Automated read-across workflow for predicting acute oral toxicity: I. The decision scheme in the QSAR toolbox.

A decision-scheme outlining the steps for identifying the appropriate chemical category and subsequently appropriate tested source analog(s) for data gap filling of a target chemical by read-across is described. The primary features used in the grouping of the target chemical with source analogues within a database of 10,039 discrete organic substances include reactivity mechanisms associated with protein interactions and specific-acute-oral-toxicity-related mechanisms (e.g., mitochondrial uncoupling). Additionally, the grouping of chemicals making use of the in vivo rat metabolic simulator and neutral hydrolysis. Subsequently, a series of structure-based profilers are used to narrow the group to the most similar analogues. The scheme is implemented in the OECD QSAR Toolbox, so it automatically predicts acute oral toxicity as the rat oral LD50 value in log [1/mol/kg]. It was demonstrated that due to the inherent variability in experimental data, classification distribution should be employed as more adequate in comparison to the exact classification. It was proved that the predictions falling in the adjacent GSH categories to the experimentally-stated ones are acceptable given the variation in experimental data. The model performance estimated by adjacent accuracy was found to be 0.89 and 0.54 while based on R2. The mechanistic and predictive coverages were >0.85.

UVCB substances II: Development of an endpoint-nonspecific procedure for selection of computationally generated representative constituents.

Substances of unknown or variable composition, complex reaction products, and biological materials (UVCBs) comprise approximately 40% of all registered substances submitted to the European Chemicals Agency. One of the main characteristics of UVCBs is that they have no unique representation. Industry scientists who are part of the scientific community have been working with academics and consultants to address the problem of a lack of a defined structural description. It has been acknowledged that one of the obstacles is the large number of possible structural isomers. We have recently proposed and published a methodology, based on the generic substance identifiers, to address this issue. The methodology allows for the coding of constituents, their generation, calculation of important characteristics of UVCB constituents, and selection of representative constituents. In the present study we introduce a statistical selection of the minimum number of generated constituents representing a UVCB. This representative sample was selected in such a way that the structural variability and the properties of concern of the UVCB were approximated within a predefined tolerable error. The aim of the statistical selection was to enable the assessment of UVCB substances by decreasing the number of constituents that need to be evaluated. The procedure, which was shown to be endpoint-independent, was validated theoretically and on real case studies. Environ Toxicol Chem 2019;38:682-694. © 2019 SETAC.

Investigating the relationship between in vitro-in vivo genotoxicity: derivation of mechanistic QSAR models for in vivo liver genotoxicity and in vivo bone marrow micronucleus formation which encompass metabolism.

Strategic testing as part of an integrated testing strategy (ITS) to maximize information and avoid the use of animals where possible is fast becoming the norm with the advent of new legislation such as REACH. Genotoxicity is an area where regulatory testing is clearly defined as part of ITS schemes. Under REACH, the specific information requirements depend on the tonnage manufactured or imported. Two types of test systems exist to meet these information requirements, in vivo genotoxicity assays, which take into account the whole animal, and in vitro assays, which are conducted outside the living mammalian organism using microbial or mammalian cells under appropriate culturing conditions. Clearly, with these different broad experimental categories, results for a given chemical can often differ, which presents challenges in the interpretation as well as in attempting to model the results in silico. This study attempted to compare the differences between in vitro and in vivo genotoxicity results, to rationalize these differences with plausible hypothesis in concert with available data. Two proof of concept (Q)SAR models were developed, one for in vivo genotoxicity effects in liver and a second for in vivo micronucleus formation in bone marrow. These "mechanistic models" will be of practical value in testing strategies, and both have been implemented into the TIMES software platform ( http://oasis-lmc.org ) to help predict the genotoxicity outcome of new untested chemicals.

Use of genotoxicity information in the development of integrated testing strategies (ITS) for skin sensitization.

Skin sensitization is an end point of concern for various legislation in the EU, including the seventh Amendment to the Cosmetics Directive and Registration Evaluation, Authorisation and Restriction of Chemicals (REACH). Since animal testing is a last resort for REACH or banned (from 2013 onward) for the Cosmetics Directive, the use of intelligent/integrated testing strategies (ITS) as an efficient means of gathering necessary information from alternative sources (e.g., in vitro, (Q)SARs, etc.) is gaining widespread interest. Previous studies have explored correlations between mutagenicity data and skin sensitization data as a means of exploiting information from surrogate end points. The work here compares the underlying chemical mechanisms for mutagenicity and skin sensitization in an effort to evaluate the role mutagenicity information can play as a predictor of skin sensitization potential. The Tissue Metabolism Simulator (TIMES) hybrid expert system was used to compare chemical mechanisms of both end points since it houses a comprehensive set of established structure-activity relationships for both skin sensitization and mutagenicity. The evaluation demonstrated that there is a great deal of overlap between skin sensitization and mutagenicity structural alerts and their underlying chemical mechanisms. The similarities and differences in chemical mechanisms are discussed in light of available experimental data. A number of new alerts for mutagenicity were also postulated for inclusion into TIMES. The results presented show that mutagenicity information can provide useful insights on skin sensitization potential as part of an ITS and should be considered prior to any in vivo skin sensitization testing being initiated.

Identifying the structural requirements for chromosomal aberration by incorporating molecular flexibility and metabolic activation of chemicals.

Modeling the potential of chemicals to induce chromosomal damage has been hampered by the diversity of mechanisms which condition this biological effect. The direct binding of a chemical to DNA is one of the underlying mechanisms that is also responsible for bacterial mutagenicity. Disturbance of DNA synthesis due to inhibition of topoisomerases and interaction of chemicals with nuclear proteins associated with DNA (e.g., histone proteins) were identified as additional mechanisms leading to chromosomal aberrations (CA). A comparative analysis of in vitro genotoxic data for a large number of chemicals revealed that more than 80% of chemicals that elicit bacterial mutagenicity (as indicated by the Ames test) also induce CA; alternatively, only 60% of chemicals that induce CA have been found to be active in the Ames test. In agreement with this relationship, a battery of models is developed for modeling CA. It combines the Ames model for bacterial mutagenicity, which has already been derived and integrated into the Optimized Approach Based on Structural Indices Set (OASIS) tissue metabolic simulator (TIMES) platform, and a newly derived model accounting for additional mechanisms leading to CA. Both models are based on the classical concept of reactive alerts. Some of the specified alerts interact directly with DNA or nuclear proteins, whereas others are applied in a combination of two- or three-dimensional quantitative structure-activity relationship models assessing the degree of activation of the alerts from the rest of the molecules. The use of each of the alerts has been justified by a mechanistic interpretation of the interaction. In combination with a rat liver S9 metabolism simulator, the model explained the CA induced by metabolically activated chemicals that do not elicit activity in the parent form. The model can be applied in two ways: with and without metabolic activation of chemicals.