Computational Genomics Studies of Genetic Adaptation in Different Environmental and Organismal Contexts
This dissertation presents a study of global patterns in the distribution of acquired resistance genes in 352 draft genomes of E. coli samples from two US West Coast hospitals. A nondeterministic clustering of genomes based on their resistance gene composition identifies two highly successful gene combinations involving beta-lactamase and aminoglycoside acetyltransferase genes that largely explain the distribution of ESBL, gentamicin, and tobramycin resistance in these samples. We name these two parallel adaptive solutions “Complementarity Groups 1 and 2” (CG1 and CG2) because we observe functional diversification of genes within these groups, driven by mutual antagonism between genes exhibiting similar resistance profiles. Mutual antagonism extending across groups drives parallel adaptive trajectories. In 761 completely assembled genomes from NCBI, representing a broader range of geographical and ecological sources, we confirm: (1) the prevalence of CG1 and CG2; (2) establish that mutual antagonism is a generalized feature of acquired resistance genes with overlapping function; (3) and verify that the observed gene-to-gene associations correspond to physical linkages. We also find that configurations placing gene pairs in high proximity and on the same strand tend to be more successful. We propose a model that explains these observations, constraining evolution through a combination of physical linkage and mutual antagonism in the context of generalized panmixia. Looking at the genomic context for antibiotic resistance genes in NCBI761, we find mosaic plasmids (with replicons belonging to different incompatibility groups), a complex network of linkages between replicons and resistance (dominated by IncF), and significant gene flow to the chromosome, particularly for ESBLs.
Also, I describe a deep mutational scanning approach for directed evolution of proteins, and the generation of TEM beta-lactamase mutant libraries using this approach. Each mutant library captures alternate sequence subspaces in the evolution of extended-spectrum resistance (a gain-of-function), and are generated by leveraging negative epistasis between their respective starting points for directed evolution. In the future the dataset generated by this approach will enable the study of higher order mutational interactions in the evolution of extended-spectrum resistance.
In addition, two other previous works are presented: (1) A comparison of the repeat landscapes in the genomes of 8 ant species highlights the role of transposable element clusters (TE islands) in facilitating the adaptation of an invasive species to new habitats. (2) A comparison of the genomes of three marine Planctomycetes inhabiting the blade of the red alga, Porphyra umbilicalis. These three OTUs represent three different genera, and contain large expansions of specific gene families and horizontally acquired genes, which appear to augment their metabolic repertoire for accessing macropolymers in the cell walls of algae, and their mechanisms for stress responses that likely help adaptation to the intertidal zone.