Exposure to beta-carbolines norharman and harman.

The aromatic beta-carbolines norharman and harman have been implicated in a number of human diseases including Parkinson's disease, tremor, addiction and cancer. It has been shown that these compounds are normal body constituents formed endogenously but external sources have been identified. Here, we summarise literature data on levels of norharman and harman in fried meat and fish, meat extracts, alcoholic drinks, and coffee brews. Other sources include edible and medicinal plants but tobacco smoke has been identified as a major source. Exposure levels from these different dietary sources are estimated to a maximum of 4 microg norharman per kg body weight (bw) per day and 1 microg harman per kg bw per day. Exposure via tobacco smoke depends on smoking habits and type of cigarettes but can be estimated to 1.1 microg/kg bw for norharman and 0.6 microg/kg bw for harman per package of cigarettes smoked. Studies on toxicokinetics indicate that inhalative exposure leads to a rapid increase in plasma levels and high bioavailability of norharman and harman. Oral bioavailability is lower but there are indications that sublingual absorption may increase dietary uptake of beta-carbolines. Endogenous formation can be estimated to be 50-100 ng/kg bw per day for norharman and about 20 ng/kg bw per day for harman but these rates may increase with high intake of precursors. Biomarker studies on plasma levels of beta-carbolines reported on elevated levels of norharman, harman or both in diseased patients, alcoholics and following tobacco smoking or consumption of beta-carboline-containing food. Cigarette smoking has been identified as major influence but dietary exposure may contribute to exposure.

Investigation of the genotoxic effects of 2-amino-9H-pyrido[2,3-b]indole in different organs of rodents and in human derived cells.

Aim of the present study was the investigation of the genotoxicity of amino-alpha-carboline (AalphaC) in human derived cells and of its organ-specific effects in laboratory rodents. This heterocyclic amine (HA) is contained in fried meat and fish in higher concentrations than most other cooked food mutagens. In the present experiments, AalphaC caused dose-dependent induction of micronuclei in the human derived hepatoma cell line HepG2 at concentrations > or =50 microM. In contrast, no significant effects were seen in Hep3B, another human hepatoma cell line, which may be explained by the concurrent lower activity of sulfotransferase (SULT), an enzyme playing a key role in the activation of AalphaC. A positive result was also obtained in the single cell gel electrophoresis (SCGE) assay in peripheral human lymphocytes, but the effect was only significant at the highest concentration (1000 microM). In Fischer F344 rats and ICR mice, the liver was the main target organ for the formation of DNA adducts (at > or =50 mg/kg bw), and in lungs and colon substantially lower levels were detected. Identical organ specificity as in the DNA adduct measurements was seen in SCGE assays with rats, whereas in mice the most pronounced induction of DNA migration was observed in the colon. Comparison of our results with data from earlier experiments indicate that the genotoxic potency of AalphaC is equal to that of other HAs, which are contained in human foods in much smaller amounts. Therefore, our findings can be taken as an indication that the human health risk caused by exposure to AalphaC is higher than that of other HAs that are formed during the cooking of meat and fish.

Morphological transformation of C3H/M2 mouse fibroblasts by, and genotoxicity of, extracts of human milk.

Breast cancer may be initiated by environmental/dietary agents and human milk may act as an ex vivo indicator of in vivo exposure of mammary epithelial cells to genotoxins. Extracts of human milk from UK-resident women (n=7) were tested for their abilities to morphologically transform C3H/M2 mouse fibroblasts. Genotoxicities were assessed in the Salmonella typhimurium reverse-mutation assay in the presence of S9 using strains TA1538 and YG1019, and in metabolically-competent human MCL-5 cells with the micronucleus and with the alkaline single cell gel electrophoresis (comet) assays. Two of the seven extracts were inactive in the transformation assay both in the presence or absence of S9, two appeared to be equally transforming either in the presence or absence of S9, and two other extracts induced increased transformation frequencies in the presence of S9. A seventh extract, tested only in the absence of S9, was inactive. Extracts were either active or inactive in at least three of the four tests applied. Four extracts were active or inactive in all four tests. The results suggest that human milk could be used as a resource for investigations of the as-yet-unidentified transforming agents previously detected in mammary lipid.

Cell transformation in vitro by food-derived heterocyclic amines Trp-P-1, Trp-P-2 and N(2)-OH-PhIP.

Heterocyclic aromatic amines (HCA) are formed upon frying of poultry, fish or meat and have been shown to induce tumours in rodent bioassays. We investigated the transforming activity of HCA in an in vitro assay using the M2/C3H mouse fibroblast cell line. An external metabolic activation system (rat liver homogenate) was required in order to observe any HCA-induced cytotoxic effects or cell transforming activity. Trp-P-1 and Trp-P-2 are shown to be among the most potent transforming HCA that have been detected in food. Metabolic activation of HCA has been shown to proceed via N-hydroxylation of the exocyclic amino group. Therefore, we tested 2-hydroxyamino-1-methyl-6-phenylimidazo[4,5-b]pyridine (N(2)-OH-PhIP) the activated metabolite of 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine. N(2)-OH-PhIP proved to be one of the most powerful compounds with transforming activity observable at a concentration as low as 30 nM. Since 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) is the most abundant HCA formed in fried and grilled food and N-hydroxylation appears to be the predominant pathway of human metabolism, these data support the hypothesis that HCA are involved in the aetiology of human cancer.

Heterocyclic amines: human carcinogens in cooked food?

During the frying of meat and fish, genotoxic heterocyclic amines (HCAs) are formed. The dietary exposure to HCAs may be implicated in the aetiology of human cancer, but there may be other factors in our diet that prevent the genotoxic effects of these compounds. Within the project described here, we plan to identify regional and individual cooking habits that affect HCA-levels in our food. These are determined with a validated analytical method and the exposure to HCAs is estimated by dietary assessment. Biomarker analysis will be employed to estimate recent or long-term exposure to HCAs. In order to identify genetically determined risk factors in humans, cell lines are genetically engineered expressing allelic variants of acetyl- and sulfotransferases implicated in HCA metabolism. Species differences of metabolism and toxicity of HCAs are assessed and the influence of the intestinal microflora on HCA-induced toxicity is evaluated. Dietary constituents that may reduce the genotoxicity of HCAs are screened for potential protective effects in in vitro and in vivo model systems. Finally, we will aim at human intervention studies to investigate if these protective factors are relevant for man. The objectives of this project are to estimate and possibly reduce the exposure levels to HCAs in Europe, to identify populations highly susceptible to HCA toxicity, and to reduce the toxic effects of HCAs by protective factors.

Bioactivation of the food mutagen 2-amino-3-methyl-imidazo[4, 5-f]quinoline (IQ) by prostaglandin-H synthase and by monooxygenases: DNA adduct analysis.

2-Amino-3-methylimidazo[4,5-f]quinoline (IQ) is a known multisite carcinogen in rodents and a potent mutagen in acetyltransferase-proficient Salmonella typhimurium strains on activation by either monooxygenases (MFO) or by prostaglandin H synthase (PHS). The primary metabolites formed by MFO- or PHS-mediated IQ-oxidation are different ([Wolz]), but secondary metabolism could ultimately result in the same DNA-binding intermediates. For further investigations, the DNA adduct pattern was now studied by means of (32)P-postlabelling analysis in vitro on PHS-activation and compared to that formed on MFO-mediated activation of IQ in hepatocytes. The C8-dG-IQ-adduct N-(deoxyguanosin-8-yl)-IQ was the major adduct in all samples, that is, in DNA isolated from S. typhimurium YG1024 treated with PHS-oxidized IQ or its nitro-derivative, from ovine seminal vesicle cells, and from hepatocytes exposed to IQ or nitro-IQ. This speaks for the formation of a common DNA-reactive species, presumably an arylnitrenium ion, generated by different pathways in these cellular model systems. The similarity of critical biochemical DNA lesions suggests that PHS can contribute to the bioactivation of IQ in vivo: this is of particular interest in extrahepatic tissues since expression of cytochrome P450 isoenzymes known to be involved in the N-oxidation of IQ is largely confined to the liver.

Heterocyclic aromatic amines induce DNA strand breaks and cell transformation.

Heterocyclic aromatic amines (HAAs), formed during the cooking of foods, are known to induce tumours in rodent bioassays and may thus contribute to human cancer risk. We tested six HAAs in a morphological transformation assay and in three in vitro genotoxicity assays. The morphological transforming abilities of HAAs were tested, in the presence of rat-liver S9, in the C3H/M2 fibroblast cell line. Concentration levels of 50 microM 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (8-MeIQx), 100 microM 2-amino-3,4,8-trimethylimidazo-[4,5-f]quinoxaline (4,8-DiMeIQx), 50 microM 2-amino-3-methylimidazo[4,5-f]quinoline (IQ), 100 microM 2-amino-9H-pyrido[2,3-b]indole (AalphaC), 100 microM 2-amino-3-methyl-9H-pyrido[2,3-b]indole (MeAalphaC) and 15 microM 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) induced maximum transformation potencies of 5.5, 6.6, 6.3, 5.2, 7.3 and 9.2 transformed foci per 10(4) surviving cells, respectively. Bacterial mutagenic activity was determined in the presence of rat-liver S9 using the Salmonella typhimurium reverse-mutation assay employing strain YG1019. Mutagenic potencies of 3800 revertants (revs)/ng with 8-MeIQx, 2900 revs/ng with 4,8-DiMeIQx, 3480 revs/ng with IQ, 1.6 revs/ng with AalphaC, 2.9 revs/ng with MeAalphaC and 5 revs/ng with PhIP were observed. Clastogenic activity in vitro was analysed by the micronucleus assay in metabolically competent MCL-5 cells. Dose-dependent induction of micronuclei was observed for all HAAs tested with 1-5.4% of cells containing micronuclei at 10 ng/ml. Micronucleus induction was in the order 4,8-DiMeIQx > 8-MeIQx > IQ > MeAalphaC > PhIP > AalphaC. DNA strand-breaking activity in MCL-5 cells was measured by the alkaline single cell-gel (comet) assay. The lowest effect doses for significant increases (P < or = 0.0007, Mann-Whitney test) in comet tail length (microm) were 45.5 microg/ml (200 microM) for PhIP, 90.9 microg/ml (410-510 microM) for 4,8-DiMeIQx, IQ, MeAalphaC and AalphaC, and 454.5 microg/ml (2130 microM) for 8-MeIQx. It is not yet clear which of these assays most accurately reflects the genotoxic potential to humans of compounds of this class of environmental carcinogens.

DNA adducts in human breast tissue: association with N-acetyltransferase-2 (NAT2) and NAT1 genotypes.

The etiology of human breast cancer is poorly understood, but circumstantial evidence points toward exogenous genotoxins as causative agents; they are believed to exert their carcinogenic action by binding to DNA. Because this binding is often preceded by metabolic activation, it is dependent on the expression and activities of metabolic enzymes of the host. Human mammary tissue samples from 42 women undergoing surgery for breast cancer or reduction mammoplasty were analyzed for DNA adducts by 32P-postlabeling analysis. With the butanol extraction method of DNA adduct enrichment, adduct levels were determined to be 0-414.6 adducts per 10(9) nucleotides, with considerable interindividual variation. To characterize the DNA adducts, we reanalyzed the adduct spots by reversed-phase high-performance liquid chromatography. Of two major adduct spots detected on TLC that accounted for up to 70% of the DNA modification, one eluted as a single peak on high-performance liquid chromatography, whereas the other was resolved into two distinct peaks of radioactivity. These major adducts were highly lipophilic in character. The N-acetyltransferase-1 (NAT1) and NAT2 genes were analyzed for common mutations using random RFLP analysis. An association between NAT2 acetylator status and adduct levels was observed; significantly elevated adduct levels occurred in the mammary DNA from women who were designated slow acetylators for NAT2 [median adduct level = 83.0 adducts per 10(9) nucleotides (range, 9.0-414.6)], as compared with the levels in individuals designated rapid acetylators for NAT2 [median adduct level = 39.7 adducts per 10(9) nucleotides (range, 0-91.0; P = 0.0053)]. On the other hand, NAT1 genotypes were not significantly associated with adduct levels. Although the agents responsible for the DNA modifications in the human breast are not known, this pilot study supports the hypothesis that DNA adduct formation in the human breast may be influenced by the NAT2 genotype.

Morphological transformation of C3H/M2 mouse fibroblasts by extracts of human mammary lipid.

Mammary lipid may act as a reservoir for genotoxins. Mammary lipid extracts (MLEs), obtained from eight UK women (21-41 years) undergoing reduction mammoplasty, were examined for their abilities to morphologically transform C3H/M2 mouse fibroblasts. Resultant transformation rates were 0.27, 0.33, 0.07, 0.29, 0.21, 0.00, 0.07, and 0.13 transformed foci/treated dish, respectively. Although the lipid-extraction procedure used was originally designed to extract heterocyclic aromatic amines (HAAs), liquid chromatography/mass spectroscopy (LC/MS) with selective ion monitoring has failed to detect HAAs in any of the lipid extracts so far examined. Genotoxicities were also assessed in S. typhimurium TA98 and in metabolically competent human (MCL-5) cells by the micronucleus and by the alkaline single-cell gel ("comet") assays. The MLEs induced bacterial mutagenicity rates ranging from 0 to 498 revertants/plate/g-lipid equivalent and micronucleus-formation rates from 0 to 20 micronuclei/500 binucleate cells/g-lipid. Median comet tail lengths (induced with MLEs of 8.0 g-lipid equivalent) ranged from 6.0 to 74.0 micrometer. The results demonstrate the presence of as-yet-unidentified transforming agents in mammary lipid.

Carcinogenicity and DNA adduct formation observed in ACI rats after long-term treatment with madder root, Rubia tinctorum L.

Madder root, Rubia tinctorum L., is a traditional herbal medicine used against kidney stones. Recently we reported that lucidin, a hydroxyanthraquinone derivative present in this plant, is mutagenic in bacteria and mammalian cells. We also demonstrated the formation of DNA adducts in tissue culture and mice after treatment with this compound. To elucidate the possible carcinogenicity of madder root, three groups of male and female ACI rats received either a normal diet or a diet supplemented with 1 or 10% drug for a total period of 780 days. Weight gain and morbidity were not different among the three groups. Non-neoplastic lesions related to the treatment were evident in the liver and kidneys of both sexes. Moreover, dose-dependent increases in benign and malignant tumour formation were observed in the liver and kidneys of treated animals. 32P-post-labelling analysis showed an increase in the overall level of DNA adducts observed in the liver, kidney and colon of rats treated with 10% madder root in the diet for 2 weeks. HPLC analysis of 32P-labelled DNA adducts revealed a peak co-migrating with an adduct obtained after in vitro treatment of deoxyguanosine-3'-phosphate with lucidin. These observations suggest that the use of madder root for medicinal purposes is associated with a carcinogenic risk.

Pancreatic DNA adducts formed in vitro and in vivo by the food mutagens 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) and 2-amino-3-methyl-9H-pyrido[2,3-b]indole (MeA alphaC).

Genotoxic heterocyclic amines have been detected in grilled or fried meat and tobacco smoke. Among these, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) and 2-amino-3-methyl-9H-pyrido[2,3-b]indole (MeA alphaC) have been shown to induce tumours in rodents in several organs. Here we report on the DNA adduct formation by PhIP and MeA alphaC in vitro and in vivo, both in rat hepatic and rat pancreatic tissues or cells. Using 32P-postlabelling analysis both compounds were shown to induce a dose-dependent DNA modification in primary rat hepatocytes that was correlated with cytotoxicity in these cells. In explanted rat pancreas maintained in dynamic short-term organ culture MeA alphaC was shown to induce covalent DNA adducts. No DNA adducts were observed with PhIP in this assay. DNA adducts were observed in the liver and the pancreas of F344 rats treated with PhIP, with a 36-times higher level of adducts in the pancreas, confirming data reported earlier. DNA adduct levels induced by feeding 32, 160 or 800 ppm MeA alphaC in the diet were dose-dependent and higher in the liver compared with other organs including pancreas. While for PhIP the N2-(desoxyguanin-8-yl)-derivative was accounting for more than 90% of DNA adducts detected, in the case of MeA alphaC the N2-(desoxyguanin-8-yl) adduct was predominant in vitro and determined in vivo as one of up to 5 DNA adducts. MeA alphaC had been reported to induce preneoplastic foci and tumours in the liver and tumours and atrophy in the pancreas. In the case of MeA alphaC, the DNA adduct formation and cytotoxicity observed by us in vitro and in vivo correlate with the organ specificity of the reported pathological lesions. In the case of PhIP our in vitro data in pancreas and liver and the low adduct levels in liver in vivo also reflect the reported lack of pathological effects in these organs. In contrast, in pancreas, in vivo extraordinarily high adduct levels induced by PhIP were observed confirming studies published earlier, in spite of the fact that this compound does not cause pancreatic lesions. This enigmatic observation is discussed and the relevant literature is reviewed.

Identification of the major hepatic DNA adduct formed by the food mutagen 2-amino-9H-pyrido[2,3-b]indole (A alpha C).

2-Amino-9H-pyrido[2,3-b]indole (A alpha C) is among the most prevalent heterocyclic amines detected in grilled or panfried meat; it was shown to be carcinogenic in mice, to induce preneoplastic foci in rat liver, and to form covalent DNA adducts in vitro and in vivo. The corresponding nitro compound 2-nitro-9H-pyrido[2,3-b]indole (N alpha C) was prepared and shown to be a direct acting mutagen in the Salmonella assay, while the amino compound required external metabolic activation with rat liver homogenate (S9). When A alpha C was incubated with S9 in the presence of calf thymus DNA, one major DNA adduct spot was detected upon 32P-postlabeling analysis. This adduct comigrated on ion-exchange TLC and reversed-phase HPLC with the major adduct detected in primary hepatocytes treated with A alpha C. In DNA isolated from livers of male F344 rats treated with 800 and 160 ppm, the formation of the same major adduct was observed with relative adduct levels of 20.6 +/- 9.6 and 1.4 +/- 1.1 adducts/10(8), respectively, as determined with the butanol extraction variant of the 32P-postlabeling assay. No DNA adducts were detected in liver DNA from rats treated with 32 ppm A alpha C or control animals. The major adduct spot was eluted and hydrolyzed and the modified base characterized by chromatographic and UV spectral comparison with a synthetic standard synthesized from acetylated guanine N3-oxide and A alpha C. Electrospray mass spectrometry and 1H- and 13C-NMR spectroscopy provided further evidence for the major adduct as N2-(guanin-8-yl)-2-amino-9H-pyrido[2,3-b]indole. A alpha C is formed especially in high-temperature preparation of food and may contribute considerably to the human carcinogenic risk that might be imposed by heterocyclic amines.

DNA adducts in marine and freshwater fish as biomarkers of environmental contamination.

The analysis of DNA modifications in aquatic animals may serve as a sensitive marker of exposure to genotoxic contaminants. This is of importance in assessing water quality regarding pollution with genotoxic compounds, food safety analysing edible aquatic animals and in terms of ecotoxicology. Covalent modification of DNA is considered a crucial event in chemical carcinogenesis and thus may be considered a biomarker of an early genotoxic effect. Measuring DNA adducts is unique in that these lesions may be considered a biomarker of both exposure and effect. A number of studies have described the analysis of the DNA isolated from the liver of both freshwater and marine fish. Considerable levels of DNA adducts have been observed in some animals from contaminated lakes or rivers. Low levels were observed in DNA from the liver of marine fish. The background levels of DNA adducts have to be determined in animals from appropriate uncontaminated control sites that are matched for species, gender, age and season of the year. It is of crucial importance to consider the species analysed since there have been reports of the non responsiveness of some species.

Characterization of the major DNA adduct formed by the food mutagen 2-amino-3-methyl-9H-pyrido[2,3-b]indole (MeAalphaC) in primary rat hepatocytes.

Cooking of proteinous food results in the formation of heterocyclic amines. Among these, 2-amino-3-methyl-9H-pyrido[2,3-b]indole (MeAalphaC) has been identified as a mutagenic pyrolysis product of soya protein and has been detected in grilled or pan fried meat. It was subsequently proven to be carcinogenic in mice and, recently, in rats and to form covalent DNA adducts in vitro and in vivo. The corresponding nitro compound, 2-nitro-3-methyl-9H-pyrido[2,3-b]indole (MeNalphaC), was prepared and shown to be a direct acting mutagen in the Ames Salmonella reversion assay. When MeNalphaC was chemically reduced in the presence of DNA a major DNA adduct was detected using the 32P-postlabelling assay. This major adduct was characterized by UV spectroscopy and mass spectrometry as N-(deoxyguanosin-8-yl)-2-amino-3-methyl-9H-pyrido[2,3-b]indole. This structure was corroborated by identification of the modified base as a guanine moiety modified at the C8 position as judged by chromatographic and spectral comparison with a standard synthesized from acetylated guanine-N3-oxide and MeAalphaC was characterized by UV/Vis and 1H NMR spectroscopy and mass spectrometry. Treatment of primary rat hepatocytes with MeAalphaC (100 microM, 24 h) resulted in adduct levels of 9.8 fmol/microg DNA as determined by 32P-postlabelling analysis. Using HPLC analysis, two major 32P-labelled adducts were observed accounting for 80 and 13% of total binding respectively. The major adduct was chromatographically indistinguishable from the synthetic deoxyguanosine-C8 adduct using both ion-exchange thin layer or reversed-phase HPLC. Although MeAalphaC is formed only in low p.p.b. levels in cooked food, the contribution to the human carcinogenic risk that might be imposed by heterocyclic amines is not to be neglected.

Characterization of DNA adducts formed by aristolochic acids in the target organ (forestomach) of rats by 32P-postlabelling analysis using different chromatographic procedures.

We report the analysis of DNA adducts in the target organ (forestomach) of male Sprague-Dawley rats treated orally with two doses (10 mg/kg body wt) per week for 2 weeks of either aristolochic acid I (AAI), aristolochic acid II (AAII) or the plant extract aristolochic acid (AA). DNA adducts were detected and quantitated using the nuclease P1-enhanced version of the 32P-postlabelling assay. For identification of adducts, reference compounds were prepared by reaction of enzymatically activated AAI and AAII with 3'-purine phosphonucleosides and analysed by the n-butanol enrichment procedure. These reference compounds were assigned to the previously characterized DNA adducts of AAI [7-(deoxyguanosin-N2-yl)-aristolactam I = dG-AAI, 7-(deoxyadenosin-N6-yl)-aristolactam I = dA-AAI] and AAII [7-(deoxyadenosin-N6-yl)-aristolactam II = dA-AAII]. Cross referencing of the carcinogen-modified nucleoside bisphosphates obtained from forestomach DNA with the synthetic standard compounds by ion-exchange chromatography and reversed-phase HPLC demonstrated that the major DNA adducts formed by AAI and AA were identical to dG-AAI and dA-AAI. Likewise, forestomach DNA isolated from AAII-treated rats showed two purine-derived adduct spots, the major one being dA-AAII, the minor one being tentatively identified as 7-(deoxyguanosin-N2-yl)-aristolactam II. A minor adduct detected in forestomach DNA of rats treated with AAI was found to be chromatographically indistinguishable from the adduct identified as dA-AAII, indicating a possible demethoxylation reaction of AAI. Quantitation of DNA adducts revealed that in in vitro reactions with 3'-phosphonucleosides the adduct levels were approximately one order higher for both AAI- and AAII-derived adducts than in forestomach DNA modified with AAI or AAII in vivo. In vitro as well as in vivo adduction by AAI was more efficient than adduction by AAII. The pattern of adduct spots obtained from forestomach DNA of rats treated with the plant extract AA reflected the composition of the extract determined by HPLC analysis. Irrespective of the aristolochic acid used to induce DNA adducts, deoxyadenosine is the major target of modification, pointing to the general importance of deoxyadenosine adducts for chemical carcinogenesis of these naturally occurring products. This study shows that the combination of two independent chromatographic systems considerably enhances the fidelity of identification of DNA adducts with the 32P-postlabelling assay.

32P-post-labelling analysis of DNA adducts formed by food-derived heterocyclic amines: evidence for incomplete hydrolysis and a procedure for adduct pattern simplification.

Food-derived aminoimidazoazarenes have been shown to be mutagenic and carcinogenic and to form covalent DNA adducts. 32P-Post-labelling analysis of DNA modified with these heterocyclic amines (HA), including 2-amino-3-methyl-imidazo[4,5-f]quinoline (IQ), 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP), 2-amino-3,4-dimethylimidazo [4,5-f]quinoline (MeIQ), 2-amino-3,7,8-trimethylimidazo[4,5-f]quinoxaline (7,8-DiMeIQx) and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx) has resulted in considerable interlaboratory variation in the characteristic patterns of DNA adduct spots, with up to six being detected for each compound. Similar complex patterns were observed when azido-derivatives of HA were photoreacted with calf thymus DNA. When deoxyguanosine 3'-monophosphate was modified with the azido derivatives and analysed using the 32P-post-labelling procedure, one major spot was observed for IQ, 4,8-DiMeIQx, 7,8-DiMeIQx or PhIP and two major spots for MeIQ or MeIQx. In each case, these adducts were chromatographically indistinguishable from the major adducts formed with DNA. No major adduct spots were observed when 3'-phosphate derivatives of deoxyadenosine, deoxycytidine or thymidine were reacted with the azido-derivatives of HA. In an attempt to identify the additional spots, azido derivatives of PhIP or IQ were reacted with the synthetic homopolymer poly(dG).poly(dC), the alternating copolymer poly(dC-dG) or a synthetic oligonucleotide (TTT-GTTTTTTCTTTCCCT): in each case a reduced number of adduct spots were detected. The introduction of an additional nuclease P1 hydrolysis step following the labelling reaction further reduced the number of adduct spots to only one or two major spots. Reversed-phase HPLC analysis showed that the number of peaks of radioactivity was also reduced to one or two, presumably corresponding to the [32P]-5'-monophosphate deoxyguanosine adducts. We suggest that many of the additional spots commonly observed in conventional 32P-post-labelling analysis of HA-modified DNA are adducted oligonucleotides that are partly resistant to hydrolysis by micrococcal nuclease and spleen phosphodiesterase but are susceptible to hydrolysis by nuclease P1.