Michael D. Fellows

Comparison of different methods for an accurate assessment of cytotoxicity in the in vitro micronucleus test: II: Practical aspects with toxic agents

Appropriate measures of cytotoxicity need to be used when selecting test concentrations in in vitro genotoxicity assays. Underestimation of toxicity may lead to inappropriately toxic concentrations being selected for analysis, with the potential for generation of irrelevant positive results. As guidance for the in vitro micronucleus test is being developed, it is clearly important to compare the different measures of cytotoxicity that can be used both with and without cytokinesis blocking. Therefore, relative cell counts (RCC), relative increase in cell counts (RICC) and relative population doubling (RPD) for treatments without cytokinesis block were compared with replication index (RI) for treatments with cytokinesis block, and the corresponding induction of micronucleated cells was evaluated. A wide range of chemicals and gamma irradiation were used, and in almost all cases, RCC underestimated cytotoxicity when compared with all other measures such that RCC would have resulted in the selection of inappropriately high concentrations for micronuclei analysis. In the absence of cytokinesis block, RICC or RPD is more comparable with RI with cytokinesis block, and therefore considered more appropriate measure of survival. Furthermore, using these estimations of cytotoxicity and the limit of 50% survival, all the mutagens and aneugens tested were appropriately identified as positive in the in vitro micronucleus assay. Accordingly, it was clear that testing beyond 50% survival was not necessary to identify the potential of these agents to induce micronuclei.

The ability of the mouse lymphoma TK assay to detect aneugens

There is some evidence that the mouse lymphoma TK assay (MLA) can detect aneugens, and this is accepted in the current International Conference on Harmonisation guidance for testing pharmaceuticals. However, whether or not it can be used as a reliable screen for aneugenicity has been the subject of debate. Consequently, aneugens with diverse mechanisms of action were tested in the MLA using 24-h exposure. No evidence of increased mutant frequency was seen with noscapine, diazepam or colchicine and increases were seen with taxol, carbendazim, econazole and chloral hydrate only at high levels of toxicity (for all but one taxol concentration survival reduced to ≤10% of control). None of these agents would be unequivocally classified as positive using currently accepted criteria. The largest increases in mutant number were seen with taxol and carbendazim; therefore, trifluorothymidine (TFT)-resistant clones resulting from treatment with them were cultured and analysed for chromosome 11 copy number using fluorescent in situ hybridisation (FISH) and loss of heterozygosity (LOH). High concentrations of these aneugens induced LOH at all loci examined indicating only one chromosome 11 was present but, perhaps surprisingly, all were found to have two copies of chromosome 11 using FISH. This would be consistent with loss of the tk(+) chromosome 11b with concomitant duplication of chromosome 11a, which has been proposed as a likely mechanism for induction of TFT-resistant clones. However, it was also surprising that analysis of centromere size showed that almost all the clones had both small and large centromeres, i.e. suggesting the presence of both chromosomes 11a and 11b. In conclusion, it appears that the TFT-resistant mutants resulting from treatment with toxic concentrations of some aneugens such as taxol and carbendazim have undergone complex genetic changes. However, these data show that the MLA cannot be used as a routine screen to detect aneugens.

Anomalous genotoxic responses induced in mouse lymphoma L5178Y cells by potassium bromate.

Potassium bromate (KBrO3) is a well-established rodent kidney carcinogen and its oxidising activity is considered to be a significant factor in its mechanism of action. Although it has also been shown to be clearly genotoxic in a range of in vivo and in vitro test systems, surprisingly, it is not readily detected in several cell lines using the standard alkaline Comet assay. However, previous results from this laboratory demonstrated huge increases in tail intensity by modifying the method to include incubation with either human 8-oxodeoxyguanosine DNA glycosylase-1 (hOGG1) or bacterial formamidopyrimidine DNA glycosylase (FPG) indicating that, as expected, significant amounts of 8-oxodeoxyguanosine (8-OHdG) were induced. The purpose of this work, therefore, was to investigate why KBrO3, in contrast to other oxidising agents, gives a relatively poor response in the standard Comet assay. Results confirmed that it is a potent genotoxin in mouse lymphoma L5178Y cells inducing micronuclei and mutation at the tk and hprt loci at relatively non-cytotoxic concentrations. Subsequent time-course studies demonstrated that substantial amounts of 8-OHdG appear to remain in cells 24h after treatment with KBrO3 but result in no increase in frank stand breaks (FSB) even though phosphorylated histone H2AX (gamma-H2AX) antibody labelling confirmed the presence of double-strand breaks. Using bromodeoxyuracil (BrdU) incorporation together with measured increases in cell numbers, L5178Y cells also appeared to go through the cell cycle with unrepaired hOGG1-recognisable damage. Since unrepaired 8-OHdG can give rise to point mutations through G:C-->T:A transversions, it was also surprising that mutation could not be detected at the Na+/K+ATPase locus as determined by ouabain resistance. Some increases in strand breakage could be seen in the Comet assay by increasing the unwinding time, but only at highly toxic concentrations and to a much smaller extent than would be expected from the magnitude of the other genotoxic responses. It was considered unlikely that these anomalous observations were due to the inability of L5178Y cells to recognise 8-OHdG because these cells were shown to express mOGG1 and have functional cleavage activity at the adducted site. It appears that the responses of L5178Y cells to KBrO3 are complex and differ from those induced by other oxidising agents.

Unusual structure–genotoxicity relationship in mouse lymphoma cells observed with a series of kinase inhibitors

During development of a novel kinase inhibitor for an anti-inflammatory therapy at AstraZeneca UK, the lead compound was found to be potently active in the mouse lymphoma assay (MLA). This was not believed to be due to primary pharmacology because structural alert relationships and a negative Ames test indicated that the compound was unlikely to form DNA adducts. A number of investigations were performed to assess whether mammalian cell genotoxicity was inherent to the chemical series. The in vitro micronucleus assay (MN(vit)) combined with a semi-automated analysis system, was used as a high-throughput screen. A number of additional compounds were selected for testing, all with different substituents around a core isoquinolinone. These modifications did not affect the kinase and non-kinase selectivity of the compounds. Several of these compounds were positive in the MN(vit), however, two compounds were found to be negative and these were also confirmed to be negative in the MLA. It was considered possible that topoisomerase II or off-target kinase inhibition may have been responsible for the observed mammalian cell genotoxicity. The present investigations show how an iterative chemical design, along with genotoxicity screening by use of a semi-automated MN(vit), can identify and remove the genotoxic hazard from pharmaceutical projects at an early stage of development, and produce high-quality molecules suitable for further progression.