While the composition of the gut microbiota varies markedly among humans, the functional ramifications of these differences have only begun to be explored. In the course of characterizing the reductive metabolism of aromatic amino acids by a common gut Firmicute, Clostridium sporogenes, we made the unanticipated discovery that this bacterium decarboxylates tryptophan to form the beta-arylamine neurotransmitter tryptamine, an enzymatic activity that is exceedingly rare among bacteria. Using a combination of genetics and biochemistry, we identify and characterize the PLP-dependent decarboxylase responsible for this activity, CLOSPO_02083, which had been misannotated as a tyrosine decarboxylase. To explore whether tryptophan decarboxylation is more widely distributed among the microbiota than previously known, we carried out a phylogeny-informed screen of ~15 putative bacterial decarboxylases. This screen revealed another novel tryptophan decarboxylase, RUMGNA_01526, which is phylogenetically distinct from CLOSPO_02083. Crystal structures of RUMGNA_01526 in its native form and bound to the inhibitor (S)-alpha-fluoromethyltryptophan, the first structures of a bacterial PLP-dependent decarboxylase, reveal the determinants of selectivity for the larger substrate Trp, including a flexible catalytic loop that controls access to the substrate-binding pocket. A computational analysis of whole-genome shotgun sequencing data from the Human Microbiome Project demonstrates that at least 10% of the human population harbors one of these two tryptophan decarboxylases in their gut community. By revealing a novel biochemical activity that is present in the gut communities of some but not all individuals, our results open a new line of investigation into the production and function of beta-arylamines by the human microbiota.

Lipopolysaccharide (LPS), a cell-associated glycolipid that makes up the outer leaflet of the outer membrane of Gram-negative bacteria, is a canonical mediator of microbe-host interactions. It is sensed by the immune system through Toll-like receptor 4 on innate immune cells, typically indicating to the host that an outside invader is present that needs to be cleared, prompting an inflammatory response. LPS is a potent immunostimulatory molecule, and when this process goes unchecked the consequences can be severe: just nanograms of LPS from a pathogenic bacterium injected into the bloodstream of a mouse will cause the mouse to die of septic shock. Puzzlingly, the human microbiota seems to avoid this response altogether and can be tolerated by the immune system despite producing the same classes of molecules that, when made by pathogens, prompt inflammation. The most prevalent Gram-negative gut bacterial taxon, Bacteroides, makes up ~50% of the cells in a typical Western gut; these cells harbor ~300 mg of LPS, making it one of the highest-abundance molecules in the intestine. However, remarkably little is known about the molecular details of microbiota-immune interactions mediated by Bacteroides LPS. As a starting point for understanding the biological function of Bacteroides LPS, we sought to characterize the biosynthesis and structure of the three classical components of LPS: lipid A, the core oligosaccharide, and the O-antigen repeating unit. We identified genes in Bacteroides thetaiotaomicron involved in the biosynthesis of its lipid A core and glycan, generated mutants that elaborate altered forms of LPS, and used mass spectrometry to interrogate the molecular features of these variants. We demonstrate that the glycan does not appear to have a repeating unit and is therefore better categorized as lipooligosaccharide (LOS) rather than LPS. Our work lays the foundation for developing a structure-function relationship for Bacteroides LPS in the context of host colonization.

While some microbial encounters can be pathogenic, most interactions between humans and microorganisms do not result in disease. Our gastrointestinal tract harbors approximately 100 trillion organisms, spanning about 1000 species that live in harmony with us. While we provide them with food and shelter, this complex intestinal bacterial community supplies essential nutrients, metabolizes indigestible compounds, protects us from pathogenic organisms and helps develop our immune system, thereby establishing a symbiotic relationship. Bacteroides are one of the most abundant bacterial genera living in the human gut and has long been a model system for studying the molecular mechanisms of commensal-host interactions. In the course of searching for surface structures that have immunomodulatory properties on the host, we serendipitously discovered a previously unknown surface protein BT1927 from Bacteroides that is capable of completely covering the surface of the bacteria by creating a tessellated surface layer. Interestingly, the formation of the entire proteinaceous capsule is controlled by inversion of a single promoter, making it phase-variable. Although only ~ 0.1% of the population produce this capsule in laboratory culture, analyzing the fecal metagenome sequence data from the Human Microbiome Project revealed that ~ 8% of the population produce a homologous capsule in humans, suggesting its significance in vivo. From a series of functional studies, we found that the BT1927-expressing subpopulation is profoundly resistant to complement-mediated killing, due in part to the BT1927-mediated blockade of C3b deposition. Our results show that the Bacteroides capsule is capable of a far greater degree of structural variation than previously known, perhaps to cope with the varying environmental conditions in the human intestine.

The human microbiome comprises a plethora of interacting bacteria. The composition of the microbiome has been correlated with conditions such as inflammatory bowel disease, colon cancer, diabetes and obesity. Manipulating the composition and function of the microbiome has huge potential for improving the human condition. Because the microbiome is an interacting, cooperating consortium, understanding how bacteria interact and assemble in communities is key to engineering a synthetic microbiome. In this thesis an exploration of the building of microbial communities is presented in two-fold.

First, bacterial members of the human microbiome are identified based on their prevalence and variation among many subjects. These bacterial species could function as potential starting points for exploring building a synthetic microbiome optimized for human health.

Second, a novel method in droplet-microfluidics is presented that uses two-dimensional droplet constriction and high-speed fluorescent particle detection to accurately count and sort droplets with sub-micron particles. This technology can be used to build defined communities of bacteria at a relevant scale, in high-throughput, with many replicates and with single cell control.

As synthetic biology strives to understand life by building it, these studies aim to provide new tools and resources for the study of microbial interactions and the assembly dynamics of their communities. Applications of this work could potentially lead to the engineering of key members of the microbiome into a synthetic consortium individually optimized for human health.