(373e) Systematic Evaluation of Two Genome-Scale Models for Clostridium Tyrobutyricum through Knowledge Matching

Clostridium tyrobutyricum is perhaps the most promising microorganism for producing high titers of C4 hydrocarbons [1,2]. While native C. tyrobutyricum is widely known for producing butyrate, several mutant strains have been developed for butanol production [3,4]. Despite its high C4 production rate, C. tyrobutyricum co-produces acetate, increasing downstream costs for utilizing the bio-C4 substrates. In addition, various publications have noted the significant production of either lactate or ethanol in certain environments. The current understanding for the strain’s coproduction lies in how the cell balances its energy and cofactor requirements when exposed to a particular environment. However, the specifics of this balance have been theorized but are not fully understood.

Genome-scale models (GEM’s) provide an overarching view of a cell’s metabolism by using the genome to create a reaction network. GEM’s can be used to predict cellular phenotypes, facilitate mutant design through in-silico experiments, and provide a platform for multi-omics data integration. Traditionally, for a GEM to be deemed relevant, it must provide quantitatively reasonable estimates based on multiple sets of experimental data. However, the GEM should also capture how the cell reacts to scenarios beyond the experimental data. GEM’s should be qualitatively evaluated based on pre-existing biological knowledge. Identifying the key qualitative results can simplify the hundreds to thousands of numerical results into a legible one that expedites the collection of novel insights [5].

To better understand the energy and cofactor balance of C. tyrobutyricum, this work conducted systematic evaluation of two separate GEM’s developed for C. tyrobutyricum ATCC 25755. The first GEM, iKB917, was developed using iCM925, a GEM for C. beijerinckii NCIMB 8052 [6], as a reference model [7]. The central carbon pathway and biomass equation were manually altered to match those of C. tyrobutyricum. The second GEM, iCT583, was constructed using the strain’s genome and manually curated to match a particular set of quantitative and qualitative experiments [1]. To compare the two models, the biomass equations were reconstructed so the elemental requirements (namely carbon, nitrogen, and phosphorous) were roughly identical. In addition, the metabolite names for iCT583 were converted to the same syntax as iKB917 for ease of analysis. For all validation, maximizing biomass production was set as the objective function.

The two C. tyrobutyricum GEM’s were initially evaluated quantitatively based on the results from their primary sources [1,4,7]. Not surprisingly, each model provided reasonable estimates for their source material when the glucose consumption rate was set. Nevertheless, neither model seemed to capture the quantitative measurements of the other. This may be due to differences in experimental media. The source for iKB917 noted significant ethanol production and no lactate production when the strain was grown on Tryptone-Glucose-Yeast medium [8]. Conversely, the source for iCT583 recorded significant lactate yields at higher pH values when grown on Reinforced Clostridial Medium under the same abiotic conditions [1].

We apply the system identification (SID)-based framework to conduct a systematic evaluation of the two GEM’s [5]. The basic idea of the SID framework is to extract the biological knowledge captured by a GEM according to the following steps: (1) designing and conducting a series of in silico experiments that represents environmental perturbation to the cells; (2) analyzing the in silico results using multivariate statistical analysis methods (such as principal component analysis) to extract how the perturbation propagates through the metabolic network; and (3) visualize the extracted knowledge against the metabolic network. For example, applying the SID framework, we found that hydrogen gas production plays a vital role in butyrate yield in both GEM’s. This qualitative result matches the current understanding of the metabolism: ferrodoxin is oxidized with the formation of hydrogen gas, and the ferrodoxin is reduced to assist in producing butyryl-CoA, the precursor for butyrate [1]. In addition to the SID-based knowledge matching, we also conducted manual examination of the reactions, metabolites, and subsystems of each GEM, contextualizing the SID results and explaining key differences between the two GEM’s.

Work Cited:

[1] Feng, J., Guo, X., Cai, F., Fu, H., & Wang, J. (2022). Model-based driving mechanism analysis for butyric acid production in Clostridium tyrobutyricum. Biotechnology for Biofuels and Bioproducts, 15(1). https://doi.org/10.1186/s13068-022-02169-z

[2] Linger, J. G., Ford, L. R., Ramnath, K., & Guarnieri, M. T. (2020). Development of Clostridium tyrobutyricum as a microbial cell factory for the production of fuel and chemical intermediates from lignocellulosic feedstocks. Frontiers in Energy Research, 8. https://doi.org/10.3389/fenrg.2020.00183

[3] Bao, T., Feng, J., Jiang, W., Fu, H., Wang, J., & Yang, S.-T. (2020). Recent advances in N-butanol and butyrate production using engineered Clostridium tyrobutyricum. World Journal of Microbiology and Biotechnology, 36(9). https://doi.org/10.1007/s11274-020-02914-2

[4] Zhang, J., Zong, W., Hong, W., Zhang, Z.-T., & Wang, Y. (2018). Exploiting endogenous CRISPR-CAS system for multiplex genome editing in Clostridium tyrobutyricum and engineer the strain for high-level butanol production. Metabolic Engineering, 47, 49–59. https://doi.org/10.1016/j.ymben.2018.03.007

[5] Damiani, A. L., He, Q. P., Jeffries, T. W., & Wang, J. (2015). Comprehensive evaluation of two genome-scale metabolic network models for Scheffersomyces stipitis. Biotechnology and Bioengineering, 112(6), 1250–1262. https://doi.org/10.1002/bit.25535

[6] Milne, C. B., Eddy, J. A., Raju, R., Ardekani, S., Kim, P.-J., Senger, R. S., Jin, Y.-S., Blaschek, H. P., & Price, N. D. (2011). Metabolic Network Reconstruction and genome-scale model of butanol-producing strain clostridium beijerinckii NCIMB 8052. BMC Systems Biology, 5(1). https://doi.org/10.1186/1752-0509-5-130

[7] Badr, K., He, Q. P., & Wang, J. (2022). Knowledge-matching based computational framework for genome-scale metabolic model refinement. Computer Aided Chemical Engineering, 919–924. https://doi.org/10.1016/b978-0-323-85159-6.50153-6

[8] Wang, Y., Li, X., Milne, C. B., Janssen, H., Lin, W., Phan, G., Hu, H., Jin, Y.-S., Price, N. D., & Blaschek, H. P. (2013). Development of a gene knockout system using mobile group II introns (Targetron) and genetic disruption of acid production pathways in clostridium beijerinckii. Applied and Environmental Microbiology, 79(19), 5853–5863. https://doi.org/10.1128/aem.00971-13