(643f) Developing Genome-Scale Whole-Plant Models for Poplar (Populus deltoides) and Switchgrass (Panicum virgatum)

Suthers, P. - Presenter, The Pennsylvania State University
Sarkar, D., The Pennsylvania State University
Maranas, C., The Pennsylvania State University
In recent years, FBA in conjunction with constraint-based models (CBMs) has been extended to include the study of multicellular organisms such as plants. Conventionally applied to microbial systems, this extension from microbial to eukaryotic metabolic modeling is not without its challenges; the first plant genome-scale model (GSM) was constructed for Arabidopsis thaliana in 2009, more than a decade after microbial GSMs were first developed. The compartmentalized nature of eukaryotic metabolism and limited understanding of metabolite transport between compartments are primary roadblocks in model construction. Similar metabolic pathways are often found in multiple organs with varying flux distributions, as plants apportion cellular functions nonuniformly between organs to form a highly connected and interdependent metabolism. Despite these challenges, plant metabolic reconstructions have been employed to (i) correctly inventory the metabolism within the plant, (ii) examine and predict the effect of nutrient and genetic perturbations, and (iii) characterize metabolite transport between tissues and organs. The present work describes genome-scale reconstruction efforts for two feedstocks used in bioenergy: Populus deltoides (poplar) and Panicum virgatum (switchgrass).

The GSM for P. deltoides covers primary and secondary metabolism for a compartmentalized plant cell based on the P. deltoides genome, accounting for 3555 metabolites and 3015 metabolic reactions. The initial draft model was constructed using the PoplarCyc database [1] and BLAST-matched Arabidopsis thaliana proteins. Due to the paucity of cellular component ontologies (CO) for poplar proteins, a constraint-based method was developed for predicting subcellular localization of reactions based on the underlying metabolic network. The developed method predicts localization scores for unannotated reactions by maximizing the flux through reactions with known (subcellular) localizations while parsimoniously filling in network gaps to ensure the production of biomass precursors. Reactions with unknown localizations were thus distributed among eight subcellular compartments – cytosol, plastid, peroxisome, vacuole, mitochondria, plasma membrane, thylakoid membrane, and inner mitochondria matrix. Among the reactions annotated thusly, localizations for ~53% of the predicted reactions were found to be same as listed in the BRENDA database for organisms other than P. deltoides and A. thaliana. For instance, 3-hydroxyacyl-CoA dehydrogenase was predicted to be located exclusively in the peroxisome, as seen in plants where the -oxidation cycle is only present in the peroxisome [2]. Subsequently, this scaffold model will be used to construct a whole-plant metabolic model encompassing the leaf, stem, and root tissues, all connected by the vascular system. The envisioned model will capture carbon and nitrogen flows between tissues as a function of its growth cycle and sequester biomass in organ-specific ratios. The model will be used to explore pathways carrying the biggest lever on target processes such as growth, nitrogen utilization, feedstock degradability, and lignin valorization.

Similarly, the metabolic model for Panicum virgatum accounts for over 2500 metabolites and 3400 reactions and provides a framework for accounting for the substantial genetic diversity present in its two cultivars (upland and lowland). Although there are differences in pathways in switchgrass (a perennial) and maize (an annual), both are members of the Panicoideae subfamily of grasses and have a form of C4carbon fixation. Therefore, we used MaizeGrow and an earlier maize leaf model [3] as a reference during model construction by performing protein-protein BLAST to systematically search for homology for the gene-product of each gene contained in the source model [4]. Each match having an Expect value of 10-10 was reverse searched into the Maize protein database; if the best result was the initial Maize protein, the switchgrass gene and the corresponding reactions were added to the model. SwitchgrassCyc and the literature were used to bring in additional reactions and gene associations such as the NAD-malic enzyme [5]. Transcript data [6] were used to localize reactions into the mesophyll and bundle sheath cells of the leaf. We describe how the modelis expanded and adapted into organs, specifically the root, stalk, and leaf connected by vascular tissues using representative data for biomass growth [7,8]. The resulting model is used to determine the metabolic changes that occur between various growth conditions, to capture carbon and nitrogen flows between tissues as a function of growth stage, pinpointing tissue-specific bottlenecks in nitrogen metabolism, and to identify key switchgrass genes involved with yield and composition and metabolic tradeoffs between target phenotypes.


[1] Zhang P, Dreher K, Karthikeyan A, Chi A, Pujar A, Caspi R, et al. Creation of a Genome-Wide Metabolic Pathway Database for Populus trichocarpa Using a New Approach for Reconstruction and Curation of Metabolic Pathways for Plants. PLANT Physiol 2010. doi:10.1104/pp.110.157396.

[2] Poirier Y, Antonenkov VD, Glumoff T, Hiltunen JK. Peroxisomal β-oxidation-A metabolic pathway with multiple functions. Biochim Biophys Acta - Mol Cell Res 2006. doi:10.1016/j.bbamcr.2006.08.034.

[3] Simons M, Saha R, Amiour N, Kumar A, Guillard L, Clement G, et al. Assessing the Metabolic Impact of Nitrogen Availability Using a Compartmentalized Maize Leaf Genome-Scale Model. PLANT Physiol 2014. doi:10.1104/pp.114.245787.

[4] Mueller TJ, Berla BM, Pakrasi HB, Maranas CD. Rapid construction of metabolic models for a family of Cyanobacteria using a multiple source annotation workflow. BMC Syst Biol 2013;7:142. doi:10.1186/1752-0509-7-142.

[5] Christin P-A, Salamin N, Kellogg EA, Vicentini A, Besnard G. Integrating Phylogeny into Studies of C4 Variation in the Grasses. PLANT Physiol 2009. doi:10.1104/pp.108.128553.

[6] Rao X, Lu N, Li G, Nakashima J, Tang Y, Dixon RA. Comparative cell-specific transcriptomics reveals differentiation of C4photosynthesis pathways in switchgrass and other C4lineages. J Exp Bot 2016. doi:10.1093/jxb/erv553.

[7] Serba DD, Daverdin G, Bouton JH, Devos KM, Brummer CE, Saha MC. Quantitative Trait Loci (QTL) Underlying Biomass Yield and Plant Height in Switchgrass. BioEnergy Res 2014. doi:10.1007/s12155-014-9523-8.

[8] Madakadze IC, Stewart K, Peterson PR, Coulman BE, Smith DL, Zhou X. Switchgrass biomass and chemical composition for biofuel in eastern Canada. Agron J 1999. doi:10.2134/agronj1999.914696x.