L-Tryptophan Production in Escherichia coli Improved by Weakening the Pta-AckA Pathway

Acetate accumulation during the fermentation process of Escherichia coli FB-04, an L-tryptophan production strain, is detrimental to L-tryptophan production. In an initial attempt to reduce acetate formation, the phosphate acetyltransferase gene (pta) from E. coli FB-04 was deleted, forming strain FB-04(Δpta). Unfortunately, FB-04(Δpta) exhibited a growth defect. Therefore, pta was replaced with a pta variant (pta1) from E. coli CCTCC M 2016009, forming strain FB-04(pta1). Pta1 exhibits lower catalytic capacity and substrate affinity than Pta because of a single amino acid substitution (Pro69Leu). FB-04(pta1) lacked the growth defect of FB-04(Δpta) and showed improved fermentation performance. Strain FB-04(pta1) showed a 91% increase in L-tryptophan yield in flask fermentation experiments, while acetate production decreased by 35%, compared with its parent FB-04. Throughout the fed-batch fermentation process, acetate accumulation by FB-04(pta1) was slower than that by FB-04. The final L-tryptophan titer of FB-04(pta1) reached 44.0 g/L, representing a 15% increase over that of FB-04. Metabolomics analysis showed that the pta1 genomic substitution slightly decreased carbon flux through glycolysis and significantly increased carbon fluxes through the pentose phosphate and common aromatic pathways. These results indicate that this strategy enhances L-tryptophan production and decreases acetate accumulation during the L-tryptophan fermentation process.


Introduction
L-tryptophan, an essential amino acid that is also a precursor of other important biomolecules, such as the neurotransmitter serotonin, is widely used in medicine, food, and animal feed [1,2]. The L-tryptophan biosynthetic pathway in microorganisms, which involves central metabolism, the common aromatic pathway and the L-tryptophan branch pathway, is long and its regulation is complicated [3][4][5]. The high industrial relevance of Escherichia coli has stimulated efforts to improve L-tryptophan yield by analyzing the underlying metabolic regulatory networks of L-in Table 1. These primers were used to amplify the genes pta, ackA and poxB using genomic DNA from E. coli FB-04 as the template. They were also used to amplify the genes pta1, ackA1, and poxB1 using genomic DNA from E. coli CCTCC M 2016009 as the template. To express Pta and its mutant Pta1, the PCR products pta and pta1, obtained from E. coli FB-04 and Amp and kan markers [21] pKD46 Amp markers, helper plasmid [21] pSTV-03 Based on plasmids of pACYC177 and pND707; p15A replicon, kan [5] marker, PR and PL promoters, carrying aroF fbr and trpE fbr D This study

5'-ttaccttagccagtttgtt-3'
SK-kan-sacB F:5'-aaacatatgattccggggatccgtcgacc-3' This study CCTCC M 2016009, respectively, using the primer set C-pta (Table 1), were digested with NdeI and HindIII, and then ligated into plasmid vector pET24a to form plasmids pET24a-pta and pET24a-pta1. These two plasmids were used to transform the expression strain E. coli BL21(DE3), resulting in strains BL21(DE3)/pET24a-pta and BL21(DE3)/pET24a-pta1. Genetic manipulations were performed in mutant alleles by using a two-step scarless gene replacement technique employing λ Red recombination [21,22]. For scarless gene replacement, plasmid pMD-SK was created by cloning two separate PCR products, kan and sacB, into the plasmid vector pMD18-T. First, kan and sacB were PCR amplified using plasmid pKD13 and the genomic DNA of B. subtilis ATCC 6051a as templates, respectively, and the primer pair SK-kan-sacB (Table 1). The two PCR products were ligated by overlap extension PCR, then the resulting kan-sacB fragment was cloned into pMD18-T, forming pMD-SK. Next, the kan-sacB fragment plus homologous sequences was PCR amplified from plasmid pMD-SK using P-pta or P-ackA primers ( Table 1). The underlined letters represent homologous sequences. The single-gene pta knockout mutant FB-04(Δpta) and single-gene ackA knockout mutant FB-04(ΔAckA) were constructed by transducing the corresponding kan-sacB fragment into cells harboring the helper plasmid pKD46, using kanamycin resistance as the positive selection marker. For the scarless insertion of pta1 in the genome of E. coli FB-04, forming mutant FB-04(pta1), the pta1 fragment, which was amplified from the E. coli CCTCC M 2016009 genome using the primer pair V-pta (Table 1), was transduced into the pta singlegene knockout mutant harboring plasmid pKD46, using 15% sucrose as the negative selection marker. PCR and DNA sequencing were also performed to verify the identities of all alleles and plasmids.

Media and culture conditions
For enzyme expression, E. coli strains BL21(DE3)/pET24a-pta and BL21(DE3)/pET24a-pta1 were grown at 37°C in LB medium. Portions of these seed cultures (10% (v/v)) were used to inoculate 100 mL portions of TB medium supplemented with 30 μg/mL kanamycin in 500 mL flasks at 37°C. After 2 h, isopropyl β-D-thiogalactopyranoside was added to a final concentration of 0.4 mM to induce protein expression. Enzyme expression was performed at 25°C for 24 h in a reciprocal shaker (200 rpm).
For flask cultivation, seed cultures of the L-tryptophan-producing strain FB-04 and its derivatives were grown in LB medium for 10 h at 37°C in a rotary shaker. Aliquots (200 μL) of these seed cultures were used to inoculate 100 mL portions of flask fermentation medium containing (per liter): 24 g K 2 HPO 4 , 9.6 g KH 2 PO 4 , 1g MgSO 4 Á7H 2 O, 5 g (NH 4 ) 2 SO 4 , 10 g glucose, 15 g yeast, 2 g citric acid, and 3 ml trace element solution. The ingredients of trace element solution have been described elsewhere [5]. These portions of fermentation medium, which were supplemented with 30 μg/mL kanamycin, were incubated in 500 mL flasks at 37°C for 48 h. During the fermentation, NH 4 OH and 10 g/L glucose were added to the medium every 8 h to maintain the pH at 6.5-7.2 and supply a source of carbon, respectively. For metabolomic analysis, cells were collected at 26 h.
For fermentor cultivations, 100 mL seed cultures of FB-04 and its derivatives were first prepared in 500 mL flasks in flask fermentation medium for 10 h at 37°C in a rotary shaker, then transferred to a 3 L BioFlo 110 fermentor (New Brunswick Scientific, USA) containing 900 mL fed-batch fermentation medium containing (per liter) (pH 6.5): 15 g K 2 HPO 4 , 2g MgSO 4 Á7H 2 O, 1.6 g (NH 4 ) 2 SO 4 , 2 g yeast, 7.5 g glucose, 2 g citric acid, 0.0129 g CaCl 2 , 0.075 g FeSO 4 Á7H 2 O. The fed-batch was performed for 54 h, and samples were taken every 3 h. The pH was maintained at 6.5 by automatically adding NH 4 OH. The temperature was kept at 37°C. The dissolved oxygen level was controlled at 20% by adjusting agitation speed. When glucose in the medium was exhausted, glucose (800 g/L) was introduced using an exponential feeding program [23], with cell growth controlled at a specific growth rate of 0.15 h -1 . All fermentation processes were repeated three times. Fermentation parameters are reported as the mean ± standard deviation of these three replicates.

Purification of wild-type Pta and its mutant Pta1
Extraction and purification of phosphate acetyltransferase were performed as previously described [24], with slight modification. All operations were performed at 4°C. E. coli cells harvested from expression cultures were suspended in 20 mM KHCO 3 , then disrupted by sonication for 15 min on ice. Cellular debris was removed by centrifugation at 12,000 × g for 30 min. At this point, saturated (NH 4 ) 2 SO 4 was added slowly to the supernatant to attain a final concentration of 50% saturation. The precipitates were stirred overnight at 4°C, collected by centrifugation, then suspended in 50 mM Tris-HCl (pH 8). This solution was dialyzed overnight against 0.2 mM ammonium sulfate in 10 mM Tris-HCl (pH 7.6). Protein purification was performed using a fast protein liquid chromatography system (AKTA FPLC system; GE Healthcare) equipped with a DEAE-Sepharose anion-exchange column (160 mm by 10 mm; Pharmacia Biotech). The dialyzed enzyme solution was diluted 4 times with 10 mM Tris-HCl (pH 8.5), and then loaded onto the DEAE-Sepharose column, which had been pre-equilibrated with 0.05 M ammonium sulfate in 10 mM Tris-HCl (pH 7.8). A linear gradient of 0.05 to 0.2 M ammonium sulfate in 10 mM Tris-HCl (pH 7.8) was used to elute the adsorbed proteins. Fractions containing enzyme were pooled and stored at -20°C.

Analytical methods
The purified phosphate acetyltransferase was analyzed by SDS-PAGE, and the Bradford method was used to determine protein concentration [25]. Phosphate acetyltransferase activity was measured as previously described [26]. The kinetic parameters (K m , V max, and k cat values) of the purified phosphate acetyltransferase were determined by using acetyl-CoA as the substrate (at concentrations of 20, 40, 80, 125, 150, 200, 250, 300, 360, and 400 μmol/L) at 25°C in 100 mM phosphate buffer (pH 7.4). K m and V max values were estimated by fitting the initial rate data to the Michaelis-Menten equation using nonlinear regression with GraphPad Prism software (GraphPad Software Inc., San Diego, CA). Enzyme assays were repeated three times. Kinetic data are reported as the mean ± standard deviation of these three replicates.
The optical density at 600 nm (OD 600 ) was measured after appropriate dilution. For detection of extracellular metabolites, the supernatant obtained after centrifugation of the fermentation broth was stored at -20°C until analyzed using high performance liquid chromatography (HPLC). The concentrations of glucose and acetate were determined using HPLC with an Aminex HPX-87H column (300 mm × 7.8 mm; Bio-Rid, Hercules, CA) at 50°C, 50 mM H 2 SO 4 as the mobile phase and a flow rate of 0.5 mL/min. The concentration of L-tryptophan was determined using the method described previously [5], with an Agilent Eclipse XDB-C18 column (15 mm × 4.6 mm).

Metabolome analysis by GC-MS
Gas chromatography-mass spectrometry (GC-MS) was used to characterize the intracellular metabolite profiles of FB-04, FB-04(Δpta) and FB-04(pta1). Five biomass samples of each strain were taken manually from the shaker, quickly filtered through a nitrocellulose filter (pore size 45 μm, Pall Corporation), and then immediately quenched in liquid nitrogen. Cells (100 mg wet weight) were suspended in 1.1 ml 90% methanol (-40°C, v/v) and mixed quickly. The cell suspension was frozen and thawed three times, sonicated for 15 min at 4°C, and then held at -20°C for 1 h. After centrifugation at 14,000 × g and 4°C for 15 min, isotopically enriched internal standard (L-phenylalanine-13 C 9 -15 N) was added into the supernatant to a final concentration of 10 μg/mL, then this mixture was dried using a moderate stream of nitrogen. Methoxylamine hydrochloride in pyridine (20 mg/mL; 30 μL) was added to the powder, and then 30 μL of N,O-Bis(trimethylsilyl)trifluoroacetamide containing 1% trimethylchlorosilane was added to the mixture. Derivatization was allowed to proceed at 70°C for 60 min.
The GC/MS analysis was performed as previously described [27], with slight modifications. The oven temperature was initially held at 70°C for 2 min, then ramped to 160°C at a rate of 6°C/min, to 240°C at a rate of 10°C/min, to 300°C at a rate of 20°C/min, and finally held at 300°C for 8 min. AMDIS software from NIST (National Institute of Standards and Technology) was employed to deconvolute the mass spectra obtained from the raw GC/MS data. The deconvoluted mass spectra were automatically matched with a standard library as previously described that included retention time and mass spectra [27]. Unmatched peaks were analyzed using NIST MS 2.0 software, which automatically searches for compound information from the NIST 11 library and the Golm Metabolome Database. The final data were subjected to principal component analysis using commercial software Simca-P 11.0 (Umetrics AB, Umeå, Sweden) to holistically observe the general clustering and trends among all samples.

Sequence comparison of genes involved in the formation of acetate in E. coli FB-04 and E. coli CCTCC M 2016009
When they are grown under the same fermentation conditions, less acetate is secreted by E. coli CCTCC M 2016009 than by E. coli FB-04 (data not shown). To address this issue, genes involved in acetate formation were analyzed. The pta, ackA, and poxB genes from E. coli FB-04 and the pta1, ackA1, and poxB1 genes from E. coli CCTCC M 2016009 were amplified using PCR and then sequenced. Comparison of the three pairs of genes (pta/pta1, ackA/ackA1, poxB/ poxB1) revealed a single base difference (c206t) between pta and pta1; the sequences of ackA and ackA1 were identical, as were those of poxB and poxB1. This difference leads to a single amino acid substitution (Pro69 to Leu) in Pta1, compared with Pta.

Enzymatic analysis of Pta and Pta1
To evaluate the activities of Pta and Pta1, the two enzymes were overexpressed in E. coli BL21 (DE3). SDS-PAGE analysis of cell lysates revealed protein bands at approximately 77 kDa, which correspond to those observed with purified recombinant Pta and Pta1 (Fig 1). The positions of these bands are consistent with their predicted molecular masses. The protein concentrations were almost the same, indicating that the difference between the expression levels of Pta and Pta1 was not significant. Purified Pta and Pta1 were subjected to kinetic analysis. The catalytic constant (k cat ) of Pta1 was 38% lower than that determined for Pta, while its K m was 190% higher than that of Pta. Thus, the k cat /K m value of Pta1 was only about 21% that of Pta ( Table 2).

The principal component differences caused by genetic modification
Principal component analysis (PCA) was carried out to gain insight into the multivariate data and evaluate biological alteration. Clustering of biological samples was based on their similarities and differences in the metabolite dataset. In the PCA score plot, each data point reflects a linear combination of the total metabolites from each sample. The distances between the groups give a measure of the overall variation among the metabolic profiles of the different strains. As shown in Fig 4, three groups of FB-04, FB-04(Δpta) and FB-04(pta1) were distributed in different areas of the PCA score plot.
Changes of carbon flux in strain carrying a pta deletion or a pta1 genomic substitution In the present study, more than 80 intracellular metabolites that showed different levels in FB-04, FB-04(Δpta) and FB-04(pta1) were identified using GC-MS. Of these 80 metabolites, 23 of them were involved in L-tryptophan biosynthesis. To identify the potential reasons for the significant diversity in metabolic profiles caused by genetic modification, the relative levels of the 19 intracellular metabolites were determined (see S1 Table). The levels of intracellular intermediates in the three strains FB-04, FB-04(Δpta) and FB-04 (pta1) are shown in Fig 5. Interestingly, metabolic intermediate levels in these three strains changed remarkably. Comparing FB-04(Δpta) with the parent strain FB-04, the overall levels of glycolysis intermediates were reduced; however, pyruvate accumulated significantly. Moreover, the overall levels of TCA cycle intermediates and pentose phosphate pathway intermediates were increased. Notably, levels of common aromatic pathway intermediates, such as 3-dehydroshikimate, shikimate and shikimate-3-phosphate, were increased in FB-04(Δpta).
Compared with FB-04, FB-04(pta1) displayed a metabolic performance similar to that of FB-04(Δpta). The overall levels of glycolysis intermediates were slightly decreased, while the overall levels of pentose phosphate pathway intermediates, TCA cycle intermediates, and common aromatic pathway intermediates were significantly increased. Moreover, compared with FB-04(Δpta), the overall levels of intermediates in glycolysis and the pentose phosphate pathway were increased in FB-04(pta1), while levels of TCA cycle intermediates were reduced. In addition, the levels of shikimate and shikimate-3-phosphate, were higher in FB-04(pta1), although that of 3-dehydroshikimate was reduced.

Discussion
L-tryptophan is mainly produced by microbial fermentation using Escherichia coli or Corynebacterium glutamicum. A randomly mutagenized E.coli strain was shown to produce up to 54.6 g/L L-tryptophan when fed L-tryptophan precursors [6]. With the recent advances in molecular technology, several studies have been conducted in an effort to construct L-tryptophan-producing strains with defined genetic modifications [15,[28][29][30]. For example, genetic modification of a classically derived L-tryptophan-producing Corynebacterium glutamicum strain increased L-tryptophan production to 58 g/L [28]. E.coli strain Dpta/mtr-Y, developed by Wang et al. [15], achieved an L-tryptophan yield of 48.68 g/L.
In this study, mutant strains FB-04(Δpta) and FB-04(ΔackA) were constructed to decrease acetate accumulation. Deletion of pta or ackA led to substantially lowered acetate formation (Table 3). Pta plays a more important role in the Pta-AckA pathway, in view of the fermentation performance of FB-04(Δpta) and FB-04(ΔackA) (Figs 2a and 3b). Reduced acetate levels were conducive to L-tryptophan biosynthesis, as improved L-tryptophan titers were observed in FB-04(Δpta) and FB-04(ΔackA), compared with FB-04 (Fig 2b). Notably, deletion of pta achieved a more significant increase in L-tryptophan production than deletion of ackA in shake-flask fermentations (Fig 2b). However, FB-04(Δpta) exhibited seriously restricted growth, which was consistent with previous findings [14,17]. To avoid the physiological defects caused by pta deletion, we identified a mutant Pta (Pta1) from E. coli CCTCC M 2016009. Kinetic analysis showed that the K m of Pta1 was 190% higher than that of Pta, and the k cat /K m value of Pta1 was only about 21% that of Pta (Table 2). These data indicate that Pta1 possesses lower catalytic activity and substrate binding affinity. In this study, we constructed FB-04(pta1), in which pta was replaced with pta1. This substitution not only resulted in a noticeably lower ability to secrete acetate, it also reversed the growth defect caused by pta deletion (Fig 3a and 3b). It has been reported that the growth defect caused by pta deletion results from disturbing acetyl-CoA flux, and that any method that relieves the oversupply of acetyl-CoA would compensate for this defect [17]. The pta1 genomic substitution might alleviate the accumulation of acetyl-CoA to a certain degree by weakening the Pta-AckA pathway. In addition to its lowered acetate accumulation and normal growth characteristics, strain FB-04(pta1) showed a substantial improvement in L-tryptophan yield over those of FB-04 and FB-04(Δpta) (Fig 3c).
To investigate the effect of Pta alteration on cells, changes in metabolic flow were explored. Because the deletion of pta reduces the flux through glycolysis and increases the flux through the TCA cycle caused by the accumulation of acetyl-coA and pyruvate [17], FB-04(Δpta) displayed decreased levels of glycolytic intermediates and increased levels of TCA cycle intermediates (Fig 5). The activity of glucose-6 phosphate dehydrogenase, the first enzyme in the pentose phosphate pathway, is known to be upregulated by pta deletion [31]. This contributes to the increased flux through the pentose phosphate pathway, which is consistent with the increased levels of pentose phosphate pathway intermediates seen in this study (Fig 5). The metabolic performance of FB-04(pta1) differed from that of FB-04 in ways similar to the metabolic differences between FB-04 and FB-04(Δpta). FB-04(pta1) displayed decreased levels of glycolytic intermediates and increased levels of TCA cycle and pentose phosphate pathway intermediates (Fig 5). In addition, the levels of pentose phosphate pathway intermediates in FB-04(pta1) were greater than those of FB-04(Δpta)). This difference may be ascribed to the more reasonable distribution of metabolic flow among the central metabolic pathways caused by the pta1 genomic substitution. Increased flux through the pentose phosphate pathway not only supplies additional erythrose 4-phosphate, an important L-tryptophan precursor, it also allows the full utilization of another precursor, phosphoenolpyruvate, which boosts carbon flux through the common aromatic pathway [32]. This is consistent with the increased levels of common aromatic pathway intermediates in FB-04(pta1) (Fig 5). The level of 3-dehydroshikimate, an intermediate in the common aromatic pathway, was lower in FB-04(pta1) than in FB-04(Δpta) ( Fig  5). The improved metabolic flow in the pentose phosphate pathway caused by the pta1 genomic substitution could increase the formation of NADPH, an essential cofactor in the common aromatic pathway [32]. This would drive carbon flux from 3-dehydroshikimate to shikimate, decreasing the level of 3-dehydroshikimate.
In view of the length of the L-tryptophan biosynthetic pathway and its complicated regulation mechanism, relying only on decreased acetate production and changes in metabolism caused by the pta1 genomic substitution would not achieve a remarkable improvement in Ltryptophan production. The metabolic alterations caused by the pta1 genomic substitution also increased levels of L-tryptophan precursors, which further contributed to L-tryptophan biosynthesis. In future studies, overexpression of the genes ppsA and tktA, which are involved in the biosynthesis of phosphoenolpyruvate and erythrose 4-phosphate, respectively [32], may significantly improve L-tryptophan production.

Conclusion
A pta gene knockout was constructed in E. coli FB-04 using λRed recombination to reduce the formation of acetate, but the resulting strain (E. coli FB-04(Δpta)) exhibited a growth defect. Then, a mutant (Pta1) that exhibits lower catalytic capacity and substrate affinity than Pta because of a single substitution (Pro69Leu) was identified in E. coli CCTCC M 2016009. This variant (pta1), was used to replace the pta gene of E. coli FB-04, forming strain FB-04(pta1).
FB-04(pta1) not only lacked the growth defect of FB-04(Δpta) and showed improved fermentation performance, it also displayed a 91% increase in L-tryptophan yield during flask fermentation, compared with FB-04, while acetate production decreased by 35%. Moreover, acetate secretion by FB-04(pta1) was slower than that by FB-04 throughout the fed-batch fermentation processes, and finally the L-tryptophan yield of FB-04(pta1) represented a 15% increase over that of FB-04. Metabolomics analysis showed that the pta1 genomic substitution slightly decreased carbon flux through glycolysis and significantly increased carbon flux through the pentose phosphate and common aromatic pathways, contributing to the biosynthesis of Ltryptophan.
Supporting Information S1 Table. The levels of intracellular metabolites involved in L-tryptophan biosynthesis in different strain. To ensure the accuracy of the data, five biomass samples of each strain were subjected to GC-MS. Most of the samples were within 95% confidence interval, except samples A1 and B1 (date not show). A1 and B1 were judged to be abnormal samples and were omitted from the analysis to ensure the reliability of the results. The rt_mz values express the mass-tocharge ratio of chromatographic retention time. (DOC)