In Silico Modeling Method for Computational Aquatic Toxicology of Endocrine Disruptors: A Software-Based Approach Using QSAR Toolbox

Our protocol uses a highly accessible and open software program to enable rapid computation of short-term adverse effects of endocrine-disrupting chemicals. This OECD QSAR Toolbox is useful for predicting the toxicity of chemical substances and provides an efficient method for assessing acute ecotoxicity of endocrine disruptors in a user-friendly manner. Upon starting the QSAR Toolbox, the user begins at the Input Toolbox stage by default and the Toolbox automatically creates a working file named Document 1.

After clicking Chemical Abstract Service Number, enter the chemical abstract service number of the target substance in the available text field and click Search. Via the Structure button, the substance can also be searched by the simplified molecular input line entry system code. To perform the in-silico assay for multiple target substances, write a simple substance list in a text editor in which each chemical abstract service number is listed in a single row and save the text file with an appropriate name.

Next, click Data and confirm that databases listed under Ecotoxicological Information are checked. Click Input, select Query, and click Yes in the dialog window to accept the selected settings. Under the Chemical Abstract Service tab, select Load List to upload the substance list saved as a text file.

Click Add at the bottom of the pop-up menu and click Execute. The QSAR Toolbox will display a message indicating the number of substances that have been retrieved for the search. For profiling, click on the Profiling Toolbox stage button and navigate to Profiling Methods in the stage option panel.

Click Unselect All in the Profiling Methods window and confirm that all of the profilers listed under Predefined and profilers related to aquatic toxicity under Endpoint Specific are checked, then click Apply, right click Parameters to run integrated 2D and 3D QSAR models as desired. For data collection, click the Data Toolbox stage button and click Gather on the actions toolbar. Select All Endpoints to gather all of the experimental data or select Choose to gather the endpoint specific experimental data.

Right click the endpoint of interest and organize the endpoint tree in the preferred manner using the available terms and arrows before clicking Okay. Then right click the endpoint of interest and select Export Data Matrix to export the gathered data into a spreadsheet. A matrix export wizard will appear, enabling the addition of other endpoints to the export list.

Finish the selection and click Export to save the file. For data gap filling, click on the Data Gap Filling Toolbox stage button and click Automated. Select Ecotoxicological Information and Fish, lethal concentration at 50%at 96 hours and click Okay.

The workflow controller window will appear. If running Batch mode, specify the range of substances over which to execute the workflow. If the prediction was successfully executed, click the Report Toolbox stage button, left click on the resulting prediction value, click Prediction, and customize the report content and appearance in the pop-up wizard.

After clicking Next, click Create Report, and save the prediction and category reports as PDF files and the data matrix as a spreadsheet file. In this quantitative analysis of the predicted acute toxicities of selected endocrine disruptors, when the predicted data points were plotted versus the experimental data points as a log log scale a positive correlation between both parameters was found for all of the fish analyzed as well as for our representative species. To enable a safety assessment of the highest reliability, further computational analysis was performed by plotting the predicted lower limit of the 95%confidence interval of the lethal concentration at 50%versus the experimentally derived values.

In this evaluation with an elevated safety threshold, 92%of the total tested endocrine disrupting compounds were shown to fall into the protective range when compared to the experimentally derived values. Here the mean levels of the predicted lethal concentration at 50%values for the estrogen receptor binders were higher than those of the non-binders. By contrast, for the experimental lethal concentration at 50%values, three levels of the non and weak binders were higher than those of the stronger estrogen receptor binders.

This procedure can only be applied to chemical substances with a unique identifier, which in this case is the SMILES code. There's one other automated workflow integrated in the OECD QSAR Toolbox which can be applied similarly and is for skin sensitization.