Claire M. Ellison

A review of the electrophilic reaction chemistry involved in covalent protein binding relevant to toxicity.

Several pieces of legislation have led to an increased interest in the use of in silico methods, specifically the formation of chemical categories for the assessment of toxicological endpoints. For a number of endpoints, this requires a detailed knowledge of the electrophilic reaction chemistry that governs the ability of an exogenous chemical to form a covalent adduct. Historically, this chemistry has been defined as compilations of structural alerts without documenting the associated electrophilic chemistry mechanisms. To address this, this article has reviewed the literature defining the structural alerts associated with covalent protein binding and detailed the associated electrophilic reaction chemistry. This information is useful to both toxicologists and regulators when using the chemical category approach to fill data gaps for endpoints involving covalent protein binding. The structural alerts and associated electrophilic reaction chemistry outlined in this review have been incorporated into the OECD (Q)SAR Toolbox, a freely available software tool designed to fill data gaps in a regulatory environment without the need for further animal testing.

Integrating (Q)SAR models, expert systems and read-across approaches for the prediction of developmental toxicity.

It has been estimated that reproductive and developmental toxicity tests will account for a significant proportion of the testing costs associated with REACH compliance. Consequently, the use of alternative methods to predict developmental toxicity is an attractive prospect. The present study evaluates a number of computational models and tools which can be used to aid assessment of developmental toxicity potential. The performance and limitations of traditional (quantitative) structure-activity relationship ((Q)SARs) modelling, structural alert-based expert system prediction and chemical profiling approaches are discussed. In addition, the use of category formation and read-across is also addressed. This study demonstrates the limited success of current modelling methods when used in isolation. However, the study also indicates that when used in combination, in a weight-of-evidence approach, better use may be made of the limited toxicity data available and predictivity improved. Recommendations are provided as to how this area could be further developed in the future.

Assessment of Methods To Define the Applicability Domain of Structural Alert Models

It is important that in silico models for use in chemical safety legislation, such as REACH, are compliant with the OECD Principles for the Validation of (Q)SARs. Structural alert models can be useful under these circumstances but lack an adequately defined applicability domain. This paper examines several methods of domain definition for structural alert models with the aim of assessing which were the most useful. Specifically, these methods were the use of fragments, chemical descriptor ranges, structural similarity, and specific applicability domain definition software. Structural alerts for mutagenicity in Derek for Windows (DfW) were used as examples, and Ames test data were used to define and test the domain of chemical space where the alerts produce reliable results. The usefulness of each domain was assessed on the criterion that confidence in the correctness of predictions should be greater inside the domain than outside it. By using a combination of structural similarity and chemical fragments a domain was produced where the majority of correct positive predictions for mutagenicity were within the domain and a large proportion of the incorrect positive predictions outside it. However this was not found for the negative predictions; there was little difference between the percentage of true and false predictions for inactivity which were found as either within or outside the applicability domain. A hypothesis for the occurrence of this difference between positive and negative predictions is that differences in structure between training and test compounds are more likely to remove the toxic potential of a compound containing a structural alert than to add an unknown mechanism of action (structural alert) to a molecule which does not already contain an alert. This could be especially true for well studied end points such as the Ames assay where the majority of mechanisms of action are likely to be known.

Definition of the structural domain of the baseline non-polar narcosis model for Tetrahymena pyriformis †

The aim of this work was to develop a high-quality 1-octanol/water partition coefficient-dependent (log P) baseline quantitative structure-activity relationship (QSAR) for the toxicity (log ) of classic non-polar narcotics to Tetrahymena pyriformis, and subsequently use this model to define the domain of applicability for baseline narcosis. The toxicities to T. pyriformis of 514 possible non-polar narcotics were assessed. A QSAR to predict toxicity was created from a training set of 87 classic non-polar narcotics (the saturated alcohols and ketones): log = 0.78 log P–2.01 (n = 87, r 2 = 0.96). This model was then used to predict the toxicity of the remaining chemicals. The chemicals from the large dataset which were poorly predicted by the model (i.e. the prediction was > ±0.5 log units from the experimental value) were used to aid the definition of structural categories of chemicals which are not non-polar narcotics. Doing so has enabled the domain for non-polar narcosis to be defined in terms of structural categories. Defining domains of applicability for QSAR models is important if they are to be considered for making predictions of toxicity for regulatory purposes. †Presented at the 13th International Workshop on QSARs in the Environmental Sciences (QSAR 2008), 8–12 June 2008, Syracuse, USA.

A review of the use of in silico methods to predict the chemistry of molecular initiating events related to drug toxicity

Introduction: Drug toxicity pathways can be extremely complex and difficult to fully understand. However, understanding specific parts of the pathway may be simpler. Every toxicity pathway starts with a molecular initiating event (MIE). If an MIE is well understood then it becomes possible to predict which compounds can partake in that particular MIE using in silico techniques. Areas covered: This review aims to describe how the use of structural alerts and the measurement/calculation of certain physicochemical properties can identify chemicals with a given MIE. For example, structural alerts can be used to identify chemicals able to form a covalent bond with a biological macromolecule. How chemistry-related MIEs relate to toxicity end points, such as hepatotoxicity, is also discussed. Expert opinion: It is emphasised that predicting that a compound can cause an MIE is not a direct prediction of toxicity. Predicting whether a compound will be toxic requires a comparison with similar compounds which cause the same MIE and that are associated with known toxicological data. It is possible to form categories of compounds that are all thought to act via the same MIE and then use read-across within the category to make a toxicity prediction.

Investigation of the Verhaar scheme for predicting acute aquatic toxicity: Improving predictions obtained from Toxtree ver. 2.6

Assessment of the potential of compounds to cause harm to the aquatic environment is an integral part of the REACH legislation. To reduce the number of vertebrate and invertebrate animals required for this analysis alternative approaches have been promoted. Category formation and read-across have been applied widely to predict toxicity. A key approach to grouping for environmental toxicity is the Verhaar scheme which uses rules to classify compounds into one of four mechanistic categories. These categories provide a mechanistic basis for grouping and any further predictive modelling. A computational implementation of the Verhaar scheme is available in Toxtree v2.6. The work presented herein demonstrates how modifications to the implementation of Verhaar between version 1.5 and 2.6 of Toxtree have improved performance by reducing the number of incorrectly classified compounds. However, for the datasets used in this analysis, version 2.6 classifies more compounds as outside of the domain of the model. Further amendments to the classification rules have been implemented here using a post-processing filter encoded as a KNIME workflow. This results in fewer compounds being classified as outside of the model domain, further improving the predictivity of the scheme. The utility of the modification described herein is demonstrated through building quality, mechanism-specific Quantitative Structure Activity Relationship (QSAR) models for the compounds within specific mechanistic categories.

Adverse Outcome Pathway (AOP) Informed Modeling of Aquatic Toxicology: QSARs, Read-Across, and Interspecies Verification of Modes of Action

Alternative approaches have been promoted to reduce the number of vertebrate and invertebrate animals required for the assessment of the potential of compounds to cause harm to the aquatic environment. A key philosophy in the development of alternatives is a greater understanding of the relevant adverse outcome pathway (AOP). One alternative method is the fish embryo toxicity (FET) assay. Although the trends in potency have been shown to be equivalent in embryo and adult assays, a detailed mechanistic analysis of the toxicity data has yet to be performed; such analysis is vital for a full understanding of the AOP. The research presented herein used an updated implementation of the Verhaar scheme to categorize compounds into AOP-informed categories. These were then used in mechanistic (quantitative) structure-activity relationship ((Q)SAR) analysis to show that the descriptors governing the distinct mechanisms of acute fish toxicity are capable of modeling data from the FET assay. The results show that compounds do appear to exhibit the same mechanisms of toxicity across life stages. Thus, this mechanistic analysis supports the argument that the FET assay is a suitable alternative testing strategy for the specified mechanisms and that understanding the AOPs is useful for toxicity prediction across test systems.

Using In Silico Tools in a Weight of Evidence Approach to Aid Toxicological Assessment

Integrated testing strategies are an important and useful approach to reduce animal usage in toxicity testing. Increased usage of integrated testing strategies is foreseen in current chemical legislation, e.g. REACH. Skin sensitisation is a well studied endpoint and many in silico models have been developed for the prediction of the skin sensitising potential of chemicals. This paper discusses the use of the OECD (Q)SAR Application Toolbox, Derek for Windows, the CAESAR global model and SMARTS rules for reactivity within a weight of evidence approach to predict skin sensitisation. Conclusions drawn from a weight of evidence approach can be used within an integrated testing strategy to reduce the requirement for in vivo tests. Using all four models in this manner enabled 76% of the conclusive predictions made regarding the test data to be in agreement with the observed toxicities. In addition, using all four models in conjunction identified areas where further information is required, as confounding results were produced. The actual data requirements for an integrated testing strategy are discussed along with what considerations need to be made for the remaining compounds that were misclassified or for which the programs contradicted one another and a definitive conclusion could not be reached.