Antibiotic resistance can evolve by mutations that change the amino acid sequence of a protein or by mutations that change the expression level of proteins. To explore the potential of changes in gene expression to confer antibiotic resistance, we developed a high-throughput assay to screen all viable gene over-expression or gene deletion mutants of E.coli against a panel of 31 antibiotics. We found 130 changes in gene expression, both positive and negative, that confer drug-specific or multi-drug resistance. These genes span a diverse range of functions and most were not previously associated with antibiotic resistance. By quantitatively adjusting gene expression and measuring resistance, we find that intrinsic antibiotic defense systems are often poorly deployed, and we identified ‘proto-resistance’ genes that confer little resistance in the wildtype strain despite enormous potential. We rationalize the abundance and diversity of hits by viewing gene-regulation as an optimization problem. As not all genes that influence drug susceptibility will respond optimally to drug exposure, there exist many possibilities for the evolution of drug resistance through regulatory mutations that deploy latent defense capabilities.

Curiously few drugs were resisted by the over-expression of their target genes, and so we measured drug resistance as drug target genes were increasingly over-expressed and observed that drug resistance can increase, decrease, or remain unchanged. We explain this counter-intuitive range of effects with mathematical models of enzyme inhibition that consider toxicity arising from gene over-expression and inhibition mechanisms that rely upon damaging an enzyme’s substrate, as is the case for sulfonamides and quinolones. These findings have important implications for identifying the molecular targets of novel pharmaceuticals: a common approach of screening for genes whose over-expression confers drug resistance could fail to identify the mechanisms of entire novel classes of drugs.