The carcinogenic air pollutant 3-nitrobenzanthrone induces GC to TA transversion mutations in human p53 sequences

Abstract

3-Nitrobenzanthrone (3-NBA) is a potent mutagen and a suspected human carcinogen present in particulate matter of diesel exhaust and ambient air pollution. Employing an assay with human p53 knock-in (Hupki) murine embryonic fibroblasts (HUFs), we examined p53 mutations induced by 3-NBA and its active metabolite, N-hydroxy-3-aminobenzanthrone (N-OH-3-ABA). Twenty-nine immortalized cultures (cell lines) from 89 HUF primary cultures exposed at passage 1 for 5 days to 2 μM 3-NBA harboured 22 different mutations in the human DNA-binding domain sequence of the Hupki p53 tumour suppressor gene. The most frequently observed mutation was GC to TA transversion (46%), corroborating previous mutation studies with 3-NBA, and consistent with the presence of persistent 3-NBA–guanosine adducts found in DNA of exposed rodents. Six of the transversions found solely in 3-NBA-treated HUFs have not been detected thus far in untreated HUFs, but have been found repeatedly in human lung tumours. ³²P-post-labelling adduct analysis of DNA from HUF cells treated with 2 μM 3-NBA for 5 days showed a pattern similar to that found in vivo, indicating the metabolic competence of HUF cells to metabolize 3-NBA to electrophilic intermediates. Total DNA binding was 160 ± 56 per 10⁷ normal nucleotides with N²-guanosine being the major adduct. In contrast, identical treatment with N-OH-3-ABA resulted in a 100-fold lower level of specific DNA adducts and no carcinogen-specific mutation pattern in the Hupki assay. This indicates that the level of DNA adduct formation by the mutagen is critical to obtain specific mutation spectra in the assay. Our results are consistent with previous experiments in Muta Mouse and are compatible with the possibility that diesel exhaust exposure contributes to mutation load in humans and to lung cancer risk.

Introduction

Epidemiological studies have shown that occupational exposure to diesel exhaust is associated with an increased risk of lung cancer and that environmental exposure to diesel exhaust may pose a significant cancer risk to the general population (1,2).

Diesel exhaust induces lung tumours in experimental animals, an effect attributed to the particulate phase that contains many adsorbed chemicals such as nitropolycyclic aromatic hydrocarbons (3). 3-Nitrobenzanthrone (3-NBA, 3-nitro-7H-benz(de)anthracen-7-one), a member of this class of compounds, has been detected in the particulate fraction of diesel exhaust and in urban air samples (4). 3-NBA is a potent mutagen (5,6) and is carcinogenic to rodents, inducing pulmonary tumours in rats after intra-tracheal instillation (7). 3-NBA requires metabolic activation to electrophilic species to exert its genotoxic activity. Both in vitro and in vivo studies show that 3-NBA reacts with DNA through initial reduction of the nitro group to N-hydroxy-3-aminobenzanthrone (N-OH-3-ABA) primarily catalysed by cytosolic nitroreductases and can be further activated by phase II enzymes, such as acetyltransferases and/or sulphotransferases to reactive esters (Figure 1 and 8–13).

Fig. 1

Potential pathways of metabolic activation and DNA adduct formation of 3-NBA and N-OH-3-ABA (adduct numbers correspond to adduct spots in Figure 2).

DNA adduct formation in rats treated orally, intra-peritoneally or intra-tracheally with 3-NBA has been demonstrated using thin-layer chromatography (TLC) ³²P-post-labelling (8,11,14). In these studies, the same adduct pattern consisting of multiple 3-NBA-specific DNA adducts was detected in various organs of rats. As shown in Figure 1, the predominant DNA adducts formed by 3-NBA in vivo are derived from purine bases reacted with aryl nitrenium and rearranged carbenium ions of 3-NBA, which have been identified as 2-(2′-deoxyguanosin-N²-yl)-3-aminobenzanthrone (dG-N²-ABA), N-(2′-deoxyguanosin-8-yl)-3-aminobenzanthrone (dG-C8-N-ABA) and 2-(2′-deoxyadenosin-N⁶-yl)-3-aminobenzanthrone (dA-N⁶-ABA) (7,15,16). These are the most likely candidates for the induction of GC:TA and AT:TA transversion mutations observed in vivo (5).

DNA damage such as DNA adduct formation is an important first step in chemical carcinogenesis (17). Several human biomonitoring studies have reported higher levels of bulky DNA adducts among subjects heavily exposed to diesel exhaust and urban air pollution (2,18,19). This correlates with increased cancer risk (1). The main metabolite of 3-NBA, 3-aminobenzanthrone (3-ABA), has been found in urine samples of workers occupationally exposed to diesel emissions (20), suggesting that exposure to 3-NBA in diesel emissions may represent a health hazard for a wider population.

Specific covalent interaction of carcinogenic substances with DNA sequences produces characteristic patterns of mutations in biological systems. Examination of mutation profiles in a defined sequence such as the p53 tumour suppressor gene can provide clues to the nature of the carcinogens responsible for these genetic alterations (21,22). This gene is particularly suited to analysis of mutation spectra because point mutations in p53 are found frequently in human tumours, are diverse in sequence location and are directly involved in the cancer process (23). Over 20 000 occurrences of human tumour p53 mutations have been registered in the International Agency for Research on Cancer (IARC) p53 mutation database (R12, http://www-p53.iarc.fr) (24), producing a naturally arising dense spectrum of mutations for inspection. However, interpretation has been stalled by the absence of appropriate data on ‘experimentally’ induced human p53 mutations in mammalian cells with which to test inferences derived from the human tumour spectrum on the origins of somatic mutations in cancer (25).

Recently, we developed a mammalian cell mutation assay using murine embryonic fibroblasts from human p53 knock-in (Hupki) mice, which harbour functional, wild-type human p53 gene sequences in place of the corresponding homologous mouse p53 sequence at the endogenous murine p53 locus (26–28). Murine fibroblasts, in contrast to human cells, readily undergo immortalization in vitro and require only one crucial genetic step such as loss of p53 function, allowing the selection for p53 mutations. With this test system, we have shown that the carcinogen benzo[a]pyrene (B[a]P), an important promutagenic component of tobacco smoke in the lung, induces strand-biased GC:TA transversions in Hupki fibroblasts (29). This type of mutation is a hallmark of p53 mutations in lung tumours from heavy smokers (30). In contrast, aristolochic acid, the putative causative agent in Balkan endemic nephropathy and aristolochic acid nephropathy, induces primarily AT:TA transversions in the Hupki assay (31–33). This selectivity of aristolochic acid for mutations at adenine residues is consistent with the extensive formation of dA–aristolochic acid adducts in target organs of exposed rats.

In this report, we used the Hupki assay to investigate the mutation pattern induced by 3-NBA in human p53 sequences. Consistent with animal studies, adducts were detected in Hupki murine embryonic fibroblast (HUF) cells treated with 3-NBA, and the mutation spectrum obtained showed a predominance of GC:TA transversions. The specific p53 mutations found solely in 3-NBA-treated cells and not in control cells have been observed repeatedly in human lung tumours.

Materials and methods

Chemicals

3-NBA and N-OH-3-ABA were synthesized as described and authenticity of both compounds was confirmed by UV, electrospray mass spectra and high-field proton NMR spectroscopy (6,15).

Mutation selection assay

Cell viability and proliferation were determined using the trypan blue exclusion assay. Cryopreserved passage 0 fibroblasts harvested from 13.5-day-old Hupki embryos (Jackson Laboratory strain #004301, Trp53^tm1/Holl, R/R in codon 72) were distributed into 24-well plates (4 × 10⁴cells/well) containing Dulbecco's modified Eagle's medium supplemented with 10% foetal calf serum (5% CO₂) and left to adhere for 24 h. Then, two sets of 24 cultures were exposed to 0.2% dimethyl sulphoxide (vehicle control); two sets (48 and 41 cultures) were exposed to 2 μM 3-NBA and one set (36 cultures) was exposed to 2 μM N-OH-3-ABA. Higher concentrations of both compounds led to precipitates in the culture medium. Each of the 5-day treatment was interrupted once to replate the growing cultures. All cells of each 24-well culture were replated into one well of a six-well plate, where they were allowed to adhere for 9 h before treatment resumed. Subsequent subculturing was performed when cells reached near confluency; following trypsinization for 4 min (0.5 mg/ml), cells were replated at a density of 1:2 to 1:4. Signs of cellular senescence became manifest by passage 4, and typically 20–40 days were required before cells were again ready for passaging. Cultures that resumed growth thereafter and regained a homogenous morphology with increasing passage number were cultivated to passage 12–18 before harvesting for DNA extraction and mutation analysis (described below). Cultures that reached this stage of immortalization could be passaged for many months subsequently, without signs of deterioration or growth arrest. These immortalized cultures are referred to here as HUF cell lines and are coded with the prefix 3NBA-J for 3-NBA, NOH3A-A for N-OH-3-ABA and CO for the vehicle control.

DNA adduct analysis by ³²P-post-labelling

DNA was isolated from primary HUFs used in the mutation selection assay after treatment with 2 μM 3-NBA or N-OH-3-ABA. Cells of all wells were immediately trypsinized and an aliquot of the cell material pooled for DNA isolation by phenol extraction and analysis of 3-NBA–DNA base adducts using the ³²P-post-labelling method with enrichment of adducted residues by butanol extraction (34,35). Briefly, 6.25 μg of DNA were digested using micrococcal endonuclease (750 mU/sample; Sigma, Deisenhofen, Germany) and spleen phosphodiesterase (62.5 mU/sample; Calbiochem, Darmstadt, Germany) for 3 h at 37°C. An aliquot (1.25 μg) of the digest was removed and diluted for determination of normal nucleotides. Adducts were enriched by butanol extraction and labelled with (γ-³²P)adenosine triphosphate (ATP) (100 μCi/sample; MP Biomedicals, Eschwege, Germany) by incubation with T4 polynucleotide kinase (USB, Staufen, Germany) for 30 min at room temperature. Efficiency of enrichment and excess of ATP were confirmed with an aliquot of the radiolabelled sample by one-dimensional TLC (polyethylenimine–cellulose, 50 mM sodium phosphate, 0.28 M ammonium sulfate, pH 6.5). ³²P-labelled adduct nucleoside bisphosphates were separated by chromatography on polyethylenimine–cellulose sheets (Macherey–Nagel, Dueren, Germany). The following solvents were used: D1, 1 M sodium phosphate, pH 6.5; D3, 3.5 M lithium formate, 8.5 M urea, pH 3.5; D4, 0.8 M lithium chloride, 0.5 M Tris, 8.5 M urea, pH 8.0; D5, 1.7 M sodium phosphate, pH 6.0. Electronic autoradiography was performed using an Instant Imager (Canberra Packard, Meriden, CT, USA). DNA adduct levels (relative adduct labelling) were calculated as counts per minute adducts per c.p.m. normal nucleotides and determined from triplicate analysis. 3-NBA-derived DNA adducts were characterized using authentic standards of N-(2′-deoxyguanosin-8-yl)-3-aminobenzanthrone-3′-phosphate, 2-(2′-deoxyguanosin-N²-yl)-3-aminobenzanthrone-3′-phosphate and 2-(2′-deoxyadenosin-N⁶-yl)-3-aminobenzanthrone-3′-phosphate as previously described (36).

Analysis of HUF cell lines

To search for mutations in sequences encoding the DNA-binding domain of the human p53 gene, DNA was extracted from each HUF cell line and used as template for polymerase chain reactions (PCRs) employing exon-flanking primers defined by Affymetrix (Santa Clara, CA, USA) that amplify splice sites and coding segments of exons 4–9, with modifications to the primer set of exon 7 as described (37). For direct fluorescent dideoxynucleotide cycle sequencing (5′ and 3′), a different subset of primers was used, which we designed by shortening the original primer sequences from Affymetrix:

• Hupki-4R-J2: 5′-GATACGGCCAGGCATTG-3′

• Hupki-5F-J: 5′-CTTGTGCCCTGACTTTC-3′

• Hupki-6R-J: 5′-GCCACTGACAACCACC-3′

• Hupki-9R-J: 5′-AACTTTCCACTTGATAAGAG-3′

Subsequent microcapillary electrophoresis analysis of the Microcon filter-purified PCR products was performed using DNA sequencing protocols recommended by Applied Biosystems International Foster City, CA, USA. Results were confirmed by re-sequencing in reverse direction based on a separate amplification. Statistical analysis for comparison of two p53 mutation spectra was performed as described (http://www.ibiblio.org/dnam/dl_hypg.htm) (38).

Results

DNA adduct analysis

Exposure of primary HUFs at passage 1 to complete medium containing 2 μM 3-NBA or its active metabolite N-OH-3-ABA for 5 days resulted in a decrease in cell survival to 44 ± 21% (3-NBA) and 68 ± 8% (N-OH-3-ABA) of untreated control (Figure 3). Cell proliferation was reduced to 26 ± 11% (3-NBA) and 49 ± 6% (N-OH-3-ABA) and nearly halted at concentrations higher than 2 μM.

Primary HUFs treated with 3-NBA or N-OH-3-ABA were analyzed by the TLC ³²P-post-labelling method using enrichment by butanol extraction. Both treatments induced 3-NBA-specific DNA adduct patterns which consisted of a cluster of up to five adduct spots (adducts 1–5, Figure 2), of which four have been characterized using authentic standards (36). No DNA adducts were observed in the control (Figure 2C). In both treatments, the dG–N²-ABA adduct (adduct 3) represented the major adduct. The dG–C8-N-ABA adduct forms two spots (adducts 4 and 5), one of which is probably a degradation product formed during the post-labelling assay (36).The deoxyadenosine-derived adducts (adducts 1 and 2, the latter not identified yet) accounted for ∼25% of the total DNA binding in both cases; however, the dA–N⁶ adduct (adduct 1) was not detectable in HUFs treated with N-OH-3-ABA.

Fig. 2

Autoradiographic profiles of DNA adducts obtained from digests of DNA from pooled primary HUFs (also used in the mutation assay) treated at passage 1 with 2 μM 3-NBA (A), N-OH-3-ABA (B) or vehicle only (C), using the butanol enrichment method of the ³²P-post-labelling assay (spot numbers correspond to adducts in Figure 1).

Fig. 3

Cell viability and DNA adduct formation (relative adduct labelling) in HUFs after treatment with 3-NBA or N-OH-3-ABA. Values represent mean ± standard deviation, determined by quadruplicate counts of three independent cultures (cell viability) or three independent ³²P-post-labelling analyses (DNA adducts).

DNA binding in HUFs by 3-NBA was ∼100-fold higher (160 ± 56 adducts per 10⁷ normal nucleotides) than by N-OH-3-ABA (1.1 ± 0.2 adducts per 10⁷ normal nucleotides) after identical treatment conditions.

These results indicate that Hupki primary murine embryonic fibroblasts are capable of converting 3-NBA and N-OH-3-ABA into electrophilic intermediates, which then bind covalently to purine nucleobases in DNA, forming adducts identical to those found in organs of rats treated by 3-NBA and N-OH-3-ABA.

Mutation assay

Of the 89 primary HUF cell cultures treated with 2 μM 3-NBA for 5 days, 76 became immortalized at passage 12–18, hereafter referred to as cell lines. During senescence, there was a slow attrition of cells and alterations in cell morphology. The cells increasingly assumed characteristics typical of senescence: enlarged, flattened appearance, irregularly shaped cell borders and aberrant nuclei. As the cultures resumed growth and developed into immortalized lines, homogeneous morphologies re-emerged. Some lines regained an elongated morphology typical of fibroblasts, while others acquired more rounded, dense forms; however, no consistent morphological type has been associated with treated and untreated, mutated and non-mutated cell lines. The 76 immortalized cultures were examined for the presence of mutations in the human DNA sequences of Hupki p53 by PCR amplification of exons 4–9 and flanking segments and direct dideoxy sequencing of PCR products.

Nineteen of the 76 cell lines derived from 3-NBA-treated cultures harboured a total of 29 mutations, consisting of 22 unique base changes (29%) in the human-derived p53 DNA-binding domain, whereas 5 of 48 cell lines from 48 unexposed cultures harboured six unique p53 mutations (12.5%) (Table I). Treatment of 36 primary HUF cell cultures with 2 μM N-OH-3-ABA for 5 days resulted in 36 immortalized lines of which four contained p53 mutations (Table I). These four mutations have previously been observed in untreated control HUFs. Some mutations in cell lines derived from 3-NBA-treated cultures were identified in more than one cell line, and eight cell lines harboured more than one mutation. All mutations but one was either at a splice site or in coding sequences, resulting in an amino acid substitution (Table I).

Table I

p53 gene mutations in HUF cell lines after treatment with 3-NBA, N-OH-3-ABA and from untreated primary cells

Cell line	Mutation			Amino acid substitution	Mutationa/codona (%)
3-NBA treated
3NBA-J-253	ag	to gg	A to G	Splice site
3NBA-J-131	CGC	to CGA	G to Tb	R 72 Rc	1/16 (6, 3)
3NBA-J-246	cag	to cat	g to t	Splice site
3NBA-J-256	CCT	to CAT	G to Tb	P 128 H	0/28 (0, 0)
3NBA-J-135, -136, -145, -148	TAC	to TGC	A to G	Y 163 C	179/255 (70, 2)
3NBA-J-134	GTG	to TTG	G to T	V 173 L	67/205 (32, 6)
3NBA-J-254	TGC	to TGG	G to Cb	C 176 W	29/377 (7, 7)
3NBA-J-102	gagtg	to tagtg	G to T	Splice site
3NBA-J-254	ATC	to TTC	A to T	I 195 F	40/157 (25, 4)
3NBA-J-226	CGA	to CTA	G to T	R 196 Lc	4/255 (1, 6)
3NBA-J-260	TAT	to AAT	A to Tb	Y 205 N	11/175 (6, 3)
3NBA-J-260	TTG	to GTG	A to Cb	L 206 V	0/22 (0, 0)
3NBA-J-120, -256	TAC	to TGC	A to G	Y 236 C	84/153 (54, 9)
3NBA-J-229, -230	GGC	to GCC	G to C	G 244 Ac	19/250 (7, 6)
3NBA-J-102, -229	GGC	to GCC	G to C	G 245 Ac	31/722 (4, 3)
3NBA-J-256	ATG	to AAG	A to Tb	M 246 K	15/140 (10, 7)
3NBA-J-102	AAC	to ACC	A to C	N 247 T	8/90 (8, 9)
3NBA-J-247	ACA	to AAA	G to Tb	T 256 K	5/47 (10, 6)
3NBA-J-260	TTT	to GTT	A to Cb	F 270 V	38/111 (34, 2)
3NBA-J-225	CGT	to CTT	G to T	R 273 Lc	154/1572 (9, 8)
3NBA-J-120, -256	GTT	to TTT	G to T	V 274 F	27/97 (27, 8)
3NBA-J-265	AGA	to ATA	G to T	R 280 I	20/250 (8, 0)
N-OH-3-ABA treated
NOH3A-A-41	TGC	to TGG	G to Cb	C 135 W	32/213 (15, 0)
NOH3A-A-60	TAC	to TGC	A to G	Y 163 C	179/255 (70, 2)
NOH3A-A-42	GGC	to GCC	G to C	G 244 Ac	19/250 (7, 6)
NOH3A-A-51	GGC	to GCC	G to C	G 245 Ac	31/722 (4, 3)
Untreated
CO-J-214, -224	TGC	to TGG	G to Cb	C 135 W	32/213 (15, 0)
CO-J-204	GCC	to CCC	G to C	A 138 P	28/111 (25, 2)
CO-J-220	AAG	to AAC	G to C	K 139 N	6/52 (11, 5)
CO-J-220	GTT	to GGT	A to Cb	V 147 G	10/47 (21, 2)
CO-A-15	TAT	to TCT	A to Cb	Y 205 S	18/175 (6, 3)
CO-J-224	AGT	to AGG	A to Cb	S 240 R	9/78 (11, 5)

a

Data from the IARC TP53 database version R12 (mutated codon in lung tumours as a fraction of all tissues).

b

Transcribed strand (template).

c

Codon: CpG site; highlighted are mutations found in 3-NBA-treated cultures and observed repeatedly in the p53 gene of human lung tumours. Four of these have a thymine 3′ to the mutated base (highlighted in pairs).

GC to TA transversions were the most common base change in cell lines derived from 3-NBA treatment (10 of 22, 45.5%, Table I and Figure 4), corresponding to the type of mutation anticipated from dG–3-NBA adducts in target sequences. Seven of the 10 GC:TA transversions occurred at base pairs where the wild-type guanine was on the non-transcribed strand (Table I). Three base substitutions were AT:TA transversions, consistent with formation of deoxyadenosine adducts.

Fig. 4

p53 mutation types in HUF cell lines derived after treatment with 3-NBA or vehicle only (22 and 52 mutations). Control mutations are pooled from several experiments [reported in (27), (29), (32), (37) and Hollstein, M.C., unpublished].

Discussion

The ability of Hupki cells to metabolize 3-NBA to active intermediates was demonstrated by DNA adduct analysis using TLC ³²P-post-labelling of genomic DNA from primary HUFs treated with 3-NBA and N-OH-3-ABA. This revealed the presence of five 3-NBA-specific adducts, of which four have been characterized using authentic standards and detected previously in vivo (5,8,11,14,36,39). The adduct patterns were similar to those found previously by Arlt et al. (40) in various human cell lines and typical for DNA adducts induced by 3-NBA and N-OH-3-ABA in vivo (11,14,39,40). Previously, HUF cells have been shown to be capable of metabolizing both aristolochic acid and B[a]P to active intermediates. Hupki embryonic fibroblasts have also been demonstrated to express a number of phase I and II metabolic enzymes, including NAD(P)H:quinone oxidoreductase, the nitroreductase responsible for the metabolism of 3-NBA (27).

After identical treatment protocols, the DNA adduct levels generated by 3-NBA in HUFs were considerably higher than those generated by N-OH-3-ABA, which is consistent with results from human cell lines (40) and most likely due to the higher reactivity and/or lower cell uptake of N-OH-3-ABA.

In previous studies with carcinogens where the Hupki assay was used successfully to generate experimental mutation patterns, DNA adduct levels in carcinogen-exposed HUFs were consistently higher than 10 adducts per 10⁷ normal nucleotides. In the present study, 3-NBA treatment resulted in an adduct level of 160 per 10⁷ normal nucleotides and led to characteristic carcinogen-associated p53 mutations, whereas the same treatment with N-OH-3-ABA showed a much lower level of DNA adducts (1.1 per 10⁷ normal nucleotides) and no specific mutation pattern. Therefore, we conclude that DNA adduct levels are crucial assay factors for the induction of carcinogen-specific p53 mutations.

DNA adducts give rise to mutations in proliferating cells when DNA adducts undergo mismatched base pairing during DNA replication. The concentration of 3-NBA used in this mutagenesis study caused a decrease in cell proliferation of ∼75%, allowing some cells to continue to divide in the presence of 3-NBA.

Two significant results obtained from the current study are the prevalence of types of mutation in HUFs after treatment with 3-NBA and their ratio compared to that of the DNA adducts examined here. As observed in previous studies, immortalized cell lines arising from HUFs harbour p53 mutations that match p53 mutations in human tumours (25,29,32,37,41). These mutations are formed in higher prevalence in the 3-NBA-treated cohort (29%) as compared to the control (12.5%) consistent with previous findings in HUFs (29,32).

Statistical analysis of mutation distribution indicates a statistically significant difference between 3-NBA-treated and control spectra (Table I and Figure 4P = 0.028). The total number of mutations compared in each group is still relatively small, however, and additional mutations studies are required to determine whether preferred mutation sites are distinct.

Unexpectedly, the predominant mutation in control HUFs is GC:CG transversion (46.2%) (29,37 and Hollstein, M.C., unpublished observations). GC:AT transitions predominate in spontaneous mutation spectra in most experimental mutation assays regardless of the reporter gene or organism. The reasons why GC:CG transversions arise frequently in cell lines from untreated HUFs are not yet clear. In standard cell culture incubators (ambient air mixed with CO₂ at 5%), cells are exposed to exceptionally high levels of oxygen in comparison to normal tissue in vivo conditions (42), which may produce free radical mutagens that contribute to the spontaneous spectrum of p53 mutations observed in Hupki cells. In clear contrast to control cell lines, 3-NBA-treated HUFs harbour GC:TA transversions as the main mutation (45.5%, Table I and Figure 4; Fisher's exact test, P < 0.001). GC:TA transversions were the major mutation found in shuttle vector plasmids propagated in human cells treated with N-acetoxy-3-aminobenzanthrone, the activated metabolite of 3-NBA (43). This class of mutation was detected at a high frequency by Arlt et al. (5) in livers of 3-NBA-treated Muta Mouse mice and in the lungs of gpt-delta transgenic mice following inhalation of diesel exhaust (44), and occurred preferentially at CpG sites, as observed previously with other mutagens (45). In our study, five of the 22 unique mutation sites identified in 3-NBA cell lines were at CpG dinucleotides. A plausible explanation for the origin of GC:TA transversions is the insertion of adenine opposite an uninformative site such as a 3-NBA–dG adduct (46) by DNA polymerase and is in agreement with results obtained by the polymerase stop assay, (43) demonstrating that activated 3-NBA preferentially binds to guanine residues.

In addition, we detected AT:TA transversions in three of the p53 mutant cell lines derived from cultures treated with 3-NBA (3/22, 13.6%, a significant increase compared to untreated HUF cell lines 0/52, Fisher's exact test, P < 0.024, Table I and Figure 4). The ratio of GC:TA to AT:TA transversions (3:1) mirrors the ratio of dG/dA adducts (75:25%) derived after treatment with 3-NBA or N-OH-3-ABA. A similar correlation in ratios of 3-NBA-derived purine adducts to transversion mutations is found in liver tissue of the Muta Mouse, where the ratio of dG/dA adducts is 6:1 and that of corresponding GC:TA and AT:TA mutations is 5:1 (5).

Analogous to our study with 3-NBA, cell lines derived from HUFs treated with B[a]P, a major mutagenic component of cigarette smoke and a constituent of diesel exhaust, also harboured GC:TA transversions as the predominant mutation (41.2%) (29,37). Nonetheless, the mutation patterns of 3-NBA and B[a]P-treated HUFs are significantly different (P < 0.05).

With respect to codon distribution, mutations in 3-NBA- and B[a]P-treated HUFs reveal few similarities (codons also found mutated in untreated HUFs excluded). Only one mutation was identical: hotspot codon 273, CGT to CTT. This mutation represents 1.9% (52/2709) of the mutations reported in lung cancer in the p53 IARC database (R12) and codon 273 is one of the five most frequently mutated codons in human tumours overall (24). Six GC:TA and AT:TA transversions specific to cell lines derived from 3-NBA-treated cultures (highlighted in Table I) have been observed repeatedly in the p53 gene of human lung tumours (IARC database, R12) (24).

When examining mutations unique to 3-NBA-exposed cultures, we noted a similarity in immediate sequence context, with a thymine present 3′ of the mutated base in four of the six mutants (highlighted in Table 1).

In summary, the most common type of base change induced by 3-NBA in human p53 sequences of Hupki cells is GC:TA as predicted from mutation studies in other test systems with non-cancer-related reporter genes. This finding encourages further studies with 3-NBA in the Hupki assay in order to explore whether this carcinogen causes GC:TA signature mutations at specific locations in the p53 gene, serving as an indicator of exposure.

Funding

The Waltraud-Lewenz PhD student fellowship (to J.B.); Yorkshire Cancer Research UK.

J. vom Brocke wishes to thank K. Mühlbauer, M. Reinbold and A. Weninger for advice and excellent technical support and T. Nedelko for the preparation of primary HUFs. Conflict of interest statement: None declared.