ExExE interactions in the transcriptomic response of E. coli exposed to a combination of antibiotic, temperature and salinity stresses

Beyond individual effects, environmental variables can interact non-additively, thereby reducing our ability to predict demographic responses under multiple simultaneous stressors. The role of gene expression-level interactions remains poorly understood, with most studies limiting their analyses to stressors in isolation, neglecting potential interactions. What is the distribution of interaction between stressors at the gene expression level? do these interactions generally amplify or buffer the individual effects of stressors? Can additional stressors restore expression levels affected by one single stress? Can interactions emerge de novo in genes that are not responsive to individual stressors? And do higher-order interactions also occur at the transcriptomic level? We addressed these questions in the context of a major environmental issue: antibiotic pollution. Third-order interactions have been reported among azithromycin (AZI), temperature, and salinity, and altered by up to 100-fold the antibiotic concentration required to halve bacterial growth rate or yield. We show that most interaction emerge de novo or are antagonistic, pushing expression back to its baseline level. A metabolic switch between carbon and nitrogen and an enhanced general stress response can explain the interactions observed at the fitness level. Overall design: Transcriptoimic analysis (mRNA) of 24 isogenic populations of Esherichia coli K12 MG1665 grown in BHI broth from an initial density around 10^8 and exposed to all combination of two concentrations of atythromycin (0 vs. 1 µg/mL), two temperature (30 and 25 °C) and two salinities (0.085 and 0.585 M), for 6 hours.