Obtaining PDC and other high-added value products from lignin by in silico genetic engineering in Novosphingobium aromaticivorans

2.1 Contraint-based modeling

A genome-scale metabolic model, also known as GEM, is a mathematical way of representing a biological system. GEM comprises three main components: the set of reactions denoted as R, the set of metabolites labeled as M, and the stoichiometric matrix named S. A vector v ∈ R ⁿ (where n is the number of reactions) can represent any possible network state.

Each vector v[i] component indicates the flux amount through reaction r _i . The internal metabolites are assumed to be steady, which leads to the metabolite equilibrium equation.

Two values, lb _i and ub _i , are assigned to each reaction r _i to restrict its possible flux.

A mode or state of the network is defined as a flux vector that satisfies equations (1) and (2). Two network states are considered equivalent if one is a non-negative multiple of the other. It is worth noting that these equivalent states will always be identified as the same mode, albeit with a slight abuse of notation. There are always infinite nonequivalent modes in non-trivial networks. The set of all possible network nodes is denoted as C, which forms the flux cone of the metabolic network.

For any given mode v ∈ C, its support, supp(v), is defined as the reactions that appear with non-zero flux in v.

Most models include a set of genes denoted by G. Each reaction r has an associated boolean expression called its gene-protein rule (GPR). The GPR specifies which genes must be activated for the flux through the reaction to be non-zero.

2.2 Coupled growth

The primary objective of metabolic network modeling is to comprehend and regulate the different behaviors exhibited by the network, with the ultimate goal of optimizing the production of bioproducts.

Combining metabolic engineering strategies can maximize flux through bioproduct synthesis. One practical approach is blocking pathways that compete with critical precursor synthesis or consume the bioproducts. A crucial step is identifying interventions that link growth (simulated by a lumped biomass reaction) with producing a specific bioproduct. This means that the synthesis of the product should occur regardless of the growth rate (strong coupling). To achieve this, sets of genes whose deactivation minimizes metabolic flux towards unwanted side products are identified. This enables the production of bioproducts even when cell growth is suboptimal or negligible [20].

The minimal cut set (MCS) method is a technique for identifying possible metabolic intervention strategies by targeting specific reactions. An MCS is the smallest set of reactions that can be inhibited (knocked out) to block certain undesired (target) flux distributions. In our specific problem, we are looking for MCSs that can block all flux vectors without bioproduct production coupled with growth. In simpler terms, we want to prevent any scenario where growth occurs without producing bioproducts.

The original MCS algorithm may not provide workable strategies at the gene level due to the one-to-many nature of GPR rules. Researchers have expanded MCS methods to address this limitation to obtain minimal gene deletion interventions, known as genetic minimal cut sets (gMCS). For further details, refer to [21], 22].

Identifying gMCSs is challenging and resource-intensive, especially for large networks requiring more than two or three gene knockouts. We have used a technique developed by the GACOP group to identify gMCSs [18]. After determining the gMCSs, we create a model by deactivating the genes within these sets. Using FBA techniques, we then analyze the solution space of these mutant strains to understand the effects of these knockouts better.

A second strategy to increase flux towards biosynthesis involves overexpressing rate-limiting enzymes, which produce intermediates specifically for biosynthesis, as well as regulatory enzymes and aromatic transporter proteins. However, genome-scale metabolic models (GEMs) do not permit the overexpression of genes, making it challenging to identify genes whose overexpression could enhance the production of bioproducts. A potential solution to this issue can be found in the Supplementary Material.

2.3 Ranking genetic strategies

Screening and evaluating multiple strategies is essential to determine the best candidates. Schneider and Klamt [19] proposed several criteria for characterizing growth-related intervention strategies. Here, we have adopted some of these criteria for the classification and scoring of the genetic strategies found. These are:

1. Number of interventions required (int): The experimental effort can be reduced by minimizing the required interventions.

2. Biomass yield Y _x/s defined as the ratio of the maximum growth rate to the sum of the substrate fluxes.

3. Minimum product yield at maximum growth rate (min Y _p/s /max x): The point of maximum mutant performance after adaptive evolution. It involves maximizing the growth rate while minimizing the production of bioproducts at a given maximum growth rate.

4. Minimum product yield (min Y _p/s ): The product yield in the mutant when growth is sub-optimal.

5. Requirement for anaerobic conditions (%O2): To make scaling up growth easier, the mutant strain should consume less oxygen than the wild-type strain.

6. Number of accessible metabolites: When a disturbance in the middle or end of a pathway can cause a buildup of unwanted intermediate metabolites. To prevent this, it is better to interrupt the pathways at their branching points, limiting the overall number of metabolites produced.

Each intervention has been given a score based on specific criteria [19], as explained in the Supplementary Material.

2.4 iNovo479 model of N. aromaticivorans

All data and the GEM used for metabolic analysis of N. aromaticivorans were implemented in [17].

iNovo479 was constructed using KEGG annotations predicted for the N. aromaticivorans DSM 12444 genome and annotation data available in the Integrated Microbial Genomes database. The model consists of 479 genes, 645 reactions, and 604 metabolites. Aromatic metabolism reactions and missing reactions in vital metabolic pathways were added manually based on the knowledge available in the literature. The biomass reaction in iNovo479 has been modified from the aerobic biomass reaction present in iRsp1095, a GEM created for another α-proteobacterium called Rhodobacter sphaeroides. The modification was done by incorporating laboratory-determined biomass composition data. Moreover, a non-growth-associated ATP maintenance reaction (NGAM) has been added to this model (refer to Figure 1 for details).

Figure 1:

Model scheme.

Metabolic engineering is necessary for the accumulation of bioproducts since, in the wild-type strain, all the energy obtained from the carbon source is devoted to the growth and maintenance of the organism, not to the biosynthesis of these products. Linz et al. [17] applied a genome engineering strategy [13] for the accumulation of 2-pyrone-4-6-dicarboxylic acid (PDC), a compound with remarkable functional properties that is replacing terephthalic acid in polyester manufacturing [23]. On the other hand, a modified iNovo479 model was generated by introducing reactions that produced potentially valuable chemicals from intermediates generated by the metabolism of aromatics, such as glutarate, zeaxanthin, citrate, or propanoate.

The results obtained in [17] have been reproduced using COBRA and dFBA. The results we obtained agree in most cases, although there are slight differences in the rate of PDC production when the ratio of aromatic:glucose is 1:1 and 9:1.

Combining metabolic engineering strategies maximizes bioproduct synthesis by blocking competing pathways and consuming critical precursors.

To find gMCS for growth-coupled PDC production, glucose and vanillic acid were used as substrates. According to [13], the microorganism cannot produce PDC by using vanillic acid as the sole carbon source in the deletion strain. A proportion of glucose must be supplied to allow the microorganism to grow and produce PDC simultaneously. Even if a source of glucose is provided, the model does not use vanillic acid by default when its pathway to central carbon metabolites is blocked. We suggest its entry through a restriction based on a three-to-two molar aromatic-to-glucose ratio excess. This is the ratio at which the highest production rate is obtained [17].

In this paper, we use a modified model [17] to obtain genetic engineering strategies leading to an accumulation of bioproducts of industrial interest, including glutarate, acetaldehyde, citrate, glycerol, phenol, and propionate.

3.1 New genetic interventions to accumulate PDC from vanillic acid and glucose

In iNovo479, Linz et al. [17] introduced a technique to block the ligI gene for PDC accumulation using vanillic acid and glucose as substrates.

Our objective was to investigate new genetic interventions that were not previously explored. We have discovered seven intervention strategies (gMCS) with at least one and a maximum of 4 interventions. Table 1 displays the biomass yield, product yields, and the change in oxygen consumption compared to the wild-type conditions for each knockout. The initial row represents the wild-type (WT) phenotype, while the remaining rows indicate the intervention gene(s) to be deleted.

Figure 1 in Supplementary Material shows the scores of all interventions found for each bioproduct according to the criteria extracted from [19].

3.2 Obtaining other bioproducts of interest by modified strains of N. aromaticivorans from vanillic acid

In the modified iNovo479 model, vanillic acid is the sole carbon source (engineering strategies do not interrupt the degradation pathway from vanillic acid to the TCA cycle). It uses a series of enzymatic reactions to synthesize bioproducts of chemical interest from intermediates that the metabolism of aromatics could generate. The model does not tend to accumulate bioproducts as the energy derived from the carbon source is directed towards supporting growth. Therefore, to ensure minimal production, knockouts must be performed based on calculated gMCS.

We have found new genetic interventions to enable the production of 6 out of the 13 bioproducts [17]: acetaldehyde, citrate, glutarate, glycerol, phenol, and propanoate. Table 2 summarizes the corresponding results. This table highlights the number of interventions (gene knockouts) for each product along with the highest Y _p/s at the maximum growth rate and lowest Y _p/s values obtained.

The results for knockouts of genes for these bioproducts are available in the Supplementary Material.

3.3 Genetic interventions to accumulate PDC from vanillic acid and glucose

The pathway for PDC production is as follows: vanillic acid is converted to protocatechuate (PCA) by tetrahydrofolate-dependent O-demethylase (LigM) and catabolized via the PCA cleavage pathway. In this pathway, the aromatic ring of PCA is cleaved by PCA 4,5-dioxygenase (LigAB) to generate 4-carboxy-2-hydroxymuconate-6-semialdehyde (CHMS), which is then converted to 2-pyrone-4,6-dicarboxylate (PDC) by CHMS dehydrogenase (LigC). PDC is catabolized to oxaloacetate and pyruvate by reactions catalyzed by enzymes (LigI, LigU, LigJ, and LigK). The biosynthesis pathway is shown in Figure 2.

Figure 2:

PDC synthesis pathway from vanillic acid in Novosphingobium aromaticivorans. Reactions blocked by gMCS 1, 2, 3, and 4 are marked with a red cross.

One of the genetic strategies obtained for PDC production is the inactivation of the ligI gene (gMSC1, Figure 2), which controls the reaction catalyzed by 2-pyrone-4,6-dicarboxylate hydrolase (LigI) and converts PDC to 4-oxalomesaconic acid (OMA). It also participates in the degradation of the latter compound. This strategy was also implemented [13], 17] for PDC accumulation using the same carbon source: vanillic acid and glucose.

Three other interventions involving the inactivation of a single gene have been obtained. Two of these genes are ligK and ligJ (gMSC2 and gMSC3, respectively), whose function is to control the reaction catalyzed by LigK (4-carboxy-4-hydroxy-2-oxoadipate aldolase) and an amidohydrolase (LigJ). Both are involved in the degradation of OMA to oxaloacetate and pyruvate. Although gMSC2 and gMSC3 have slightly higher scores than gMSC1, they have more accessible metabolites. LigK and LigJ disrupt the PDC-to-oxaloacetate pathway at an intermediate point, not at the branch point, as LigI does.

Blockage of ligI (gMSC1) ligK (gMSC2) or ligJ (gMSC3) targets the direct accumulation of PDC or its downstream metabolite, OMA, to decrease the flux of PDC catabolism into the tricarboxylic acid (TCA) cycle and increase its accumulation.

The fourth intervention is the inactivation of pckA gene. This gene encodes phosphoenolpyruvate carboxykinase (PEPCK), which catalyzes the oxaloacetate to phosphoenolpyruvate (PEP) reaction. This intervention results in a higher biomass yield as it does not directly interrupt the flow of PEP into the TCA cycle. However, it has a lower overall score as the product yield is lower. This intervention targets the accumulation of oxaloacetate and the decrease of PDC flux towards this compound.

Perez et al. [13] showed experimentally that disruption of ligI gene is sufficient to prevent PDC catabolism. Here, it has been proposed that, in addition to deletion of this gene, deletions of genes involved in the OMA to oxaloacetate pathway (ligK and ligJ) or decreased flux of PDC degradation (pckA) could also serve as a genetic strategy to maximize PDC production from vanillic acid.

As shown in Table 1, the strategy of blocking ligK or ligJ, as opposed to blocking ligI, produces a scenario in which a higher biomass yield (Y _x /s) is observed, although with a slightly lower PDC yield (Y _p /x). This could be because, by interrupting ligK- or ligJ-catalyzed reactions, the PDC degradation pathway is halted at an intermediate point, allowing the accumulation of downstream metabolites-primarily oxalomesaconic acid (OMA), which, instead of being wholly channeled into the TCA cycle, could be used to increase biomass yield (Y _x /s) by incorporation into alternative metabolic pathways, thus compensating for the observed slight decrease in PDC yield (Y _p /x).

In other words, although part of the substrate is “lost” in the form of OMA (and, potentially, 2-keto-4-carboxyl-hexanedioic acid), these intermediate metabolites could serve as a carbon source for growth. Furthermore, according to the ranking criteria used in the study [19], strategies based on ligK and ligJ inactivation score higher than ligI, as the balance between product yield and growth robustness is overall more favorable.

In summary, the preference for blocking ligK or ligJ lies in the fact that, although some of the substrate accumulates as downstream metabolites (which could be used as a carbon source for growth), the net effect is an increase in biomass, which favors the stability and efficiency of the PDC production process. It is important to emphasize that this hypothesis would require experimental validation to confirm whether, indeed, these metabolites can be redirected to provide energy for cell growth.

In addition, three interventions involving four gene knockouts have been found. These three interventions have three genes in common: sucC, sucD, and cysE.

Succinyl-CoA synthetase or succinate-CoA ligase is an enzyme that catalyzes the reversible reaction of succinate to succinyl-CoA in the TCA cycle. The sucD gene encodes the alpha subunit, while sucC encodes the beta subunit.

On the other hand, cysE encodes the enzyme serine O-acetyltransferase. This enzyme catalyzes the reversible reaction of serine with acetyl-CoA to give O-acetyl-l-serine. On the other hand, serine biosynthesis occurs via l-serine ammonium-lyase, with ammonium and pyruvate as substrates (Figure 3).

Figure 3:

PDC synthesis pathway from vanillic acid in Novosphingobium aromaticivorans. Reactions blocked by gMCS 5,6, and 7 are marked with a red cross. S-AMDHP: S-aminomethyldihydrolipoylprotein; THF: tetrahydrofolate; mTHF: 5,10-methylenetetrahydrofolate; DHP: dihydrolipoylprotein; PCA: protocatechuic acid; DE3M: d-erythro-3-methylmalate; R2MM: R-2-methylmalate; a-KB: alpha-ketobutyrate; G6P: glucose-6-phosphate; F6P: fructose-6-phosphate; F16B: fructose-1,6-biphosphate; DHAP: dihydroxyacetone phosphate; G3P: glyceraldehyde 3-phosphate; 13BG; 1,3-biphosphoglycerate; 3PG: 3-phosphoglycerate; 2PG: 2-phosphoglycerate; PEP: phosphoenolpyruvate.

The other genes that appear (one in each genetic intervention) belong to the same biological process: the glycine cleavage system (GCS). This multi-enzyme system composed of four proteins (P, H, T, and L) catalyzes the reversible oxidation of glycine [27].

The gcvT gene (gMCS5) encodes the tetrahydrofolate aminomethyltransferase or T protein. It acts on GCS secondarily by degrading aminomethyl to ammonia in the presence of tetrahydrofolate (THF). On the other hand, the gcvPB and gcvPA genes (gMCS6 and gMCS7, respectively) encode lipoyl-protein oxidoreductase (P-protein), the first enzyme of the GCS system. With the deletion of succinyl-CoA synthetase, there is a decrease in the flux of the TCA cycle (Figure 3). This results in the accumulation of metabolites targeted to this cycle, including PDC and the metabolites included in its degradation pathway. The flux towards serine biosynthesis from pyruvate increases, as does glycolysis.

On the other hand, the formation of threonine from aspartate decreases, making synthesis rather than degradation to keto butyric acid the predominant fate of this essential amino acid protein. The route of its biosynthesis via pyruvate and acetyl-CoA is activated to recover the latter compound. The O-demethylation of vanillic acid by LigM uses THF as a cofactor (Figure 2), so the deletion of the gvcT gene and the consequent absence of the T protein will result in the accumulation of THF, thus increasing its flux towards vanillic acid demethylation. In addition, deletion of the P protein will block this pathway from the start and similarly lead to increased THF availability.

These three interventions result in higher biomass and product yields with lower oxygen consumption, outperforming single-gene interventions.

In gMCS5/6/7 strategies, the deletion of genes such as sucC, sucD, gcvT (or its variants gcvPB/gcvPA) and cysE disrupts the normal flow of the Krebs cycle, preventing metabolites downstream of PDC from being channeled to central metabolism. Thus, PDC accumulation would be favored, as the reactions that generally degrade or transform these metabolites are blocked.

In this scenario, it is essential to remember that vanillate is the primary source of PDC formation, whereas glucose is used exclusively for cell growth. Thus, even though glucose-derived pyruvate is not channeled into the Krebs cycle due to the blockage of this pathway, it can be employed through these alternative pathways to generate energy and, consequently, favor biomass synthesis.

Simulations predict that, in gMCS5/6/7, by preventing complete degradation of PDC, higher minimum PDC yields (min Y _p /s) are obtained compared to other strategies. This parameter is defined as the minimum amount of product (in this case, PDC) obtained per mole of substrate consumed, measured under suboptimal growth conditions or at the point of maximum growth rate. This indicator is crucial because it ensures that, even in the least favorable scenarios for production, a minimum threshold of bioproduct synthesis is reached, allowing the robustness and feasibility of the metabolic engineering strategy to be assessed.

However, it is important to underline that this hypothesis is based on the simulations performed and the interpretation of the in silico models. It would be exciting for future studies to have experimental parts or simulations presenting metabolite profiles-for example, showing the fraction of vanillate converted to PDC in gMCS5/6/7 versus other strategies-to confirm whether, indeed, the accumulation of downstream metabolites translates into increased PDC formation and whether those metabolites can contribute to the generation of energy for growth.

3.4 Glutarate production

Glutaric acid is a metabolite of l-lysine catabolism in Pseudomonas species, in which l-lysine is converted to glutaric acid via the 5-amino valeric acid (VAA) pathway (Table 3). In order for N. aromaticivorans to produce this compound, four reactions involving four enzymes were introduced into iNovo479: a 2-oxidoreductase, a 5-aminopentanamide, an amidohydrolase, a 2-oxoglutarate aminotransferase, and an NAD+ oxidoreductase. From this modified GEM, genetic strategies for glutarate production were sought. As a result, three interventions with four genes per intervention were obtained. These three strategies are the same as those found for PDC production: deletion of the sucC, sucD, cysE and gcvT, gcvPB or gcvPA genes.

Succinyl-CoA is an essential compound in multiple metabolic pathways, including the TCA cycle and the lysine biosynthesis pathway shown in Figure 4. While it is possible for Esterichia coli to survive without succinyl-CoA, these cells have a more prolonged latency phase and slower growth rate, highlighting the importance of this compound in the TCA cycle [28].

Figure 4:

Glutarate synthesis pathway from vanillic acid in Novosphingobium aromaticivorans. Reactions blocked by gMCS 1,2 and 3 are marked with a red cross. HTHDP: (2S,4S)-4-hydroxy-2,3,4,5-tetrahydrodipicolinate; THDP: 2,3,4,5-tetrahydrodipicolinate; SAOH: N-succinyl-2-l-amino-6-oxoheptanedioate; SDAH: N-succinyl-LL-2,6-diaminoheptanedioate; DAH: LL-2,6-diaminoheptanedioate; mDAH: meso-2,6-diaminoheptanedioate; 5-APA: 5-aminopentanamide; 5-AP: 5-aminopentanoate; OG: 2-oxoglutarate; 5-OP: 5-oxopentanoate; S-AMDHP: S-aminomethyldihydrolipoylprotein; THF: tetrahydrofolate; mTHF: 5,10-methylenetetrahydrofolate; NH3: ammonia; DHP: dihydrolipoylprotein.

Inactivation of these two genes could lead to a decrease in succinyl-CoA synthetase activity, accumulation of succinyl-CoA, and an increase in intracellular l-lysine available for glutaric acid production. Deletion of cysE would lead to an accumulation of serine, a decrease in its biosynthesis, and an increased bioavailability of ammonium and pyruvate. The ammonium influx would be directed to the synthesis of l-glutamate, a precursor of lysine. As in PDC production, the three genes that do not match in the interventions belong to the GCS biological process and aim to increase the bioavailability of THF.

On the other hand, all three strategies score equally (Figure 1 in Supplementary Material), although it is worth noting that strains with ΔgcvPA and ΔgcvPB with loss of function of Protein P have a lower number of accessible metabolites.

3.5 Citrate production

We have been able to obtain three strategies for citrate that are identical to those used for glutarate. Citrate is formed in the first reaction of the TCA cycle from oxaloacetate, acetyl-CoA, and water by citrate synthase (Table 4). With the deletion of serine O-acetyltransferase (cysE), acetyl-CoA accumulation would occur. In contrast, the deletion of succinyl-CoA synthetase (sucC and sucD) reduces the flux of the TCA cycle, and metabolites preceding succinyl-CoA in this cycle (including citrate) accumulate. On the other hand, deletion of the first two genes involved in the GCS will lead to THF accumulation. Citrate synthetase is inhibited by succinyl-CoA, which resembles acetyl-CoA and acts as a competitive inhibitor of acetyl-CoA and a non-competitive inhibitor of oxaloacetate, so blocking succinyl-CoA synthetase could reduce the yield of citrate production [29].

3.6 Propionate production

Propionic acid (propionate) is a commercially valuable carboxylic acid produced through microbial fermentation (Table 5). Propionic acid is mainly used in the food industry but has recently found applications in the cosmetics, plastics, and pharmaceutical industries. Propionate can be produced through several metabolic pathways, classified into three main groups: fermentative, biosynthetic, and amino acid catabolic pathways [30].

The metabolic pathway that allows propionate synthesis in Novosphingobium arovativorans is the degradation of threonine. This amino acid, synthesized from aspartate, is degraded to alpha-ketobutyrate and subsequently to propionate. In turn, aspartate can be obtained from oxaloacetate by 2-oxoglutarate aminotransferase.

Twenty-nine interventions have been obtained, two of which are from a single gene and the rest from four genes (Figure 5).

Figure 5:

Propionate synthesis pathway from vanillic acid in Novosphingobium aromaticivorans. Reaction blocked by gMCS2 is marked with a red cross.

The gMSC with the highest score is gMSC2 (Figure 1 in Supplementary Material), corresponding to deleting the pckA gene. Blocking this gene, which catalyzes the reaction of oxaloacetate to PEP by PEPCK, would lead to an accumulation of oxaloacetate and an increased flux toward aspartate synthesis. This intervention results in a higher biomass yield and a considerable increase in product yield compared to other gMSCs, although oxygen consumption is somewhat higher than under wild-type conditions. In C. glutamicum this gene was blocked for oxaloacetate and aspartate accumulation [31].

The other gMSC involving a single gene is gMSC1 (Δadk) which encodes adenylate kinase (AK). The kinases represent a ubiquitous family of catalysts that play key roles in cellular metabolism, including the synthesis of RNA and DNA molecules and in energy metabolism. AK catalyzes the reversible transfer of the phosphate group between ATP and AMP, so its blockage leads to increased production of AMP, a metabolite required in the propionate synthesis reaction. AK is known to be essential for cell growth and survival [32], so blocking this gene may compromise the survival of N. aromaticivorans.

3.7 Acetaldehyde production

Acetaldehyde, like propionate, can be obtained through the metabolism of the amino acid threonine but with the action of different enzymes (Table 6). In this case, the action of treoninaldolase, an enzyme capable of splitting this amino acid into glycine and acetaldehyde, is involved. Threonine, in turn, is synthesized from aspartate.

The highest scoring gMCS is gMCS2 (pckA (Figure 1 in Supplementary Material), which encodes PEPCK. Its deletion will lead to an accumulation of oxaloacetate, an increase in aspartate synthesis by the action of 2-oxoglutarate aminotransferase, and a consequent increase in flux to threonine synthesis.

Another proposed strategy is to block the glycine formation reaction (gMSC3, glyA). This increases the flux towards its synthesis from threonine metabolism, obtaining acetaldehyde from its degradation.

3.8 Glycerol production

Twenty-four strategies for maximizing glycerol synthesis are obtained, all formed by cutsets of 4 genes. The highest scoring is gMCS11 (Figure 1 in Supplementary Material), with the highest minimum product yield at maximum biomass (Table 7).

The gene repeated in all interventions is pyk. This gene encodes pyruvate kinase, which catalyzes the reaction of pyruvate formation from PEP. Glycerol is formed from 2-phospho-d-glycerate, and the action of 2-phospho-d-glycerate hydro-lyase with PEP as substrate synthesizes this metabolite. Therefore, pyk blockage targets the accumulation of PEP and an increase in glycerol synthesis. The only amplification target genes lead to the synthesis of sn-glycerol-3-phosphate, the glycerol precursor (Table 12 in Supplementary Material).

3.9 Phenol production

Twenty-four strategies consisting of cutsets of 4 genes are obtained. The highest scoring is gMCS18 (Figure 1 in Supplementary Material), with the highest minimum product yield at maximum biomass. As in glycerol production, the gene repeated in all interventions (maximum overlap) is pyk, which encodes pyruvate kinase. Phenol is formed from tyrosine from the reaction catalyzed by the heterologous enzyme l-tyrosine phenol-lyase. Tyrosine is synthesized through a series of reactions starting with the reaction of PEP with Shikimate 3-phosphate, catalyzed by 3-phosphoshikimate 5-O-(1-carboxyvinyl)-transferase. The blockage of pyk and the resulting accumulation of PEP will increase this compound’s flux towards tyrosine and phenol formation (Table 8).

In all bioproduction strategies, there is an accumulation of formate, a metabolite resulting from the demethylation of vanillic acid. This format forms reducing power (NADH) through the reaction catalyzed by an NAD+ oxidoreductase, thus compensating for reducing carbon flux by the TCA cycle. In addition, decreasing power is required to form some bioproducts such as glycerol (Figure 6).

Figure 6:

Glycerol synthesis pathway from vanillic acid in Novosphingobium aromaticivorans. 2-PDG: 2-phospho-d-glycerate; 3-PDG: 3-phospho-d-glycerate; 3-PDGP: 3-phospho-d-glycerol phosphate; DG3P: d-glyceraldehyde 3-phosphate; GP: glycerone phosphate; G3P: sn-glycerol 3-phosphate; Pi: orthophosphate.

3.10 General discussion

In this project, vanillic acid has been used as a carbon source for synthesizing bioproducts in the same way as implemented in [17]. However, there is a wide range of aromatic compounds that can be obtained from lignin depolymerization besides vanillic acid: p-hydroxybenzoic acid, syringic acid, syringaldehyde, protocatechuic acid, vanillin, ferulic acid, etc. For microbial valorization, the depolymerization of lignin into these low molecular weight compounds is crucial to ensure their bioavailability for further valorization by N. aromaticivorans. On the other hand, it is also of utmost importance to tailor depolymerization strategies to produce a suitable product spectrum. For N. aromaticivorans this spectrum is quite broad: it has been experimentally demonstrated to degrade up to 17 different aromatic compounds [33].

We found 120 intervention strategies (gMCS) ranging from a minimum of 1 to a maximum of 4 genes that guarantee a minimum production of bioproducts while growing the N. aromaticivorans strain. Three of these genetic interventions are repeated to produce four different bioproducts. These strategies correspond to the blocking of genes encoding serine O-acetyltransferase (cysE), succinyl-CoA synthetase (sucC and sucD), and the P-protein (gcvPA and gcvPB) or T-protein (gcvT) of GCS (see Figure 7), which are common in the production of PDC, glutarate, citrate and propionate. These strategies target the accumulation of TCA cycle intermediates, a decrease in TCA cycle flux, and an increase in THF bioavailability. The other strategy is blocking the pckA gene that encodes phosphoenolpyruvate carboxykinase, which is involved in producing PDC, propionate, and acetaldehyde. This strategy seeks to accumulate oxaloacetate, either to decrease the flux from PDC to oxaloacetate (in the case of PDC) or to increase the flux to threonine synthase, the precursor of acetaldehyde and propionate.

Figure 7:

Genetic strategies that are common in different bioproducts.