"Identifying Anticipatory Regulation of E. coli Under Stress Conditions Using Publicly Available High-Throughput Data Sets"
Studies in both yeast and bacteria have observed that microorganisms under stress conditions regulate genes that are not directly related to the encountered stress. Many of these co-regulated genes are important for adaptation to stresses that frequently co-occur or follow the original cue stresses. This phenomenon of co-regulation of genes for expected stresses, which we refer to as anticipatory regulation, provides cells with evolutionarily maintained fitness advantages. Understanding anticipatory regulation can provide insights into the evolutionary wiring of regulatory pathways, and assist in designing antibiotic treatment strategies that exploit the regulatory patterns of target microbes.
To identify instances of anticipatory regulation, or cases where changes in response to a cue stress will affect fitness of microbes under a subsequent environmental stress, we connect gene expression profiles from cue stresses to phenotypic fitness landscapes of anticipated stresses. We combine this curated database of gene expression and metadata from COLOMBOS and the Gene Expression Omnibus (GEO) with phenotypic fitness scores of a knockout collection to predict anticipatory or sensitizing regulatory interactions.
Our analysis method properly enriches for apparent cross-protection between exposure to a given stress as a cue stimulus and subsequent growth under the same stimulus. We noticed different regulation patterns in stresses related to essential nutrients, e.g. starvation of carbon source, than stresses that are not related to essential nutrients. We have also identified several stress pairs that exhibit anticipatory or sensitizing regulation according to the above computational method, including anticipatory regulation involving anaerobic stress and certain antibiotics exposure, and sensitizing cases between different membrane permeabilizing agents. Experimental testing of these computationally predicted stress/response pairs is currently underway with preliminary data to be explored.