Investigation of Metabolic Capabilities of Recombinant Lactococcus Lactis for Production of Hyaluronic Acid Using Constraint Based Genome Scale Models

Background:

Hyaluronic acid (HA), a naturally occurring polysaccharide, is one of the chief components of the extracellular matrix in eukaryotes and the extracellular capsule in Streptococci. It is a linear non-sulphated glucosaminoglycan that contains alternating units of β-1,3-N-acetyl glucosamine and β-1,4-glucuronic acid. Its excellent water binding capacity paves way for its wide application in various fields including ophthalmic surgery, viscoelastic supplementation in arthritis and drug delivery scaffolds.

At first, HA was primarily extracted from rooster comb. However, the convoluted protocol of extraction and the chances of cross-species infection from avian viruses discouraged its application in biomedical areas. On the other hand, microbial sources like Streptococci, which are natural producers of HA, were more attractive as far as industrial production was concerned. However, the pathogenicity of Streptococci limited their use in biomedical applications. Hence, several alternatives are being developed by incorporation of HA synthesizing ability in â??saferâ? organisms such as Bacillus subtilis, Lactococcus lactis, Agrobacterium sp. and Escherichia coli.

Our research group focuses on Lactococcus lactis, a lactic acid bacterium, for production of HA. Currently, Lactococci are seen as ideal avenues for metabolic engineering due to the simplicity of their metabolism. They are widely used in many food fermentation industries and are classified as GRAS (Generally Regarded As Safe), making them more favourable for industrial production and medical applications.

A major issue concerning the recombinant microorganisms developed for HA production is the low HA yield and molecular weight when compared to Streptococci. Several research groups worldwide are studying the effect of process parameters and media constituents on HA yield and molecular weight to optimise the production strategy. However, to successfully engineer the cell to harness its maximum capabilities, it is also important to understand its metabolism in detail.

Redirection of fluxes towards HA pathway by modifying the metabolic network of the organism is another approach to contrive the cell for improved HA production. The general view is that flux towards HA biosynthesis pathway competes with glucose metabolism fluxes (glycolysis, PPP etc.) for glucose. Under normal culture conditions, it is known that the

glycolysis pathway is the major glucose consuming pathway. Hence glycolytic fluxes are central in the understanding of HA biosynthesis. The precursors of HA are also formed from glycolytic intermediates (glucose-6-phosphate and fructose-6-phosphate). On this note, there are successful studies of enhancement of HA production upon partial inhibition of glycolysis.

We here use a genome-scale metabolic model to analyse fluxes of Lactococcus lactis for producing HA in silico. Constraint-based modelling techniques such as Flux Balance Analysis (FBA) have been proven to be an effective tool to interrogate metabolic networks and understand the metabolic capabilities of organisms. Apart from predicting a theoretical maximum HA yield, the model also enables the analysis of metabolic fluxes under different conditions. This would lead towards a better understanding of the obscure factors playing a role in production of HA. Further, we will also identify strategies for overproducing HA, through gene deletions as well as over-expression of critical genes.

Methods:

The most recent version of the genome-scale metabolic model of Lactococcus lactis MG1363 was published in 2013 (Flahaut et al.,
Appl Microbiol Biotechnol.
97(19):8729). The metabolic model involves a complete catalogue of all (known) reactions taking place in the organism, complete with stoichiometry and enzyme information. Based on constraints on the flux values of each reaction and an overall cellular objective, FBA involves the formulation of a linear programming problem, which is solved to obtain a set of permissible fluxes that will achieve a given objective (e.g. maximum cellular growth). Since the original model did not specify precise upper and lower bounds for the reaction fluxes, we discarded the flux bounds in the original model and replaced them with some extracellular bounds determined experimentally for the current system of recombinant Lactococcus lactis capable of producing HA. Following this, we evaluated the ability of the model in reproducing trends observed in laboratory experiments. We also systematically analysed the model for possible bottlenecks and over-expression targets that increase flux towards HA synthesis. The aim was to analyse if there are any important reactions that affect flux towards HA production in the entire reactome of the cell, apart from the already known pathway reactions. Using a technique called FSEOF (Flux

Scanning Based on Enforced Objective Flux), the set of reactions whose flux increased upon forced increased of HA flux were collected. FSEOF enforces objective flux to various fractions of maximum HA flux (say 0.2, 0.3, â?¦, 0.8) and performs an FBA to maximise biomass, for each case. Across these simulations, FSEOF identifies reactions that carry an increased flux, as HA production increases. The premise is that these reactions, which uniformly show an increase in flux, with increase in HA production, may be potential over-expression targets. However, there are chances that some of these increased co-incidentally along with increase in HA flux and these need to be weeded out to get the actual over-expression target list.

Results:

The model was able to reproduce the experimentally observed trends of the effect of certain factors on HA. For example, the model predicted that an increase in glutamine uptake flux, a decrease in lactate production flux, an increase in acetate production flux and a decrease in flux towards fructose-6-phosphate individually increased HA formation flux. The stoichiometric effect of currency metabolites on HA was also checked. Increase in ATP formation fluxes in the system showed increase in HA production flux and vice versa.

FSEOF analysis identified several amino acid biosynthesis reactions and nucleotide metabolism reactions apart from HA pathway reactions. The involvement of glutamine in HA pathway could be a possible reason why amino acid biosynthesis reactions were among the identified reactions. When we sought to identify why nucleotide metabolism reactions were in the list, we found that the model used the ribose sugar backbone from nucleotides as the carbon source to make the HA monomers under the conditions where the cell had to maximise HA production. Similarly, the actual effect of the amino acid biosynthesis reactions on HA production are to be evaluated too, to know if apart from increasing glutamine flux, any other amino acid is involved in an indirect way in the HA pathway. A few inorganic reactions such as phosphate uptake were also among the identified reactions. They need to be screened to check if they are co-incidentally increasing.

Our model, like most other genome-scale models, has certain limitations. This model is restricted to analysis of the stoichiometry and metabolic network of the cell alone. In reality, there are other important networks like regulatory networks and signalling networks that play along with the metabolic network in bringing about changes. Another limitation is that we can analyse

fluxes towards HA monomer production alone. The molecular weight cannot be determined from this model.

Overall, our simulations suggest alternate ways to improve the production of HA in L. lactis, through a combination of over-expression of genes in critical pathways such as nucleotide and amino acid metabolism apart from the usual targets in the HA pathway. Our model also provides several hypotheses for experimental verification, which we intend to carry out in the near future.