Mitochondrial DNA Content Increase in Response to Cigarette Smoking

Mitochondria are intracellular organelles that produce ATP by the coupling of oxidative phosphorylation to respiration. They provide the major source of energy to the cell and are also the major source of endogenous reactive oxygen species (ROS). Mitochondria also play a critical role in apoptosis, or programmed cell death, by the release of soluble factors into the cytoplasm, thereby providing a critical role in the maintenance of cell proliferation. Aging effects and cancer development have been linked to exposure to ROS in humans, particularly through injury to both mitochondrial and nuclear DNA (1). The mitochondrial genome is highly susceptible to DNA damage inflicted by ROS and mutagens and exhibits higher rates of mutation than does the nuclear genome, and DNA damage persists longer in the mitochondrial genome (2, 3). The absence of histones, which serve packaging and protective purposes for nuclear DNA, and the error-prone replication and repair of mitochondrial genes all contribute to the vulnerable nature of mitochondrial DNA (mtDNA; ref. 4).

Mitochondrial function also declines with age, particularly in tissues with high energy requirements (5-7). The decrease in mitochondrial function associated with age is also accompanied by rearrangements in mtDNA (8-11). These and other DNA alterations are likely to be the result of the combined effects of oxidative damage and reduced efficiency of mtDNA repair systems with increasing age (2-4). Age-related mitochondrial alterations also include an increase in mtDNA content (12). This increase in mtDNA content is accompanied by mtDNA deletions, evidence of oxidative damage, and decreased transcription of mitochondrial-specific proteins in specific tissues (13, 14). An increase in mtDNA content has been shown to be a response to endogenous and exogenous oxidative stress in in vitro studies of human cell systems, as well as a characteristic of senescent cells (15, 16). This response is likely due to transcription of nuclear genome-encoded genes that are translocated into the mitochondria in response to altered mitochondrial membrane potential due to production of ROS (16).

Smoking has been shown to inhibit mitochondrial enzyme activity in platelets and cause mitochondrial dysfunction in alveolar macrophages (17, 18). Smoking-related damage to respiratory chain function in lymphocytes has been correlated with measures of oxidative damage in smoking individuals (19). Studies of lung tissue from smokers have shown elevated measures of DNA damage and increased DNA mutation when compared with lung tissue from nonsmokers (20-22). Two small retrospective studies of mtDNA content in lung tissue from patients with and without lung cancer showed an inconsistent association of increased mtDNA content with smoking and age (21, 22). Other small studies have provided conflicting evidence for and against mtDNA injury related to tobacco exposure in human cohorts (23, 24). Tobacco smoke itself is a complex mixture of thousands of compounds with varied biological effects, but studies that have examined the effect of tobacco smoke as a whole have shown that tobacco smoke induces mitochondrial damage and depolarization, as well as mtDNA damage (25), and that this effect can be blocked by treatment with antioxidants (26, 27). Increases in mtDNA content and declines in mitochondrial function are associated with aging and are observed in response to DNA-damaging agents, including tobacco smoke. However, the effect of smoking on the mitochondrial genome for a large, broad population has not been well defined. To clarify the nature of mitochondrial genotoxic effects after smoking exposure and smoking cessation, we examined normalized mtDNA quantity in cells from saliva of subjects recruited at baseline of a prospective head and neck cancer screening study.

We did a cross-sectional study of mtDNA content in relation to tobacco exposure in upper aerodigestive tract mucosa. Local community organizations were contacted for participation in a screening study for upper aerodigestive tract malignancy. Informed consent was obtained by institutional review board–approved protocol, and 604 participants were recruited. All enrolled subjects were administered a confidential survey. Cigarette smoking histories included current smoking status (never, former, current), duration of smoking, and number of cigarettes per day. Clinical and demographic characteristics included age, gender, race, income, presence of pulmonary or cardiac disease, and alcohol drinking history.

Cumulative ethanol consumption was calculated separately for (a) average drinks per weekend over a yearly period × years of exposure and (b) average drinks/5-day workweek over a yearly period × years of exposure. A complete head and neck examination was done, and a gargle, rinse, and oral cavity brush specimen was obtained. Oral rinse samples were obtained by swishing and gargling for 15 seconds with 25 mL of sterile 0.9% NaCl. The 25-mL swish and gargle samples were centrifuged at 2,500 rpm for 15 minutes, the cell pellet was retained and placed in 1% SDS/proteinase K (0.5 mg/mL) at 48 C for 48 hours, phenol-chloroform was extracted, and ethanol was precipitated.

A Perkin-Elmer/ABI 7900 thermocycler was used to perform real-time PCR amplification for β-actin, a nuclear gene, and cytochrome c oxidase I (Cox I), a mitochondrial gene to provide normalization of mtDNA content relative to nuclear genomic copy number (28, 29). Primers were custom-made and obtained from Invitrogen (Carlsbad, CA): Cox I forward primer, 5′-TTCGCCGACCGTTGACTATTCTCT-3′ and reverse primer, 5′-AAGATTATTACAAATGCATGGGC-3′ (97-bp product); Cox II forward primer, 5′-CCCCACATTAGGCTTAAAAACAGAT-3′ and reverse primer, 5′-ACCGCTACACGACCGGGGGTATA-3′ (80-bp product); β-actin forward primer, 5′-ACCCACACTGTGCCCATCTAC-3′ and reverse primer, 5′-TCGGTGAGGATCTTCATGAGGTA-3′ (103-bp product). Taqman probes (Applied Biosystems, Foster City, CA) included Cox I probe 6-FAM-AACGACCACATCTACAACGTTATCGTCAC-TAMRA, Cox II 6-FAM-CAATTCCCGGACGTCTAAACCAAACCACTTTC-TAMRA, and β-actin probe 6-FAM-ATGCCCTCCCCCATGCCATCC-TAMRA. PCR amplifications were carried out in buffer containing 16.6 mmol/L ammonium sulfate, 67 mmol/L Trizma (pH 8.3), 2.5 mmol/L MgCl₂, 10 mmol/L β-mercaptoethanol, 0.1% DMSO, 600 nmol/L each of forward and reverse primers, 200 nmol/L Taqman probe, 0.6 unit platinum Taq polymerase, and 2% Rox reference dye. DNA (500 pg) was used to amplify Cox I or Cox II, and 10 ng were used to amplify β-actin. PCR reactions were done in triplicate for each gene. Standard curves where obtained using Adult Retinal Pigmented Epithelia-19 (ARPE-19) cell line DNA from untreated control cells. Mitochondrial to nuclear DNA ratios were calculated by dividing the mtDNA quantity for each gene by the corresponding β-actin quantity. Intra-assay variation was low: for triplicate amplifications, the mean SD was 23% of the measured quantity for the triplicates with a 95% confidence interval of ±2%. A representative curve is shown in Supplementary Fig. 1.

Of note, primers were chosen to avoid amplification of genomic pseudogenes that are homologous to mitochondrial genes; this was confirmed by BLAST results for all primers (http://www.ncbi.nlm.nih.gov/BLAST/). In addition, to exclude the possibility that random nicking due to tobacco related may lead to an apparent increase in amplification of these shorter targets due to lack of DNA supercoiling, we treated 100 matched samples from subjects with varied smoking exposures with and without PvuII digestion to linearize mtDNA before amplification and noted no difference in the smoking-related increase in mtDNA/nuclear DNA. Finally, mtDNA content was compared with mitochondrial mass as measured by amount of Cox II protein within samples by performing Western blot analysis on samples with a variety of mtDNA/nuclear DNA ratios, showing excellent agreement between mtDNA content and Cox II protein normalized to β-actin controls (Fig. 1). The expression of Cox II was assessed by Western blotting analysis of total protein of homogenized saliva. The protein concentrations were determined using Bio-Rad (Hercules, CA) protein dye with bovine serum albumin as standard. Thirty micrograms of each sample were loaded onto 13% SDS-polyacrylamide gels and electrophoresed to resolve proteins. The proteins were then transferred onto nitrocellulose membrane. Membrane was blocked in PBS supplemented with 5% nonfat dry milk for 1 hour at room temperature. Membrane was first probed with anti-Cox II antibodies (Molecular Probes, Eugene, OR) with 1:500 dilution and then with peroxidase-conjugated rabbit anti-mouse IgG (Santa Cruz Biotechnology, Santa Cruz, CA). Cox II was visualized with detection reagents (Amersham Biosciences, Piscataway, NJ). Control of protein loading was achieved by reprobing with anti-actin (Sigma, St. Louis, MO).

All statistical procedures were done using SAS Institute Software, version 8.2 (Carey, NC). Sample means and SDs were calculated for continuous variables, and frequency responses were determined for categorical data. Univariate tests were done for Cox I and Cox II as well as other demographic and clinical variables. Standard t tests were done to determine associations between Cox I and Cox II ratios and smoking status, including Spearman correlation to examine association with smoking dosage as measured by pack-years. Data analysis revealed that values for both Cox I and Cox II ratios were skewed towards higher values. Therefore, simple linear models would not ideally fit the skewed distribution of Cox I and Cox II; thus, data transformation was done for multiple variable regression analysis. Based on exploratory data analyses, the cubic root transformation for Cox I and Cox II ratio was chosen to approximate a normal distribution. For multiple variable regression model selection, clinical and demographic characteristics included age, gender, race, income, presence of pulmonary or cardiac disease, current smoking and ethanol consumption status, cumulative tobacco exposure and ethanol consumption, and time of onset/time of cessation of smoking and alcohol consumption (Table 1). Multivariate multiple regression was applied to estimate the relationship between two correlated responses and a set of possible predictors or risk factors listed in Table 1. Hence, we had looked at the joint effect of a set of predictors on the values for Cox I and Cox II. The models were fit by using the stepwise improvement technique. The Ps for F statistics reflecting variables' contribution to the model were compared with the value of 0.25. The final regression model had the following risk factors: age of a participant, smoking habit, drinking status, and income. There is a significant association between the set of responses and the set of risk factors (P < 0.0001). We chose the simplest way to assess strength of effect by examining the univariate tests. Table 3 lists the estimated regression coefficients with corresponding Ps for their effect on the response.

We determined the ratio of mtDNA to nuclear DNA in cells from the saliva of 604 individuals with varied exposures that conferred an elevated risk for head and neck cancer. The study population surveyed ranged from 18 to 94 years with a mean age of 61.8 years and median age of 63 years. The population was predominantly male (64%), including 384 men and 220 women; 246 never smokers, 289 ex smokers, and 69 current smokers were included. Current smokers had a mean age of 56 years, never smokers had a mean age of 59 years, and ex smokers had a mean age of 64 years (Table 2). Current smokers had a mean 40 pack-years and median 36 pack-years history of smoking exposure with a range from 1 to 84 pack-years. Ex smokers had a mean 35 pack-years and median 22 pack-years history of smoking exposure with a range of 1 to 228 pack-years, with an interval since smoking cessation characterized by a mean of 21 years and a median of 20 years, and a range of 1 to 66 years. Ethanol consumption of at least one drink per month for 1 year occurred for 444 participants (74%), with a median of 26,533 cumulative lifetime drinks for all participants.

mtDNA content was measured by calculating the ratio of the mitochondrial gene (Cox I or Cox II) to a nuclear gene (β-actin). Average mtDNA levels for current smokers showed a 31% increase for Cox I and 29% increase for Cox II relative to never smokers (pairwise t test, P < 0.001; Fig. 1). In addition, mean mtDNA levels for ex smokers were increased 31% for Cox I and 34% for Cox II relative to never smokers (pairwise t test, P < 0.001).

To characterize the relationship of age and smoking status to mtDNA content, we stratified mean Cox I and Cox II values by age and smoking status. A representative graph of these results for Cox I (Fig. 2) shows elevated mtDNA levels for current smokers and former smokers in every age category, indicating that elevated Cox I and Cox II values were related to smoking exposure independent of age. It is of interest that in the oldest age category, >70 years, the Cox I levels are greater than in any other age category, consistent with age-related increase in mtDNA content. Analysis was done to determine if a dose-response relationship existed between tobacco consumption and mtDNA content; Spearman correlation shows a positive correlation between pack-years of smoking and Cox I values (P < 0.001) and Cox II values (P = 0.001; Supplementary Fig. 2).

To further investigate these relationships, multiple variable regression analysis was done to estimate the relationship between Cox I and Cox II ratios and risk factors and demographic predictors as detailed in Materials and Methods. A four-variable model using the following variables: age, smoking status, ethanol consumption status, and income, was significantly predictive for mtDNA content (P < 0.0001 for Cox I and Cox II, R² = 0.09 and 0.07, respectively). For each of the four variables included in the model, we examined the variable estimates and Ps (Table 3). For Cox I and Cox II, both smoking status (P < 0.0001 for both) and age (P < 0.0001 and P = 0.002, respectively) were significantly associated with increased mtDNA content (Fig. 3). Although less significant, higher income was associated with a lower Cox I and Cox II ratio (P = 0.03 and P = 0.004, respectively). In addition, ethanol consumption was negatively associated with mtDNA content for Cox II (P = 0.04), although this was not associated with Cox I (P = 0.07). It is unclear if this effect is the result of biological activity or a confounding predictor related to ethanol consumption status; nevertheless, the effect reaches statistical significance only for Cox II ratios.

We also investigated the relationship of smoking cessation interval to mtDNA content. With smoking cessation intervals ranging from 1 to 66 years in ex smokers with a median of 20 years, there was no association of smoking cessation interval and mtDNA content as measured by both Cox I and Cox II. In addition, there was no relationship found between mtDNA levels and interval since ethanol intake cessation. It is also noted that there are those individuals with long smoking cessation intervals that still have persistent elevation in mtDNA content, which may indicate increased susceptibility to mitochondrial injury from oxidative exposures or decreased capacity for mtDNA repair after oxidative injury.

We evaluated substantive changes in mtDNA content related to smoking exposure in a population with varied smoking exposure and age range. A dose-dependent association of smoking with mtDNA increase independent of other factors was shown, with persistent alterations in mtDNA content for decades after smoking cessation. The susceptibility of the mtDNA to oxidative and mutagenic damage is well documented. Compared with the nuclear genome, the mitochondrial genome is more susceptible to DNA damage and possesses a mutation rate that has been reported to be as much as two orders of magnitude greater than that of nuclear DNA (3, 30). It is not surprising that exposure to extrinsic mutagenic and oxidative agents, such as cigarette smoke, augments mutagenicity of mtDNA and results in the generalized decline of mitochondrial respiratory chain function (13, 14). An increase in mtDNA content is closely associated with DNA damage and reduced respiratory chain function secondary to oxidative damage (17-22, 25-28). This increase in mitochondrial mass and mtDNA is thought to be a compensatory mechanism for damage to oxidative damage to mtDNA and respiratory chain components.

The increase of mtDNA as a consequence of aging has been suggested in various tissues in smaller studies and is confirmed in this study of exfoliated cells from the upper aerodigestive tract in saliva. Multiple investigators have hypothesized that this increase occurs in response to oxidative damage that occurs with aging and showed that the increase in mitochondrial mass is a response to the decreased function of mitochondria as a result of this injury (12-14). It is possible, therefore, that the genomic injury from other sources, including tobacco smoke, should elicit a similar response, as confirmed by in vitro studies (15, 16). It is noted that in smoking individuals compared with nonsmokers >70 years of age in our study, the elevation of mtDNA is least marked, presumably due to the dominant effect of cumulative age-related mtDNA alteration in very elderly individuals. It is clear that those individuals with greater cumulative smoking exposure tend to be older, and that this would make the effects of these factors also more difficult to distinguish in the elderly population. In addition, inferences from a cross-sectional study are not as strong as those from a prospective study examining mtDNA alteration in smokers and nonsmokers. For example, in this elderly group, it is possible that surviving smokers are a select who are somehow more resistant to untoward effects of cigarette smoking, and this may be manifest is less of an increase in mtDNA content than smokers who died before that age and therefore did not make it into the study.

Higher income was associated with lower mtDNA levels, which may suggest that income level is a confounding predictor in terms of oxidative exposure, whereby those participants with a higher income had less exposure to secondhand smoke or other oxidative exposures unrelated to smoking. Alternatively, a bias in reporting of smoking exposure may be dependent upon reported income level (i.e., that those participants with a lower income level may have underreported smoking exposure). Similarly, the weak, borderline significant negative association with ethanol intake may be due to bias in reporting.

The independence of smoking-related mtDNA alterations is suggested by a dose-dependent positive effect of smoking on mtDNA content and persistence of this elevation when compared with a similar population of nonsmokers. In addition, despite a lower mean age of current smokers compared with never smokers (56 versus 59 years), current smokers had a 30% increase in mtDNA content when compared with never smokers. The association between cigarette smoking and increased mtDNA content was consistently observed in age-stratified analyses, and cigarette smoking persisted as a significant independent predictor of mtDNA elevation even after adjustment for age and other factors.

The elevation of mtDNA content in current smokers may be due to both acute and chronic injury to mitochondria. However, we were able to show persistence of mtDNA elevation in ex smokers with mean cessation intervals of several decades. The persistence of these effects two decades or more after smoking cessation is consistent with long-term genotoxic damage, because short-term damage to cellular and mitochondrial membranes and protein components would cease after replacement of these components with newly synthesized molecules. mtDNA elevation seems to be a long-term compensatory response to chronic injury from tobacco exposure. This is consistent with observations that decreased energy production from damaged mitochondria induces nuclear signals that lead to compensatory replication.

Mitochondria are the site of critical steps in the pathway to apoptosis or programmed cell death. Release of the enzyme cytochrome c and other soluble factors from mitochondria into the cytosol induces apoptosis, and defects in apoptosis have been closely linked to cancer development. In addition, somatic mutations in mtDNA have been described in most cancers, although their role in mitochondrial dysfunction and cancer progression has not been firmly established.

Although the response of mitochondria to oxidative stress is well documented, the mechanism by which mitochondrial mass and mtDNA increases in response to tobacco exposure is not well delineated. The mechanism of response of mitochondria to agents like tobacco smoke may yield important insights into mechanisms of mitochondrial function and its relation to carcinogenesis, particularly related to critical role of mitochondria in the apoptotic pathway. It is possible that an increase in mtDNA content seems to be a marker for tobacco exposure and age-related oxidative damage but may also be an indicator of altered mitochondrial function.

Individual variation in elevation of mtDNA content is shown by the characterization of individuals with elevated mtDNA content despite lack of smoking exposure or extended smoking cessation intervals, and individuals with low mtDNA content despite prolonged smoking exposure and advanced age. This indicates that other factors, including possible hereditary or environmental factors in addition to those we investigated, may influence the net exposure to mitochondrial-damaging agents in tissue and therefore mitochondrial alterations in response to age, tobacco, and other exposures. It is an intriguing concept that mtDNA content may provide a net measure of mtDNA injury that incorporates both degree of exposure to ROS and other DNA-damaging agents as well as the innate constitutive ability of a particular cell to inactivate, prevent, or repair oxidative damage.

Finally, whereas our data are consistent with other in vitro data, a change in β-actin expression as a result of smoking is not formally excluded. In addition, it is possible that cells with increased mitochondrial content are more easily exfoliated, resulting in the association of smoking with increased mtDNA content; however, the persistence of this effect over smoking cessation would argue against this phenomenon.

Relative risk for oral cavity and oropharynx cancer persists up to 15 years after smoking cessation, and elevated risk for lung cancer can be measured up to 30 years after smoking cessation (31). Although cancer risk decreases significantly a decade after smoking cessation, mtDNA content that persists after smoking cessation may reflect in some part the cumulative genotoxic damage induced by smoking. Further studies that track the relationship between mtDNA content, risks for damage from oxidative exposure, and tobacco-related cancer risk in parallel will help elucidate this relationship.