Mutational Signature of the Proximate Bladder Carcinogen N-Hydroxy-4-acetylaminobiphenyl: Inconsistency with the p53 Mutational Spectrum in Bladder Cancer¹

Historically, aromatic amines are among the first chemical carcinogens that have been implicated in human cancer (1). In 1895, Rehn (2) reported an increased incidence of urinary bladder cancer in German dyestuff workers, which according to Leichtenstern (3), could be ascribed to their exposure to naphthylamines. Epidemiological studies followed these reports and showed an unambiguous association between urinary bladder cancer and exposure to 2-naphthylamine, benzidine and 4-ABP³ (4). This was further confirmed in experimental animal models wherein aromatic amine administration led to urinary bladder tumorigenesis (5, 6, 7). On the basis of these observations, the industrial use of carcinogenic aromatic amines was substantially curtailed, but by no means prohibited, in most parts of the world (8). Today, occupational exposure to aromatic amines still occurs in chemical and mechanic industries, e.g., in rubber, coal, gas, cable, and cosmetic productions, and aluminum transformation (9). Also, tobacco smoke constitutes a major source of aromatic amines because both main stream and side stream smoke contain nanogram quantities of 2-toluidine, 2-naphthylamine, and 4-ABP (10). It is widely believed that 4-ABP is a putative tobacco carcinogen responsible for bladder cancer development in smokers (1, 11).

As a model compound, 4-ABP has been extensively studied to explore the underlying mechanism of aromatic amine-induced bladder cancer (1). 4-ABP undergoes biotransformation to yield reactive metabolites, which are capable of binding covalently to DNA and forming promutagenic DNA adducts (12). The metabolic pathway of 4-ABP consists of an initial N-oxidation step catalyzed by the cytochrome P450 enzyme, CYP1A2, in the liver (13). The resulting hydroxyarylamine is also generated both via N-methylation or peroxidation but to a lesser extent (14, 15). This direct-acting mutagen can either interact with hemoglobin and form stable sulfinamide adducts (16) or circulate freely or in conjugation with acetate, sulfate, or glucuronate (17, 18, 19). The glucuronate conjugates can subsequently, be excreted through the kidney, where they are hydrolyzed at the acidic pH of urine (18, 19). Upon arrival in the urinary bladder, these hydroxylated amines can either bind directly to the DNA of the uroepithelial cells or be further bioactivated into highly reactive species by O-acetylation (12). N-Acetylation of 4-ABP and O-acetylation of N-hydroxy-4-ABP, respectively, are catalyzed by two related NATs, NAT1 and NAT2. It has been shown that NAT1 is the main contributor to the O-acetylation of N-hydroxylated amines in human bladder (20). NAT1 and its murine homologue each have a ubiquitous tissue distribution in humans and mice and are expressed early in the development of the mouse (21). Analyses in both rodent tissues and humans have demonstrated the presence of efficient activating systems for 4-ABP in virtually all tissues studied (18, 20, 22).

The increasingly popularized theory of “fingerprinting” suggests that specific carcinogens leave a characteristic mutational signature on DNA (23). Studies have shown that the p53 tumor suppresser gene is frequently mutated in human urinary bladder cancer (24, 25, 26). However, no distinct pattern of mutation has thus far been observed in the p53 mutational spectrum in aromatic amine-associated bladder cancers relative to bladder cancers not linked to occupational or lifestyle exposure to aromatic amines (27, 28, 29). Also, scant data are available on the mutational signature of 4-ABP and consist mostly of data from in vitro treated shuttle vectors and more recently from exon 3 of the HPRT gene (30, 31, 32, 33). The HPRT study, however, presented only 13 sequenced mutations from N-OH-AABP-treated cells (34). Thus, a comprehensive study of aromatic amine mutagenesis on a chromosomal gene is still missing. This inspired us to investigate the mutational spectrum of arylamines in a relevant system wherein both DNA adduct formation and mutation induction can be simultaneously determined. The transgenic Big Blue mouse system offers a promising venue to study mutations induced by DNA-damaging agents in vivo. Recoverable lambda-LIZ shuttle vectors carrying mutational target genes, e.g., LacI, cII, and cI inserted into the genome of these mice, can be screened after the animal or its derived cell line is exposed to a carcinogen of interest (34, 35). The cII mutagenesis assay in this system has the advantage of targeting a transgene whose CpGs are methylated (36). This is of importance because most mutational hotspots in the p53 gene are at methylated CpG sites (37). In the present study, we analyzed the mutational spectrum and mutation frequency of the cII gene in embryonic fibroblasts of Big Blue mice treated with N-OH-AABP. We also mapped the formation of DNA adducts along the cII locus using the TD-PCR technique.

Early-passage embryonic fibroblasts of Big Blue mice were treated with increasing doses of N-OH-AABP (Midwest Research Institute, Kansas City, MO) dissolved in DMSO for 24 h (for several hours before treatment and during the course of treatment, cultures were kept in serum-free DMEM). Subsequently, cells were washed with serum-free DMEM, and viability was checked by trypan blue dye exclusion. To allow DNA replication and fixation of possible mutations, the cells were cultured in DMEM plus 10% fetal bovine serum for 5 days, whilst the medium was changed every other day. The cells were harvested by trypsinization, washed with isotonic saline, pelleted by centrifugation, and finally preserved at −80°C until further analysis. All experimental settings were in triplicates.

Genomic DNA was isolated using standard phenol and chloroform extraction and ethanol precipitation techniques (38). The DNA was dissolved in TE buffer (10 mm Tris-HCl, 1 mm EDTA, pH 7.5) and kept at −80°C until further analysis.

Spontaneous and induced mutation frequencies were measured using the lambda select-cII mutation detection system for Big Blue rodents (Stratagene, La Jolla, CA). The assay was performed by rescuing the λ LIZ shuttle vectors from the genomic DNA (∼5 μg) and packaging them into viable phage particles using Transpack packaging extract according to the manufacturer’s instructions (Stratagene). Subsequently, the phages were preadsorbed to G1250 Escherichia coli and then plated on TB1 agar plates. The plates were incubated at 24°C for 48 h and at 37°C for overnight (regarded as selective and nonselective conditions, respectively). The cII mutant frequency was calculated as the ratio of the number of plaques formed on the selection plates to that on the nonselection plates. As recommended by the manufacturer (Stratagene), a minimum of 3 × 10⁵ rescued phages were screened for each experimental setting. The intra- and interassay variations were 8.2 and 14.3%, respectively.

Putative mutant cII plaques were replated under the selective conditions on a secondary TB1 agar plate. The verified plaques were PCR amplified using the lambda select-cII sequencing primers according to the manual instructions (Stratagene). The PCR products were purified with QIA quick PCR purification kits (Qiagen GmbH, Hilden, Germany) and sequenced in both directions using a Big Dye terminator cycle sequencing kit on an ABI-377 DNA Sequencer (ABI Prism; PE Applied BioSystems, Foster City, CA). Confirmed mutations on both sense and antisense strands were considered for further evaluation.

TD-PCR for the entire cII gene in both directions was performed as described earlier with some modifications (38, 39). Briefly, genomic DNA (∼2 μg) was used as a template, and single-stranded products were made by repeated primer extensions. The extension protocol consisted of: forward primer, U1, 5′-AATCGAGAGTGCGTTGCTT-3′, T_m = 49.9°C; and reverse primer, L1, 5′-GACCTCAGAACTCCATCTGG-3′, T_m = 48.2°C in a mixture of Vent^(exo-) DNA polymerase (New England Biolabs, Inc., Beverly, MA) and a thermocycler setting of 3 min at 95°C, 3 min at 53°C, 5 min at 72°C, and 10 cycles of (1 min at 95°C, 3 min at 53°C, and 2 min at 72°C). Subsequently, DNA was precipitated in the presence of glycogen and subjected to homopolymeric ribotailing and adapter ligation. The ligated fragments were PCR amplified using the following primers: forward primer, U2, 5′-GCGTTGCTTAACAAAATCGCAATGCT-3′, T_m = 63.1°C; and reverse primer, L2, 5′-CTCCATCTGGATTTGTTCAGAACGCT-3′, T_m = 60.7°C. The thermocycler was set for 3 min at 95°C, 2 min at 60°C, 3 min at 72°C, 18 cycles of (45 s at 95°C, 2 min at 60°C, and 10 min at 72°C), 45 s at 95°C, 2 min at 60°C, and 10 min at 72°C. Final primer extension of the PCR products was performed with fluorescence infrared dye-labeled primers [IRD-700, IRD-800] (LI-COR, Inc., Lincoln, NE): forward primer, U3, 5′-GCAATGCTTGGAACTGAGAAGACAGC-3′, T_m = 61.4°C; and reverse primer, L3, 5′-TTCAGAACGCTCGGTTGCCGC-3′, T_m = 64.0°C. The thermocycler setting was 2 min at 95°C and 6 cycles of (45 s at 95°C, 2 min at 60°C, and 3 min at 72°C). Gel electrophoresis of the labeled product was done in an IR² Long Ranger 4200 system with simultaneous detection (LI-COR, Inc.).

Cytological examination of the embryonic fibroblasts of the Big Blue mouse treated with increasing concentrations of N-OH-AABP (10, 20, 40, 80, 160, 320, and 640 μm) showed a dose-dependent decrease in cell viability at the end of 24-h treatment (Fig. 1 A). The survival rate was expressed as the ratio of the viable cells in N-OH-AABP-treated samples to that in DMSO-treated samples. As the survival rate declined drastically at the 640 μm N-OH-AABP dose (>95% decrease), this set of samples was excluded from further analysis.

Treatment of Big Blue mouse embryonic fibroblasts with N-OH-AABP increased the cII mutation frequency in a dose-dependent fashion (Fig. 1,B). The fold increases in the cII mutation frequency over the background at 10, 20, 40, 80, 160, and 320 μm N-OH-AABP doses were 2.0, 3.2, 5.0, 5.4, 8.9, and 12.8, respectively. On the basis of these data, a mutational spectrum was determined for the 320 μm N-OH-AABP and for the DMSO-treated samples. Sequence analyses of 141 induced cII mutant plaques (isolated from N-OH-AABP-treated cells) and 145 spontaneously derived cII mutant plaques (isolated from DMSO-treated cells) revealed that 13 (9.2%) and 12 (8.3%) of the plaques, respectively, were nonmutant cII. Overall, single base substitutions comprised 87% of the induced cII mutations and 74% of the spontaneous cII mutations. Of these, 63 and 36%, respectively, occurred at guanine residues along the cII gene. In both induced and spontaneous cII mutations, 28 of all mutations (including insertions and deletions) were at CpG sites. However, C to T transitions at CpG sites were much more prevalent in the spontaneous cII mutations (36% versus 7%). Whereas G to T transversions predominated in the induced cII mutations (47%; including G to T and C to A mutations), insertion was the most spontaneously derived cII mutation (19%; Fig. 2). A single G insertion/deletion event in the homopolymeric runs of six guanines along the nucleotide positions 179–184 appeared to be a major jackpot mutation accounting for >21% of all spontaneous cII mutations. The same stretch of guanines was also among the N-OH-AABP-induced mutational hotspots because >5% of all base substitutions including 8% of all G to T transversions and 13% of all G to A transitions in the induced cII mutations occurred at this location. Another jackpot mutation was a G to C transversion at nucleotide position 211, which contributed to >5% of all spontaneous mutations. The spectra of spontaneous and N-OH-AABP-induced cII mutations are shown in detail in Fig. 3.

By and large, the N-OH-AABP-induced mutational spectrum was characterized by several common mutation sites (defined as four or more mutations at one position). They were localized to nucleotide positions 42, 122, 125, 132, 167, and 179–184. Most of these hotspots were at guanine bases within 5′AG sequences. This sequence context of N-OH-AABP-induced mutations (Fig. 4) was strikingly different from that of the BPDE-induced mutations established in our previous work (40). As shown in Fig. 4, 58% of all G to T mutations induced by BPDE occurred at methylated CpG dinucleotides, which were also the preferential binding sites for BPDE (40), whereas only 24% of N-OH-AABP-induced G to T transversions were at such sites (P < 0.001; χ² test).

To determine DNA adduct formation, a subset of cell cultures was treated with 320 μm N-OH-AABP for 1, 2, and 24 h and subsequently harvested and processed as described above. TD-PCR analysis of these samples showed that DNA adduct formation was already initiated at specific nucleotide positions (97, 153, 165, and 205–207) after 1 h of treatment (Fig. 4). At the end of 24 h of treatment, DNA adducts accumulated at nucleotide positions 97, 125, 132, 179–184, and 205–207. At the same time, removal of DNA adducts at nucleotide positions 153 and 165 were observed as well (Fig. 5). The observed adduct hotspots at nucleotide positions 125, 132, 179–184, and 206 correlated with the presence of frequent mutation sites (Fig. 3).

Our mutational spectrum analysis showed that more than one-fifth of all spontaneous mutations were accounted for by a jackpot insertion/deletion mutation in a stretch of six guanines at nucleotide positions 179–184. This phenomenon, already reported by others, has been ascribed to the slippage of DNA polymerase at nucleotide repeats during DNA synthesis (41, 42, 43, 44). These jackpot mutations are likely to occur early in development and undergo clonal expansion such that most cells from a single tissue contain the same type of mutation (45). The G insertion also occurred frequently in the N-OH-AABP-induced mutations. Because this run of guanines is a site of frequent adduct formation (Fig. 5), it is possible that a fraction of these insertion events was triggered by 4-ABP-dG adducts, causing a frameshift mutation. It has been shown that almost all small insertions/deletions in various model systems occur at monotonic runs of two or more identical bases or at repeats of 2–8 bp DNA motifs, either in tandem or nontandem repeat sequences (46). The p53 gene in human lung cancer frequently harbors frameshift mutations at codons 151–159, wherein three base substitution hotspots are encompassed (codons 154, 157, and 158). The hypermutability of this GC-rich sequence with multiple runs and direct repeats was assumed to be carcinogen mediated (47). In an intensive search to address this issue, our group has unequivocally shown that the p53 mutational hotspots in lung cancer colocalize with the preferential binding sites of tobacco-derived carcinogens, polycyclic aromatic hydrocarbons (48, 49).

Although mutations at CpG dinucleotides occurred at the same frequency in spontaneous and induced cII mutations, C to T transitions at these sites were more frequent in the spontaneously derived cII mutations (36% versus 7% of all mutations at CpG sites). These mutations are assumed to arise from either the spontaneous or enzyme-mediated deamination of 5-methylcytosine (5mC) yielding 5-methylcytosine→T, or the cytosine-5-DNA methyltransferaseinduced deamination of cytosine resulting in C→U→T (50, 51, 52).

On the other hand, the most prevalent N-OH-AABP-induced cII mutation was the G to T transversion. This type of induced mutation has also been observed in other studies where single- and double-stranded DNA of shuttle vectors has been exposed to 4-ABP or its acetoxy metabolites (30, 31, 32). Also, a G to T transversion, together with a 7-bp repeat insertion, was a dominant type of mutation induced by N-OH-AABP in exon 3 of the HPRT gene in human lymphoblastoid cells (33). Of course, we acknowledge that the cII mutagenesis assay in the Big Blue mouse system is mainly suitable for detecting small point mutations. Practically, large insertions/deletions within the cII gene, especially those interfering with the adjacent O gene, which is essential for lytic growth (53), will go undetected in this system. Also, G to C transversions commonly observed in shuttle vector studies (31, 33) were <10% in our study. It is worth mentioning that the proximate form of another aromatic amine, N-hydroxy-2-acetylaminofluorene, induced mostly G to T transversions in the cII as well as in the HPRT gene (54). Both in vitro and in vivo, 4-ABP derivatives have been shown to react mainly with the C8 position of guanine, forming a major DNA adduct, dG^8-ABP Refs. 20, 55, 56). Theoretical and spectroscopic analyses have documented that dG^8-ABP can readily adopt a “syn” conformation around the guanine-deoxyribose linkage (57, 58). This is most pronounced in destabilized or unwound DNA helices such as those formed during replication. The syn conformation may place the O6 and N7 atoms of the modified guanine in a position to mispair with N1 and N2 of a guanine or with an N6 and N1 of an adenine (iminotautomer) in the complementary strand. This in turn can give rise to G to C or G to T transversions, respectively (59, 60). Our data support the theory of guanine susceptibility for 4-ABP-derived mutations because 63% of all induced base substitutions in our study were at G residues along the cII gene. This was also reiterated in the HPRT study (33), where a stretch of Gs at nucleotide position 207–212 appeared to be a major mutational hotspot for N-OH-AABP.

Moreover, our mapping data of N-OH-AABP-induced DNA adducts along the cII gene were mostly in good agreement with the mutation data. In fact, of five established DNA adductation sites, four including those at nucleotide positions 125, 132, 179–184, and 205–207 were major mutation sites for N-OH-AABP, especially for G to T transversions. The other preferential DNA adduction site (nucleotide position 97) as well as those whose DNA adducts were totally removed by the end of treatment (nucleotide positions 153 and 165) did not harbor any mutations.

To explore the significance of our findings for human cancer, we analyzed the updated database of p53 mutations (61) in human bladder cancer. Compared with the predominantly G to T transversions induced by N-OH-AABP in the cII gene, the p53 mutations were mostly G to A transitions (53%; including G to A and C to T), whereas G to T transversions were only of minor prevalence (11%; Fig. 6,C). When we stratified all bladder cancer cases according to occupational exposure to aromatic amines and/or exposure to tobacco smoke, this pattern of mutations still remained unchanged (Fig. 6, A, B, D, and E). Additionally, of five established mutational hotspots for the p53 gene in bladder cancer, four including codons 175, 248, 273, and 285 had mainly G to A transitions (on average, 72%) and the other (codon 280) had mostly G to C transversions (57%) and G to A transitions (38%). The majority of mutations in three of these hotspots (codons 175, 248, and 273) were at a methylated CpG site. In our study, however, neither the preferential DNA adduction sites nor the induced mutational hotspots seemed to be biased toward methylated CpG dinucleotides (Fig. 3). Altogether, it appears that the mutational spectrum of the p53 gene in human bladder cancer is in sharp contrast with the mutational signature of N-OH-AABP in the cII gene. Of course, we should interpret these comparative analyses cautiously because: (a) statistically, the small number of subjects in the stratified populations might have overshadowed the subtle effects of aromatic amine exposure and/or tobacco smoking; (b) methodologically, although the cII gene is similar to p53 in its content of methylated CpG dinucleotides, the target cells used in our study (mouse embryonic fibroblasts) are different from the target cells of human bladder cancer (urothelial stem cells). Nevertheless, we believe that the basic mechanisms of mutagenesis induced by N-OH-AABP, i.e., targeted adduct formation, repair, and replicative bypass, should be conserved across mammalian species and cell types. The N-acetyltransferases that convert N-OH-AABP to the N-acetoxy derivative are conserved between humans and mice and are expressed in a wide range of tissues (21); (c) whereas transcription-coupled repair can significantly impact upon induced mutations in endogenous genes in mammalian cells (62, 63, 64), it is almost certain that the cII transgene is not transcribed after its integration into the genome of the rodent (65). However, transcription-coupled repair is not expected to influence the type of mutations produced; and (d) most studies have thus far shown mutations in the p53 gene only in advanced-stage bladder neoplasia (24, 25, 66, 67, 68). Taken together and given the similar pattern of mutations induced by all aromatic amines tested thus far, the mechanistic basis for the involvement of this class of chemicals in human bladder cancer still remains to be determined. Currently, work in our laboratory is under way to map the location of N-OH-AABP-induced DNA adducts along the p53 gene in human uroepithelial cells.

Aromatic amines also have been implicated in the development of lung cancer in cigarette smokers. The aromatic amine fraction was shown to contribute to 58% of cigarette smoke mutagenicity (69). 4-ABP is present in cigarette smoke at a concentration of ∼5 ng/cigarette (10, 70). 4-ABP-DNA adducts have been found in human lung (71, 72). However, the rates of microsomal 4-ABP N-oxidation were below the limit of detection, which was consistent with the lack of detectable CYP1A2 in human lung (72). N-Hydroxy-4-ABP O-acetyltransferase activity was detected in cytosols, and NAT1 and NAT2 contributed to this activity. These data suggested that 4-ABP-DNA adducts in human lung may result from some environmental exposure to 4-nitrobiphenyl. The bioactivation pathways seemed to involve metabolic reduction to N-hydroxy-4-ABP and subsequent O-acetylation by NAT1 and/or NAT2. Alternatively, 4-ABP inhaled with cigarette smoke may be the subject of peroxidation by myeloperoxidase in human lung. The myeloperoxidase activity appeared to be the highest peroxidase activity measured in mammalian tissue and is consistent with the presence of neutrophils and polymorphonuclear leukocytes surrounding particulate matter derived from cigarette smoke (72).

Lung cancer is characterized by a high frequency of G to T transversions in the p53 gene (30%), which is different from most other cancers where G to T transversions are only on the order of 10%. These G to T transversions have been ascribed to polycyclic aromatic hydrocarbons, such as benzo[a]pyrene, present in cigarette smoke (48, 73). However, theoretically, aromatic amines would also be candidates for producing G to T transversions in lung cancer. To further understand mechanistic differences in the induction of G to T transversions by polycyclic aromatic hydrocarbon compounds and by aromatic amines, we conducted a nearest neighbor base analysis of these mutational events (Fig. 4). It became apparent that the sequence specificities of N-OH-AABP- and BPDE-induced mutations in the cII gene differ from one another in that the preferred 5′ base of BPDE-induced G to T transversions is 5-methylcytosine, and a 5′thymine is disfavored. This might be attributable to the preferential formation of BPDE adducts at methylated CpG dinucleotides (37). In contrast, N-OH-AABP has no preference for 5-methylcytosine as a 5′ neighbor base of G to T transversions. Instead, adenines and thymines are more commonly represented (Fig. 4). Both N-OH-AABP and BPDE have a similar specificity for the 3′ nearest neighbor bases. Altogether, the sequence specificity of BPDE seems to be much more consistent with the p53 mutational spectrum in lung cancer, where >50% of the G to T transversions occur at 5-methylcytosine as the 5′ nearest neighbor base (40).

In summary, we present the first comprehensive analysis of N-OH-AABP mutagenesis on a chromosomal gene. We conclude that N-OH-AABP leaves a distinctive mutational signature in the cII gene in the Big Blue mouse system. The characteristic mutagenicity of N-OH-AABP in this system is well reflected by its DNA adduct formation profile because all but one major DNA adduction sites are sites of common induced mutations. However, the specific pattern of N-OH-AABP-induced mutations in the cII gene is largely at odds with the mutational spectra of the p53 gene in human bladder and lung cancers, thereby questioning the involvement of aromatic amines in human carcinogenesis.

We thank Dr. Hsui-Hua Chen for expertise and excellent assistance in TD-PCR analysis.