Molecular modeling of enzymes and receptors involved in carcinogenesis: QSARs and compact-3D

A Comparison between COMPACT and Hazardexpert Evaluations for 80 Chemicals Tested by the NTP/NCI Rodent Bioassay

A total of 80 structurally diverse chemicals for which rodent carcinogenicity data are known, have been evaluated by two predictive computer-based systems, COMPACT and Hazardexpert. The results demonstrate that these two predictive test procedures are, to some extent, complementary, in that COMPACT can identify carcinogens which require metabolic activation, whereas Hazardexpert is able to provide an alert for potentially direct-acting agents. In this study, COMPACT was able to identify correctly most rodent carcinogens, and the Hazardexpert system showed a high concordance with non-carcinogens, in particular. It is suggested that some form of combined approach to the use of predictive methods in chemical safety evaluation may represent a way forward for the future.

Development and Uses of Offline and Web-Searchable Metabolism Databases - The Case of Benzo[a]pyrene

BACKGROUND

The present work describes development of offline and web-searchable metabolism databases for drugs, other chemicals, and physiological compounds using human and model species, prompted by the large amount of data published after year 1990. The intent was to provide a rapid and accurate approach to published data to be applied both in science and to assist therapy.

METHODS

Searches for the data were done using the Pub Med database, accessing the Medline database of references and abstracts. In addition, data presented at scientific conferences (e.g., ISSX conferences) are included covering the publishing period beginning with the year 1976.

RESULTS

Application of the data is illustrated by the properties of benzo[a]pyrene (B[a]P) and its metabolites. Analysis show higher activity of P450 1A1 for activation of the (-)- isomer of trans-B[a]P-7,8-diol, while P4501B1 exerts higher activity for the (+)- isomer. P450 1A2 showed equally low activity in the metabolic activation of both isomers.

CONCLUSION

The information collected in the databases is applicable in prediction of metabolic drug-drug and/or drug-chemical interactions in clinical and environmental studies. The data on the metabolism of searched compound (exemplified by benzo[a]pyrene and its metabolites) also indicate toxicological properties of the products of specific reactions. The offline and web-searchable databases had wide range of applications (e.g. computer assisted drug design and development, optimization of clinical therapy, toxicological applications) and adjustment in everyday life styles.

Significant increase in phenacetin oxidation on L382V substitution in human cytochrome P450 1A2

Human CYP1A2 is an important drug-metabolizing enzyme, similar in sequence to CYP1A1 but with distinct substrate specificity. We have previously shown that residue 382 affected CYP1A1 and CYP1A2 specificities with alkoxyresorufins. To determine whether this residue is also important for the metabolism of other substrates, we have investigated phenacetin oxidation by single (T124S, T223N, V227G, N312L, and L382V) and multiple (L382V/T223N, L382V/N312L, L382V/T223N/N312L, and L382V/T124S/N312L) mutants of CYP1A2. The enzymes were expressed in Escherichia coli and purified. All the CYP1A2 mutants that contained the L382V substitution displayed much higher activities than the wild-type enzyme, with k(cat) values 3-fold higher, in contrast to other mutants, for which k(cat) decreased. Likewise, a significant increase in specificity, expressed as the k(cat)/K(m) ratio, was observed for the mutants containing the L382V substitution. The efficiency of coupling of reducing equivalents to acetaminophen formation was decreased for all the single mutants except L382V, for which the coupling increased. This effect was also observed with multiple CYP1A2 mutants containing the L382V substitution. Low activities of the four other single mutants were likely caused by dramatically increased uncoupling to water. In contrast, the increase in activity of the L382V-containing mutants resulted from decreased water formation. This finding is consistent with molecular dynamics results, which showed decreased phenacetin mobility leading to increased product formation. The results of these studies confirm the importance of residue 382 in CYP1A2-catalyzed oxidations and show that a single residue substitution can dramatically affect enzymatic activity.

Functional interaction of nitrogenous organic bases with cytochrome P450: A critical assessment and update of substrate features and predicted key active-site elements steering the access, binding, and orientation of amines

The widespread use of nitrogenous organic bases as environmental chemicals, food additives, and clinically important drugs necessitates precise knowledge about the molecular principles governing biotransformation of this category of substrates. In this regard, analysis of the topological background of complex formation between amines and P450s, acting as major catalysts in C- and N-oxidative attack, is of paramount importance. Thus, progress in collaborative investigations, combining physico-chemical techniques with chemical-modification as well as genetic engineering experiments, enables substantiation of hypothetical work resulting from the design of pharmacophores or homology modelling of P450s. Based on a general, CYP2D6-related construct, the majority of prospective amine-docking residues was found to cluster near the distal heme face in the six known SRSs, made up by the highly variant helices B', F and G as well as the N-terminal portion of helix C and certain beta-structures. Most of the contact sites examined show a frequency of conservation < 20%, hinting at the requirement of some degree of conformational versatility, while a limited number of amino acids exhibiting a higher level of conservation reside close to the heme core. Some key determinants may have a dual role in amine binding and/or maintenance of protein integrity. Importantly, a series of non-SRS elements are likely to be operative via long-range effects. While hydrophobic mechanisms appear to dominate orientation of the nitrogenous compounds toward the iron-oxene species, polar residues seem to foster binding events through H-bonding or salt-bridge formation. Careful uncovering of structure-function relationships in amine-enzyme association together with recently developed unsupervised machine learning approaches will be helpful in both tailoring of novel amine-type drugs and early elimination of potentially toxic or mutagenic candidates. Also, chimeragenesis might serve in the construction of more efficient P450s for activation of amine drugs and/or bioremediation.

Quantitative structure-activity relationships (QSARs) within the cytochrome P450 system: QSARs describing substrate binding, inhibition and induction of P450s

Quantitative structure-activity relationships (QSARs) within substrates, inducers and inhibitors of cytochromes P450 involved in xenobiotic metabolism are reported, together with QSARs associated with induction, inhibition and metabolic rate. The importance of frontier orbitals and shape descriptors, such as planarity (estimated by the area/depth(2) parameter) and rectangularity (estimated by the length/width parameter) is discussed, particularly in the context of the COMPACT system which discriminates between several P450 families associated with the activation and detoxication of xenobiotics. The use of parameters, particularly those derived from homology modelling of mammalian (especially human) P450s that are involved in exogenous metabolism, in generating QSARs for P450 substrates is discussed in the context of explaining differences in the binding affinities of human P450 substrates which are pharmacologically active.

Discovery of cytochrome P450 1B1 inhibitors as new promising anti-cancer agents

Human cytochrome P450 (CYP)1B1 is a major enzyme for carcinogenic estrogen metabolism and involved in the metabolic activation of procarcinogens of the polycyclic aromatic hydrocarbons (PAHs). CYP1B1 is known to be expressed at a high frequency in various human cancers, but not in normal tissues. It also plays an important role in the metabolism of various anti-cancer drugs. These findings suggest inhibition of CYP1B1 as a new oncological therapeutic strategy. Several natural and synthetic compounds have been studied in an effort to find the isoform-specific inhibitors of the CYP1 subfamily. A survey of the inhibitors of CYP1B1 and other related inhibitors of the CYP1 subfamily is provided in this review.

P450 BM3: the very model of a modern flavocytochrome

Flavocytochrome P450 BM3 is a bacterial P450 system in which a fatty acid hydroxylase P450 is fused to a mammalian-like diflavin NADPH-P450 reductase in a single polypeptide. The enzyme is soluble (unlike mammalian P450 redox systems) and its fusion arrangement affords it the highest catalytic activity of any P450 mono-oxygenase. This article discusses the fundamental properties of P450 BM3 and how progress with this model P450 has affected our comprehension of P450 systems in general.

Human carcinogens: an evaluation study via the COMPACT and HazardExpert procedures

The results of computer-optimized molecular parametric analysis of chemical toxicity (COMPACT) and HazardExpert evaluations on 14 established human carcinogens are reported. The concordances between COMPACT and carcinogenicity (71%) and between HazardExpert and carcinogenicity (57%) are significantly improved when taken in combination, where all 14 carcinogens are correctly identified by the two systems used in conjunction. However, if a negative energy of the highest occupied molecular orbital (E(HOMO)) value is regarded as evidence of electrophilic reactivity likely to give rise to mutagenicity and carcinogenicity, then 13/14 (93%) of the carcinogens are correctly identified by combination with the COMPACT procedure alone. It is possible, therefore, to establish likely carcinogenicity arising from either P450 mediation (CYP1 and CYP2E) or compound electrophilicity via the employment of a straightforward approach to molecular and electronic structure calculation, a process that can be performed in a relatively short time frame (i.e., less than 1 hour per chemical) and at a low cost.

Summary of information on human CYP enzymes: human P450 metabolism data

This chapter is an update of the data on substrates, reactions, inducers, and inhibitors of human CYP enzymes published previously by Rendic and DiCarlo (1), now covering selection of the literature through 2001 in the reference section. The data are presented in a tabular form (Table 1) to provide a framework for predicting and interpreting the new P450 metabolic data. The data are formatted in an Excel format as most suitable for off-line searching and management of the Web-database. The data are presented as stated by the author(s) and in the case when several references are cited the data are presented according to the latest published information. The searchable database is available either as an Excel file (for information contact the author), or as a Web-searchable database (Human P450 Metabolism Database, www.gentest.com) enabling the readers easy and quick approach to the latest updates on human CYP metabolic reactions.

Application of process chemistry and SAR modelling to the evaluation of health findings of lower olefins

Epidemiology studies show increased leukemia mortality among styrene butadiene rubber (SBR) workers but not among butadiene monomer production employees. A detailed review of the SBR manufacturing process indicates that sodium dimethyldithiocarbamate (DMDTC) introduced into the SBR manufacturing process for a period in the 1950s coincides with increased leukemia mortality. Using the Computer-Optimized Molecular Parametric Analysis of Chemical Toxicity (COMPACT), we assessed the enzyme (cytochrome P450) substrate specificity of an olefin series including 1,3-butadiene (BD) and also modeled its interaction with DMDTC. These analyses showed correlation of a structural/electronic parameter--the COMPACT radius--with the presence or absence of cytogenetic activity and also found that DMDTC would inhibit the oxidative metabolism of BD at least at high concentrations. Both DMDTC and its diethyl analog (DEDTC) bind with CYP 2E1 and CYP 2A6. Both of these isoforms are important in the initial oxidative metabolism of butadiene and other olefins. In co-exposure studies in mice of DMDTC with BD or with epoxybutene (EB), we found that there was a reduced increase in genotoxic activity based on micronuclei induction compared with BD or EB exposure alone. Treatment with DMDTC significantly increased the protein carbonyl contents of hepatic microsomes compared with that of controls, a finding that may be related to DMDTC's activity as a prooxidant. Co-exposure with DMDTC and EB increased hepatic microsomal carbonyls to levels significantly greater than those of DMDTC-treated mice, while EB administration in the absence of DMDTC did not change protein carbonyls relative to those of controls. The increase in hepatic microsomal protein carbonyls suggests that DMDTC may modulate EB metabolism towards the formation of reactive intermediates that react with proteins. The present molecular modeling and mechanistic studies suggest that co-exposure of BD and DMDTC is a plausible biological hypothesis regarding increased leukemia risk among SBR workers.

Orientation of caffeine within the active site of human cytochrome P450 1A2 based on NMR longitudinal (T1) relaxation measurements

Longitudinal (T1) relaxation studies were performed in order to examine the interaction of caffeine with the heme of human P450 1A2. Addition of caffeine to this P450 resulted in a small, incomplete conversion of the heme from high spin to low spin, as shown by changes in the optical spectrum. Determination of a relatively large dissociation constant (Ks = 2.6 mM) as well as the relative instability of the P450 after 2 h at room temperature necessitated the performance of these experiments at high concentrations (25 mM) of caffeine. The relaxation measurements on the three sets of methyl hydrogens led to the determination of the corresponding distances between the iron and the methyl groups on the bound caffeine as well as the position and orientation of caffeine within the active site of P450 1A2. The three methyl groups were found to be nearly equidistant from the iron (> or = 4.79-4.89 A), with slight preference for the N-3 position, and thus, the average position of caffeine was parallel to the heme. In vitro incubations with P450 1A2 and 5 mM caffeine led primarily to paraxanthine formation (N-3 demethylation), as expected. However, with 25 mM substrate, the overall extent of oxidation was doubled and there was more equivalent oxidation at each of the four potential sites on caffeine. This latter observation was consistent with the lack of selective positioning of the N-3 methyl group of caffeine relative to the heme.

Cytochrome P450 substrate specificities, substrate structural templates and enzyme active site geometries

The structural characteristics of human cytochrome P450 substrates are outlined in the light of extensive studies on P450 substrate specificity. Templates of superimposed substrates for individual P450 isozymes are shown to fit the corresponding enzyme active sites, where contacts with specific amino acid residues appear to be involved in the interaction with each structural template. Procedures leading to the evaluation of likely P450 specificity, binding affinity and rate of metabolism are described in the context of key examples in which molecular modelling appears to rationalize experimentally observed findings.

Structural determinants of cytochrome P450 substrate specificity, binding affinity and catalytic rate

The structural characteristics of cytochrome P450 substrates are summarised, showing that molecular descriptors can discriminate between chemicals of differing P450 isozyme specificity. Procedures for the estimation of P450 substrate binding interaction energies and rates of metabolism are described, providing specific examples in both individual compounds binding to P450s, including those of known crystal structure, and within series of structurally related chemicals. It is demonstrated that binding energy components are primarily hydrophobic/desolvation and electrostatic/hydrogen-bonded in nature, whereas electronic factors are of importance in determining variations in reaction rates. It is thus shown that the prediction of P450 substrate binding affinities and catalytic rates may be feasible, provided that sufficient structural information is available for the relevant enzyme-substrate complex.

A combined COMPACT and HazardExpert study of 40 chemicals for which information on mutagenicity and carcinogenicity is known, including the results of human epidemiological studies

The COMPACT approach for defining structural criteria for substrates and inducers of cytochrome P450 (CYP) enzymes which mediate the formation of reactive intermediates is discussed in the context of prediction of potential carcinogenicity. This is broadened to encompass structural studies on mammalian P450s, including those relevant to genetic polymorphism in man. The use of the COMPACT system, in parallel with the structure alert program HazardExpert (now incorporated into the Pallas system), for evaluating human carcinogenicity data is reported, as an example of the possible employment of a battery of short-term test procedures for safety evaluation. In particular, the importance of using the log P value (as a measure of compound lipophilicity) to assess the likelihood of a potentially toxic compound reaching the site of activation, is emphasized by the finding that most procarcinogens requiring metabolic activation by P450s are lipophilic in nature.

Molecular modelling of CYP2E1 enzymes from rat, mouse and man: an explanation for species differences in butadiene metabolism and potential carcinogenicity, and rationalization of CYP2E substrate specificity

Molecular modelling of substrates of cytochrome P4502E1 (CYP2E1) within the putative active site region of CYP2E1 constructed from the CYP102 crystal structure is reported. Structural characteristics of CYP2E1 substrates, such as molecular size, energy levels and polarity, calculated via molecular orbital procedures provide correlations with toxicity and carcinogenicity; and species differences in CYP2E1-mediated metabolism are rationalized in terms of interactions with putative active site amino acid residues, including Thr-437 and Phe-181. In particular, the activation of buta-1,3-diene can be explained by active site modelling with CYP2E1 enzymes sequenced from rat, mouse and man, where there is a non-conservative change T437H between rodent and human isozymes, together with a conservative change I438V between mouse and rat CYP2E1.

Extrapolation from animals to man. The integration of pharmacokinetics and pharmacodynamics

This paper has focused on the difficulties of extrapolating toxicological or pharmacological data obtained from animals to those expected in man. For some drugs, under certain conditions, there may be no problem, but for many, this is clearly not the case. Differences in apparent activity are impossible to reconcile without "normalizing" the dose for differences in pharmacokinetics and metabolism. The increasing use of artificial intelligence and expert systems in drug investigations may provide a greater insight into why these differences may occur and allow prediction but, in the end, they must be tested in the experiments undertaken. The use of kinetic dynamic relationships in different species will certainly help in this regard and, wherever possible, should be included in experimental design to build up a database of experience since such information is sadly lacking. But we must interpret with caution the data produced by those that continue to extrapolate animal data to humans without some attempt to discuss in detail the validity of their assumptions.

Molecular modelling of CYP3A4 from an alignment with CYP102: identification of key interactions between putative active site residues and CYP3A-specific chemicals

1. A structural model of CYP3A4 is reported on the basis of a novel amino acid sequence alignment between the CYP3 family and CYP102, a bacterial P450 of known crystal structure. 2. Construction of the CYP3A4 model from CYP102 is facilitated by the relatively high sequence homology between the two protein (52% homology; 27% identity) with many conservative amino acid changes, yielding a structure of low internal energy. 3. A considerable number of specific substrates, and some specific inhibitors, are shown to occupy the putative CYP3A4 active site via interactions with the same amino acid residues in almost all cases investigated. 4. The CYP3A4 model rationalizes the known positions of metabolism for many substrates of this major human P450 such that the route of metabolism in novel development compounds can be predicted.

FRAME Recommendations for the Application of the Three Rs to the Regulatory Toxicity Testing of Food Additives

Direct food additives are tested for genotoxicity, acute and subchronic toxicity, carcinogenicity and teratogenicity. International guidelines differ in the types of tests required, the duration of the tests, the species of animals to be used, the number of animals recommended and the method of housing experimental animals. This lack of harmonisation is wasteful in terms of animal use and creates additional and, perhaps, unnecessary work for the food industry. In addition, unlike other chemicals, food additives pose a special problem for toxicity testing due to repeated low-dose, life-time human exposure, which is difficult to model in animal studies. In an assessment of the extent to which the Three Rs (reduction, refinement and replacement) can be applied to food additive toxicity testing, it was concluded that differences in regulatory requirements and testing protocols can be improved in both the short term and longer term. Suggestions for improvements to existing alternative approaches for food toxicity testing are made.

Molecular modelling of CYP1A subfamily members based on an alignment with CYP102: rationalization of CYP1A substrate specificity in terms of active site amino acid residues

1. Using a novel amino acid sequence alignment, proteins of the CYP1A subfamily have been produced from the CYP102 crystal structure template via residue replacement and energy minimization procedures. 2. Known substrates and inhibitors of CYP1A1 and CYP1A2 are shown to fit their respective active sites via key interactions with complementary amino acid residues. Substrates used in the modelling studies include: caffeine, PhIP, oestradiol, 2,4- and 2,5-diaminotoluenes, Glu-P-1, phenacetin, acetanilide, 7-methoxy and 7-ethoxyresorufins, 11-methyl cyclopenta[a]phenanthren-17-one, 7-ethoxycoumarin, aflatoxin B1, benzo[a]pyrene, benzo[a]pyrene-7,8-diol and 1'-hydroxy 3-methylcholanthrene. 3. A number of aspects relating to CYP1A substrate specificity and metabolism can be explained in terms of the enzyme models, as it is found that key interactions with active site amino acid residues direct CYP1A-mediated metabolism in the known positions.

An approach to QSAR of 16-substituted pregnenolones as microsomal enzyme inducers

In this study, we attempt to correlate quantitatively the structure of eight 16-substituted pregnenolones with microsomal enzyme inducing activity. We also performed some electrostatic potential calculations to get further insight into the properties of these substituents. It was found that pregnenolone-16 alpha-carbonitrile is the most active steroidal inducer among the pregnenolone derivatives tested. The receptor-inducer interaction is facilitated by a favourable electronic effect of the 16 alpha-substituents. The orientation of the electronegative area at position 16 seems to influence activity. Lipophilic and volume effects of the 16 alpha-substituents do not seem to be important for microsomal enzyme induction. However, substituent length has some influence on drug metabolising enzyme activity, probably interfering with receptor-inducer interactions.

Molecular orbital-generated QSARs in a homologous series of alkoxyresorufins and studies of their interactive docking with P450s

1. Molecular and electronic structural parameters have been determined, by molecular orbital (MO) calculations, for a homologous series of 8 alkoxyresorufins (methoxy- to octoxy-). 2. Quantitative structure-activity relationships (QSARs) between these structural parameters and the rates of metabolism of the alkoxyresorufins in hepatic microsomes from the 3-methylcholanthrene (MC)-, and phenobarbital (PB)-pretreated mouse, and the beta-naphthoflavone (beta NF)-pretreated rat have been established. 3. The most significant single relationship is between beta NF-induction of cytochrome P4501 (CYP1A) and the total nucleophilic superdelocalizability (sigma SN) for the eight compounds in the series. 4. For double regressions, the electronic charge on the alkoxy oxygen, Q(O), or alpha-carbon Q(C), is important when combined with the hydrophobic substituent constant (pi). 5. These findings indicate that the rates of metabolism of these alkoxyresorufins are dependent upon their ability to cross cellular membranes, to fit the relevant CYP1A binding site, and on their ability to accept electrons from a donor nucleophilic species. 6. A different set of parameters correlated with CYP2B activity, namely, parameters of overall shape, which indicates that the way in which the alkoxyresorufins fit the CYP2B site, determines their differences in specificity. 7. Computer graphic interactive docking studies of the alkoxyresorufins with their affinity-specific cytochromes P450, namely, methoxy- with CYP1A2; ethoxy- with CYP1A1; pentoxy- with CYP2B1; and benzyloxy- with CYP3A, have also been undertaken to show the specific interactions of the alkoxyresorufins with the binding sites of the individual P450s.

A quantitative structure-activity relationship study on a series of 10 para-substituted toluenes binding to cytochrome P4502B4 (CYP2B4), and their hydroxylation rates

Molecular structural and molecular orbital calculations (AM1 method) are reported on a series of 10 para-substituted toluene derivatives and this structural information has been used to rationalize the differences between both rates of hydroxylation catalysed by cytochrome P4502B4 and binding to the same cytochrome P450, via the generation of quantitative structure-activity relationships (QSARs). It was found that the rate constant for hydroxylation can be described by a two-variable expression involving the dipole moment and volume of the solvent-accessible molecular surface (r = 0.98), whereas binding free energies are well characterized by combinations of molecular volume and various electronic frontier orbital parameters (r = 0.98 and 0.99). This study represents an advance on a previous evaluation by White and McCarthy (Arch Biochem Biophys 246: 19-32, 1986) who used empirical physico-chemical parameters to obtain similar results which were generally of lower statistical significance to those of the present work. The QSAR expressions suggest that both binding to P450 and metabolism for this series of compounds are dependent on the relative ability of the molecules to desolvate and occupy the heme binding site, together with electronic properties of the whole molecule and of the methyl group which undergoes hydroxylation.

Molecular modelling of members of the P4502A subfamily: application to studies of enzyme specificity

1. Using the recently published crystal structure of a bacterial P450, namely 102 (also termed P450bm3), as a template molecular models of mammalian 2A1, 2A4, 2A5 and 2A6 were constructed. 2. Substrate interaction studies demonstrated that in keeping with known catalytic activities the putative binding sites of mouse hepatic P4502A4 and 2A5 oriented testosterone for 15 alpha-hydroxylation and coumarin for 7-hydroxylation respectively. 3. Substrate interaction studies with the putative binding site of human liver P4502A6 demonstrated that coumarin was oriented for 7-hydroxylation. However, in keeping with previous site-directed mutagenesis studies with P4502A4 and 2A5, changing a single phenylalanine residue to leucine in 2A6 gave rise to a mutant enzyme, which could bind testosterone as a substrate for 15 alpha-hydroxylation rather than coumarin. 4. Substrate interaction studies with the putative binding site of rat hepatic P4502A1 suggested that this isoenzyme would hydroxylate coumarin at the 3- rather than at the 7-position. 5. The results of these molecular modelling studies demonstrate that apparently minor modifications to P4502A subfamily amino acid sequences can result in major alterations in enzyme specificity. 6. Molecular modelling is thus a useful technique that can aid in elucidating substrate specificities of P450 isoenzymes and species differences in xenobiotic metabolism. The technique can also be utilized to complement site-directed mutagenesis studies in order to identify critical structural features of P450s and other enzymes.

Regulatory Genotoxicity Testing: A Critical Appraisal

This review considers current approaches to regulatory genotoxicity testing, focusing on how the use of animals can be further replaced, reduced and refined. The complementary roles of in vitro and in vivo testing, and the justification for using animals, are discussed in detail. Recommendations are made for improvements and further work, in the light of the considerable current controversy surrounding the composition and deployment of testing strategies, and the interpretation of the data generated, particularly for carcinogenicity prediction. The major problems are the oversensitivity of in vitro tests and the insensitivity of in vivo assays. On the basis of an analysis of some published databases, it is concluded that there is insufficient support for using in vivo genotoxicity assays for screening. Also, it is questionable whether the scientific benefits of using such assays always outweigh the costs to the animals involved. The considerable efforts being made to harmonise in vivo protocols and to develop improved methods for detecting genotoxicity are discussed. It is recommended that more emphasis be placed on characterising genotoxins in vitro, especially for mechanisms of activity, to optimise the benefits of any confirmatory animal tests.. Also, regulatory agencies are urged to require better-designed and more-scientifically sound protocols, in which animal numbers are minimised and data interpretation, particularly that of negative results, is facilitated. Lastly, in the development and validation of transgenic rodent systems, emphasis should be placed on developing protocols in which other acute toxicity and metabolism endpoints can be measured simultaneously with in vivo mutagenesis, while minimising animal numbers.