Transcriptomic and computational approaches for interrogating metabolic interactions in the coral microbiome

Ecosystems comprise large groups of highly interdependent organisms. Cnidarians, such as sea anemones and corals, are keystone species in many marine ecosystems, especially coral reefs. Each individual cnidarian also constitutes an ecosystem unto itself, a "holo- biont", consisting of the host animal and accompanying microbial symbionts. To interro- gate cnidarian holobionts, I used computational approaches to analyze the transcriptomes of three cnidarians and build mechanistic models of their microbial symbionts. In par- ticular, I analyzed and annotated the transcriptomes of the cauliflower coral Pocillopora damicornis, the lined sea anemone Edwardsiella lineata, and the starlet sea anemone Ne- matostella vectensis, providing information about the molecular functions expressed by these organisms, and allowing development of a corresponding set of public databases: PocilloporaBase, EdBase, and an updated version of StellaBase, that facilitate access to the corresponding datasets. Additionally, I developed a method to infer the phylogenetic antiquity of transcripts. This method also allowed me to identify transcripts from other organisms (e.g., microbes) belonging to the anemone or coral holobiont. In parallel – in order better to understand the microbial symbionts that share envi- ronments with cnidarian hosts, I also developed new computer-simulation approaches for modeling metabolic interactions between different microbial species. These approaches are based on genome-scale stoichiometric reconstructions of metabolic networks and on Flux Balance Analysis (FBA). In addition to contributing to the development and testing of a new FBA-based platform for modeling communities in structured environments (Compu- tation Of Microbial Ecosystems in Time and Space, or COMETS), I used this platform for specific in silico experiments on microbial symbiosis. In particular, I computed all pairwise interactions between 582 different prokaryotic models, and identified global patterns of pu- tative positive (cross-feeding) vs. negative (food competition) interactions in this matrix of species pairs. I found that about 7% of the pairs yielded a greater biomass when grown together than when grown separately as monocultures. Despite existing challenges, such as the limitations of gap-filling steps in model construction and the need for a better knowl- edge of nutrient composition in natural environments, this approach could in the future help forecast shifts in the coral holobiont under likely scenarios of marine environmen- tal changes. In general, this work demonstrates how the integration of high-throughput sequencing technology and mechanistic systems-biology simulations, can provide unique tools to analyze interactions between microbes, and to mitigate or reverse adverse changes in marine ecosystems