Adrenal proteins bound by a reactive intermediate of mitotane.

PURPOSE: Mitotane (o,p'-DDD), is the only adrenolytic agent available for the treatment of adrenocortical carcinoma. Previous studies have shown that mitotane covalently binds to adrenal proteins following its metabolism in adrenocortical tissue to a reactive acyl chloride intermediate. It was the objective of this study to compare the electrophoresis separation patterns of such adducts following activation of mitotane by various adrenocortical sources. METHODS: With the use of a 125I-labeled analog of mitotane, 1-(2-chlorophenyl)-1-(4-iodophenyl)-2,2-dichloroethane, gel electrophoresis patterns were obtained for homogenates from bovine, canine and human adrenocortical preparations as well as from a human adrenal preparation. Western immunoblotting analysis was used to test the resulting patterns for adducts of cytochrome P-450scc and adrenodoxin. RESULTS: The electrophoresis separations were similar for all preparations, with bands at apparent molecular weights of 49.5 and 11.5 kDa being the most pronounced. Radiolabeling of the proteins of a human adrenal cancer cell line NCI H-295 was weak, but a band at 11.5 kDa was detected. Western immunoblotting analyses indicated that the band at 49.5 kDa corresponded in molecular weight to that of adrenal cytochrome P-450scc, but the band at 11.5 kDa did not correspond to adrenodoxin. CONCLUSIONS: The similarity of the results with canine and bovine adrenal preparations to that of human material offers useful systems for studying mitotane and its analogs. This should aid in understanding the mechanism of action of mitotane and in the design of compounds for the treatment of adrenocortical carcinoma.

Bovine adrenal cortex transformations of mitotane [1-(2-chlorophenyl)-1-(4-chlorophenyl)-2,2-dichloroethane; o,p'-DDD] and its p,p'- and m,p'-isomers.

The adrenalytic activity of mitotane (o,p'-DDD) has made it useful in the treatment of adrenocortical carcinoma and Cushing's syndrome. In support of a study to develop mitotane analogs as more effective therapeutic agents and as a basis for understanding the toxicity of related compounds in the adrenals, the biotransformations of o,p'-DDD in adrenocortical homogenate preparations have been studied and compared with those of its m,p'- and p,p'-isomers. Aliphatic oxidation to the corresponding acetic acid derivative, o,p'-, m,p'- or p,p'-DDA, was the major transformation for all the preparations. In the comparisons of the DDD isomers, the order of both DDA formation and apparent covalent binding was o,p'- > m,p'- > p,p'-DDD. There was also evidence for alpha-hydroxylation at the benzylic carbon with subsequent loss of water to form ethylene derivatives. This was a minor pathway for o,p'-DDD, but was the major pathway for the other two isomers. Thus, while the total yields of metabolites of o,p'- and m,p'-DDD were similar and at least twice that of the p,p'-isomer, their distribution of metabolites differed significantly. The effects of the three isomers on cell growth and cortisol production with the human adrenocortical carcinoma cell line, NCI H-295, followed the same order as their DDA formation and tissue binding. It is proposed that hydroxylation by the adrenal cortex at the beta-carbon leads to an adrenalytic effect, whereas hydroxylation at the alpha-carbon would represent an alternate deactivation pathway.

Metabolic activation and binding of mitotane in adrenal cortex homogenates.

Mitotane [1-(2-chlorophenyl)-1-(4-chlorophenyl)-2,2-dichloroethane, o,p'-DDD] is an adrenocorticolytic agent of value in the treatment of adrenocortical carcinoma and Cushing's syndrome. In support of a program to develop agents superior to mitotane, it is the purpose of this study to explore the relationship of the metabolism of mitotane to its binding to adrenal cortex tissue from several sources. The objective was to detect the mitotane moiety responsible for its covalent binding in various test systems. Studies were conducted with an 125l-labeled analog of mitotane, 1-(2-chlorophenyl)-1-(4-iodophenyl)-2,2-dichloroethane, prior to a comparison to results with lower specific activity [14C]mitotane. With dog adrenal cortical whole homogenates, the majority of covalent binding was to proteins with an additional one-sixth of the total bound radioactivity associated with a phospholipid fraction. No radioactivity was associated with DNA. The rank order of species in regard to metabolism and protein binding was bovine > dog > rat adrenal homogenates > human normal adrenal or tumor homogenates. The percentage of radioactivity recovered from the hydrolysates of those fractions was uniformly high. In addition, the only metabolite present in the hydrolysates corresponded to 1-(2-chlorophenyl)-1-(4-iodophenyl)acetic acid from the iodo analog of o,p'-DDD and the corresponding o,p'-dichlorodiphenylacetic acid (o,p'-DDA) from o,p'-DDD. Our results are consistent with an acyl chloride being the reactive intermediate formed from the dichloromethyl moiety of mitotane, which leads to both DDA metabolite formation and binding to adrenal cortical bionucleophiles.

Substituent effects on the genotoxicity of 4-nitrostilbene derivatives.

4-Nitrostilbene and twelve of its derivatives (eleven E-stilbenes and two Z-stilbenes) were examined for possible quantitative structure-activity relationships of their in vitro and in vivo genotoxicity. Relative mutagenicity was studied with and without S9 activation in Salmonella strains TA98 and TA100, as well as in the nitroreductase deficient strains TA98/NR and TA100/NR. Chromosomal aberrations in the bone-marrow cells of mice following intraperitoneal administration of the nitrostilbenes were observed as an indicator of in vivo genotoxicity. All of the compounds were active in TA98 and TA100 without S9 activation, with the exception of 4-amino-4'-nitrostilbene in TA100. Mutagenic activity was greatly reduced or eliminated in the NR strains, which is consistent with metabolic activation of the compounds by bacterial reductase. The presence of S9 lowered the activity of most of the nitrostilbenes presumedly by enzymatic detoxication. Hammet values of substituents, partition coefficients and frontier orbital energies (ELUMO and EHOMO) were studied for correlations with mutagenicity of the eleven E-stilbenes. Correlations could be established between mutagenicity in TA98 without S9 activation and the Hammet values. The same mutagenicity could also be correlated to ELUMO. Rationales for these correlations include the concept that electron-withdrawing groups which lower ELUMO should facilitate the reduction of the nitro group, leading to the proximate mutagen hydroxylamine. The correlations are also explained by the concept that electron-withdrawing groups should help stabilize the hydroxylamine intermediate and make the ultimate mutagenic species, the nitrenium ions, more reactive toward DNA. The relationship between mutagenicity and electronic effects of substituent groups found in vitro could not be extended to the in vivo results. However, except for the dinitrostilbenes, where insolubility prevented their testing, all the nitrostilbenes produced a statistically significant increase in chromosomal aberrations compared to the negative solvent control.

Substituent effects on the in vitro and in vivo genotoxicity of 4-aminobiphenyl and 4-aminostilbene derivatives.

4-Amino-4'-substituted biphenyls and 4-aminostilbenes substituted in the 3' or 4' position were studied for their in vitro and in vivo genotoxicity. The in vitro mutagenicity of the biphenyls with and without S9 activation was established with Salmonella strains TA98 and TA100 and that of the stilbenes with the same strains plus TA98/1,8-DNP6. The in vivo genotoxicity assay with both series of compounds was for chromosomal aberrations in the bone-marrow cells of mice following intraperitoneal administration of the chemicals. Hammett values of substituents, partition coefficients and frontier orbital energies (ELUMO and EHOMO) of the compounds were used for correlations with mutagenicity. The Salmonella mutagenicity in TA98 and TA98/1,8-DNP6 with S9 was correlated to Hammett sigma + values for the 4-aminostilbene substituents, showing a strong trend of increasing mutagenicity with an increase in the electron-withdrawing capability of the substituent. Hydrophobicity of the stilbenes, however, had little effect on their relative mutagenicity. The 4-aminobiphenyls showed a correlation between their mutagenicity and Hammett sigma + values of their 4'-substituents in stain TA98 with S9, although the trend was not as strong as for the stilbenes. But unlike the stilbenes, TA98 mutagenicity of the biphenyls could also be correlated to hydrophobicity, and structure-activity correlations for the biphenyls was substantially improved when both sigma + and hydrophobicity data were included. For strain TA100 with S9, little correlation was found between mutagenicity of the stilbenes and any of the parameters. However, a limited correlation did exist between the mutagenicity of the biphenyls and their hydrophobicity. There was also limited correlations of the mutagenicity for the stilbenes in TA98 and TA98/1,8-DNP6 with S9 to ELUMO or EHOMO. The in vivo genotoxicity results for the biphenyls and stilbenes could not be correlated to electronic effects as for the in vitro results, nor could they be explained by hydrophobicity. However, it is interesting to note that 3'-substituted 4-aminostilbenes were all substantially more genotoxic in vivo than their corresponding 4'-substituted counterparts. The most genotoxic compound in vivo in either series was 4-aminostilbene which would not have been predicted from the in vitro results.

Chromosomal aberrations in mouse lymphocytes exposed in vitro and in vivo to benzidine and 5 related aromatic amines.

Mouse lymphocytes were exposed in vitro for 2 h or in vivo for 24 h to benzidine and related aromatic amines to test for chromosome aberrations (CA) and mitotic indices. Uninduced mouse S9 was used to activate the amines for the in vitro tests to be consistent with the in vivo tests. Contrary to a previous report, no difference could be established in the genotoxicity of benzidine following activation with uninduced S9 compared to induced S9. There were concentration related increases in CA for benzidine and all the amines in vitro except for 4,4'-diaminostilbene which exhibited the greatest cellular toxicity towards cultured lymphocytes. Benzidine and its derivatives showed significant increases in CA in vivo compared to its negative control. The CA values for 4-aminostilbene were significantly higher than the other amines in both in vivo and in vitro studies. These genotoxicity results for 4-aminostilbene are consistent with our previous report of the pronounced CA effects in murine bone-marrow cells but would not be predicted from Salmonella mutagenicity tests.

ortho-Substituent effects on the in vitro and in vivo genotoxicity of benzidine derivatives.

Benzidine and its 3,3'-diamino, 3,3'-dimethyl, 3,3'-dimethoxy, 3,3'-difluoro, 3,3'-dichloro, 3,3'-dibromo, 3,3'-dicarbomethoxy and 3,3'-dinitro derivatives together with 2-nitrobenzidine and 3-nitrobenzidine were compared for their in vitro and in vivo genotoxicity. Relative mutagenicity was established with Salmonella strains TA98, TA98/1,8-DNP6 and TA100 with and without S9 activation. All the derivatives in the presence of S9 were more mutagenic than benzidine with 3,3'-dinitro- and 3-nitro-benzidine having the greatest mutagenicity. Mutagenicity in all 3 strains with S9 activation could be correlated to electron-withdrawing ability of substituent groups, as measured by the basicity of the amines. This correlation was explained on the basis that electron-withdrawing groups could favor the stability of the mutagenic intermediate N-hydroxylamine and also enhance the reactivity of the ultimate mutagenic species, the nitrenium ion. Mutagenicity was also correlated to the energy of the lowest unoccupied molecular orbitals (ELUMO). Hydrophobicity was found to have very limited effect on the relative mutagenicity of our benzidine derivatives. The in vivo endpoint was chromosomal aberrations in the bone-marrow cells of mice following intraperitoneal administration of benzidine and its derivatives. In contrast to the in vitro results, while all the amines were genotoxic in vivo, only the 3-nitro derivative had a significant increase in toxicity over benzidine.

Quantitative structure-activity relationships for the mutagenicity of propylene oxides with Salmonella.

A quantitative structure-activity relationship approach was used to investigate the mutagenicity of a series of seventeen-monosubstituted propylene oxides in Salmonella typhimurium strains TA100 and TA1535. Mutagenicity in strain TA100, using a liquid suspension assay, was found to correlate with chemical reactivity, as measured by the rates of reaction with two model bionucleophiles, nicotinamide and 4-(4-nitrobenzyl)pyridine. However, since the reactivity of three of the epoxides did not correlate to their Taft sigma * values, as a measure of the electronic effects of substituent groups, neither was their mutagenicity predicted by this substituent constant. The relative mutagenicity for the propylene oxides was different in the liquid suspension assay than that determined by the standard plate incorporation assay and also differed between the two bacterial strains. The assay differences were attributed to epoxide stability. The differences between strains was observed to be due to the response of the error-prone repair system, found only in TA100, to the stronger alkylating agents.

Chromosomal aberrations in mouse lymphocytes exposed in vivo and in vitro to aliphatic epoxides.

Mouse lymphocytes in vivo or in vitro were exposed for 24 h to 4 aliphatic epoxides, glycidyl 1-naphthyl ether, glycidyl 4-nitrophenyl ether, 1-naphthyl-propylene oxide and trichloropropylene oxide (TCPO), and tested for the induction of chromosomal aberrations (CA). These epoxides were among the most genotoxic aliphatic epoxides in our previous studies. With the exception of TCPO, the test epoxides caused significant increases in CA in vivo compared to a negative control. There were concentration related increases in CA for all 4 epoxides in vitro and TCPO produced the greatest cellular toxicity and genotoxic effects towards cultured lymphocytes. The difference in the order of genotoxicity for the two test systems can be explained on the basis of a much shorter half-life for TCPO than for the other epoxides.

Comparison of the adrenalytic activity of mitotane and a methylated homolog on normal adrenal cortex and adrenal cortical carcinoma.

Mitotane is an important adrenalytic drug for the treatment of adrenal cancer whose use is limited by toxicity. Reports from another laboratory indicated that a methylated homolog of Mitotane (Mitometh) tested in guinea pigs possessed comparable adrenalytic activity but was less toxic than Mitotane. This observation prompted us to undertake a comparative study of these two drugs on the basis that Mitometh may be a superior agent for the treatment of adrenal cancer. Preliminary studies in guinea pigs failed to show a significant adrenalytic effect for either Mitotane or Mitometh. Thus, we extended the study to 13 mongrel dogs weighing 12-15 kg that were treated daily with Mitometh or Mitotane (50-100 mg/kg) for 6 or 12 days. Cortisol decreased to undetectable levels and adrenocorticotropic hormone (ACTH) rose to 10 times the baseline levels within 72 h in Mitotane-treated animals. Despite the achievement of similar drug levels, Mitometh treatment in dogs failed to suppress cortisol or increase ACTH. To determine whether these differences were due to differences in bioavailability, we measured the relative concentration of Mitotane and Mitometh in homogenates of adrenal cortex obtained from Mitotane- and Mitometh-treated dogs. The adrenal concentration of Mitometh determined in Mitometh-treated dogs was 5 times higher than the concentration of Mitotane measured in Mitotane-treated animals. Whereas the adrenal glands of Mitotane-treated dogs showed hemorrhage and necrosis, the Mitometh-treated animals showed no adrenal damage. Despite the lack of adrenalytic activity, Mitometh maintained its toxicity as demonstrated by microscopic evidence of hepatic necrosis and an increase in hepatic enzymes. The adrenalytic effects of both agents was also studied in vitro using a human functioning adrenal cortical carcinoma cell line, NCI-H295. Whereas Mitotane strongly suppressed cell growth, Mitometh had a weaker effect. We conclude that Mitometh is not likely to be effective in the therapy of adrenal cancer. Moreover, the results of this study are supportive of the view that metabolic transformation of Mitotane is in some way linked to its adrenalytic action.

Enantioselective detoxication of optical isomers of glycidyl ethers.

The detoxication of the enantiomers of glycidyl 4-nitrophenyl ether (GNPE), (-)-(R)- and (+)-(S)-GNPE, and glycidyl 1-naphthyl ether (GNE), (-)-(R)- and (+)-(S)-GNE, by rat liver glutathione transferase and epoxide hydrolase was studied. Enantioselectivity was observed with both enzymes favoring the (R)-isomers as determined by the formation of conjugate, diol, and remaining substrate measured by HPLC. Enantiomers of GNE were detoxified by cytosolic epoxide hydrolase but those of GNPE were not. Substantial nonenzymatically formed conjugates of enantiomers of GNPE were detected showing (S)-GNPE the more reactive of the pair.

Molecular analysis of hypoxanthine phosphoribosyl transferase mutants induced by glycidyl 1-naphthyl ether in mouse spleen cells in vivo.

Treatment of C57BL/6J mice with an epoxide, glycidyl 1-naphthyl ether (GNE), resulted in an average of a 3.4-fold increase in frequency of 6-thioguanine-resistant mutants of mouse spleen T-lymphocytes. In similar experiments with the epoxide trichloropropylene oxide, no increase in mutant frequency was found. To determine the kind and location of mutations in the coding region of the hypoxanthine phosphoribosyl transferase (HPRT) gene, 26 GNE-induced mutants and 17 spontaneous mutants were analyzed by direct sequencing of polymerase chain reaction amplified cDNA. Among the GNE-induced mutants, HPRT cDNA was present in 22, while that from 4 could not be detected. Among the spontaneous mutants, HPRT cDNA was present in 15 and absent in 2. Among GNE-induced mutants, base substitution in HPRT occurred in 15 of 22 mutants analyzed. Nine of 15 base substitutions involved TA base pairs, primarily TA-->CG transitions. Base substitutions were found throughout exons 3-7 but 46% of substitutions were located in exon 3 and one frameshift mutation involving a GC base pair in exon 3 was also observed. Among the spontaneous mutants, base substitutions of HPRT occurred in 7 of 15 mutants analyzed with 6 of 7 base substitutions involving a TA base pair and another 2 of the 15 mutants showed a 4 base pair deletion. The base substitution spectrum in GNE-induced mutants was different from that of the spontaneous mutants.

Base-pair mutations caused by six aliphatic epoxides in Salmonella typhimurium TA100, TA104, TA4001, and TA4006.

Salmonella typhimurium strains TA100, TA104, TA4001, and TA4006 were used to detect the base-pair mutations caused by six aliphatic epoxides: chloropropylene oxide, glycidyl 1-naphthyl ether, glycidyl 4-nitrophenyl ether, 1-naphthyl-propylene oxide, styrene oxide, and trichloropropylene oxide. Dose-mutagenicity relationships could be established for all six epoxides in strains TA100 and TA104 but not in strains TA4001 and TA4006. These results, together with the lack of sensitivity of the TA100 revertants to DL-1,2,4-triazole-3-alanine, indicate CG-->TA transitions and/or CG-->AT transversions are of major importance for mutations induced by these epoxides in Salmonella TA100 and possibly TA104. In addition, since the reproducibility of the effect of the triazole on TA104 reversions was poor, TA-->AT transversions were not eliminated as also contributing to the mutagenicity of these epoxides in this Salmonella strain.

The genotoxicity of enantiomeric aliphatic epoxides.

The (R)- and (S)-optical isomers of 9 epoxides, benzyloxymethyloxirane, epichlorohydrin, glycidol, glycidyl 3-nitrobenzenesulfonate, glycidyl 4-nitrobenzoate, glycidyl tosylate, styrene oxide, glycidyl 1-naphthyl ether and glycidyl 4-nitrophenyl ether, have been compared for their in vivo and in vitro genotoxicity. The in vitro short-term test employed was the Ames mutagenicity assay with Salmonella strain TA100. The in vivo tests were chromosomal aberrations (CA) as well as sister-chromatid exchange (SCE) in bone-marrow cells of mice following intraperitoneal administration of these epoxides. Differences in mutagenicity between isomers were established with TA100 for all the compounds. While 13 of the isomers were genotoxic compared to a negative control by CA measurements, only in the case of glycidyl 4-nitrobenzoate could a significant difference be found between isomers by this test. However, with SCE evaluations, differences were detected between the (R)- and (S)-isomers for all the pairs of compounds with the exception of those for benzyloxymethyloxirane and glycidyl 4-nitrophenyl ether. At least in part, differences in the patterns of genotoxicity among compounds can be related to their differences in reaction pathways.

The in vivo and in vitro genotoxicity of aromatic amines in relationship to the genotoxicity of benzidine.

Benzidine and 12 related aromatic amines have been studied for the effects of substituent groups and pi orbital conjugation on their genotoxicity as measured by their mutagenicity in vitro with Salmonella and by chromosomal aberrations (CA) in vivo in the bone-marrow cells of mice. The in vitro studies indicated increases in mutagenicity with increases in the electron withdrawing ability of para' substituents. Mutagenicity also increases with increased conjugation as shown by the degree of planarity of the biphenyl compounds and by comparing the mutagenicities of biphenyl amines to stilbenes as well as to ethylene bridged diphenyl compounds. The relative in vitro mutagenicity results were not predictive of relative in vivo CA results. The 3 most genotoxic compounds in vivo were the conjugated amines without substituents in the para' position. The CA values for 4-aminostilbene were exceptionally high. These in vivo results indicate increased genotoxicity for benzidine analogs without substitution in the para' position.

Genotoxicity of chryseno[4,5-bcd]thiophene and its sulfone derivative.

Our recent syntheses of chryseno[4,5-bcd]thiophene together with its potential sulfone metabolite, chryseno[4,5-bcd]thiophene-4,4-dioxide, have made these compounds available for genotoxicity testing. Such toxicity testing is of interest as this thiophene is an isoster of the established carcinogen benzo[a]pyrene and is one of the thiaarenes which are potential environmental contaminants found in fossil fuels. Although the thiophene was less mutagenic than benzo[a]pyrene in Salmonella strains TA98 and TA100 after S9 activation, it exhibited in vivo chromosomal aberration activity equal to that of benzo[a]pyrene in the bone-marrow cells of mice. A reduced activity with Salmonella as well as in the bone-marrow cell assay for the sulfone does not support its role as the key active metabolic intermediate for the genotoxicity of the thiophene. Our molecular orbital calculations would be consistent with the concept of activation through a diol-epoxide mechanism and offers an explanation for the reduced genotoxicity of the sulfone via this mechanism. These genotoxicity studies support the concern that sulfur isosters of established carcinogenic polycyclic aromatic hydrocarbons could themselves be toxic.

Comparison of the isolation of adducts of 2'-deoxycytidine and 2'-deoxyguanosine with phenylglycidyl ether by high-performance liquid chromatography on a reversed-phase column and a polystyrene-divinylbenzene column.

2'-Deoxycitidine (dCyd) and 2'-deoxyguanosine (dGuo) were subjected to reaction with phenylglycidyl ether (PGE) in methanol in order to study the formation of the corresponding 2'-deoxynucleoside adducts. Separation methods were developed on analytical and semi-preparative scales using high-performance liquid chromatography with photodiode-array detection on a reversed-phase column and on a polystyrene-divinylbenzene column. The use of the latter column was prompted by decomposition of the preparatively isolated dGuo-PGE adducts on the reversed-phase column. The use of a polystyrene-divinylbenzene column solved this problem and also revealed the presence of one more peak in both the dCyd- and dGuo-PGE reaction mixtures. The adducts of dCyd and dGuo were isolated on preparative reversed-phase and polystyrene-divinylbenzene columns and characterized by UV, fast atom bombardment mass and 360 MHz 1H NMR spectrometry. The adducts of dCyd were the diastereomers of N-3-(2-hydroxy-3-phenoxypropyl)-2'-deoxycytidine and N4-(2-hydroxy-3-phenoxypropyl)-2'-deoxycytidine whereas those of dGuo were the two diastereomers of N-7-(2-hydroxy-3-phenoxypropyl)-2'-deoxyguanosine and a third peak which appeared to be mainly N2-(2-hydroxy-3-phenoxypropyl)-2'-deoxyguanosine.

Mutagenicity in Salmonella and sister chromatid exchange in mice for 1,4-, 1,3-, 2,4-, and 3,4-dimethylphenanthrenes.

The mutagenicity in Salmonella and in vivo sister chromatid exchange in the bone-marrow cells of mice was determined for 1,4-, 1,3-, 2,4-, and 3,4-dimethylphenanthrene (DMPh) with the objective to study the relative importance of substitution at the 1 and 4 positions of this series of methylated phenanthrenes. For both tests, 1,4- DMPh was decidedly more genotoxic than the remaining regioisomers. While the well recognized role of steric crowding in the bay region is a factor in this enhanced genotoxicity, equally important is substitution at the 1 position with its potential to inhibit detoxication through 9,10-diol formation.

DNA strand breaks in liver for four aliphatic epoxides in mice.

Four aliphatic epoxides, 1-naphthyl glycidyl ether (NGE), 1-naphthylpropylene oxide (NPO), 4-nitrophenyl glycidyl ether (NPGE), 3,3,3-trichloropropylene oxide (TCPO) and two of their precursors, 1-allylnaphthalene (AN) and 3,3,3-trichloropropylene (TCP), were selected for DNA strand-break analysis in liver in vivo with mice. The four epoxides selected were among the most mutagenic aliphatic epoxides in our previous structure-mutagenicity studies with the Ames test and had been evaluated for their in vivo genotoxicity as measured by sister-chromatid exchange (SCE) and chromosome aberrations (CA). A significant increase in the percentage of unwound DNA was observed at a 4-h exposure time for all the compounds at high doses except for AN. TCPO, the least genotoxic compound in bone marrow, had the greatest liver toxicity after 1-h exposure while NGE showed the most toxicity after 6 h. As might be expected from their corresponding epoxides, AN but not TCP exhibited significant SCE activity in the bone marrow of mice. This study reemphasizes the importance of evaluating the stability of direct-acting alkylating agents in comparing test results and in establishing the relative order of genotoxicity for such compounds.