Date of Award
Doctor of Philosophy (PhD)
Complex microbial communities colonize every habitat investigated to date, including soil, animals, water, and humans, as well as the structures we live in. It has been hypothesized that there is network of exchange allowing both bacterial organisms and functions to seamlessly move between and within these environments. These microbial communities often serve essential and beneficial functions for the host organism or environment. This is particularly evident in the human gut, where microbial communities consistently provide a set of services to its human host, including protecting against enteric pathogens, liberating nutrients from food, and signaling immune system regulation. However, these communities can also serve as reservoirs of antibiotic resistance (AR) genes readily available for transfer to pathogenic bacteria, compromising treatment of infectious disease. The goal of my thesis was to (1) understand how AR genes move through this potential network of exchange between environments and (2) how therapeutic levels of antibiotics effect a dynamically developing microbial community and its encoded reservoir of AR genes within the human gut environment. In order to accomplish this goal, I developed and optimized two computational methods, Resfams and ShortBRED, to study antibiotic resistance in diverse and dynamic microbial communities using high-throughput shotgun sequencing methods. I employed Resfams, an AR-specific profile hidden Markov model (pHMM) database, to demonstrate
that despite the apparent environmental source of clinical AR genes, the AR functions encoded in soil microbial communities are significantly distinct from clinical, human-associated microbial communities. I was also able to show that using a consensus and probabilistic model approach, like Resfams, significantly reduces bias when studying diverse microbial communities over the current best practice of pairwise-sequence alignment (BLAST). To study how therapeutic levels of specific antibiotics modulate microbial communities and encoded AR functions, I studied the developing preterm infant gut microbiota, a population which almost universally receives early and prolonged antibiotic therapy at birth. By analyzing 401 fecal samples from 84 longitudinally-sampled preterm
infants, I show that meropenem, cefotaxime, and ticarcillin-clavulanate significantly reduce species richness, and significantly enrich or deplete specific bacterial species and AR genes. In contrast, vancomycin and gentamicin, two of the antibiotics most commonly administered to preterm infants, result in a varied response in species richness. I show this response is predictable with 85% accuracy based on the relative abundance of only two bacterial species and two AR genes prior to treatment. Further, I show that AR genes enriched following specific antibiotic treatments are generally unique to the specific treatment and highly correlated with the abundance of a single bacterial species. In addition to AR genes relevant to specific antibiotics likely providing a protective function, I show that all antibiotic treatments also result in widespread collateral microbiome impact, through enrichment of AR genes with no known direct activity against the specific antibiotic treatment. Together, these results demonstrate that while movement of AR functions between environmental and human-associated communities is limited, antibiotic pressure applied to dynamic microbial communities can result in significant and predictable responses. Further work to interrogate
the rates of AR movement between alternative nodes in the network of gene exchange as
well as animal models to confirm our predicted response of developing microbial communities to antibiotic treatment is warranted.
Chair and Committee
Barak A Cohen, Jeffrey I Gordon, James J Havranek, Jonathan A Myers, James B Skeath, Barbara B Warner
Gibson, Molly Krisann, "Characterizing and Modeling Antibiotic Resistance Dynamics in Diverse Microbial Communities" (2015). Arts & Sciences Electronic Theses and Dissertations. 572.