Mitochondrial DNA in Mortal and Immortal Human Cells

Mitochondrial DNA was quantitated in total DNA of various normal and mutant strains of human diploid fibroblasts (finite replicative lifespan) and permanent cell lines, using Southern-transfer hybridization to 32P-labeled pure mtDNA probe and saturation hybrid- ization to ‘H-labeled cRNA copied from mtDNA. In six normal fibroblast strains, mtDNA copy number in- creased during serial passage roughly in proportion to cell volume or protein content, whereas normalized mtDNA content per pg of protein depended upon in uiuo donor age but not passage level (“in uitro” age). Copy numbers for mtDNA varied much more widely in individual fibroblast clones than in mass cultures, but were not well correlated with longevity or growth rate. Five mutant fibroblast strains associated with reduced replicative lifespan, and four permanent cell lines, were also examined; in each group, mtDNA values were observed both lower and higher than any obtained for normal fibroblasts. No evidence was found of petite-type deletions from human mtDNA, either at late passage or in individual clones of fibroblasts.

Mitochondria, the cellular organelles responsible for oxidative energy production, possess a unique combination of autonomy and interdependence with the cell nucleus. Human mitochondria contain multiple circular copies of their own DNA (mtDNA), which encodes the mitochondrial ribosomal and transfer RNAs, plus about 13 protein chains. Nuclear genes, however, direct the synthesis of another 300-400 mitochondrial proteins (Anderson et al., 1981;Tzagoloff, 1982). It is not known to what extent or in what manner replicative synchrony is maintained between mitochondrial and nuclear genomes, although several levels of interaction are possible (Storrie and Attardi, 1972;Clayton, 1982).
In principle, since a balanced collaboration between nuclear and mitochondrial DNA is required for synthesis of mitochondrial ribosomes and three multimeric proteins, mitochondrial function could be impaired by any gross imbalance in the copy number per cell of mitochondrial DNA molecules or portions thereof. Nuclear DNA replication, on the other hand, may be modulated through mitochondrial synthesis of ATP. Clearly, a deficiency of intact mtDNA genomes can limit the * This research was supported by grants from the National Institutes of Health (AG-03314 and AG-03787) and the Veterans Administration. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. rate of cell proliferation, as in "petite" mutations of yeast (Slonimski and Lazowska, 1977), and it has been suggested that imbalance in either direction could profoundly alter cellular growth parameters (Belcour and Begel, 1978;Shmookler Reis et al., 1980;Hartung, 1982).
We wished to investigate the relationship between mitochondrial DNA and both the growth rate and growth potential of cultured human cells. We have prepared total DNA from a variety of sources, including several normal diploid fibroblast strains at early and late passage. These cells have a limited growth potential or proliferative lifespan which is inversely related to the age of the tissue donor (Hayflick, 1977;. Exhaustion of that potential, commonly described as in uitro aging, is accompanied by a progressive slowing of the cell cycle and hence of growth rate (Cristofalo, 1976;Macieira-Coelho, 1977). We have also examined mutant fibroblast strains, from individuals with premature aging and/ or chromosomal breakage syndromes, and several "immortal" human cell lines (i.e. with unlimited replicative lifespans). These DNAs were then probed for mtDNA sequences in Southern-blot hybridizations (Southern, 1975) and in saturation hybridizations (Gillespie and Spiegelman, 1965), allowing determination of the number of mtDNA molecules per nuclear equivalent.
Alternative methods of quantitation based on prior isolation of mitochondria (Bogenhagen and Clayton, 1974) and/or mtDNA (Smith et al., 1971) were excluded as liable to underestimation, due to inevitable losses during preparation or analysis. Such losses could vary systematically with in uitro aging, producing artifactual differences, since fragmented or fragile mitochondria might be preferentially lost during purification, whereas endonuclease nicking of DNA (in uiuo or during mtDNA preparation) would decrease the yield of mtDNA identified as covalently closed circular DNA molecules.
We here report data on copy numbers, methylation, and integrity of mitochondrial DNA in normal and mutant fibroblast strains and in permanent aneuploid cell lines, and we attempt to correlate these data with other cellular parameters including short term growth rate, long term growth potential, and age of donor. Methods DNA Purification-Cells were rinsed 3X with phosphate-buffered saline and lysed in situ on Petri dishes upon addition of SDS' to 0.5% plus EDTA at 20 mM and Tris-HC1 at 60 mM, pH 8.3. Proteinase K was added at 100 pg/ml and SDS increased to 1% (w/v); samples were then incubated 4 h at 37 "C before and after addition of a further 100 pg/ml of Proteinase K. DNA was extracted twice with ultrapure phenol (Bethesda Research Laboratories), freshly saturated with 0.3 M Tris-HC1, pH 8.3,20 mM EDTA, 0.18 M NaCl; it was then extracted twice with ch1oroform:octanol (201, v/v). After precipitation and rinsing in 70% ethanol, 0.1 M NaOAc, pH 6, the DNA pellets were redissolved and digested with heat-treated RNase A (100 pg/ml, 1 h at 37 "C) followed by heat-treated pronase (250 pg/ml, twice for 4 h each at 37 "C). Extractions of DNA were repeated as above and DNA was ethanol-precipitated, rinsed twice in 70% ethanol, and redissolved in Tris/EDTA buffer (10 mM Tris-HC1, 1 mM EDTA, pH 8.3).
Saturation Hybridization with cRNA-Complementary [3H]cRNA was transcribed from HeLa mtDNA by Escherichia coli RNA polymerase in the presence of [3H]GTP and purified as previously described (Shmookler Reis and Biro, 1978). The cRNA product, specific activity 1.8 x lo6 cpmlpg, was hybridized at 16-to 100-fold excess against filter-immobilized DNA.
Total cellular ["CIDNA (or unlabeled cellular DNA brought to 2000 cpmlpg by addition of normal fibroblast ["CIDNA at 48,000 cpm/pg) was denatured by boiling 10 min in 10 mM Tris-HC1, 1 mM EDTA, pH 8.3. It was then diluted rapidly into 6 X SSC at 0 "C, bound to nitrocellulose filters (Millipore HAWP, prewashed in 2 X SSC at 95 "C and 6 X SSC at 20 "C) at 0.3 pg of DNA per filter, and rinsed, dried, and baked in uacw as described by Gillespie and Spiegelman (1965). Hybridization was for 18 h at 60 'C, in 3 X SSC containing 0.2% SDS and 65 pg/ml of E. coli tRNA. Filters were washed 3 X 15 min in 4 X SSC at 60 "C, then incubated for 30 min at 20 "C with 20 pg/ml of RNase A in 2 x SSC and rinsed 2 X 10 min in 2 X SSC at 20 "C. Filters were air-dried and counted in toluene plus 0.3% (w/v) diphenyloxazole. In some experiments, filters were dried and counted prior to RNase treatment, then rinsed 3 X in toluene, dried, and rewet with water for RNase digestion, etc., as above.
Determination of Cell Vofumes-Cell suspensions were obtained at the time of confluence, immediately following harvest with 0.125% trypsin, and were analyzed for median cell volume on a Coulter Counter (Model ZF and Channelyzer) calibrated with polystyrene microsphere standards.
The abbreviations used are: SDS, sodium dodecyl sulfate; mtDNA, mitochondrial DNA; MPD, mean population doublings; bp, base pairs; kb, kilobase pairs. Protein Assay-Total cell protein was determined on cell suspensions (obtained and counted as in the preceding section) by the method of Lowry et al. (1951), using bovine serum albumin standards.

RESULTS AND DISCUSSION
mtDNA in Early versus Late Passage Cells from Young and Old Donors-Total DNA was prepared from six normal fibroblast strains at early and late levels of mean population doubling with 97 f 2% (S.E.) yield (Shmookler Reis and  and digested with HpaII, MspI, or EcoRI restriction endonucleases, electrophoresed in agarose gels, and transferred to nitrocellulose filter sheets. The filters were then hybridized to pure mtDNA probe, prepared as covalently closed circular DNA molecules from isolated HeLa cell mitochondria and labeled with [32P]dCTP by nick translation. The resulting autoradiographs (Fig. 1) were then used to evaluate copy number for the mitochondrial genome and fragments thereof, by comparison to hybridization standards (lanes at left of Fig. 1, containing serial dilutions of pure HeLa mtDNA digested with HpaII). Hybridization to restriction fragments in these mtDNA standards is proportional to fragment length, confirming that the probe is essentially uniformly labeled. (Apparent exceptions, at 2.2 kb and 0.5 kb, are due to comigration of two or more fragments of similar length (Brown and Goodman, 1979).) The HpaII pattern obtained in most samples ( Fig. 1) corresponds to the principal form of human mtDNA, which is polymorphic for cleavage sites of numerous restriction enzymes (Brown and Goodman, 1979;Brown, 1980). The variant pattern observed in two of our cell strains ( Fig. 1, i andj, and Fig. Be) is identical with Brown's "Morph 2" seen in 1 out of 21 individuals previously surveyed (Brown, 1980).
The results of several such experiments are summarized in Table I. Values are expressed as "% mtDNA," obtained by dividing the quantity of mtDNA in each channel (determined autoradiographically) by the amount of total cell DNA loaded. Several of these determinations have been confirmed by saturation hybridization (see below). Since the human diploid genome contains -6.5 X lo9 base pairs, while human mtDNA contains 16,569 bp, each 0.1% mtDNA corresponds to approximately 400 copies of the mitochondrial genome per cell. In each of the "young donor" cell strains, mtDNA increased during serial passage by 30-50%. At the same time, however, the cell volume and protein per cell also increased at late passage (Table I) confirming a well established concomitant of in vitro aging (Simons, 1967;Cristofalo and Kritchevsky, 1969;Haslam and Goldstein, 1974;Mitsui and Schneider, 1976). In strain A2 for example, while mtDNA/cell increased by -50% at late passage, cell protein content increased by 34% and cell volume increased by 44% (Table I). Thus, although late passage cells generally had gained mtDNA copies per cell, mtDNA expressed per unit volume of cytoplasm or per mass of protein remained relatively constant during serial culture.
The fraction of mtDNA in cell strains derived from three old donors was initially higher than in three young donor strains (1.2% k 0.06 (S.E.) versus 0.73% f 0.09, respectively; p < 0.02), but did not increase as much at late passage. Indeed, when mtDNA is considered in relation to cell protein or cell volume serviced by the mitochondria, the difference between old and young donors at early passage is even more pronounced (Table I, right-hand column). In early passage cells, old donors averaged 22.1 k 2.0 (S.E.) copies of mtDNA per pg of protein, compared to 10.6 f 0.7 for young donors ( p < 0.005). Thus, fibroblasts from old donors, which tend to have shorter replicative lifespans in vitro (Goldstein et al., 1969;Martin et al., 1970;Hayflick, 1977;, may be FIG. 1. Mitochondrial DNA in six strains of human diploid fibroblasts from normal donors, at early and late passage in vitro. Total cellular DNA samples were digested with HpaII restriction enzyme (12 units per pg of DNA, 2 h at 37 "C), electrophoresed in 1.5% agarose gel, Southern-transferred, hybridized on nitrocellulose filters against [32P]mtDNA probe, and autoradiographed. Lams at the left are hybridization standards: serial dilutions containing 25 ng, 12 ng, 6 ng, and 3 ng of HpaII-cleaved HeLa cell mtDNA. Lanes a-1 correspond to entries in Table I distinguished from young donors' fibroblasts even at early passage. This implies that the mtDNA/protein ratio reflects an age-dependent condition of the donors' cells in vivo which persists through in vitro cultivation.

Mitochondrial DNA Copy Number in Permanent Cell Lines and Mutant Cell
Strains-In view of the limited variation in mtDNA content observed in normal diploid cell strains, we wished to extend our survey to cells with grossly abnormal growth characteristics. We first examined mtDNA in several permanent human cell lines (Fig. 2, lanes a-d, and Table 11). The value obtained for HeLa cells (7200 mtDNA genomes per cell) was in reasonably good agreement with an earlier estimate of 8800 copies per cell, based on [3H]thymidine incorporation into mtDNA of HeLa TK-cells (Bogenhagen and Clayton, 1974). The range of copy number values obtained in these lines, considered either per cell or per mass of protein, was greater than had been found for diploid cell strains throughout their lifespans (Table  11; compare to Table I), suggesting that mtDNA levels cannot be important in the escape from mortality of permanent cell lines. It should also be noted that all four permanent lines grew a t a similar rate (doubling time -24 h), somewhat faster than even the most vigorous diploid fibroblasts (doubling time 2 28 h), implying that mtDNA levels per se do not govern the growth rate of these cells.
Genetically determined syndromes which resemble premature senescence have been found to give rise to cultured fibroblasts with a marked reduction in growth rate and replicative lifespan (Martin et al., 1970;. We have examined mtDNA in fibroblasts from three donors with the Hutchinson-Gilford progeria syndrome and one donor with Werner syndrome (Fig. 2, lanes f-j). Bloom syndrome fibroblasts, characterized by abnormally high levels of DNA breakage and sister-chromatid exchange , were also assessed (Fig. 2, lane e ) . Both Werner syndrome (Salk et al., 1981) and Bloom syndrome  are associated with an unusually high incidence of malignancy, as well as elevated frequency of chromosomal aberrations. There were no consistent changes in mtDNA correlated with either abbreviated lifespan (Table 111, WS2, P5, P11, and P18) or chromosomal instability (Table 111, BL and WS2). Three progeria strains and a Werner syndrome strain were low to average in mtDNA copy number, for their doubling level, compared to normal strains (Tables  I and 111). Cell volumes and protein contents per cell, however, tended to be higher in early passage progeria fibroblasts than in normal strains, perhaps reflecting their slow growth; after adjustment for this their mtDNA concentrations were rather low, in contrast to the high mtDNA concentrations associated with aged donors. Since both immortal cell lines and "short-lived" mutant strains can carry a complement of mtDNA either greater or less than that found in normal diploid fibroblasts at any age, we conclude that the relatively moderate variation in mtDNA copy number seen during the replicative lifespan of diploid fibroblasts is not responsible for the limit on that lifespan.
Clonal Heterogeneity of mtDNA Levels-When seven individual clones isolated from the A2 mass culture were examined for mtDNA sequences (Fig. 3 and Table IV), values were found ranging from 0.4% to 1.3% of the genome (1600-5200 copies/cell). The level of mtDNA/cell in these clones was not correlated with either cell volume or maximal replicative lifespan, but showed a weak inverse correlation with growth rate ( r = -0.65) (Table IV). These data support the conclusion that mtDNA number is not important in determining cellular lifespan in vitro. Indeed, if there were any advantage in  possessing either high or low mtDNA levels, then given the substantial clonal heterogeneity within a mass culture the population should have shifted toward the mtDNA levels found in "fitter" clones. Since the mass cultures changed little in mtDNA proportion, compared to the heterogeneity they comprise, such selection must not occur significantly. The moderate inverse correlation between mtDNA number and cell growth rate, however, suggests that the replication rate of mtDNA genomes may lag slightly behind nuclear replication in the fastest growing lineages. Tables I-IV were obtained from densitometer tracings of Southerntransfer autoradiographs (Figs. 1-3) within the linear range of the films. Several such experiments, with appropriate hybridization standards, were used in each determination. However, in addition, many of these results were confirmed by saturation hybridization (Gillespie and Spiegelman, 1965;Shmookler Reis and Goldstein, 1980). 3H-labeled complementary RNA, copied from pure mtDNA template, was hybridized in increasing excess to total cell [14C]DNA immobilized on small nitrocellulose filters.

Saturation Hybridization-The mtDNA levels in
The plateau level of maximal hybridization, approached at high cRNA concentration (ranging from 16-100-fold excess over complementary mtDNA strands), can be most accurately determined by linear extrapolation in a double reciprocal plot of the data (Bishop, 1972).
Representative results from such saturation hybridizations are shown in Fig. 4 a Further details, including maximum MPD, and references for these strains can be found in Shmookler Reis and  Passage level in mean population doublings.
Young versus old donor difference is significant a t p < 0.02. Young versus old donor difference is significant a t p < 0.005.    late passage (17 and 53 MPD) were analyzed in Fig. 4, a and b, confirming a moderate ("50%) increase in mtDNA content per cell a t late passage (see Table I, c and d ) . A heterologous DNA control and a mtDNA standard are also shown. In contrast, comparison of the HeLa and Chang cell lines (Fig.  4, c and d ) indicated a 4-fold difference in their mtDNA levels per cell, in close agreement with the results of Southerntransfer quantitations (Table 11).
Are There Petite-type Deletion Mutations in Human mtDNA?-Senescence of the fungus, Podospora anserina, is associated with the random amplification of specific segments of mtDNA at the expense of integral mtDNA genomes (Jamet-Vierny et al., 1980), analogous to p-("petite") deletion mutations of yeast mtDNA (Slonimski and Lazowska, 1977). In view of this, and the observation by Smith and Vinograd (1972) of small polydisperse circular DNA molecules in HeLa and WI-38 cells, we undertook Southern-transfer hybridizations (Figs. 1-3) in a manner which would favor identification of amplified regions in mtDNA. These experiments revealed no departures from stoichiometry of mtDNA restriction fragments in early or late passage mass cultures, in isolated clones, or in permanent cell lines or mutant cell strains (Figs. 1-3). Apparent departures from molar equivalence of fragments were due to co-migration of two or more restriction fragments (Brown and Goodman, 1979, and see above).
In order to ascertain whether any mtDNA sequences could be identified a t sizes smaller than the mitochondrial genome, we electrophoresed undigested total DNA from the DS fibroblast strain a t several passage levels and probed the resulting Southern-transfers with [3"P]mtDNA under conditions of very high stringency. In addition to intact monomeric and dimeric mtDNA molecules, we observed a number of smaller "extrachromosomal" bands hybridizing to [32P]mtDNA probe (Fig. 5). The most prominent of these was approximatdy 0.65 kb in length and probably corresponds to the principal species of "D-loop strand" (putative primer for mtDNA H-strand synthesis) released from nicked mtDNA circles (Robberson and Clayton, 1973;Ojala and Attardi, 1978;Clayton, 1982). Hybridization to this band indicated 1000-2000 copies per cell (perhaps more if it is imperfectly homologous to the mtDNA probe or under-represented there), thus comprising a substantial fraction of the mtDNA copy number in DS fibroblasts, 3000-4000 per cell. While there was only a moderate decrease in the amount of 0.65-kb band hybridization during in vitro aging, it was considerably diminished in senescent fibroblasts maintained for 5 weeks after the cessation of mitosis (Fig. 5e). In contrast, mtDNA hybridization to a 2.3kb band from undigested DNA did not change substantially (Fig. 5) and mtDNA genome number per cell remained essentially constant (data not shown) in those postmitotic senescent cells. The preferential loss of the 0.65-kb band, in cells which have essentially ceased dividing and which maintain a stable level of mtDNA, thus suggests that D-loops may play a role not only in regulation of transcription (Clayton, 1982), but also in nuclear control of mtDNA replication.
Although the significance of these small mtDNA molecular species is still speculative, they are clearly not analogous to the mtDNA fragments amplified in senescing Podospora cultures. Extrachromosomal circular DNA molecules have been observed recently in nuclei of monkey and human cells (Krolewski et al., 1982;Calabretta et al., 1982;Shmookler Reis et al., 1983a) but these are not homologous to mtDNA (Shmookler Reis et al., 1983b). Thus, the observation of small DNA molecules hybridizing to mtDNA is of particular interest. They may form part of the small circular DNA characterized by Smith and Vinograd (1972) which appeared to be cytoplasmic although not localized to mitochondria.
Methylation of mtDNA-Previous estimates have indicated very low levels of methylation (not exceeding 0.2-0.6% of cytosines) in mtDNA from mouse and hamster cells (Nass, 1973). We have used restriction cleavage with the enzymes MspI and HpaII to estimate the extent of methylation in human mtDNA (Waalwijk and Flavell, 1978;Singer et al., 1979;. These endonucleases share a common recognition sequence, -C-C-G-G-, but differ in their responses to 5-methylcytosine ( [3H]RNA counts/min hybridized prior to RNase digestion (but after extensive rinsing) were approximately 4X higher than values after RNase, butwere otherwise consistent with these data.
In many experiments repeated with the same DNA samples, the patterns and quantities of mtDNA hybridization measured after MspI cleavage were very close to those measured after HpaII cleavage. HpaII digestions, however, generally left 2-5% of hybridizing mtDNA at and near the top of the gel in positions corresponding to undigested mtDNA (see Figs. 1 and 3) indicating molecules with all -C-C-G-G-sites modified to -C-"C-G-G-; such material was not seen after digestion with MspI (Fig. 2) or EcoRI (not shown) and occurred at much lower levels in HeLa mtDNA, isolated as covalently closed circular DNA molecules, following HpaII digestion ( Fig. 1). One fibroblast clone (Fig. 3, lane i) had -10% of its mtDNA at high molecular weight after HpaII digestion and interestingly this was a clone shown previously to have unusually high -CpG-methylation of total nuclear DNA (see Shmookler Reis and , "clone 30").
The uncleaved mtDNA was unaffected by higher ratios of HpaII enzyme to DNA (not shown), but was less abundant in late passage DNA samples than at early passage for each of the fibroblast strains from young donors (Fig. 1, lanes a-f). This relationship was not apparent, however, in fibroblast strains from old donors (Fig. 1, lanes g-l).
If -C-C-G-G-sites are typical of all CpGs in mtDNA, then 2-5% of sites methylated would correspond to 0.06-0.14% of cytosines methylated, far below the fraction observed in total DNA from human fibroblasts (2.2-4.5%: Shmookler Reis and , but in good agreement with earlier estimates for mammalian mtDNA (Nass, 1973). Our calculation of mitochondrial cytosine methylation is based on where "CpG/CpG = 2-5% (this report), CpG/C = 2.5% (Anderson et al., 1981), and TpG/"C = 3.9 (Ehrlich and Wang, 1981). Since each mtDNA molecule contains 22 potential sites for HpaIIIMspI (Brown, 1980), the appearance of a substan- FIG. 5. Hybridization of mtDNA probe to submonomer length mitochondrial DNA. DNA samples were electrophoresed in 1.5% agarose gels following HpaII digestion or without digestion by restriction enzymes and were probed for mtDNA sequences as described in the legend to Fig. 1. a, DNA from DS fibroblasts, 5 weeks after mitotic arrest (MPD 58) due to senescence plus confluence: 0.2 pg of DNA digested 2 h at 37 "C with 3 units of HpaII; b, 2 pg of DNA from DS fibroblasts at 17 MPD, undigested; c, 2 pg of DNA from DS fibroblasts at 36 MPD, undigested; d, 2 pg of DNA from DS fibroblasts at 53 MPD, undigested; e, 1.2 pg of DNA from DS fibroblasts at 58 MPD plus 5 weeks postmitotic (as in a), undigested. tial fraction (2-5%) of totally uncleaved mtDNA molecules, with no apparent partially cleaved/partially methylated forms, indicates a highly nonrandom distribution of methylations. Such a distribution might arise, for example, if nuclear methylases "escape" into the cytoplasm and enter a fraction of mitochondria, in which they fully methylate appropriate sites in the mtDNA. We have no indication, however, of how a vulnerable fraction of mitochondria might be specified, or of why in uitro senescence should alter that fraction.

CONCLUSIONS
We have quantitated mitochondrial DNA in normal human diploid fibroblasts and obtained values ranging from 2400-6000 copies/cell, entirely within the range (1100-8800 copies/ cell) spanned by permanent aneuploid cell lines analyzed previously, human HeLa and mouse L cells (Bogenhagen and Clayton, 1974), and concurrently (this report, Table 11). The average number of mtDNA genomes per cell varied relatively little during the replicative lifespan of any given diploid cell strain (Table I), increasing in 5 of 6 strains by 10-50% at late passage. When normalized to cell protein content, which also increased by 30-80% in those five strains which gained mtDNA, the concentration of mtDNA genomes remained remarkably constant over the period of in uitro culture. A number of conclusions may be drawn from these results. Firstly, the data begin to answer the question of how closely mtDNA replication is kept in synchrony with nuclear DNA replication: it would appear to be regulated not by direct coupling to the nuclear DNA replication, but rather by the cell mass to be serviced by mitochondria.
Secondly, mitochondrial genome number is unlikely to be a major determinant of cellular senescence in uitro, in view of the moderate alterations observed in mtDNA number during the replicative lifespan, relative to interindividual and interclonal variation, or to values seen in immortal lines. Indeed, when genome numbers per cell are adjusted for changes in cell protein or cell volume, as discussed above, virtually all passage-dependent change in mtDNA copy number is eliminated. Observations by high voltage electron microscopy* have indicated a similar constancy during in uitro culture, in the ratio of mitochondria/cytoplasm (estimated from projected mitochondrial area per cell area). There is, however, a latepassage increase in glycolysis  and a decrease in protonmotive capacity (Goldstein and Korczack, 1981). In the light of the present data, these changes are unlikely to result from deficiencies or deletions in mtDNA, but instead may reflect impairments in mitochondrial structure (Lipetz and Cristofalo, 1972;Johnson, 1979), mitochondrial gene expression, or metabolic regulation.
Fibroblasts from old donors displayed significantly higher levels of mtDNAlcel1 protein or mtDNAlcel1 volume than cells from young donors, regardless of passage level in uitro. This suggests that a molecular correlate of in uiuo senescence may be maintained throughout fibroblast isolation and propagation in uitro. Further studies on additional strains from young and old subjects will be necessary to resolve this question.
The integrity of the mitochondrial genome does not appear to alter during in uitro senescence, since restriction fragments maintain the same relative proportions and mobilities. This is consistent with our earlier report of an essentially invariant pattern of mitochondrial protein synthesis for 24 human cell lines and strains (Yatscoff et al., 1978). In addition to chromosome-size mtDNA (16,569-bp monomeric form), several discrete species of small mtDNA molecules were seen in similar amounts at all passage levels prior to mitotic arrest of the culture. The disappearance of the 0.65-kb DNA species, coincident with prolonged cessation of mitosis (due to saturation arrest plus senescence), suggests its involvement in mtDNA replication. Unlike "petite" mutations in yeast (Slonimski and Lazowska, 1977) and "SEN-DNA" of fungi (Jamet-Vierny et al., 1980), these small mtDNA molecules do not appear to proliferate in lieu of intact mtDNA genomes, nor do they ultimately become the principle form of mtDNA.
Approximately 2-5% of mtDNA molecules are essentially fully methylated at -CCGG-sites, while the remainder are fully unmethylated. The methylated fraction appears to be dependent on fibroblast doubling level in uitro provided that the fibroblasts were derived from young normal donors. It is unlikely that such mtDNA methylation would have any functional significance in the control of mitochondrial gene expression (see, for example, Razin and Riggs, 1980), since entire genomes were methylated uniformly.
In summary, the evidence indicates that mtDNA is not a major determinant of either short term growth rate or long term proliferative potential of human diploid fibroblasts, since neither quantity, integrity, nor methylation pattern of mtDNA is substantially altered during serial passage of mass * S. Goldstein, E. J. Moerman, and K. R. Porter, manuscript in preparation.

Mitochondrial DNA in
Human Cells 9085 cultures in oitro. Substantial variation is seen, however, in individual fibroblast clones and in permanent cell lines, ranging from -1000--8000 mtDNA copies per cell, indicating that mtDNA genome number is not tightly regulated in cultured human cells.