Computational new approach methods guide focused testing and enhance understanding of chlorantraniliprole toxicity across species.

Diamide insecticides, specifically chlorantraniliprole (CHL), have been rising in popularity over the past decade, becoming one of the most widely used insecticide classes globally. These insecticides target the ryanodine receptor (RyR), primarily for control of lepidopteran agricultural pests. Field studies have revealed that some lepidopteran species have developed mutations where a methionine in a particular position (e.g., I4790M) increases resistance to CHL. The toxicity data for CHL across species is limited, as is the case for many chemicals, which creates an opportunity to apply both traditional toxicity test methods and new approach methods (NAMs) to address data gaps. Here, the U.S. Environmental Protection Agency's Sequence Alignment to Predict Across Species Susceptibility (SeqAPASS) tool was used to query the RyR to generate susceptibility predictions for species exposed to CHL to fill those data gaps. These SeqAPASS results generated testable hypotheses that were used to guide focused acute aquatic toxicity studies using Daphnia magna, Daphnia pulex, Pimephales promelas, and Danio rerio. The fish species were not sensitive to CHL, whereas D. magna and D. pulex were found to be sensitive to CHL at environmentally relevant concentrations, despite having the methionine residue in the position of the I4790M resistance mutation. Additional SeqAPASS results showed that many other species, including beneficial pollinators and Lepidoptera, are predicted as likely susceptible to CHL. This study provided multiple lines of evidence toward the unlikelihood for the I4790M mutation to be the primary cause of resistance across species, filled knowledge gaps concerning CHL toxicity across species, and generated predictions of susceptibility for nontarget species that are not generally amenable to toxicity testing. This work presents a case example that demonstrates how NAMs can be used in combination with other types of data to direct targeted testing and build confidence in predictive approaches for their use in risk assessment.