Cell Sorting Experiments Link Persistent Mitochondrial DNA Damage with Loss of Mitochondrial Membrane Potential and Apoptotic Cell Death*

In order to understand the molecular events following oxidative stress, which lead to persistence of lesions in the mtDNA, experiments were performed on normal human fibroblast (NHF) expressing human telomerase reverse transcriptase (hTERT). The formation and repair of H2O2-induced DNA lesions were examined using quantitative PCR. It was found that NHF hTERTs show extensive mtDNA damage (∼4 lesions/10 kb) after exposure to 200 μm H2O2, which is partially repaired during a recovery period of 6 h. At the same time, the nDNA seemed to be completely resistant to damage. Cell sorting experiments revealed persistent mtDNA damage at 24 h only in the fraction of cells with low mitochondrial membrane potential (ΔΨm). Further analysis also showed increased production of H2O2 by these cells, which subsequently undergo apoptosis. This work supports a hypothesis for a feed-forward cascade of reactive oxygen species generation and mtDNA damage and also suggested a possible mechanism for persistence of lesions in the mtDNA involving a drop in ΔΨm, compromised protein import, secondary reactive oxygen species generation, and loss of repair capacity.

Reactive oxygen species (ROS) 1 are ubiquitous toxicants generated primarily by the mitochondrial respiration and by other biochemical processes within cells (1). In addition to these sources, ROS are also produced in pathophysiological conditions such as inflammation and ischemic reperfusion injury (2). Furthermore, certain environmental pollutants (like pesticides and herbicides), as well as smoking, can induce ROS either directly or indirectly (1,3,4). It is well known that ROS can damage biomolecules and have been postulated as an important cause of aging, cancer, and various human degenerative diseases (3,5). One of the most widely used compounds to produce oxidative stress, both in vitro and in vivo biological studies, is hydrogen peroxide (H 2 O 2 ). Unlike other ROS, H 2 O 2 is not charged and is therefore freely diffusible within the cell. It can arise spontaneously or be generated from superoxide by superoxide dismutase. More importantly, in the presence of metal ions such as copper or iron, H 2 O 2 can undergo Fenton chemistry and give rise to the potent hydroxyl radical. This radical is extremely reactive and can damage different components of the cell (6). The responses of cultured cells to H 2 O 2 vary according to cell type and are known to be both time-and, especially, dose-dependent (for review see Ref. 7). For example, low concentrations (3-15 M) of H 2 O 2 were shown to activate a mitogenic response (8), whereas higher doses can have either cytostatic or cytotoxic effects. It was observed that concentrations ranging from 50 to 150 M promote DNA damage (9,10) replicative senescence (11)(12)(13), sustained p21 levels, cell cycle arrest, transient elevation of p53 protein (14), and temporary growth arrest followed by increased resistance to subsequent oxidative stress (8,15). In contrast, H 2 O 2 doses of 200 M and higher were demonstrated to induce apoptosis (16), necrosis (16,17), protein oxidation (18), as well as lesions in both nuclear and mitochondrial genomes (9,10).
With regard to DNA damage, we and others have demonstrated that the mtDNA of different cell types and organisms is more prone to oxidative injury than the nDNA (9, 10, 19 -22). Additionally, Yakes and Van Houten (9) demonstrated that depending on the intensity of the damage (dose and time of exposure), nuclear and mitochondrial genomes are differentially damaged and repaired. Although the precise molecular reason(s) that lead to the persistence of lesions in the mitochondrial genome is not clear, several explanations have been offered to clarify the increased vulnerability of the mtDNA to damage. These include the lack of a compact nucleosome structure as compared with the nDNA, the limited mitochondrial repair pathways, as well as the proximity of the mtDNA to the main source of ROS generation (9,23). More work is needed to understand the molecular events underlying the susceptibility of the mtDNA to oxidative damage.
Whereas the initial results obtained regarding the kinetics of damage and repair in both mitochondria and nuclear genomes were based on SV40-transformed fibroblasts (9), data from the literature raise concerns as to the suitability of the SV40transformed cell model. In fact, a significant number of physiological changes after viral transformation have been reported. For example, loss of p53 function with consequent alteration in cell cycle control (24) decreased DNA repair (25,26), and diminished catalase activity (27) was found.
In an effort to better understand the basis of the increased susceptibility and persistent damage observed in mtDNA after oxidative stress, we have used a non-transformed diploid fibroblast cell line expressing the telomerase gene (NHF hTERT) to follow the formation and removal of H 2 O 2 -induced lesions. With this approach it was observed that these cells accumulate large amounts of mtDNA damage, which can be completely repaired after a 15-min treatment but not a 60-min treatment.
Cell sorting experiments revealed that persistent lesions in the mtDNA correlate with loss of ⌬m, increased ROS generation, and cell death. Interestingly, we also show that the nDNA of the NHF hTERTs seems totally resistant to H 2 O 2 -induced damage, suggesting that SV40-transformation makes the nuclear genome more prone to oxidative injury.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-The normal human fibroblasts transfected with the telomerase gene (NHF hTERT) used throughout this study were a gift from Dr. Carl Barrett (NCI, National Institutes of Health). The cells were routinely grown and sub-cultured in 75-cm 2 flasks at 37°C and 5% CO 2 in DMEM/F-12 medium (Invitrogen) supplemented with 10% fetal bovine serum (HyClone), SAP (0.2 mM serine, 0.1 mM aspartic acid, and 1.0 mM pyruvate), 50 units/ml penicillin/streptomycin (Invitrogen), and 800 g/ml G418 (Invitrogen) for selection of the stable transfectants.
H 2 O 2 Treatments-Cells were seeded in 100-mm Petri dishes 15-18 h prior to the experiments. Immediately before treatment cells were washed once with DMEM/F-12 without any supplements, and the conditioned medium was saved for later use. A 30 mM stock of H 2 O 2 (30% Sigma) was prepared in PBS (Invitrogen) and used to generate the 200 M solution (in DMEM/F-12 alone) with which the cells were challenged. Monolayer cultures were exposed to H 2 O 2 for 15 or 60 min and harvested immediately or allowed to recover for up to 24 h (in this case, in the original conditioned medium).
DNA Isolation, QPCR Assay, and DNA Damage Analysis-High molecular weight DNA was extracted, and QPCR was performed and analyzed as described elsewhere (22). Briefly, genomic DNA was isolated, and specific primers were used to amplify a fragment of the mitochondrial and/or nDNA. The assay is based on the premise that DNA lesions (including oxidative damage such as strand breaks, base modifications, and abasic sites) block the progression of the polymerase so that only undamaged templates can participate in the PCR. Thus, amplification is inversely proportional to DNA damage: the more lesions encountered on the target DNA, the less the amplification. Note that for the mitochondrial genome both a short (221 bp) and a long fragment (8.9 kb) are routinely amplified. The rationale underlying this procedure is that the probability of introducing a lesion in a short segment is low, and therefore amplification of this segment gives an accurate estimation of the copy number of mtDNA in the sample. The data obtained from the small fragment were subsequently used to normalize the results of the 8.9-kb target (more details can be obtained in Ref. 22). Amplification of treated samples was then compared with controls, and the relative amplification was calculated. These values were next used to estimate the average number of lesions per 10 kb of the genome, using a Poisson distribution (22,28). It is important to mention that a "negative" number of lesions was obtained when amplification of the treated sample was higher than that of the matched control. This phenomenon can occur by means of induced repair activity in the treated sample (9,29). Results presented here are the mean of two sets of PCR for each target gene of at least three different biological experiments. Student's unpaired t test was performed to evaluate statistical significance.
Mitochondrial Membrane Potential Analysis-Changes in the ⌬m were measured by flow cytometry using JC-1 (5,5Ј,6,6Ј-tetrachloro-1,1Ј,3,3Ј-tetra-ethylbenzimidazolylcarbocyanine iodide; Molecular Probes). JC-1 was chosen because of its specificity regarding the mitochondrial membrane (30 -32). At various time points (every 2 h up to 24 h and then at 36 h and 48 h) after the challenge with H 2 O 2 , control and treated cells were harvested, washed twice with PBS, and incubated with 10 M JC-1 for 10 min (37°C; 5% CO 2 atmosphere). For each sample, 10,000 cells were examined on an FL-1 (530 nm) versus FL-2 (585 nm) dot plot using a BD Biosciences FACSort. JC-1 has dual emission depending on the state of the ⌬m, forming aggregates with a high 585 nm fluorescence in cells (which indicates a normal ⌬m). Loss of mitochondrial membrane integrity results in reduction in 585 nm fluorescence with a concurrent gain in 530 nm fluorescence as the dye shifts from an aggregate to a monomeric state (33). Therefore, retention of the dye in the cell can be monitored. The data were converted to density plots using the CellQuest software.
Cell Cycle Analysis with Propidium Iodide-Cells were treated as described above and at various times after the treatments were subjected to cell cycle analysis by flow cytometry using propidium iodide (PI, Molecular Probes). Briefly, at each time point both control and treated cells were harvested, washed twice with PBS, and incubated at room temperature for 15 min with 1 mg/ml RNase, 1 l of Triton X-100(Sigma), and 20 g/ml of PI. Samples were then immediately analyzed by fluorescence-activated cell sorter.
Caspase Activation Assay-Measurement of caspase 3 activation was performed fluorimetrically using the Bio-Rad Apopain Kit, according to instructions of the manufacturer. Briefly, 24 h after the challenge with H 2 O 2 cells were harvested and subjected to total protein extraction using CHAPS buffer with 5 mM dithiothreitol (Cell Signaling) and 1 mM phenylmethylsulfonyl fluoride. After 5 rounds of quick-freezing (dry ice embedded in ethanol) and thawing (37°C bath), protein concentrations were determined, and subsequently, 50 l of each sample was used in the Apopain assay. Pro-caspase/apopain activity was monitored using a fluorogenic peptide substrate, Ac-DEVD-AFC. Apopain enzymatically cleaved the AFC from the peptide and released it in a free form, which upon exposure to near UV light produced a blue-green fluorescence (detected at 500 -550 nm). A calibration curve was generated first by measuring fluorescence of AFC released from a serial dilution of the standard substrate (AC-DEVD-AFC). A positive control consisting of protein extract of Jurkat cells treated with a Fas ligand, a model system for apoptosis, was also included. Duplicate fluorescence readings of the samples were taken every 30 min after reaction was started for up to 3 h until the readings stabilized. Caspase activity was calculated comparing results from the samples to the calibration curve. Data were reported as fluorescence readings.
Sorting Procedure Based on JC-1 Staining and Flow Cytometry-Sorting of the normal and depolarized (low ⌬m) JC-1-stained cells was accomplished using a BD Biosciences FACSVantage SE. In these experiments, both control and 60-min treated cells were initially gated in forward-scatter versus a side-scatter dot plot to exclude debris. Sort gates were set up on an FL-1 (530:30 nm) versus FL-2 (575:26 nm) dot plot with increased stringency for only the depolarized cells. An all-sort cell population (cells gated on the initial forward-scatter versus sidescatter dot plot) was also collected to control for the introduction of DNA lesions and/or any change in the ⌬m because of the sorting procedure. The data were converted to density plots using CellQuest software. Individual sorted cell populations were then immediately subjected to DNA extraction and QPCR analysis or placed back into culture for viability assays.
H 2 O 2 Detection from Sorted Cells-Cells were sorted as described above, and supernatants were immediately collected by centrifugation and subjected to analysis using the Amplex Red Kit (Molecular Probes). The assay is based on the detection of H 2 O 2 using 10-acetyl-7-hydroxiphenoxazine (Amplex Red reagent), a highly stable and specific probe for this compound. In the presence of horseradish peroxidase, Amplex Red reacts with H 2 O 2 at a 1:1 stoichiometry, producing a highly fluorescent product, resorufin, which can be detected using excitation and emission wavelengths of 530 and 590 nm, respectively (34). Briefly, 100 l of the supernatants was loaded in a 96-well microtiter plate and incubated for 30 min (protected from light and at room temperature) in the presence of phosphate buffer containing 400 M of Amplex Red reagent and 2 units/ml horseradish peroxidase. Background fluorescence was corrected by subtracting the values derived from PBS alone. Controls including catalase were also included to ensure that the fluorescent signal was due to H 2 O 2 . For each sample, all measurements were made in duplicate. The mean of those readings was then used to estimate H 2 O 2 concentration using a standard H 2 O 2 curve.

RESULTS
The purpose of this study was to investigate the formation and repair of H 2 O 2 -induced lesions in both nuclear and mitochondrial genomes of normal human fibroblasts transfected with the telomerase gene. In addition, we also wanted to understand the apparent loss of DNA repair activity in cells showing persistent mtDNA damage and to gain some insights into the biological outcome(s) of this permanent damage.
H 2 O 2 -induced Damage-Quantitation of lesions introduced both in the mitochondrial and nuclear genomes was performed using QPCR, and the details of the assay have been published previously (9, 22, 28). Fig. 1A shows the relative amplification for the mtDNA of NHF hTERT cells after treatment with 200 M H 2 O 2 and repair periods of up to 24 h. As can be seen, a 15-min exposure to H 2 O 2 gives rise to an ϳ70% decrease in amplification as compared with the untreated control. Note that amplification was completely restored after a 1.5-h incubation in conditioned medium. In contrast, cells challenged for 60 min showed a marked reduction in amplification, corresponding to ϳ4 lesions/10 kb. Unlike the 15-min treatment, the 60-min treatment leads to lesions that persisted even after long recovery periods (24 h). During the first 6 h, ϳ50% of the damage from a 60-min treatment could be repaired. However, no further removal could be observed up to 24 h. These data suggest that the 60-min treatment passed some threshold of damage, such that some other deleterious events have taken place inside the mitochondria.
Surprisingly, the nDNA of NHF hTERT cells did not show any decrease in amplification (when compared with untreated controls) even after a 60-min exposure to H 2 O 2 (Fig. 1B). This result suggested an absence of lesions detected in the two genomic regions (␤-globin and HPRT) studied. These data differ significantly from the findings of Yakes and Van Houten (9), which demonstrated the same type of treatment to introduce 0.5 lesions/10 kb into the nuclear genome of SV40-transformed fibroblasts. Interestingly, we were able to perceive a very low amount of nDNA damage 24 h after the treatments, which was coupled to the observation that a vast proportion of cells (ϳ50%, data not shown) were detaching from the dishes. Together, these results indicate that the biology of a specific population of cells has changed after H 2 O 2 treatment, culminating in cell death. Moreover, the absence of nDNA damage suggests that, in this scenario, the cell death is exclusively a consequence of persistent lesions in the mtDNA. Mitochondrial Membrane Potential as an Indicator of Apoptosis-We next sought to determine whether the fraction of cells that were detaching from the dishes was dying by apoptotic death. We reasoned that cells with extensive and persistent mtDNA damage might not be able to maintain the ⌬m across the inner membrane, believed to play a role in the apoptotic process (reviewed in Ref. 35). The integrity of the ⌬m was evaluated using JC-1, a fluorophore recognized as the most specific probe for the detection of changes in the ⌬m of different cell types as compared with other available dyes (30 -32). Cells were treated as described under "Experimental Procedures" for 15 or 60 min and were analyzed every 2 h after the exposure to H 2 O 2 . No differences in the ⌬m were observed when comparing the 15-min with the control at any time monitored ( Fig. 2A, upper right panel). However, a decline in the ⌬m was detected at around 14 h in cells treated with H 2 O 2 for 60 min, with the highest proportion of cells (ϳ70%) with decreased ⌬m values observed at 24 h ( Fig. 2A). These data suggest that the cells treated for long periods of time with H 2 O 2 might be undergoing apoptosis.
Caspase Activation and Apoptotic Population Detection by PI-To confirm that a fraction of the cells in our experiments were undergoing apoptotic death, other apoptosis-specific assays were conducted including analysis of caspase-3 activation as well as the staining of the nDNA with PI. Caspase 3 activation was followed using a fluorimetric assay as described under "Experimental Procedures." Fig. 2B shows that only cells that were exposed for 60 min to H 2 O 2 had increased caspase-3 activation as compared with the negative control. DNA staining with PI also indicated a significant percentage of apoptotic cells beginning at 24 h with a peak at 72 h (Fig. 2C). Taken together, these data indicate that a fraction of NHF hTERTs, when exposed to H 2 O 2 for 60 min, undergoes apoptosis; this process was exclusively a consequence of extensive damage to the mitochondrial genome. Like many other studies with fibroblasts, these same cells did not show DNA laddering at any time monitored (24,36,48, and 72 h; data not shown).
Sorting Procedure by Flow Cytometry and JC-1 Staining-Integrity of the ⌬m plays a role not only in apoptosis but also in trafficking various proteins into the mitochondria. In fact, nuclear encoded proteins that act in this organelle have to be targeted, unfolded, and refolded inside the mitochondrion by different protein complexes attached to the outer and inner mitochondrial membranes. This import mechanism also depended on the maintenance of a high ⌬m (reviewed in Ref. 36). Considering that repair proteins that act in mitochondria are nuclear encoded, we next sought to evaluate whether the persistent mtDNA damage observed only in the cells treated for 60 min correlated with the drop in ⌬m. To determine whether this was the case, a procedure relying on JC-1 staining and flow cytometry was adopted.
By using a FACSVantage SE flow cytometer and JC-1, both control and H 2 O 2 -treated cells were sorted into separate populations based on having low (depolarized cells) or high ⌬m values (polarized cells). Fig. 3 shows the gating scheme used to individually isolate polarized and depolarized populations. Immediately after sorting, DNA was isolated, and damage was evaluated with the QPCR assay. Fig. 4A illustrates the results from the mtDNA damage analysis on the various sorted populations of cells. Interestingly, only the depolarized population from the treated cells showed significant mtDNA damage. This is the first evidence that directly links depolarization of the ⌬m with persistent lesions in this genome. As described under "Experimental Procedures," an all-sort cell population where cells passed through the sorter but were not physically separated was included. Note that no significant differences in the amount of mtDNA damage were observed between the depolarized and all-sort treated cells, which is probably due to stringency of the gated depolarized population that excluded some depolarized cells (those falling between the two sort gates) that were included in the all-sort. In addition, it is important to notice that the majority of the treated (Ն60%) cells were falling in the depolarized gate (example depicted in Fig. 3).
H 2 O 2 Released from Sorted Cells and Viability Assessment-Our current working hypothesis predicts that a damaged mitochondrion generates more ROS than an intact organelle. The mtDNA encodes 13 polypeptides involved in the function of the electron transport chain (ETC). Incomplete flow of electrons could be caused by the lack of expression of vital mitochondrial genes from a damaged genome and would be expected to give rise to increased levels of ROS. In this way, a high steady-state level of ROS could be generated inside the organelle, which may in turn continuously injure the mtDNA in a feed-forward cascade. Eventually, this vicious cycle results in loss of mitochondria function and apoptotic cell death. Thus we next sought to determine whether ROS were being released from the sorted cells, especially from the population showing both low ⌬m and permanent mtDNA damage. As described under "Experimental Procedures," supernatants of sorted cells were collected and used to perform the Amplex Red assay, which is specific for H 2 O 2 . As can be observed, treated depolarized cells release significantly more H 2 O 2 than polarized counterparts 24 h after the treatments (Fig. 4B). It could be argued that this result reflects the presence of the initially supplied H 2 O 2 ; however, the exogenous compound showed a half-life of 30 min in the presence of NHF hTERT cells. Moreover, H 2 O 2 detection reached background levels by 4 h after the treatments (data not shown).
Finally, it was important to confirm that the cells with persistent mtDNA damage were the inviable cells after the treatments. In an effort to investigate this phenomenon, sterile sorting was conducted, and both H 2 O 2-treated and non-treated polarized and depolarized populations were re-cultured to access viability. Cells were monitored for the next 72 h and counted with a Coulter counter. It was found that 80% of the control polarized cells were able to attach and grow, whereas the portion of treated cells with normal ⌬m showed ϳ66% attachment. Only 5% of the depolarized treated cells was able to attach as compared with 13% of the depolarized controls (data not shown), suggesting that these populations are mainly composed of non-viable cells. DISCUSSION The increased susceptibility of the mtDNA to H 2 O 2 -induced damage, as compared with the nDNA, has been documented previously (9,10,23); however, the specific molecular causes of these phenomena are not well understood. It has been hypothesized that a combination of biological determinants, such as the proximity of the mtDNA to ROS, the lack of a compact nucleosome structure to protect the genome, and a paucity of mtDNA damage-processing pathways relative to those present in the nucleus play a role (reviewed in Ref. 23). In an attempt to understand these phenomena better, we have used telomerase-expressing normal diploid fibroblasts and found that repair of ROS-induced lesions in the mitochondrial genome depends on a threshold of dose and time and that persistent mtDNA damage leads to loss of ⌬m and subsequent cell death.
The present study confirmed earlier findings (9, 10), H 2 O 2 induces a large amount of mtDNA lesions (Fig. 1A) and little or no nDNA damage (Fig. 1B). Surprisingly, the mitochondrial genome of hTERT-transfected cells appeared to be more vulnerable to H 2 O 2 than previously reported (9, 10, 19 -22) for several other cell types, whereas the nuclear genome of the same cells was completely resistant to damage. This is likely a reflection of the fact that hTERT-transfected fibroblasts have increased levels of intracellular iron, making the mtDNA more prone to damage. 2 The fact that the lesions induced in the mitochondrial genome after a 60-min treatment are repaired up to 6 h (ϳ50% of lesions are removed) agrees with previous reports (23,(37)(38)(39) suggesting that repair in this organelle is capable of removing oxidative DNA lesions. However, these data also indicate that removal of damage is clearly dependent on the total amount of lesions induced over a period of time, after which additional events take place in the mitochondrion that seem to compromise repair function.
DNA damage caused by ROS is removed by the base excision repair (BER) machinery both in the nucleus and in the mito- chondria. Although oxidative stress induces a wide variety of lesions, many are removed from the mtDNA through the action of a glycosylase with an associated abasic site-lyase activity, followed by the activity of ␥-polymerase and DNA ligase (reviewed in Refs. 23 and 37-39). Because the mitochondrial genome encodes polypeptides exclusively involved in the production of ATP through oxidative phosphorylation, BER pro-teins (as well as the other 1000 estimated proteins that function in this organelle) are nuclear encoded and therefore must be targeted to the mitochondrion.
Trafficking of nuclear encoded proteins from the cytosol to the mitochondria is achieved by the concerted action of at least four translocation complexes (see Ref. 36 for review). In this scenario, the ⌬m provides one of the major driving forces for the vectorial transport of polypeptide chains by these import machines (40). The results obtained during this study provide evidence that a majority (ϳ70%) of cells exposed to H 2 O 2 for 60 min have lost their ⌬m within 24 h of the treatments ( Fig.  2A). This led us to hypothesize that the presence of mtDNA lesions and the loss of ⌬m are intimately linked and may result in lack of repair activity through a compromised import mechanism. Data to support this concept were obtained from our sorting experiments (Fig. 4A). In fact, this approach clearly showed that only those cells that have low ⌬m also showed mtDNA damage, suggesting that a compromised import mechanism was involved in the lack of repair activity in mitochondria.
However, one cannot rule out other putative causes that may impair the overall repair activity in the mitochondria. For example, Graziewicz et al. (18) showed that the human ␥-polymerase, the only polymerase present and active in mitochondria, is prone to oxidative damage. This might also be the case in our experiments. Along the same line, another BER protein that could be the target of oxidative attack is the human endonuclease III homologue hNTH1 (41). Although there is no direct evidence that hNTH1 is sensitive to oxidative damage, one could envision a mechanism analogous to the inactivation of aconitase. This protein localizes in the mitochondrial matrix and, like hNTH1, has an iron-sulfur [4Fe-4S] cluster in its active site. Yan et al. (42) showed that aconitase is modified and inactivated by oxidative stress, which is believed to occur because of the release of one iron atom from the cluster (43). It is worth mentioning that if oxidation of hNTH1 is taking place in our experiments, one would expect not only the enzyme to decrease repair efficiency but also to release iron, which could further contribute to Fenton chemistry and higher the levels of mtDNA damage. In this context, preliminary data following mitochondrial protein oxidation revealed that several proteins are oxidized in this organelle in NHF hTERT cells exposed to H 2 O 2 (data not shown).
Alternatively, saturation of repair activity by increased promotion of lesions, even in the presence of functional repair machinery, could be responsible for the persistence of lesions in the mitochondrial genome. Inappropriate electron flow on an already damaged ETC represents a potential source of additional waves of ROS (9). In this case, a vicious cycle is established where the basal level of ROS generation as well as of mtDNA lesions is enhanced, possibly saturating the active repair system. The fact that the depolarized sorted cells (those that also show the mtDNA damage) release increased amounts of H 2 O 2 as compared with the polarized counterparts (Fig. 4B) is consistent with this model.
It is worth pointing out that regardless of the mechanism(s) contributing to the maintenance of lesions in the mitochondrial genome, our experiments indicate that the biological effect of these persistent lesions is apoptosis (Fig. 2, A-C). Accordingly, Dobson et al. (44) demonstrated that targeting an 8-oxoguanine-specific glycosylase not only enhances the repair of menadione-induced mtDNA lesions but also the viability of treated cells. In addition, it has been proposed that the differential susceptibility of glial cell types to oxidative stress and apoptosis correlates with increased oxidative mtDNA damage (45). Although reports are available showing that H 2 O 2 induces FIG. 4. Persistent mitochondrial DNA damage is restricted to treated depolarized cells, which also show significantly increased levels of ROS. A, QPCR analysis of the mitochondrial target of NHF hTERT cells after the sorting procedure. Cells were treated as described previously, allowed to repair for 24 h, stained with JC-1 for 10 min, and subjected to flow cytometric sorting using the FACSVantage SE. Controls included 1) cells harvested immediately after the treatments (data not shown); 2) aliquot harvested 24 h after the challenge and kept with JC-1 for the same period as sorted cells but not submitted to this procedure (data not shown); and 3) cells that passed through the sorter but no separation of the two distinct populations conducted (all-sort). Data represent the amount of lesions per 10 kb in the populations that were sorted (n ϭ 2). B, H 2 O 2 detection in sorted cells. After sorting, cells were harvested, and the supernatant was collected for H 2 O 2 measurements. To ensure comparability of results, the ratio between cell number and volume collected was constant so that the dilution was the same in all samples (i.e. the volume of flow cytometer fluid per cell was identical). Aliquots of the supernatant were then subjected to the Amplex Red coupled to horseradish peroxidase assay and H 2 O 2 concentration calculated based on a standard H 2 O 2 curve. Numbers represent the mean Ϯ S.E. (n ϭ 2). Unpaired Student's t test applied for statistical significance comparing depolarized samples to polarized counterparts (p values Ͻ 0.05). apoptotic death, the mechanisms through which this oxygen byproduct promotes apoptosis are still debated (see Ref. 7 for review). Interestingly, down-regulation of mitochondrially encoded mRNA, rRNA, and DNA was reported in apoptotic cells (46 -48). All these events are consistent with lesions being present in the mitochondrial genome, as previously proposed by Yakes and Van Houten (9).
A striking finding in this study is the total absence of nDNA damage in the same cells that show a large number of mtDNA lesions. In fact, detectable levels of nDNA damage were only observed 24 h after the treatments, which is when apoptosis has been initiated. Because the mtDNA showed increased susceptibility to oxidative damage in hTERT-transfected cells, one might also expect the nDNA to be more prone to insult. It might be argued that the nuclear genome, unlike the mitochondrial one, is better protected by histones and, thus, less accessible to oxidative attack. In addition, the presence of additional and overlapping repair pathways (such as the nucleotide excision repair) (49) could account for the lack of damage accumulation. An appealing alternative relates to the fact that the cells used in this study are not virus-transformed and, therefore, have intact p53 function. Although p53 has been claimed to interact directly and promote the activity of the BER machinery (50), we have no evidence that this is the case in our experiments. One could envision that an active p53 is necessary and sufficient to decrease nDNA damage resulting from oxidative stress. Further p53 analysis should help to shed light onto this problem.
In summary, the findings of the sorting experiments lead us to suggest a chain of events, involving persistent mtDNA damage, drop in ⌬m, loss of repair activity and apoptosis. Because mtDNA lesions were observed within 5 min of H 2 O 2 exposure (data not shown), it is most likely that mtDNA damage is the initial event triggering this cascade. Although our results do not implicate a drop in the ⌬m as the sole event to impair mitochondrial repair activity, they provide a novel insight to account for the increased susceptibility of the mitochondrial genome to oxidative attack as compared with the nDNA. An increased understanding of the kinetics of damage and repair (or lack of repair) in the mitochondrial genome should provide new insights into the biological/physiological significance of mtDNA damage and, eventually, mutagenesis.