Heterocyclic amines in cooked foods - role in human health | HC AMINES | Project | Results | FP5 | CORDIS

The alkaline single cell gel electrophoresis (comet assay) was found suitable to determine the genotoxicity of aromatic amines (AAF, PhIP, IQ) activated by genetically engineered V79 lung fibroblasts of the Chinese hamster stably co-expressing cytochrome P450 dependent monooxygenases (rCYP1A2; hCYP1A2) and conjugating enzymes (rSULT1C1; hSULT1A1*1; hNAT2*4). With this convenient in vitro method 11 plant-derived beverages, 15 fruit and 13 vegetable juices were investigated for their protection against the DNA damaging action of aromatic amines. Of these 39 food components only 5 (banana, apple [Granny Smith], orange, green beans, cucumber) were inactive; the majority halved the genotoxicity of the aromatic amines at concentrations of 0.4-2 % (v/v; IC50-value); beer (yeast-wheat, alcohol-free), red wine, green tea, blueberry, plum, red grape, sweet cherry and spinach were exceptionally active with IC50-values of 0.15-0.20 %. The influence of the food components on the two successive steps of metabolic activation could independently be determined with diagnostic substrates (benzo[a]pyrene-7,8-dihydrodiol; N-hydroxy derivatives of PhIP and IQ) and indicated that the inhibition of one or both enzymatic steps can be made responsible for the antigenotoxicity of the food component. This assumption was confirmed by direct measurement of the activity of CYP1A2 and its inhibition by selected plant-derived food components. Our in vitro method for the detection of antigenotoxic food components combines experimental simplicity with considerable human relevance since mammalian cells expressing human enzymes were used as test organisms. Furthermore the experimental design very closely mimics the human cellular situation since the biotransformation required to activate the genotoxins takes place inside the indicator cell and not outside as it is the case when employing bacteria as test organisms (AMES test).

To elucidate if dietary factors have an impact on the metabolism of HAs in humans and might exert protective properties intervention trials were carried out with green tea (GT) and with Brussel sprouts (BS). It has been shown earlier that GTand BS reduce the carcinogenic effect of HA in animal models via modulation of their metabolism. Endpoints were (1) changes in the bacterial urinary mutagenicity pattern caused by consumption of HA rich chicken meal. (2) Induction of the HA protective enzymes glutation-S-transferase and glucuronosyltransferase (UGT and GST) (3) Investigation of PhIP metabolites caused by a chicken meal given before and after the intervention phase (4) Investigation of the sensitivity and repair of HA induced DNA- damage of peripheral lymphocytes. The cells were isolated before and after the intervention trails and exposed in vitro either to IQ or to HA mix which represented the HA composition of fried beef. GT.: No changes in the urinary excretion patterns were seen; also no changes in the activity of UGT (measured in the urine after paracetamol dairy) were seen. Preliminary data suggest that no changes of the 5-OH PhIP excretion take place and the sensitivity of peripheral lymphocytes were even enhanced after the intervention. This later experiment is currently repeated BS: The urinary mutagenicity caused by HAs from a chicken meal was significantly decreased after consumption of the vegetables. Data from UGT measurements are not yet available. Results from 5-OH determination in urine are not conclusive and have to be repeated as problems were encountered in the stability of this metabolite. Results from in vitro experiments with lymphocytes are not yet available. The experimental work is still in progress. Preliminary data suggest that BS but not GT might have an effect on the detoxification of HAs in humans. Final conclusions are not yet available as the experimental work is still in progress.

Most in vivo models (micronucleus assays, chromosome aberrations, etc....) with rodents are not able to detect the genotoxic effects with HAs. One possible alternative are DNA- adduct measurements but these approaches are costly and will miss possible adverse effects of putative protective compounds. Therefore we developed a protocol for SCGE assays with rats which enable to measure HA induced DNA migration in different inner organs which are targets for tumour induction by HAs (liver, colon, etc.). Optimal exposure times were established (4 hours) and the doses required to detect effects were investigated, they rank between 50-100 mg/kg bw. Three different heterocyclic amines were tested. IQ caused pronounced effects in liver and colon, which were more or less identical, whereas AáC was only active in the liver and caused only marginal effect in the lungs and colonic tissue. With PhIP, a much weaker effect was seen under identical experimental conditions in both organs (the effects were ca. 60% lower than those seen with IQ). However, effects were statistically significant. Additionally, also HA mixtures were tested which represented the composition of HAs of fried beef and fried chicken. Also with these mixtures clear-cut positive effects were obtained. Beef mix induced HA migration in hepatic and colonic tissue, whereas chicken mix was only active in the liver. These differences can be explained by the high amount of PhIP in the chicken mix. The SCGE model was used subsequently in a number of chemoprevention studies with Brassica vegetables, prebiotics (long and short chain inulin) and probiotics (different Lactobacilli strains). The most pronounced protection was found with the different vegetables and with the Lactobacilli strains complete inhibition of DNA- damage was observed. Also the prebiotics exerted chemoprotective activities. With the vegetables, additional experiments were conducted in which it was shown that the reduction of DNA- damage (measured in the colon and in the liver) was paralleled by a reduction of the formation of preneoplastic lesion in these organs (GST, foci in hepatic tissue, aberrant crypt foci in the colon). The extent of cancer protection was identical to the extent of inhibition of DNA- damage measured in the SCGE assays. This observation shows that the SCGE model indeed has predictive value in regard to cancer prevention caused by HAs.

PhIP DNA adducts were most prevalent at sites in the intestine were PhIP induced most tumours and less adducts were seen at a later time point when the mice were less susceptible to PhIP (Steffensen et al, 2001). The major mechanism of tumour induction by PhIP and IQ was due to induction of loss of the wt apc allele Andreassen et al, 2001, 2002). From those tumours apparently retaining the wt apc allele we verified 25 Apc mutations from 804 PhIP induced tumours screened. Of these were 60% G to T transversions, and 16% G deletions, indicating that these are the predominant types of PhIP-induced mutations in the Apc gene in Min/+ mice. Most of the mutations were located between codon 989 and 1156 corresponding to the first part of the beta-catenin binding region. We also identified two Apc truncation mutations in spontaneously formed intestinal tumours (n=606) from untreated mice. These mutations were one C to T transition and one T insertion, which were different from those induced by PhIP (Møllersen et al, 2003, submitted). Published mutations in apc of mice induced with mutagens other than PhIP, i.e. ethyl-nitroso-urea showed a different mutation spectrum from that of PhIP. Furthermore mice with miss-match repair deficiency (knock-out) showed mutations mainly affecting cytosine and which were also different from that of PhIP. We used a database of reported human APC mutations (T. Soussi, personal communication) comprising of 790 sporadic colorectal adenomas/carcinomas. Of these were 9.6% G to T transversions and 3.2% were G deletions, which are mutations compatible with those induced by PhIP. From the human APC region between codons 996-1166, corresponding to the murine Apc mutation sensitive area between codons 989-1156 described. PhIP appears to induce a mutation spectrum that is specific relative to a carcinogen of a different chemical class and from those occurring spontaneously and following knocking out of DNA repair genes. About 10% of the mutations seen in APC of human colorectal cancers are similar to those induced by PhIP in the corresponding gene in the mice. We cannot exclude that also other carcinogens forming bulky DNA adducts in intestinal epithilium might cause the G to T mutations similar to those induced by PhIP and those seen in humans. Our findings accord with the hypothesis that dietary PhIP may cause APC mutations in humans. However, only a minor fraction of the APC mutations seen in human colorectal tumours are probably due to PhIP and other HAs. It should be kept in mind that PhIP and other IQ could induce tumours via other mechanisms such as loss of the wt APC allele and mutations in other genes.

In an analysis of ten urinary samples collected after a meal of fried meat we were able to detect only the amino-b-carbolines, norharman and harman. Qualitatively, we were able to detect PhIP and AaC smoker's urinary samples, but not in urinary samples from non-smokers. This is the first time AaC has been reported in human urine, thus, this HA contributes to the HA burden, as proposed. The detection limits of our method are in the same range as compared to published procedures and levels or HA detected in urinary samples. In order to be able to analyse the less abundant HAs increased analytical sensitivity is necessary.

The objective is to estimate population ranges of HA intake by combining data on HA-content in cooked foods and information about food selection and cooking practices from dietary surveys, which can be used in epidemiological studies to assess the cancer risk with HAs. The estimation of the intake range by combining food survey data with food frequency data has started. Data from a Swedish food survey, the Malmö Diet & Cancer Study (also participating in the EPIC study) will be combined with HA values and the daily uptake of HAs estimated. To assess the intake of HAs, food questionnaires that address cooking methods are required. Among the factors that influence the exposure to HAs are types of food, cooking method, portion size and intake frequency. Six different food questionnaires (food frequency, dietary record, 24-h recall) including the one that will be used in the EPIC study were evaluated, and a new food questionnaire was developed, which provides information on eating habits and cooking techniques with a focus on cooked meat and fish. The questionnaire has been distributed to the other partners to be used in other studies on the exposure/excretion of dietary HAs.

Numerous homo- and heterocyclic aromatic amines are carcinogenic. This activity is mediated by mutagenic metabolites, which are usually formed in a 2-step reaction involving a cytochrome P450 (mainly CYP1A2) and a conjugating enzyme (an acetyltransferase, NAT, or a sulfotransferase, SULT). Target cells of standard mutagenicity test systems do not express these enzymes and, therefore, are insensitive towards most aromatic amines. We have gene-technically modified Chinese hamster V79 cells (a standard target cell in genetic toxicology) for expression of human CYP1A2, alone or in combination with one of the following human enzymes (wild-type forms): NAT1, NAT2, SULT1A1, SULT1A3 and SULT1B1 (i.e. the members of these enzyme classes being expressed in colon mucosa, a putative target of the carcinogenicity of heterocyclic amines). Since NAT2 and SULT1A1 are genetically polymorphic, we expressed not only the wild-type forms (NAT2*4 and SULT1A1*1, respectively), but also the major other alloenzymes (NAT2*5B, NAT2*6A, NAT2*7B and SULT1A1*2). A substantial number of carcinogenic homo- and heterocyclic aromatic amines scored strongly positive in the novel test system (using gene mutations at the hprt locus at the endpoint), but was inactive or only very weakly mutagenic in the standard V79 test system. Activation required the presence of both, CYP1A2 and an appropriate conjugating enzyme. Heterocyclic aromatic amines could be classified into two groups, NAT-dependent mutagens and SULT-dependent mutagens. SULT-dependent mutagens may deserve special attention, as laboratory animals used for carcinogenicity studies (rat and mouse) are poor models for human with regard to tissue distribution and substrate specificity of SULTs. In particular, expression of many SULT forms in extrahepatic tissues (such as colon mucosa or mammary gland) is low in rat and mouse, but high in human, suggesting an increased risk of the latter species. For this reason, we have started the gene-technical construction of humanized animal models for SULT enzymes. The results with cell lines with polymorphic enzymes led to unexpected results, requiring revisiting some common views and predictions on genotype-risk associations. Again, humanized animal models representing different human genotypes will be useful in this regard.

Animal experiments in which the impact of vegetables on HA induced specific DNA- damage was studied lead to a important discovery: with garden cress (Lepidium sativum) a pronounced protective effect was observed, which was paralleled by an increase of the enzyme glucuronosyltransferase (UGT), whereas other xenobiotic drug metabolising enzymes in the liver and colon (GST, CYP1A2, etc.) were not significantly affected. These findings suggest that UGT plays a major role in the detoxification of IQ and probably also of HAs and it is known that this enzyme can be induced by dietary constituents. Furthermore we demonstrated that the protection against DNA- damage indeed leads to prevention of the induction of preneoplastic lesions in the colon and in the liver. In subsequent experiments we found that commonly consumed Brassica vegetables caused protection in colon and hepatic tissue against cancer induction by heterocyclic amines and against this effects were paralleled by UGT induction. These results are of direct human relevance; as it is known that UGT can be induced in humans by dietary constituents. Therefore it is possible to design nutritional recommendations which may protect humans against HA induced DNA- damage. Our findings also indicate that the reduced colon cancer rats seen in epidemiological studies in individuals who consume Brassica vegetables might be due to protection against HAs.

The impact of the human intestinal microflora and of major representatives (Lactobacilli/Bacteroides strains) on the genotoxic/carcinogenic properties of HAs were investigated.In a first experiment, we found that the presence of the microflora- leads to a strong enhancement of the genotoxic activity of IQ (a widespread HA). In germ free rats, the extent of damage and colon cells was substantially (ca. 70%) lower than that seen in rats harbouring human microflora and in normal rats. This important discovery shows that the microflora plays a crucial role in the conversion of HAs to carcinogenic metabolites, which has been completely under estimated in the pont. Our observation was confirmed by further experiments with HA- mixtures representing fried meats. In a second experimental series the DNA-damage caused by HAs was investigated comparatively in rats harbouring microfloras from meat consumers and vegetarians (7 days adventists). In these experiments, no significant differences were seen in the extent of DNA-damage of colons, but a significantly lower extent of DNA-migration was seen in the livers of animals, which harboured microfloras of vegetarians. In further experiments we investigated the effects of Lactobacilli commonly contained in fermented food (Streptococcus thermophili, Bifidobacterium longum, Lactobacillus bulgarius) on the induction of DNA-damage in colon and livers (both organs are targets for tumour induction by HAs). The damage was induced by HA mixtures which are representative for fried meat and chicken. Complete protection was found with all Lactobacilli strains when the rats were received orally 108 cells per kg/bw in both organs with meat mix. On the contrary no effects were seen with the chicken mix. The protective effect seen with meat mix was dose dependent and persisted up to 12 hours. The observation suggests that consumption of Lactobacilli containing fermented foods protects against HAs and may explain the decrease colon cancer rats seen in humans who consume elevated levels of fermented foods and excrete high elevated Lactobaccili in their faces. On the other hand, also Bacteriodes strains which occur frequently in the normal colonic microflora where found to amplify the DNA- damaging effects of HAs in bacterial in vitro assays (Ames test) but also in rat strains harbouring Bacteroides fragilis and Bacteroides tetragonicom. The extent of DNA- damage was increased approximately 2 fold compared to that seen in germ free animals under the same conditions. It is known that the Bacteroides contents in the human microflora depend on the diet and can be modulated. Increase uptake of milk and dietary products will lead to an increase of the Lactobacilli flora in the intestinal tract and to a decrease of the Bacteroides population. Therefore it will be possible to develop dietary recommendations, which increase the amount of HA protective bacteria in the intestinal contents.

A low dietary dose of EPAX 3000TG with a high n-3/n-6 PUFA ratio of 1.45 was not sufficient to significantly reduce the intestinal tumourigenesis in the Min mice. The dose dependent relationships indicated that the chemo preventive effect of n-3 PUFA from fish oil is dependent on a critical amount of fish oil, which might be difficult to achieve in the human diet.

The project partners from Austria, Sweden and Spain evaluated the eating habits and cooking habits for each country. The main dishes containing HAs are in Austria: Wiener Schnitzel, pork chop, Hamburger, roasted pork, turkey cutlet, sauce bolognaise, roasted chicken, ragout pork, goulash, ragout turkey, fried chicken, minced meat, roasted pork cutlet, boiled beef, beef steak, pork medallions, grilled chicken cutlet, meat with juice, beef roasted with onions, roasted beef; in Sweden: panfried (fish (12,0 g/day, meat 16,3 g/day, minced meat 11,4 g/day, offals 2,2 g/day, chicken 0,6 g/day, sausage 6,8 g/day) oven roasted (fish 0,3 g/day, meat 9,4 g/day, minced meat 2,0 g/day, chicken 8,0 g/day, sausage 1,9 g/day) and in Spain: white fish (hake, cod, sole coated with flour and deep fried), griddled veal (steak, hamburger), pork (loin, coated with flour and egg, and deep fried), griddled pork (loin, ''bontifarra'', Catalan sausage), griddled chicken, roasted chicken, chicken coated with flour and egg, deep fried), griddled white fish (hake, sole), fried pork (loin, ''bontifarra'', Catalan sausage, meatballs), roasted white fish (hake, sole), grilled lamb (chops), fried white fish (hake, anglerfish), griddled blue fish (salmon, sardine), ''Calamars a la romana'' (squid rings coated with batter, deep fried), blue fish (anchovy, sardine, coated with flour and deep fried), stewed veal, stewed white fish (hake, sole), griddled cuttlefish, stewed rabbit, fried chicken, griddled lamb.

AáC, is a heterocyclic amine, which has been discovered later than other cooked food mutagens, and to data no information on organ specific genotoxic effects are available, and it was not known if this compound causes DNA- damage in human cells. Our data show that AáC, a widespread HA, causes genotoxic effect (DNA-migration) in the liver and to a much lesser extent in the colon of rats. In lung tissue, only moderate not significant genotoxicity was measured. The genotoxic potency of the compound is similar to that of IQ (which is a potent rodent carcinogen) but a different pattern of organ specificity was observed. AáC was in the liver more than 2.5 folds stronger than in the colon, whereas the effects of IQ are identical in both organs. The differences can be explained by the fact that AáC requires sulfotransferases for its metabolic activation, which is present in rats only in the liver but not in the colon. AáC caused also genotoxic effects (micronucleus induction and DNA- migration) in human derived cells, i.e. in human hepatoma cell lines Hep G2 and Hep G3 and also in peripheral human lymphocytes. This observation indicates that the compound is converted by humans to DNA reactive metabolites and suggests that exposure to AáC leads to an increase cancer risk in humans. Furthermore it can be expected that dietary compounds which cause inhibition of sulfotransferase (resveratrol, certain flavonoids) might protect against this compound.

Preparation of HAs test solution (TR.1). Fifty bottles were filled at random with the test solution, whose concentration range was 0.8-2 µg/g for each HAs. A weight evolution study was performed by weighting all the bottles everyday during the first week, and afterwards four measurements were carried out in a period of three months. Purity of methanol and of the HAs stock standard solutions were checked. Some bottles were discarded because a decrease in weight was observed. The methanol used as solvent was found to be free of interference, and the stock solutions of HAs could be considered good enough to be used without further purification. Homogeneity study of HAs test solution (TR.2). The test solution was found to be homogeneous. The homogeneity was estimated with a good precision by examining the coefficient of variation of the replicate measurements obtained on samples within and between bottles. These coefficients of variations are very similar to the coefficient of variation due to the analytical method (<6%). Stability study of HAs test solution (TR.3). For stability study, sixteen bottles from the set of test solution were selected using a random sampling. Four bottles were kept at +40ºC, four at +25ºC, four at +4ºC and four at -18ºC. After 1, 2, 3 and 6 months, one bottle (five replicate analysis) of each temperature will be analysed using the HAs LC-UV method. The test solution was considered to be stable and no instability could be demonstrated in the time period studied. The test solution was considered suitable for the interlaboratory exercise.

Attempts were made to develop methods, which can be used as possible biomarkers for HA- induced colon cancer in humans. These methods include: (1) urinary mutagenicity measurements (2) analysis of specific HA metabolites (i.e.5- OH- PhIP) in the urine (3) development of hair analysis for PhIP- which will provide information on consumption of HAs (4) analysis of Lactobacilli/ Bacteriodes concentrations in human faces. These microorganisms have an impact on the genotoxicity/ carcinogenicity of HAs. (5) analysis of the susceptibility of peripheral blood cells towards HAs and of differences in the repair capacity of HA induced DNA- damage in cells of high vs. low risk individuals. Methods (1) and (3) have been established, also method (4) is available, but improvements of the measurements by use of real time- PCR for the quantification of the micro-organism are in progress. The development/standardisation of method (2) is in progress. Method (2) is available but for the stabilisation of urinary metabolites in the samples have to be developed. The validation of the different biomarkers requires studies with cancer/ non-cancer patients and the recruitment of the participants has been started but progress is low and only few individuals could be tested so far. Therefore results might be delayed and might be available after the end of the project. The outcome of the project will provide information which biomarkers can be used to identify individuals with elevated risks for colon cancer and provide also important basic information on the involvement of HAs in the aetiology of human colon cancer.

Humans are exposed to carcinogenic Heterocyclic Amines (HA) through the diet. The content of HA in the diet vary greatly dependent on meat type, cooking conditions and cooking temperature. It is therefore difficult to obtain an accurate measure of the exposure. HA's are metabolised in the body and excreted mainly in the urine within 24 hours. HA's are activated by N-hydroxylation and detoxified by ring hydroxylation followed by conjugation. There is a large interindividual variation in the metabolism of HA in the human population. In order to improve the risk assessment of exposure to HA, biomarkers for both exposure and activation capacity of individuals are needed. In the present study, the incorporation of 2-amino-1-methyl-6-phenylimidazo [4,5-b] pyridine (PhIP) into hair has been evaluated as a marker for the long-term exposure to PhIP. In mice PhIP was incorporated into hair in a dose dependent manner. In humans given ordinary diet, PhIP was detected at levels from less than 50 to 5000 pg/g hair. The incorporation of PhIP into hair is dependent on the content of eumelanin. PhIP in hair is a promising biomarker for the long-term exposure. We have proposed 5-OH-PhIP as a biomarker for the genotoxic dose of PhIP, since this compound is formed from the activated metabolite, together with DNA adducts in vitro experiments. 5-OH-PhIP is excreted in the urine from rats exposed to PhIP in a dose dependent manner. 5-OH-PhIP therefore seems to be a urinary biomarker for the active dose of PhIP.

Heterocyclic amines in cooked foods - role in human health

Deliverables

Share this page

Download