Metabolic Activation and DNA Interactions of Carcinogenic N-Nitrosamines to Which Humans Are Commonly Exposed

Carcinogenic N-nitrosamine contamination in certain drugs has recently caused great concern and the attention of regulatory agencies. These carcinogens—widely detectable in relatively low levels in food, water, cosmetics, and drugs—are well-established and powerful animal carcinogens. The electrophiles resulting from the cytochrome P450-mediated metabolism of N-nitrosamines can readily react with DNA and form covalent addition products (DNA adducts) that play a central role in carcinogenesis if not repaired. In this review, we aim to provide a comprehensive and updated review of progress on the metabolic activation and DNA interactions of 10 carcinogenic N-nitrosamines to which humans are commonly exposed. Certain DNA adducts such as O6-methylguanine with established miscoding properties play central roles in the cancer induction process, whereas others have been linked to the high incidence of certain types of cancers. We hope the data summarized here will help researchers gain a better understanding of the bioactivation and DNA interactions of these 10 carcinogenic N-nitrosamines and facilitate further research on their toxicologic and carcinogenic properties.


Introduction
Peter Magee and John Barnes reported in 1956 that N-nitrosodimethylamine 1 (NDMA, Figure 1), a simple water-soluble compound with only 11 atoms, readily induced liver tumors in rats [1]. This was remarkable because most carcinogenesis studies at the time had been performed with higher molecular weight non-water-soluble compounds such as polycyclic aromatic hydrocarbons. Sakshaug et al., and Ender and Ceh made the connection between NDMA formation from nitrite-treated herring meal and liver toxicity in farm animals and provided evidence for the occurrence of this carcinogen in smoked fish and meat [2,3]. Thus, concern arose that N-nitrosamines in food treated with nitrite could be a carcinogenic hazard to humans. Sen and colleagues demonstrated the presence of N-nitrosamines in cured meat products [4] and this was followed by a surge in interest in the N-nitrosamine contamination of foods, which persists to the present.
In the meantime, cancer researchers investigated the powerful carcinogenic properties of multiple structurally diverse N-nitrosamines. Druckrey and co-workers and Lijinsky and colleagues demonstrated the carcinogenicity and frequent organoselectivity of multiple N-nitrosamines [5,6]. A review published in 1984 by Preussmann and Stewart summarizes the carcinogenicity of more than 200 N-nitrosamines, which commonly affect specific organs in laboratory animals, independent of the route of administration [7]. A book by Lijinsky also summarizes the extensive carcinogenicity data [5]. Bogovski and Bogovski published a summary of the carcinogenic activity of nitroso compounds in different animal species; NDMA induced tumors in 16 different animal species and NDEA in 26 ranging from rainbow trout to cynomolgus monkey [8]. Such a remarkable database hardly exists for any other type of carcinogen. Thus, there was intense interest in the possible role of N-nitrosamines in human cancer in the latter part of the 20th century. The International Agency for Research on Cancer (IARC) held a series of regular meetings dedicated to this subject between 1966 and 1991. Figure 2 shows the participants at the 1983 meeting in Banff, Canada, a clear indication of the high interest in the topic. Fortunately, methods were developed to decrease levels of N-nitrosamines in food, beer, and other consumer products and interest in the topic waned somewhat in the early part of this century. Recently, however, concern regarding N-nitrosamine contamination of consumer products has re-emerged as they were found in certain pharmaceutical agents and drinking water [9]. It is worth noting that the carcinogenic potency of most N-nitrosamines is so great that they are excluded from the widely used Threshold of Toxicological Concern concept in the risk assessment of exposure to potential carcinogens in food and other consumer products [10]. Thus, there was intense interest in the possible role of N-nitrosamines in human cancer in the latter part of the 20th century. The International Agency for Research on Cancer (IARC) held a series of regular meetings dedicated to this subject between 1966 and 1991. Figure 2 shows the participants at the 1983 meeting in Banff, Canada, a clear indication of the high interest in the topic. Fortunately, methods were developed to decrease levels of N-nitrosamines in food, beer, and other consumer products and interest in the topic waned somewhat in the early part of this century. Recently, however, concern regarding N-nitrosamine contamination of consumer products has re-emerged as they were found in certain pharmaceutical agents and drinking water [9]. It is worth noting that the carcinogenic potency of most N-nitrosamines is so great that they are excluded from the widely used Threshold of Toxicological Concern concept in the risk assessment of exposure to potential carcinogens in food and other consumer products [10].
All N-nitrosamines require metabolism to exert their carcinogenic properties. The electrophiles produced in these simple metabolic pathways, generally catalyzed by cytochrome P450 enzymes, readily alkylate DNA initiating the carcinogenic process. These critical pathways are the subject of this review of the 10 N-nitrosamines illustrated in Figure 1.
primarily targeting the liver but it can also induce kidney tumors when given at high doses [7]. The cancer risk of NDMA exceeds many known strong carcinogens including asbestos, benzo[a]pyrene, and polychlorinated biphenyls [9]. A linear dose-response relationship was observed at low dose rates (lower than 1 ppm) in a study with 4080 rats. No indicated safe threshold concentration was observed for NDMA in the rats [21,22]. NDMA is tumorigenic in multiple animal models including Syrian golden hamsters, mastomys, guinea pigs, rabbits, and ducks [7]. The toxicity of NDMA in humans has been demonstrated by some unfortunate poisonings [20]. The currently available data suggest that the IARC classification of NDMA should be re-evaluated. The U.S. government classifies NDMA as "reasonably anticipated to be a human carcinogen" [23].

Metabolism
The bioactivation of NDMA to reactive intermediates is primarily catalyzed by P450 2E1 in human liver microsomes [24]. The oxidation of the methyl group (α-methyl hydroxylation) leads to α-hydroxyNDMA (11, Scheme 1), an unstable and mutagenic intermediate that spontaneously decomposes generating two reactive species-formaldehyde 12 and methyl diazohydroxide 14 [24,25]. Formaldehyde can be oxidized sequentially producing formic acid 13 and CO2 [26,27]; methyl diazohydroxide will spontaneously form the highly electrophilic methyldiazonium ion 15 and alkylate DNA [28] or be solvolyzed to methanol [29]. The evolution of 15 N2 from 15 N-labeled NDMA metabolism suggested that approximately 33-67% of NDMA was metabolized in vitro via α-methyl hydroxylation [30][31][32]. A deuterium isotope effect has been observed with a slight reduction of the NDMA metabolic oxidation rate for NDMA-d6 with the ratio of VH/VD = ~1.2. The effect was more prominent based on the concentrations of DNA adducts formed in the liver and kidneys after oral administration [33]. The methylating species or its immediate precursor 11, although highly reactive, were sufficiently stable to pass out of rat hepatocytes [34]. The enzymatic denitrosation of NDMA has also been observed in liver microsomes and is considered a detoxification pathway. The major denitrosation product of NDMA is methylamine 17, formed via the proposed intermediate methyliminium ion 16; the other product is formaldehyde [35][36][37]. Scheme 1. NDMA metabolism by α-hydroxylation and denitrosation.

Methyl DNA Adducts Formed by NDMA Metabolism
The mutagenicity and genotoxicity of NDMA are well-established [38]. It alkylates DNA and protein via the two reactive intermediates-methyldiazonium ion 15 and formaldehyde 12. Methyl DNA adducts formed by the methyldiazonium ion are considered to play a major role in carcinogenesis by NDMA [39]. Formaldehyde, on the other hand, can also form DNA adducts such as cross-links or hydroxymethylene adducts [40]. Scheme 1. NDMA metabolism by α-hydroxylation and denitrosation.

Methyl DNA Adducts Formed by NDMA Metabolism
The mutagenicity and genotoxicity of NDMA are well-established [38]. It alkylates DNA and protein via the two reactive intermediates-methyldiazonium ion 15 and formaldehyde 12. Methyl DNA adducts formed by the methyldiazonium ion are considered to play a major role in carcinogenesis by NDMA [39]. Formaldehyde, on the other hand, can also form DNA adducts such as cross-links or hydroxymethylene adducts [40].
A comprehensive analysis of methyl DNA adducts in liver DNA was conducted in rats treated with NDMA by a single i.p. dose of 10 mg/kg [41]. Adducts were detected after deglycosylation. Methyl DNA adducts (  (31, 33 and 35), and methyl DNA phosphate adduct 36, were formed 2 h after administration. The preferential methylation sites in liver DNA were Gua-N7, Gua-O 6 , and Ade-N3. The half-lives of methyl DNA adducts were relatively short, ranging from 4 to 17 h for the methyl base adducts and up to 7 days for the methyl phosphate adduct Tp(Me)T 37 [41]. The adduct profile in NDMA-exposed Syrian golden hamsters was different from that in the rats. O 6 -Me-Gua was the most persistent adduct, whereas N7-Me-Gua was extensively excised, especially in the liver. Other minor methyl adducts including N1-, N3-, N7-Me-Ade, and N3-Me-Gua were also detectable in most hamster tissues [42]. rats treated with NDMA by a single i.p. dose of 10 mg/kg [41]. Adducts were detected after deglycosylation. Methyl DNA adducts (Figure 3), including N3-and N7-Me-Ade (19 and 20), O 2 -and N3-Me-Cyt (21 and 22), N3-, O 6 -, N7-Me-Gua (23, 24 and 27) and Me-Fapy-Gua (29), O 2 -, N3-, and O 4 -Me-Thy (31, 33 and 35), and methyl DNA phosphate adduct 36, were formed 2 h after administration. The preferential methylation sites in liver DNA were Gua-N7, Gua-O 6 , and Ade-N3. The half-lives of methyl DNA adducts were relatively short, ranging from 4 to 17 h for the methyl base adducts and up to 7 days for the methyl phosphate adduct Tp(Me)T 37 [41]. The adduct profile in NDMA-exposed Syrian golden hamsters was different from that in the rats. O 6 -Me-Gua was the most persistent adduct, whereas N7-Me-Gua was extensively excised, especially in the liver. Other minor methyl adducts including N1-, N3-, N7-Me-Ade, and N3-Me-Gua were also detectable in most hamster tissues [42]. In the rat study, after a 2 mg/kg dose of NDMA by stomach tubing daily (workdays only) for up to 24 weeks, N7-Me-Gua 27 predominated in the liver, exceeding that in the kidneys and lungs [43] by approximately 16 times. This is consistent with the liver being the primary tumor site observed in carcinogenicity studies with low doses of NDMA. The other most commonly quantified adduct, O 6 -Me-Gua 24, accounted for 6.7% and 12.0% of N7-Me-Gua, 4 h and 24 h after a single s.c. dose of 0.055 mmol/kg NDMA, respectively [44]. Although the levels of N7-Me-Gua remained constant in the rat liver, increased excision of O 6 -Me-Gua was observed after chronic administration of NDMA [45,46]. This was due to the enhanced activity of O 6 -alkylguanine-DNA alkyltransferease (AGT) in the rat liver after repeated low-dose exposure to this carcinogen [47][48][49]. However, a clear interspecies difference has been noted, with a partially depleted AGT activity observed in mice chronically treated with low doses of NDMA [49]. The formation and persistence of O 6 -Me-Gua were also greatly affected by the co-treatment with ethanol, a known inhibitor of NDMA hepatic metabolism [50]. A remarkable 10-fold increase was observed in mammary gland DNA adducts in rats co-exposed to NDMA and ethanol. This is probably due In the rat study, after a 2 mg/kg dose of NDMA by stomach tubing daily (workdays only) for up to 24 weeks, N7-Me-Gua 27 predominated in the liver, exceeding that in the kidneys and lungs [43] by approximately 16 times. This is consistent with the liver being the primary tumor site observed in carcinogenicity studies with low doses of NDMA. The other most commonly quantified adduct, O 6 -Me-Gua 24, accounted for 6.7% and 12.0% of N7-Me-Gua, 4 h and 24 h after a single s.c. dose of 0.055 mmol/kg NDMA, respectively [44]. Although the levels of N7-Me-Gua remained constant in the rat liver, increased excision of O 6 -Me-Gua was observed after chronic administration of NDMA [45,46]. This was due to the enhanced activity of O 6 -alkylguanine-DNA alkyltransferease (AGT) in the rat liver after repeated low-dose exposure to this carcinogen [47][48][49]. However, a clear interspecies difference has been noted, with a partially depleted AGT activity observed in mice chronically treated with low doses of NDMA [49]. The formation and persistence of O 6 -Me-Gua were also greatly affected by the co-treatment with ethanol, a known inhibitor of NDMA hepatic metabolism [50]. A remarkable 10-fold increase was observed in mammary gland DNA adducts in rats co-exposed to NDMA and ethanol. This is probably due to a reduced hepatic first-pass effect, which also resulted in slightly decreased concentrations of the liver DNA adducts [51].
3.1.4. Mutagenicity and Genotoxicity of Methyl DNA Adducts DNA methylation, especially as related to the activation of proto-oncogenes, has been linked to the induction of pulmonary neoplasia by N-nitrosomethylamines [52]. There is convincing evidence that the formation and removal of O 6 -Me-Gua 24 ( Figure 3) in DNA is related to the risk of tumor induction in different organs [53,54]. O 6 -Me-Gua, formed from alkylating agents such as NDMA, is possibly one of the most extensively studied 7 of 47 of all DNA adducts. Its persistence in specific rat tissues where tumors developed after treatment with N-methyl-N-nitrosourea was established in early studies, supporting the hypothesis that it caused miscoding in DNA [55]. Elegant experiments by the Essigmann group demonstrated unequivocally that O 6 -Me-Gua causes G to A transition mutations, which is consistent with G-A transition mutations in the ras oncogene in laboratory animals treated with NDMA and other methylating carcinogens [56,57]. The repair enzyme AGT can remove the methyl group or other alkyl groups, returning the DNA to its unmodified structure [58].
Methylated thymidines such as O 4 -Me-Thd 34 ( Figure 3) are poorly repaired [59]. The persistence of the Thd adducts likely contributes to the mutagenicity and carcinogenicity of NDMA. Two adducts, O 2 -Me-Thd 30 and O 4 -Me-Thd, blocked DNA synthesis and induced A to G transitions mediated by human DNA polymerase κ (pol κ) [60]. This result echoes that of Singer et al. published in 1983 [61].
The methyl Fapy-dGuo adduct 28 blocked eukaryotic high-fidelity polymerases but can be efficiently bypassed by translesion polymerases. Misreplication products, accounting for 8-29% of total extension products, included C to T/G/A mutations and an interesting one-nucleotide deletion [62].
Wang et al. first investigated the mutagenicity of methyl DNA phosphate adduct 36 ( Figure 3) [63]. Due to the stereochemistry of the phosphorus atom, two diastereomers are formed upon DNA phosphate methylation. The (S)-isomer can be efficiently bypassed, causing TT to GT and GC mutations in the flanking TT dinucleotide site. This mutation was induced in an AGT (also termed as Ada)-dependent manner. On the other hand, the (R)-isomer moderately blocked the replication of the DNA synthesis [63].

Methyl DNA Adducts in Human Tissues
The major DNA adducts caused by NDMA incubated with cultured esophagus from human patients were O 6 -Me-Gua 24 and N7-Me-Gua 27, in a ratio of 0.3. A 10-fold interindividual variation was observed for adduct levels due to NDMA in human esophageal DNA [64]. In the liver DNA of an NDMA-poisoning victim, O 6 -Me-Gua and N7-Me-Gua were detected at levels of 273-317 and 1363-1373 µmol/mol Gua, respectively. These adducts were not detected in the liver and kidney DNA from unrelated cases [65].
Methyl DNA adducts have been widely detected in human tissues likely due to exposure to various methylating agents that can originate from multiple sources both endogenously and exogenously [66][67][68]. For example, Foiles et al. developed a monoclonal antibody for O 6 -Me-Gua and used it in a competitive enzyme-linked immunosorbent assay coupled with HPLC resulting in the first identification of O 6 -Me-Gua in human placental DNA [69]. In Japanese donors, this adduct was detected in both leukocyte and liver DNA. In contrast, O 4 -Me-Thy 35 was only detected in the liver DNA [70]. Similarly, in the liver and leukocyte DNA of humans who were not exposed to known alkylating agents, O 6 -Me-Gua and O 4 -Me-Thy were detected in nearly all the liver DNA samples, at levels of 1.1-6.7 and 0.1-14 adducts/10 7 nucleotides, respectively. Only O 6 -Me-Gua was detected in peripheral leukocyte DNA, accounting for 3.6% of that in the liver [71]. Using a radioimmunoassay with a monoclonal antibody against O 6 -Me-dGuo, this adduct has been detected at relatively high levels in the esophagus and liver DNA of esophageal cancer patients in China who were considered exposed to high levels of dietary N-nitrosamines [72].
Peter Magee would frequently conclude his lectures by noting that NDMA is a potent hepatocarcinogen in rats, but we have insufficient data to conclude that it is a human carcinogen. Now we have massive biochemical, molecular biological, and laboratory animal testing data in support of its potential human carcinogenicity, resulting in increased regulatory vigilance that has decreased NDMA exposure in most situations to extremely low levels. This appears to be an excellent example of cancer prevention through research and application.

N-Nitrosomethylethylamine (NMEA)
3.2.1. Exposure and Carcinogenicity NMEA (2, Figure 1) has been found in a variety of processed foods [73,74]. It can be formed as a byproduct during water disinfection with chlorine [75]. It is also one of the contaminants found in some drug products at trace levels [76]. NMEA has been documented as a tobacco constituent at very low concentrations [73,77].
The carcinogenicity of NMEA was demonstrated in rats in 1967. Hepatocellular carcinomas are the primary cancer type induced by NMEA administered in drinking water at doses of 1 or 2 mg/kg body weight per day [6]. As part of an extensive program on the chemistry and biology of N-nitroso compound [5], Lijinsky et al. showed that deuterated NMEA-d 3 38 (Figure 4) led to a high incidence of esophageal tumors when compared to undeuterated NMEA administered at an identical dose [78,79].
hepatocarcinogen in rats, but we have insufficient data to conclude that it is a human carcinogen. Now we have massive biochemical, molecular biological, and laboratory animal testing data in support of its potential human carcinogenicity, resulting in increased regulatory vigilance that has decreased NDMA exposure in most situations to extremely low levels. This appears to be an excellent example of cancer prevention through research and application.

N-Nitrosomethylethylamine (NMEA)
3.2.1. Exposure and Carcinogenicity NMEA (2, Figure 1) has been found in a variety of processed foods [73,74]. It can be formed as a byproduct during water disinfection with chlorine [75]. It is also one of the contaminants found in some drug products at trace levels [76]. NMEA has been documented as a tobacco constituent at very low concentrations [73,77].
The carcinogenicity of NMEA was demonstrated in rats in 1967. Hepatocellular carcinomas are the primary cancer type induced by NMEA administered in drinking water at doses of 1 or 2 mg/kg body weight per day [6]. As part of an extensive program on the chemistry and biology of N-nitroso compound [5], Lijinsky et al. showed that deuterated NMEA-d3 38 (Figure 4) led to a high incidence of esophageal tumors when compared to undeuterated NMEA administered at an identical dose [78,79].

Metabolism
It is generally considered that NMEA is bioactivated by two α-hydroxylation pathways (Scheme 2). When the α-hydroxylation occurs on the ethyl group (as illustrated by 41), NMEA is converted to methyldiazonium ion 15, the same intermediate as formed from NDMA metabolism; the other product is acetaldehyde 42. When the α-hydroxylation occurs on the methyl group (as illustrated by 43), the ethyldiazonium ion 45 is formed along with formaldehyde 12. The two diazonium ions 15 and 45 can alkylate DNA and form methyl or ethyl DNA adducts correspondingly. A detailed discussion of the pathway by which α-hydroxylation occurs in the ethyl group can be found in the metabolism study of NDEA in Section 3.4.

Metabolism
It is generally considered that NMEA is bioactivated by two α-hydroxylation pathways (Scheme 2). When the α-hydroxylation occurs on the ethyl group (as illustrated by 41), NMEA is converted to methyldiazonium ion 15, the same intermediate as formed from NDMA metabolism; the other product is acetaldehyde 42. When the α-hydroxylation occurs on the methyl group (as illustrated by 43), the ethyldiazonium ion 45 is formed along with formaldehyde 12. The two diazonium ions 15 and 45 can alkylate DNA and form methyl or ethyl DNA adducts correspondingly. A detailed discussion of the pathway by which α-hydroxylation occurs in the ethyl group can be found in the metabolism study of NDEA in Section 3.4.

DNA Adducts Formed by NMEA Metabolism
After a single i.p. injection of NMEA to F344 rats, N7-Me-Gua 27 ( Figure 3) was readily detected in the liver, kidneys, and esophageal DNA. However, O 6 -Me-Gua 24 was only quantifiable in the liver and kidneys. N7-Et-Gua 54 ( Figure 5) was also detected in the hepatic DNA [80]. The level of N7-Me-Gua in hepatic DNA exceeded N7-Et-Gua by  times. N7-Me-Gua occurred at the highest concentration in the liver, followed by the kidneys (15-fold lower), esophagus (100-fold lower), and lung (200-fold lower) [81]. In another study in which F344 rats were administered [1-ethyl- 14  were also identified as minor products resulting from NMEA metabolism [82].

DNA Adducts Formed by NMEA Metabolism
After a single i.p. injection of NMEA to F344 rats, N7-Me-Gua 27 ( Figure 3) was readily detected in the liver, kidneys, and esophageal DNA. However, O 6 -Me-Gua 24 was only quantifiable in the liver and kidneys. N7-Et-Gua 54 ( Figure 5) was also detected in the hepatic DNA [80]. The level of N7-Me-Gua in hepatic DNA exceeded N7-Et-Gua by  times. N7-Me-Gua occurred at the highest concentration in the liver, followed by the kidneys (15-fold lower), esophagus (100-fold lower), and lung (200-fold lower) [81]. In another study in which F344 rats were administered [1-ethyl-14 C]NMEA 39 ( Figure 4) by a single i.p. injection, N7-Et-Gua and O 6 -Et-Gua 52 ( Figure 5) were the two major ethyl DNA adducts identified in the liver after 4 h exposure. N3-Et-Gua 51, N3-Et-Ade 47, and N7-Et-Ade 48 were also identified as minor products resulting from NMEA metabolism [82].  Quantitative differences of methylation and/or ethylation in hepatic DNA were clearly observed in F344 rats administered NMEA or the related N-nitrosamines NDMA and NDEA by i.p. injections. Although DNA methylation by NMEA was comparable to that observed upon the administration of an equimolar mixture of NDMA plus NDEA, DNA ethylation by NMEA was~4-fold lower [83]. Thus, the ratio changes of methyl/ethyl DNA adducts formed by NMEA in the tissue DNA of rats may explain the shift of organ specificity of NMEA-d 3 carcinogenicity reported by Lijinsky et al. [78,79]. After a single i.p. injection of [methyl-14 C]NMEA-d 3 40 ( Figure 4) to F344 rats, the levels of N7-Me-Gua decreased by~30% in the liver but were 160% greater in the esophagus compared to non-deuterated [methyl-14 C]NMEA. The increase in esophageal DNA methylation seems to correlate with the increased esophageal carcinogenicity of NMEA-d 3 because of the increased α-hydroxylation of its ethyl group since pharmacokinetic differences between NMEA and NMEA-d 3 have been ruled out [80].  Figure 1) has been detected in food such as smoked meat (2-56 µg/kg) [73,84], malt (5.6-11.3 ppb), and beer (trace-6.0 ppb) [85]. It is also one of the carcinogenic components of tobacco smoke (22-460 ng/cigarette) [84] and some smokeless tobacco products (30-550 ng/g) [86,87]. It may also be formed endogenously from nitrosation of sarcosine and dimethylglycine (but not creatine) [88][89][90]. Human exposure to NSAR has been demonstrated by the detection of urinary NSAR, ranging from 0.1 to 3.4 µg/day, in multiple countries [73].
The carcinogenicity of NSAR has been demonstrated in mice and rats. The dietary administration of NSAR caused nasal tumors in mice; oral exposure from drinking water caused esophageal tumors in rats. Liver tumors were observed in newborn mice administered NSAR by i.p. injection [73,84]. Rat studies administering a combination of sarcosine ethyl ester hydrochloride and NaNO2 suggested esophageal carcinogenicity of the NSAR derivative N-nitrososarcosine ethyl ester 70 ( Figure 6) [91,92].

Metabolism
NSAR is not extensively metabolized in rats; 88% of dosed NSAR is excreted unchanged in the urine [93]. Metabolic activation of NSAR starts from α-methyl hydroxylation (Scheme 3). The reactive intermediate carboxymethyldiazonium ion 78 and formaldehyde 12 are formed after the spontaneous decomposition of α-hydroxyNSAR 76. Both products resulting from NSAR α-hydroxylation are able to react with DNA [94]. By analogy to N-nitroso-2-oxopropylpropylamine metabolism [95][96][97][98], it also seems possible for NSAR to form the methyldiazonium ion 15 via the intramolecular attack by the diazotate Figure 6. Structures of agents that carboxymethylate DNA.

Mutagenicity and Genotoxicity of Carboxymethyl DNA Adducts
KDA 74 ( Figure 6) caused GC to AT transitions in the p53 gene in equal amounts to GC to TA and AT to TA transversions. This contrasts with the methylating agent methylnitrosourea which causes predominantly GC to AT transitions. The difference is hypothesized to be due to the different mutagenicity of O 6 -CM-dGuo 82 compared with O 6 -Me-dGuo [105].
O 6 -CM-dGuo is not a substrate of bacterial and mammalian AGT [103] but can be repaired by human O 6 -methylguanine-DNA methyltransferase (MGMT) [116]. It may also be repaired by nucleotide excision repair (NER) [117]. In E. coli cells, O 6 -CM-dGuo strongly impeded DNA replication. It caused exclusive G to A transitions during the bypass that uniquely requires the involvement of DNA polymerases IV and V [118]. Pol η and ζ were also likely involved in bypassing O 6 -CM-dGuo lesions [119].
In a primer extension assay, N 4 -CM-dCyd 80 ( Figure 7) strongly blocked the extension reaction, causing relatively few C to A or T mutations; N 6 -CM-dAdo 79 blocked the extension reaction to a lesser extent but caused a higher frequency of A to T mutations [114]. Pol η was able to readily bypass N 6 -CM-dAdo lesions with high fidelity. However, it bypassed N 4 -CM-dCyd lesions inefficiently with a substantial frequency of dCMP and dAMP misincorporation [120]. N3-CM-Thd 83 and O 4 -CM-Thd 84 were highly resistant to the bypass of the yeast Pol η. Preferential T to C mutations caused by those two lesions were observed [120].
The five carboxymethyl DNA adducts Figure 7) were investigated individually for their mutagenicity and genotoxicity in human HEK293T cells. Among the five adducts, O 6 -CM-dGuo was moderately genotoxic by blocking DNA replication and weakly mutagenic by inducing 6.4% G to A mutations. In contrast, two Thd adducts, N3-CM-Thd and O 4 -CM-Thd, were strongly mutagenic, inducing 81% T to A mutations and 68% T to C mutations, respectively [119].

Carboxymethyl DNA Adducts in Human Tissues
Based on the evidence of human exposure to NSAR and its potential metabolic activation mechanism, it is reasonable to anticipate the detection of carboxymethyl DNA adducts in humans. In the blood DNA of healthy volunteers restricted to a standardized high-meat diet, O 6 -CM-Gua 81 ( Figure 7) was detected at levels of 35-80 adducts/10 8 nucleotides [104]. The formation of O 6 -CM-Gua in the colonic exfoliated cells of healthy volunteers has been positively associated with the consumption of red meat. Levels of O 6 -CM-Gua were also positively correlated with the concentrations of apparent total fecal N-nitroso compounds [121]. These results provide some evidence supporting the etiologi-cal role of dietary N-nitroso compounds such as NSAR or nitrosated glycine derivatives, which can decompose to form carboxymethylating and methylating agents, in human gastrointestinal carcinogenesis. However, further studies are required.
vation mechanism, it is reasonable to anticipate the detection of carboxymethyl DNA adducts in humans. In the blood DNA of healthy volunteers restricted to a standardized high-meat diet, O 6 -CM-Gua 81 (Figure 7) was detected at levels of 35-80 adducts/10 8 nucleotides [104]. The formation of O 6 -CM-Gua in the colonic exfoliated cells of healthy volunteers has been positively associated with the consumption of red meat. Levels of O 6 -CM-Gua were also positively correlated with the concentrations of apparent total fecal Nnitroso compounds [121]. These results provide some evidence supporting the etiological role of dietary N-nitroso compounds such as NSAR or nitrosated glycine derivatives, which can decompose to form carboxymethylating and methylating agents, in human gastrointestinal carcinogenesis. However, further studies are required.

Exposure and Carcinogenicity
Human exposure to NDEA (4, Figure 1) occurs through water, food, and cosmetics [18,73]. Some batches of drugs including irbesartan, losartan, and valsartan have been recalled due to contamination with NDEA since 2018 [16]. It has also been identified in tobacco smoke, with greater concentrations found in sidestream smoke compared to mainstream smoke [73]. However, the levels of NDEA in current cigarette smoke and smokeless tobacco products are extremely low [77].

Exposure and Carcinogenicity
Human exposure to NDEA (4, Figure 1) occurs through water, food, and cosmetics [18,73]. Some batches of drugs including irbesartan, losartan, and valsartan have been recalled due to contamination with NDEA since 2018 [16]. It has also been identified in tobacco smoke, with greater concentrations found in sidestream smoke compared to mainstream smoke [73]. However, the levels of NDEA in current cigarette smoke and smokeless tobacco products are extremely low [77].
The carcinogenicity of NDEA has been demonstrated in multiple laboratory animal species including mice, rats, Syrian golden hamsters, guinea pigs, rabbits, dogs, pigs, and monkeys [73]. Tumors caused by NDEA primarily occurred in the liver, often with lung metastases in mice and rats after oral administration [73]. Peto et al. conducted a large study of rats treated with NDEA in drinking water for their lifetime starting from 6 weeks of age. Various types of liver cancers were observed primarily in the NDEA-treated rats; esophageal cancers were also observed exclusively with NDEA but not NDMA treatment. A few nasopharyngeal tumors were also caused by NDEA in the rats [21,22]. The IARC upgraded NDEA to Group 2A in 1987, but as is the case with NDMA, re-evaluation may be appropriate.

Metabolism
NDEA metabolic activation for carcinogenicity is principally catalyzed by P450 2E1 and P450 2A6 [24,122,123]. When the P450-catalyzed hydroxylation occurs at the α-carbon of the ethyl group of NDEA, the electrophilic ethyldiazonium ion 45 (Scheme 4) is formed after decomposition of the unstable intermediate ethyl diazohydroxide 44. Intermediate 44 reacts with DNA producing ethyl DNA adducts such as N7-Et-Gua and O 6 -Et-Gua. Acetaldehyde 42 is also formed by NDEA α-hydroxylation [124]. When the β-carbon is hydroxylated, the reactive intermediate 2-hydroxyethyldiazonium ion 100 can be formed by a secondary α-hydroxylation on the other ethyl group of NDEA (as illustrated by 90). The 2-hydroxyethyldiazonium ion 100 alkylates DNA and forms DNA adducts such as N7-HOEt-Gua 69 ( Figure 5) that have been detected in the hepatic DNA of NDEA-treated rats [82]. The denitrosation reaction competes with bioactivation in NDEA metabolism, suggesting a dual role played by P450s in the biotransformation of NDEA [35].
Acetaldehyde 42 is also formed by NDEA α-hydroxylation [124]. When the β-carbon is hydroxylated, the reactive intermediate 2-hydroxyethyldiazonium ion 100 can be formed by a secondary α-hydroxylation on the other ethyl group of NDEA (as illustrated by 90). The 2-hydroxyethyldiazonium ion 100 alkylates DNA and forms DNA adducts such as N7-HOEt-Gua 69 ( Figure 5) that have been detected in the hepatic DNA of NDEA-treated rats [82]. The denitrosation reaction competes with bioactivation in NDEA metabolism, suggesting a dual role played by P450s in the biotransformation of NDEA [35].  [125,126]. In addition, trace levels of N7-HOEt-Gua 69 were also detected in the hepatic DNA of rats exposed to NDEA, comprising less than 2% of N7-Et-Gua 54 [82].
Variation in ethyl DNA base adduct accumulation has been investigated in the liver DNA of rats exposed to NDEA in drinking water for 77 days [127,128]. O 4 -Et-Thd 60 accumulated in the first 28 days and persisted at such levels until the end of the experiment. O 6 -Et-dGuo 53, in contrast, did not accumulate after repeated exposure to NDEA. The  [125,126]. In addition, trace levels of N7-HOEt-Gua 69 were also detected in the hepatic DNA of rats exposed to NDEA, comprising less than 2% of N7-Et-Gua 54 [82].
Variation in ethyl DNA base adduct accumulation has been investigated in the liver DNA of rats exposed to NDEA in drinking water for 77 days [127,128]. O 4 -Et-Thd 60 accumulated in the first 28 days and persisted at such levels until the end of the experiment. O 6 -Et-dGuo 53, in contrast, did not accumulate after repeated exposure to NDEA. The concentration of O 6 -Et-dGuo was highest after 2 days of administration of NDEA and decreased throughout the 77-day study course [127,128]. This was attributed to a rapid repair mechanism for O 6 -Et-dGuo when compared with O 4 -Et-Thd [129].

Mutagenicity and Genotoxicity of Ethyl DNA Adducts
The dynamic change in the accumulation of O 6 -Et-dGuo and O 4 -Et-Thd after continued exposure to NDEA clearly indicates the different repair mechanisms of these ethyl DNA adducts. O 6 -Et-dGuo can be repaired by AGT without the excision of the DNA base [130]. On the contrary, O 4 -Et-Thd is not a good substrate of rat liver AGT; the half-life of O 4 -Et-Thd reaches 11-19 days in vivo [41,129,131,132]. A similar result was also observed for O 2 -Et-Thd 56 ( Figure 5) [133]. The DNA repair protein AGT repairs O 4 -Et-Thd, however, to a lesser extent compared to its analog O 4 -Me-Thd 34 ( Figure 3) [134][135][136].
Diverse mutagenic consequences have been observed for the 3 regioisomers of ethyl thymidine adducts. O 4 -Et-Thd, a major-groove lesion, induces a large number of A to G transitions when incorporated into polynucleotides. However, O 2 -Et-Thd (a minorgroove lesion) was only slightly mutagenic and N3-Et-Thd 58 ( Figure 5) did not induce mutations [137]. O 4 -Et-Thd is also genotoxic by moderately blocking DNA replication with a bypass efficiency of 20-33% in human cells [138].
The DNA polymerases responsible for bypassing the three regioisomeric ethyl thymidine lesions have been extensively investigated. DNA polymerase I and RNA polymerase II can recognize the ethyl Thd lesions and direct the dGMP misincorporation opposite to O 4 -Et-Thd but not O 2 -Et-Thd [139,140]. Human Pol η similarly can bypass all 3 ethyl Thd lesions and generate the replication product with a substantial frequency of A to G transitions caused by O 4 -Et-Thd [141][142][143]. In E. coli cells, both Pol IV and Pol V are essential for the misincorporation of dCMP opposite to O 2 -Et-Thd, whereas Pol V is necessary for the T to A transversions caused by this lesion [144]. In human cells, Pol η and ζ are the predominant polymerases mainly responsible for bypassing O 2 -and O 4 -Et-Thd, and causing the respective T to A/G and T to C mutations [138,145].

Ethyl DNA Adducts in Human Tissues
Ethyl DNA adducts have been detected in various human tissues. They may arise from the metabolism of the N-nitrosamines NDEA and NMEA, or from structurally unknown directly acting ethylating agents present in tobacco and tobacco smoke [146]. Levels of O 2 -, N3-, and O 4 -Et-Thd (56, 58, and 60, Figure 5) in the leukocyte DNA of 20 smokers were 44.8 ± 52.0, 41.1 ± 43.8, and 48.3 ± 53.9 adducts/10 8 nucleotides, respectively, significantly exceeding those in the 20 nonsmokers. The formation of each ethyl thymidine adduct was statistically associated with that of the other two ethyl Thd adducts [147]. O 4 -Et-Thd was detected in the lower respiratory tract DNA of smokers but not in nonsmokers [148]. O 4 -Et-Thd levels in the lung DNA of smokers (3.8 adducts/10 8 nucleotides) were higher (p < 0.01) than in nonsmokers (1.6 adducts/10 8 nucleotides) [149]. The smoking-related formation of this adduct in lung DNA was further confirmed in Hungarian lung cancer patients [150].
Similarly, levels of N3-Et-Ade 47 and N7-Et-Gua 54 ( Figure 5) were also significantly higher in the leukocyte DNA from smokers (16.0 ± 7.8 and 9.7 ± 8.3 adducts/10 8 nucleotides, respectively) than those from nonsmokers (5.4 ± 2.6 and 0.3 ± 0.8 adducts/10 8 nucleotides, respectively). The levels of N3-Et-Ade and N7-Et-Gua were positively correlated [151]. N3-Et-Ade and N7-Et-Gua were also detected in the salivary DNA of smokers and nonsmokers. The occurrence of N7-Et-Gua in the saliva of smokers (14.1 ± 8.2 adducts/10 8 nucleotides) was significantly higher than that of nonsmokers (3.8 ± 2.8 adducts/10 8 nucleotides, p < 0.0001). The levels of N7-Et-Gua were also strongly associated with tobacco smoking [152]. However, there are some contradictory results that question the robustness of applying ethyl DNA adducts as biomarkers for smoking-related cancer etiology studies. We have quantified the levels of N7-Et-Gua in human leukocyte DNA from 30 smokers and 30 nonsmokers. No statistical difference was observed in the levels of this adduct in the leukocyte DNA from the two subject groups (smokers: 49.6 ± 43.3 fmol/µmol Gua; nonsmokers: 41.3 ± 34.9 fmol/µmol Gua) [153].
In addition to human tissue DNA, ethyl DNA adducts have been detected in human urine showing a potential correlation with smoking status. Urinary excretion of N3-Et-Ade has been observed to increase 5-8-fold with tobacco smoking [154,155] but not from dietary origin [156]. At 50%, the mean level of N-terminal N-ethylvaline in the hemoglobin of smokers was also significantly higher than in the nonsmokers [157].
Other than smoking, dietary exposure to ethylating agents derived from N-nitrosamines or possibly other sources may play an important role in human carcinogenesis, especially in some particular geographic areas. In Linxian, the incidence of esophageal cancer was comparably higher than its surrounding area in the same region of China. In the esophageal and hepatic DNA of esophageal cancer patients from Linxian, relatively high levels of O 6 -Et-dGuo 53 ( Figure 5) were detected by radioimmunoassay. This appeared to be consistent with the relatively high exposure levels of dietary N-nitrosamines in those who lived in this area [72].
The endogenous formation of ethylating agents may be important in the formation of ethyl DNA adducts in the human liver. In the liver DNA of 15 autopsy specimens, O 4 -Et-Thd occurred at levels of 0.5-140 adducts/10 7 nucleotides [71]. It was not detected in the peripheral leukocyte DNA in the same study [70,71]. N7-Et-Gua was detected in 25 of 26 human hepatic DNA samples, occurring at a level of 42.2 ± 43.0 fmol/µmol Gua [158].
3.5. N-Nitroso-di-n-propylamine (NDPA) 3.5.1. Exposure and Carcinogenicity Human exposure to NDPA (5, Figure 1) may arise from the consumption of drinking water, food, and beverages, contact with pesticides and wastewater, or from endogenous formation due to the use of nitrite-or secondary amine-containing food or drugs [73,159].
The carcinogenicity of NDPA has been demonstrated in laboratory animal studies with rats, mice, hamsters, and monkeys. After the oral administration of NDPA to rats, tumors primarily occurred in the liver, nasal cavity, and esophagus [73,159].

N-Nitroso-di-n-Propylamine (NDPA)
3.5.1. Exposure and Carcinogenicity Human exposure to NDPA (5, Figure 1) may arise from the consumption of drinking water, food, and beverages, contact with pesticides and wastewater, or from endogenous formation due to the use of nitrite-or secondary amine-containing food or drugs [73,159].
The carcinogenicity of NDPA has been demonstrated in laboratory animal studies with rats, mice, hamsters, and monkeys. After the oral administration of NDPA to rats, tumors primarily occurred in the liver, nasal cavity, and esophagus [73,159].
Considering the structural similarity of NDPA and other N-alkylnitrosamines such as NDMA and NDEA, P450-catalyzed denitrosation is likely to occur through a radical mechanism ultimately forming propylamine, propionaldehyde, and nitrate [37,169].
The carcinogenicity of NDELA has been extensively studied in rats, mice, and hamsters [73,179]. After oral administration, NDELA induced primarily liver tumors in rats. It also induced lung tumors in mice and some nasal tumors in rats. In hamsters, NDELA treatment resulted in nasal cavity tumors and tracheal tumors regardless of the administration pathways (s.c. injection, topical application, and oral swabbing) [73,179].

Metabolism
The distribution of NDELA in Osborne-Mendel rats has been studied using two administration pathways. After oral administration, NDELA was absorbed and distributed rapidly and reached a peak concentration at 8 h; after topical application, NDELA was slowly absorbed but rapidly distributed as when dosed orally. NDELA was excreted mainly in the urine as the unchanged form and one metabolite [180,181]. The excreted NDELA also represented a high percentage (60-90%) in the urine of male Sprague-Dawley rats administered NDELA in drinking water [182]. A similar high urinary excretion rate (73-89%) of unchanged NDELA was also determined in rats treated percutaneously and intratracheally [19]. However, the absorption rate of NDELA was significantly lower in Syrian golden hamsters. After s.c. injection, 49% and 11% of the dose appeared in the urine and feces, respectively, in 16 h; 34% and 6% after oral swabbing; and only 21% and 4% were detected after skin application [183].
An investigation of the rat urinary metabolites of NDELA suggested only one compound containing the nitroso moiety, which was identified as N-nitroso-(2-hydroxyethyl)glycine 135 (NHEG, Scheme 6). It represented 6% of the dosed NDELA in the rats [184]. The glucuronide of NDELA was also identified in rat urine after gavage [185]. However, no sulfate derivatives were observed even though NDELA sulfate was considered a possible activated metabolite that could react with DNA to form 2-hydroxyethyl adducts (as shown in Figure 5) [185][186][187]. N-Nitroso-2-hydroxymorpholine 137 (NHMOR) was also observed as a minor metabolite in vitro in rat liver S9 supernatant [188,189]. The formation of NHEG 135 and NHMOR 137 strongly implies the important role that β-hydroxylation may play in the bioactivation of NDELA in vivo (Scheme 6). This reaction is catalyzed primarily by P450 2E1 [188]. NHMOR can be further metabolized by α-hydroxylation on the two methylene groups. When the α-hydroxylation occurs on the 3-carbon, the major metabolite has been identified as glyoxal 133 formed via 139 and 140; when the α-hydroxylation occurs on the 5-carbon, the major metabolite is 2-acetoxyacetaldehyde 144 formed via 142 and 143 [188]. It is noteworthy that NHMOR also arises from NMOR metabolism which is discussed in detail in Section 4.3.2.
However, several lines of evidence raise questions regarding the importance of βhydroxylation in the carcinogenesis of NDELA. Although it is a stable precursor to poten- The formation of NHEG 135 and NHMOR 137 strongly implies the important role that β-hydroxylation may play in the bioactivation of NDELA in vivo (Scheme 6). This reaction is catalyzed primarily by P450 2E1 [188]. NHMOR can be further metabolized by α-hydroxylation on the two methylene groups. When the α-hydroxylation occurs on the 3-carbon, the major metabolite has been identified as glyoxal 133 formed via 139 and 140; when the α-hydroxylation occurs on the 5-carbon, the major metabolite is 2-acetoxyacetaldehyde 144 formed via 142 and 143 [188]. It is noteworthy that NHMOR also arises from NMOR metabolism which is discussed in detail in Section 4.3.2.
However, several lines of evidence raise questions regarding the importance of βhydroxylation in the carcinogenesis of NDELA. Although it is a stable precursor to potential DNA alkylating agents such as glyoxal, NHMOR was inactive or marginally carcinogenic to rats or mice when administered in drinking water [190]. Only the glyoxal-deoxyguanosine adduct 145 (N1,N 2 -glyoxal-dGuo, Figure 11) was observed in vitro in NHMOR-incubated calf thymus DNA (up to 48 h) and in vivo in the liver DNA of rats given NHMOR at a single dose by gavage for 4 h [191]. The levels of N1,N 2 -glyoxal-dGuo formed by NHMOR were also lower than those formed by NDELA [192]. Those findings taken together suggest that additional metabolic activation pathways such as α-hydroxylation are involved in the carcinogenesis of NDELA [193]. The formation of NHEG 135 and NHMOR 137 strongly implies the important role that β-hydroxylation may play in the bioactivation of NDELA in vivo (Scheme 6). This reaction is catalyzed primarily by P450 2E1 [188]. NHMOR can be further metabolized by α-hydroxylation on the two methylene groups. When the α-hydroxylation occurs on the 3-carbon, the major metabolite has been identified as glyoxal 133 formed via 139 and 140; when the α-hydroxylation occurs on the 5-carbon, the major metabolite is 2-acetoxyacetaldehyde 144 formed via 142 and 143 [188]. It is noteworthy that NHMOR also arises from NMOR metabolism which is discussed in detail in Section 4.3.2.
However, several lines of evidence raise questions regarding the importance of βhydroxylation in the carcinogenesis of NDELA. Although it is a stable precursor to potential DNA alkylating agents such as glyoxal, NHMOR was inactive or marginally carcinogenic to rats or mice when administered in drinking water [190]. Only the glyoxal-deoxyguanosine adduct 145 (N1,N 2 -glyoxal-dGuo, Figure 11) was observed in vitro in NHMORincubated calf thymus DNA (up to 48 h) and in vivo in the liver DNA of rats given NHMOR at a single dose by gavage for 4 h [191]. The levels of N1,N 2 -glyoxal-dGuo formed by NHMOR were also lower than those formed by NDELA [192]. Those findings taken together suggest that additional metabolic activation pathways such as α-hydroxylation are involved in the carcinogenesis of NDELA [193].  Catalyzed primarily by P450 2E1, hydroxylation occurs on the α-carbon of NDELA and forms the 2-hydroxyethyldiazonium ion 127 and glycolaldehyde 130 after spontaneous decomposition [188,194]. The 2-hydroxyethyldiazonium ion forms the carbocation 128 after the loss of H 2 O and subsequently yields the solvolysis product ethylene glycol 129, which can also undergo microsome-mediated oxidation to glyoxal 133. The carbocation 128 also undergoes elimination and forms acetaldehyde 42 [188,195]. O 6 -(2-Hydroxyethyl)-2deoxyguanosine 68 (O 6 -HOEt-dGuo, Figure 5) arising from NDELA metabolism has been detected in rat liver DNA; it is not derived from NHMOR metabolism [191]. Deuteration on the α-carbon of NDELA greatly decreased the formation of glycolaldehyde 130 and O 6 -HOEt-dGuo 68, whereas β-deuteration oppositely affected the formation of those products. This further suggests that O 6 -HOEt-dGuo mainly results from NDELA αhydroxylation [191,194]. One unexpected finding was that glycolaldehyde 130, other than being classically converted to glycolic acid 131 and oxalic acid 132, was also transformed to glyoxal 133 via the catalysis of P450 2E1 [188,194]. The collective evidence indicates the necessity of the α-hydroxylation of NDELA for DNA adduct formation.

DNA Adducts Formed by NDELA Metabolism
O 6 -HOEt-dGuo 68 and N7-HOEt-Gua 69 ( Figure 5) were characterized in the reaction mixture of N-nitroso-3-acetoxy-2-hydroxymorpholine 138 (Scheme 6) with dGuo. However, neither of these adducts was detected in the reaction mixture of N-nitroso-5-acetoxy-2hydroxymorpholine 141 with dGuo. This suggested that the α-hydroxylation of NHMOR on the 5-carbon was unlikely to produce DNA adducts, whereas NHMOR α-hydroxylation on the 3-carbon might yield a carcinogenic outcome [195]. However, this seems not to be supported by an in vivo study, in which no 2-hydroxyethyl guanine adducts were detected in the liver DNA of rats treated with NHMOR [188]. In contrast, 2-hydroxyethyl guanine adducts were tentatively identified in the hydrolysates of the liver DNA of rats treated with NDELA by gavage [185]. One adduct was later characterized as O 6 -HOEt-dGuo using the synthesized authentic standard. It was detected in vivo in the liver DNA of rats treated with NDELA [191,194,196]. Taken together, 2-hydroxyethyl adducts formed by NDELA are reasonably considered to result from the α-hydroxylation pathway rather than the β-hydroxylation pathway.
NDBA was classified by IARC in 1978 and reaffirmed in 1987 as a group 2B carcinogen. It is carcinogenic to the esophagus and bladder of laboratory animals including mice, rats, Syrian golden hamsters, and guinea pigs. It also causes liver and forestomach tumors [73,[198][199][200][201][202]. In 1983, Lijinsky and Reuber reported that NDBA, even though much weaker than NDPA, induced liver tumors in 60% of rats administered by gavage for 83 weeks. It also induced forestomach (50%) and bladder tumors (35%) [203]. Even after a short-term (2 weeks) exposure to NDBA, preneoplastic lesions were positively found in the liver, esophagus, forestomach, and bladder of rats 52 weeks post-treatment [204]. Mice with a p53 gene knockout had increased susceptibility to esophageal and bladder carcinogenesis caused by NDBA [205].

Metabolism
By analogy to other N-nitrosodialkylamines, NDBA requires metabolic activation to exert its carcinogenicity. Four metabolic pathways can occur in NDBA metabolism (Scheme 7). They are α-, β-, γ-, and δ-hydroxylation of the butyl group of NDBA, among which αhydroxylation has been suggested to be primarily involved in NDBA carcinogenesis [206]. short-term (2 weeks) exposure to NDBA, preneoplastic lesions were positively found in the liver, esophagus, forestomach, and bladder of rats 52 weeks post-treatment [204]. Mice with a p53 gene knockout had increased susceptibility to esophageal and bladder carcinogenesis caused by NDBA [205].
In the urine of rats treated with NDBA, the end products of NDBA-GSH conjugates were also detected. They are N-acetyl-S-butyl-L-cysteine 153 resulting from αhydroxylation of NDBA and N-acetyl-S-3-hydroxybutyl-L-cysteine 158 and N-acetyl-S-3oxobutyl-L-cysteine 159 that are hypothesized to result from γ-hydroxylation of NDBA followed by a secondary α-hydroxylation [213].

DNA Adducts Formed by NDBA Metabolism
The butyl carbocation 150 (Scheme 7) resulting from the α-hydroxylation of NDBA is considered to be the alkylating agent which attacks DNA forming n-butyl DNA adducts. In the liver DNA of rats treated with 185 mg/kg NDBA by a single i.p. dose, O 6 -(nbutyl)guanine 123 (O 6 -n-Bu-Gua, Figure 10) has been detected at a concentration of 0.34 µmol/mol Gua [214].
The sequential α-hydroxylation of NDBA metabolites such as BBN 160 can form the other alkylating species and react with DNA to cause lesions. In the urothelial and hepatic DNA of rats, both O 6 -n-Bu-Gua 123 and O 6 -(4-hydroxybutyl)guanine 124 (O 6 -(4-OH-n-Bu)-Gua, Figure 10) were detected at 17.9 ± 7.23 and 12.2 ± 7.01 µmol/mol Gua, respectively, after 24 h treatment with a single oral dose of 120 mg BBN. O 6 -n-Bu-Gua did not accumulate after repeated exposure to a lower dose of BBN; O 6 -(4-OH-n-Bu)-Gua was not detected in the same study [215].

Metabolic Activation and DNA Interactions of Carcinogenic Cyclic N-Nitrosamines
Three carcinogenic cyclic N-nitrosamines (NPYR, NPIP, and NMOR, Figure 1) are discussed here due to their common human exposure. They are introduced in order based on their increasing structural complexity.  Figure 1) is a simple symmetric cyclic N-nitrosamine. It has been extensively investigated as a model compound for studies of N -nitrosonornicotine (NNN), and in its own right due to its common occurrence in food (1.5 ± 0.2 ng/g) and water (5.5 ± 2.6 ng/L) [9]. It is also present in tobacco smoke and smokeless tobacco products [73,77,216]. Endogenous nitrosation is considered to be another possible pathway to form NPYR, due to the relatively high human exposure to its precursor pyrrolidine, which is excreted to the extent of~20 mg per day in the urine [217].
NPYR is a strong hepatic carcinogen in mice and rats. The IARC reviewed some early data on NPYR carcinogenicity in 1978 [73]. Since then, some new data have been reported. In rats administered NPYR in drinking water, it primarily caused hepatocellular carcinomas, many of which metastasized [218,219]. A dose-response relationship of hepatic tumor formation was observed in rats administered NPYR in drinking water [220]. NPYR also induced lung tumors in A/J mice after i.p. injection [221]. In Syrian golden hamsters, NPYR induced tracheal and nasal cavity tumors when administered intraperitoneally [222] but tracheal papillomas and hepatic neoplastic nodules when fed in the diet [223].
α-Hydroxylation of NPYR predominated in explanted esophagus from rats and humans [236], cultured human colon [237], and cultured bladder cells from rats and humans [238]. The rate of NPYR -hydroxylation by rat hepatic microsomes was 1.43 nmol/min/mg protein [228]; this rate was lower in human hepatic microsomes (0.68 nmol/min/mg protein) [239]. The apparent Km for rat lung microsomes and post-microsomal supernatant was approximately 20 mM but was much smaller for rat liver microsomes (0.36 mM) [229,240]. NPYR -hydroxylation was inducible in rats and hamsters that were pre-treated with Aroclor or ethanol [228,241,242].

DNA Adducts Formed by NPYR Metabolism
As depicted in Scheme 8, the butanal carbocation 172 is considered to form the oxonium ion 170 and the isomeric carbocation 174 after rearrangement. All 3 reactive species can react with DNA and form the corresponding DNA adducts. Crotonaldehyde 177, the elimination product of the isomeric carbocation 174, also reacts with DNA.
α-Hydroxylation of NPYR predominated in explanted esophagus from rats and humans [236], cultured human colon [237], and cultured bladder cells from rats and humans [238]. The rate of NPYR α-hydroxylation by rat hepatic microsomes was 1.43 nmol/min/mg protein [228]; this rate was lower in human hepatic microsomes (0.68 nmol/min/mg protein) [239]. The apparent K m for rat lung microsomes and post-microsomal supernatant was approximately 20 mM but was much smaller for rat liver microsomes (0.36 mM) [229,240]. NPYR α-hydroxylation was inducible in rats and hamsters that were pre-treated with Aroclor or ethanol [228,241,242].

DNA Adducts Formed by NPYR Metabolism
As depicted in Scheme 8, the butanal carbocation 172 is considered to form the oxonium ion 170 and the isomeric carbocation 174 after rearrangement. All 3 reactive species can react with DNA and form the corresponding DNA adducts. Crotonaldehyde 177, the elimination product of the isomeric carbocation 174, also reacts with DNA.

DNA Adducts Formed by an NPYR-Derived Oxonium Ion
The NPYR-derived oxonium ion 170 reacts with DNA forming a group of tetrahydrofuranyl DNA adducts, some of which have been identified both in vitro and in vivo. Quantitation of those adducts has been performed by reducing them to the corresponding 4-hydroxybutyl adducts (see Table 1).
In the reaction of dGuo and calf thymus DNA with α-acetoxyNPYR 165, N 2 -(tetrahydrofuran-2-yl)-2 -deoxyguanosine 185 (N 2 -THF-dGuo, Figure 12) was characterized as the first identified tetrahydrofuranyl DNA adduct derived from NPYR metabolism [253,254]. The levels of this adduct exceeded those of other adducts formed by α-acetoxyNPYR in the in vitro DNA hydrolysate samples [253]. After neutral thermal hydrolysis, 2-hydroxyTHF 171 (Scheme 8) was released as the major product from the treated DNA in vitro; N 2 -THF-dGuo 185 is considered the major precursor of this product [235]. The reduction of N 2 -THF-dGuo forms N 2 -(4-hydroxybutyl)-2 -deoxyguanosine 190 (N 2 -(4-HOB)-dGuo, Figure 12), which occurred in relatively low abundance compared to other types of NPYR DNA adducts in the liver DNA of rats. However, it remains one of the most abundant adducts among those formed by the oxonium ion 170 [247,252].

Mutagenicity and Genotoxicity of NPYR-Derived DNA Adducts
NPYR is mutagenic towards E. coli and S. typhinurium after hepatic microsomal activation [261,262]. A few DNA adducts formed by NPYR metabolism such as N1,N 2 -propano-dGuo can also arise from crotonaldehyde. The mutagenicity and genotoxicity of these types of DNA adducts have been extensively reviewed before [263][264][265]. However, the mutagenicity and genotoxicity of the other types of NPYR DNA adducts, such as 184-186 and their reduced forms 189-193, still warrant investigation.

Human DNA Adducts Related to NPYR Metabolism
The N1,N 2 -propano-dGuo adducts have been detected in human tissues [266,267]. We have quantified (6S,8S)-N1,N 2 -propano-dGuo 194 and (6R,8R)-N1,N 2 -propano-dGuo 195 in human liver, lung, and white blood cells [268]. (6S,8S)-N1,N 2 -Propano-dGuo 194 occurred at a mean (±SD) concentration of 6.70 ± 2.92 and 7.19 ± 4.14 fmol/µmol dGuo, respectively, in liver and lung DNA. However, the detection rates were low; only 4 out of 23 liver samples and 16 out of 45 lung samples were positive. Similarly, (6R,8R)-N1,N 2 -propano-dGuo 195 occurred at mean (±SD) concentrations of 7.87 ± 4.47 and 12.8 ± 7.6 fmol/µmol dGuo, respectively, in the same liver and lung DNA. No statistically significant difference was observed between the occurrence of the two diastereomers. Neither of the two isomers was detected in the DNA of 11 human white blood cell samples [268]. However, considering the fact that the most abundant NPYR DNA adduct, N7,8-butano-Gua 179, has not been detected in the human tissues, the origin of N1,N 2 -propano-dGuo is likely not due to the exposure of NPYR but rather to other pathways such as the endogenous formation of crotonaldehyde from lipid peroxidation [269]. Presently, there is no direct evidence for the presence of NPYR-DNA adducts specifically in human tissues. However, the studies on NPYR-DNA damage did lead to research on crotonaldehyde-DNA interactions and their detection in human tissue samples as noted above.  Figure 1) has been reported in water, spices, and foods such as cheese, smoked fish, and processed meat [73,270]. The average concentration of NPIP was 0.5 ± 0.1 ng/g in food and 7.9 ± 4.0 ng/L in potable water [9]. The formation of NPIP by the nitrosation of amine precursors such as piperidine has been associated with concentrations of sodium nitrite during the production of dry fermented sausages [271]. The presence of NPIP in tobacco products has been documented but is currently at extremely low levels [73,77].
The IARC reviewed the carcinogenicity data of NPIP in 1978 and reaffirmed its Group 2B classification in 1987. It has shown carcinogenic effects in multiple laboratory animals such as mice, rats, hamsters, and monkeys [73]. In mice, NPIP induced primarily liver and lung tumors when administered in the diet and caused lung adenomas when administered in the drinking water or by i.p. injection. In rats, esophageal and hepatic tumors were induced by NPIP administered in drinking water. However, the target organ shifted to the nasal cavity when administered by s.c. injection. Carcinomas of the esophagus and pharynx were mainly observed by i.v. injection of NPIP [73]. The ability of NPIP to cause esophageal tumors is striking when compared to its close analog NPYR, which never causes esophageal tumors in rats. This was confirmed by Gray et al. in a large 2-year study. The incidence of hepatic and esophageal tumors in rats was dose-responsive to the concentration of NPIP in the drinking water [220]. In hamsters, the trachea appears to be the primary target organ by NPIP after s.c. injection. In monkeys, hepatocellular carcinomas were observed after oral dosing with NPIP [73].
The metabolic bioactivation pathways illustrated in Scheme 9 were also supported by the structure-carcinogenicity studies. Methyl substitution at the α-carbons of NPIP significantly decreased tumor formation in rats administered NPIP or its methyl analogs in drinking water [284]. The β-or γ-substitutions (methyl and hydroxy) did not affect the carcinogenicity of NPIP in rats [284,285].
The metabolic bioactivation pathways illustrated in Scheme 9 were also supported by the structure-carcinogenicity studies. Methyl substitution at the α-carbons of NPIP significantly decreased tumor formation in rats administered NPIP or its methyl analogs in drinking water [284]. The βor γ-substitutions (methyl and hydroxy) did not affect the carcinogenicity of NPIP in rats [284,285].

Human DNA Adducts Related to NPIP Metabolism
In 2019, Totsuka et al. reported their DNA adductome investigation of esophageal cancer patients in China [287]. They found that a distinctive pattern of NPIP-derived DNA adduct N 2 -THP-dGuo 224 ( Figure 14) formed in the tissues of esophageal cancer patients in the high-incidence area versus those in the low-incidence area. The difference in the occurrence of this adduct was also statistically significant (p < 0.01) in the peripheral blood samples of patients from the two areas. The exposure to NPIP was not likely due to smoking or drinking since all the samples were from patients who were nonsmokers and nondrinkers. Drinking water and vegetables were potential sources that have been preliminarily investigated [287].
NPIP induced preferentially AT to CG transversions in gpt delta transgenic rats. However, the predominant somatic mutation of esophageal tumor samples was CG to TA transitions; no distinctive mutational pattern was identified to associate with esophageal cancer in the high-incidence area. In the p53 genes of~90% of patients, GC to AT transitions predominated, in agreement with the findings from their previous study in 2005 [288]. There was only a weak correlation observed between one of the mutational signatures (signature 17 in the COSMIC database) and the N 2 -THP-dGuo levels among 19 subjects (r = 0.44; p < 0.05). Taken together, this study suggests that NPIP-induced DNA adducts can at least partially contribute to esophageal carcinogenesis [287].  Figure 1) has been detected in water, food, cosmetics, and occupational airspaces [289][290][291][292][293][294][295]. It has also been documented to occur in smokeless tobacco products but at very low concentrations [77]. Similar to other N-nitrosamines, NMOR can be formed easily by nitrosation of its parent amine morpholine with NaNO 2 . It is one of the most rapidly formed nitrosamine products [296]. When morpholine was administered in combination with nitrite to rats by gavage, the formation of NMOR was readily measured in the urine, with the extent of morpholine nitrosation being 0.5-12% depending on the doses [297]. Mice exposed to 15 NO 2 and administered morpholine by gavage also formed NMOR with the highest concentrations detected in the skin followed by the stomach [298]. Human exposure to morpholine and its analogs has been considered substantial from food and drugs [270,299,300], tobacco usage [301], and rubber and tire manufacturing [302,303]. Thus, the in vivo formation of NMOR has been considered likely to occur in humans. Low levels of NMOR have been detected in human urine and gastric juice [304][305][306].
The carcinogenicity data of NMOR was reviewed by IARC in 1978 and 1987, and it was classified as a group 2B carcinogen. NMOR primarily induces liver, bile duct, and kidney tumors in mice and rats when given in the drinking water or by i.v. injection; it causes tumors of the respiratory system (trachea and nasal cavity) of hamsters by s.c. injection [73]. After 1978, there were some new carcinogenicity data reported. When NMOR was administered in the drinking water to A/J mice, lung tumors occurred at a relatively low total dose of 53-55 µmol/mouse. It induced 100% liver tumor incidence in F344 rats at a high total dose of 1.1 mmol/rat [190]. It is interesting to note that 2-/6-methylation of NMOR shifts its organospecificity. In Syrian golden hamsters treated by gavage, NMOR induced primarily nasal cavity tumors whereas N-nitroso-2-methylmorpholine induced tumors in the nasal cavity and liver [307]; N-nitroso-2,6-dimethylmorpholine caused liver and pancreas tumors [308]. In rats, NMOR is a potent liver carcinogen whereas N-nitroso-2,6-dimethylmorpholine is carcinogenic to the esophagus and nasal cavity [7,309].

Metabolism
After a single i.p. injection, NMOR was rapidly distributed throughout the rat tissues and metabolized to a relatively high extent with 24% of the dose excreted unchanged in the urine after 18 h [310]. It is metabolized via both αand β-hydroxylation pathways (Scheme 10). NMOR shifts its organospecificity. In Syrian golden hamsters treated by gavage, NMOR induced primarily nasal cavity tumors whereas N-nitroso-2-methylmorpholine induced tumors in the nasal cavity and liver [307]; N-nitroso-2,6-dimethylmorpholine caused liver and pancreas tumors [308]. In rats, NMOR is a potent liver carcinogen whereas N-nitroso-2,6-dimethylmorpholine is carcinogenic to the esophagus and nasal cavity [7,309].

Metabolism
After a single i.p. injection, NMOR was rapidly distributed throughout the rat tissues and metabolized to a relatively high extent with 24% of the dose excreted unchanged in the urine after 18 h [310]. It is metabolized via both -and β-hydroxylation pathways (Scheme 10).

Scheme 10. Mechanisms of NMOR and NHMOR metabolism.
Hydroxylation on the 2-carbons of NMOR appears to be the major biotransformation pathway, since NHEG 135, derived from the β-hydroxylation product NHMOR 137 via the intermediate 136, has been detected predominantly among the rat urinary metabolites, accounting for 33% or 37% of total metabolites [311]. It has been used as a reliable biomarker to monitor the in vivo formation of NMOR [297]. However, NHEG can also be formed by NDELA and N,N- Hydroxylation on the 2-carbons of NMOR appears to be the major biotransformation pathway, since NHEG 135, derived from the β-hydroxylation product NHMOR 137 via the intermediate 136, has been detected predominantly among the rat urinary metabolites, accounting for 33% or 37% of total metabolites [311]. It has been used as a reliable biomarker to monitor the in vivo formation of NMOR [297]. However, NHEG can also be formed by NDELA and N,N-dinitrosopiperazine, although to the lesser extent of <10% and 22% from each compound, respectively [184,312]. It is noteworthy that the group 2B carcinogen NDELA 6 is also formed by the reduction of intermediate 136 during NMOR metabolism, accounting for 12% of total urinary metabolites [311,313]. NDELA can be reversibly oxidized to form the same products 135, 136, and 137 (see Scheme 6) as in the case of NMOR.
When the hydroxylation occurs on the 3-carbon of NMOR, the unstable product N-nitroso-3-hydroxymorpholine 228 (α-hydroxyNMOR) is formed. By analogy to NPIP metabolism, a potential α-hydroxyNMOR acetate ester 234 has been suggested to be similarly formed under near-ultraviolet light irradiation conditions [314]. α-HydroxyNMOR decomposes quickly to form the diazohydroxide 229, which can also result from the regiochemically activated precursors α-acetoxy-N-nitrosomorpholine 225, N-nitrosomorpholine  [311,318]. The major ultimate product arising from NMOR α-hydroxylation was (2-hydroxyethoxy)acetic acid 237 in rat urine. A significant isotope effect was observed in the formation of 237 when the α-carbons of NMOR were deuterated (16% vs. 3.4%) [311]. It has been shown that 1,4-dioxan-2-one 238 exists in aqueous solutions in equilibrium with 237 with an equilibrium constant K OA of 0.034 ± 0.002 M. It appears to be a common non-carcinogenic metabolite of the carcinogens NMOR, dioxane, and diethylene glycol [319].

DNA Adducts Formed by NMOR Metabolism
A study by Stewart et al. in 1974 using [ 14 C]NMOR suggested that six radioactive DNA adducts are formed by NMOR metabolism. One of those compounds was putatively identified to be N7-HOEt-Gua 69 ( Figure 5 Figure 5) with a faster rate in duplex DNA than in nucleosides. Interestingly, the two 2-oxoethoxyethyl adducts 246 and 250, occurring at the exocyclic amino groups, also formed the corresponding intramolecular ring closure products 247 and 251, respectively, during chemical standard synthesis. However, 247 and 251 were not detected in the treated nucleosides or calf thymus DNA [317]. The same group also identified the first cross-link DNA adduct 6-(2-(2-((9H-purin-6-yl)amino)ethoxy)ethoxy)-9H-purin-2-amine 254 ( Figure 15) derived from α-hydroxyNMOR upon reduction. It was proposed to be formed by the carbocation 231 (Scheme 10) attacking the O 6 -OH of guanine first, followed by the pendant aldehyde reacting with N 6 -NH 2 of adenine [322]. We conducted an in vitro study of [2-(carbethoxynitrosamino)ethoxy]ethanal 227 (Scheme 10) reacting with dGuo and demonstrated the formation of N1,N 2 -glyoxal-dGuo 145 ( Figure 11) probably via glyoxal resulting from NMOR α-hydroxylation [309]. In the reaction mixture of NHMOR with dGuo, adduct 145 was also formed. The proposed mechanism of the formation of this adduct by NHMOR is depicted in Scheme 11 [315]. Scheme 11. Proposed mechanism of N1,N 2 -glyoxal-dGuo formation by NHMOR.

Concluding Remarks
The structural simplicity of the powerful carcinogen NDMA provided a unique opportunity to understand the role of metabolism and DNA adduct formation in carcinogenesis. This research has clearly demonstrated the main requisite steps in tumor for- We conducted an in vitro study of [2-(carbethoxynitrosamino)ethoxy]ethanal 227 (Scheme 10) reacting with dGuo and demonstrated the formation of N1,N 2 -glyoxal-dGuo 145 ( Figure 11) probably via glyoxal resulting from NMOR α-hydroxylation [309]. In the reaction mixture of NHMOR with dGuo, adduct 145 was also formed. The proposed mechanism of the formation of this adduct by NHMOR is depicted in Scheme 11 [315]. We conducted an in vitro study of [2-(carbethoxynitrosamino)ethoxy]ethanal 227 (Scheme 10) reacting with dGuo and demonstrated the formation of N1,N 2 -glyoxal-dGuo 145 ( Figure 11) probably via glyoxal resulting from NMOR α-hydroxylation [309]. In the reaction mixture of NHMOR with dGuo, adduct 145 was also formed. The proposed mechanism of the formation of this adduct by NHMOR is depicted in Scheme 11 [315]. Scheme 11. Proposed mechanism of N1,N 2 -glyoxal-dGuo formation by NHMOR.

Concluding Remarks
The structural simplicity of the powerful carcinogen NDMA provided a unique opportunity to understand the role of metabolism and DNA adduct formation in carcino-Scheme 11. Proposed mechanism of N1,N 2 -glyoxal-dGuo formation by NHMOR.

Concluding Remarks
The structural simplicity of the powerful carcinogen NDMA provided a unique opportunity to understand the role of metabolism and DNA adduct formation in carcinogenesis. This research has clearly demonstrated the main requisite steps in tumor formation by N-nitrosamines and other related organic carcinogens: formation of electrophilic intermediates catalyzed by cytochrome P450 enzymes, binding of those intermediates to sites in DNA, the role of DNA repair systems, mutagenesis resulting from certain unrepaired DNA adducts such as O 6 -Me-Gua, and consequent permanent mutations in critical growth control genes. This sequence of events is also quite well-established for NDEA and NMEA and forms the basis for the investigation of more structurally complex N-nitrosamines as described here. For these other N-nitrosamines, there are still gaps in our understanding of their metabolic activation pathways and the DNA adduct structures, repair mechanisms, and mutagenic properties of the persistent adducts. This review has presented an overview of the current status of this research.
Studies of the metabolic activation and DNA adduct formation of carcinogenic Nnitrosamines have long been part of the critical mass of research on these compounds. The focus of many cancer researchers on N-nitrosamines produced an overall heightened awareness of their ease of formation, levels of contamination in consumer products, and carcinogenic effects among other topics. This heightened awareness led to significant decreases in human N-nitrosamine exposures from the 1970s to the present. However, constant vigilance is necessary to maintain these lower exposure levels. The recent manufacturing errors leading to N-nitrosamine contamination of drugs would have been unimaginable in the 1980s and 1990s when the ease of N-nitrosamine formation under certain conditions was prominent in the eyes of chemists and toxicologists.
There are approximately 200 different N-nitrosamines that have been documented to be carcinogenic in more than 30 animal species [323]. Humans exposed to these compounds are, no doubt, susceptible to their carcinogenic effects. The continuous evaluation of potential N-nitrosamine contamination such as N-nitrosomethyl-n-alkylamines in medicines is in progress by other groups [324]. Among those many N-nitrosamines, the compounds listed in Figure 1 comprise the most commonly detected carcinogenic N-nitrosamines from human daily exposure, occurring in food, water, drugs, and cosmetics. An understanding of the metabolism of these carcinogens and their mechanisms of DNA interactions is necessary basic knowledge for evaluating and controlling their potential carcinogenic effects in humans, and some metabolites and/or DNA adducts that are specific to carcinogen exposure in humans may serve as biomarkers for cancer etiology studies.
With advances in bioanalytical methods, especially in the development of mass spectrometry, metabolism studies have evolved from radioisotope-labeling technology to highresolution mass spectrometry. Furthermore, mass spectrometry-based metabolomics studies may provide rich chemical information on structures of potential metabolites of interest [325]. Similarly, DNA adductomics [326] is also emerging as a useful tool to investigate DNA adducts that are potential biomarkers for cancer etiology studies, as illustrated in the NPIP section [287].
In summary, the recent occurrence of NDMA and NDEA in some batches of FDAapproved drugs has heightened global awareness of the carcinogenic effects of N-nitrosamines, which comprise a significant number of carcinogens to which humans are exposed on a daily basis through food, water, drugs, cosmetics, and tobacco products. In this review, we provide a comprehensive and updated review of 10 N-nitrosamine carcinogens with a focus on their mechanisms of metabolic activation and DNA interactions. A better understanding of the metabolism and DNA adduct formation of N-nitrosamines can hopefully provide clues for relevant cancer etiology and prevention studies.