A Genome-Wide Association Analysis Reveals a Role for Recombination in the Evolution of Antimicrobial Resistance in Burkholderia multivorans

Abstract

Cystic fibrosis (CF) lung infections caused by members of the Burkholderia cepacia complex, such as Burkholderia multivorans, are associated with high rates of mortality and morbidity. We performed a population genomic study of 111 B. multivorans sputum isolates from a single CF patient through three stages of infection including the initial incident infection, deep sampling of a one-year period of chronic infection, and deep sampling of a post-transplant recolonization. We reconstructed the evolutionary history of the population and used a lineage-controlled genome-wide association study (GWAS) approach to identify genetic variants associated with antibiotic resistance. We found that the incident isolate was more susceptible to agents from three antimicrobial classes ({beta}-lactams, aminoglycosides, quinolones), while the chronic isolates diversified into distinct genetic lineages with reduced antimicrobial susceptibility to the same agents. The post-transplant reinfection isolates displayed genetic and phenotypic signatures that were distinct from sputum isolates from all CF lung specimens. There were numerous examples of parallel pathoadaptation, in which individual loci, or even the same codon, were independently mutated multiple times. This set of loci was enriched for functions associated with virulence and resistance. Our GWAS approach identified one variant in the ampD locus (which was independently mutated four times in our dataset) associated with resistance to {beta}-lactams, and two non-synonymous polymorphisms associated with resistance to both aminoglycosides and quinolones, affecting an araC family transcriptional regulator, which was independently mutated three times, and an outer member porin, which was independently mutated twice. We also performed recombination analysis and identified a minimum of 14 recombination events. Parallel pathoadaptive loci and polymorphisms associated with {beta}-lactam resistance were over-represented in these recombinogenic regions. This study illustrates the power of deep, longitudinal sampling coupled with evolutionary and lineage-corrected GWAS analyses to reveal how pathogens adapt to their hosts.