Introduction

The native metabolic machinery of most natural microorganisms presents unfavorable atomic economics in biorefinery with respect to the generation of the 2C-compound, acetyl CoA, a central metabolic intermediate and precursor for a range of desirable bioproducts. In the fundamental biochemical infrastructure for sugar catabolism in almost all organisms such as the Embden-Meyerhof-Parnas pathway (EMP) and variants, acetyl CoA is formed through the oxidative decarboxylation of pyruvate. The intrinsic carbon loss associated with this process significantly limits the yield of acetyl CoA-derived products in commonly employed industrial platform strains. Thus, maximizing carbon conservation in bioproduction has received significant attention recently (Keasling 2010; Wang et al. 2018b).

Phosphoketolases (PKTs) (EC 4.1.2.9, EC 4.1.2.22) from the pentose phosphate pathway of lactic acid bacteria and the d-fructose-6-phosphate (F6P) shunt of Bifidobacteria, are versatile tools employed to facilitate carbon conservation in bioproduction. PKTs act in conjunction with phosphotransacetylases and enzymes of the pentose phosphate pathway to catalyze an alternative non-oxidative glycolytic (NOG) pathway with fructose-6-phosphate and/or xylulose-5-phosphate as substrates, bypassing pyruvate decarboxylation and leading to increased generation of stoichiometric amounts of acetyl CoA per glucose molecule (3 mole acetyl CoA/mole glucose compared with 2 mole acetyl CoA/mole glucose obtainable from the conventional pathway) (Bogorad et al. 2013). The application of PKTs in significantly improving the yield of acetyl CoA-associated products such as acetate, acetone, polyhydroxybutyrate, mevalonate, and l-glutamate from a variety of microbial producers has been demonstrated (Bergman et al. 2016; Bogorad et al. 2013; Chwa et al. 2016; de Jong et al. 2014; Henard et al. 2017; Kocharin et al. 2013; Sonderegger et al. 2004; Wang et al. 2018a; Yang et al. 2016). For instance, the expression of PKT in Corynebacterium glutamicum has been estimated to improve l-glutamate yield (about 98% by weight) beyond the maximum theoretical yield from the conventional pathway (about 82% by weight) (Chinen et al. 2007). To achieve the 98% yield, the NOG/EMP pathway metabolic flux ratio should be 4:1 (Kozlov et al. 2006). PKTs and 6-phosphofructokinase (PFK) are the key enzymes determining the flux ratio. However, the low activity of characterized PKTs compared with that of PFK significantly limits their potential applications. In most industrial platform strains, the specific enzyme activity of PFK ranges from 148 - 263 U/mg with Km value between 0.011 – 0.05 mM (Babul 1978; Hofmann and Kopperschlager 1982; Kotlarz and Buc 1982; Nissler et al. 1983; Welch and Scopes 1981), while PKTs from both Bifidobacteria and Lactobacilli that have been employed in metabolic engineering for improving the yield of acetyl CoA-derived products display specific enzyme activity in the range of 0.7–30 U/mg with Km value between 1.4 and 39 mM (Grill et al. 1995; Posthuma et al. 2002; Sgorbati et al. 1976; Suzuki et al. 2010b; Yevenes and Frey 2008). Therefore, improving PKT activity is very crucial and can be an effective strategy in metabolic engineering (Henard et al. 2015). Previous attempts have focused on PKT overexpression through the use of strong promoters (Chinen et al. 2007; Henard et al. 2017; Wang et al. 2018b). However, this may increase the cell burden for protein synthesis, and protein overexpression sometimes results in inactive inclusion body formation (Baneyx and Mujacic 2004; García-Fruitós et al. 2005; Gonzalez-Montalban et al. 2005; Strandberg and Enfors 1991). Thus, improving the specific activity of PKTs may be an alternative and better strategy. However, until now, there has been no such report on improving PKT specific activity.

Structure-guided protein rational design and directed enzyme evolution are successful proven approaches in engineering and improving enzyme properties. While rational design requires an in-depth knowledge of enzyme catalytic mechanism, directed evolution offers the possibility of improving enzyme activity without detailed understanding of enzyme structure-function relationships (Bloom et al. 2005; Chen 2001). However, a major challenge in directed evolution is the mutant selection method. Growth-coupled selection has been previously employed to improve the efficiency of directed evolution. For instance, using a growth-coupled evolution strategy, by coupling the xylose utilization ability of the engineered host with the specific activity of xylose isomerase, Lee et al obtained a Piromyces sp. xylose isomerase mutant which exhibited 77% increased enzyme activity (Lee et al. 2012). In another instance, Atsumi and Liao obtained several M. jannaschii citramalate synthase mutants that led to increased growth of a host auxotrophic for l-isoleucine (Atsumi and Liao 2008).

In this study, a growth-coupled evolution strategy was established for improving the specific activity of PKTs, utilizing Corynebacterium glutamicum with defective PFK as hosts, and the industrial relevance of the obtained mutant enzymes in improving the yield of acetyl CoA-derived products with l-glutamate as an example was demonstrated.

Materials and methods

Chemicals and enzymes

Isopropyl-β-d-thiogalactopyranoside (IPTG), l-cysteine hydrochloride, thiamine pyrophosphate, magnesium chloride, xylulose-5-phosphate (X5P), hydroxylamine hydrochloride, trichloroacetic acid, and acetyl phosphate (ACP) were supplied by Sinopharm Chemical Reagent Co. Ltd (Tianjin, China). d-fructose 6-phosphate (F6P) and phosphoenolpyruvate (PEP) were supplied by Solarbio (Beijing, China). 3-morpholinopropanesulfonic acid (MOPS) and adenosine-5′-monophosphate (AMP) were supplied by Amresco (USA). Hexadecyltrimethylammonium bromide (CTAB) and oxaloacetate (OAA) were purchased from Sigma (USA). Adenosine-5′-triphosphate (ATP) was from ThermoFisher Scientific (USA), while erythrose-4-phosphate (E4P) was from Santa Cruz Biotechnology Inc. (USA). Other pure chemicals used in this study were of analytical grade or better. The His SpinTrap columns used for the purification of the PKTs were purchased from GE Healthcare Corporation (USA). DNA polymerase was obtained from Transgene (Beijing, China). Restriction endonucleases were purchased from Fermentas (USA). T4 DNA ligase was purchased from New England Biolabs Inc. (Beijing, China).

Strains and plasmids

The main bacterial strains and plasmids used in this study are listed in Table 1. Other strains and plasmids were constructed based on them. Escherichia coli DH5α was used for general cloning and cultivated aerobically at 37 °C in Luria–Bertani (LB) broth. Kanamycin at 50 μg/ml was added to the LB broth as required.

Table 1 Bacterial strains, plasmids, and primers used in this study
Full size table

Z188Δpfk and 13032Δpfk were constructed by the knockout of pfk gene in Z188 and 13032 via a two-step homologous recombination using the suicide plasmid pK18mobsacB according to previously described method (Niebisch and Bott 2001; Schafer et al. 1994). Briefly, the primer pairs Δpfk-F1/Δpfk-R1 and Δpfk-F2/Δpfk-R2 (Table 1) were used to amplify the left and right homologous arms of pfk from the genomic DNA of Z188 or 13032, for recombination. The two fragments were capable of annealing to each other by their overlap sequence. The final amplified fragment was digested with EcoRI and PstI and inserted into the corresponding sites of pK18mobsacB. The resulting plasmid was then introduced into Z188 or 13032 for pfk inactivation.

E. coli-C. glutamicum shuttle vector pTRCmob was used for Bifidobacterium adolescentis—source PKT (BA-PKT, encoded by fxpk) expression in C. glutamicum (Liu et al. 2007). The codon-optimized version of fxpk was synthesized by Genewiz Company (Suzhou, China) and inserted in pTRCmob vector between EcoRI and BamHI sites under the control of the derepressed promoter Ptrc. The resultant plasmid was named pTR-fxpk. To improve the expression of fxpk, Ptrc was mutated by introducing a mutant site using two complementary primers Ptrc-1 and Ptrc-2 (Table 1) to amplify the whole plasmid and through self-recombination, plasmid pTR1-fxpk was obtained.

To construct fxpk mutant library, a standard error-prone PCR protocol involving the use of Taq polymerase with varied concentrations of manganese ions (0.2 mM, 0.5 mM, and 0.8 mM) was employed, using primer pair fxpk-F1/R1 (Table 1) and pTR-fxpk as template. The amplified products were mixed, purified, double-digested, and inserted into EcoRI and BamHI sites of the pTR-fxpk plasmid. Nine parallel libraries were constructed.

For the reintroduction of the fxpk mutants PM21, PM41, PM71, PM81, and PM82 into pTR-fxpk, fxpk-R1 was used as a common reverse primer for the mutant PCR amplification. The forward primers were PM21-F for PM21, PM41-F for PM41, PM71-F for PM71, PM81/PM82-F for PM81, and PM82, respectively (Table 1). The PCR products were inserted and ligated to replace the wildtype fxpk fragment in pTR-fxpk after double-digestion with EcoRI and BamHI, and purification. With pTR-fxpk and the constructed plasmids as templates, fxpk-F1/R2 (Table 1) as primers, PCR was performed to amplify wildtype fxpk and selected mutants for ligation into pET30a after double-digestion with NdeI and XhoI. The constructed pET30a plasmids were used to overexpress PKTs for enzyme analyses.

Site-directed mutagenesis of fxpk on plasmids pTR-fxpk and pET-fxpk was performed. Site mutations were introduced individually into the wildtype gene using a standard procedure involving the PCR amplification of the whole plasmid using complementary primers containing the mutant sites (Table S1) (Miyazaki and Takenouchi 2002). The PCR products were transformed directly into E. coli DH5α, and positive transformants were verified and obtained.

To characterize BA-PKT and its mutants, the constructed plasmids were transformed into corresponding hosts, Z188Δpfk or 13032Δpfk for growth analysis, Z188 for l-glutamate production analysis, and BL21 (DE3) for enzyme activity analysis.

Selection of fxpk mutant library

The constructed 9 fxpk mutant libraries were transferred individually into E. coli DH5α cells, each with about 104 colonies. After cultivation on LB agar plates, transformants were obtained, scraped off, the plasmids extracted and transformed into Z188Δpfk through electroporation as described previously (Ruan et al. 2015; van der Rest et al. 1999). The Z188Δpfk libraries were allowed to grow for about 48 hours on LBG (LB broth supplemented with 5 g/l glucose) agar plates at 30 oC, before being scraped off and harvested. The process for the enrichment of the positive BA-PKT mutants with improved activities is shown in Fig. 1. The constructed 9 libraries of Z188Δpfk expressing BA-PKT variants were inoculated into 9 wells of a 24-deep-well microtiter plate (each library in one well). Three additional wells were used to cultivate the original Z188Δpfk (pTR-fxpk) as control. Each inoculated well contained 1 ml CGXII medium (Keilhauer et al. 1993). Serial transfer was repeated every 48 hours with an inoculum size of 1 % (v/v) cells from the previous culture. After seven rounds of serial transfers, cells from the final culture were plated on solid agar plates, and 3 larger colonies randomly selected from each sample were isolated for further growth analysis in 1 ml CGXII medium in 24 deep-well culture plate. The variants with the highest growth were preliminarily selected. The plasmid vector from the selected variants was purified, the fxpk mutants sequenced, ligated into fresh plasmid vector, and retransformed into fresh starting strain background. A second growth rate measurement against the control was performed to confirm that the growth rate improvement was due to the fxpk mutants and not background adaptation of the host strain or unwanted mutations on the plasmid backbone. To evaluate cell densities, optical density at 600 nm (OD600) was monitored using a microplate reader (SpectraMax 190 from Molecular Devices, China) with a sample volume of 200 μL.

Fig. 1
figure 1

Workflow of the growth-coupled evolution process for the selection of BA-PKT mutants. Nine BA-PKT mutant libraries were constructed by random mutagenesis. The mutant libraries were individually transformed into host Z188Δpfk cells and serially transferred in CGXII medium in nine wells of 24 deep-well microtiter plate. Three additional wells were used to cultivate the control Z188Δpfk (pTR-fxpk) strain expressing wildtype BA-PKT without mutagenesis

Full size image

Expression and purification of phosphoketolases

E. coli BL21 (DE3) was used as a protein expression host. The recombinant strains expressing the PKTs were cultivated in LB medium. When the OD600 of the culture reached 0.6–0.8, IPTG was added to a final concentration of 0.4 mM. Then, the cultures were cultivated for an additional 18 h at 20 °C. Afterwards, the cells were harvested by centrifugation (4 °C, 7000g, 5 min), and washed twice with 30 mL of buffer A (100 mM sodium phosphate buffer, pH 6.5). For protein purification, the cell pellets were resuspended in buffer A and disrupted by sonication (Scientz-IID, Ningbo Scientz Biotech. Co. Ltd). Then, cellular debris was removed by centrifugation at 12,000 rpm and 4 °C for 10 min. The PKT enzyme supernatants with 6 × His fusion tags were purified using His SpinTrap columns, and the bound proteins were eluted with concentration-gradient imidazole in buffer A. The pure enzymes were desalted using an Amicon Ultra-4 centrifugal concentrator (10 kDa) at 4 °C and exchanged into buffer B (100 mM sodium phosphate buffer, 1% glycerol, pH 6.5). Enzyme purity was estimated by 12% SDS–PAGE. Protein concentrations were determined according to the method of Bradford (Bradford 1976) using BCA Protein Assay Kit (ThermoFisher Scientific, USA).

Phosphoketolase enzyme activity assay

PKT enzyme activity measurement was through colorimetric assay as reported previously (Meile et al. 2001) with some modifications. The standard reaction mixture of 100 μL comprises sodium phosphate buffer (49 mM, pH 6.5), l-cysteine hydrochloride (0.7 mM), magnesium chloride (0.1 mM), thiamine pyrophosphate (0.1 mM), F6P or X5P as substrate (30 mM), and, finally, the purified enzyme (20 μg) to initiate the reaction. To investigate the effect of tested ligands on PKT activity, E4P, ACP, OAA, and PEP were added to the reaction mixture at a final concentration of 5 mM, respectively, while AMP and ATP were added at a final concentration of 3 mM, respectively. After setting up the reaction, mixing and pre-incubation at 30 °C in a rotary shaker for 10 min, 80 μl of hydroxylamine hydrochloride (2 M) was added and the mixture was further incubated at room temperature for 10 min. Thereafter, 55 μl of 15% (w/v) trichloroacetic acid, 55 μl of 4 M HCl and 55 μl of FeCl3.6H2O (5% (w/v) in 0.1 M HCl) were added sequentially for the final ruby color development of the ferric hydroxamate. The generated hydroxamate was spectrophotometrically quantified at 505 nm by comparing with a series of freshly prepared acetyl phosphate standards. One unit of PKT activity is defined as the amount of enzyme forming 1 μmol of acetyl phosphate per minute from F6P or X5P. Specific activity is expressed as units per milligram of protein. PKT activity is expressed in micromoles per minute per milligram of protein.

The kinetic parameters of the purified PKTs for F6P were determined by measuring the activities at different F6P concentrations (0.5–50 mM). The initial-velocity data obtained were fitted to the equation V = Vmax[S]/([S] + Km) (where V, initial velocity; Vmax, maximum velocity; [S], F6P concentration; and Km, Michaelis constant) by using the GraphPad Prism software (hyperbolic fitting) version 5.

For comparative measurements of crude activity of PKT enzyme in whole C. glutamicum cells harboring pTR-fxpk and pTR1-fxpk, cells were cultured in CGXII for 60 h, harvested by centrifugation (4 °C, 7000g, 5 min), washed twice with 30 mL of buffer A, and pretreated with CTAB to release enzymes as reported previously (Orban and Patterson 2000). Enzyme activity was measured using 30 mM F6P as substrate as described above.

l-glutamate production assay

For l-glutamate production, the seed medium contained (l−1) 50 g of glucose, 10 g of (NH4)2SO4, 0.8 g of MgSO4·7H2O, 700 μl of H3PO4 (v/v), 3 g of corn flour, 10 g of urea, 1 g of tryptone, 0.5 g of yeast extract. The fermentation medium comprises (l−1) 80 g of glucose, 10 g of (NH4)2SO4, 0.8 g of MgSO4·7H2O, 700 μl of H3PO4 (v/v), 84 g of MOPS, 3 g of corn flour, 10 g of urea. The media were adjusted to pH 7.0 using NaOH.

Seed cultures of Z188 and its derivatives expressing PKT and its mutants were prepared by transferring overnight cultures prepared in LBG medium into 24-deep-well plates containing 1 ml seed medium at an initial OD600 of 0.1. Cells at the exponential growth phase were transferred into 1 ml fermentation medium in 24-deep-well plates with an initial OD600 of 0.1. The cells were cultivated aerobically at 30 °C with shaking at 850 rpm for 33 h. The culture supernatant after appropriate dilution, was used for extracellular l-glutamate and residual glucose concentration measurements with the aid of SBA-40D Biosensor Analyzer (Institute of Biology, Shandong Province Academy of Sciences, China) as described previously (Wang et al. 2018c).

Flexible docking

The model structure of the wildtype PKT was constructed with the crystal dimeric structure of B. longum PKT (PDB ID: 3AI7) bound to TPP (thiamine pyrophosphate) as template (Takahashi et al. 2010; Zhang and Liu 2013) using Discovery Studio 4.1 software (BIOVIA, formerly Accelrys, USA). The protein active site was determined from PDB site records, and the active site was defined by a sphere of 11-Å radius. Flexible dockings of the ligand F6P into PKT active site were performed with Discovery Studio. Other parameters were kept as default settings in flexible docking.

Nucleotide accession number

The nucleotide sequence for the codon-optimized BA-PKT fragment used in the study had been deposited in GenBank database with the accession number MN081868.

Results

Evolution of phosphoketolase mutants conferring increased Z188Δpfk cell growth in glucose minimal medium

To improve PKT specific activity, in our study, we designed a growth-coupled PKT evolution strategy by inactivating PFK to block the EMP pathway and couple cell growth with PKT activity. We constructed Z188Δpfk, and expressed the wildtype fxpk using plasmids pTR-fxpk and pTR1-fxpk, with promoters Ptrc and its mutant promoter where part of the Ptrc sequence was replaced resulting in increased strength (TAACAATTTCACACA changed to TAACAATCTCACACA). We then compared the growths of Z188, Z188Δpfk, Z188Δpfk (pTR-fxpk), and Z188Δpfk (pTR1-fxpk) in CGXII medium. As shown in Fig. 2a, the OD600 of Z188Δpfk (pTR1-fxpk) at 60 h was about 1.4 fold higher compared with that of Z188Δpfk (pTR-fxpk). The growth rate of the unmodified parent strain was much higher (exceeded 14 fold, from 4 h to 12 h) than Z188Δpfk (pTR1-fxpk). Moreover, at 60 h, the Ptrc mutant led to about 1.5-fold improvement in PKT crude activity (Z188Δpfk (pTR1-fxpk); 12.46 ± 0.34 U/OD600 vs Z188Δpfk (pTR-fxpk); 8.36 ± 0.34 U/OD600). The results indicate that the growth of Z188Δpfk can be coupled to the level of BA-PKT enzyme activity under the designed experimental condition in our study. Moreover, the large growth gap between the parent strain Z188 and Z188Δpfk strain with significantly increased expression of PKT implies an immense potential for enhancing PKT enzyme activity.

Fig. 2
figure 2

Growth-coupled selection process for phosphoketolase mutants. a Comparison of cell growth of strains Z188, Z188Δpfk, and Z188Δpfk expressing wildtype BA-PKT with native Ptrc promoter (Z188Δpfk (pTR-fxpk)) and mutant Ptrc promoter with increased strength (Z188Δpfk (pTR1-fxpk)). b Growth-based selection of BA-PKT mutants. Cultures were serially transferred seven times in CGXII medium for strain enrichment. M1-M9; pools of cells from BA-PKT mutant library. Four cultures M2, M4, M7, and M8 were selected for further analyses based on their improved growth. c Growth assay of 3 clones selected from each of the 4 evolved BA-PKT mutant cultures. Based on sequence analyses, M21, M41, M71, M81, and M82 were selected for further analyses. d Growth curves for the reconstructed strains (M21N, M41N, M71N, M81N and M82N) expressing the selected BA-PKT mutants

Full size image

To apply the growth-based evolution strategy in selecting better PKT mutants, pTRCmob was used to construct nine parallel fxpk mutant libraries each with about 104 variants. The libraries were transformed into Z188Δpfk. Selection of mutants on the basis of growth improvement was thereafter carried out. The pools of cells from the mutant library were serially subcultured in CGXII medium for seven transfers. The growth rates for the pools of cells from the PKT mutant library (M1-M9) compared with the strain expressing the wildtype PKT are shown in Fig. 2b. It was observed that the growth of strains from M2, M4, M7, and M8 progressively improved during the evolution process. Following the selection process, mutant strains from these four libraries were subcultured on solid CGXII media plates for single colony isolation. Three colonies with large sizes were randomly selected from each library and further tested for growth (Fig. 2c). The fxpk mutants from two strains with better growth among the three selected colonies of each sample were sequenced. Except for M81 and M82, the other sequenced mutant fxpk fragments from the same library were the same (Fig. 2c and Table 2).

Table 2 Nucleotide and amino acid changes identified in BA-PKT mutants
Full size table

To eliminate the possibility of any adaptive changes which may arise from the genomic DNA and plasmid of the selected strains in the course of the evolution, the fxpk genes from mutants M21, M41, M71, M81, and M82 (Fig. 2c) were amplified with a high-fidelity DNA polymerase and cloned into pTR-fxpk (substituting original fxpk). The freshly constructed expression plasmids were retransformed into Z188Δpfk host strain with clean background. The growth curves of the retransformed strains (M21N, M41N, M71N, M81N, and M82N) in CGXII medium are shown in Fig. 2d. From the observed growth rate, it was evident that the PKT mutants improved the growth of the host strain.

Enzyme assay of evolved phosphoketolase mutants

To understand the basis for the growth improvement conferred by the evolved phosphoketolase mutants on Z188Δpfk in CGXII medium, in vitro BA-PKT enzyme assay was conducted on the pure enzymes using a colorimetry-based coupled enzyme assay system as reported previously (Meile et al. 2001). Since the BA-PKT used in this study is dual-substrate specific for F6P and X5P, we compared the activity of the wildtype enzyme using X5P or F6P as substrates. The wildtype enzyme displayed much higher activity towards F6P (12.07 ± 0.09 U/mg) compared with X5P (0.04 ± 0.00 U/mg) up to 302-fold, implying that the BA-PKT employed in this study displays higher substrate affinity towards F6P. This observation is consistent with a previous report in which the activity of X5P/F6P phosphoketolase was found to be much higher for F6P compared with X5P (Servinsky et al. 2012). In addition, the activities of five mutants with X5P as substrate were still a hundred-fold lower than the corresponding activities with F6P as substrate (Table S2). Since both F6P and X5P act as substrates for PKTs and facilitate the same theoretical yield improvement of acetyl CoA-derived products via the NOG pathway (Bogorad et al. 2013), and due to the much lower specific enzyme activity recorded from X5P, F6P was chosen as the substrate for the enzyme assays conducted thereafter. As shown in Fig. 3, in a similar trend observed with the growth rate, the specific enzyme activities of the PKT mutants improved significantly.

Fig. 3
figure 3

Enzyme activity assay of both evolved and constructed BA-PKT mutants with F6P as substrate. M21, M41, M71, M81, and M82 represent the BA-PKT mutants from the corresponding strains. For other mutants, the mutant sites are indicated. The F6P concentration was 30 mM. Error bars indicate standard deviation from three independent measurements

Full size image

The cumulative amino acid substitutions identified in the mutants were analyzed to identify the ones that are essential for the improved activity of the mutant BA-PKTs. To accomplish this, single mutations of each amino acid substitution were introduced into the wildtype BA-PKT, and the resulting single-site mutants were expressed and purified and their enzyme activities evaluated with 30 mM F6P as substrate. The SDS-PAGE bands of the representative PKTs which showed that they were of high purity and correspond to the reported 90-kDa molecular weight (Chinen et al. 2007; Henard et al. 2017; Meile et al. 2001) are shown in Fig. S1. From the enzyme assay results (Fig. 3), 11 of the 14 tested single-site mutants showed improved activity. The best three PKTs with mutations at position 2 (T2A), position 6 (I6T), and position 260 (H260Y), yielded 32.81 ± 0.50%, 44.10 ± 7.46%, and 52.41 ± 5.07% increased specific activities compared with the wildtype. This suggests that these three mutations contribute significantly to the improved activity of the BA-PKT mutants. To determine whether the combination of these three mutations will yield a synergistic effect on improved PKT activity, double and triple site mutations were introduced into the wildtype. As depicted in Fig. 3, the combined double (T2A/I6T) and triple mutations (T2A/I6T/H260Y), resulted in much more improved enzyme activity compared with single amino acid substitution at the respective sites, exhibiting 75.54 ± 3.00% and 76.37 ± 2.33% increased enzyme activities, respectively, compared with the wildtype. Notably, we observed that the combined mutations of the three site mutants (T2A/I6T/H260Y) did not yield a significant increase in the PKT specific activity compared with the double mutant (T2A/I6T), although the single-site mutant (H260Y) yielded significantly increased specific activity compared with the wildtype. The underlying reason for this observation seems to be complicated.

Kinetics analysis of phosphoketolase mutants

Since the three PKT mutants H260Y, T2A/I6T, and T2A/I6T/H260Y yielded the best improvement in enzyme activity (Fig. 3), they were selected for further kinetics analysis. Their Vmax values displayed 48.96 ± 7.40%, 59.12 ± 4.84% and 73.27 ± 3.25% respective increases more than the wildtype (Table 3). The Kcat/Km values of these mutant enzymes also improved significantly, with 60.34 ± 7.97%, 71.45 ± 5.22% and 80.16 ± 3.38% increases over that of the wildtype. The Km values of both the wildtype and mutant enzymes compare with previous reports (Bogorad et al. 2013; Meile et al. 2001). It is also noteworthy that there was no significant change in the Km values for the mutant enzymes compared with the wildtype, implying that both the mutant enzymes and the wildtype may demonstrate similar affinities towards F6P. To study the possible mechanism underlying the enzyme kinetics change, we built a model structure for the BA-PKT in this study with TPP as co-factor and performed the flexible dockings of the ligand F6P into the BA-PKT. As shown in Fig. S2, the F6P interacts with TPP at the enzyme active site pocket, but the mutant sites were far away from the active center. This may provide some explanation for the observed insignificant change in the Km of the mutant enzymes compared with the wildtype, with changes reflecting only in the respective Vmax values.

Table 3 Kinetic constants of purified BA-PKT and mutants
Full size table

Phosphoketolase mutants enhance l-glutamate yield

To demonstrate the potential of the mutant BA-PKTs in improving the production of acetyl CoA-derived products, we tested their effects on l-glutamate yield. Strain Z188 (pTR-fxpk) and its derivatives expressing the enriched BA-PKT mutants, the best three single-site mutants and the best two site-combined mutants, were constructed. The l-glutamate production and cell growth were tested in 24-deep-well culture plate. As shown in Fig. S3, no significant difference was observed in the growth rates of the tested strains. As illustrated in Fig. 4, in agreement with previous studies which improved the yield of acetyl CoA-associated products by PKT overexpression, utilizing the promoter Ptrc to express the wildtype BA-PKT, the l-glutamate yield increased by 9.02 ± 0.66%, compared with the strain without PKT. With the use of the mutant PKTs, l-glutamate titer and yield further increased to varying levels, with the triple mutant enzyme, T2A/I6T/H260Y displaying the highest percentage increase of 16.67 ± 2.24% and 18.19 ± 0.53%, respectively, compared with the wildtype BA-PKT. The results indicate that in addition to PKT overexpression, the obtained best PKT mutant with improved specific activity can highly enhance the production of l-glutamate. To further improve the l-glutamate yield from the PKT mutants, fermentation in jar reactors under optimized conditions is recommended in the future.

Fig. 4
figure 4

l-glutamate production assay of Z188 and its derivatives expressing BA-PKT variants. l-glutamate production titers (gray bars) and l-glutamate yield (black bars). PKT, M21, M41, M71, M81, and M82 represent Z188 derivatives expressing the corresponding PKT variants. For Z188 derivatives expressing other mutant BA-PKTs, the mutant sites are indicated. Error bars indicate standard deviation from three independent measurements. Cells were cultured for 33 hours before l-glutamate production was quantified

Full size image

Discussion

PKTs are promising enzymes for improving the production of a range of beneficial acetyl CoA-associated products. However, a significant factor limiting their potential applications lies in their low specific activities compared with that of PFK. Previous strategies employed strong promoters to improve PKT expression level (Bogorad et al. 2013; Chinen et al. 2007; Kocharin et al. 2013; Krusemann et al. 2018; Wang et al. 2018b). As a result, intracellular concentrations of the enzyme were increased and the yields of target acetyl CoA-associated products were significantly improved. For instance, by overexpressing BA-PKT in Escherichia coli, Bogorad et al. improved acetate yield from xylose by 500% (Bogorad et al. 2013). Similarly, by overexpressing PKT from Aspergillus nidulans in Saccharomyces cerevisiae, Kocharin et al. improved the yield of polyhydroxybutyrate from glucose by 50% (Kocharin et al. 2013). However, these studies failed to address the challenge of PKT low specific activity, which is the central focus of this study.

To improve the specific activity of enzymes, structure-guided protein design and directed enzyme evolution are two major enabling technologies widely applied. Rational protein design approach focuses on modification of residues located at or close to the enzyme active site, residues located at protein interfaces and residues which play important roles in stabilization of protein conformations, to generate new enzymes with improved functions (Bloom et al. 2005; Chen 2001). The demerit of this strategy is that it requires an exhaustive knowledge of enzyme active sites and structure-function relationships. Directed evolution on the other hand screens and selects new enzymes of desired functions from a constructed library of the parent protein, and thus do not require detailed protein structural and functional data. The application of growth-coupled evolution as an effective approach in selecting new enzymes of improved activity and other desirable characteristics had been demonstrated previously (Atsumi and Liao 2008; Lee et al. 2012; Liu et al. 2018; Pan et al. 2013). The key is to effectively couple the cell growth of the engineered strain solely to the activity of the target enzyme. In this study, by deleting pfk in Z188, carbon flux through the EMP pathway should be reduced and re-routed by heterologously expressed BA-PKT towards the alternative NOG pathway. Through this installed mechanism, a platform is created for coupling strain growth to improved BA-PKT activity. By means of evolution in glucose minimal medium, the obtained BA-PKT mutants were all confirmed by reintroduction into host cells, to be positively coupled with increased host cell growth (Fig. 2d), and the specific enzyme activities of all the tested PKT mutants improved significantly (Fig. 3), indicating the high efficiency of the established growth coupled PKT directed evolution strategy in our study. The growth-coupled evolution strategy was also tested in the model C. glutamicum strain ATCC 13032. Similar results as recorded for Z188 were obtained (Fig. 2a and Fig S4). This confirms that the PKT evolution strategy should also be applicable in other commonly accessible C. glutamicum model strains.

As a result of the installed growth-coupled evolution strategy, 5 better BA-PKT mutants M21, M41, M71, M81, and M82 were obtained. To determine if the mutants affected parameters such as feedback inhibition or allosteric activation, we investigated the function of 6 potential effectors E4P, ACP, OAA, PEP, ATP, and AMP on the enzymatic activities of wildtype BA-PKT and its mutants. E4P and ACP, the products of the reaction catalyzed by BA-PKT with F6P as substrate, were found to have an insignificant feedback inhibition effect on BA-PKTs’ activity (Fig. S5a). No previous report assessed the effects of E4P and ACP on PKT activity. As OAA, PEP, ATP, and AMP had been previously reported as the allosteric regulators of PKTs from both bacterial and eukaryotic origin (Glenn et al. 2014; Glenn and Smith 2015), their effects on both wildtype and mutant BA-PKTs isolated in this study were tested (Fig. S5). Less than 15% of BA-PKTs’ activity was reduced by 5 mM PEP, whereas 40% activities of both C. neoformans Xfp2 and L. plantarum Xfp were inhibited. The BA-PKTs were neither sensitive to 5 mM OAA, which is identical to L. plantarum Xfp; however, the activity of C. neoformans Xfp2 was inhibited by 50%. The BA-PKTs retained about 60% of the activity at 3 mM ATP, whereas only about 10% activity of C. neoformans Xfp2 was retained in the presence of 3 mM ATP, and R graminis Xfp activity was totally inhibited at 0.1 mM ATP (Glenn et al. 2014; Glenn and Smith 2015; Whitworth and Ratledge, 1977). AMP is an allosteric activator of C. neoformans Xfp2 activity; however, 3 mM AMP did not demonstrate a significant effect on BA-PKTs and L. plantarum Xfp’s activities. Altogether, the wildtype and mutant BA-PKTs are better than previously reported enzymes in resistance to inhibition by the six tested effectors. The overall effect of the mutant sites on the BA-PKT activity in the presence of the tested metabolites seemed not to be significant with the exception of ATP, which showed an observable different effect on the BA-PKTs.

Subsequent mutant site analysis of the BA-PKT mutants led to the construction of mutants, T2A, I6T, H260Y, T2A/I6T, and T2A/I6T/H260Y with 32.81 ± 0.50%, 44.10 ± 7.46%, 52.41 ± 5.07%, 75.54 ± 3.00% and 76.37 ± 2.33% respective improved specific enzyme activities compared with the codon-optimized wildtype BA-PKT (12.07 ± 0.09 U/mg). Kinetic analysis of the best three mutants H260Y, T2A/I6T, and T2A/I6T/H260Y revealed the Vmax values between 25–29 U/mg, and the Km values of 9.7–9.9 mM. Compared with the previously reported PKTs often employed in metabolic engineering, especially PKTs from B. adolescentis (Specific activity; 0.7 U/mg), B. animalis (Specific activity; 4.28 U/mg, Km; 10 mM), B. breve (Specific activity; 14.0 U/mg, Km; 9.7 mM), B. longum (Specific activity; 14.6 U/mg, Km; 26 mM) and L. plantarum (Specific activity; 1.8 U/mg, Km; 24 mM) activities for F6P (Bogorad et al. 2013; Grill et al. 1995; Meile et al. 2001; Suzuki et al. 2010a; Yevenes and Frey 2008), the best three codon-optimized BA-PKT mutants obtained in our study, especially T2A/I6T/H260Y, might serve as better candidates for metabolic engineering.

Based on the catalytic mechanism of PKTs as proposed previously, the N-terminal domain (residues 2–378) and the middle domain (residues 379–612) interact with the cofactor TPP and the substrate, F6P to facilitate the catalytic mechanism. The reported key enzyme active site residues are His64, His97, Gly155, Glu479, and His553, respectively (Suzuki et al. 2010a; Takahashi et al. 2010; Zhang and Liu 2013). In our efforts to understand the basis for the improved specific activities of the BA-PKT mutants, we constructed the 3D model structure of BA-PKT based on the crystal structure of B. longum phosphoketolase (PDB ID: 3AI7 (Takahashi et al. 2010; Zhang and Liu 2013)) which shows 59% amino acid sequence similarity. From our analysis, we discovered that all of the mutations including the mutants with higher activities (principally T2A, I6T, and H260Y) were located far away from the enzyme active site, and no clear interaction was disclosed by our preliminary structure analysis (Fig. S2). Therefore, the residues are generally hard to be predicted by structure-based rational design strategy, suggesting that directed evolution approach is a powerful alternative for the study of enzyme structure-function relationship. It deserves to mention that of the 11 beneficial mutant sites found in this study, 4 were located within the first 20 amino acids of the N-terminal domain of the protein. Moreover, two mutants out of the 4 residues, T2A, I6T, together with the double mutant, T2A/I6T displayed significantly improved enzyme activities, indicating the potential important contribution of the short sequence to the enzyme function.

To demonstrate the applicability of the BA-PKT mutants in improving the production of acetyl CoA-associated products, we tested their effects on l-glutamate production by Z188. Previously, by overexpressing PKT from B. animalis JCM1190 using the strong constitutively active cspB promoter (Peyret et al. 1993) in C. glutamicum, the l-glutamate yield of the strain improved by 9% (Chinen et al. 2007). In this study, using Ptrc for expressing wildtype BA-PKT in Z188, l-glutamate yield improved by 9.02 ± 0.66%, which is in agreement with the improved yield reported previously as a result of PKT overexpression (Wang et al. 2018b; Yang et al. 2016). However, with the expression of the best BA-PKT mutant T2A/I6T/H260Y, l-glutamate yield further improved by 18.19 ± 0.53% (Fig. 4). The results indicate that improving the specific activity of PKTs is a promising strategy for improving acetyl CoA-associated metabolites production.

In summary, in this study, we established an efficient growth-coupled evolution strategy for significantly improving the activities of PKTs in C. glutamicum, leading to better yield of l-glutamate. To further improve the specific activity of the PKT mutants to a comparable ratio to that of PFK, the key enzyme of the competing pathway, further rounds of growth-coupled evolution should be conducted. For the aim of enhanced yield of target products, optimization of downstream pathway to form a strong flux pulling force, coupled with optimization of the fermentation conditions is recommended. The established strategy can be also applied for improving the performance of other metabolism limiting enzymes for target applications.