Michael Novak

Two related classes of carcinogens, arylamines (AAs) and heterocyclic arylamines (HAAs), have played an important role in the development of our understanding of chemical carcinogenesis. AAs are largely industrial products or by‐products, but significant exposure also occurs through inhalation of cigarette smoke. HAAs are produced during broiling or frying of meats and fish and are viewed as significant cancer risks in most human populations. AAs and HAAs are properly characterized as procarcinogens because they require metabolic activation into their ultimate carcinogenic forms. Both classes are activated predominately by cytochrome P450 (CYP450) mediated oxidation into N ‐arylhydroxylamines that are further metabolized into acetic or sulfuric acid esters of the hydroxylamines by the action of N ‐acetyltransferases (NATs) or sulfotransferases (SULTs). These esters appear to be the ultimate carcinogens in most cases. Research on the metabolism of AAs and HAAs, the structures of DNA adducts derived from their carcinogenic metabolites, and the relationship of DNA adduct conformations and persistence to observed mutations are reviewed. Evidence for the proposal that the reactive intermediate responsible for the formation of DNA adducts is a nitrenium ion derived from the hydroxylamine esters is the main topic of this review. Future prospects for research in this area are considered. Introduction: Metabolic Activation of AAs and HAAs: The Structures of DNA Adducts: The Nitrenium Ion Hypothesis: The Hypothesis Tested: Concluding Remarks:

The 4-(benzothiazol-2-yl)phenylnitrenium ion 11 is generated from hydrolysis or photolysis of O-acetoxy-N-(4-(benzothiazol-2-yl)phenyl)hydroxylamine 8, a model metabolite of 2-(4-aminophenyl)benzothiazole 1 and its ring-substituted derivatives that are being developed for a variety of medicinal applications, including antitumor, antibacterial, antifungal, and imaging agents. Previously, we showed that 11 had an aqueous solution lifetime of 530 ns, similar to the 560 ns lifetime of the 4-biphenylylnitrenium ion 12 derived from the well-known chemical carcinogen 4-aminobiphenyl. We now show that the analogy between these two cations extends well beyond their lifetimes. The initial product of hydration of 11 is the quinolimine 16, which can be detected as a long-lived reactive intermediate that hydrolyzes in a pH-dependent manner into the final hydrolysis product, the quinol 15. This hydrolysis behavior is equivalent to that previously described for a large number of ester metabolites of carcinogenic arylamines, including 4-aminobiphenyl. The major azide trapping product (90% of azide products) of 11, 20, is generated by substitution on the carbons ortho to the nitrenium ion center of 11. This product is a direct analogue of the major azide adducts, such as 22, generated from trapping of the nitrenium ions of carcinogenic arylamines. The azide/solvent selectivity for 11, k_az/k_s, is also nearly equivalent to that of 12. A minor product of the reaction of 11 with N₃⁻, 21, contains no azide functionality but may be generated by a process in which N₃⁻ attacks 11 at the nitrenium ion center with loss of N₂ to generate a diazene 25 that subsequently decomposes into 21 with loss of another N₂. The adduct derived from attack of 2′-deoxyguanosine (d-G) on 11, 28, is a familiar C-8 adduct of the type generated from the reaction of d-G with a wide variety of arylnitrenium ions derived from carcinogenic arylamines. The rate constant for reaction of d-G with 11, k_d-G, is very similar to that observed for the reaction of d-G with 12. The similar lifetimes and chemical reactivities of 11 and 12 can be rationalized by B3LYP/6-31G(d) calculations on the two ions that show that they are of nearly equivalent stability relative to their respective hydration products. The calculations also help to rationalize the different regiochemistry observed for the reaction of N₃⁻ with 11 and its oxenium ion analogue, 13. Since 8 is the likely active metabolite of 1 and a significant number of derivatives of 1 are being developed as pharmaceutical agents, the similarity of the chemistry of 11 to that of carcinogenic arylnitrenium ions is of considerable importance. Consideration should be given to this chemistry in continued development of pharmaceuticals containing the 2-(4-aminophenyl)benzothiazole moiety.

Quinol esters 2b, 2c, and 3b and sulfonamide 4c were investigated as possible precursors to 4-alkylaryloxenium ions, reactive intermediates that have not been previously detected. These compounds exhibit a variety of interesting reactions, but with one possible exception, they do not generate oxenium ions. The 4-isopropyl ester 2b predominantly undergoes ordinary acid- and base-catalyzed ester hydrolysis. The 4-tert-butyl ester 2c decomposes under both acidic and neutral conditions to generate tert-butanol and 1-acetyl-1,4-hydroquinone, 8, apparently by an S_N1 mechanism. This is also a minor decomposition pathway for 2b, but the mechanism in that case is not likely to be S_N1. Decomposition of 2c in the presence of N₃^- leads to formation of the explosive 2,3,5,6-tetraazido-1,4-benzoquinone, 14, produced by N₃^--induced hydrolysis of 8, followed by a series of oxidations and nucleophilic additions by N₃^-. No products suggestive of N₃^--trapping of an oxenium ion were detected. The 4-isopropyl dichloroacetic acid ester 3b reacts with N₃^- to generate the two adducts 2-azido-4-isopropylphenol, 5b, and 3-azido-4-isopropylphenol, 11b. Although 5b is the expected product of N₃^- trapping of the oxenium ion, kinetic analysis shows that it is produced by a kinetically bimolecular reaction of N₃^- with 3b. No oxenium ion is involved. The sulfonamide 4c predominantly undergoes a rearrangement reaction under acidic and neutral conditions, but a minor component of the reaction yields 4-tert-butylcresol, 17, and 2-azido-4-tert-butylphenol, 5c, in the presence of N₃^-. These products may indicate that 4c generates the oxenium ion 1c, but they are generated in very low yields (ca. 10%) so it is not possible to definitively conclude that 1c has been produced. If 1c has been generated, the N₃^--trapping data indicate that it is a very short-lived and reactive species in H₂O. Comparisons with similarly reactive nitrenium ions indicate that the lifetime of 1c is ca. 20−200 ps if it is generated, so it must react by a preassociation process. Density functional theory calculations at the B3LYP/6-31G*//HF/6-31G* level coupled with kinetic correlations also indicate that the aqueous solution lifetimes of 1a−c are in the picosecond range.

Azide trapping shows that the 4‘-substituted-4-biphenylyloxenium ions 1b−d are generated during hydrolysis of 4-aryl-4-acetoxy-2,5-cyclohexadienones, 2c and 2d, and O-(4-aryl)phenyl-N-methanesulfonylhydroxylamines, 3b and 3c. In addition, the 4‘-bromo-substituted ester, 2d, undergoes a kinetically second-order reaction with N₃^- that accounts for a fraction of the azide adduct, 5d. Since both first-order and second-order azide trapping occurs simultaneously in 2d, the second-order reaction is not enforced by the short lifetime of 1d, which has similar azide/solvent selectivity to the unsubstituted ion, 1a. In contrast the 4‘-CN and 4‘-NO₂ ions 1e and 1f cannot be detected by azide trapping during the hydrolysis of the dichloroacetic acid esters 2e‘ and 2f‘ even though ¹⁸O labeling experiments show that a fraction of the hydrolysis of both esters occurs through C_alkyl−O bond cleavage. These esters exhibit only second-order trapping by azide. Correlations of the azide/solvent selectivities of 1a−d with the calculated relative driving force for hydration of the ions (ΔE of eq 4) determined at the pBP/DN*//HF/6-31G* and BP/6-31G*//HF/6-31G* levels of theory suggest that 1e and 1f have lifetimes in the 1−100 ps range. Ions with these short lifetimes are not in diffusional equilibrium with nonsolvent nucleophiles, and must be trapped by such nucleophiles via a preassociation mechanism. The second-order trapping that is observed in these two cases is enforced by the short lifetime of the cations, and may occur by a concerted S_N2‘ mechanism or by internal azide trapping of an ion sandwich produced by azide-assisted ionization. Comparison of azide/solvent selectivities of the oxenium ions 1a−c with the corresponding biphenylylnitrenium ions 8a−c shows that 4‘-substituent effects on reactivity in both sets of ions are similar in magnitude, although the nitrenium ions are ca. 30-fold more stable in an aqueous environment than the corresponding oxenium ions. The magnitude of the 4‘-substituent effects for electron-donating substituents suggest that both sets of ions are more accurately described as 4-aryl-1-imino-2,5-cyclohexadienyl or 4-aryl-1-oxo-2,5-cyclohexadienyl carbocations. Calculated structures of the oxenium ions are also consistent with this interpretation.

The title compounds serve as potential precursors to aryloxenium ions, often proposed, but primarily uncharacterized intermediates in phenol oxidations. The uncatalyzed and acid-catalyzed decomposition of 4-acetoxy-4-phenyl-2,5-cyclohexadienone, 2a, generates the quinol, 3a. ¹⁸O-Labeling studies performed in ¹⁸O−H₂O, and monitored by LC/MS and ¹³C NMR spectroscopy that can detect ¹⁸O-induced chemical shifts on ¹³C resonances, show that 3a was generated in both the uncatalyzed and acid-catalyzed reactions by C_alkyl−O bond cleavage consistent with formation of an aryloxenium ion. Trapping with N₃^- and Br^- confirms that both uncatalyzed and acid-catalyzed decompositions occur by rate-limiting ionization to form the 4-biphenylyloxenium ion, 1a. This ion has a shorter lifetime in H₂O than the corresponding nitrenium ion, 7a (12 ns for 1a, 300 ns for 7a at 30 °C). Similar analyses of the product, 3b, of acid- and base-catalyzed decomposition of 4-acetoxy-4-methyl-2,5-cyclohexadienone, 2b, in ¹⁸O−H₂O show that these reactions are ester hydrolyses that proceed by C_acyl−O bond cleavage processes not involving the p-tolyloxenium ion, 1b. Uncatalyzed decomposition of the more reactive 4-dichloroacetoxy-4-methyl-2,5-cyclohexadienone, 2b‘, is also an ester hydrolysis, but 2b‘ undergoes a kinetically second-order reaction with N₃^- that generates an oxenium ion-like substitution product by an apparent S_N2‘ mechanism. Estimates based on the lifetimes of 1a, 7a, and the p-tolylnitrenium ion, 7b, and the calculated relative stabilities of these ions toward hydration indicate that the aqueous solution lifetime of 1b is ca. 3−5 ps. Simple 4-alkyl substituted aryloxenium ions are apparently not stable enough in aqueous solution to be competitively trapped by nonsolvent nucleophiles.

Generation of the 4-biphenylyloxenium ion, 1a, from hydrolysis of 4-acetoxy-4-phenyl-2,5-cyclohexadienone, 2a, is demonstrated by common ion rate depression and azide trapping. The ion is less selective with a shorter lifetime (12 ns at 30 °C) in aqueous solution than the corresponding nitrenium ion, 6a (ca 0.3 μs at 30 °C). Its lifetime is considerably longer than those of structurally related carbenium ions. Unlike 6a, 1a reacts with N₃^-, in part, at the distal ring to generate 4‘-azido-4-hydroxybiphenyl, 5. These results are rationalized by calculations at the HF/6-31G* and pBP/DN*//HF/6-31G* levels that show that 1a is planar but is less stable than 6a relative to their hydration products by ca. 12 kcal/mol. The p-tolyloxenium ion, 1b, has not been unequivocally demonstrated to be formed in H₂O, but kinetic data show that it must be at least 10⁴-fold less stable than 1a relative to their respective 4-acetoxy derivatives. It is also calculated to be ca. 19 kcal/mol less stable than the corresponding nitrenium ion, 6b, relative to their hydration products.

The aqueous solution reactions of the title compound, 1, were examined for comparison to those previously reported for another model carcinogen N-pivaloyloxy-2-amino-α-carboline, 2. Both of these are models for the ultimate carcinogenic metabolites of 2-amino-α-carboline (AαC), a food-derived heterocyclic amine mutagen and carcinogen. The present study was undertaken to determine the effect of the N-acetyl group on the chemistry of such compounds. The N-acetyl group slows down N−O bond cleavage by a factor of (5.5 × 10³)-fold. This allows other reactions not observed in 2, or in other model carcinogens, to be observed. Among these are acyl-transfer reactions to the aqueous solvent, both uncatalyzed and catalyzed by N₃^-. In addition, the conjugate acid of 1, 1H ⁺, is subject to a spontaneous decomposition not previously observed in other esters of heterocyclic hydroxylamines or hydroxamic acids. This reaction yields the hydroxylamine, 5, and does so without the intermediacy of the hydroxamic acid, 3, and with ¹⁸O exchange from the solvent into the hydroxylamine O. This unique reaction may be caused by an intramolecular proton donation by the pyridyl N−H to the amide carboxyl that catalyzes an intramolecular nucleophilic attack by the carboxyl O of 1H ⁺. A nitrenium ion pathway can still be detected for 1, but, unlike 2 and related esters, this reaction is in competition with other processes throughout the pH range of the study.