Modeling Lactococcus lactis using a genome-scale flux model

Abstract

Background: Genome-scale flux models are useful tools to represent and analyze microbial metabolism. In this work we reconstructed the metabolic network of the lactic acid bacteria Lactococcus lactis and developed a genome-scale flux model able to simulate and analyze network capabilities and whole-cell function under aerobic and anaerobic continuous cultures. Flux balance analysis (FBA) and minimization of metabolic adjustment (MOMA) were used as modeling frameworks.

Results: The metabolic network was reconstructed using the annotated genome sequence from L. lactis ssp. lactis IL1403 together with physiological and biochemical information. The established network comprised a total of 621 reactions and 509 metabolites, representing the overall metabolism of L. lactis. Experimental data reported in the literature was used to fit the model to phenotypic observations. Regulatory constraints had to be included to simulate certain metabolic features, such as the shift from homo to heterolactic fermentation. A minimal medium for in silico growth was identified, indicating the requirement of four amino acids in addition to a sugar. Remarkably, de novo biosynthesis of four other amino acids was observed even when all amino acids were supplied, which is in good agreement with experimental observations. Additionally, enhanced metabolic engineering strategies for improved diacetyl producing strains were designed.

Conclusion: The L. lactis metabolic network can now be used for a better understanding of lactococcal metabolic capabilities and potential, for the design of enhanced metabolic engineering strategies and for integration with other types of 'omic' data, to assist in finding new information on cellular organization and function.

Figure 1

Plotting model and experimental data for anaerobic growth, chemostat conditions, when maximizing for biomass formation. Growth rate and conversion yields for lactate (L), formate (F), ethanol (E) and acetate (A) on glucose (S) are plotted against glucose uptake rate. Model results are from simulations NX5 to NX10 (see Table 5 in Additional file 5). Experimental data is from Thomas, T.D., et al (1979) [30]. Ethanol and formate predictions are overlapped.

Figure 2

Plotting model and experimental data for anaerobic growth, chemostat conditions, when minimizing for substrates uptake. Glucose uptake rate and conversion yields for lactate (L), formate (F), ethanol (E) and acetate (A) on glucose (S) are plotted against growth rate. Model results are from simulations NS3, NS5, NS7, NS8 to NS11 (see Table 6 in Additional file 5). Experimental data is from Thomas, T.D., et al (1979) [30]. Acetate and formate predictions are overlapped.

Figure 3

Model predictions for glucose and amino acid uptake rates versus the growth rate. Model results are from simulations NS3, NS5, NS7, NS8 to NS11. Amino acids were grouped into six families: Histidine (His), Aromatic (Phe, Trp, Tyr), Serine-family (Cys, Gly, Met, Ser, Thr), Pyruvate-family (Ala, Ile, Leu, Val), Aspartate-family (Asp, Asn, Lys) and Glutamate-family (Arg, Glu, Gln, Pro) [47].

Figure 4

The pyruvate metabolism of Lactococcus lactis. LDH: lactate dehydrogenase; PDH: pyruvate dehydrogenase complex; PFL: pyruvate formate-lyase; ADHE: acetaldehyde dehydrogenase ; ADHA: alcohol dehydrogenase; PTA: phosphotransacetylase; ACKA: acetate kinase, ALS/ILV B: catabolic and anabolic 2-acetolactate synthase; ALDB-acetolactate decarboxylase; BUTA-diacetyl reductase; BUTB: acetoin reductase; NOX: NADH – oxidase. ..