A mechanism-based kinetic analysis of succinimide-mediated deamidation, racemization, and covalent adduct formation in a model peptide in amorphous lyophiles

The succinimide intermediate generated during deamidation of asparagine-containing peptides and proteins has been implicated as having a role in the formation of multiple types of degradants in addition to hydrolysis products, including racemization products and, more recently, amide-linked, nonreducible protein and peptide aggregates. The formation of alternative degradants may be particularly important in solid-state formulations. This study quantitatively examines the role of the succinimide intermediate in hydrolysis, racemization, and covalent, amide-linked adduct formation in amorphous lyophiles. The degradation of a model peptide, Gly-Phe-L-Asn-Gly, and its L- or D-succinimide intermediates were examined in lyophiles containing hydroxypropyl methylcellulose and varying amounts of excess Gly-Val. Disappearance of the starting reactants and formation of up to 10 degradants were monitored when lyophiles were exposed to either 27°C/40% relative humidity (RH) or 40°C/75 RH using a stability indicating high-performance liquid chromatography method. Terminal degradant profiles were the same when the starting reactant was either Gly-Phe-L-Asn-Gly or its succinimide intermediate. Nucleophilic attack occurred preferentially at the α-carbonyl of the succinimide intermediate at ratios of approximately 2:1 for both water and the N-terminus of Gly-Val as the attacking nucleophiles. A mechanism-based kinetic model analysis indicates that hydrolysis, racemization, and covalent, amide-linked adduct formation all proceed via the succinimide intermediate.     

Mechanism of covalent adduct formation of aucubin to proteins

The iridoid glucoside aucubin can irreversibly bind to proteins through the formation of its aglycone. In view of a possible involvement of these protein adducts in the toxicity of aucubin, we investigated the mechanism of binding of aucubin to proteins. [3H]aucubin in itself did not result in binding to protein whereas it covalently bound to rat serum albumin as a function of exposure time and dose in the presence of beta-glucosidase. The rate and extent of protein binding were significantly increased in the presence of the imine-trapping agent sodium cyanide. Oral administration of [3H]aucubin to rats showed that the total radioactivity in plasma remained at a similar level for up to 6 h once peak level was reached, suggesting that a considerable amount of radioactivity might be covalently associated with plasma proteins. The levels of radioactivity in the liver and kidney after oral dosing were higher than those after i.v. dosing. These results indicate that the open-chain aglycone of aucubin can form an imine bond with a nucleophilic site of the protein and these irreversible bindings may partially contribute to its biological and toxic effects.     

Effects of benzene metabolite treatment on granulocytic differentiation and DNA adduct formation in HL-60 cells

 Reactive metabolites of benzene (BZ) play important roles in BZ-induced hematotoxicity. Although reactive metabolites of BZ covalently bind to DNA, the significance of DNA adduct formation in the mechanism of BZ toxicity is not clear. These studies investigated the covalent binding of the BZ metabolites hydroquinone(HQ) and 1,2,4-benzenetriol(BT) using the DNA [³²P]postlabeling method and explored the potential relationship between DNA adduct formation and cell differentiation in human promyelocytic leukemia (HL-60) cells, a model system for studying hematopoiesis. Maturation of HL-60 cells to granulocytes, as assessed by light and electron microscopy, was significantly inhibited in cells that were pretreated with HQ or BT prior to inducing differentiation with retinoic acid (RA). The capacity of RA-induced cells to phagocytose sheep red blood cells (RBC) and to reduce nitroblue tetrazolium (NBT), two functional parameters characteristic of mature, differentiated neutrophils, was also inhibited in cells pretreated with HQ or BT. These BZ metabolite treatments induced DNA adduct formation in HQ- but not in BT-treated cells. These results indicate that whereas HQ and BT each block granulocytic differentiation in HL-60 cells, DNA adducts were observed only following HQ treatment. Thus DNA adduct formation may be important in HQ but not in BT toxicity. Received: 24 April 1995 / Accepted: 5 July 1995     

Application of a two-state kinetic model to the heterogeneous kinetics of reaction between cysteine and hydrogen peroxide in amorphous lyophiles

The bimolecular reaction between cysteine (CSH) and hydrogen peroxide (H(2)O(2)) in amorphous PVP and trehalose lyophiles has been examined at different reactant and excipient concentrations and at varying pH and temperature. Initial rates of product formation and complete reactant and product concentration-time profiles were generated by HPLC analyses of reconstituted solutions of lyophiles stored for various periods of time. While only cystine (CSSC) forms in aqueous solutions, cysteine sulfinic (CSO(2)H) and sulfonic (CSO(3)H) acids are significant degradants in amorphous solids. The formation of alternative degradants was consistent with the solution reaction mechanism, which involves a reactive sulfenic acid (CSOH) intermediate, coupled with the restricted mobility in the amorphous solid-state, which favors reaction of CSOH with the smaller, mobility-advantaged H(2)O(2) over its reaction with cysteine. Complex rate laws (i.e., deviations from 1st order for each reactant) observed in initial rate studies and biphasic concentration-time profiles in PVP were successfully fitted by a two-state kinetic model assuming two reactant populations with different reactivities. The highly reactive population forms CSSC preferentially while the less reactive population generates primarily sulfinic and sulfonic acids. Reactions in trehalose could be described by a simple one-state model. In contrast to the reaction in aqueous solutions, the 'pH' effect was minimal in amorphous solids, suggesting a change in the rate-determining step to diffusion control for the model reaction occurring in amorphous lyophiles.     

Studies on the formation of methoxychlor-protein adduct in rat and human liver microsomes. Is demethylation of methoxychlor essential for cytochrome P450 catalyzed covalent binding?

Previous studies demonstrated that liver microsomal monooxygenases metabolize the pesticide methoxychlor into phenolic estrogenic derivatives. Additionally, methoxychlor is activated by the hepatic cytochrome P450 monooxygenase to bind covalently to microsomal proteins (Bulger WH, Temple JE and Kupfer D, Toxicol Appl Pharmacol 68: 367-374, 1983). The current study examines, in liver microsomes from control and phenobarbital-treated rats and humans, whether demethylation of methoxychlor is essential for covalent binding and whether demethylated methoxychlor metabolites are on the pathway of formation of the reactive intermediate and protein adduct. Using 3H-methoxyl-labeled and 14C-ring-labeled methoxychlor, it was demonstrated that demethylation is not essential for covalent binding. Namely, the major portion of the methoxychlor moiety in the protein adduct was found to contain intact methoxyls. Nevertheless, in the absence of methoxychlor, both the mono- and bis-demethylated methoxychlor metabolites could undergo monooxygenase-mediated covalent binding to proteins. This was demonstrated in incubations of purified 14C-labeled mono- and bis-demethylated methoxychlor metabolites with liver microsomes, in the presence of NADPH. Additionally, the dehydrochlorinated metabolite of methoxychlor, containing a double bond, underwent covalent binding, which exhibited characteristics similar to those of methoxychlor. These findings demonstrated that the protein adduct from relatively brief incubation periods contains a methoxychlor derivative with intact methoxyls. The possibility that the activation of methoxychlor involves modification of the side chain, which is the active site that binds to proteins, is discussed.     

Metabolism and hemoglobin adduct formation of acrylamide in humans.

Acrylamide (AM), used in the manufacture of polyacrylamide and grouting agents, is produced during the cooking of foods. Workplace exposure to AM can occur through the dermal and inhalation routes. The objectives of this study were to evaluate the metabolism of AM in humans following oral administration, to compare hemoglobin adduct formation on oral and dermal administration, and to measure hormone levels. The health of the people exposed under controlled conditions was continually monitored. Prior to conducting exposures in humans, a low-dose study was conducted in rats administered 3 mg/kg (1,2,3-13C3) AM by gavage. The study protocol was reviewed and approved by Institute Review Boards both at RTI, which performed the sample analysis, and the clinical research center conducting the study. (1,2,3-13C3) AM was administered in an aqueous solution orally (single dose of 0.5, 1.0, or 3.0 mg/kg) or dermally (three daily doses of 3.0 mg/kg) to sterile male volunteers. Urine samples (3 mg/kg oral dose) were analyzed for AM metabolites using 13C NMR spectroscopy. Approximately 86% of the urinary metabolites were derived from GSH conjugation and excreted as N-acetyl-S-(3-amino-3-oxopropyl)cysteine and its S-oxide. Glycidamide, glyceramide, and low levels of N-acetyl-S-(3-amino-2-hydroxy-3-oxopropyl)cysteine were detected in urine. On oral administration, a linear dose response was observed for N-(2-carbamoylethyl)valine (AAVal) and N-(2-carbamoyl-2-hydroxyethyl)valine (GAVal) in hemoglobin. Dermal administration resulted in lower levels of AAVal and GAVal. This study indicated that humans metabolize AM via glycidamide to a lesser extent than rodents, and dermal uptake was approximately 6.6% of that observed with oral uptake.     

Identification of the major covalent adduct formed in vitro and in vivo between acetaminophen and mouse liver proteins

Improved analytical methodology has been developed for the structural characterization of covalently bound drug-protein adducts and has been applied to an investigation of the conjugates formed in vivo and in vitro between [14C]acetaminophen and mouse liver proteins. The major adduct released by acid hydrolysis of hepatic protein samples, which accounted for approximately 70% of the bound radioactivity in vivo and in vitro, was identified as 3-cystein-S-yl-4-hydroxyaniline, a derivative whose structure reflects the predominance of acetaminophen thioether adducts in drug-modified proteins. It is concluded that the reactive, electrophilic metabolite of acetaminophen, which most likely is N-acetyl-p-benzoquinoneimine, binds with a high degree of selectivity to cysteinyl thiol groups on protein, formally in a Michael-type addition reaction. Cysteine residues thus represent primary target sites for arylation by the reactive metabolite of acetaminophen, and proteins rich in free thiols may be especially vulnerable to damage by this toxic intermediate.     

Varying effects of sulfhydryl nucleophiles on acetaminophen oxidation and sulfhydryl adduct formation

The effects of glutathione, cysteine, N-acetylcysteine, cysteamine, α-mercaptopropionylglycine and methionine on the NADPH-dependent metabolism and covalent binding of acetaminophen have been examined in mouse liver microsomal incubations. With the exception of methionine, all of the nucleophiles decreased covalent binding by forming adducts with the electrophilic metabolite of acetaminophen. The adducts were measured quantitatively by high pressure liquid chromatography. In contrast to glutathione, N-acetylcysteine and α-mercaptopropionylglycine, both cysteamine and cysteine in high concentrations also decreased covalent binding of acetaminophen through another mechanism, inhibition of the formation of the reactive acetaminophen metabolite. These results indicate that both inhibition of metabolite formation and detoxification of metabolite by sulfhydryl adduct formation are mechanisms that can be important in reducing acetaminophen toxicity in overdosed patients treated with these nucleophiles.     

Enzymatic biomarker measurement and study of DNA adduct formation in benzo

The aim of this study was to improve the knowledge on the metabolic pathways involved in benzo[a]pyrene (B[a]P) activation and on the relationship between adduct levels and enzymatic biomarker activities. With this purpose, a model to assess pollutant exposure via food supply has been developed for the sentinel organism, Mytilus galloprovincialis. Mussels were fed for 4 weeks with B[a]P-contaminated feed (50 mg/kg dry weight mussel). Bioaccumulation was studied by determination of B[a]P concentration in whole mussel by GC/MS analysis. Different biomarkers of pollutant exposure were measured to assess the metabolic state of the exposed organisms. CYP1A-like immunopositive protein titration and B[a]P hydroxylase (BPH) activity were assessed as indicators of phase I biotransformation. Glutathione-S-transferase (GST) activity was measured as an indicator of the conjugation activities. Catalase (CAT) and DT-diaphorase (DTD) activities were assessed as potential biomarkers of oxidative stress, whereas acetylthiocholine esterase (AChE) activity was measured as an indication of possible neurotoxicity of B[a]P exposure. DNA adduct levels were determined in digestive gland DNA by applying the 32P-postlabeling technique with nuclease P1 enhancement. For the developed conditions of exposure, B[a]P concentration reached in whole mussel tissues was very high (>500 mg/kg d.w. mussel) and significant B[a]P-induced changes were recorded for each enzymatic biomarkers. BPH and CAT activities were significantly increased by B[a]P exposure, whereas GST in the gills, DTD and AChE were significantly depressed. On the other hand, no change in CYP1A-like immunopositive protein content was observed. Induction and increase with time of bulky B[a]P-related DNA adducts were demonstrated in the digestive gland, although at low levels (0.269+/-0.082 adduct/10e8 dNps at maximum) by the 32P-postlabeling assay. DNA adduct level was significantly correlated with whole mussel tissue B[a]P concentration, so were all the enzymatic biomarkers measured except to GST activity in both gill and digestive gland tissues. BPH, DTD, CAT and AChE displayed a strong correlation with adduct levels. These results demonstrate the neurotoxicity and the genotoxicity of B[a]P exposure in the mussel. The induction of bulky DNA adducts in mussels demonstrates the existence of activation pathways already identified in vertebrates. It validates also the suitability of this model for further studies on B[a]P metabolism in mussels. Our results support the proposal of BPH, AChE, DTD and CAT activities as suitable biomarkers of PAH exposure for these sentinel species.     

The effects of changes in glutathione levels through exogenous agents on intracellular cysteine content and protein adduct formation in chronic alcohol-treated VL17A cells

Alcohol-mediated liver injury is associated with changes in the level of the major cellular antioxidant glutathione (GSH). It is interesting to investigate if the changes in intracellular GSH level through exogenous agents affect the intracellular cysteine content and the protein adduct formation indicative of oxidative insult in chronic alcohol treated liver cells. In VL-17A cells treated with 2?mM N-acetyl cysteine (NAC) or 0.1?mM ursodeoxycholic acid (UDCA) plus 100?mM ethanol, an increase in cysteine concentration which was accompanied by decreases in hydroxynonenal (HNE) and glutathionylated protein adducts were observed. Pretreatment of 100?mM ethanol treated VL-17A cells with 0.4?mM buthionine sulfoximine (BSO) or 1?mM diethyl maleate (DEM) had opposite effects. Thus, altered GSH level through exogenous agents may either potentiate or ameliorate chronic alcohol-mediated protein adduct formation and change the cysteine level in chronic alcohol treated VL-17A cells. The gene expression of non-treated and ethanol-treated hepatocytes in 2 microarray datasets was also compared to locate differentially expressed genes involved in cysteine metabolism. The study demonstrates that increased protein adducts formation and changes in cysteine concentration occur under chronic alcohol condition in liver cells which may increase alcohol-mediated oxidative injury.     

Proteomic analysis of acrylamide-protein adduct formation in rat brain synaptosomes

Evidence suggests that the neurological defects (gait abnormalities, foot splay, and skeletal muscle weakness) associated with acrylamide (ACR) intoxication are mediated by impaired neurotransmission at central and peripheral synapses. ACR can form adducts with nucleophilic residues on proteins and thereby alter corresponding structure and function. To evaluate protein adduction in nerve terminals as a possible mechanism of action, recombinant N-ethylmaleimide sensitive factor (NSF) was exposed in vitro to ACR (10 micromol) and mass spectrometry (MS) was used to identify adduct sites. MS analyses demonstrated that ACR formed adducts with sulfhydryl groups on cysteine residues (carbamoylethylcysteine, or CEC) of NSF. Ex vivo incubation of whole brain synaptosomes with ACR (0.001-1.0 M) produced concentration-dependent increases in CEC that were inversely correlated to reductions in neurotransmitter release that occurred over the same neurotoxicant concentration range. In synaptosomes isolated from rats intoxicated at a higher (50 mg/kg per day x 3, 5, 8, or 11 days) or a lower (21 mg/kg per day x 14, 21, or 28 day) ACR dose rate, CEC levels increased progressively up to a moderate level of neurotoxicity. To identify protein adducts, synaptosomal proteins labeled by ex vivo 14C-ACR exposure were separated by gel electrophoresis and probed by immunoblot analysis. Results showed that NSF and the SNARE protein, SNAP-25, were tentative ACR targets. Subsequent experiments indicated that ACR exposure increased synaptosomal levels of the 7S SNARE core complex, which is consistent with inhibition of NSF, SNAP-25 function, or both. These data suggest that adduction of cysteine residues on NSF and certain SNARE proteins might be causally involved in the nerve terminal dysfunction induced by ACR.     

Effects of nitroprusside and nitroglycerin on cGMP content and PGI2 formation in aorta and vena cava

Nitroprusside (NP) and nitroglycerin (NG) are potent vasodilators that are used clinically on the basis of their abilities to cause relaxation of smooth muscle. In vitro, both agents cause activation of guanylate cyclase, resulting in increased intracellular cGMP. They also have effects on arachidonate metabolism. Despite apparent similarities in their mechanisms of action, the two drugs have different therapeutic applications based in part on differences in their effectiveness on the arterial and venous systems in vivo. To understand better their target tissue preference, slices of aorta and vena cava were incubated with the agents; cGMP and the vasodilatory prostanoid, prostacyclin, were quantified. NP was more effective in increasing the cGMP content of aorta than of vena cava; it was more active than NG in both tissues. Prostaglandin formation by vascular tissue was influenced by the preliminary equilibration period. Under optimal conditions, it appeared that NG enhanced prostacyclin formation in aorta more than did NP. This in vitro model for NP and NG action may be useful in studying the mechanisms of action of these and other vasoactive agents.     

Effects of butylated hydroxyanisole and butylated hydroxytoluene on dna adduct formation and arylamine N‐acetyltransferase activity in human bladder tumour cells

In this study, butylated hydroxyanisole (BHA) and butylated hydroxytoluene (BHT) were used to determine the inhibition of arylamine N‐acetyltransferase (NAT) activity and DNA adduct formation in human bladder tumour cell line T‐24. The activity of NAT was measured by high‐performance liquid chromatography, assaying for the amounts of N‐acetyl‐2‐aminofluorene and N‐acetyl‐p‐aminobenzoic acid and remaining 2‐aminofluorene and p‐aminobenzoic acid. Human bladder tumour cell line T‐24 cytosols and intact cells were used for examining NAT activity and carcinogen–DNA adduct formation. The results demonstrated that NAT activity and 2‐aminofluorene–DNA adduct formation in human bladder tumour cells were inhibited and decreased by BHA and BHT in a dose‐dependent manner. The effects of BHA and BHT on the values of the apparent K_m and V_max also were determined in both systems examined. The results indicated that BHA and BHT decreased the apparent values of K_m and V_max from human bladder tumour cells in both cytosol and intact cells. Copyright © 2002 John Wiley & Sons, Ltd.     

Effects of intermittent exposure to aflatoxin B1 on DNA and RNA adduct formation in rat liver: dose-response and temporal patterns.

We studied the effects of intermittent exposure to aflatoxin B1 (AFB1) on hepatic DNA and RNA adduct formation. Fisher-344 male rats were fed 0.01, 0.04, 0.4, or 1.6 ppm of AFB1 intermittently for 8, 12, 16, and 20 weeks, alternating with 4 weeks of dosing and 4 weeks of rest. Other groups of rats were fed 1.6 ppm of AFB1 continuously for 4, 8, 12, and 16 weeks. Control rats received AFB1-free NIH-31 meal diet. AFB1-DNA and -RNA adducts were measured by HPLC with fluorescence detection. The data are presented as total DNA or RNA adducts. The DNA and RNA adduct levels increased or decreased depending on the cycles of dosing and rest. Rats removed from treatment 1 month after 1 or 2 dosing cycles (8 and 16 weeks of intermittent exposure) showed approximately a twofold decrease in DNA adduct levels and a two- to elevenfold decrease in RNA adduct levels compared with rats euthanized immediately after the last dosing cycle (12 and 20 weeks of intermittent exposure). Our data indicate that DNA and RNA adducts increased linearly, from 0.01 ppm to 1.6 ppm of AFB1 after 12 and 20 weeks of intermittent treatment. A linear dose response was also apparent for DNA but not for RNA adducts after 8 and 16 weeks of treatment. As biomarkers of exposure, AFB1-RNA adducts were three to nine times more sensitive than AFB1-DNA adducts but showed greater variability. These results suggest that binding of AFB1 to hepatic DNA is a linear function of the dose, regardless of the way this is administered. The dose-response relationship for RNA adducts depends on the length of the no-dosing cycles and on the turnover rate of RNA.     

Immunochemical detection of flucloxacillin adduct formation in livers of treated rats

Flucloxacillin is a semi-synthetic penicillin widely used in the prophylaxis and treatment of staphylococcal infections. Severe liver reactions, characterised by delayed cholestatic hepatitis and a prolonged course of recovery, are associated with flucloxacillin therapy. Clinical findings are suggestive of an immune mediated reaction but there exists little supporting experimental evidence. The formation of drug modified hepatic protein adducts has been proposed to play an important role in the hepatotoxicity of many drugs. The aim of the present study was to investigate whether flucloxacillin treatment results in adduct formation in vivo. Flucloxacillin was conjugated to rabbit serum albumin by formation of a penicilloyl determinant and used as an immunogen to raise a polyclonal antiserum specific for flucloxacillin-modified proteins. Antibody specificity was confirmed by competitive enzyme-linked immunosorbent assay (ELISA) with free drug. The antiserum was used in combination with western blotting to detect adduct formation in the livers of flucloxacillin treated rats. Western blot analysis of rat liver subcellular fractions revealed the formation of six flucloxacillin adducts in various subcellular fractions. These studies demonstrate for the first time that treatment with flucloxacillin results in the formation of hepatic protein adducts.     

Effects of dipyridamole in vivo on ATP and cAMP content in platelets and arterial walls and on atherosclerotic plaque formation

Summary In rabbits receiving an atherogenic diet for 2 months, the ATP content of platelet rich plasma (PRP) and arterial tissue was significantly elevated as compared to normal rabbits. This increase in ATP levels of platelets from atherosclerotic rabbits was paralleled by higher basal as well as PGI₂-induced cAMP levels. In arterial tissues, an increase was only obtained in PGI₂-stimulated cAMP content.Treatment with dipyridamole (DPD) for 4 weeks resulted in a reduction of the ATP content in platelets and arterial tissue from atherosclerotic rabbits to values seen in normal animals. Again, the reduction of ATP content was reflected in a decrease of basal as well as PGI₂-induced cAMP levels in platelets, whereas in arterial tissue a decrease was only obtained in PGI₂-induced cAMP content. At the same time, DPD treatment enhanced atherosclerotic plaque formation in the aortic wall.The enhanced atherosclerotic plaque formation seen in DPD treated atherosclerotic rabbits may be linked to the inhibition of adenosine uptake, resulting in a decrease of the adenine nucleotide pools of arterial wall cells. The decrease also caused a reduction in PGI₂-induced cAMP content. This effect may be linked to altered proliferative activity, since in many cell types, stimulation of cAMP levels results in reduced proliferation rates.     

Effects of solvent on DNA adduct formation in skin and lung of CD1 mice exposed cutaneously to benzo(a)pyrene

The effect of solvent polarity and lipophilicity on DNA adduct formation by polycyclic aromatic hydrocarbons in skin and lung has been studied in CD1 mice exposed cutaneously in vivo to benzo(a)pyrene (∼0.01–7.0 μg/animal) in either tetrahydrofuran or n-dodecane. The nature and amounts of DNA adducts, measured as 7R,8S,9R-trihydroxy-10S-(N²-deoxyguanosyl-3′-phosphate)-7,8,9,10-tetrahydrobenzo(a)pyrene, in relation to exposure dose and treatment regime was determined by ³²P-postlabelling. In skin DNA there was a linear relationship between exposure dose and adduct formation with both solvents, though the amount of adduct formed was significantly lower from treatment with benzo(a)pyrene in n-dodecane than in tetrahydrofuran. The amounts of adducts measured in skin DNA ranged from 67 amol adducts/μg DNA at the lowest exposure dose of benzo(a)pyrene in n-dodecane to 3.5 fmol adducts/μg DNA (1 adduct in 5 × 10⁷ nucleotides to 1 adduct in 9 × 10⁵ nucleotides) at the highest dose. In tetrahydrofuran the corresponding levels were 89 amol adducts/μg DNA (1 adduct in 3 × 10⁷ nucleotides) to 16.9 fmol adducts/μg DNA (1 adduct in 2 × 10⁵ nucleotides). DNA adducts could not be detected in lung tissue following cutaneous treatment of animals with benzo(a)pyrene in n-dodecane. Cutaneous treatment of animals with benzo(a)pyrene in tetrahydrofuran, however, resulted in adducts in lung DNA at a level of 88 amol/μg DNA from exposures only at the highest dose (6.72 μg/animal). The difference in octanol-water partition coefficient, log P_ow between n-dodecane compared to tetrahydrofuran is considered to be the most likely reason for the reduction in the bioavailability of benzo(a)pyrene and/or its metabolites and hence the degree of genotoxicity in tissues. The results suggest that other paraffinic hydrocarbon solvents may moderate the genotoxicity of polycyclic aromatic hydrocarbons in vivo. The assessment of the genotoxicity in vivo of mixtures of compounds should be carried out on complete mixtures of substances of interest in order to take account of these possible antagonistic or synergistic effects. Received: 10 February 1999 / Accepted: 10 May 1999