Computational Toxicology最新文献

We present machine learning studies devoted to the creation of predictive models for toxicity evaluation of imidazolium- and pyridinium-containing ionic liquids. New created predictive models were developed using the OCHEM. The predictive ability of the models was tested by cross-validation, giving a coefficient of determination q² = 0.77–0.82. The models were applied to screen a virtual chemical library to the toxicity of ILs in Danio rerio and Daphnia magna bioassays. Models were used to predict toxicity for 25 ILs, which were then synthesized and tested in vivo. The in vivo toxicity studies found that D. magna is a more sensitive aquatic test organism than D. rerio – 67 % of the studied ILs are classified as extremely toxic with an LC₅₀ range from 0.005 to 0.01 mg/l. At the same time, only one IL 1-dodecylpyridinium bromide with an LC₅₀ of 0.08 mg/l is classified as extremely toxic, and 76 % are classified as slightly and moderately toxic compounds using D. rerio as a test organism. The most toxic ILs 5 and 19 were docked into the human AChE active center and demonstrated calculated binding energy values −9.5 and −9.3 kcal/mol that is comparable with the complexation of the human AChE inhibitor Donepezil, which provides insight into the potential molecular mechanisms of ILs toxicity. The created QSTR models are a successful tool for the toxicity analysis of new promising ILs. QSTR models demonstrated not only high predictive indicators but also a high percentage of correctly predicted toxicity values in vivo studies.

Adverse Outcome Pathways (AOPs) provide a basis for non-animal testing, by outlining the cascade of molecular and cellular events initiated upon stressor exposure, leading to adverse effects. In recent years, the scientific community has shown interest in developing AOPs through crowdsourcing, with the results archived in the AOP-Wiki: a centralized repository coordinated by the OECD, hosting nearly 512 AOPs (April, 2023). However, the AOP-Wiki platform currently lacks a versatile querying system, which hinders developers' exploration of the AOP network and impedes its practical use in risk assessment. This work proposes to unleash the full potential of the AOP-Wiki archive by adapting its data into a Labelled Property Graph (LPG) schema. Additionally, the tool offers a visual network query interface for both database-specific and natural language queries, facilitating the retrieval and analysis of graph data. The multi-query interface allows non-technical users to construct flexible queries, thereby enhancing the potential for AOP exploration. By reducing the time and technical requirements, the present query engine enhances the practical utilization of the valuable data within AOP-Wiki. To evaluate the platform, a case study is presented with three levels of use-case scenarios (simple, moderate, and complex queries). AOPWIKI-EXPLORER is freely available on GitHub (https://github.com/Crispae/AOPWiki_Explorer) for wider community reach and further enhancement.

Sequencing depth and biological replication represent key experimental design considerations in toxicogenomics and risk assessment. However, their relative impacts on differential gene expression analysis remain unclear. Using an 8-dose chemical (Prochloraz) perturbation RNA-seq dataset in A549 cells, we systematically subsampled sequencing depth (5–100 %) and replicates (2–4) to evaluate effects on number of differentially expressed genes. While dose was the primary variance driver, replication had a greater influence than depth for optimizing detection power. With only 2 replicates, over 80% of the ∼2000 differential genes were unique to specific depths, indicating high variability. Increasing to 4 replicates substantially improved reproducibility, with over 550 genes consistently identified across most depths, representing 30% of the total differential genes. Higher replicates also increased the rate of overlap of benchmark dose pathways and precision of median benchmark dose estimates. However, key gene ontology pathways related to DNA replication, cell cycle, and division were consistently captured even at lower replicates. Thus, replication enhanced confidence but did not fundamentally expand biological findings. Our study delineates key trade-offs between sequencing depth and replication for toxicogenomic experimental design. While additional replicates fundamentally improve reproducibility, gains from depth exhibit diminishing returns. Prioritizing biological replication over depth provides a cost-effective approach to enhance interpretation without sacrificing detection of core gene expression patterns. Altogether, this study provides important insights into the experimental design of toxicogenomics experiments.

Advances in the development and application of in silico models in toxicology has been recognised by two OECD guidance documents (69: Guidance Document On The Validation Of (Quantitative) Structure-Activity Relationship [(Q)SAR] Models and 386: (Q)SAR Assessment Framework: Guidance for the regulatory assessment of (Q)SAR models, predictions, and results based on multiple predictions) published in 2007 and 2023 respectively. The former outlines criteria for appropriate model validation, whilst the latter provides guidance around assessing predictions derived from them. The concepts and criteria described within these guidelines have been used to establish a framework to support both model builders and those applying them to support regulatory decisions. Herein we demonstrate how to meet those criteria and propose where further guidance is essential for ensuring the consistent, confident, and safe application of in silico models in support of regulatory decisions.

Glyphosate is a widely used herbicide known for its effectiveness in weed control; and it is an inhibitor of the plant enzyme 5-enolpyruvylshikimate-3-phosphate synthase. Currently, it is one of the most extensively used non-specific herbicides in agroindustry. However, toxic effects of glyphosate have recently been reported, including endocrine disruption, metabolic alterations, teratogenic, tumorigenic, and hepatorenal effects. Additionally, there are environmental concerns related to possible interactions with proteins from microorganisms, aquatic organisms, and mammals.

Research on the description of these interactions has gained interest, primarily with the aim of generating recommendations in terms of its use and possible regulations. On the other hand, computational methods have emerged to identify potential targets or unintended targets among numerous possible receptors. Several programs, online services, and databases are available for use in these methods.

In this study, we employed a set of online tools for computational target fishing to identify receptors of glyphosate. A set of thirteen targets were selected using six fishing tools. Furthermore, docking procedures were performed to investigate the expected interactions and binding energies. Certain associations with diseases are also reported.

Read-across is a well-established data-gap filling technique used within analogue or category approaches. Acceptance remains an issue, mainly due to the difficulties of addressing residual uncertainties associated with a read-across prediction and because assessments are expert-driven. Frameworks to develop, assess and document read-across may help reduce variability in read-across results. Data-driven read-across approaches such as Generalised Read-Across (GenRA) include quantification of uncertainties and performance. GenRA also offers opportunities on how New Approach Method (NAM) data can be systematically incorporated to support the read-across hypothesis. Herein, a systematic investigation of differences in expert-driven read-across with data-driven approaches was pursued in terms of building scientific confidence in the use of read-across. A dataset of expert-driven read-across assessments that made use of registration data as disseminated in the public International Uniform Chemical Information Database (IUCLID) (version 6) of Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) Study Results were compiled. A dataset of ∼5000 read-across cases pertaining to repeated dose and developmental toxicity was extracted and mapped to content within EPA’s Distributed Structure Searchable Toxicity database (DSSTox) to retrieve chemical name and structural identification information. Content could be mapped to ∼3600 cases which when filtered for unique cases with curated quantitative structure–activity relationship-ready SMILES resulted in 389 target-source analogue pairs. The similarity between target and the source analogues on the basis of different contexts – from structural similarity using chemical fingerprints to metabolic similarity using predicted metabolic information was evaluated. An attempt was also made to quantify the relative contribution each similarity context played relative to the target-source analogue pairs by deriving a model which predicted known analogue pairs. Finally, point of departure values (PODs) were predicted using the GenRA approach underpinned by data extracted from the EPA’s Toxicity Values Database (ToxValDB). The GenRA predicted PODs were compared with those reported within the REACH dossiers themselves. This study offers generalisable insights on how read-across is already applied for regulatory submissions and expectations on the levels of similarity necessary to make decisions.

Bacterial infections caused by resistant strains, especially those conferring multi-drug resistance (MDR), have become a severe health problem worldwide. Novobiocin (NB) is a widely used antibiotic that inhibits the action of DNA gyrase in Escherichia coli (E. coli). The drug's efficiency is hindered by its strong binding with the resistance causing efflux pump AcrAB-TolC on recurrent exposure. Consequently, the discovery of alternate/substitute analogue compounds for the parent drug with higher selectivity could counter drug resistance. In this work, we identified potent analogues of drug NB against the gyrase B enzyme by performing high throughput virtual screening of forty analogues that includes drug-likeness properties, pharmacokinetic parameters analysis, molecular docking, and molecular dynamics (MD) simulations. Our comprehensive pharmacological profiling with intrinsic analysis of selectivity and safety resulted in the identification of four potential compounds, C4 (ZINC218812366), C6 (ZINC221968665), C8 (ZINC49783724) and C10 (ZINC49783727), have better inhibitory and binding capacity against the primary target gyrase B subunit and reduced interaction with the counterpart of resistant target AcrB. These findings provide proof of concept for developing lead compounds targeting gyrase B and help in combatting AcrB-mediated drug resistance.

Animal toxicity testing is time and resource intensive, making it difficult to keep pace with the number of substances requiring assessment. Machine learning (ML) models that use chemical structure information and high-throughput experimental data can be helpful in predicting potential toxicity. However, much of the toxicity data used to train ML models is biased with an unequal balance of positives and negatives primarily since substances selected for in vivo testing are expected to elicit some toxicity effect. To investigate the impact this bias had on predictive performance, various sampling approaches were used to balance in vivo toxicity data as part of a supervised ML workflow to predict hepatotoxicity outcomes from chemical structure and/or targeted transcriptomic data. From the chronic, subchronic, developmental, multigenerational reproductive, and subacute repeat-dose testing toxicity outcomes with a minimum of 50 positive and 50 negative substances, 18 different study-toxicity outcome combinations were evaluated in up to 7 ML models. These included Artificial Neural Networks, Random Forests, Bernouilli Naïve Bayes, Gradient Boosting, and Support Vector classification algorithms which were compared with a local approach, Generalised Read-Across (GenRA), a similarity-weighted k-Nearest Neighbour (k-NN) method. The mean CV F1 performance for unbalanced data across all classifiers and descriptors for chronic liver effects was 0.735 (0.0395 SD). Mean CV F1 performance dropped to 0.639 (0.073 SD) with over-sampling approaches though the poorer performance of KNN approaches in some cases contributed to the observed decrease (mean CV F1 performance excluding KNN was 0.697 (0.072 SD)). With under-sampling approaches, the mean CV F1 was 0.523 (0.083 SD). For developmental liver effects, the mean CV F1 performance was much lower with 0.089 (0.111 SD) for unbalanced approaches and 0.149 (0.084 SD) for under-sampling. Over-sampling approaches led to an increase in mean CV F1 performance (0.234, (0.107 SD)) for developmental liver toxicity. Model performance was found to be dependent on dataset, model type, balancing approach and feature selection. Accordingly tailoring ML workflows for predicting toxicity should consider class imbalance and rely on simple classifiers first.

To reach conclusions during chemical safety assessments, risk assessors need to ensure sufficient information is present to satisfy the decision criteria. This often requires data to be generated and, in some cases, insufficient knowledge is present, or it is not feasible to generate new data through experiments. Read-across is a powerful technique to fill such data gaps, however the expert-driven process can be time intensive and subjective in nature resulting in variation of approach. To overcome these barriers a prototype software application has been developed by Lhasa Limited to support decision making about the toxicity and potency of chemicals using a read-across approach. The application supports a workflow which allows the user to gather data and knowledge about a chemical of interest and possible read-across candidates. Relevant information is then presented that enables the user to decide if read-across can be performed and, if so, which analogue or category can be considered the most appropriate. Data and knowledge about the toxicity of a compound and potential analogues include assay and metabolism data, toxicophore identification and its local similarity, physico-chemical and pharmacokinetic properties and observed and predicted metabolic profile. The utility of the approach is demonstrated with case studies using N-nitrosamine compounds, where the conclusions from using the workflow supported by the software are concordant with the evidence base. The components of the workflow have been further validated by demonstrating that conclusions are significantly better than would be expect from the distribution of data in test sets. The approach taken demonstrates how software implementing intuitive workflows that guide experts during read-across can support decisions and how validation of the methods can increase confidence in the overall approach.

This paper presents an ordinary differential equation (ODE) model of endogenous H₂O₂ metabolism in hepatocytes that is unique, at the time of writing, in its ability to accurately compute intracellular H₂O₂ concentration during incidents of oxidative stress and in its usefulness for constructing PBPK/PD models for ROS-generating xenobiotics. Versions of the model are presented for rat hepatocytes in vitro and mouse liver in vivo. A generic method is given for using the model to create PBPK/PD models which predict intracellular H₂O₂ concentration and oxidative-stress-induced hepatocyte death; these are identifiable from in vitro data sets reporting cell mortality following xenobiotic exposure at various levels. The procedure is demonstrated for the trivalent arsenical dimethylarsinous acid (DMA^III), which is produced in liver as part of the arsenic elimination pathway. This is the first model of H₂O₂ metabolism in hepatocytes to feature values for the endogenous rates of H₂O₂ production by mitochondria and other organelles which are inferred from the physiology literature, and to feature a detailed, realistic treatment of GSH metabolism; the latter is achieved by incorporating a minimal version of Reed and coworkers’ pioneering model of GSH metabolism in liver. Model simulations indicate that critical GSH depletion is the immediate trigger for intracellular H₂O₂ rising to concentrations associated with apoptosis ($>1\mu M$), that this may only occur hours after the xenobiotic concentration peaks (“delay effect”), that when critical GSH depletion does occur, H₂O₂ concentration rises rapidly in a sequence of two boundary layers, characterized by the kinetics of glutathione peroxidase (first boundary layer) and catalase (second boundary layer), and that intracellular H₂O₂ concentration $>1\mu M$ implies critical GSH depletion. There has been speculation that ROS levels in the range associated with apoptosis simply indicate, rather than cause, an apoptotic milieu. Model simulations are consistent with this view. In a result of interest to the wider physiology community, the delay effect is shown to provide a GSH-based mechanism by which cells can distinguish transient elevations in H₂O₂ concentration, of use in intracellular signaling, from persistent ones indicative of either pathology or the presence of toxins, the second state of affairs eventually triggering apoptosis.