Data-driven risk assessment of antibiotic mixtures against gram-negative and gram-positive bacteria: QSAR, joint toxic action, and component contribution

The increasing demand and use of antibiotics due to epidemic outbreaks under climate change leads to more antibacterial substances entering the environment. The residual drugs, commonly found as mixtures, can specially target bacteria, posing a threat to the ecosystem and human health. Therefore, it is necessary to assess the risk of antibiotic mixtures using bacteria as the model organisms. In this study, selecting sulfonamides, sulfonamides potentiators, and tetracyclines as the representative antibiotics, the individual and combined toxicity of these agents were tested against Gram-positive (Bacillus subtilis) and Gram-negative bacteria (Aliivibrio fischeri and Escherichia coli). The quantitative structure–activity relationship models were constructed by setting E_bind (lowest interaction energy between antibiotic and target protein) and K_ow (octanol–water partition coefficient) as the structural descriptors, which provide reliable and robust tools for predicting the toxicity of single agents and binary mixtures. Furthermore, the hormetic effects, characterized by low-dose stimulation and high-dose inhibition, were observed in Gram-negative bacteria, and the dose-responses in Gram-positive bacteria all exhibited S-shaped. Within the hormetic phenomena of antibiotic mixtures, the pattern of joint toxic action and component contribution generally changed with the transition from stimulatory to inhibitory actions. These results not only demonstrate the influence of bacterial species and test endpoints on the toxicity of antibiotics, but also clarify the pivotal role of hormesis in the joint effects of antibiotics. This study provides a data-driven methodology for evaluating the combined toxicity of mixtures, which will promote the development of the risk assessment of antibiotics and other environmental pollutants.