Endogenous versus Exogenous DNA Adducts: Their Role in Carcinogenesis, Epidemiology, and Risk Assessment

Formaldehyde-DNA Adducts and Mutagenicity

Formaldehyde is a very reactive compound, directly targeting diverse nucleophiles of DNA and protein. Typically, formaldehyde can induce DNA adducts including N²-hydroxymethyl-deoxyguanosine (dG), N⁶-hydroxymethyl-deoxyadenosine (dA), and N⁴-hydroxymethyl-deoxycytosine (dC) in vitro. Those DNA adducts are considered to be promutagenic, as the adduction occurs on the amino groups participating in Watson-Crick base pairing. Numerous experiments have demonstrated that formaldehyde exposure induces mutations in a variety of test systems ranging from bacteria to laboratory animals. In Escherichia coli, exposure to 4 mmol/l formaldehyde for 1 h induced large insertions (41%), large deletions (18%), and point mutations (41%) in the xanthine guanine phosphoribosyl transferase gene (Crosby et al., 1988). The point mutations were transversions at GC base pairs, as revealed by DNA sequencing. GC → TA transversions were also found in E. coli Lac+ WP3104P and in S. typhimurium His+ TA7005 after formaldehyde exposure (Ohta et al., 1999, 2000). DNA sequencing demonstrated that AT → CG transitions were the predominant point mutations in formaldehyde-induced human lymphoblast TK6 X-linked hypoxanthine guanine phosphoribosyl-transferase (HPRT) mutants (Liber et al., 1989). Formaldehyde exposure did not induce gene mutations in the HPRT locus in Chinese hamster V79 cells, but other genotoxic effects such as DNA-protein cross-links, sister chromatid exchange, and micronuclei were found (Merk and Speit, 1998). In addition, several point mutations have been identified by DNA sequence analysis of p53 complementary DNA from formaldehyde-induced squamous cell carcinomas in nasal tissues of rats, including ₃₉₆C → A (codon 132), ₃₉₈G → T (codon 133), ₆₃₈G → T (codon 213), ₈₁₂G → A (codon 271), and ₈₄₂G → C (codon 281) (Recio et al., 1992).

Endogenous Formaldehyde-DNA Adducts

As mentioned above, formaldehyde is an essential metabolic intermediate generated in all living cells. The endogenous formaldehyde concentration in human blood is ∼100μM, which induces endogenous formaldehyde-DNA adducts. The number of endogenous N²-hydroxymethyl-dG adducts in rats was measured by LC-MS/MS and found to be ∼1–7 adducts/10⁷ dG (Lu et al., 2010a). The number of endogenous N⁶-hydroxymethyl-dA adducts in rats was also quantified and shown to be ∼1–3 adducts/10⁷ dA across a range of tissues (Cheng et al., 2008; Lu et al., 2010a). The formation of formaldehyde-DNA adducts in humans has also been demonstrated. The adduct number of endogenous N⁶-hydroxymethyl-dA was determined as ∼0.7 adducts/10⁷ adducts in human leukocytes (Wang et al., 2009). We now have additional data on endogenous N²-hydroxymethyl-dG adducts in primates where higher numbers of endogenous adducts have been measured in bone marrow (Moeller et al., 2011).

Relationships of Exogenous/Endogenous Formaldehyde-DNA Adducts

Quantifying the number of adducts caused by exposure to formaldehyde has proven difficult because it is confounded by the presence of a substantial natural background of formaldehyde adducts. Recently, our laboratory has employed [¹³CD₂]-formaldehyde for exposure, coupled with highly sensitive MS analyses (Lu et al., 2010a; Lu et al., 2010b; Moeller et al., 2011). This unique strategy allowed us to distinguish DNA adducts originating from endogenous and exogenous sources of formaldehyde based on the different transitions used in selected reaction monitoring (SRM) mode (Fig. 1). The exogenous DNA adducts have a 3 Da mass increase because of the use of [¹³CD₂]-formaldehyde. Through this approach, we have been able to accurately quantify both endogenous and exogenous adducts induced by formaldehyde. The first molecular dosimetry study was carried out by exposing rats to 10 pmm [¹³CD₂]-formaldehyde for 1 day (6 h/day) or 5 days (6 h/day) (Lu et al., 2010a). The number of exogenous N²-HO¹³CD₂-dG in 1-day and 5-day nasal DNA samples from rats exposed by inhalation to 10 ppm [¹³CD₂]-formaldehyde was 1.28 ± 0.49 and 2.43 ± 0.78 adducts/10⁷ dG, respectively. The number of endogenous N²-HOCH₂-dG adducts was 2.63 ± 0.73 and 2.84 ± 1.13 adducts/10⁷ dG. In addition, 3.95 ± 0.26 and 3.61 ± 0.95 N⁶-HOCH₂-dA endogenous adducts/10⁷ dA were present, but no exogenous dA adducts were detected. The ratios of exogenous versus endogenous dG adducts were 0.57 ± 0.28 and 1.06 ± 0.40 for the 1-day– and 5-day–exposed groups, respectively. This study also examined sites distant from the point of contact but did not detect exogenous adducts in any distant tissues, such as bone marrow, spleen, thymus, and white blood cells, whereas endogenous dG and dA adducts were present in all tissues analyzed. In order to increase the likelihood of measuring [¹³CD₂]-labeled exogenous adducts in sites distant to the point of contact, approximately five times more DNA was analyzed.

Open in a separate window

FIG. 1.

Formation of hydroxymethyl DNA adducts induced by formaldehyde.

More recently, we have expanded our studies on molecular dosimetry of formaldehyde-DNA adducts by exposing rats to 0.7, 2, 5.8, 9.1, and 15.2 ppm [¹³CD₂]-formaldehdye for 1 day (6 h/day), which modeled the exposures in a previous cell proliferation and carcinogenicity study (Monticello et al., 1996). Formaldehyde induced nasal squamous cell carcinomas in a highly nonlinear fashion, with no nasal carcinomas at 0.7 and 2 ppm but 1, 22, and 47% carcinomas at 6, 10, and 15 ppm formaldehyde. The exogenous dG adducts were 0.039 ± 0.019, 0.19 ± 0.08, 1.04 ± 0.24, 2.02 ± 0.43, and 11.15 ± 3.01 for 0.7, 2, 5.8, 9.1, and 15.2 ppm exposure, respectively, which clearly illustrated a nonlinear exposure-response, as demonstrated by the fact that 21-fold exposure increase (0.7–15.2 ppm) induced 287-fold higher amounts of exogenous DNA adducts in rat nasal epithelium (Fig. 2). In addition, the ratios of exogenous versus endogenous were determined to be 0.011 ± 0.001, 0.033 ± 0.006, 0.19 ± 0.04, 0.66 ± 0.17, and 2.78 ± 1.08, respectively, for the [¹³CD₂]-formaldehdye exposures, clearly demonstrating that endogenous DNA adducts predominate at low-dose exposures.

Open in a separate window

FIG. 2.

Molecular dosimetry of N²-hydroxymethyl-dG adducts in rats exposed to formaldehyde.

Moreover, we have carried out a primate study that exposed monkeys to 1.9 or 6.1 ppm isotope-labeled formaldehyde for 2 days (6 h/day). The exogenous dG adducts in nasal DNA of monkeys after the 2-day exposures were similar to that of rats exposed to 2 ppm for 1 day, whereas the exogenous dG adducts were ∼2.5-fold lower in monkeys exposed to 6.1 ppm [¹³CD₂]-formaldehdye for 2 days (6 h/day) compared with rats exposed to 5.8 ppm for 1 day (Fig. 3). These data demonstrated that lower numbers of exogenous DNA adducts were formed in nasal turbinates of nonhuman primates than rats after inhalation exposure to formaldehyde. Moreover, our preliminary data indicate that primates have two- to threefold higher endogenous dG adducts in tissues we analyzed. This substantially higher background of endogenous adducts further reduces the ratio of exogenous/endogenous formaldehyde adducts in primates exposed to low concentrations.

Open in a separate window

FIG. 3.

Formation of exogenous dG adducts in rats and monkeys exposed to 1.9 and 6.1 ppm formaldehyde (1-day exposure for rats and 2-day exposure for monkeys).

Furthermore, we performed a study to determine the half-life of exogenous formaldehyde-DNA adducts. Rats were exposed to 10 ppm [¹³CD₂]-formaldehyde for 1 day (6h) and then sacrificed at 0, 6, 12, 24, 48, or 72 h, respectively. The data showed a rapid loss of nearly half of the adducts during the first 6 h postexposure, followed by a constant rate of loss over the rest of the 72-h experiment (Lu, unpublished data). The initial decrease in [¹³CD₂] adducts is thought to result from cell death because of cytotoxicity, followed by a half-life of ∼94 h. When 10-ppm exposures occur daily, this scenario repeats itself. That explains why 5 days of exposure only doubled the number of exogenous adducts.

VC DNA Adducts and Mutagenicity

VC is a procarcinogen that requires metabolic activation, with the primary enzyme involved being Cytochrome P450 2E1, which converts VC to chloroethylene oxide (CEO). CEO covalently binds to DNA, inducing four DNA adducts (Fig. 4.). The most prevalent DNA adduct is 7-(2-oxoethyl)guanine (OEG), which represents ∼98% of the adducts formed by VC. OEG in not promutagenic, i.e., it does not cause mispairing during DNA replication (Barbin et al., 1985). It is lost primarily by chemical depurination, but some repair by glycosylases is also possible. In either case, abasic sites represent a repair intermediate. With daily exposure, OEG adducts reach steady-state concentrations in ∼7 days and have a T_½ of 4 days (Mutlu, unpublished data). VC also induces three exocyclic etheno DNA adducts 1,N⁶-ethenodeoxyadenosine (ϵdA), 3,N⁴-ethenodeoxycytosine (ϵdC), and N²,3-ethenoguanine (ϵG) (Bartsch et al., 1994; Guengerich and Persmark, 1994). ϵG has been measured using gas chromatography/mass spectrometry (GC/MS) and gas chromatography/high resolution mass spectrometry (GC/HRMS) following derivatization with or without immunoaffinity enrichment (Fedtke et al., 1989, 1990a,b; Morinello et al., 2001, 2002a,b). Most recently, ϵG was analyzed following neutral thermal hydrolysis and HPLC enrichment with LC-MS/MS pseudoSRM (Mutlu et al., 2010). Most research analyzing ϵdA and ϵdC has utilized immunoaffinity-³²P-postlabeling (Nair et al., 1995; Reddy, 1993). Recently, an LC-MS/MS method was applied using either immunoaffinity (Ham et al., 2004) or HPLC (Gao, unpublished data) enrichment.

Open in a separate window

FIG. 4.

Metabolism of VC and formation of DNA adducts from CEO.

All three etheno adducts are promutagenic, as summarized in Table 4. Mutations have been induced in site-specific mutagenesis assays (Basu et al., 1993; Cheng et al., 1991; Shibutani et al., 1996; Singer et al., 1991; Zhang et al., 1995) as well as following in vitro exposure to CEO or chloroacetaldehyde (Guengerich, 1992; Oesch and Doerjer, 1982).The human K-ras gene is activated by a GC → AT base substitution at the second base of codon 13. This mutation was present in 5/6 human VC-induced angiosarcomas (Marion et al., 1996) and is consistent with the known promutagenic properties of ϵG and ϵdC. Hollstein et al. (1994) reported AT → TA mutations in the p53 tumor suppressor gene in 2/4 human hepatic angiosarcomas analyzed in VC workers. These mutations are consistent with the miscoding properties of ϵdA.

Endogenous VC DNA Adducts

Etheno adducts have been identified in unexposed humans and rats in several studies (Barbin et al., 2003; Bartsch et al., 1994; Morinello et al., 2002a; Nair et al., 1995). In vitro studies demonstrated that incubating nucleosides or nucleobases with lipid peroxidation products resulted in the formation of ϵA, ϵC, and ϵG (el Ghissassi et al., 1995; Ham et al., 2000, 2004; Nair et al., 1996). Most recently, endogenous OEG was also identified in [¹³C₂]-VC–exposed rats (Mutlu, unpublished data).

A review of human studies on DNA adducts associated with inflammation and lipid peroxidation was published by Nair et al. (2007). This paper includes data from many studies on ϵdA and ϵdC, including nondiseased tissues. It reported values ranging from not detected to 29 ϵdA and ϵdC adducts per 10⁹ parent bases. Median values of endogenous etheno adducts present in a variety of tissues from 12 human autopsy cases with no known exposure to VC ranged from 1.0 to 2.9 ϵdA/10⁸ dA and 1.9 to 4.8 ϵdC/10⁸ dC (Barbin et al., 2003). The study by Barbin also examined aldehydic lesions such as abasic sites and found 3.7–11.3 aldehydic DNA lesions (ADL)/10⁶ nt. Ethenoguanine has been less well studied, with ∼15 ϵG/10⁸ G in both human liver and colon (Hussain et al., 2000; Swenberg et al., 1999).

Several studies have examined DNA from unexposed rodent tissues and demonstrated the presence of endogenous etheno adducts. Using immunoaffinity/³²P-postlabeling, control rat liver had ∼0.4–5.8 ϵdA/10⁹ dA and 0.6–40 ϵdC/10⁹ dC (Guichard et al., 1996; Nair et al., 2005). Guichard et al. (1996) also reported ∼2 ϵdA/10⁸ dA and 9 ϵdC/10⁸ dC in lung and ∼3 ϵdA/10⁸ dA and 10 ϵdC/10⁸ dC in kidney. Ham et al. (2004) developed an LC-MS/MS assay for ϵdA and reported ∼0.8 ϵdA/10⁸ dA in control and Aag null mouse liver but 0.1 ϵdC/10⁸ dC in control mouse lung compared with 0.7 ϵdC/10⁸ dC in the lung tissues of Aag knockout mice. Morinello et al. (2002b) utilized immunoaffinity GC-HRMS to first show endogenous ϵG in rat liver and brain DNA. Liver had ∼4 ϵG/10⁸ G, whereas brain had ∼5.3 ϵG/10⁸ G in control animals. It is of interest that human tissues tended to have greater amounts of endogenous etheno adducts than rodents.

Relationships of Exogenous/Endogenous VC DNA Adducts

Several studies have compared endogenous DNA adducts identical to those induced by VC (Gao, unpublished data; Mutlu et al., 2010). The most informative studies have utilized 1100 ppm [¹³C₂]-VC exposures for 5 or 20 days (6 h/day) that provided accurate information on both endogenous and exogenous DNA adducts in the same DNA sample (Gao, unpublished data; Morinello et al., 2002a,b; Mutlu et al., 2010). Morinello et al. (2002b) examined adult rat liver and brain DNA and demonstrated that 5.2 ± 1.2 endogenous ϵG/10⁸ G were present in hepatocyte DNA, whereas 55 ± 4.0 [¹³C₂]-ϵG/10⁸ G were formed following 5 days of exposure to 1100 ppm [¹³C₂]-VC (6 h/day) and 110 ± 20 [¹³C₂]-ϵG/10⁸ G were formed following 20 days of exposure. In contrast, 5.4 ± 1.2 endogenous ϵG/10⁸ G were present in brain, but no [¹³C₂]-ϵG was detectable following 5 or 20 days of exposure. This strongly supports the more comprehensive epidemiology studies that did not find a causal association between brain tumors and VC exposure. We hypothesize that the chemical instability of CEO formed in the liver does not allow it to be transported to the brain where it can form DNA adducts, whereas the endogenous adducts are formed from oxidative stress arising within the brain.

Table 5 provides quantitative data for OEG, ϵG, and ϵdA from a study that also exposed rats to 1100 ppm [¹³C₂]-VC for 5 days (6 h/day) (Gao, unpublished data; Mutlu, unpublished data; Mutlu et al., 2010). Results similar to those of Morinello et al. (2002b) were shown in adult liver DNA (4.1 ± 2.8 endogenous ϵG/10⁸ G vs. 18.9 ± 4.9 [¹³C₂]-ϵG/10⁸ G). Using sequential sacrifices 2, 4, and 8 weeks later, they demonstrated that [¹³C₂]-ϵG adducts had little or no repair, with a half-life of 150 days. In contrast, endogenous hepatic OEG was present at 2 ± 1 OEG/10⁶ G at the end of the 5-day exposure and over the next 8 weeks. [¹³C₂]-OEG was present at the end of the 5-day exposure at 104 ± 23 [¹³C₂]-OEG/10⁶ G, 4 ± 3 [¹³C₂]-OEG/10⁶ G 2 weeks later and was not detectable 4 or 8 weeks postexposure. The half-life of [¹³C₂]-OEG was calculated to be 4 days. In our laboratory, Gao developed a nano-UPLC-MS/MS assay for ϵdA to analyze same tissues (Gao, unpublished data). She found 4.9 ± 0.6 endogenous ϵdA/10⁸ dA in the 1100 ppm [¹³C₂]-VC–exposed rat liver compared with 5.1 ± 0.6 [¹³C₂]-ϵdA/10⁸ dA. No [¹³C₂]-ϵdA adducts could be detected in liver samples from rats killed 2, 4, or 8 weeks postexposure. The study design did not allow accurate calculation of the half-life of [¹³C₂]-ϵdA adducts, as the postexposure times were too long. This is consistent with other data suggesting that ϵdA is rapidly repaired with a half-life of ∼24 h (Ham et al., 2004). These [¹³C₂]-VC studies are the only data that discriminate between endogenous and exogenous VC adducts, as all other studies have been measuring a mixture of endogenous and exogenous adducts. It is clear that markedly different half-lives exist for ϵdA, OEG, and ϵG.

Ethylene Oxide DNA Adducts and Mutagenicity

Ethylene oxide is a direct-acting alkylating agent that binds to DNA at several positions (Fig. 5). Numerous groups have conducted research on the DNA and globin adducts of EO, with most data collected during the last 20 years. The advancement of analytical methods such as derivatization and MS, HPLC with fluorescence detection, and the application of ³²P-postlabeling of DNA adducts provided methods with adequate sensitivity to accurately measure the DNA and globin adducts of EO. Törnqvist et al. (1986) developed the Edman degradation method for GC/MS analysis of N-terminal valine adducts of EO in 1986. This method clearly demonstrated the presence of hydroxyethylvaline adducts in unexposed humans and animals. Subsequently, GC/MS and GC/HRMS methods for detecting N7-(2-hydroxyethyl)guanine (7HEG), the major DNA adduct formed by EO, were developed (Föst et al., 1989; Wu et al., 1999), and ³²P-postlabeling of DNA was used to determine EO DNA adducts (Eide et al., 1999). More recently, LC-MS/MS and HPLC coupled with accelerator MS have been used to measure exogenous 7HEG (Marsden et al., 2007; Rusyn et al., 2005). Whereas 7HEG makes up ∼95% of the DNA adducts, EO forms several additional minor DNA adducts including N3-(2-hydroxyethyl)dA, N3-(2-hydroxyethyl)dU, and O⁶-(2-hydroxyethyl)dG (La et al., 2010). 7HEG is not considered a promutagenic DNA adduct but can depurinate and produce an abasic site that could be mutagenic if present during DNA replication. However, inhalation exposures of rats to 100 ppm EO for 4 weeks (6 h/day, 5 days/week) did not result in an increase in abasic sites (Rusyn et al., 2005). When mice were exposed to ethylene or EO by inhalation for 4 weeks (6 h/day, 5 days/week) and HPRT mutant frequencies were determined, there was no increase in mutations at 40, 1000, or 3000 ppm ethylene, which is equivalent to 4.5, 9, or 10 ppm EO. The lowest exposure that increased HPRT mutant frequencies was 50 ppm EO (Swenberg et al., 2008). More recently, EO mutagenicity was evaluated in the pSP189 shuttle vector being replicated in human Ad293 cells (Tompkins et al., 2009). Even when plasmids containing up to 29,000 7HEG adducts/10⁸ nt were evaluated, no other EO DNA adducts could be detected using sensitive LC-MS/MS methods and no increases in mutations were found. Only when very high exposures to EO were used and N1-hydroxyethyldA (N1-HEdA), O⁶-hydroxyethyldG (O⁶-HEdG), and N3-hydroxyethyldU (N3-HEdU) could be measured were significant increases in mutations apparent. The half-lives of these three adducts are much shorter than 7HEG because of active DNA repair pathways. This plethora of data suggests that EO is a relatively weak mutagen.

Open in a separate window

FIG. 5.

Metabolism of ethylene and formation of DNA adducts from EO.

Endogenous Ethylene Oxide DNA Adducts

Endogenous amounts of 7HEG in tissues from unexposed rats and mice range from 1 to 4 7HEG/10⁸ nt (Marsden et al., 2007; Wu et al., 1999), whereas human liver had 58.5 ± 21.4 7HEG/10⁸ nt and human lymphocytes had 48 ± 31 7HEG/10⁸ nt (Wu et al., 1999). Thus, humans appear to have nearly 10-fold higher amounts of endogenous 7HEG than are found in rats and mice. Endogenous 7HEG arises from multiple sources, including the generation of ethylene from methionine oxidation, lipid peroxidation, and bacterial metabolism. Endogenous ethylene is metabolized to EO in the liver, circulates in the blood to expose all tissues, and is exhaled in expired breath. Of considerable interest is the recent study by Marsden (Marsden et al., 2009), who found that ip injections of [¹⁴C₂]-EO not only formed [¹⁴C₂]-labeled exogenous 7HEG but also caused dose-related increases in unlabeled, endogenous 7HEG. They went on to demonstrate that the increase in endogenous 7HEG was because of increased oxidative stress. This may be the result of competition for detoxication pathways between endogenous and injected [¹⁴C₂]-EO. Additional evidence for endogenous 7HEG was provided by the demonstration of 7HEG in human urine (Cushnir et al., 1993; Huang et al., 2008). The amounts of 7HEG in urine were lowest in nonsmokers with no occupational exposure to EO.

Relationships of Exogenous/Endogenous Ethylene Oxide DNA Adducts

Numerous studies have measured 7HEG in various tissues of rats, mice, and humans exposed to EO. What must be clearly understood is that such biomarker measurements actually represent exposure to both endogenous and exogenous EO. To some degree, the source can be ferreted out by comparing exposed individuals to unexposed controls. However, the use of stable isotope exposures, followed by adduct purification and MS, is the only way to measure both endogenous and exogenous 7HEG in DNA at the same time. To date, this has not been reported for EO. As mentioned earlier, Marsden et al. (2007, 2009) exposed mice to [¹⁴C₂]-EO ip at doses of 0, 0.0001, 0.0005, 0.001, 0.005, 0.01, 0.05, or 0.1 mg/kg per day for 3 consecutive days and harvested the tissues 4 h after the last dose. The authors concluded that the responses in liver, spleen, and stomach were dose dependent and not significantly nonlinear. The authors also pointed out that several of the lowest doses induced numbers of adducts that were below the limit of quantitation for the AMS method (∼20–50 7HEG/10¹² nt). To put that in perspective with endogenous 7HEG, this amount is 1000 times lower than endogenous 7HEG observed in control rodent tissues and 10,000 times lower than measured in normal human tissues. The molecular dose of 7HEG created by such low exposures implies that only 1/1000 7HEG adducts in a mouse and only 1/10,000 adducts in a human would arise from exogenous exposure, assuming that rodents and humans form equivalent molecular doses from equal exposures. The endogenous EO adducts outnumber the exogenous adducts by such a vast margin that the exogenous adducts are not likely to be causal for EO-induced mutations or cancer. When looked at from the perspective of the total number of endogenous DNA adducts in a cell, it is clearly implausible.