Spatio-Temporal Consumer-Resource and Genome-Scale Metabolic Modeling of the Rhizosphere Microbiome
Authors:
Ilija Dukovski1* ([email protected]), Jing Zhang1, Markus de Raad2, Melisa Osborne1, Rowan Nelson1, Manish Kumar4, Haroon Qureshi1, Hui Shi1, Sierre Ternoey1, Phineas McMillan1, Kshitiz Gupta3, Jeffrey Motschman3, William Hynes3, Anup Singh3, Karsten Zengler4, Daniel Segrè1, Adam M. Deutschbauer2, Trent R. Northen2
Institutions:
1Boston University; 2Lawrence Berkeley National Laboratory; 3Lawrence Livermore National Laboratory; 4University of California–San Diego
URLs:
Goals
The goal of the program is to understand the interactions, localization, and dynamics of grass rhizosphere microbial communities at the molecular level (genes, proteins, metabolites) to enable accurate predictions and interventions to effectively manage and harness microbes to achieve DOE missions in sustainable energy and carbon cycling.
Abstract
One of the challenges with dealing with the wealth of experimentally generated data in the project is to formulate a theoretical framework with predictive capability. Researchers approach this problem by employing predictive modeling strategies that integrate the increasingly available genomic information with mechanistic and machine learning modeling methodologies.
Our current modeling efforts are focused on two distinct strategies: the first, based on flux balance analysis (FBA), is aimed at pushing the boundaries of completeness and accuracy in simulating the spatio-temporal dynamics of rhizosphere communities; the second, based on experimentally parametrized differential equation models, is aimed at exploring tradeoffs between detail and scalability for increasingly large communities.
For the first strategy, researchers merge genome-scale reconstructions of cellular metabolism, assembled through the KBase platform (Arkin et al 2018) and manual curation of specific rhizosphere strains, with a spatio-temporal biophysical model of bacterial propagation implemented in the software COMETS (Computation Of Microbial Ecosystems in Time and Space; Harcombe et al 2014; Dukovski et al 2021) COMETS is freely available at https://www.runcomets.org/). Layouts for specific scenarios can be simulated through a Python macro-language. Researchers are using COMETS to build detailed biophysical models of the rhizosphere microbiome, starting from synthetic communities composed of 17 common rhizosphere microbiome members. These models will help researchers understand the interplay between metabolic and biophysical processes in shaping the spatial organization of microbial biomass metabolism around plant roots, and the nature of microbiome-plant interaction. These COMETS models will be used to simulate effects of genetic modifications, such as gene knockouts, to prioritize experimental community editing targets.
A second strategy is based on the usage of Consumer Resource Modeling (CRM; Mehta and Marsland 2021). This methodology provides a fast way of building coarse grained models of complex microbial communities, trading detail for scalability and speed, extending ecosystem models to virtually thousands of species (Mehta and Marsland 2021; Silverstein et al 2023). While CRMs may not be able to accurately capture many nonlinear features of microbial metabolism (such as diauxic shifts) and intracellular perturbations, they provide a valuable theoretical framework for studying complex bacterial communities. Researchers have developed a CRM of a synthetic community composed of the same 17 rhizosphere bacteria being simulated in COMETS. After estimating and optimizing the CRM parameters based on detailed exometabolomics data for individual microbes, researchers show that the CRM can predict pairwise interactions, and help map the role of metabolic cross-feeding across species in multi-species consortia. Furthermore, the CRM allowed researchers to predict new communities with maximal diversity.
Future efforts will focus on CRM predictions of environmental perturbations that could drastically alter community structure and dynamics, and at comparing COMETS and CRM simulations, in an attempt to develop a unified scalable strategy for effective community prediction and editing.
References
Arkin, A. P., et al. 2018. “KBase: The United States DOE Systems Biology Knowledgebase,” Nature Biotechnology 36, 566–69.
Dukovski, I., et al. 2021. “A Metabolic Modeling Platform for the Computation of Microbial Ecosystems in Time and Space (COMETS),” Nature Protocols 16, 5030–82.
Harcombe, W. R., et al. 2014. “Metabolic Resource Allocation in Individual Microbes Determines Ecosystem Interactions and Spatial Dynamics,” Cell Reports 7, 1104–15.
Mehta, P., and R. Marsland III. 2021. “Cross-Feeding Shapes both Competition and Cooperation in Microbial Ecosystems,” arXiv:2110.04965.
Silverstein, M., et al. 2023. “Metabolic Complexity Drives Divergence in Microbial Communities,” BioRxiv. DOI:10.1101/2023.08.03.551516.
Funding Information
This material by m-CAFEs Microbial Community Analysis and Functional Evaluation in Soils, ([email protected]) a Science Focus Area led by Lawrence Berkeley National Laboratory is based upon work supported by the U.S. DOE, Office of Science, Office of Biological and Environmental Research under contract number DE-AC02-05CH11231.