Muconic acid production from glucose and xylose in Pseudomonas putida via evolution and metabolic engineering – Nature.com

Introducing the d-xylose isomerase pathway into muconate-producing P. putida

Three xylose metabolic pathways were considered to enable the production of muconate from this substrate36, including the isomerase pathway in which xylose is metabolized to xylulose-5-P (X5P) in the pentose phosphate pathway (PPP)38, the Weimberg pathway that feeds xylose to the TCA cycle via -ketoglutarate38,39, and the Dahms pathway40, which shares the initial three steps with the Weimberg pathway, after which -ketoglutaric semialdehyde is converted by an aldolase into pyruvate and glycolaldehyde. Among these, the d-xylose isomerase pathway, in which xylose is metabolized via the d-xylose isomerase (xylA) and xylulokinase (xylB) to xylulose-5-phosphate (X5P), is ideal for achieving a high theoretical muconate yield since X5P can be further converted to E4P and subsequently enter the shikimate pathway (Fig.1a)35. We integrated the isomerase pathway into a strain previously engineered to produce muconate from glucose, CJ5223, by overexpressing codon-optimized versions of the E. coli d-xylose isomerase (xylA), xylulokinase (xylB), and d-xylose:H+ symporter (xylE), together with a transaldolase (tal) and a transketolase (tkt) to improve carbon flux within the PPP (Fig.1a)35. We also deleted hexR, which encodes a transcriptional regulator that controls expression of genes important for sugar metabolism, since we had previously found this to improve the conversion of glucose to muconate32.

Thompson et al. previously reported that employing both the asbF and aroE pathways can help to maximize net precursor assimilation and metabolite flux toward muconate25. Thus, an engineered chorismate pyruvate-lyase (ubiC-C22)41 with relieved product inhibition was integrated to enhance muconate production through the shikimate pathway via aroE (Fig.1a). We had previously deleted pgi-1 and pgi-2, which encode redundant glucose-6-P isomerases, to disrupt the EDEMP cycle, a combination of the Entner-Doudoroff, gluconeogenic Embden-Meyerhoff-Pernass, and the pentose phosphate pathways42. The purpose of disrupting the EDEMP cycle is to prevent it from cycling to generate pyruvate independent of PEP during growth on glucose, which could enable the cell to redirect carbon toward growth at the expense of muconate production, despite deletion of the genes encoding the pyruvate kinases (pykA, pykF) and PEP carboxylase (ppc)3. This strategy is beneficial for muconate production from glucose as the sole carbon source, but in this case, deletion of pgi-1 and pgi-2 would decrease the maximum theoretical muconate yield of both asbF- and aroE-catalyzed muconate biosynthesis pathways when xylose is converted via the PPP (Fig.1b).

Considering that the xylose fraction in the mixture of glucose and xylose (xylose/glucose+xylose%, moles) in corn stover hydrolysate ranges from 34 to 38% (Supplementary Fig.1), the modeling predicted maximum theoretical yield of muconate with pgi-1 and pgi-2 deleted to be lower than if one or both are present (Fig.1b). To test the hypothesis that glucose-6-phosphate isomerase (encoded by pgi-1 and pgi-2) activity is necessary for xylose flux to enter the EDEMP cycle, we built strains based on JE322635, a P. putida KT2440-dervied strain that was previously engineered to utilize xylose using the d-xylose isomerase pathway but is otherwise wild-type, generating strains LC041 (pgi-1), LC345 (pgi-2), LC347 (pgi-1 pgi-2). In plate reader cultivation on M9 medium containing xylose, LC347 failed to grow, whereas both LC041 and LC345 demonstrated reduced growth rates and increased growth lags (Supplementary Fig.2). LC345, with pgi-1 intact, exhibited a lower growth rate and longer growth lag compared to LC041, which contains only pgi-2, suggesting that Pgi-1 contributes less to the overall glucose-6-phosphate isomerase activity than Pgi-2. Since the EDEMP cycle would be expected to compete with muconate biosynthesis and reduce the muconate yield, we thus restored pgi-1 to enable xylose flux into the EDEMP cycle and improve the maximum theoretical yield, generating strain QP328 (Fig.1a and Table1).

Strain QP328 was cultivated in shake flasks on a mixture of glucose and xylose to examine their conversion to muconate. Although the xylose isomerase pathway has been shown to be efficient in wild-type P. putida35, the xylose utilization rate of QP328, however, was very low compared to that of glucose (Fig.1c). Since glucose and xylose can be utilized simultaneously in the P. putida KT2440 wild-type background upon introduction of the same xylose isomerase pathway35, we hypothesized that a bottleneck in xylose transport or metabolism was present in our muconate-producing strain.

To improve xylose utilization by QP328, we conducted ALE by serial passaging of the strain on M9 medium supplemented with 10mM xylose as a sole carbon and energy source. As the populations were passaged, higher OD600 values were achieved more rapidly. After 7 passages (~50 generations), all four lineages achieved turbidity in 24 days compared to 14 days at the beginning of the ALE, and the evolution was terminated. The evolved populations of the four lineages were plated onto an LB agar plate, and three isolates on each plate were chosen for shake flask pre-screening (12 isolates in total). In most cases, all triplicates from the same lineage exhibited similar growth and muconate production, so it was assumed that they likely represented the same genotype and only one from each lineage was saved. In lineage 1, however, one replicate performed differently, thus two isolates were saved (Supplementary Fig.3). To identify mutations that may contribute to improved xylose utilization, the genomes of all five isolates were sequenced. All five isolates had mutations in xylose transporter xylE (A62V, A62V and A455V, T34I, L214P, S205F, for isolates 1, 2, 3, 4, 5, respectively). Four of the isolates (1, 35) had mutations that likely inactivated aroG-D146N (frameshift +7bp, frameshift +2bp, M1N and L2H, frameshift 16 bp, for isolate 1, 3, 4, 5, respectively) (Fig.2a). The five isolates were then evaluated in shake flasks on glucose, xylose, and a mixture of glucose and xylose. As expected, strains with mutations in aroG-D146N grew better but produced less muconate. Isolate 2 achieved the highest muconate yield and the lowest biomass yield, and was designated QP478 (Supplementary Fig.4). QP478 demonstrated substantially improved growth on xylose compared to QP328 in a plate reader, in which the growth of QP328 was negligible while QP478 reached a OD600 of 0.5 in 72h (Fig.2b).

a Mutations identified in ALE by whole-genome sequencing of the five isolates (the Sankey diagram was built using SankeyMATIC online tool). b Growth curve of reverse engineered strains LC091 and LC100, with the unevolved strain QP328 and the evolved strain QP478 for comparison. A represents absolute growth rate, and all A presented here are the average values of three independent growth curves. c Shake-flask experiments of reverse engineered strains LC091 and LC100, comparing to QP478, on M9 medium supplemented with 30mM xylose. Yield of LC100 was compared to LC091 using two-tailed Student t test (P<0.0001). d Shake flask experiments comparing thereverse engineered strains LC091and LC100, and theevolved strain QP478 on M9 medium supplemented with 30mM glucose and 15mM xylose. % Molar yield was calculated as [mM muconate/mM (glucose+xylose)100], and% carbon yield was calculated as [mM muconate6/mM (glucose6+xylose5)100]. Error bars here represent the standard deviation of three biological replicates. Source data are provided as a Source Data file.

The mutations identified in QP478 are listed in Supplementary Data1. Of those, mutations we hypothesized might be related to the improved growth on xylose included: (1) two missense mutations in the xylose transporter gene, xylE, where alanine residues were replaced with valines, A62V and A455V; (2) a GA point mutation 10bp upstream of the 35 element of a putative promoter (Supplementary Fig.5) predicted by the BPROM 70 promoter prediction program43 upstream of PP_2569, which is annotated as a metabolite major facilitator superfamily (MFS) transporter in the Uniprot database; and (3) a 227.8 kB region of the genome from PP_5050 to PP_5242 that appeared to be duplicated (Fig.2a).

To understand the contribution of the mutations that led to improved growth on xylose during ALE, we created strains that individually restored the wild-type sequences into the evolved strain QP478. The A62V and A455V mutations were restored to wild type separately in xylE, generating LC093 and LC078, respectively. The GA mutation in the promoter region of PP_2569 was restored, generating LC061. In plate reader experiments, restoring either xylE-A455V or xylE-A62V led to decreased growth rate and increased growth lag of LC078 and LC093, respectively (Supplementary Fig.6ac). The restoration of the GA mutation in PPP_2569 led to slightly decreased growth rate (Supplementary Fig.6a, b). In shake flasks, only LC093 demonstrated significantly lower muconate yield compared to QP478 (Supplementary Fig.6f, g, j), but all three strains LC061, LC078, and LC093 exhibited slower growth and muconate production (Supplementary Fig.6gj), which is consistent with the results from the plate reader experiments (Supplementary Fig.6ac). The reduced muconate productivity, caused by decreased growth rates and/or increased growth lag, indicated that all these mutations contributed to improved cell growth on xylose of QP478 (Supplementary Fig.6).

We also performed the reverse experiment, engineering the ALE mutations into the parent strain QP328 to obtain a rationally engineered strain containing only mutations that contribute to improved production of muconate. We first reverse engineered the unevolved strain QP328 with the three point mutations. The A62V and A455V XylE mutations were introduced into the unevolved strain QP328, generating LC091. The GA mutation in PPP_2569 was engineered in QP328 and LC091, generating LC092 and LC100, respectively. Strains LC091, LC092, and LC100, together with QP328 and QP478, were evaluated in a plate reader containing M9 medium with 30mM xylose. Interestingly, introducing the two XylE mutations enabled cell growth on xylose in LC091 (Fig.2b), which exhibited a comparable growth rate but higher final biomass compared to QP478 on xylose alone and xylose and glucose mixture (Fig.2b). Introducing the GA mutation in PPP_2569 to QP328 also enabled cell growth of LC092 on xylose, while at a much lower rate compared to LC091 (Fig.2b). Unexpectedly, introducing the GA mutation in PPP_2569 to LC091 led to decreased growth and lower biomass of LC100 on both xylose and mixed substrates (Fig.2b).

We next evaluated LC091, LC100, and QP478 in shake flasks containing M9 medium with 30mM xylose. LC091 reached almost twice the biomass yield (OD600) but achieved lower muconate yield compared to QP478 (Fig.2c). Moreover, LC100 reached a comparable muconate yield to QP478, albeit at a lower rate (Fig.2c). RT-qPCR indicated that the GA mutation in PPP_2569 increased the expression of PP_2569 (Supplementary Fig.7). We also evaluated LC091, LC100, and QP478 on M9 medium with a mixture of glucose and xylose. Consistent with the results on xylose only, the muconate yield of LC091 is lower than LC100 and QP478 on the mixture; LC100 still exhibited much slower growth than QP478 on the mixture, though it utilized glucose and xylose simultaneously and reached comparable muconate yield (Fig.2d). The difference in these strains suggested that a gene or genes in the PP_5050PP_5242 duplicated region might be important for the improved performance of QP478.

The only genetic difference between strains LC091 and LC100 is the GA mutation in the promoter region that led to higher expression of a putative MFS transporter PP_2569 (Supplementary Fig.5). To rationalize how the mutation could occur in ALE, and to examine how metabolism could be affected by PP_2569, intracellular and extracellular metabolomics experiments were conducted with LC091 and LC100 grown on xylose. Selected metabolites from early log phase, mid-log phase, and late log phase were analyzed (Supplementary Fig.8), and Z-scores were plotted as Fig.3a. Generally, we observed that LC091 accumulated more metabolites in both the EDEMP pathway and the TCA cycle, both intracellularly and extracellularly, while LC100 demonstrated a greater accumulation of shikimate pathway-related metabolites (Fig.3a). There were three prominent exceptions of shikimate pathway-related compounds that were more abundant in LC091, namely l-tyrosine, l-phenylalanine, and l-tryptophan, which are all chorismate-derived aromatic amino acids (Fig.3bd). In Pseudomonas aeruginosa, it was previously reported that l-tyrosine and l-tryptophan could strongly inhibit the native DAHP synthases AroF-1 and AroF-244. Thus, we posit that AroF-1 and AroF-2 could be less inhibited in LC100 compared to LC091. Although a feedback-resistant DAHP synthase aroGD146N has been overexpressed in our strains, previous studies showed that the native DAHP synthases AroF-1 and AroF-2 alone led to ~30% PCA and phenol production compared to additional overexpression of aroGD146N, in P. putida45 and Pseudomonas taiwanensis46, respectively. We reasoned that the improved muconate production of LC100 compared to LC091 might be because of lower inhibition of AroF-1 and AroF-2 (Fig.3be), and the reduced PEP concentration and increased accumulation of DAHP in LC100 support this interpretation (Fig.3a).

a Heatmap of the selected intracellular and extracellular metabolites. Z-scores were calculated using the average intensities of intracellular and extracellular metabolites separately, and a time zero control from uninoculated medium was also included for extracellular metabolite analysis. be Intensities of selected metabolites from late log phase, In represents intracellular, Ex represents extracellular, and N.D. represents not detected. The intracellular intensity signal was collected using the cell pellet from a 1mL cell culture, and the extracellular intensity signal was collected using 20L of the corresponding supernatant. f Structure of XylE (PDB ID: 4GBY)48 with residues of ALE-derived mutation locations labeled in green. d-xylose is shown in black ball-and-sticks. g Growth curve of strain LC111 compared to JE3692 on M9 medium supplemented with 30mM xylose. Source data are provided as a Source Data file.

Moreover, a further explanation could be that PP_2569 is able to transport the aromatic amino acids extracellularlyhowever, we did not observe the extracellular accumulation of these amino acids (Fig.3bd). Instead, we found substantial extracellular accumulation of anthranilic acid in LC100 (Fig.3e). Anthranilic acid is a precursor of l-tryptophan and direct product of chorismate, which is an important node in the shikimate pathway from which all of the aromatic amino acids are derived. Chorismate was reported to be unstable intracellularly47 and was not detected in our intracellular metabolomics samples. Based on the current results, we posited that PP_2569 might be able to export anthranilic acid, which could lead to decreased l-tryptophan intracellular accumulation (Fig.3d, e). The increased accumulation of anthranilic acid may also reduce the flux from chorismate toward other aromatic amino acids, leading to the reduced accumulation of l-tyrosine and l-phenylalanine in strain LC100. Further work is warranted to investigate the mechanistic basis of this beneficial ALE-derived mutation.

Separately, to investigate the mechanism of mutations in XylE, five mutations from isolates 15 (A62V, A62V, and A455V, T34I, L214P, S205F, for isolates 1, 2, 3, 4, 5, respectively) were labeled in the structure (PDB ID: 4GBY)48, highlighting that the mutations were all located in the transmembrane domains (Fig.3f). Previously, Jiang et al. demonstrated that introducing two mutations (G58W and L315W) in XylE could prevent the binding of two inhibitors through conformational changes49. Notably, the G58W site is in the same transmembrane domain to A62V and close to T34I in the structure. Since the non-muconate-producing strain LC345 with wild-type XylE and similar genetic background to QP328 grew well on xylose (Supplementary Fig.2a), we hypothesized that the mutations in XylE that occurred in ALE might be induced by inhibitor(s) from the muconate-producing background strain. We also introduced the mutations xylE-A62V and A455V to strain JE3692, previously reported to grow on lignocellulosic hydrolysates, generating LC111. The two strains were evaluated in a plate reader on xylose. LC111 demonstrated improved growth with a reduced lag time and higher growth rate compared to JE3692 (Fig.3g). The slight improvement may be caused by the trace amounts of inhibitor(s) from the native pathways, and this may also suggest that introducing xylE-A62V, A455V can improve xylose utilization for the production of other non-shikimate pathway-related products.

The 227.8 kB duplication was identified based on approximately twofold higher sequencing coverage from PP_5050 to PP_5242 compared to the rest of the genome (Fig.4a). However, it was challenging to identify the exact location of the duplicated region based on the sequencing data due to the short read length of Illumina sequencing50,51. We thus deleted the original region of PP_5050PP_5242 in QP478, using the known sequences outside the region as homologous arms, to generate LC171, and deleted a portion of the duplication from PP_5084PP_5242 to generate strain LC173. On 30mM xylose in M9 medium, LC171 with the whole region deletion exhibited a lower growth rate, while LC173 with partial deletion showed comparable growth with a slightly longer growth lag and increased growth rate relative to QP478 (Fig.4b). Based on these results, we concluded that the duplication is important to the performance of QP478, and the potential beneficial gene(s) should be in the region PP_5050PP_5083, which remains intact in LC173. Thus, we next sought to identify the gene(s) in this region that contributed to the improved growth on xylose.

a Identification of the duplicated region from next-generation sequencing data, which presented ~2 the number of sequencing reads in QP478, and complete and partial deletions of these duplicated regions in LC171 and LC173, respectively. The graphs of coverage were generated in Geneious Prime 2020.0.4. b Growth curves of QP478, LC171, and LC173 on M9 medium with 30mM xylose. represents growth lag, A represents absolute growth rate, andboth are the average values of three independent growth curves. c Overexpressing candidate genes in reverse engineered strain LC100 at the pykF site. d Growth curves of QP478, LC100, LC199 and LC224 on M9 medium with 30mM xylose. e Maximum specific growth rates extracted from panel d. f Growth lag values extracted from panel d. gi Profiles in shake-flask experiments of strain LC224 on M9 medium with 30mM xylose, 30mM glucose+15mM xylose, and30mM glucose, respectively. For shake flask experiments, % molar yield was calculated as [mM muconate/mM (glucose+xylose)100], and% carbon yield was calculated as [mM muconate6/mM (glucose6+xylose5)100]. Error bars here represent the standard deviation of three biological replicates. Source data are provided as a Source Data file.

Since glucose and xylose were both utilized at similarly low rates in LC100 (Fig.2d), we reasoned that the slow growth might manifest in part(s) of the pathway shared by both sugars. Three candidate genes within PP_5050PP_5083 were selected for overexpression in LC100, including one related with sugar metabolism, PP_5056 (gpmI, 2,3-bisphosphoglycerate-independent phosphoglycerate mutase), and two in the shikimate pathway, PP_5078 (aroB, 3-dehydroquinate synthase) and PP_5079 (aroK, shikimate kinase) (Fig.4c). Two other genes outside this region but related with sugar metabolism were also tested, including PP_5085 (maeB, malic enzyme B) and PP_5150 (rpiA, ribose-5-phosphate isomerase A). Overexpression cassettes of the five genes were then inserted individually at the pykF site, generating strains LC147 (gpmI), LC150 (maeB), LC151 (rpiA), LC199 (aroK), and LC224 (aroB). All of these genes were driven by the Ptac promoter except for gpmI, which was driven by Plac promoter after two unsuccessful attempts to insert the gene into the genome using Ptac. The resulting strains were then evaluated with LC100 and QP478 in a plate reader containing M9 medium and 30mM xylose (Fig.4df). Overexpression of the three genes related to sugar metabolism in LC100 did not reduce the growth lag, while strains LC150 and LC151 demonstrated higher growth rates and greater final biomass accumulation (Supplementary Fig.9). Considering that higher biomass yield may reduce muconate yield, as we observed with strain LC091 (Fig.2bd), we decided to not pursue these targets further. Strains LC199 and LC224, which overexpress aroK and aroB, respectively, both demonstrated improved growth rates and reduced growth lag compared to LC100 (Fig.4df). LC224 grew even faster than QP478 with a similar lag time and higher growth rate (Fig.4df).

To investigate the potential additive effect of overexpressing aroK and aroB, we also expressed aroK and aroB in an operon-like pattern as aroKB in LC100, generating strain LC168.Strains LC199, LC224, LC168, and QP478 were evaluated in shake-flask experiments with M9 medium containing glucose and xylose to examine muconate production. The aroB overexpression strain LC224 outperformed its evolved counterpart QP478 with a higher muconate yield and improved growth rate (Supplementary Fig.10a, c), suggesting that the reaction of DAHP to 3-dehydroquinate (3DHQ) was rate limiting in LC100. Overexpressing aroK in LC100 (generating LC199) increased the growth rate slightly (Supplementary Fig.10b, d). Strain LC168did not exhibit improvement compared to LC224 (Supplementary Fig.10c, e).

To investigate if aroB overexpression alone can lead to better strain performance, we overexpressed aroB in QP328, generating strain LC349. In the plate reader evaluation of strains LC349, QP328 and LC224, LC349 exhibited highest growth rate on glucose, and slightly lower growth rate on mixture of glucose and xylose compared to LC224, while not surprisingly much slower growth on xylose relative to LC224 (Supplementary Fig.11ac), probably due to the lack of mutations in xylE. Interestingly, in shake flasks experiment on mixture of glucose and xylose, LC349 outperformed QP328 with a much higher muconate yield, which was still significantly lower than LC224 (Supplementary Fig.11df). The muconate production of LC349 is slower than LC224, as it took up to 41h for LC349 and 26.5h for LC224 to reach maximum muconate titer(Supplementary Fig. 11e, f).

We next examined the performance of LC224 on M9 medium containing glucose, xylose, or a mixture of the two. Muconate yields were highest on xylose, lowest on glucose, and intermediate on the mixture (Fig.4gi), reflecting the benefit of introducing xylose into the pentose phosphate pathway to supply E4P. Interestingly, both glucose and xylose utilization rates were higher on the mixture of glucose and xylose compared to on glucose or xylose alone (Fig.4gi).

Bioreactor cultivations of LC224 and QP478 were conducted in fed-batch mode to maintain sugar (glucose and xylose) concentrations lower than 10gL1 (Supplementary Fig.12ac). Glucose and xylose were simultaneously utilized in both strains from the start of the cultivation (Fig.5a, b); however, sugar utilization rates were higher in LC224 than QP478. LC224 utilized 46gL1 glucose and 20gL1 of xylose by the end of the cultivation while QP478 utilized 34gL1 and 10gL1, respectively. The muconate titer was almost threefold higher in LC224 compared to QP478, a 26.8gL1 and 9.3gL1, respectively (Fig.5a, b). Muconate yields reached 50% (molar yield) in LC224 while the yields were 25.9% in QP478 (Fig.5a, b). These improvements in LC224 were also reflected in the overall muconate production rate (0.28gL1h1), which was substantially higher than that achieved in QP478 (0.10gL1h1) (Fig.5a, b).

a, b Bacterial growth, glucose and xylose utilization, and muconate titers, yields, and rates from QP478 and LC224 in 96.6-h cultivations. c Bacterial growth, glucose and xylose utilization, and muconate titer, yields, and rate from LC224 in 191-h cultivation. For a and b, data points represent the average values of biological duplicates, error bars represent the absolute difference between the data generated from duplicates at each time point; for (c), data points represent the values of singlets. Metabolic yields (mol%) at each time point were calculated as the amount of muconate (in moles) produced divided by the glucose and xylose (in moles) utilized. Metabolic yields (mol%) are corrected based on the dilution factor generated by the volumes of base and substratefeeding. Final carbon yields (carbon%) listed were calculated as [mM muconate6/mM (glucose6+xylose5)100]. Rates (gL1h1) at each time point were calculated as the muconate concentration divided by the cultivation time. All the titer (T), rate (R), and yield (Y) valueslisted were at the last time point. Final yields (mol%, carbon%) listed in (c) have alsobeen corrected based on the quantified evaporated volume. Source data are provided as a Source Data file.

The muconate titer, rate, and yield achieved in bioreactor cultivations were 26.8gL1, 0.28gL1h1, and 49.9% (Fig.5b), respectively, at 96.6h. This yield represents almost 100% of the maximum theoretical based on our strain design and metabolic modeling (vida supra). To explore whether the titer could be further improved, we conducted another bioreactor experiment where LC224 was cultivated for 191h (Fig.5c). The resulting muconate titer increased to 33.7gL1 and at a yield of 46%, 92% of the theoretical maximum when corrected for evaporation. It is also noteworthy that while LC224 reached stationary phase at ~54h, the cells continued utilizing sugars and producing muconate, which demonstrated that the muconate production here was not growth coupled, suggesting thatthe muconate titer and yieldcould be further improved if the experiment had continued (Fig.5b, c).

To better understand the differences between the unevolved parent QP328, the evolved strain QP478, and the rationally engineered strain LC224, intracellular and extracellular metabolomics experiments were conducted. Selected metabolites related to sugar metabolism and muconate production are presented in Fig.6. Compared to QP328, the strains QP478 and LC224 exhibited reduced accumulation of metabolites in the EDEMP cycle, and greater accumulation of metabolites in the shikimate and the muconate pathways (Fig.6a, b), which is consistent with the mutations that evolved in QP478 and were engineered in LC224 (Figs.2 and 4). The intensities of DAHP, the joint node of sugar metabolism and shikimate pathway, however, demonstrated theopposite pattern compared to other metabolites in shikimate pathway (Fig.6b). LC224 and QP478 accumulated less DAHP compared to QP328, which is consistent to our objective above regarding the duplication and aroB overexpression.

a Simplified metabolic pathway. Metabolites in the EDEMP cycle arelabeled blue, in the shikimate and muconate pathway are labeled green, the joint node DAHP is labeled purple, the extracellular anthranilic acid (ANA) is labeled brown. QA quinic acid, ANA anthranilic acid. b Intensity of selected metabolites in the three strains QP328 (blue), QP478 (orange), and LC224 (red). Intracellular intensities have been normalized by lyophilized biomass. Error bars represent the standard deviation of three biological replicates. Source data are provided as a Source Data file.

Specifically, the DAHP level in LC224 was much lower compared to QP478, which may suggest the aroB activity in LC224 driven by tac promoter was higher than that of QP478. Except for DAHP, LC224 accumulated a higher amount of metabolites in the shikimate pathway and fewer metabolites in the EDEMP cycle relative to QP478 (Fig.6a, b), suggesting greater flux entering the shikimate pathway in LC224 and enabling greater muconate biosynthesis.

Although its precursor 3DHQ was not detected in any samples, quinic acid (quinate, QA) was substantially accumulated in LC224 (Fig.6b), which may suggest an overflow of carbon resulting from overexpression of aroB. Shikimic acid (shikimate, SA) accumulation in LC224 is evidence of muconate biosynthesis through the aroE pathway, while SA accumulation was much lower in QP478 relative to LC224, which may be related to the aroK duplication in QP478. The accumulation of anthranilic acid in the culture media likely represents another case of overflow metabolism. More catechol (CAT) accumulated in LC224 (Fig.6b), which could represent new bottlenecks associated with increased flux to muconate. Together, these results illustrate that engineering to generate LC224 broadly recapitulated the evolved strain QP478 and suggest additional targets for further improvement.

Read the original here:

Muconic acid production from glucose and xylose in Pseudomonas putida via evolution and metabolic engineering - Nature.com