Studies in Bacterial Genome Dynamics

Abstract

Bacteria are among the most genetically dynamic organisms on Earth, continuously reshaping their genomes through gene gain, loss, and modification. Understanding how new genes arise, how they spread, and how selective pressures like antibiotic use reshape this diversity remain central questions in bacteriology. This thesis investigates these questions through three complementary studies unified by a shared interest in the fluid dynamics of bacterial genomes. First, we explore a mechanism by which the streamlining bias that trims bacterial genomes can also serve as a source of evolutionary innovation. We propose and characterize a model of “deletion- born fusion genes,” in which adaptive deletions fuse distant gene fragments into novel open reading frames. Unlike other gene birth mechanisms that begin with rare, neutral intermediates, these fusions reach high frequency by hitchhiking on the selective advantage of the deletion itself. We document examples in the Lenski Long-Term Evolution Experiment and in the Mycobacterium tuberculosis– bovis divergence. Finally, we develop a scalable screen to efficiently detect these genes across multi- million bacterial genome collections. Second, we collate 1,817 high-quality genomes from the British National Collection of Type Cul- tures, spanning isolates collected from 1885 to the present, and trace how human antibiotic use re- shaped the frequency and mobility of genomic resistance. We find that functional resistance genes cir- culated in clinically relevant isolates before the age of antibiotics, but were generally rare and chromo- somally encoded. Following the clinical introduction of specific antibiotics, corresponding genomic resistance significantly increased in frequency and became progressively associated with multiple mo- bile genetic elements. These findings suggest that anthropogenic use of antibiotics did not generate resistance de novo, but both amplified and refined pre-existing genetic potential. Third, we introduce the phylogeny-colored de Bruijn graph (pcDBG), a data structure that re- colors each unitig in a compacted pangenome graph with the phylogenetic branch where its pres- ence/absence pattern most parsimoniously changed state. Applied to over 1,000 Staphylococcus aureus ST8 genomes, we show that co-inherited sequence blocks extend roughly one kilobase before decay- ing, and that mapping evolutionary signals onto the USA300 reference recovers the known mobile element landscape without prior annotation. The pcDBG provides a general framework for charac- terizing the spatial organization of evolutionary history across bacterial pangenomes. Together, these studies suggest a view of bacterial genomes as dynamic mosaics, continually being reshaped both by their environment and the pressures we impose on them. The genomes we sequence today are not endpoints, but fleeting snapshots of a continuous evolutionary process that long pre- dates humanity and will long outlast us.