (18 ).DNA samples were digested with a variety of restriction endonucleases according to the manufacturers' instructions (Promega). Digested samples were electrophoresed at 1.3 V/cm for 16-20 h.Southern blotting and hybridization to labelled purified placental mtDNA and 18S rDNA were as described in Molecular Cloning (19 ). Amplification, cloning and sequencing of the duplication-triplication junction was as described previously for the duplication junction (11 ). Quantitation of the relative amounts of pd and wild-type mtDNA was determined using a BioRad GS 250 molecular imager. The apparently normal sized mtDNA molecules that co-existed with the duplicated molecules in the proband's tissues and subsequently in trans-mitochondrial cell lines are referred to as wild-type throughout the report, although only a portion of the large non-coding region and a few hundred nucleotides around the two sequence blocks that form the duplication junction were sequenced (data not shown). In vitro labelling of mitochondrial translation products was performed as described by Chomyn et al. (6 ). Other biochemical assays were as described previously (20 ).
The original patient material was provided by Drs D.Davidson, R.Roberts and R.Swingler. The 143B TK- cells and 143B.206 TK- cells ([rho]o) were gifts of Drs G.Attardi and M.King. Dr D.Tosh helped us establish reproducible measurements of pyruvate and lactate, and Dr B.Mc.Stay provided the 18S rDNA probe. Pamela Moonie provided valuable technical assistance. I.J.H. holds a Royal Society University Fellowship. This work was supported by the Muscular Dystrophy Group, The Medical Research Council, The EU Human Capital and Mobility Programme and Tenovus Tayside. I.J.H. would like to dedicate this paper to the memory of Anita Harding, Professor of Neurology.
Human Molecular Genetics
Pages
Introduction
Results
Generation of osteosarcoma cybrids containing extremes of pd and wild-type mtDNAs
Detection of a partially triplicated mtDNA species in osteosarcoma cybrid cells
Transfer of mitochondria containing pd mtDNA from osteosarcoma cells to lung carcinoma cells results in the loss of rearranged mtDNAs and appearance of wild-type molecules
Retransfer of pd mtDNA to osteosarcoma cells can precipitate mtDNA depletion
Biochemical analysis of osteosarcoma cybrids containing rearranged mtDNA
Discussion
Segregation of pd and wild-type mtDNAs in opposing directions
The appearance of pt mtDNA molecules in culture probably reflects the presence of an additional origin of heavy strand replication, conferring replicative advantage
Systematic loss of pd mtDNA molecules in the lung carcinoma cell background suggests that intramolecular recombination occurs in human mitochondria
Introduction of pd mtDNA into an osteosarcoma nuclear background can disturb mtDNA copy number control
Implications of the mild respiratory impairment associated with pd mtDNA
Materials And Methods
Pedigree analysis
Cell culture
DNA and biochemical analyses
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Tue Jul 15 11:13:42 BST 1997. Part of the OUP Journals World Wide Web service. Copyright Oxford University Press, 1996