Search

The in vitro micronucleus assay is a globally significant method for DNA damage quantification used for regulatory compound safety testing in addition to inter-individual monitoring of environmental, lifestyle and occupational factors. However, it relies on time-consuming and user-subjective manual scoring. Here we show that imaging flow cytometry and deep learning image classification represents a capable platform for automated, inter-laboratory operation. Images were captured for the cytokinesis-block micronucleus (CBMN) assay across three laboratories using methyl methanesulphonate (1.25–5.0 μg/mL) and/or carbendazim (0.8–1.6 μg/mL) exposures to TK6 cells. Human-scored image sets were assembled and used to train and test the classification abilities of the “DeepFlow” neural network in both intra- and inter-laboratory contexts. Harnessing image diversity across laboratories yielded a network able to score unseen data from an entirely new laboratory without any user configuration. Image classification accuracies of 98%, 95%, 82% and 85% were achieved for ‘mononucleates’, ‘binucleates’, ‘mononucleates with MN’ and ‘binucleates with MN’, respectively. Successful classifications of ‘trinucleates’ (90%) and ‘tetranucleates’ (88%) in addition to ‘other or unscorable’ phenotypes (96%) were also achieved. Attempts to classify extremely rare, tri- and tetranucleated cells with micronuclei into their own categories were less successful (≤ 57%). Benchmark dose analyses of human or automatically scored micronucleus frequency data yielded quantitation of the same equipotent concentration regardless of scoring method. We conclude that this automated approach offers significant potential to broaden the practical utility of the CBMN method across industry, research and clinical domains. We share our strategy using openly-accessible frameworks. Supplementary Information The online version contains supplementary material available at 10.1007/s00204-021-03113-0.

In vitro genotoxicity studies are a quick and high throughput approach to assess the genotoxic potential of chemicals; however, the reliability of these tests and their relevance to in vivo effects depends on the choice of representative cell line and optimisation of assay conditions. For chemicals like urethane that require specific metabolic activation to cause genotoxicity, it is important that in vitro tests are conducted using cell lines exhibiting the activity and induction of CYP450 enzymes, including CYP2E1 enzyme that is important in the metabolism of urethane, at a concentration representing actual or perceived chemical exposure. We compared 2D MCL-5 cells and HepG2 cells with 3D HepG2 hanging drop spheroids to determine the genotoxicity of urethane using the micronucleus assay. Our 2D studies with MCL-5 did not show any statistically significant genotoxicity [99% relative population doubling (RPD)] compared to controls for concentrations and time point tested in vitro. HepG2 cells grown as 2D indicated that exposure to urethane of up to 30 mM for 23 h did not cause any genotoxic effect (102% RPD) but, at higher concentrations, genotoxicity was produced with only 89–85% RPD. Furthermore, an exposure of 20–50 mM for 23 h using 3D hanging drop spheroid assays revealed a higher MN frequency, thus exhibiting in vitro genotoxicity of urethane in metabolically active cell models. In comparison with previous studies, this study indicated that urethane genotoxicity is dose, sensitivity of cell model (2D vs. 3D) and exposure dependent.

Thein vitromicronucleus assay is a globally significant method for DNA damage quantification used for regulatory compound safety testing in addition to inter-individual monitoring of environmental, lifestyle and occupational factors. However it relies on time-consuming and user-subjective manual scoring. Here we show that imaging flow cytometry and deep learning image classification represents a capable platform for automated, inter-laboratory operation. Images were captured for the cytokinesis-block micronucleus (CBMN) assay across three laboratories using methyl methanesulphonate (1.25 – 5.0 µg/mL) and/or carbendazim (0.8 – 1.6 µg/mL) exposures to TK6 cells. Human-scored image sets were assembled and used to train and test the classification abilities of the “DeepFlow” neural network in both intra- and inter-laboratory contexts. Harnessing image diversity across laboratories yielded a network able to score unseen data from an entirely new laboratory without any user configuration. Image classification accuracies of 98%, 95%, 82% and 85% were achieved for ‘mononucleates’, ‘binucleates’, ‘mononucleates with MN’ and ‘binucleates with MN’, respectively. Successful classifications of ‘trinucleates’ (90%) and ‘tetranucleates’ (88%) in addition to ‘other or unscorable’ phenotypes (96%) were also achieved. Attempts to classify extremely rare, tri- and tetranucleated cells with micronuclei into their own categories were less successful (≤ 57%). Benchmark dose analyses of human or automatically scored micronucleus frequency data yielded quantitation of the same equipotent dose regardless of scoring method. We conclude that this automated approach offers significant potential to broaden the practical utility of the CBMN method across industry, research and clinical domains. We share our strategy using openly-accessible frameworks.

Current in vitro genotoxicity tests can produce misleading positive results, indicating an inability to effectively predict a compound’s subsequent carcinogenic potential in vivo. Such oversensitivity can incur unnecessary in vivo tests to further investigate positive in vitro results, supporting the need to improve in vitro tests to better inform risk assessment. It is increasingly acknowledged that more informative in vitro tests using multiple endpoints may support the correct identification of carcinogenic potential. The present study, therefore, employed a holistic, multiple-endpoint approach using low doses of selected carcinogens and non-carcinogens (0.001–770 µM) to assess whether these chemicals caused perturbations in molecular and cellular endpoints relating to the Hallmarks of Cancer. Endpoints included micronucleus induction, alterations in gene expression, cell cycle dynamics, cell morphology and bioenergetics in the human lymphoblastoid cell line TK6. Carcinogens ochratoxin A and oestradiol produced greater Integrated Signature of Carcinogenicity scores for the combined endpoints than the “misleading” in vitro positive compounds, quercetin, 2,4-dichlorophenol and quinacrine dihydrochloride and toxic non-carcinogens, caffeine, cycloheximide and phenformin HCl. This study provides compelling evidence that carcinogens can successfully be distinguished from non-carcinogens using a holistic in vitro test system. Avoidance of misleading in vitro outcomes could lead to the reduction and replacement of animals in carcinogenicity testing. Electronic supplementary material The online version of this article (10.1007/s00204-020-02902-3) contains supplementary material, which is available to authorized users.

Human exposure to carcinogens occurs via a plethora of environmental sources, with 70–90% of cancers caused by extrinsic factors. Aberrant phenotypes induced by such carcinogenic agents may provide universal biomarkers for cancer causation. Both current in vitro genotoxicity tests and the animal-testing paradigm in human cancer risk assessment fail to accurately represent and predict whether a chemical causes human carcinogenesis. The study aimed to establish whether the integrated analysis of multiple cellular endpoints related to the Hallmarks of Cancer could advance in vitro carcinogenicity assessment. Human lymphoblastoid cells (TK6, MCL-5) were treated for either 4 or 23 h with 8 known in vivo carcinogens, with doses up to 50% Relative Population Doubling (maximum 66.6 mM). The adverse effects of carcinogens on wide-ranging aspects of cellular health were quantified using several approaches; these included chromosome damage, cell signalling, cell morphology, cell-cycle dynamics and bioenergetic perturbations. Cell morphology and gene expression alterations proved particularly sensitive for environmental carcinogen identification. Composite scores for the carcinogens’ adverse effects revealed that this approach could identify both DNA-reactive and non-DNA reactive carcinogens in vitro. The richer datasets generated proved that the holistic evaluation of integrated phenotypic alterations is valuable for effective in vitro risk assessment, while also supporting animal test replacement. Crucially, the study offers valuable insights into the mechanisms of human carcinogenesis resulting from exposure to chemicals that humans are likely to encounter in their environment. Such an understanding of cancer induction via environmental agents is essential for cancer prevention. Electronic supplementary material The online version of this article (10.1007/s00204-017-2102-y) contains supplementary material, which is available to authorized users.

Use of imaging flow cytometry to assess induced DNA damage via the cytokinesis block micronucleus (CBMN) assay has thus far been limited to radiation dosimetry in human lymphocytes using high end, 'ImageStream X' series imaging cytometers. Its potential to enumerate chemically induced DNA damage using in vitro cell lines remains unexplored. In the present manuscript, we investigate the more affordable FlowSight® imaging cytometry platform to assess in vitro micronucleus (MN) induction in the human lymphoblastoid TK6 and metabolically competent MCL-5 cells treated with Methyl Methane Sulfonate (MMS) (0-5 µg/ml), Carbendazim (0-1.6 µg/ml), and Benzo[a]Pyrene (B[a]P) (0-6.3 µg/ml) for a period of 1.5-2 cell-cycles. Cells were fixed, and nuclei and MN were stained using the fluorescent nuclear dye DRAQ5™. Image acquisition was carried out using a 20X objective on a FlowSight® imaging cytometer (Amnis, part of Merck Millipore) equipped with a 488 nm laser. Populations of ∼20000 brightfield cell images, alongside DRAQ5™ stained nuclei/MN were rapidly collected (≤10 min). Single, in-focus cells suitable for scoring were then isolated using the IDEAS® software. An overlay of the brightfield cell outlines and the DRAQ5 nuclear fluorescence was used to facilitate scoring of mono-, bi-, tri-, and tetra-nucleated cells with or without MN events and in context of the cytoplasmic boundary of the parent cell.To establish the potential of the FlowSight® platform, and to establish 'ground truth' cell classification for the supervised machine learning based scoring algorithm that represents the next stage of our project, the captured images were scored manually. Alongside, MN frequencies were also derived using the 'gold standard' light microscopy and manual scoring. A minimum of 3000 bi-nucleated cells were assessed using both approaches. Using the benchmark dose approach, the comparability of genotoxic potency estimations for the different compounds and cell lines was assessed across the two scoring platforms as highly similar. This study therefore provides essential proof-of-concept that FlowSight® imaging cytometry is capable of reproducing the results of 'gold standard' manual scoring by light microscopy. We conclude that, with the right automated scoring algorithm, imaging flow cytometry could revolutionise the reportability and scoring throughput of the CBMN assay.

Genetic toxicology testing has a crucial role in the safety assessment of new and existing substances of societal value by reducing/eliminating human exposure to potential somatic and germ cell mutagens. Genetic toxicology assays have historically been used in a qualitative manner to arrive at the binary decision of ‘yes’ or ‘no’ with regards to the mutagenic potential. However, the field is currently at a crossroads, with new methods being developed and new proposals being made to use genetic toxicity data in a more quantitative manner. Technological advances have made it possible to perform high-content, high-throughput and high-precision analysis to increase the number of “scored” events leading to increased statistical precision of the endpoint under evaluation. Automated flow cytometry and image analysis are providing significant advantages for the evaluation of gene mutations as well as cytogenetic damage both in vitro and in vivo. In addition, statistical methods such as the benchmark dose (BMD) approach can be used to identify point of departure (PoD) metrics for use in human health risk assessments, including estimation of reference dose (RfD) and margins of exposure (MOE) from in vivo data. Here we provide new data to compare different in vitro micronucleus approaches, observing that the flow based assay performs very well in defining a PoD for methyl methanesulfonate. We also present reanalysis of published in vivo Pig-a gene mutation data, to show how covariate analysis increases precision and reduces the effects of outliers when defining BMD values. Furthermore, we show how in vivo BMD metrics can be used to define RfD values, and then provide comparisons to other human exposure limit values such as permitted daily exposure (PDE). Finally, the principles of empirical correlation using BMD metrics are presented, with methods for derivation of BMD values for endpoint B, when using data from only endpoint A. These developments are opening the possibility of genetic toxicity data being used as an apical endpoint to define negligible risk in human health risk assessments. Expert groups consisting of stakeholders representing academia, industry and the government are now developing guidance on transforming genetic toxicology testing from a qualitative to a quantitative science, keeping in mind the 3R principles of animal welfare.

The use of manual microscopy for the scoring of chromosome damage in the in vitro micronucleus assay is often associated with user subjectivity. This level of subjectivity can be reduced by using automated platforms, which have added value of faster with high-throughput and multi-endpoint capabilities. However, there is a need to assess the reproducibility and sensitivity of these automated platforms compared with the gold standard of the manual scoring. The automated flow cytometry-based MicroFlow® and image analysis-based Metafer™ were used for dose response analyses in human lymphoblastoid TK6 cells exposed to the model clastogen, methyl methanesulfonate (MMS), aneugen, carbendazim, and the weak genotoxic carcinogen, ochratoxin A (OTA). Cells were treated for 4 or 30 h, with a 26- or 0-h recovery. Flow cytometry scoring parameters and the Metafer™ image classifier were investigated, to assess any potential differences in the micronucleus (MN) dose responses. Dose response data were assessed using the benchmark dose approach with chemical and scoring system set as covariate to assess reproducibility between endpoints. A clear increase in MN frequency was observed using the MicroFlow® approach on TK6 cells treated for 30 h with MMS, carbendazim and OTA. The MicroFlow®-based MN frequencies were comparable to those derived by using the Metafer™ and manual scoring platforms. However, there was a potential overscoring of MN with the MicroFlow® due to the cell lysis step and an underscoring with the Metafer™ system based on current image classifier settings. The findings clearly demonstrate that the MicroFlow® and Metafer™ MN scoring platforms are powerful tools for automated high-throughput MN scoring and dose response analysis. Electronic supplementary material The online version of this article (doi:10.1007/s00204-016-1903-8) contains supplementary material, which is available to authorized users.

A recent shift by the scientiˆc and regulatory communi-ty, towards accepting the existence of non-linear doseresponses for certain DNA reactive genotoxic agents, hasunveiled a myriad of questions regarding their biologicalbasis. The mechanisms responsible for `genotoxic toler-ance' at low doses are wide ranging but poorly under-stood, yet this information is essential when analysingnon-linear dose responses for hazard and risk assessment.For DNA reactive genotoxins, non-linear dose responsescan arise from many diSerent biological mechanisms, in-cluding DNA repair. Recent work from our group exploredthe contributory role of DNA repair to nonlinear genotoxicdose responses, in human cells exposed alkylating agents.Here we discuss the involvement of the repair enzymesmethylpurine DNA-glycosylase and methyl-guaninemethyl-transferase in modulating the non-linear doseresponses observed in human cells exposed to ethylmethanesulfonate (EMS) and N-methyl-N-nitrosourea, re-spectively. We also discuss the exposure of binary mix-tures, and how combinations of the dissimilar actingagents Benomyl and EMS at their no observed genotoxiceSect levels, induce a signiˆcant increase in micronuclei.Key words: alkylating agents, thresholds, mode of action,DNA repair, mixtures

The use of manual microscopy for the scoring of chromosome damage in the in vitro micronucleus assay is often associated with user subjectivity. This level of subjectivity can be reduced by using automated platforms, which have added value of faster with high-throughput and multi-endpoint capabilities. However, there is a need to assess the reproducibility and sensitivity of these automated platforms compared with the gold standard of the manual scoring. The automated flow cytometry-based MicroFlow® and image analysis-based Metafer™ were used for dose response analyses in human lymphoblastoid TK6 cells exposed to the model clastogen, methyl methanesulfonate (MMS), aneugen, carbendazim, and the weak genotoxic carcinogen, ochratoxin A (OTA). Cells were treated for 4 or 30 h, with a 26- or 0-h recovery. Flow cytometry scoring parameters and the Metafer™ image classifier were investigated, to assess any potential differences in the micronucleus (MN) dose responses. Dose response data were assessed using the benchmark dose approach with chemical and scoring system set as covariate to assess reproducibility between endpoints. A clear increase in MN frequency was observed using the MicroFlow® approach on TK6 cells treated for 30 h with MMS, carbendazim and OTA. The MicroFlow®-based MN frequencies were comparable to those derived by using the Metafer™ and manual scoring platforms. However, there was a potential overscoring of MN with the MicroFlow® due to the cell lysis step and an underscoring with the Metafer™ system based on current image classifier settings. The findings clearly demonstrate that the MicroFlow® and Metafer™ MN scoring platforms are powerful tools for automated high-throughput MN scoring and dose response analysis.