Team:Aachen/Modeling


Introduction

Is it possible to go from the C1 molecule methanol to the C6 polysaccharide glycogen? Using stoichiometric modeling we can make predictions for the conversion pathway and decide on a host organism.

To model through which reactions molecules are converted within a biochemical network, all reactions have to be represented in a mathematical form. This is achieved by constructing a really big matrix that represents every reaction in the network and the stoichiometric coefficients of that reaction. This matrix is multiplied with a vector that contains all reaction rates. In the steady-state of the biochemical network, the product of the matrix with the vector is the null vector.

Gathering information to understand how the biochemical network operates works by setting constraints onto the reaction rates. All reactions have lower- and upper-bounds that also determine if the reaction is reversible. To test a certain scenario, for example the growth on glucose, the upper and lower bound will be set to less extreme values. The equation system is then optimized to maximize a so called objective function, for example the biomass reaction rate. The solution of this optimization process then gives information about the flux distribution within the biochemical network.[1]

Key Achievements

  • elemental balances were checked for all solutions
  • tested different methanol uptake pathways in different genome-scale models
  • modeled the stoichiometric feasibility of methanol -> glycogen conversion

Model and Methods

We evaluated three candidate organisms for our conversion pathway using their genome-scale stoichiometric models:

  • Escherichia coli genome scale model iJO1366 [2]
  • Saccharomyces cerevisiae genome scale model iND750 [3]
  • Pseudomonas putida genome scale model PpuMBEL1071 [4]

The models were loaded with COBRApy [5] and solved with GLPK [6]. Escher [7] was used for visualization.


The metabolic network is modeled under the steady-state assumption, i.e., all metabolite concentrations and reaction rates are constant

Methanol Uptake Pathways

At the beginning, we searched for organisms that are worth considering for our purposes. We could either start with an methylotrophic microorganism that is by nature able to grow on methanol, or engenineer a model organism like E.coli to do so. We chose to use E. coli because you can hardly find tools for genetically engineering methylotrophs like Bacillus methanolicus. Furthermore, the physiology of these strains is less understood.

Wild type E. coli is not able to utilize methanol as a carbon source. By the heterologous expression of three enzymes, the ribulose monophosphate pathway (RuMP) can be introduced to E. coli [8].


Aachen Reaction mdh.png
Reaction of the Methanol Dehydrogenase
Methanol is oxidized and NADH is generated.
Aachen Reaction hps.png
Formation of the C6-body ba the hexulose-6-phosphate synthase (hps)
Formaldehyde is assimilated with ribulose-5-hosphate to hexulose-6-phosphate.
Aachen Reaction phi.png
Isomerization to fructose-6-phosphate
Isomerization to fructose-6-phosphate.


A disadvantage of the RuMP is the consumption of one ATP per assimilated formaldehyde molecule.


Recently an ATP-neutral assimilation route was found in-silico [9]. To construct the "major mode" configuration of the methanol condensation cycle (MCC) in E. coli, the Mdh, Hps and Phi have to be expressed together with the phosphoketolase Xpk.

Aachen Reaction xpk.png
Reaction of the Phosphoketolase
Methanol is oxidized and NADH is generated.

To test if the introduction of the four reactions is sufficient to enable methanol assimilation, we added missing metabolites and reactions to the iJO1366 model. In all test scenarios we introduced the metabolites D-arabino-hex-3-ulose 6-phosphate and D-fructofuranose 6-phosphate.


Aachen Modeling Performance1.png
Theoretical growth rate with different uptake pathways
The unmodified iJO1366 model is not able to grow methanol (EX_meoh_e=-10) as the sole carbon source. By addition of missing reactions from the ribulose monophosphate pathway (RuMP), a maximal biomass reaction rate of 1.2 is possible. With reactions for the major-mode methanol condensation cycle (MCC) the maximal biomass reaction rate increases to 1.4.

After we considered the number of genes that have to be introduced recombinantly and the presence of native formaldehyde degradation routes, we chose E. coli as our host organism.

Conversion of Methanol to Glycogen

In the iJO1366 model, the glycogen metabolism is modeled with two metabolites glycogen_c and bglycogen_c that represent unbranched and branched polysaccharide chains. Due to the steady-state assumption, the model does not have exchange-reactions for glycogen. In order to model glycogen formation, we introduced exchange reactions for glycogen_c and bglycogen_c and optimized the model with the glycogen_c exchange reaction as the objective function.

When the glucose exchange reaction was set to a lower bound of 0 and methanol exchange to -10 (methanol as the only carbon source) and the model was optimized for the glycogen exchange reaction, a formation rate of 1.67 was observed. This means that all carbon atoms from methanol can theoretically be converted to glycogen. At the same time the ATP synthesis and consumption is fully balanced between five reactions: ATPS4rpp, PPKr, ATPM, PFK, GLGC.

Aachen Modeling toGlycogen.png
Conversion of Methanol to Glycogen
The iJO1366 model was constrained to a methanol uptake rate of -10 and no CO2 uptake. Reactions that were introduced to the model are named with a double underscore (__). The glycogen formation rate of 1.67 shows that theoretically all carbon atoms can be converted from methanol to glycogen. All non-zero ATP utilizing reactions are balanced and shown.

References

  1. Scarlat,N.,etal.,Theroleofbiomassandbioenergyinafuture bioeconomy:Policiesandfacts. EnvironmentalDevelopment(2015), http://dx.doi.org/10.1016/j.envdev.2015.03.006
  2. http://www.jbc.org/content/239/12/4018.full.pdf
  3. Reconstruction and validation of Saccharomyces cerevisiae iND750, a fully compartmentalized genome-scale metabolic model, Duarte et. al 2004, http://www.ncbi.nlm.nih.gov/pubmed/15197165
  4. Sohn SB, Kim TY, Park JM, Lee SY. In silico genome-scale metabolic analysis of Pseudomonas putida KT2440 for polyhydroxyalkanoate synthesis, degradation of aromatics and anaerobic survival. Biotechnol J. 2010 Jul;5(7):739-50. doi:10.1002/biot.201000124. PubMed PMID: 20540110.
  5. Ali Ebrahim, Joshua A Lerman, Bernhard O Palsson and Daniel R Hyduke (2015) COBRApy: COnstraints-Based Reconstruction and Analysis for Python
  6. http://www.gnu.org/software/glpk/
  7. Zachary A. King, Andreas Dräger, Ali Ebrahim, Nikolaus Sonnenschein, Nathan E. Lewis, and Bernhard O. Palsson (2015) Escher: A web application for building, sharing, and embedding data-rich visualizations of biological pathways, PLOS Computational Biology
  8. Müller JE, Meyer F, Litsanov B, Kiefer P, Potthoff E, Heux S, Quax WJ, Wendisch VF, Brautaset T, Portais JC, Vorholt JA. Engineering Escherichia coli for methanol conversion. Metab Eng. 2015 Mar;28:190-201. doi:10.1016/j.ymben.2014.12.008. Epub 2015 Jan 14. PubMed PMID: 25596507
  9. Bogorad IW, Chen CT, Theisen MK, Wu TY, Schlenz AR, Lam AT, Liao JC. Building carbon-carbon bonds using a biocatalytic methanol condensation cycle. Proc NatlAcad Sci USA. 2014 Nov 11;111(45):15928-33. doi: 10.1073/pnas.1413470111. Epub 2014 Oct 29. PubMed PMID: 25355907; PubMed Central PMCID: PMC4234558.