Genotoxicity testing of biotechnology-derived products. Report of a GUM task force. Gesellschaft für Umweltmutationsforschung.

Various aspects of genotoxicity testing of biotechnology-derived products are discussed based on information gathered from a questionnaire which was sent to about 30 predominantly European companies. Feedback was received from 13 companies on 78 compounds, mostly recombinant proteins but also on a number of nonrecombinant proteins, which had been assessed for genotoxicity in a total of 177 tests. Four of the 78 compounds appeared to elicit reproducible genotoxic effects. For one of these compounds, the activity could be related to a nonpeptidic linker molecule. No scientifically convincing rationale for the other three compounds could be established, although, at least for two compounds, their activity may be connected with the enzymatic/hormonal activity. In addition to the survey, published reports on genotoxicity testing of biotechnology products were reviewed. The data are discussed relative to whether genotoxicity testing is a valuable exercise when assessing potentially toxic liabilities of biotechnology-derived compounds. It is concluded that genotoxicity testing is generally inappropriate and unnecessary, a position which is in accordance with the available guidelines addressing this area. For the 'average' protein, electrophilic reactions are difficult to envision. Indirect reactions via DNA metabolism and growth regulation seem possible for only very specific proteins such as nucleases, growth factors, cytokines. No information on testing of different types of biotechnology-derived products (e.g., ribozymes, antisense-oligonucleotides, DNA vaccines) has been received in the questionnaires. Discussion of their potential to cause genotoxic changes was based on literature reports. Even for those products for which concerns of genotoxic/tumourigenic potential cannot be completely ruled out, e.g., because of their interaction with DNA metabolism or proliferation control, the performance of standard genotoxicity assays generally appears to be of little value. All information, including also information on the occurrence of genotoxic impurities, has been utilized to formulate a decision tree approach for the genotoxicity testing of biotechnology-derived products.

Rapid assay of intrinsic radiosensitivity based on apoptosis in human CD4 and CD8 T-lymphocytes.

PURPOSE: An assay for radiosensitivity has numerous applications in the clinic. Avoidance of acute responses, prediction of normal tissue toxicity, and individualization of patient radiotherapy are included among these. We have developed a rapid assay (about 24 h) able to predict intrinsic radiosensitivity of CD4 and CD8 T-lymphocytes based on radiation-induced apoptosis. METHODS AND MATERIALS: Fresh blood samples (1-2 ml in heparinized tubes) were irradiated with 0-, 2-, and 8-Gy X rays at a dose rate of approximately 3 Gy/min. Following irradiation, the cells were collected and prepared for flow-cytometric analysis and cell sorting. In conjunction with the CellQuest software available with the FACSVantage cell sorter (Becton-Dickinson), two T-lymphocyte types were analyzed on the basis of their cell-specific antigens (CD4 and CD8), and DNA was stained with DAPI. Following the separation of these cell types, radiation-induced cell death was assessed. Cytotoxicity was characterized by gradual degradation of internucleosomal DNA which results in a sub-G1 peak on the DNA histogram, and by the associated loss of surface antigens causing an intermediate positive peak in the antibody histogram. Using the assay, we investigated the interdonor variation in a cohort of 45 healthy adult blood donors and 5 children [one had immunodeficiency, centromeric instability, and facial anomalies syndrome (ICF), and one had ataxia telangiectasia (AT)]. Intradonor variation was assessed with 10 different experiments from a single donor. RESULTS: CD4 and CD8 T-lymphocyte radiosensitivities were correlated (r = 0.63 and 0.65 for 2 and 8 Gy, respectively) in 45 adult donors. Both for CD4 and CD8 cells, 2 and 8 Gy irradiation responses showed a good correlation (r = 0.77 for both). Interdonor variation was significantly higher than intradonor variation (p < 0.0005) for all CD4 and CD8 data. We observed a decrease in the antigen fluorescence of dying cells, a phenomenon referred to as antigen-ebb. Antigen-ebb was clearly observed in both cell types, and correlated significantly with cytotoxicity. A trend was observed between radiosensitivity and donor age, but there was no correlation for gender. Blood from a 4-year-old girl presenting with ICF demonstrated compromised radiation-induced cytotoxicity in her CD4 T-lymphocytes, and an 11-year-old boy presenting with AT demonstrated compromised radiation-induced cytotoxicity in both his CD4 and CD8 T-lymphocytes. CONCLUSION: We conclude that the assay provides a rapid means of determining radiosensitivity, can discriminate differences in radiation-induced cytotoxicity between individuals, and can be used as a rapid screen for genetically hypersensitive patients. Antigen-ebb offers interesting possibilities for molecular biological investigations, permitting characterization and isolation of abnormal but vital cells in the absence of clastogenic agents.

The vicinal chloroalcohols 1,3-dichloro-2-propanol (DC2P), 3-chloro-1,2-propanediol (3CPD) and 2-chloro-1,3-propanediol (2CPD) are not genotoxic in vivo in the wing spot test of Drosophila melanogaster.

In this study, the vicinal chloroalcohols 1,3-dichloro-2-propanol (DC2P), 3-chloro-1,2-propanediol (3CPD) and 2-chloro-1,3-propanediol (2CPD) were investigated for genotoxicity in the wing spot test of Drosophila. DC2P is an important starting material in many processes of synthesis in chemical industry. 3CPD as well as some related glycerol chlorohydrins were identified in protein hydrolysates industrially used for the production of food items such as seasonings, sauces and soups. The wing spot test is a somatic mutation and recombination test (SMART) and is a sensitive in vivo assay for the detection of mutagens and promutagens. The test was applied here in its standard version with normal bioactivation and in a variant with increased cytochrome P450-dependent bioactivation capacity. All three compounds were clearly non-genotoxic in these in vivo assays. The results are in agreement with recent findings which strongly suggest that positive genotoxicity results in in vitro testing of vicinal chloroalcohols such as DC2P are due to directly acting genotoxic intermediates arising from a chemical reaction with the culture medium rather than from enzymatic biotransformation.

Inhibitory effects of heavy metals on cytochrome P4501A induction in permanent fish hepatoma cells.

The interactions in vitro of heavy metals Cd(II), Co(II), Cu(II), Ni(II), Pb(II), and Zn(II) with cytochrome P4501A (CYP1A) induction response and enzyme activity were studied in fish hepatoma cells PLHC-1. Cells were simultaneously exposed to heavy metals and to 3-methylcholanthrene (3-MC), an inducer of CYP1A. Heavy metals were added to the cells in different concentrations. Cytotoxicity were measured in the neutral red (NR) assay, relative CYP1A protein contents in an enzyme-linked immunosorbent assay (ELISA), and CYP1A activities in the ethoxyresorufin-O-deethylase (EROD) assay. All metals had a more pronounced effect on EROD activity than on CYP1A protein content and cytotoxicity. For the most active metal Cd(II), a 50% inhibition of EROD activity was observed at significantly lower concentrations (2.2 x 10(-5) M) than a 50% reduction of CYP1A protein (5.3 x 10(-5) M), and a 50% cytotoxicity (1.4 x 10(-4) M). The inhibitory potency of the metals had the following order: Cd(II) > Ni(II) > Cu(II) > Co(II) = Zn(II) > Pb(II). In a second set of experiments, lysates of 3-MC-induced cells were exposed to heavy metals. Cd(II) and Cu(II) caused a 50% inhibition of EROD activity at significantly lower concentrations than in the experiments with living cells, at 8.2 x 10(-6) M and 1.3 x 10(-5) M, respectively, whereas the effect by Co(II) occurred at a significantly higher concentration (8.2 x 10(-4) M). The results indicate that Cd(II) and Cu(II) in particular may affect the CYP1A system of the liver of fish at low concentrations through direct inhibition of the CYP1A enzyme activity. CYP1A induction response in fish liver is increasingly being used in biomonitoring programs. In the environment, interactions of CYP1A-inducing and CYP1A-inhibiting components (such as heavy metals) can be expected and must be taken into consideration.

Detection of genotoxic activity in native hospital waste water by the umuC test.

The genotoxic potential of the waste water of a hospital was evaluated by the umuC test. Within 2 years over 800 native waste water samples were analysed. Genotoxic activity was found in 13% of the samples. The highest genotoxic activity occurred in the morning hours, but genotoxic samples were detected also during the day and at night. 96% of the genotoxic waste water samples revealed a genotoxic potential without growth inhibition of test bacteria monitored as OD600, in the same way as antineoplastic drugs like mitomycin C or cisplatin. 4% of the genotoxic waste water samples showed combined cytotoxic and genotoxic activities as seen in control experiments using glutaraldehyde containing disinfectants and certain antibiotics.

Targeting of heterologous membrane proteins into proliferated internal membranes in Saccharomyces cerevisiae.

Overproduction of chimeric proteins containing the HMG2/1 peptide, which comprises the seven transmembrane domains of Saccharomyces cerevisiae 3-hydroxy-3-methylglutaryl-CoA reductase isozymes 1 and 2, has previously been observed to induce the proliferation of internal endoplasmic reticulum-like membranes. In order to exploit this amplified membrane surface area for the accommodation of heterologous microsomal proteins, we fused sequences coding for human cytochrome P4501A1 (CYP1A1) to sequences encoding the HMG2/1 peptide and expressed the hybrid genes in yeast. The heterologous hybrid proteins were targeted into strongly proliferated membranes, as shown by electron microscopic and immunofluorescent analysis. Fusion proteins comprising the whole CYP1A1 polypeptide (HMG2/1-CYP1A1) exhibited 7-ethoxyresorufin-O-deethylase activity, whereas fusion proteins lacking the N-terminal 56 amino acids of CYP1A1 (HMG2/1-delta CYP1A1) were inactive and appeared to be unable to incorporate protoheme. Similar amounts of heterologous protein were detected in cells expressing HMG2/1-CYP1A1, HMG2/1-delta CYP1A1 and CYP1A1, respectively. Replacement of the N-terminal membrane anchor domain of human NADPH-cytochrome P450 oxidoreductase by the HMG2/1 peptide also resulted in a functional fusion enzyme, which was able to interact with HMG2/1-CYP1A1 and the yeast endogenous P450 enzyme lanosterol-14 alpha-demethylase.

Distinct area distribution differences of micronuclei induced by clastogenic and aneuploidogenic chemicals in the bone marrow of the CD-1 mouse.

To distinguish between aneuploidogenic and clastogenic effects of test chemicals, area distributions of micronuclei (MN) in polychromatic erythrocytes (PE) from the mouse bone marrow were measured using an image analysis system. Triethylenemelamine (TEM), cytosine-beta-D-arabinofuranoside (ara-C), urethane (URT), cyclosphamide (CP), mitomycin C (MMC), colcemid (COL) and tubulazole C (TUB) were investigated for the induction of micronucleus area distributions. The area distribution of micronuclei of untreated mice was also determined. Reproducible small differences between the clastogens and the aneuploidogens were observed after measuring 1100-1200 micronuclei. A common feature of the distribution curves was a shoulder region in the same area range for all clastogens. The aneuploidogens COL and TUB showed a plateau (= wide peak) in this clastogenic shoulder region. For all clastogens, the integrated area of shoulder over a fitted function (shoulder strength) was evaluated. MMC and CP, thought to have some aneuploidogenic potential, showed an increased shoulder strength compared to TEM, ara-C and URT. The control area distribution had no similarities to the area distribution of either clastogens or aneuploidogens. In a further experiment, we attempted to correlate the size of micronuclei determined after treatment with the aneuploidogenic chemicals to the size of whole chromosomes. Micronuclei found by image analysis which bear chromosome-like structures (judged by light microscopy) were manually identified. This selection of micronuclei was area-distributed to determine the mean size of these micronuclei. None of the peaks and plateaus in the area distributions obtained with the aneuploidogenic chemicals could be attributed to the size of a chromosome.

DNA recombination induced by aflatoxin B1 activated by cytochrome P450 1A enzymes.

Mutations in tumor suppressor genes are intricately associated with the etiology of neoplasia. Often, such mutations are followed by the loss of the second, functional alleles of tumor suppressor genes, a phenomenon known as loss of heterozygosity. Loss of heterozygosity may occur by different molecular mechanisms, including mitotic recombination, and it is conceivable that these molecular events are influenced by endogenous as well as exogenous factors. To test whether mitotic recombination is induced by certain carcinogens, we genetically engineered a Saccharomyces cerevisiae tester strain so that it metabolizes two important classes of carcinogens, polycyclic aromatic hydrocarbons and heterocyclic arylamines. This was accomplished by expressing human cDNA's coding for the cytochrome P450 (CYP) enzymes CYP1A1 or CYP1A2 in combination with NADPH-CYP oxidoreductase in a strain heterozygous for two mutations in the trp5 gene. Microsomes isolated from the transformed yeast strains activated various xenobiotics to powerful mutagens that were detected in the Ames test. Of these, the mycotoxin aflatoxin B1, when activated intracellularly in the strains containing either human CYP enzyme, significantly induced mitotic recombination. These results are discussed in light of possible mechanisms that are involved in aflatoxin B1-mediated hepatocarcinogenesis. Similarly, benzo[a]pyrene-trans-7,8-dihydrodiol and 3-amino-1-methyl-5H-pyrido[4,3-b]indole were activated to recombinagenic products, whereas benzo[a]pyrene and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline were negative in this assay. Our results argue that the constructed yeast strains may be a valuable tool for the investigation of drug-induced mitotic recombination.

Effects of ionizing radiation and caffeine treatment on cyclin dependent kinase complexes in V79 hamster cells.

Exponentially growing V79 Chinese hamster lung fibroblasts irradiated with 7 Gy X-rays undergo cell cycle arrest in the S and G2 phases. These arrests are released, probably on completion of DNA repair. A premature release occurs after treatment of irradiated cells with caffeine. This release is accompanied by increased activity of the p34cdc2 serine/threonine protein kinase complex [Hain et al. (1993) Cancer Res. 53, 1507-1510]. We have investigated in V79 cells whether the association of p34cdc2 with its regulatory subunits cyclin A and B is affected by irradiation and subsequent caffeine treatment and found that this was not the case. The phosphorylation of p34cdc2 as assayed by mobility shift on SDS polyacrylamide gels was increased as early as 0.5 h after irradiation and decreased after subsequent caffeine treatment. A novel protein p40, detected with anti-PSTAIRE antibodies, appeared several fold more abundant than p34cdc2. Its phosphorylation state also changed after irradiation and after subsequent caffeine treatment.

Characterization of the trp5-27 allele used to monitor drug-induced mitotic gene conversion in the Saccharomyces cerevisiae tester strain D7.

Mitotic gene conversions, among other recombinagenic events, can play an important role in the multistep process of carcinogenesis. The ability of chemicals to induce such gene conversions can easily be monitored in the Saccharomyces cerevisiae tester strain YHE2, a derivative of strain D7. For the detection of drug-induced gene conversions, two mutations in the TRP5 locus are used, trp5-12 and trp5-27. Here we report on the characterization of the stable allele trp5-27. Our analysis revealed two relevant mutations in trp5-27: (a) a transition C to T at position 121 after ATG that results in an amber stop codon and abolishes gene expression and (b) a transversion A to T at position 1555 that creates an ochre stop codon. Simultaneous amber and ochre suppression with the suppressors SUP3 and SUP11, respectively, was capable of relieving the tryptophan-requiring phenotype of strains carrying the trp5-27 allele. These findings have implications on the length of gene conversion tracts in conversion events between trp5-12 and trp5-27: conversion tracts can cover several kilobases, if the site of the mutation in trp5-12 lies outside of the positions mutated in trp5-27. Conversely, the maximal length is limited to 1435 bp, if the mutation in trp5-12 is located between the positions mutated in trp5-27.

High promutagen activating capacity of yeast microsomes containing human cytochrome P-450 1A and human NADPH-cytochrome P-450 reductase.

Yeast Saccharomyces cerevisiae strains have been constructed that co-express cDNAs coding for the human cytochrome P-450 enzymes CYP1A1 or CYP1A2 in combination with human NADPH-cytochrome P-450 reductase (oxidoreductase). Microsomal fractions prepared from the strains were able to efficiently activate various drugs to Salmonella mutagens. These experiments demonstrated that a functional interaction occurred between the respective human enzymes in the yeast microsomes. For every drug tested, the microsomes containing CYP enzymes and oxidoreductase were 2- to 4-fold better in activation than the corresponding microsomes that contained CYP alone. Interestingly, co-expression of CYP1A2 with oxidoreductase resulted in a decrease of 7-ethoxyresorufin-O-deethylase activity, a problem which is related to this specific substrate. Using the microsomes, it was demonstrated that aflatoxin B1 was activated to a mutagen not only by CYP1A2 but also by CYP1A1. In contrast, benzo[a]pyrene was exclusively activated by CYP1A1 whereas CYP1A2 was inactive. The drug 3-amino-1-methyl-5H-pyrido[4,3-b]indole (Trp-P-2) was activated by CYP1A2 and to a lesser extent by CYP1A1. A strong substrate specificity was observed with the two structurally related heterocyclic arylamines 2-amino-3,4-dimethylimidazo[4,5-f]quinoline (MeIQ) and 2-amino-3,4-dimethylimidazo[4,5-f]quinoline (MeIQ) and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx). MeIQx was activated efficiently by both CYP enzymes, whereas MeIQ was only activated by CYP1A2 and not by CYP1A1. The fact that microsomes from vector transformed control strains were unable to activate any of the drugs studied underlines the suitability of these microsomes for metabolic studies. Moreover, the presence of suitable marker genes in the yeast strains will enable us to study mitotic recombination and gene conversion events induced by drugs that require metabolic activation.

Reciprocal mitotic recombination is the predominant mechanism for the loss of a heterozygous gene in Saccharomyces cerevisiae.

The loss of a functional copy of a heterozygous tumor suppressor gene represents an important step during neoplastic transformation. In order to learn more about the genetic events that lead to spontaneous and drug-induced loss of heterozygosity, a diploid Saccharomyces cerevisiae strain was constructed that allows the detection of the loss of a heterozygous gene by means of direct selection. The strain contains a single functional URA3 gene copy inserted at the ADE2 locus located on the right arm of chromosome 15. In addition, the chromosome contains two other phenotypic marker genes, HIS3 which is located distal from URA3, and PHO80 which is closely linked to the centromere. The homologous chromosome lacks all three marker genes. Loss of the heterozygous copy of URA3 can easily be detected by 5-fluoro-orotic acid resistance of the resulting clones. Simple phenotypic tests of the resistant clones further allows one to distinguish whether the loss of the URA3 gene copy occurred by crossing over, chromosomal loss, or point mutation and gene conversion. Loss of heterozygosity was found to be induced in a dose-dependent fashion by UV radiation and by several chemical agents. All the tested mutagens induced loss of heterozygosity predominantly by crossing over.

Caffeine, estradiol, and progesterone interact with human CYP1A1 and CYP1A2. Evidence from cDNA-directed expression in Saccharomyces cerevisiae.

Heterologous expression of cytochrome P-450 cDNAs in yeast is a potent instrument for the study of enzyme-specific parameters and can be used to answer questions with regard to substrate specificity as well as drug interaction in a background with no interfering activities. Two cDNAs of human CYP1A1 and CYP1A2 were expressed in yeast Saccharomyces cerevisiae, and microsomes of transformed strains contained substantial amounts of functional heterologous enzymes. Enzyme kinetics with 7-ethoxyresorufin as substrate resulted in KM values of 0.017 and 1.67 microM and Vmax values of 840 and 387 pmol/mg/min for CYP1A1 and CYP1A2, respectively. Both heterologous enzymes showed an overlapping substrate specificity pattern assayed with different phenoxazone ethers and caffeine. Caffeine was shown to be metabolized by CYP1A2 and CYP1A1. Both enzymes formed paraxanthine and minor amounts of theobromine; however, trimethyluric acid was exclusively formed by CYP1A1. The fact that theophylline was not formed by either enzyme anticipates the involvement of additional enzyme(s) in the primary metabolism of caffeine. Inhibition studies with caffeine, phenacetin, 17 beta-estradiol, and progesterone as inhibitors of the CYP1A1 and CYP1A2 catalyzed O-deethylation of 7-ethoxyresorufin suggest all compounds as possible substrates of CYP1A enzymes. 17 beta-estradiol inhibited CYP1A1-catalyzed paraxanthine and trimethyluric acid formation. In contrast 17 beta-estradiol did not inhibit CYP1A2-catalyzed formation of primary caffeine metabolites. These data clearly demonstrate the capacity of human CYP1A1 and CYP1A2 to metabolize caffeine. Furthermore, possible consequences of CYP1A enzyme inhibition by caffeine, phenacetin, 17 beta-estradiol, and progesterone will be discussed.

Structure/activity relationships of the genotoxic potencies of sixteen pyrrolizidine alkaloids assayed for the induction of somatic mutation and recombination in wing cells of Drosophila melanogaster.

Sixteen pyrrolizidine alkaloids (PAs) were examined for their genotoxic potency in the wing spot test of Drosophila melanogaster following oral application. This in vivo assay tests for the induction of somatic mutation and mitotic recombination in cells of the developing wing primordia. All PAs tested except the C9-monoester supinine were clearly genotoxic. Depending on their chemical structure, however, genotoxicity of the PAs varied widely in a range encompassing about three orders of magnitude. In general, macrocyclic diester-type PAs were the most and 7-hydroxy C9-monoester types the least genotoxic representatives studied, while open diesters were intermediate in this respect. Stereoisomeric PAs mostly showed similar, but sometimes also clearly unequal genotoxicity. An increasing number of hydroxy groups in the PA molecule seemed to reduce its genotoxic potency. With respect to the structure/activity relationships, there appears to be a good correlation between hepatotoxicity of PAs in experimental rodents and genotoxicity in the wing spot test of Drosophila. This suggests that PAs are bioactivated along similar pathways in the mammalian liver and in the somatic cells of Drosophila. The genotoxic potential of PAs in the Drosophila wing spot test and their carcinogenic potential in mammals also seem correlated, although the information in the literature on carcinogenicity of the non-macrocyclic PAs with moderate to low genotoxic potency is concededly limited. Comparisons with other genotoxicity tests suggest that the wing spot test is particularly suitable for genotoxins like PAs, on the one hand because of the versatile metabolic bioactivation system of Drosophila and on the other hand also because of its excellent sensitivity to the crosslinking agents among the genotoxins.

The genotoxicity of the anti-cancer drug mitoxantrone in somatic and germ cells of Drosophila melanogaster.

The novel antineoplastic drug mitoxantrone was studied for its genotoxic effects in Drosophila melanogaster. In male germ cells, the clinical preparation Novantrone, the dihydrochloride salt of mitoxantrone, did not induce sex-linked recessive lethal mutations in feeding and injection experiments with adult flies, although statistically the results were inconclusive rather than truly negative. However, the free base mitoxantrone was weakly, but significantly genotoxic in this test (0.14% lethals/mM exposure concentration); this is most probably the result of prolonged exposure. On the other hand, both forms of mitoxantrone assayed were clearly genotoxic in the somatic mutation and recombination test of the wing. This test assays the cells of the proliferating imaginal wing discs of larvae. Depending on the feeding method used, the overall clone induction frequency was in the range of about 2-6 x 10(-5) per cell and cell generation and per mM exposure dose. Correction of these frequencies according to mean clone size led to slightly higher estimates (by about 5-25% higher). Although the majority of the clone induction events are due to mitotic recombination, a significant proportion can be attributed to mutational events (gene and chromosome mutations). The genotoxicity of mitoxantrone seems to depend mainly on impaired DNA synthesis in cycling cells owing to the compound's ability to inhibit topoisomerase II by intercalation into DNA.

Genotoxicity of 2-amino-3-methylimidazo[4,5-f]quinoline (IQ) and related compounds in Drosophila.

The potent food mutagen and carcinogen 2-amino-3-methylimidazo[4,5- f]quinoline (IQ) and the structurally related heterocyclic aromatic amines 2-aminoimidazo[4,5-f]quinoline (demethyl-IQ) and 2-amino-1-methylimidazo[4,5-f]quinoline (iso-IQ) were assayed for genotoxicity in the wing somatic mutation and recombination test (SMART) as well as in the sex-linked recessive lethal (SLRL) test in Drosophila melanogaster. In addition, 3-methyl-2-nitroimidazo[4,5-f]quinoline (nitro-IQ), 2-nitrofluorene and 1,8-dinitropyrene were also assayed in the wing spot test. IQ was clearly mutagenic in the SLRL test with highest activity in spermatids. Iso-IQ was more active than IQ whereas demethyl-IQ was inactive in this test. The same pattern of results was obtained in the wing SMART: iso-IQ produced greater than 2-fold higher frequencies of spots than IQ and demethyl-IQ was clearly negative. In addition, nitro-IQ exhibited an approximately equal genotoxic activity as IQ. 2-Nitrofluorene and 1,8-dinitropyrene were both inactive in the wing spot test. These data provide good evidence for a correlation of genotoxic effects in germinal and somatic cells, and for the practical advantage of the wing spot test in Drosophila. Moreover, the results show structure-activity relationships among the heterocyclic aromatic amines and nitro compounds similar to those found in Salmonella.

Genetic toxicity of six carcinogens and six non-carcinogens in the Drosophila wing spot test.

Six rodent carcinogens, 5 of which are also human carcinogens, and 6 compounds recognized as non-carcinogens were tested for their genotoxic activity in the Drosophila melanogaster wing spot test. 72-h-old larvae trans-heterozygous for the recessive wing cell markers 'multiple wing hairs' (mwh) and 'flare' (flr3) were fed various concentrations of the test compounds for a period of 48 h. With amitrole and 4-aminobiphenyl, larvae of the same age were also given an acute treatment of 6 h with higher concentrations, and, in addition, 48-h-old larvae were fed for a longer period of 72 h. Repeats of all experiments document the good reproducibility of the results in the wing spot test. Amitrole and 4-aminobiphenyl were genotoxic after both 48-h and 72-h treatments, but their activity could not be detected following acute exposure of only 6 h. Chlorambucil and melphalan were clearly genotoxic. The carcinogens sodium arsenite and sodium arsenate, however, which are highly toxic to Drosophila, could only be tested at low exposure levels and were negative under these treatment conditions. The 6 non-carcinogens (ascorbic acid, 2-aminobiphenyl, mannitol, piperonyl butoxide, stannous chloride and titanium dioxide) were all definitely non-genotoxic in the Drosophila wing spot test. The data for the non-carcinogens demonstrate that non-genotoxic compounds can be identified in the wing spot test with a reasonable experimental effort.

Knowledge-based battery design of short-term tests based on dose information.

A construction of batteries of short-term tests (STTs) is described which is based on a classification of 73 chemicals in regard to their carcinogenicity. The 73 chemicals were studied within the U.S. National Toxicology Program (Ashby and Tennant, 1988). The batteries are validated using the classification of 35 additional chemicals. They are defined by logically structured combinations of rules. The single rules are defined by the z-scores of the logarithmic values of the limiting doses obtained from the 4 in vitro STTs used in the study by Ashby and Tennant. The limiting dose is defined as the lowest effective dose or the highest ineffective dose (Waters et al., 1987). The batteries are constructed by minimizing the number of disagreements with the classification by Ashby and Tennant. Compared with the results obtained from single STTs, 2 batteries of 3 STTs have higher concordances with the carcinogenicity data, namely 70% for the NTP data and 74-77% for the independent test data. In addition, a theoretical result shows that the proposed battery design, for a large enough learning set of chemicals, leads to results which are replicated with high probability on a large enough validation set. Based on the first results obtained with a limited number of chemicals it is concluded that the knowledge-based battery design is worth further development.