The binding of protein to DNA is central to the regulation of gene expression and the organization of chromosomal DNA. To date, there exist few techniques for the determination of genome-wide protein binding in prokaryotes, and none that are simultaneously simple, high-resolution, and rapid. I describe XATAC-seq, an adaptation of the eukaryotic assay for transposase-accessible chromatin with sequencing (ATAC-seq), combining formaldehyde crosslinking of DNA-protein complexes, adapter-loaded transposase treatment for next-generation sequencing library generation, and high-throughput sequencing to interrogate these genome-wide binding patterns in bacteria. The technique captures the binding of both major classes of prokaryotic DNA-binding proteins–transcription factors and nucleoid-associated proteins–genome-wide at the resolution of individual binding sites. XATAC-seq was applied to determine the protein occupancy landscapes of several bacterial species. Remarkably, the landscapes show a high degree of fidelity to specific nucleoid-associated proteins and demonstrate several conserved characteristics, including extended domains of high enrichment and preferential enrichment of AT-rich regions. This has led to the speculation that these nucleoid-associated proteins are members of a common high-level functional group, and that this class of nucleoid-associated protein is prevalent among a significantly wider range of prokaryotes than previously realized. In particular, the Mga protein of Streptococcus pyogenes is proposed to serve the high-level function of suppression of ectopic expression in an analogous fashion to the H-NS protein in E. coli. This work represents the first assessment of protein occupancy landscapes in gram-positive bacteria and a significant technical improvement over existing techniques for assaying genome-wide protein binding in prokaryotes.

The essential role of microbiota in macroscopic organisms has led to a coining of the term “metaorganism”, which encompasses a multicellular host organism and the various microbial communities found within it. Humans depend on the many microbial communities found within and around them to survive. However, reproducible and mechanistic studies of these communities have remained frustratingly elusive in many instances due to the challenges of replicating in vivo community behavior in an in vitro laboratory setting. This dissertation seeks to address this problem and demonstrate the benefits for human health that can be obtained through microbiome studies. The first chapters present the development of reproducible synthetic soil and skin microbial communities, to be used for in vitro study of community dynamics and targeted interventions on microbiome composition. Studies with these tools show that final community diversity and composition can be modulated by minor adjustments in starting inoculum, with certain organisms driving the final state of the community. This knowledge of community dynamics can be used, for example, to support design of soil microbial communities to increase agricultural plant growth in a sustainable fashion. Additionally, growth of the skin community with cosmetics compounds probes the effect on these compounds on skin microbiota, which has implications for development of skin pathologies. Finally, a later chapter presents the use of a sialidase isolated from the gut microbiome to reduce cardiovascular disease development in a humanized mouse model by removing the sialic acid N-glycolylneuraminic acid from the endothelium. This study takes a novel enzyme and applies it in a new way, seeking to use microbiome studies to benefit the host organism on multiple levels. Together this body of work bridges the in vitro and in vivo aspects of microbiome research to implement novel strategies for partnering with the microbiome to support and advance human health.

As I am preparing my dissertation, a global Coronavirus pandemic is spreading, affecting the way we travel, work, and communicate. Transmitted primarily via respiratory droplets, this virus spread across the entire globe in a matter of months, highlighting the interconnectedness of modern life and the microscopic world which invisibly underlies all our interactions.

In fact the vast majority of life on earth is microbial, and for the vast majority of time microorganisms were the only life present. Animals emerged in a robust microbial world, and our evolution was shaped by and from these microorganisms.

Even during this global pandemic, it is easy to view ourselves as distinct, independent organisms. But recent evidence challenges this notion. Roughly 1% of our body weight is comprised of trillions of microorganisms, collectively referred to as the human microbiome. Our microbiomes are necessary for healthy nutrition, immune function, and even reproduction. How our microbiomes are formed, maintained, shared, and perturbed are active areas of research. My doctoral research focused on understanding and improving the molecular tools used to evaluate microbial communities and applying those tools to better understand human microbiomes.

This dissertation begins with three reviews summarizing what we know about human microbiomes and what is still lacking. Chapters 2 and 3 describe benchtop and computational advances to the human microbiome field, respectively. The final chapter includes two research articles currently under review describing the application of these tools to gain novel insight about human oral microorganisms.

Throughout my research, I found that human saliva served as a valuable tool for developing these novel techniques because it is simple to collect, high biomass, and relatively easy to manipulate. So although it was not my original intention, I have come to appreciate saliva as a valuable resource for pushing the boundary of human microbiome research. It is my hope that this research improves our ability to evaluate microbial communities, allowing for future research on altering microbiomes to improve human and environmental health.

Systems biology strives to understand complex multi-component biological processes and capture knowledge of their function through models. With metabolic and gene expression models (ME-models), we can mathematically and simultaneously represent the majority of these processes, including transcription, translation, and metabolism. This enables us to compute the molecular constituents of a cell as a function of genetic and environmental parameters. ME-models represent an improvement in current capabilities to predict phenotypes, as demonstrated by the reconstruction and validation of a ME-model for the acetogen Clostridium ljungdahlii. C. ljungdahlii can grow autotrophically on carbon monoxide (CO), and/or carbon dioxide + hydrogen (CO2+H2) and fix these gases into multicarbon organics, an ability that can be redirected to produce biocommodities. The C. ljungdahlii ME-model was able to improve growth rate predictions, identify previously unknown secretion products, and compute the transcriptome of C. ljungdahlii accurately.

ME-models offer the opportunity to systematically explore the interface between protein and function. First, perturbations of tRNA co-expression in ME-models revealed unique organization solutions to two different selective pressures: Optimization of growth through minimal co-expression of tRNAs, and efficiency of resources through optimal grouping of tRNAs. Second, because of the incorporation of protein translocation and membrane function, a ME-model was able to recapitulate acetate production during glucose consumption due to membrane overcrowding. Third, a ME-model highlighted how variations in nickel availability impacts metalloproteins, thereby controlling growth and secretion rates of fermentation products. Thus, three features that could constrain the proteome of an organism – genome architecture, ultra-structure, and media requirements – were successfully interrogated using ME-models.

The human microbiota exhibit a highly dynamic composition over the course of life and changes in the human gut microbiota have been associated with human health or disease. Reprogramming of the gut microbiota by interventions that counter these changes and promote long-lasting health has been an emerging topic in microbiome research. Predicting changes in the gut microbiome is therefore crucial for the nature and design of these interventions. Here, we report on a new method based on deep learning to forecast changes in the microbiome. We processed and analyzed nine time-course datasets of the human gut microbiome, identifying the main microorganisms present in these microbial communities at any given time. We then used an encoder-decoder neural network to train a model that successfully predicts the progression of the microbiome composition over time given only five time points of context data. Our results demonstrate the ability to predict the fate of the human gut microbiome into the future, providing the foundation for rational intervention design.

Gut dysbiosis can have an impact on a host’s susceptibility to disease due to the human reliance on microbial functions. Due to the major effects of the gut microbiome on various aspects of human health, there is a need to better understand the mechanisms behind what creates a healthy microbiome and how a microbiome can transition to a diseased state. Synthetic microbial gut communities have been previously developed as a method to better understand microbial interactions and functions, but most of them consisted of only a small number of species with a limited ability to model the complexity of the gut. One recent study incorporated 119 different strains from the most prevalent taxa in the human gut into a synthetic community but displayed an imbalanced final composition between fast and slow-growing strains, reducing its efficacy in representing slow-growing strains that are necessary for microbiome function. With the findings from previous studies in mind, this master’s thesis focuses on the development of a reproducible synthetic community with high complexity and a balanced composition named hCom as a study of how the human gut microbiome functions. This project aims to establish a model of the gut microbiome with increased α-diversity that allows for the study of both fast and slow-growing strains by incorporating the individual rates of change of growth for each strain into the design of the community. To assemble the community, four groups were made by separating the strains based on their rates of change of growth and added to the community at a 10-fold difference in concentration between each group. Upon analyzing the final composition of the community, it was found that the hCom design did successfully increase the relative-abundance of slow-growing strains while decreasing the relative-abundance of fast-growing strains, however one genus with medium-slow growing strains demonstrated dominance over the community, thus reducing the overall balance. Altogether the results of this thesis were successful in achieving its research goal of reducing the dropout of slow-growing strains, but there is still more work to be done in creating an even composition of the microbial community. By continuing to improve on the composition of the community, the information learned from experiments on the hCom can be applied to the native gut microbiome and the host-microbiome relationship, as well as aid in the development of treatments for human disease in the future.