PAHs are products of the incomplete combustion of organic material, which are widespread environmental pollutants found in air, soil and water. Inhalation of tobacco smoke is a major source of exposure for smokers and to a lesser extent, for passive smokers. The most studied PAH is benzo[a]pyrene, which is metabolically activated to BPDE, which forms a major adduct with DNA at the N2 position of guanine.59
When 32P-postlabelling, fluorescence or immunological methods (with antibodies raised against BPDE-modified DNA) have been used to compare DNA from smokers with that from nonsmokers, the adduct levels detected by this technique are frequently higher in the former, as reviewed earlier.11 In more recent studies this trend has continued, with the phenomenon observed with 32P-postlabelling studies of larynx,33 bladder,60 anal epithelium,61 and with immunohistochemical analysis of liver62 and oral tissue.63 However, in an immunohistochemical analysis of formalin-fixed cervical tissue, there was no correlation between PAH-DNA adducts and smoking status,64 in contrast to earlier reports of smoking-related DNA adducts in human cervix.11 A similar analysis of human placentas also found no difference in extent of DNA damage between smokers and nonsmokers.65
Adduct levels are also consistently higher in lung tissue of smokers compared with nonsmokers.32, 54, 66 Varkonyi et al.67 reported a correlation between adduct levels in blood mononuclear cells and lung tissue from lung cancer patients, and they also reported the detection of adducts chromatographically similar to those derived from hydroquinone and benzenetriol, metabolites of benzene. Gyorffy et al.68 compared DNA adducts in normal lung and tumor tissue by both 32P-postlabelling and immunoassay (with antibodies to BPDE-DNA) and found a statistically significant correlation between levels in the two samples for both smokers and nonsmokers by both methods. Associations between normal lung and blood DNA adducts correlated only for nonsmokers, and the levels in normal lung did not correlate between the two methods, even though levels were significantly elevated in smokers by both methods of analysis. These discrepancies are not so surprising, given the very different ways of detecting DNA adducts inherent in the 32P-postlabelling and immunoassay methods, and continuing uncertainty about the nature of the smoking-related adducts detected by 32P-postlabelling.54
Several studies investigating environmental exposures to carcinogens in general populations have demonstrated the influence of smoking, as well as other lifestyle factors, including dietary habits and air pollution, on adduct levels in peripheral blood cells.69–75 In other studies, a significant correlation with smoking was not apparent.76–78 With several possible sources of adduct-forming compounds evident from these studies, it is perhaps not surprising that the influence of smoking on adduct levels is not observed in some populations, but is in others.
Even though the association between smoking and levels of bulky adducts in peripheral blood is inconsistent, high-adduct levels appear to be predictive of lung and bladder cancer risk.79, 80 A pooled analysis of studies of adducts in blood cells found the sources of interindividual variation largely unexplained.81
Some studies have suggested that higher levels of PAH-DNA adducts in peripheral blood cells is associated with increased risk of breast cancer,82 but the evidence that this is related to smoking is also not consistent.83
Several studies have found that increased levels of DNA damage and DNA adducts are associated with sperm motility and impaired male fertility, but there are inconsistencies in the evidence for a link with smoking. For example, in one study, levels of bulky DNA adducts detected by 32P-postlabelling showed a significant inverse correlation with sperm concentration and motility, but bulky adducts were only 1.2-fold higher in smokers, which was not statistically significant.84 In another study, occupational exposure to PAHs, but not smoking, was significantly associated with higher levels of PAH-DNA adducts detected by immunofluorescence.85 In contrast, another study has reported significantly higher levels of BPDE-DNA adducts, detected by fluorescence imaging, in the sperm of smokers than in nonsmokers.86
Aromatic and heterocyclic aromatic amines
4-Aminobiphenyl (4-ABP) is an aromatic amine with a variety of environmental and occupational sources of exposure, including tobacco smoke, and it is a known bladder carcinogen in both experimental animals and humans. Its major DNA adduct is formed at the C-8 position of guanine. Sensitive and specific mass spectrometry and immunologic methods for detecting 4-ABP-DNA adducts have been applied to a number of human tissues (see Table 1). In breast tissue, adduct levels in tumor-adjacent tissue, but not in tumor tissue itself, correlated with smoking status.87 In bladder biopsies from cancer patients, adduct levels were associated with current smoking and with tumor grade.88 However, in a small study of human pancreas (n = 12), there was no correlation of adducts with smoking, age or gender.28 A more recent study has called into question the conclusions regarding the presence of 4-ABP-DNA adducts in breast tissue, based on immunohistochemistry. Using a specific LC-MS method, Gu et al.89 failed to detect the presence of 4-ABP-DNA adducts in any of 70 breast biopsies of tumor-adjacent tissue. Although there was no systematic enquiry into the smoking status of these patients, 37 were described as never smokers, 13 as former smokers and 4 as current smokers.
Heterocyclic amines, formed in cooked meat and also present, in some cases, in tobacco smoke, also form adducts predominantly with the C-8 position of guanine in DNA. Bessette et al.90 investigated the formation of a number of adducts formed by these compounds in the saliva of 37 human volunteers. The C-8-dGuo adduct of 2-amino-1-methyl-6-phenylimidazo[4,4-b]pyridine (PhIP) was detected in 13/29 ever smokers (former and current smokers) and 2/8 never smokers. In contrast, adducts formed by 2-amino-9H-pyrido[2,3-b]indole (AαC) and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx) were detected only in the saliva of three current smokers, and 4-ABP-DNA adducts were also detected in two current smokers. However, in another study by these investigators,91 in which the number of subjects was not specified, adducts formed by these carcinogens were not detectable in buccal cell DNA from smokers.
Acrolein (2-propenal), a highly reactive α,β-unsaturated aldehyde, occurs widely in cooked foods and in the environment. Its ubiquity is attributed to incomplete combustion of petrol, wood, and plastic, to smoking tobacco, frying of foods in oils, endogenous lipid peroxidation, and to endogenous polyamine metabolism. Cigarette smoke contains about 180 μg of acrolein per cigarette4 and is considered to account for a large proportion of total human exposure to acrolein.92 Acrolein is mutagenic in bacteria and in cultured human cells. There is inadequate evidence for its human carcinogenicity.93 Chen94 has reviewed the use of 32P-postlabelling- and MS-based methods for the analyses of acrolein-derived DNA adducts in human tissues.
Acrolein reacts with deoxyguanosine in DNA to form two pairs of stereoisomers of cyclic 1,N2-propanodeoxyguanosine adducts (Acr-dGuo). Of the α-OH-Acr-dGuo and γ-OH-Acr-dGuo isomers, the α-isomer is particularly mutagenic in human cells and induces predominantly G to T transversions.39
It is noteworthy that in an earlier study using 32P-postlabelling /HPLC, Nath et al.38 found that the mean Acr-dGuo levels in gingival tissue DNA from 11 smokers (4 male and 7 female) was significantly higher than that in 12 nonsmokers (8 male and 4 female) (1.36 ± 0.90 μmol/mol guanine in smokers versus 0.46 ± 0.26 μmol Acr-dGuo/mol guanine in nonsmokers; p = 0.003).
Because Acr-dGuo adducts, like PAH-DNA adducts, act to induce predominantly G to T transversions in human cells, Feng et al.95 hypothesized that Acr-dGuo adducts could be responsible for TP53 mutations in cigarette-related lung cancer. They mapped the distribution of Acr-dGuo adducts at the sequence level in the TP53 gene of lung cells and found that the Acr-dGuo binding pattern is similar to the TP53 mutational pattern in human lung cancer. They also found that acrolein greatly reduces the DNA repair capacity for damage induced by BPDE. They suggested that acrolein is a major etiological agent for lung cancer caused by cigarette smoke and that it contributes to lung carcinogenesis through two detrimental effects: DNA damage and inhibition of DNA repair.
To determine whether Acr-dGuo adducts could be detected in human lung, Zhang et al.39 developed a specific liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) method for the quantitative analysis of these DNA adducts. Thirty DNA samples from normal lung tissue obtained at surgery were analyzed, and Acr-dGuo adducts were detected in all samples. Both α-OH- and γ-OH-Acr-dGuo were observed in most of the samples; total adduct concentrations ranged from 16 to 209 adducts/109 nucleotides. However there was no difference in adduct levels between confirmed current smokers (N = 5) and nonsmokers (N = 9), nor was there any relationship of adduct levels to self-reported time since cessation of smoking, gender, age, urinary nicotine and cotinine.
Both these adducts were reported to be detectable by mass spectrometry in buccal cell DNA of smokers.91 In a later study96 Acr-dGuo adducts were analyzed in peripheral blood leukocytes from 25 smokers and 25 nonsmokers whose smoking history was known and whose smoking status was confirmed by exhaled carbon monoxide. The predominant isomer in all samples was γ-OH-Acr-dGuo, while α-OH-Acr-dGuo was detected in only three subjects. Again, there was no significant difference between the total Acr-dGuo levels in smokers (7.4 ± 3.4 adducts/109 nucleotides) and nonsmokers (9.8 ± 5.5 adducts/109 nucleotides). However, the mean level of γ-OH-Acr-dGuo was significantly higher in nonsmokers (9.7 ± 5.5) than in smokers (7.0 ± 2.5). Based on this and other studies of acrolein-derived mercapturic acids in the urine of smokers and nonsmokers, the authors concluded that glutathione conjugation effectively removes acrolein from external exposures such as cigarette smoking, protecting leukocyte DNA from damage.
Acetaldehyde is one of the most prevalent carcinogens in tobacco smoke, and also found widely in the environment, for example in food and fuel combustion products. It is also formed endogenously during the catabolism of threonine and metabolism of ethanol. It reacts with DNA to form primarily N2-ethylidene-dGuo, which when DNA is subjected to enzyme hydrolysis in the presence of NaBH3CN is converted to N2-ethyl-dGuo, detectable by LC-ESI-MS/MS-SRM.37 N2-ethyl-dGuo was detected thus in the blood leukocyte DNA of 25 smokers at levels of 1,310 ± 720 (range: 124–7,700) and 1,120 ± 1,140 (range: 138–5,760) fmol/μmol dGuo at two baseline points.94
Theoretically, subsequent reaction of N2-ethylidene-dGuo in DNA would give rise to 1,N2-propano-dGuo (1,N2-εdGuo, i.e., Acr-dGuo). This adduct has been detected in human cells exposed in vitro to micromolar concentrations of acetaldehyde. This raises the possibility that such a “double adduct” could play a role in the genotoxicity of acetaldehyde, and that this carcinogen could, in part, account for the presence of 1,N2-εdGuo (Acr-dGuo) in human DNA.97
There is evidence for a direct-acting ethylating agent in tobacco smoke, whose structure is as yet unknown.98 In a study of nontumorous lung tissue from lung cancer patients, O4-ethylthymidine was detectable in 10 of 13 smokers, but only 3 of the 11 nonsmokers.32 In another small study, the adduct was detected in cells obtained by sputum induction from two of four smokers, but from none of three nonsmokers.40 Both these studies used an immunoenriched 32P-postlabelling method. In a third study using this technique, Anna et al.99 found the adduct to be 1.7-fold higher in normal lung tissue, from lung cancer patients at surgery, of smokers than of long-term ex-smokers. There was no correlation with levels of bulky adducts determined by 32P-postlabelling, in contrast to the study by Godschalk et al.,32 where a correlation between the two adduct types was observed (R = 0.65, p < 0.05). A mass spectrometry method has been developed to measure 7-ethylguanine in DNA; when leukocyte DNA from 30 smokers and 30 nonsmokers was compared, the difference in adduct levels was not significant (49.6 ± 43.3 vs. 41.3 ± 43.3).41
The representative tobacco nitrosamine NNK is metabolically activated to 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL). Both NNK and NNAL convert to methanediazohydroxide, which reacts with DNA to form the methyl adducts O6-methyl-dGuo, 7-methyl-dGuo and O4-methyl-dThd. Such adducts can also be formed by many methylating agents, but a more specific pathway leading to adducts from both NNK and NNN is one that yields 4-hydroxy-1-(3-pyridyl)-1-butanone (HPB) on neutral thermal or acid hydrolysis of DNA. Such adducts, including O6-, 7-, N2 dGuo, O2-dThD and O2-dCyd, are formed at multiple sites in DNA.18
HPB-releasing DNA adducts were at significantly higher levels in lung DNA from 21 self-reported smokers (404 ± 258 fmol/mg DNA) than in 11 self-reported nonsmokers (59 ± 56 fmol/mg).100 However adduct levels in esophagus did not differ significantly between smokers and nonsmokers, leading to speculation that the alkaloid myosine, which occurs in many food items as well as in tobacco, might also be a source of HPB-releasing adducts.101, 102 in another study, HPB-releasing adducts were detectable in only 6 of 58 samples of pancreatic DNA (four of them from smokers).103
Formaldehyde is an industrial chemical with a wide array of uses and is produced in high volumes. It occurs in humans endogenously, is rapidly metabolized and is also formed through the metabolism of many xenobiotic agents. Formaldehyde is ubiquitous in the environment and has been detected in indoor and outdoor air; in treated drinking water, bottled drinking water, surface water, and groundwater; on land and in the soil; and in numerous types of food.104 It is also a constituent of tobacco smoke; based on an analysis of 48 cigarette brands under ISO conditions, it has been established that mainstream cigarette smoke contains 14–28 μg/cigarette of formaldehyde.105 Formaldehyde is classified as a human carcinogen and is genotoxic and forms DNA adducts and crosslinks.106
In a study of leukocyte DNA samples from 32 smokers and 30 nonsmokers, Wang et al.36 used LC-ESI-MS/MS to quantify the formaldehyde-DNA adduct N6-hydroxymethyldeoxyadenosine (N6-HOMe-dAdo). N6-HOMe-dAdo was detected in 29 of 32 smoker samples (mean ± SD, 179 ± −205 fmol/μmol dAdo). In contrast, it was detected in only 7 of 30 nonsmoker samples (15.5 ± 33.8 fmol/μmol dAdo; p < 0.001). The authors speculated on the source of the elevated levels of formaldehyde-DNA adducts in smokers and suggested that inhalation of formaldehyde in cigarette smoke as the simplest explanation.
Oxidative damage to DNA
Exposure to tobacco smoke can provoke the formation of reactive oxygen species (ROS) that can oxidize DNA, causing single-strand and double-strand breaks, and generating genotoxic electrophiles by lipid peroxidation of polyunsaturated fatty acids. These highly reactive species include 2,3-epoxy-4-hydroxynonanol, and can form a variety of mutagenic DNA adducts, including the exocyclic adducts 1,N6-ethenoadenine (εAde), 3,N4-ethenocytosine (εCyt) and N2,3-ethenoguanine (N2,3-εGua).107, 108 Malondialdehyde (MDA) is another reactive electrophile generated by lipid peroxidation and forms a cyclic pyrimidopurinone N-1,N2-malondialdehyde-2′-deoxyguanosine DNA adduct (M1dG). Another consequence of the attack of DNA by ROS is the formation of 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxodG) which is formed by hydroxylation of deoxyguanosine in DNA.
In a study of several types of DNA adduct in tumor-adjacent but uninvolved lung tissues of 13 smoking and 11 nonsmoking obtained at surgery from lung cancer patients, levels of εAde and εCyt, determined by immunoenriched 32P-postlabelling, did not differ between smokers and nonsmokers, but large interindividual variations were observed (80- and 250-fold differences for εAde and εCyt, respectively).32
Whether levels of 8-oxodG are higher in smokers than in nonsmokers remains uncertain. For example, Singh et al.,31 using liquid chromatography–tandem mass spectrometry selected reaction monitoring (LC–MS/MS SRM), found no significant differences in 8-oxodG adduct levels in lymphocyte DNA from individuals, from three European countries, that could be attributed to smoking status. Earlier studies (reviewed by Singh et al.) have produced conflicting results, but several studies support the view that there is no consistent and reliable evidence that 8-oxodG levels are higher in smokers than nonsmokers.
There is also a lack of consistency in the case of M1dG adducts in smokers compared to nonsmokers. However, this may be related to the diversity of tissues and organs that have been examined.
Using an immunohistochemical method, employing a monoclonal antibody specific for MDA-DNA adducts, Zhang et al.35 observed a mean 1.3-fold (p = 0.02) excess of adduct-related staining in oral mucosa cells from 25 smokers compared to 25 age-, race- and sex-matched nonsmokers.
In a study of laryngeal biopsies, levels of malondialdehyde-DNA adducts with chromatographic properties consistent with exocyclic MDA-deoxyadenosine (MDA-dA) and MDA-deoxyguanosine (MDA-dG) were detected in laryngeal biopsies from 30 patients, 13 with larynx cancer and 17 controls.33 Of the 30, 9 were nonsmokers and 21 were smokers. Mean total MDA-DNA adducts, MDA-dA and MDA-dG adducts per 108 nucleotides in smokers were higher than in nonsmokers (6.7 ± 1.0 vs. 4.0 ± 1.4, p < 0.05; 3.3 ± 0.7 vs. 2.1 ± 1.0, p < 0.05; 3.4 ± 0.4 vs. 1.9 ± 0.5, p = 0.078, respectively) after controlling for age, sex, larynx cancer, residence and alcohol consumption. There was evidence of a positive dose-response relationship between cigarette consumption and adduct levels but this did not achieve statistical significance.
Munnia et al.34 measured the relationship between bronchial M1dG adducts (using 32P-postlabelling) and tobacco smoking, cancer status, and selected polymorphisms in 43 subjects undergoing diagnostic bronchoscopic examination. M1dG-DNA adducts were higher in current smokers than in never smokers [frequency ratio (FR) = 1.51, 95% confidence interval (CI): 1.01–2.26]. MDA-DNA adducts were also higher in lung cancer cases with respect to controls, but only in smokers (FR = 1.70, 95% CI: 1.16–2.51). Subjects with GA and AA cyclin D1 (CCND1) genotypes showed higher levels of MDA-DNA adducts than those with the wild-type genotype [FR = 1.51 (1.04–2.20) and 1.45 (1.02–2.07)].
Singh et al.31 found no significant difference between smokers and nonsmokers in the levels of lymphocyte M1dG adducts measured by an immunoslot blot assay employing a murine M1dG monoclonal primary antibody. The adduct level (per 108 nucleotides) in smokers was 32.8 ± 30.5 and in nonsmokers, 30.4 ± 23.8.
In a cross-sectional study comparing the prevalence of M1dG adducts, detected by 32P-postlabelling, in the peripheral leukocytes of groups of subjects experiencing various degrees of air pollution in Thailand, Peluso et al.109 concluded that “formation of DNA damage [M1dG adducts] tended to be associated with tobacco smoking, but without reaching statistical significance.” Adduct levels per 108 nucleotides were 3.7 ± 0.4 in nonsmokers, 4.2 ± 0.7 in ex-smokers and 4.8 ± 0.4 in current smokers.
In a study of M1dG adduct levels, analyzed by 32P-postlabelling, in leukocytes of pathologists occupationally exposed to formaldehyde and nonexposed control subjects, Bono et al.110 found no significant differences in adduct levels per 108 nucleotides between nonsmokers (n = 27, 3.8 ± 0.9) and smokers (n = 13, 4.5 ± 1.3, p = 0.494).
Using 32P-postlabelling of breast fine-needle aspirate samples, Peluso et al.111 measured M1dG adduct levels in 22 patients with breast cancer, at different clinical stages, and 13 controls. Multivariate analysis revealed that increasing severity of breast cancer was significantly and positively associated with M1dG adduct levels but that there were no significant effects of age and smoking habit on adduct levels (MR (mean ratio) = 1.58, 95% CI: 0.92–2.72 and MR = 1.68, 95% CI: 0.88–3.20, respectively).