Use of human-derived liver cell lines for the detection of environmental and dietary genotoxicants; current state of knowledge.

This article gives an overview of the results of genotoxicity tests, which have been conducted within the last 5 years with the human liver cell line HepG2. It is an update of an earlier review from 1998 (by Knasmüller et al.). In addition, a number of publications are discussed which are relevant for the use of human derived liver cell lines in genetic toxicology. They concern the establishment of new endpoints, the development of new cell lines and possible pitfalls and problems. HepG2 cells have been used to test a wide variety of compounds over the last years. The most interesting observations are that the cells are highly sensitive toward polycyclic aromatic hydrocarbons and that genotoxic effects are seen with a number of carcinogenic mycotoxins, that give negative results in other in vitro assays. Carcinogenic metals such as As and Cd caused positive results as well, whereas only marginal or negative results were seen with nitrosamines. The low sensitivity toward these latter carcinogens is probably due to a lack of cytochrome P4502E1 which catalyses their activation. Also, a number of structurally different synthetic pesticides as well as bioactive plant constituents ("natural pesticides") have been tested and with some of them genotoxic effects were found. In most experiments, the formation of micronuclei was used as an endpoint; however also the single cell gel electrophoresis assay is increasingly used. Several transfectant lines of HepG2 have been constructed which express increased levels of phase I enzymes (such as CYP1A1, CYP1A2, CYP2E1 etc.); furthermore, cell lines became available which express human glutathione-S-transferases. These new clones might be particularly useful for the investigation of specific classes of genotoxicants and also for mechanistic studies. Apart from HepG2 cells, a number of other human derived liver cell lines have been isolated, but so far no data from genotoxicity experiments are available, except for Hep3B cells, which were compared with HepG2 and found to be less sensitive in general. Studies with HepG2 clones of a different origin indicate that the cells differ in regard to their sensitivity toward genotoxicants; also medium effects and the cultivation time might affect the outcome of genotoxicity studies. Overall, the results support the assumption that HepG2 cells are a suitable tool for genotoxicity testing.

Cafestol and kahweol, two coffee specific diterpenes with anticarcinogenic activity.

Epidemiological studies have found an inverse association between coffee consumption and the risk of certain types of cancers such as colorectal cancers. Animal data support such a chemopreventive effect of coffee. Substantial research has been devoted to the identification of coffee components that may be responsible for these beneficial effects. In animal models and cell culture systems, the coffee diterpenes cafestol and kahweol (C+K) were shown to produce a broad range of biochemical effects resulting in a reduction of the genotoxicity of several carcinogens including 7,12-dimethylbenz[a]anthracene (DMBA), aflatoxin B(1) (AFB(1)), benzo[a]pyrene (B[a]P) and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP). Different mechanisms appear to be involved in these chemoprotective effects: an induction of conjugating enzymes (e.g. glutathione S-transferases, glucuronosyl S-transferases), an increased expression of proteins involved in cellular antioxidant defense (e.g. gamma-glutamyl cysteine synthetase and heme oxygenase-1) and an inhibition of the expression and/or activity of cytochromes P450 involved in carcinogen activation (e.g. CYP2C11, CYP3A2). In animal models, the C+K-mediated induction of conjugating and antioxidant enzymes has been observed in hepatic, intestinal and kidney tissues. In the small intestine, these inductions were shown to be mediated by Nrf2-dependent transcriptional activation. In vitro investigations obtained in cell cultures of human origin indicate that the effects and mechanisms observed in animal test systems with C+K are likely to be of relevance for humans. In human liver epithelial cell lines transfected to express AFB(1)-activating P450s, C+K treatment resulted in a reduction of AFB(1)-DNA binding. This protection was correlated with an induction of GST-mu, an enzyme known to be involved in AFB(1) detoxification. In addition, C+K was found to inhibit P450 2B6, one of the human enzymes responsible for AFB(1) activation. Altogether, the data on the biological effects of C+K provide a plausible hypothesis to explain some of the anticarcinogenic effects of coffee observed in human epidemiological studies and in animal experiments.

Effect of polymorphism in the human glutathione S-transferase A1 promoter on hepatic GSTA1 and GSTA2 expression.

The patterns of expression of glutathione S-transferases A1 and A2 in human liver (hGSTA1 and hGSTA2, respectively) are highly variable, notably in the ratio of hGSTA1/hGSTA2. We investigated if this variation had a genetic basis by sequencing the proximal promoters (-721 to -1 nucleotides) of hGSTA1 and hGSTA2, using 55 samples of human liver that exemplified the variability of hGSTA1 and hGSTA2 expression. Variants were found in the hGSTA1 gene: -631T or G, -567T, -69C, -52G, designated as hGSTA1*A; and -631G, -567G, -69T, -52A, designated as hGSTA1*B. Genotyping for the substitution -69C > T by polymerase chain reaction restriction fragment length polymorphism (PCR-RFLP), showed that the polymorphism was widespread in Caucasians, African-Americans and Hispanics, and that it appeared to conform to allelic variation. Constructs consisting of the proximal promoters of hGSTA1*A, hGSTA1*B or hGSTA2, with luciferase as a reporter gene, showed differential expression when transfected into HepG2 cells: hGSTA1*A approximately hGSTA2 > hGSTA1*B. Similarly, mean levels of hGSTA1 protein expression in liver cytosols decreased significantly according to genotype: hGSTA1*A > hGSTA1-heterozygous > hGSTA1*B. Conversely, mean hGSTA2 expression increased according to the same order of hGSTA1 genotype. Consequently, the ratio of GSTA1/GSTA2 was highly hGSTA1 allele-specific. Because the polymorphism in hGSTA1 correlates with hGSTA1 and hGSTA2 expression in liver, and hGSTA1-1 and hGSTA2-2 exhibit differential catalysis of the detoxification of carcinogen metabolites and chemotherapeutics, the polymorphism is expected to be of significance for individual risk of cancer or individual response to chemotherapeutic agents.

Glutathione transferase activities in renal carcinomas and adjacent normal renal tissues: factors influencing renal carcinogenesis induced by xenobiotics.

In general, the biological activation of nephrocarcinogenic chlorinated hydrocarbons proceeds via conjugation with glutathione. It has mostly been assumed that the main site of initial conjugation is the liver, followed by a mandatory transfer of intermediates to the kidney. It was therefore of interest to study the enzyme activities of subgroups of glutathione transferases (GSTs) in renal cancers and the surrounding normal renal tissues of the same individuals (n = 21). For genotyping the individuals with respect to known polymorphic GST isozymes the following substrates with differential specificity were used: 1-chloro-2,4-dinitrobenzene for overall GST activity (except GST theta); 7-chloro-4-nitrobenzo-2-oxa- 1,3-diazole for GST alpha; 1,2-dichloro-4-nitro-benzene for GST mu; ethacrynic acid and 4-vinylpyridine for GST pi; and methyl chloride for GST theta. In general, the normal tissues were able to metabolize the test substrates. A general decrease in individual GST enzyme activities was apparent in the course of cancerization, and in some (exceptional) cases individual activities, expressed in the normal renal tissue, were lost in the tumour tissue. The GST enzyme activities in tumours were independent of tumour stage, or the age and gender of the patients. There was little influence of known polymorphisms of GSTM1, GSTM3 and GSTP1 upon the activities towards the test substrates, whereas the influence of GSTT1 polymorphism on the activity towads methyl chloride was straightforward. In general, the present findings support the concept that the initial GST-dependent bioactivation step of nephrocarcinogenic chlorinated hydrocarbons may take place in the kidney itself. This should be a consideration in toxicokinetic modelling.

Failure to demonstrate chemoprevention by the monoterpene perillyl alcohol during early rat hepatocarcinogenesis: a cautionary note.

The monoterpene perillyl alcohol (PA) is being considered as a useful chemopreventive and therapeutic agent against human cancers. However, no data are available on the effects of PA in the first stages of hepatocarcinogenesis. To study such effects, putatively initiated cells and preneoplastic foci in hepatocarcinogenesis were used as a model. Male Wistar rats were treated with a single dose of N:-nitrosomorpholine (NNM). Between days 4 and 91 after NNM, subgroups of rats received either PA (1 g/kg body wt/day) or phenobarbital (PB) (50 mg/kg body wt/day) in the diet. Since PA treatment reduced food intake, one control group was fed ad libitum, while a second control was pair fed between days 4 and 91. In order to enhance any treatment effects, all groups, including the controls, were treated with the potent tumor promoter PB after day 91 until the end of the experiment at day 266. Rats were killed at multiple time points and putatively initiated cells and preneoplastic foci were identified by staining positively for placental glutathione S-transferase (G+). The following results were obtained. (i) A few days after NNM treatment single G+ cells emerged; a considerable portion of which developed into foci. (ii) Treatment with PB resulted in an increase in number and size of G+ foci. (iii) PA treatment failed to reduce the number of G+ cells; it somewhat lowered rates of apoptosis in G+ foci and clearly increased their average size. (iv) Eighty-seven days of PA revealed no protective effect on day 266, but, similar to PB treatment, increased the growth of foci. In conclusion, PA exerted no detectable chemopreventive effect in the early stages of rat hepatocarcinogenesis. It rather exerted a PB-like tumor promoting activity. These data argue against a recommendation of PA as a chemopreventive agent for healthy humans.

Search for compounds that inhibit the genotoxic and carcinogenic effects of heterocyclic aromatic amines.

Over the last 30 years approximately 160 reports have been published on dietary compounds that protect from the mutagenic and carcinogenic effects of heterocyclic aromatic amines (HAAs). In the first section of this review, the current state of knowledge is briefly summarized. Based on the evaluation of the available data, various protective mechanisms are described, and the use of different methodologies for the detection of protective effects is critically discussed. In most antimutagenicity studies (>70%) bacterial indicators (predominantly Salmonella strain TA98) were used, and about 600 individual compounds and complex mixtures have been identified that attenuate the effects of HAAs. The most frequently used in vivo method to detect protective effects are adduct measurements; anticarcinogenic dietary factors were identified by aberrant crypt foci assays and liver foci tests with rats. The mechanisms of protection include inactivation of HAAs and their metabolites by direct binding, inhibition of enzymes involved in the metabolic activation of the amines, induction of detoxifying enzymes, and interaction with DNA repair processes. The detection spectrum of conventional in vitro mutagenicity assays with metabolically incompetent indicator cells is limited. These procedures reflect only simple mechanisms such as direct binding of the HAAs to pyrroles and fibers. It has been shown that these compounds are also effective in rodents. More complex mechanisms, namely, interactions with metabolic activation reactions are not adequately represented in in vitro assays with exogenous enzyme homogenates, and false-negative as well as false-positive results may be obtained. More appropriate approaches for the detection of protective effects are recently developed test systems with metabolically competent cells such as the human Hep G2 line or primary hepatocytes. SCGE tests and DNA adduct measurements with laboratory rodents enable the detection of antigenotoxic effects in different organs, including those that are targets for tumor induction by the amines. Medium term assays based on aberrant crypt foci in colon and liver foci tests have been used to prove that certain compounds that prevented DNA damage by HAAs also reduced their carcinogenic effects. These experiments are costly and time consuming and, due to the weak induction capacity of the amines, only pronounced anticarcinogenic effects can be detected. Over the years, a large bulk of data on HAA protective compounds has accumulated, but only for a few (e.g., fibers, pyrroles, constituents of teas, and lactic acid bacteria) is there sufficient evidence to support the assumption that they are protective in humans as well.

Effect of fumonisin B1 on the levels and fatty acid composition of selected lipids in rat liver in vivo.

The modulating role of fumonisin B1 (FB1) on lipid biosynthesis was evaluated in a short-term (21 day) experiment using male Fischer rats fed high dietary levels (50, 100 and 250 mg FB1/kg) and in a long-term (2 yr) experiment using male BD IX rats fed low dietary levels (1, 10 and 25 mg FB1/kg) of FB1. The total serum and liver cholesterol was significantly (P < 0.01) increased in the rats fed 250 mg FB1/kg diet for 21 days, while the liver phospholipids, sphingomyelin and phosphatidylethanolamine (PE) were significantly decreased (P < 0.01) and increased (P < 0.05), respectively. In the long-term study, only PE was significantly (P < 0.05) increased in all the FB1-treated animals. Fatty acid (FA) analysis of PE indicated that C18:2n-6 was significantly increased (P < 0.05 to P < 0.01) in the FB1-treated rats of the short-term study, while it was markedly (not significantly) increased in phosphatidylcholine (PC). The same pattern was observed in the PC and PE fractions of the liver of the FB1-treated rats from the long-term studies, but the changes were not significant due to the small number (three rats per group) of rats analysed. The levels of C22:5n-6 and C22:6n-3 were also markedly decreased and increased respectively in the 10 and 25 mg FB1/kg-treated groups. When the FAs were determined in the total lipids in a larger number of rats (four to six animals per group) the level of C18:2n-6 was significantly increased in the 10 (P < 0.01) and 25 (P < 0.05) mg FB1/kg-treated groups. Similar effects were noticed in plasma PC with respect to the C18:5n-6 and C22:55n-6 in both the long- and short-term treated groups, except that C20:4n-6 was also lower in both cases. The total n-6 FAs and polyunsaturated FAs were significantly (P < 0.01) and markedly reduced in PC and PE, respectively, of the rats fed the 250 mg FB1/kg diet. In the long-term experiment the n-6/n-3 ratio was significantly (P < 0.01) decreased in PE and markedly lowered in PC due to a significant (P < 0.05) increase in the n-3 FAs of both phospholipid fractions. The sphinganine/sphingosine ratio was significantly (P < 0.05) altered in the liver of the rats fed the 100 and 250 mg FB1/kg diets for 21 days, while in the long-term study no significant changes were noticed in either the liver or sera. The present data indicate that FB1 affects lipid biosynthesis in rat liver and plasma differently, depending on the dietary level and duration of treatment. Alterations to the n-3 and n-6 FA biosynthetic pathways, detected in rats fed relatively low dietary levels of FB1, are likely to be important mediators for FB1-induced effects on hepatocyte cell proliferation.

Chemoprotection against the formation of colon DNA adducts from the food-borne carcinogen 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) in the rat.

The mutagenic heterocyclic aromatic amine, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP), is a pyrolysis product in cooked foods that has been shown to be a rat colon carcinogen and has been implicated in the etiology of human colon cancer. In order to identify chemoprotection strategies that could be carried out in humans, a pilot study was conducted in which PhIP-DNA-adduct levels were quantified in the colons of male F344 rats that had been subjected to 16 different putative chemoprotection regimens, followed by a gavage of PhIP (50 mg/kg) and sacrifice 24 h later. The 16 treatments (Oltipraz, benzylisothiocyanate, diallyl sulfide, garlic powder, ethoxyquin, butylated hydroxyanisole, glutathione, indole-3-carbinol, alpha-angelicalactone, kahweol/cafestol palmitates, quercetin, green tea, black tea, tannic acid, amylase-resistant starch, and physical exercise) comprised sulfur-containing compounds, antioxidants, flavonoids, diterpenes, polyphenols, high dietary fiber, etc. The strongest inhibition of PhIP-DNA adduct formation in the colon was observed upon pretreatment with black tea, benzylisothiocyanate, and a mixture (1:1) of kahweol:cafestol palmitates, which resulted in 67, 66, and 54% decreases in colon PhIP-DNA adduct levels, as compared with controls. Preliminary studies on their mechanism of action indicated that only kahweol:cafestol caused a substantial induction of glutathione S-transferase isozymes (GSTs) that are thought to be important in the detoxification of PhIP. Notably, this induction occurred in the liver rather than in the colon.

Inhibition instead of enhancement of lipid peroxidation by pretreatment with the carcinogenic peroxisome proliferator nafenopin in rat liver exposed to a high single dose of corn oil.

Oxidative stress is discussed as a possible hepatocarcinogenic mechanism of peroxisome proliferators (PP) in rodents and is suggested to result from the induction of peroxisomal beta-oxidation (PBOX) by PP. The induced PBOX is assumed to produce excessive H2O2 from the degradation of fatty acids, ultimately leading to oxidative stress and lipid peroxidation. In the present short term-study, we attempted to stimulate lipid peroxidation in male Wistar rats by (1) inducing PBOX enzymes with the peroxisome proliferator nafenopin at 90 mg/kg body weight per day in the diet for 10-11 days, and (2) by supplying the induced PBOX with an abundant amount of fatty acid as substrate, using a corn oil gavage at 20 ml/kg body weight. The corn-oil gavage alone, i.e. without preceding nafenopin treatment, enhanced liver triacylglycerol nine- to tenfold and hepatic lipid peroxidation, measured as thiobarbituric acid reactive substances (TBARS), was increased 50% compared with controls. Both observations were made after 18 h when the peak elevations occurred. Upon pretreatment with nafenopin, associated with a sevenfold induction of PBOX, the corn oil gavage however caused only a threefold maximal increase in hepatic triacylglycerol, also at the 18 h time-point; TBARS remained almost at control levels, as monitored at seven time points over 24-25 h. These results suggest that nafenopin reduces rather than enhances lipid peroxidation, despite the provision, in a short term study, of high doses of substrate to the induced enzyme system that is hypothetically causing oxidative stress in the liver.

Hepatocarcinogenic potential of di(2-ethylhexyl)phthalate in rodents and its implications on human risk.

The plasticizer di(2-ethylhexyl) phthalate (DEHP), to which humans are extensively exposed, was found to be hepatocarcinogenic in rats and mice. DEHP is potentially set free from objects made of synthetic materials (e.g., those used in medicine). Chronically, the greatest amounts are transferred to persons undergoing hemodialysis (up to 3.1 mg/kg b.w. per day) who would thus be considered the individuals most endangered by tumorigenesis. Although toxicokinetics seem to play a certain unclear role in the course of DEHP-related toxicity, toxicodynamic factors appear more decisive. DEHP is a representative of "peroxisome proliferators" (PP), a distinct group of substances that, in rodents, do not only induce peroxisomes but also specific enzymes in other organelles, organ growth, and DNA synthesis. The cluster of the characteristic effects of PP is generally, although perhaps not quite appropriately summarized as "peroxisome proliferation," and is strongest in the liver. The lowest observed effect level (LOEL) and the no observed effect level (NOEL) of peroxisome proliferation in the rat, as determined by the induction of specific enzymes (peroxisomal beta-oxidation, carnitine-acetyl-transferase, cytochrome P-452), DNA synthesis, and hepatomegaly, may be assumed as 50 and 25 mg/kg b.w. per day, respectively. DEHP and other carcinogenic PP are neither genotoxic nor tumor initiators, but they appear to be tumor promoters, also implicating a threshold level for the carcinogenic effect. Although a causal relationship between a particular effect of peroxisome proliferation and hepatocarcinogenesis is as yet unknown, peroxisome proliferation as a whole phenomenon appears to be associated with the potential of tumor induction, as shown by comparison of the relative strength of individual PP and by comparison of species and organ specificities. Likewise, LOEL and NOEL of rodent carcinogenesis, that is, 300 and 50 to 100 mg/kg b.w. per day, respectively, are above but not too far from the corresponding values for the investigated parameters of peroxisome proliferation. Thus, with respect to dose alone, worst-case exposure in hemodialysis patients is at least 16-fold below the LOEL of any characterized PP-specific effect of DEHP and approximately 100-fold below that of DEHP-related tumorigenesis. Also, primates are less responsive to PP than rats with respect to the investigated biochemical and morphological parameters. If this lower primate responsiveness is extrapolated to estimate carcinogenicity in humans, we might thus arrive at an even larger safety margin than when based on exposure alone. Doses of PP hypolipidemics that had clearly induced several indicators of peroxisome proliferation in rats did not cause any clear-cut enhancements in the peroxisomes of patients, even though most of these hypolipidemics were considerably stronger PP than DEHP. Thus, an actual threat to humans by DEHP seems rather unlikely. Accordingly, hepatocarcinogenesis was neither enhanced in workers exposed to DEHP nor in patients treated with hypolipidemics.

Inhibition of repairable DNA-damage in Escherichia coli K-12 cells recovered from various organs of nitrosamine-treated mice by vitamin A, phenethylisothiocyanate, oleic acid and triolein.

The influence of various dietary constituents--phenethylisothiocyanate (PEITC), oleic acid (OA), triolein (TO), and vitamin A (ROL)--on the genotoxic activity of nitrosamines (NDMA, NDELA, NPYR) was investigated. For this purpose differential DNA repair assays with Escherichia coli K-12 strains were performed in vitro and in vivo with mice. Under in vitro conditions (liquid holding), all compounds reduced nitrosamine induced DNA-damage in the indicator bacteria in the dose range 1-10 micrograms/ml, the ranking order of efficiency being PEITC greater than OA greater than ROL greater than or equal to TO. In animal-mediated assays, acute oral treatment with PEITC (17-150 mg/kg), 2 h before nitrosamine administration, resulted in a marked decrease of nitrosamine genotoxicity in liver, kidneys, lungs and in the blood. Also in other organs (spleen, testes) an increase in differential survival (which serves as a measure for repairable DNA damage) occurred. With ROL only a comparatively moderate antigenotoxic effect was obtained at a high dose level (250 mg/kg) under identical experimental conditions. OA (2000 mg/kg) and TO (16,000 mg/kg) were completely inactive. Upon repeated treatment (consecutive oral administration of the putative antigenotoxins over 4 days, a final treatment 24 h before nitrosamine administration) PEITC (150 mg/kg/day), ROL (80 mg/kg/day) and OA (2000 mg/kg/day) had no influence on the genotoxic effects of the nitrosamines. Repeated treatment with TO (4000-16,000 mg/kg/day) resulted in a moderate dose-dependent reduction of NDMA-induced DNA-damage in the indicator bacteria, whereas in combination with NPYR only a marginal effect was observed. Biochemical experiments indicated that the antigenotoxic effects of PEITC seen under in vivo conditions were due to inhibition of alpha-hydroxylation of the nitrosamines, whereas ROL and TO appeared not to interfere strongly with this metabolic activation step. Our results indicate that in vitro assays do only partly reflect the antigenotoxic properties of the different food constituents in vivo and that animal-mediated DNA repair assays with E. coli strains are an appropriate approach to study the effects of modifiers of nitrosamine genotoxicity in the living animal.