Aynur O. Aptula

Quantitative structure–activity relationships (QSARs) in toxicology: a historical perspective

Abstract The history of the use of quantitative structure–activity relationships (QSARs) in toxicology, both for environmental, and human health effects is described. A particular emphasis is made on the science in response to the United States Toxic Substance Control Act of 1976. Specifically, the basic concepts and objectives of QSARs for toxicity are reviewed. QSARs for environmental and human health effects are discussed separately. Environmental, and more specifically, ecotoxicity, QSARs have focused historically on modeling congeneric series and non-specific effects in aquatic organisms through the use of the logarithm of the 1-octanol/water partition coefficient to describe hydrophobicity, and hence uptake. Compounds that do not fit these QSARs (namely the outliers) have been explained by differences in mechanism of acute toxicity, especially as a result of electro(nucleo)philic interactions. In light of this, mechanisms of acute toxicity are discussed. QSAR approaches to receptor-mediated effects, such as those exhibited by environmental estrogens, and competitive binding to the estrogen receptor, are different from those typically applied to model acute toxic endpoints. Several of these approaches, including three-dimensional QSAR techniques, are reviewed. Human health effects include both local and systemic effects. Local effects (e.g. corrosivity and skin sensitization) are often modeled by multivariate QSAR methods such as linear regression and discriminant analysis. The prediction of systemic effects such as mutagenesis and carcinogenesis requires consideration of the endpoint and a more mechanistic basis for modeling. Approaches to predict these endpoints include the use of expert systems.

Local lymph node data for the evaluation of skin sensitization alternatives: a second compilation.

Background: Development, evaluation and validation of alternatives to skin sensitisation testing require the availability of reliable databases with which comparative analyses can be conducted to establish performance characteristics. To facilitate this we have published previously a database comprising results from local lymph node assays (LLNAs) conducted with 211 chemicals. That database embraced a substantial range of chemistry, and of relative skin sensitising potency, and has found application in the assessment of new or refined methods. Objective: In this paper we describe a second compilation to extend the LLNA database. Methods: This second data compilation was derived from previously conducted LLNA studies involving an additional 108 chemicals. In addition, the first database contained a small number of inaccuracies, affecting results recorded with a few chemicals. In this paper these have been corrected. Results: The inclusion of 108 new substances has served to extend and consolidate the areas of chemistry covered by the database. In addition, the entire dataset was evaluated for pre and prohaptens which will facilitate the choice of chemicals for alternative assay developments. Conclusions: It is anticipated that the new revised and extended database totalling over 300 chemicals will now serve as the primary resource to support the development and evaluation of new approaches to hazard identification and potency assessment.

Multivariate Discrimination between Modes of Toxic Action of Phenols

A set of 221 phenols, for which toxicity data to the ciliate Tetrahymena pyriformis were available, was subjected to stepwise linear discriminant analysis (LDA) in order to classify their toxic mechanisms of action. The compounds were a priori grouped into the following four mechanisms according to structural rules: polar narcotics, weak acid respiratory uncouplers, pro-electrophiles and soft electrophiles. Hydrophobicity with and without correction for ionisation (log K o w , log D o w u), acidity constant (pK a ), frontier orbital energies (E L U M O , E H O M O ) and hydrogenbond donor and acceptor counts were used as molecular descriptors. LDA models employing 3-6 variables achieved 86-89% overall correct classification of the four mechanisms, with more varied performance for respiratory uncouplers and pro-electrophiles. For the latter, a separate model was developed that discriminated compounds undergoing metabolic activation from compounds with different mechanisms very accurately. Model validation was performed by evaluating the simulated external prediction through LDA models built from complementary subsets.

Reactivity Profiling: Covalent Modification of Single Nucleophile Peptides for Skin Sensitization Risk Assessment

The molecular basis of chemical allergy is rooted in the ability of an allergen (hapten) to modify endogenous proteins. This mechanistic understanding aided development of screening assays which generate reproducible quantitative and qualitative reactivity data. Such assays use model peptides with a limited number and type of protein nucleophiles, and the data does not reflect the specificity, variety, and complexity of hapten interactions with multiple nucleophiles. Building on these developments, we extended the standardized approach to maximize the type and the amount of information that can be derived from an in chemico assay. We used a panel of six single nucleophile peptides and individually optimized the incubation conditions to favor chemical modification. Employing liquid chromatography tandem mass spectrometry (LC-MS/MS) technique, we simultaneously obtained multiple quantitative and qualitative measurements (% peptide depletion, adducts formation, and peptide dimerization for Cys-containing peptide). Using these methods, we obtained reactivity data for 36 chemicals of known skin sensitizing potency. By optimizing incubation conditions, we ensured detection of all reactive chemicals. We explored the LC-MS/MS approach to generate kinetic data for 10 chemicals allowing further characterization of reactivity and a potentially more robust quantitative reactivity descriptor. Our ultimate aim is to integrate this dataset with available physicochemical data and outputs from other predictive assays, all addressing different key steps in the induction of sensitization, to help us make decisions about the safe use of chemicals without using animal tests. The epidermal protein target sites, modification of which may be immunogenic and lead to induction of skin sensitization, are currently unknown. Increasing the understanding of this process may help further refine in chemico reactivity assays as well as aid the interpretation of the reactivity data.

An evaluation of selected global (Q)SARs/expert systems for the prediction of skin sensitisation potential

Skin sensitisation potential is an endpoint that needs to be assessed within the framework of existing and forthcoming legislation. At present, skin sensitisation hazard is normally identified using in vivo test methods, the favoured approach being the local lymph node assay (LLNA). This method can also provide a measure of relative skin sensitising potency which is essential for assessing and managing human health risks. One potential alternative approach to skin sensitisation hazard identification is the use of (Quantitative) structure activity relationships ((Q)SARs) coupled with appropriate documentation and performance characteristics. This represents a major challenge. Current thinking is that (Q)SARs might best be employed as part of a battery of approaches that collectively provide information on skin sensitisation hazard. A number of (Q)SARs and expert systems have been developed and are described in the literature. Here we focus on three models (TOPKAT, Derek for Windows and TOPS-MODE), and evaluate their performance against a recently published dataset of 211 chemicals. The current strengths and limitations of one of these models is highlighted, together with modifications that could be made to improve its performance. Of the models/expert systems evaluated, none performed sufficiently well to act as a standalone tool for hazard identification.

Prediction of hERG K+ blocking potency: Application of structural knowledge

Modelling of QT-prolongation has been performed using data for 19 structurally diverse hERG K+ channel blocking drugs taken from literature. The modelling used hydrophobicity corrected for ionisation (log D) and various 2D and 3D physico-chemical molecular descriptors. Stepwise regression produced a two parameter, interpretable and transparent QSAR with good statistical fit, including log D and the maximum diameter of molecules (D max). Two strategies were applied for model validation: (i) a scrambling procedure, i.e., training the total set of 19 chemicals after randomising the hERG K+ channel blocking activity data and (ii) use of external validation sets. Validation of the models showed them to be stable and statistically significant. The effect of molecular size on QT-prolongation side effect is discussed.

Read-across to rank skin sensitization potential: subcategories for the Michael acceptor domain.

Background:  Eliminating animal testing for skin sensitization is a significant challenge in consumer safety risk assessment. To be able to perform resilient risk assessments in the future, one will need alternative approaches to fill the data gaps.

Global (Q)SARs for skin sensitisation–assessment against OECD principles‖

As part of a European Chemicals Bureau contract relating to the evaluation of (Q)SARs for toxicological endpoints of regulatory importance, we have reviewed and analysed (Q)SARs for skin sensitisation. Here we consider some recently published global (Q)SAR approaches against the OECD principles and present re-analysis of the data. Our analyses indicate that “statistical” (Q)SARs which aim to be global in their applicability tend to be insufficiently robust mechanistically, leading to an unacceptably high failure rate. Our conclusions are that, for skin sensitisation, the mechanistic chemistry is very important and consequently the best non-animal approach currently applicable to predict skin sensitisation potential is with the help of an expert system. This would assign compounds into mechanistic applicability domains and apply mechanism-based (Q)SARs specific for those domains and, very importantly, recognise when a compound is outside its range of competence. In such situations, it would call for human expert input supported by experimental chemistry studies as necessary. ‖Presented at the 12th International Workshop on Quantitative Structure-Activity Relationships in Environmental Toxicology (QSAR2006), 8–12 May 2006, Lyon, France.

Experimental reactivity parameters for toxicity modeling: application to the acute aquatic toxicity of SN2 electrophiles to Tetrahymena pyriformis.

A diverse set of 60 haloaliphatic compounds were evaluated for reactivity with cysteine thiol groups in the previously described RC(50) assay using glutathione (GSH) as a model nucleophile. Reactivity was quantified by the RC(50) value, the concentration of test compound that produced 50% reaction of the GSH thiol groups in 120 min. Under standard conditions, RC(50) values are mathematically proportional to reciprocal rate constants. Quantitative structure-activity relationship (QSAR) analysis correlating acute aquatic toxicity (IGC(50)) to Tetrahymena pyriformis with RC(50) values was carried out. It was found that subdivision of the compounds into subdomains according to their reaction mechanism characteristics enabled toxicity-reactivity relationships to be identified. The largest subdomain consisting of 22 compounds in which a primary halogen is alpha to a carbonyl or other electronegative unsaturated group and which can be confidently assigned as S(N)2 electrophiles fits the equation pIGC(50) (mM) = 0.94 (+/-0.07) pRC(50) (mM) + 1.34 (+/-0.07), n = 22, r(2) = 0.889, r(2)(adj) = 0.884, s = 0.27, and F = 161. Compounds in which the halogen is not alpha to an unsaturated group are not reactive in the GSH assay and do not exhibit reactive toxicity to T. pyriformis. Compounds tested in which the halogen is alpha to an unsaturated nonelectronegative group were found to be less toxic in the assay than predicted by the above QSAR equation. Within a subdomain of 21 compounds having a halogen alpha to an electronegative unsaturated group that, in the absence of experimental evidence, could not be confidently assigned as S(N)2 electrophiles, 2-bromoalkanoates of general structure R(1)CHBrCO(2)R(2), 2-bromopropionamide, and 2-haloalkanoic acids of general formula R(1)CHXCO(2)H (nine compounds in total) are all well-predicted by the above equation. Of the other 12 compounds of this subdomain, eight are substantially less toxic than predicted by the above equation and are considered to react differently, whereas the alpha-halonitriles (four compounds) are more toxic than predicted and fit a correlation of their own: pIGC(50) = 1.01 (+/-0.05) pRC(50) + 2.04 (+/-0.05), n = 4, r(2) = 0.995, r(2)(adj) = 0.992, s = 0.08, and F = 381, with a similar slope but larger intercept. An explanation in terms of their physical chemistry and possible involvement of released cyanide ion is suggested.

Structure-activity relationships for abiotic thiol reactivity and aquatic toxicity of halo-substituted carbonyl compounds

Using abiotic thiol reactivity (EC50) and Tetrahymena pyriformis toxicity (IGC50) data for a group of halo-substituted ketones, esters and amides (i.e. SN2 electrophiles) and related compounds a series of structure–activity relationships are illustrated. Only the α-halo-carbonyl-containing compounds are observed to be thiol reactive with the order I > Br > Cl > F. Further comparisons disclose α-halo-carbonyl compounds to be more reactive than non-α-halo-carbonyl compounds; in addition, the reactivity is reduced when the number of C atoms between the carbonyl and halogen is greater than one. Comparing reactivity among α-halo-carbonyl-containing compounds with different β-alkyl groups shows the greater the size of the β-alkyl group the lesser the reactivity. A comparison of reactivity data for 2-bromoacetyl-containing compounds of differing dimensions reveals little difference in reactivity. Regression analysis demonstrates a linear relationship between toxicity and thiol reactivity: ; n = 19, s = 0.250, r 2 = 0.926, r 2(pred) = 0.905, F = 199, Pr > F = 0.0001. ‖Presented at the 12th International Workshop on Quantitative Structure--Activity Relationships in Environmental Toxicology (QSAR2006), 8--12 May 2006, Lyon, France.