Key points are not available for this paper at this time.
Neurotoxicity is frequently observed in the global aquatic environment, threatening aquatic ecosystems and human health. However, a very limited proportion of neurotoxic effects (∼1%) has been explained by known chemicals of concern. Here, we integrated machine learning, nontargeted analysis, and in vitro biotesting to identify neurotoxic drivers of acetylcholinesterase (AChE) inhibition in estuarine waters along the coast of China. Machine learning was used to predict AChE inhibitors in a large chemical space. The prediction output was profiled into a suspect screening list to guide high-resolution mass spectrometry (HRMS) screening of AChE inhibitors in estuarine water samples. Ultimately, 60 chemicals with diverse known and presently unknown structures were identified, explaining 82.1% of the observed AChE inhibition. Polyunsaturated fatty acids were unexpectedly found to be neurotoxic drivers, accounting for 80.5% of the overall effect. This proof-of-concept study demonstrates that machine learning-based toxicological prediction can achieve a virtual fractionation role to pinpoint HRMS features with the bioactivity potential. Our approach is expected to enable rapid and comprehensive screening of organic pollutants associated with various in vitro end points for large-scale monitoring of water quality.
Wang et al. (Wed,) studied this question.