Genome-scale metabolic reconstructions of Bifidobacterium adolescentis L2-32 and Faecalibacterium prausnitzii A2-165 and their interaction

Abstract

Background: The gut microbiota plays an important role in human health and disease by acting as a metabolic organ. Metagenomic sequencing has shown how dysbiosis in the gut microbiota is associated with human metabolic diseases such as obesity and diabetes. Modeling may assist to gain insight into the metabolic implication of an altered microbiota. Fast and accurate reconstruction of metabolic models for members of the gut microbiota, as well as methods to simulate a community of microorganisms, are therefore needed. The Integrated Microbial Genomes (IMG) database contains functional annotation for nearly 4,650 bacterial genomes. This tremendous new genomic information adds new opportunities for systems biology to reconstruct accurate genome scale metabolic models (GEMs).

Results: Here we assembled a reaction data set containing 2,340 reactions obtained from existing genome-scale metabolic models, where each reaction is assigned with KEGG Orthology. The reaction data set was then used to reconstruct two genome scale metabolic models for gut microorganisms available in the IMG database Bifidobacterium adolescentis L2-32, which produces acetate during fermentation, and Faecalibacterium prausnitzii A2-165, which consumes acetate and produces butyrate. F. prausnitzii is less abundant in patients with Crohn's disease and has been suggested to play an anti-inflammatory role in the gut ecosystem. The B. adolescentis model, iBif452, comprises 699 reactions and 611 unique metabolites. The F. prausnitzii model, iFap484, comprises 713 reactions and 621 unique metabolites. Each model was validated with in vivo data. We used OptCom and Flux Balance Analysis to simulate how both organisms interact.

Conclusions: The consortium of iBif452 and iFap484 was applied to predict F. prausnitzii's demand for acetate and production of butyrate which plays an essential role in colonic homeostasis and cancer prevention. The assembled reaction set is a useful tool to generate bacterial draft models from KEGG Orthology.

Figure 1

Method summary. (I) The gene assigned with KO for each studied organism was downloaded from the IMG database or KEGG. (II) The KO was mapped with the reaction data set. (III) A draft model was exported to MS-Excel format by our function saveDraftModel, the draft model was mapped to KEGG maps using our function DrawPathway. (IV) The draft was curated manually from literature and other gene annotations in IMG files such as TIGRFAMs and Pfam. After this, the model was simulated using RAVEN and MOSEK. (V) The community interaction design described how the organisms share growth medium components. (VII) Community interaction was converted to XML format. (VIII) Both optCom and FBA models were generated from XML files.

Figure 2

The effect of lactate production on the iBif452 Model. (A) The biomass decreases with increasing the lactate production. (B) The production of acetate, formate, and ethanol decrease with increasing the lactate production.

Figure 3

The effect of glucose and acetate on the growth of iFap484. Glucose and acetate uptake rate varied between 0 to 1 mmol/gDW/h.

Figure 4

Simulation summary results using OptCom and FBA methods when iBif452 and iFap484 grow together on glucose. The black numbers are fluxes predicted by OptCom, the red numbers are fluxes predicted by the FBA method. ATP non-growth association maintenance was fixed at 0.4 and 0.5 mmol/gDW/h in the iFap484 and iBif452 models respectively. The unit of ATP, acetate, lactate, ethanol, formate and butyrate is mmol/gDW/h. The unit of biomass is (1/h).

Figure 5

Abundances analysis of iFap484 and iBif452. How the amount of butyrate and the total glucose consumption change with different compositions of iFap484 and iBif452.