(711b) Elucidating Microbiome-Virome Interactions and Metabolic Transactions in Bovine Rumen through in silico genome-Scale Modeling

The complex microbial ecosystem of the cattle rumen has a significant impact on host energy utilization. In addition to the complex syntrophic, mutualistic, and competitive interactions within the rumen microbiome, viruses have been shown to impact microbial populations through a myriad of processes, including cell lysis and reprogramming of host metabolism via Auxiliary Metabolic Genes (AMGs). Although advancements in high-throughput sequencing are providing access to the vast diversity and functions of this complex microbial ecosystem, our understanding of factors that shape rumen microbial communities is in its infancy, and little is known about the processes shaping the distribution of rumen viruses or the modulation of microbe-driven processes in the rumen. It is well established that bovine dietary variations significantly alter the ruminal microbiome. However, a comprehensive and systematic computational study to decipher the effect of diet utilization and viral AMGs on the microbiome is still missing. With a gradual increase in computational capability and the abundance of in silico genome-scale metabolic reconstruction tools, metabolic networks combined with constraint-based modeling provides opportunities to identify microbe-microbe and phage-microbe interactions. In this work, we investigated how rumen microbial and viral community structure and function respond in varied dietary treatments, as well as the distribution and ecological drivers of rumen viruses alongside their bacterial counterparts.

We developed a simplified rumen community metabolic model, with Prevotella ruminicola, Methanobrevibacter gottschalkii,and Ruminococcus flavefaciensas representative organisms for starch and protein digestion, methane production, and fiber digestion, respectively. We reconstructed the draft models for each of these species from the ModelSEED database¹, and then performed extensive manual curation, including chemical and charge-balancing, eliminating thermodynamically infeasible cycles and ensuring network connectivity. After our manual refinement processes, the models of P. ruminicola(517 genes, 1175 metabolites, 1107 reactions), M. gottschalkii (338 genes, 1000 metabolites, 987 reactions),and R. flavefaciens(492 genes, 1142 metabolites, 1137 reactions) were integrated into a community model using multi-level optimization framework^2,3. The community model was used to estimate metabolite secretion profiles and community compositions as a function of diet and host-specific variations.

Metabolic interactions among rumen community members have been only partially known to the scientific community. To enrich our understanding of the inter-species interactions in the ecosystem, we have customized and utilized a GapFind-GapFill procedure⁴that attempts to introduce new metabolic functions to the model of an organism, and based on the source of the de novo function, determines the interaction between difference species in the community. We have discovered 20 novel interactions involving the transfer of fatty acids, vitamins, coenzymes, amino acids and sugars among the organisms. To elucidate the functional role of the virome on the microbial ecosystem, we computationally identified the metabolic functions of the viruses associated with the community members. We identified viral metabolic functions that drive nucleotide synthesis, reducing power generation, the reprogramming of the bacterial carbon metabolism to pentose phosphate pathway and folate biosynthesis use, and viral replication. The identified functions of viral AMGs were incorporated into the model as regulatory information. The metabolic hubs and bottlenecks in the community were identified and the metabolite pools for important energy currencies in the community were predicted. Our results show that viruses have affected the core rumen microbiome structure and metabolism. Also, the virome dynamics is highly correlated with microbial density and dietary factors. Our model serves to discover unidentified metabolite transactions and answer key ecological questions of ruminant nutrition through diet-virome-microbiome interactions, while promising to develop novel strategies for methane mitigation and increasing nutritional efficiency of domesticated bovine species.

References:

1. Henry CS, DeJongh M, Best AA, Frybarger PM, Linsay B, Stevens RL. High-throughput generation, optimization and analysis of genome-scale metabolic models. Nature biotechnology. Sep 2010;28(9):977-982.
2. Zomorrodi AR, Maranas CD. OptCom: a multi-level optimization framework for the metabolic modeling and analysis of microbial communities. PLoS computational biology. Feb 2012;8(2):e1002363.
3. Zomorrodi AR, Islam MM, Maranas CD. d-OptCom: Dynamic Multi-level and Multi-objective Metabolic Modeling of Microbial Communities. ACS synthetic biology. Apr 18 2014;3(4):247-257.
4. Satish Kumar V, Dasika MS, Maranas CD. Optimization based automated curation of metabolic reconstructions.BMC bioinformatics. // 2007;8:212.