The number of mitochondrial deoxyribonucleic acid genomes in mouse L and human HeLa cells. Quantitative isolation of mitochondrial deoxyribonucleic acid.

Abstract Mitochondrial DNA can be isotopically labeled to the virtual exclusion of nuclear DNA labeling in cell lines lacking the major soluble thymidine kinase (EC 2.7.1.21) but retaining a mitochondrial thymidine kinase. This characteristic provides a means for the selective assay of mitochondrial DNA during isolation and a determination of the cellular content of mitochondrial DNA. The simplest of the three isolation procedures compared in this study is shown to yield approximately two-thirds of the total mitochondrial DNA in the tissue culture cells examined. Mouse L cells containing predominantly a 107 dalton closed circular mitochondrial DNA species have 1100 ± 250 molecules per cell. A second line of mouse L cells which has mitochondrial DNA of molecular weight 2 x 107 contains approximately 900 molecules per cell. HeLa cells have at least four times the mitochondrial DNA mass per mitochondrial volume as L cells.

DNA can be isotopically labeled to the virtual exclusion of nuclear DNA labeling in cell lines lacking the major soluble thymidine kinase (EC 2.7.1.21) but retaining a mitochondrial thymidine kinase. This characteristic provides a means for the selective assay of mitochondrial DNA during isolation and a determination of the cellular content of mitochondrial DNA. The simplest of the three isolation procedures compared in this study is shown to yield approximately two-thirds of the total mitochondrial DNA in the tissue culture cells examined.
Mouse L cells containing predominantly a 10' dalton closed circular mitochondrial DNA species have 1100 f 250 molecules per cell. A second line of mouse L cells which has mitochondrial DNA of molecular weight 2 X 10' contains approximately 900 molecules per cell. HeLa cells have at least four times the mitochondrial DNA mass per mitochondrial volume as L cells.
Closed circular animal mtDNA possesses a number of distinctive characteristics which have stimulated a wide range of studies (l-3).
The majority of such studies of mtDNA structure and function have examined properties of that fraction of mtDNA which can be purified from isolated mitochondria. Therefore, the proper interpretation of much of this work relies upon knowledge of the quantity of mtDNA in a cell and the efficiency with which it can be isolated.
Reported attempts at mtDNA quantitation have related mtDNA content to the number of mitochondria or the mass of mitochondrial protein isolated. The studies reviewed by Nass (3) and by Borst and Kroon (4) indicate a yield of 0.2 to 1.8 pg of mtDNA per mg of mitochondrial protein, or a content of roughly 2 to 10 mtDNA molecules per organelle, depending upon the animal cell or tissue examined. However, in all of these studies it was impossible to selectively assay mtDNA during the often lengthy purification from a * This research was supported by Grants NP-9A from the American Cancer Society and CA-12312 from the National Cancer Institute.
$ Senior Dernham Fellow in Oncology (D-203) of the American Cancer Society, California Division. several hundred-fold excess of nuclear DNA. We report here an analysis of the isolation of mtDNA from animal cells in which mtDNA is isotopically labeled to a much higher specific activity than nuclear DNA. This analysis has enabled us to estimate both the total cellular mtDNA content and the yields of several types of mtDNA purification procedures.
The L and HeLa thymidine kinase minus cell lines used in this study were derived by growth in increasing quantities of RrdUrd] (5), a procedure which selects for cells lacking the ability to incorporate exogenous thymidine into nuclear DNA due to the loss of the major cellular thymidine kinase (EC 2.7.1.21). Such cell lines retain a mitochondrial-specific activity (6, 7) which allows mtDNA labeling with exogenous radioactive thymidine. Two L cell lines were compared in this study. The LMTK-cell line contains predominantly monomer length (approximately 5 pm) mtDNA molecules, while the mtDNA from LDTK-cells is almost exclusively in the form of dimer length molecules (8). The radioactive labeling technique has also been used to quantitate the mtDNA content of HeLaTKcells. The mitochondria were collected from the 1.0 to 1.5 M sucrose interface by pipetting, diluted with 2 volumes of mannitol-sucrose buffer, and pelleted by centrifugation as above. ~)m,ti ,'?ten" Procedure-Cells were homogenized in MnRSB as above.
One-sixth volume of 2.0 M sucrose, 35 mM EDTA, 50 mM Tris, pH 7.5 (instead of concentrated mannitol-sucrose buffer), was added immediately to stabilize mitochondria against osmotic rupture.
Nuclei were pelleted from the homogenate as in the "two step" procedure. The supernatant was layered over 15 ml of 1.5 M sucrose, 5 m&r EDTA, 10 mM Tris, pH 7.5. The tubes were centrifuged at 22 krpm in an SW 27 rotor for 30 min.

Mitochondria
were collected from the interface of the one step sucrose gradient and pelleted as in the "two step" procedure. "No Gradient" Procedure-The most direct mtDNA purification procedure used mitochondria isolated without a sucrose gradient step.
This procedure represents a modification of the mitochondrial isolation method of Schneider (10).
In this "no sucrose" procedure cells were homogenized and nuclei pelleted as in the "one step" procedure. Mitochondria were pelleted from the supernatant of the nuclear pellet at 15 krpm for 20 min in the Beckman JA-20 rotor. The mitochondrial pellet was resuspended in 20 ml of mannitol-sucrose buffer and mitochondria were repelleted as above.

Isolation of DNA
The final mitochondrial pellet was dispersed in 2 ml of 10 mM NaCl, 10 mM EDTA, 50 mM Tris, pH 7.5. Lysis was achieved by the addition of 50 ~1 of 25yo SDS and incubation for 3 to 5 min at 37".
Then 0.5 ml of 7 M CsCl was added before incubation of the tubes on ice for 10 min.
Chilling precipitates cesium dodecyl sulfate and some protein, which were removed by centrifugation for 10 min at 10 krpm in the JA-29 rotor. represents labeling of nuclear DNA rather than a loss of mtDNA, an aliquot of the washed nuclear pellet was lysed with SDS and centrifuged in a buoyant alkaline CsCl density gradient.
Approximately 75% of the original radioactivity was recovered in the alkaline CsCl gradient (Fig. IA).
At least 95% of the radioactivity in the nuclear fraction did not show the differential buoyant density characteristic of mtDNA in alkaline CsCl (Fig.  1B) (5, 13). This observation was reproduced in two additional preparative alkaline CsCl gradients with nuclear DNA from HeLaTK-cells and a second preparation of LMTK-nuclei (data not shown).
The nuclear labeling observed in LMTKcells resulted from an average nuclear DNA specific activity of 28 f 4 cpm per pg (Table I). Similar nuclear DNA labeling was observed in LDTK-and HeLaTK-cells.
mtDNA Isolation Procedure-One of the objectives of this study was to devise a procedure for purification of mtDNA in high yield. The following considerations are relevant to this objective.
(a) The data of Robberson and Clayton (14) indicated that some mtDNA loss was caused by a DNase treatment of isolated mitochondria or by simple incubation of mitochondria at 37" in DNase buffer without added nuclease.
(b) The discontinuation of this enzyme treatment, which had been a standard step in mtDNA purification procedures published by several groups (8,(15)(16)(17), led to an unavoidable contamination of mtDNA with nuclear DNA. (c) In the case of TK-cells, in which nuclear DNA is labeled to a low level relative to mtDNA, this significant contamination by nuclear DNA mass provided only a slight radioactive contamination in the upper band of the ethidium bromide-CsCl gradient. When this study was begun, two sucrose gradient procedures (see "Experimental Procedure") for purification of mitochondria were in use in this laboratory (6, 18). Early mtDNA labeling experiments with LMTK-cells (data not reported) revealed that the "two step" procedure isolat,ed roughly 65% of the yield of mtDNA radioactivity obtained with the "one step" procedure or 36% of that obtained with the "no gradient" procedure (Fig. 2). This presumably resulted from loss of mtDNA due to mitochondrial damage or aggregation during the first mitochondrial pelleting of the "two step" procedure.
The use of the "two step" isolation method was discontinued in further experiments. In addition, experimentation utilizing the "one step" procedure provided the following data.
The use of CaRSB is preferred because it will not stimulate activity of any Mgzf-dependent endogenous DNase during mitochondrial isolation (19). Furthermore, the level of nuclear DNA contamination of mtDNA was consistently lower when L cells were homogenized in CaRSB instead of MgRSB, as judged by the comparison of upper band fluorescence intensity in ethidium bromide-CsCl gradients.
(5) The conditions of mitochondrial lysis were thoroughly tested in order to maximize the yield of mtDNA.
Lysis at 25", instead of 37", resulted in a 15qib reduction in the yield of mtDNA.
In all cases, SDS was used at a 0.6% concentration.
Varying the duration of SDS treatment between 0.5 and 30 min had no effect in two trials.
(c) Collection of a cesium dodecyl sulfate pellet appeared to cause a slight (<lo%) loss of mtDNA during isolation from LMTK-or LDTK-cells. However, since this step removes a large amount of protein, its omission may result in extensive loss of free ethidium bromide from solution due to binding by protein, resulting in poor control over the dye concentration in the ethidium bromide-CsCl gradient.
Quantitation of mtDNA-The radioactivity in the cytoplasmic fraction of LMTK-and LDTK-cells represents mtDNA with a slight nuclear DNA contamination.
The maximum level of this 2. Radioactivity profiles of ethidium bromide-CsCl buoyant density gradients of mtDNA obtained from LMTKcells treated identically through the first mitochondrial pelleting and divided into equal parts for further purification by the "no gradient" procedure (O--O) and by the "two step" procedure (C---0).
In this experiment the yield of the "no gradient" procedure was 72y0 of total mtDNA; that of the "two step" procedure was 26% of total mtDNA. Both gradients contained approximately 680/O of the total radioactivity in seven lower band fractions.
nuclear DNA contamination has been estimated with experiments using thymidine kinase plus LA9 cells in which only 4% of the total taH]thymidine incorporation is in the cytoplasmic fraction. Then, assuming a maximum mtDNA loss of 5% in the nuclear fraction, our data show that LMTK-and LDTKincorporate 4300 and 3600 cpm per lo* cells, respectively, into mtDNA (Table II). The B-hour pulse-l-hour chase labeling protocol used results in mtDNA specific activities of 2300 cpm per pg in LMTK-and 1200 cpm per pg in LDTK-cells. These data estimate the cellular content of mtDNA as 1.9 pg per lOa LMTK-cells and 3.0 pg per lo* LDTK-cells. The yield of mtDNA by the "no gradient" procedure is approximately 65% from LMTK-cells and 30% from LDTK-. The reasons for this lower yield from the LDTK-cell have not been established. For either cell line, the "one step" sucrose gradient procedure was approximately 50% as efficient as the "no gradient" procedure.
Identical labeling conditions provided mtDNA specific activities in a ratio of approximately 2 : 1 for LMTK-and LDTKcells. Since in our routine labeling experiments mtDNA was not mass labeled, it was essential to determine whether a selective loss of newly replicated mtDNA during isolation from LDTK-cells could explain this discrepancy. LDTK-and LMTK-cell cultures were treated identically through ten generations of growth in [3H]thymidine medium and through mtDNA isolation.
The specific activity of isolated LDTK-mtDNA was 47% that of LMTK-.
Therefore, we conclude that isolated LDTK-mtDNA can be considered representative of total cell mtDNA.
We note that this ratio of specific activities of mtDNA is identical with the ratio of total mitochondrial thymidine kinase in these cell lines (7).
These labeling experiments demonstrate that the LDTKcell contains more, but significantly less than twice as much mtDNA as an LMTK-cell. It was of interest to determine Stereological analysis (12) of thin sections of LMTK-, LDTK-, and LA9 cells revealed that all contained 5 f 1 y0 of the total cell volume within mitochondria.
However, the LDTK-cell was approximately 15% larger than the LMTKcell by comparison of mean cell sizes in spinner cultures at comparable cell densities with the use of the Coulter model B cell counter.
Thus, the mtDNA content per unit mitochondrial volume is approximately 400/, greater in the LDTK-cell than in the LMTK-cell. Similar experiments were also performed with HeLaTKcells. For the purposes of this report, we have not attempted to accurately determine the loss of mtDNA during purification from HeLaTK-cells. Instead, the minimum cell content of mtDNA has been estimated as the amount isolated in ethidium bromide-CsCl gradients.
HeLaTK-cells grown in monolayer culture contained at least 15 pg of mtDNA per lo8 cells labeled to a specific activity of 1640 cpm per pg by the 5-hour pulse-l-hour chase labeling procedure (Table I). Even after considering the larger mitochondrial content of HeLaTK-(7% of cell volume), the mtDNA content of HeLaTK-cells is at least four times as large as that of either TK-L cell line. This result is consistent with our qualitative observation that the mitochondria of He-LaTK-cells are uniformly more well developed (i.e. contain more cristae per unit volume) than the mitochondria of L cells.

DISCUSSION
This study has attempted to quantitate the mtDNA contents of L and HeLa cells grown in tissue culture. This measurement has been complicated in earlier studies (15-17) by the large ratio of nuclear DNA to mtDNA.
To compensate for this difficulty these previous studies have resorted to lengthy mtDNA isolation procedures which have involved the use of DNase to remove contaminating nuclear DNA. We have circumvented this problem with the use of cultured cells which, due to the lack of the major cellular thymidine kinase activity (6), label mtDNA to a high specific activity relative to nuclear DNA.
Thus, by labeling thymidine kinase minus L and HeLa cell lines with [3H]thymidine, we have selectively assayed mtDNA during a brief isolation from cultured cells in order to measure the cellular content of mtDNA and the efficiency with which it can be isolated by our methods.
This study has estimated the mtDNA content of LMTK-cell as 1.9 f 0.4 pg per lo8 cells. This amount is the equivalent of 1100 f 250 mtDNA molecules per cell, which is in excellent agreement with the analysis by Nass of 1300 mtDNA genomes per wild type L cell (17). This agreement between our data and those of Nass may be fortuitous in light of the lengthy mtDNA purification procedure used in her study.
We have further shown that the yield of mtDNA from LMTK-cells may be as high as 75% with the "no gradient" isolation procedure or 35% to 50% with the "one step" sucrose procedure.
These relatively high yields add credence to those studies of mtDNA structure and replication which have used procedures which isolate the majority of the mtDNA population.
The "no gradient" procedure has the advantage of requiring only 90 min for the entire mtDNA isolation process. However, this method involves considerable nuclear DNA contamination of the upper band of the ethidium bromide-CsCl gradient, and is not suitable for mtDNA isolation at high purity without rebanding of closed circular DNA in ethidium bromide-CsCl.
With one rebanding step we have consistently obtained mtDNA preparations containing only a 4% nuclear DNA contaminant (Fig. 1B). Our experiments with LDTK-and HeLaTK-cells have re vealed two significant relationships in comparison with LMTKcells. First, the two L cell lines contain comparable masses of mtDNA per unit of mitochondrial volume. The somewhat larger mtDNA mass of the LDTK-cell may reflect a physiological balance between optimal gene dosage and a requirement for a minimal concentration of mtDNA molecular units. Secondly, the HeLaTK-cell line was found to contain at least four times the mtDNA mass per mitochondrial volume as either L cell line. Throughout this study we presented measurements of mtDNA content on a per cell basis. This method is convenient for measurements using tissue culture cells. Moreover, Hoffmann and Avers (20) have suggested that the mitochondrial volume of an animal cell may be organized in only one or a few giant branched organelles, as appears to be the case in yeast. However, our data do indicate that the most reasonable procedure by which 7995 to compare the mtDNA contents of different cell lines may be at the level of cellular mtDNA mass per mitochondrial volume.