(217h) Parallel Mapping of Cryptic and Multidrug Resistance Alleles in Bacteria

Antibiotic resistance in microbes constitutes an impending crisis in clinical, agricultural, and industrial settings. Here we use genome engineering tools enabled by synthetic biology to examine factors underlying the development of antibiotic resistance in bacteria. A barcoded, genome-scale library of strains simulating overexpression or knockout of over 4,000 genes in E. coli (the TRMR library) was challenged in parallel at sub-inhibitory concentrations with eight different clinically-relevant antibiotics. The selected populations were rapidly characterized by next-generation sequencing and the approximately 200,000 individual fitness measurements were analyzed in the QIIME (Quantitative Insights into Microbial Ecology) bioinformatics pipeline. We discovered that the most prevalent alleles conferred resistance to multiple antibiotics (multi-drug resistant, or MDR alleles) and several of these isolated alleles were previously unknown or uncharacterized. Interestingly, chemically similar antibiotics or antibiotics with similar mechanisms of action did not produce similar allelic responses on the genome level. Also, in spite of the prevalence of MDR alleles, the depth of characterization enabled by next-generation sequencing allowed the use supervised learning to determine a genomic "fingerprint" for each antibiotic of interest. The same methodologies of rapid selection, high-throughput sequencing, and bioinformatic analysis are broadly applicable to experiments on chemical tolerance for any inhibitory chemical, from antibiotics to toxic metabolites to next-generation biofuels.