- Email: [email protected]
13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli
Journal of Bioscience and Bioengineering VOL. xx No. xx, 1e6, 2016 www.elsevier.com/locate/jbiosc
13
C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli Keisuke Wada, Yoshihiro Toya, Satomi Banno, Katsunori Yoshikawa, Fumio Matsuda, and Hiroshi Shimizu* Department of Bioinformatic Engineering, Graduate School of Information Science and Technology, Osaka University, 1-5 Yamadaoka, Suita, Osaka 565-0871, Japan Received 15 June 2016; accepted 1 August 2016 Available online xxx
Mevalonate (MVA) is used to produce various useful products such as drugs, cosmetics and food additives. An MVAproducing strain of Escherichia coli (engineered) was constructed by introducing mvaES genes from Enterococcus faecalis. The engineered strain produced 1.84 mmol/gDCW/h yielding 22% (C-mol/C-mol) of MVA from glucose in the aerobic exponential growth phase. The mass balance analysis revealed that the MVA yield of the engineered strain was close to the upper limit at the biomass yield. Since MVA is synthesized from acetyl-CoA using NADPH as a cofactor, the production of MVA affects central metabolism in terms of carbon utilization and NADPH requirements. The reason for this highly efficient MVA production was investigated based on 13C-metabolic flux analysis. The estimated flux distributions revealed that the fluxes of acetate formation and the TCA cycle in the engineered strain were lower than those in the control strain. Although the oxidative pentose phosphate pathway is considered as the NADPH generating pathway in E. coli, no difference of the flux was observed between the control and engineered strains. The production/consumption balance of NADPH suggested that additional requirement of NADPH for MVA synthesis was obtained from the transhydrogenase reaction in the engineered strain. Comparison between the measured flux distribution and the ideal values for MVA production proposes a strategy for further engineering to improve the MVA production in E. coli. Ó 2016, The Society for Biotechnology, Japan. All rights reserved. [Key words:
13
C-metabolic flux analysis; Mevalonate; Escherichia coli; [1-13C] glucose; Flux distribution; Redox balance; Flux balance analysis]
Mevalonate (MVA) is a valuable precursor for the synthesis of drugs, cosmetics, and food additives (1e4), and can be produced by fermentation using microorganisms. MVA is synthesized from acetyl-CoA (AcCoA) in central carbon metabolism, and is subsequently used for the synthesis of isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) as essential building blocks of isoprenoids (5). High MVA-producing organisms have been obtained by screening techniques (6,7). It has been reported that Saccharomycopsis fibuligera ADK8107 produces 19 g L1 of MVA in 12 days (7). Since this yeast strain consumes MVA for isoprene synthesis, loss of the MVA product is natural for their survival and maintenance (8,9). Bacteria used for MVA production, such as Escherichia coli, do not possess the biosynthetic pathway via MVA, which have been investigated to avoid MVA consumption (9). E. coli produces IPP and DMAPP using an alternative pathway from pyruvate (Pyr) and glyceraldehyde 3-phosphate (GAP) (10). The heterologous enzymes of the mevalonate pathway from Streptococcus pneumoniae, Staphylococcus aureus, and Enterococcus faecalis have been studied for the efficient production of MVA and its derivatives in E. coli (1,4,9,11). Tabata and Hashimoto (9) achieved 47 g L1 of MVA
production for 2 days under fed-batch culture by introducing mvaES genes from E. faecalis as these genes function most effectively in E. coli. Since MVA is synthesized from AcCoA using NADPH as a cofactor (12), MVA production does affect central metabolism in terms of NADPH requirements and carbon utilization. It is important to optimize the flux distribution in central carbon metabolism for further improvement of MVA production because NADPH is generated by some reactions in these metabolic pathways. However, currently, little is known about the flux distribution in the central carbon metabolism of the MVA-producing strains. In the present study, an MVA producing E. coli (engineered) strain was also constructed by introduction of mvaES genes from E. faecalis. The effect of MVA production on the central carbon metabolism was investigated by 13C-metabolic flux analysis (13CMFA). 13C-MFA is a method to experimentally estimate the flux distribution in pathways based on mass balance constraints and isotope labeling measurements (13,14). The flux distribution suggested that additional requirements of NADPH for MVA synthesis are obtained from the transhydrogenase reaction. The flux distributions provide a strategy for rational engineering to improve the production of MVA.
* Corresponding author. Tel.: þ81 6 6879 7446; fax: þ81 6 6879 4359. E-mail addresses: [email protected] (K. Wada), ytoya@ist. osaka-u.ac.jp (Y. Toya), [email protected] (S. Banno), yoshikawa@ ist.osaka-u.ac.jp (K. Yoshikawa), [email protected] (F. Matsuda), [email protected] (H. Shimizu).
MATERIALS AND METHODS Bacterial strains and culture condition All E. coli strains were grown as pre-pre-cultures overnight at 37 C in Lennox medium (10 g L1 tryptone, 5 g L1 yeast extract, 5 g L1 NaCl, 1 g L1 glucose). Pre-pre-cultures were transferred to
1389-1723/$ e see front matter Ó 2016, The Society for Biotechnology, Japan. All rights reserved. http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
WADA ET AL.
J. BIOSCI. BIOENG.,
Construction of the plasmid and strain The E. coli strains used in this study are derived from MG1655(DE3) (15). The plasmid pCOLADuet-1 (Merck KGaA, Darmstadt, Germany) was used as the vector. Acetoacetyl-CoA synthase/3hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase gene (mvaE) and HMG-CoA synthase mvaS genes derived from E. faecalis were synthesized by GeneArt (Thermo Fisher Scientific, MA, USA) according to E. coli codon usage. These genes were trimmed with NdeI and XhoI, and then subcloned into pCOLADuet-1 to generate pCOLADuet-1/mvaES (mvaE and mvaS expression plasmid, T7 promoter, KmR). E. coli MG1655(DE3) was transformed with pCOLADuet-1 (control) and pCOLADuet-1/mvaES (engineered), respectively. Analyses of dry cell weight and extracellular metabolites Cell growth was monitored by measuring OD600 using the UVmini-1240 (Shimadzu, Kyoto, Japan). Dry cell weight was calculated using a conversion coefficient of 0.3 gDCW L-1 OD-1 600 . The concentrations of glucose and organic acids (lactate, formate, acetate, succinate and MVA) in the supernatant of the culture broth were determined by an HPLC system (Shimadzu) equipped with an Aminex HPX-87X column (BioRad, Hercules, CA, USA), UV/vis detector (SPD-20A), and a refractive index detector (RID-10A). The column temperature was set to 65 C, and 1.5 mM H2SO4 was used as the mobile phase with a flow rate of 0.5 ml min1. The supernatant of the culture broth was obtained by centrifugation at 18,000 g for 5 min at 4 C, and then filtered through a Millex HV 0.45-mm filter (Merck KGaA). Flux prediction by simulation analysis Constraint-based Flux Balance Analysis (FBA) was performed by using the core metabolic model of E. coli K-12, named as ecoli_core_model obtained from BiGG Models (16). All calculations were performed using a solver for linear programming, the GNU Linear Programming Kit in Matlab and the COBRA Toolbox (17). The MVA synthesis pathway was added to this model. The glucose uptake rate was set to the experimental value of 7.74 and 8.21 mmol/gDCW/h for control and engineered strains, respectively. In both strains, the non-growth-associated ATP maintenance requirement rate was set to the default value (8.39 mmol/gDCW/h), which was estimated using experimental data from continuous cultures with glucose as a carbon source (18). To illustrate the relationship between the biomass yield and the MVA yield, the maximum growth rate was calculated using an objective function as a maximization of biomass production. The maximum and minimum fluxes of MVA production were calculated using an objective function as the maximization or minimization of MVA production with a constraint of each fixed growth rate from zero to the maximum value. The MVA production rate and the growth rate were converted to carbon yields using coefficient 6 C-mol/mol-MVA and 63.1 C-mol/gDCW, respectively. The flux distribution with theoretical MVA production yield was calculated by FBA using an objective function as a maximization of MVA production. To minimize the Euclidian norm of internal fluxes, the optional parameter minNorm of optimizeCbModel function was set to ‘one’. The glucose uptake rate was set to the experimentally measured value (8.21 mmol/gDCW/h) of the engineered strain.
Control 10
OD600
20 1
15 10
0.1
5 0.01
0
2 OD600
4
6 8 10 Time (h) Glucose
12
0
MDV5e6h ¼ where MDVxh is
MDV6h OD6h MDV5h OD5h OD6h OD5h
(1)
13
C-entrichment at x h, and ODxh is the cell concentration at x h.
13
13 C-metabolic flux analysis C-MFA was performed using the OpenMebius software (20), which is based on the elementary metabolite units framework (21) in Matlab (MathWorks Inc., Natick, MA, USA). The building block and cofactor requirements for E. coli biomass formation was referred to in a previous study (22). The metabolic pathway model of E. coli that was used in this study has been described previously (19). The following MVA synthesis pathway was added. The amino acid fragments were chosen according to the previous study (23). Metabolic fluxes were estimated by minimizing the residual sum of squares between the experimentally measured and model predicted 13C-enrichment using the fmincon optimization solver in the Matlab toolbox. The standard deviation (SD) of 13C-enrichment was set to 0.01. The 95% confidence intervals were calculated using a grid search method as described in previous studies (24).
Malic enzyme activity measurement The crude cell extract was prepared from M9 batch cultures that were harvested at 2 h after the addition of IPTG. Cells were washed twice with 0.9% NaCl, and disrupted by sonication. The activity of two types of malic enzymes (SfcA and MaeB) was measured as described in a previous study (25). Enzyme activities were measured spectrophotometrically at 30 C by monitoring NADPH production at 340 nm with the UVmini-1240 system (Shimadzu). The reaction was started by addition of 10 mL of crude cell extract. The protein concentration in the supernatant was determined by using the Bradford method.
RESULTS AND DISCUSSION Growth characteristics of the MVA-producing strain The control and MVA-producing (engineered) strains were constructed by transforming E. coli MG1655(DE3) with pCOLADuet-1 and pCOLADuet-1/mvaES harboring acetoacetyl-CoA synthase/HMGCoA reductase gene (mvaE) and HMG-CoA synthase gene (mvaS) from E. faecalis, respectively. The codon usage of mvaES genes and its expression vector were modified from the previous report (9). The control and engineered strains were grown in batch cultures on glucose under aerobic conditions. IPTG was added at 4 h after inoculation for inducing the expression of mvaES. Fig. 1 shows the cell growth and metabolite production of the two strains. Growth parameters are summarized in Table 1. These specific rates were calculated at the exponential phase between 4 and 6.5 h. Both
B
25 Concentration (mM)
A
13 Analysis of 13C-enrichment of proteinogenic amino acids C-enrichment of proteinogenic amino acids was measured by GCeMS. Broth culture (7.5 ml) was taken from the flask, and centrifuged at 7000 g, for 5 min at 4 C. The cell pellet was then washed twice with 0.9% NaCl, and hydrolyzed in 6 mol/L HCl at 105 C for 18 h. The resulting proteinogenic acids were derivatized with N-(tert-butyldimethylsilyl)N-methyl-trifluoroacetamide containing tert-butyldimethylchlorosilane in acetonitrile at 105 C for 1 h, and then analyzed by a GCeMS [Agilent 7890A GC and 5975C Mass Selective Detector (Agilent Technologies, Santa Clara, USA)] equipped with a DB5MSþDG column (Agilent Technologies). The analytical conditions used are described elsewhere (19). The data obtained from GCeMS were corrected by reduction of the natural abundance ratio of C, H, O, N, and Si isotopes. Cumulative 13C-enrichment from 5 to 6 h after cultivation was calculated using the following equation:
Engineered 10
25 20
1
15 10
0.1
5 0.01
Succinate
0
2
4
6 8 Time (h) Acetate
10
12
Concentration (mM)
50 ml of M9 medium (17 g L1 Na2HPO4$12H2O, 3.1 g L1 KH2PO4, 1 g L1 NH4Cl, 0.5 g L1 NaCl, 0.25 g L1 MgSO4$7H2O, 15 mg L1 CaCl2$2H2O, 8.1 mg L1 FeCl3, 1 mg L1 MnSO4$4H2O, 1.7 mg L1 ZnCl2, 0.43 mg L1 CuCl2$2H2O, 0.6 mg L1 CoCl2$6H2O, 0.6 mg L1 Na2MoO4$2H2O, and 4 g L1 glucose) to form pre-cultures with an initial optical density of 0.05 at 600 nm (OD600). Pre-culture was performed for reducing the risk of interfusion of rich culture medium. After 24 h, pre-cultures were inoculated in 50 ml of M9 medium containing [1-13C] glucose as the main culture with an initial OD600 of 0.05. [1-13C] glucose was purchased from Cambridge Isotope Laboratories (Andover, MA, USA). Kanamycin (0.03 g L1) and isopropylthiogalactoside (IPTG, 24 mg L1) was added whenever necessary. Dry cell mass was determined by filtration, washing of cells, and drying at 70 C until a constant biomass of an equivalent of 25 mL culture volume at an OD600 of 1.0 was obtained.
OD600
2
0
MVA
FIG. 1. Growth characteristics of control (A) and engineered (B) strains in batch cultures grown on glucose under aerobic conditions. Closed squares, biomass; circles, glucose; triangles, acetate; open squares, succinate; diamonds, MVA. Data shown are mean SD (n ¼ 3).
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
VOL. xx, 2016
ANALYSIS OF MVA-PRODUCING E. COLI
3
TABLE 1. Growth characteristics. Strain
Growth rate (1/h)
Specific rate (mmol/gDCW/h) Acetate production
Succinate production
Mevalonate production
Mevalonate
7.74 0.17 8.21 0.22
4.41 0.08 0.42 0.24
0.20 0.00 0.08 0.04
N.D. 1.84 0.07
N.D. 0.22 0.00
0.43 0.01 0.48 0.01
Control Engineered
Yield (C-mol/C-mol)
Glucose uptake
The growth characteristics of the control and engineered strains during 4e6.5 h. The data are presented as means SD (n ¼ 3). N.D. represents not detected.
strains converted glucose to biomass, acetate, succinate, and MVA. No other by-products were detected by HPLC. The growth rate and glucose uptake rate in the engineered strain were slightly faster than in the control strain. Acetate production rate in the engineered strain was reduced to 10% of the control strain. While no MVA production was observed for the control strain, the engineered strain produced MVA with the productivity at 1.84 mmol/gDCW/h; corresponding to the yield at 22% MVA Cmol/Glucose C-mol.
Evaluation of MVA production based on mass balance analysis One molecule of MVA is synthesized from three molecules of AcCoA (carbon backbone) using two molecules of NADPH as a cofactor. Since AcCoA and NADPH are also required for cell growth, the capacity of the MVA production yield should be restricted by the biomass yield. The yields of MVA and biomass of the strains were evaluated in terms of mass balances. To reveal the relationship between biomass yield and MVA yield, the maximum and minimum rates of MVA production at each growth rate were calculated using an E. coli core metabolic pathway model. The growth rate and the MVA production rate were normalized to the yield of carbon atoms. The gray area in Fig. 2 indicates the feasible MVA yield at each biomass yield; thus, the measured data must be located in this area. The open square symbol represents a metabolic state in which the cells can most efficiently use glucose for cell growth with no MVA production. In contrast, the open triangle symbol represents a state in which the cells can most efficiently convert glucose to MVA without growth. The measured carbon yields of biomass and MVA were 22 (Cmol/C-mol) and 62 (C-mol/C-mol) in the engineered strain, respectively, and were 0 (C-mol/C-mol) and 58 (C-mol/C-mol) in the control strain, respectively. Open and closed circles in Fig. 2 represent the MVA production of the engineered and control strains, respectively. It is confirmed that both data are located
MVA yield (MVA C-mol/Substrate C-mol)
100
Control Engineered
80
Theoretical maximum yield of biomass Theoretical maximum yield of MVA
60 40
(62, 22) 20 (58, 0) 0 0
20 40 60 80 100 Biomass yield (Biomass C-mol/Substrate C-mol)
FIG. 2. Feasible range of the MVA yield against the biomass yield. The gray area indicates the feasible MVA production yield. Open and closed circles indicate the measured metabolic states of engineered and control strains. Open square and open triangle represent metabolic states in which the cells can most efficiently use glucose for cell growth or MVA production.
within the feasible area in Fig. 2. The MVA yield of the engineered strain was close to the upper limit at the biomass yield. Comparison of 13C-flux distributions between control and engineered strains The metabolic flux distributions of control and engineered strains were estimated by 13C-MFA. The cells were cultured using [1-13C] glucose as sole carbon source. The 13C-enrichments of proteinogenic amino acids were measured by GCeMS. In this study, the proteinogenic amino acids produced at the exponential phase between 5 and 6 h were used for the flux estimations. The malic enzyme reaction was not considered in the 13CMFA due to the low specific activity (<0.01 mmol/gDCW/h) by the enzyme assay. The measured and estimated 13C-enrichments are shown in Table S1. The threshold of chi-squared test for the goodness of fit was 88 (the number of degrees of freedom of the reaction model was 20 and the number of independent data was 88). Both control and engineered strains passed the chi-squared test as the values in these strains were 42 and 20, respectively. This suggests that the estimated flux distributions can statistically explain the experimentally measured 13C-enrichment of amino acids. The absolute flux values (mmol/gDCW/h) of the best-fit estimation in each reaction are shown on the pathway map in Fig. 3. The best-fit values of metabolic fluxes were estimated by minimizing the residual sum of squares between the experimentally measured and model-predicted 13C-enrichment. For example, the glucose (7.7 mmol/gDCW/h) was metabolized through glycolysis and the pentose phosphate (PP) pathway with fluxes of 5.0 and 2.7 mmol/gDCW/h in the control strain, respectively. The flux distribution of the control strain was similar to the previously reported results of wild type E. coli (26). Whereas no significant differences were observed in the fluxes of glycolysis, PP pathway, and the ED pathway between control and engineered strains, there were differences in the fluxes in the pathways downstream of AcCoA. Although the acetate formation flux was 4.4 mmol/gDCW/h in the control strain, the flux decreased to 0.4 mmol/gDCW/h and the MVA formation flux increased to 1.8 mmol/gDCW/h in the engineered strain. The TCA cycle flux of the engineered strain was lower than that of the control strain. It seems that AcCoA was diverted to the MVA synthesis pathway in the engineered strain due to overexpression of enzymes in the pathway. NADPH generation in the entire central carbon metabolism does affect MVA production since MVA synthesis requires NADPH as a cofactor. NADPH is produced by glucose-6-phosphate dehydrogenase, phosphogluconate dehydrogenase, isocitrate dehydrogenase, and NADPþ-dependent malic enzyme, and is consumed for cell growth and MVA synthesis. Although the oxidative PP pathway is considered as the NADPH-generating pathway in E. coli, the flux was not significantly different between the control and engineered strains. The isocitrate dehydrogenase flux of the engineered strain was lower than that of the control strain since the AcCoA was diverted from the TCA cycle to MVA synthesis. No enzymatic activity of the NADPþ-dependent malic enzyme was confirmed in both strains. Because the imbalance between NADPH production and consumption rates is maintained by transhydrogenase reactions to achieve homeostasis in cells (27), the flux of the transhydrogenase reaction was calculated based on the mass balance. Fig. 4 shows the NADPH balance in the control and engineered
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
4
WADA ET AL.
J. BIOSCI. BIOENG.,
FIG. 3. (A) Metabolic flux distributions in the central metabolic pathways of the control and engineered strains during exponential growth phase. (B) Simulated metabolic flux distribution with maximal theoretical MVA yield. The numbers on the arrows indicate absolute flux values (mmol/gDCW/h). Asterisks indicate biomass formation.
strains calculated from flux distributions. The total NADPH production/consumption rate in the engineered strain was 1.5 times higher than that in the control strain. In other words, the turnover rate of NADPH in the engineered strain was higher than that of the control strain. This difference was related to the high requirement of NADPH for MVA synthesis in the engineered strain. The balance of production/consumption NADPH rate suggested that the additional requirement of NADPH for MVA synthesis was obtained from the transhydrogenase reaction in the engineered strain. He et al. reported the flux responses to the overproduction of fatty acids in E. coli (26). The fatty acids are synthesized from AcCoA using NADPH as a cofactor, as with MVA. The observed decrease in acetate production and the increase in transhydrogenase flux
converting NADH to NADPH in the engineered strain in the present study were in accordance with the flux responses in the fatty acidproducing strain. On the other hand, the growth reduction, the increasing of the PP pathway flux, and the decreasing of the anaplerotic pathway flux in the fatty acid-producing strain were not observed in the MVA-producing strain. It has been reported that the overproduction of fatty acids in E. coli requires significantly higher maintenance energy than in the control strain because the overproduction of free fatty acids induces cellular membrane stress (28). In the case of MVA production, the growth reduction did not occur due to the non-toxicity of MVA. Since the role of anaplerotic reactions in the TCA cycle is to supply the precursors of biomass components, the small difference in the flux between the control
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
VOL. xx, 2016
15
ANALYSIS OF MVA-PRODUCING E. COLI
Control
Engineered
mmol/gDCW/h
12 9
5
suggested that the additional requirements of NADPH for MVA synthesis were obtained from the transhydrogenase reaction. Furthermore, comparison of the experimentally obtained and ideal fluxes could be useful for planning a strategy for metabolic modification for future engineering towards increasing the MVA production. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.jbiosc.2016.08.001.
6 ACKNOWLEDGMENTS
3 This work was partially supported by the Grant in Aid for Scientific Research (A) No. 24246134 and by the Grant in Aid for Scientific Research (B) No. 16H04576.
0
References
Citrate synthase
Cell synthesis
G6P dehydrogenase
MVA synthesis
6PG dehydrogenase
Transhydrogenase
FIG. 4. Estimated NADPH balance in control and engineered strains using the best-fit metabolic flux values. Transhydrogenases function as equilibrators of NADPH balance. Error-bars indicate 95% confidence intervals of each flux.
and engineered strains is consistent. Furthermore, a fatty acid (C16:0) requires 8 AcCoA and 14 NADPH molecules, whereas an MVA requires 3 AcCoA and 2 NADPH (26). MVA production requires much less NADPH than fatty acid production; thus, the MVAproducing strain can cover the NADPH needs for MVA synthesis by increasing the transhydrogenase flux without increasing PP pathway flux. Prediction of engineering for further MVA production The potential for improvement of MVA production by further engineering was investigated by comparing current and ideal flux distributions. The ideal flux distribution with theoretical maximum MVA production yield was calculated by flux balance analysis using an objective function as a maximization of MVA production (Fig. 3B). The glucose uptake rate was set to the experimentally measured value (8.2 mmol/gDCW/h) of the engineered strain. This metabolic state corresponds to the position of the open triangles in Fig. 2. The uptake glucose is catabolized only through glycolysis and is used for MVA synthesis without competitive pathways such as acetate formation and the TCA cycle in this ideal metabolic state. Repression of the citrate synthase (CS) or the PP pathway would be required to improve MVA production because excess fluxes were observed in these pathways by 13CMFA. However, deletion of CS is undesirable for bioproduction due to gene essentiality. A switching system of CS expression might be effective. Soma et al. (29) already developed a switch for CS expression during the exponential growth phase to efficiently produce isopropanol in E. coli, and succeeded in increasing productivity. This system may be useful for effective MVA production by achieving precise control of the CS flux. The fed-batch process development is also suitable to achieve high titer production of MVA using this strain. The cells enter the stationary phase by CS flux repression. It is expected that a high yield and titer production of MVA can be achieved by maintaining this state for a long period, since the carbons are not consumed for the biomass synthesis. In the present study, we evaluated the metabolic state of an MVA-producing E. coli strain based on 13C-MFA. The metabolic flux
1. Martin, V. J., Pitera, D. J., Withers, S. T., Newman, J. D., and Keasling, J. D.: Engineering a mevalonate pathway in Escherichia coli for production of terpenoids, Nat. Biotechnol., 21, 796e802 (2003). 2. Haratake, A., Ikenaga, K., Katoh, N., Uchiwa, H., Hirano, S., and Yasuno, H.: Topical mevalonic acid stimulates de novo cholesterol synthesis and epidermal permeability barrier homeostasis in aged mice, J. Invest. Dermatol., 114, 247e252 (2000). 3. Wang, C., Yoon, S. H., Shah, A. A., Chung, Y. R., Kim, J. Y., Choi, E. S., Keasling, J. D., and Kim, S. W.: Farnesol production from Escherichia coli by harnessing the exogenous mevalonate pathway, Biotechnol. Bioeng., 107, 421e429 (2010). 4. Yoon, S. H., Lee, S. H., Das, A., Ryu, H. K., Jang, H. J., Kim, J. Y., Oh, D. K., Keasling, J. D., and Kim, S. W.: Combinatorial expression of bacterial whole mevalonate pathway for the production of beta-carotene in E. coli, J. Biotechnol., 140, 218e226 (2009). 5. Miziorko, H. M.: Enzymes of the mevalonate pathway of isoprenoid biosynthesis, Arch. Biochem. Biophys., 505, 131e143 (2011). 6. Tamura, G., Ando, K., Kodama, K., and Arima, K.: Production of mevalonic acid by fermentation, Appl. Microbiol., 16, 965e972 (1968). 7. Kuzuyama, T., Hemmi, H., and Takahashi, S.: Mevalonate pathway in bacteria and archaea, pp. 493e516, in: Mander, L. and Liu, H.-W. (Eds.), Comprehensive natural products II, vol. 1. Elsevier, Oxford (2010). 8. Kuzuyama, T., Dairi, T., Yamashita, H., Shoji, Y., and Seto, H.: Heterologous mevalonate production in Streptomyces lividans TK23, Biosci. Biotechnol. Biochem., 68, 931e934 (2004). 9. Tabata, K. and Hashimoto, S.: Production of mevalonate by a metabolicallyengineered Escherichia coli, Biotechnol. Lett., 26, 1487e1491 (2004). 10. Kuzuyama, T.: Mevalonate and nonmevalonate pathways for the biosynthesis of isoprene units, Biosci. Biotechnol. Biochem., 66, 1619e1627 (2002). 11. Zheng, Y., Liu, Q., Li, L., Qin, W., Yang, J., Zhang, H., Jiang, X., Cheng, T., Liu, W., Xu, X., and Xian, M.: Metabolic engineering of Escherichia coli for highspecificity production of isoprenol and prenol as next generation of biofuels, Biotechnol. Biofuels, 6, 57 (2013). 12. Hedl, M., Sutherlin, A., Wilding, E. I., Mazzulla, M., McDevitt, D., Lane, P., Burgner, J. W., 2nd, Lehnbeuter, K. R., Stauffacher, C. V., Gwynn, M. N., and Rodwell, V. W.: Enterococcus faecalis acetoacetyl-coenzyme A thiolase/3hydroxy-3-methylglutaryl-coenzyme A reductase, a dual-function protein of isopentenyl diphosphate biosynthesis, J. Bacteriol., 184, 2116e2122 (2002). 13. Wiechert, W.: 13C metabolic flux analysis, Metab. Eng., 3, 195e206 (2001). 14. Wittmann, C.: Fluxome analysis using GC-MS, Microb. Cell Fact., 6, 6 (2007). 15. Tokuyama, K., Ohno, S., Yoshikawa, K., Hirasawa, T., Tanaka, S., Furusawa, C., and Shimizu, H.: Increased 3-hydroxypropionic acid production from glycerol, by modification of central metabolism in Escherichia coli, Microb. Cell Fact., 13, 64 (2014). 16. King, Z. A., Lu, J., Dräger, A., Miller, P., Federowicz, S., Lerman, J. A., Ebrahim, A., Palsson, B.Ø., and Lewis, N. E.: BiGG models: a platform for integrating, standardizing and sharing genome-scale models, Nucleic Acids Res., 44, D515eD522 (2015). 17. Becker, S. A., Feist, A. M., Mo, M. L., Hannum, G., Palsson, B.Ø., and Herrgard, M. J.: Quantitative prediction of cellular metabolism with constraintbased models: the COBRA Toolbox, Nat. Protoc., 2, 727e738 (2007). 18. Feist, A. M., Henry, C. S., Reed, J. L., Krummenacker, M., Joyce, A. R., Karp, P. D., Broadbelt, L. J., Hatzimanikatis, V., and Palsson, B.Ø.: A genomescale metabolic reconstruction for Escherichia coli K-12 MG1655 that accounts for 1260 ORFs and thermodynamic information, Mol. Syst. Biol., 3, 121 (2007). 19. Okahashi, N., Kajihata, S., Furusawa, C., and Shimizu, H.: Reliable metabolic flux estimation in Escherichia coli central carbon metabolism using intracellular free amino acids, Metabolites, 4, 408e420 (2014).
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
6
WADA ET AL.
20. Kajihata, S., Furusawa, C., Matsuda, F., and Shimizu, H.: OpenMebius: an open source software for isotopically nonstationary 13C-based metabolic flux analysis, Biomed. Res. Int., 2014, 627014 (2014). 21. Antoniewicz, M. R., Kelleher, J. K., and Stephanopoulos, G.: Elementary metabolite units (EMU): a novel framework for modeling isotopic distributions, Metab. Eng., 9, 68e86 (2007). 22. Ingraham, J. L., Maaløe, O., and Neidhardt, F. C.: Growth of the bacterial cell pp. 128. Sinauer Associates Inc., Sunderland, MA, USA (1983). 23. Antoniewicz, M. R., Kelleher, J. K., and Stephanopoulos, G.: Accurate assessment of amino acid mass isotopomer distributions for metabolic flux analysis, Anal. Chem., 79, 7554e7559 (2007). 24. Antoniewicz, M. R., Kelleher, J. K., and Stephanopoulos, G.: Determination of confidence intervals of metabolic fluxes estimated from stable isotope measurements, Metab. Eng., 8, 324e337 (2006). 25. Bologna, F. P., Andreo, C. S., and Drincovich, M. F.: Escherichia coli malic enzymes: two isoforms with substantial differences in kinetic properties, metabolic regulation, and structure, J. Bacteriol., 189, 5937e5946 (2007).
J. BIOSCI. BIOENG., 26. He, L., Xiao, Y., Gebreselassie, N., Zhang, F., Antoniewiez, M. R., Tang, Y. J., and Peng, L.: Central metabolic responses to the overproduction of fatty acids in Escherichia coli based on 13C-metabolic flux analysis, Biotechnol. Bioeng., 111, 575e585 (2014). 27. Sauer, U., Canonaco, F., Heri, S., Perrenoud, A., and Fischer, E.: The soluble and membrane-bound transhydrogenases UdhA and PntAB have divergent functions in NADPH metabolism of Escherichia coli, J. Biol. Chem., 279, 6613e6619 (2004). 28. Youngquist, J. T., Lennen, R. M., Ranatunga, D. R., Bothfeld, W. H., Marner, W. D., 2nd, and Pfleger, B. F.: Kinetic modeling of free fatty acid production in Escherichia coli based on continuous cultivation of a plasmid free strain, Biotechnol. Bioeng., 109, 1518e1527 (2012). 29. Soma, Y., Tsuruno, K., Wada, M., Yokota, A., and Hanai, T.: Metabolic flux redirection from a central metabolic pathway toward a synthetic pathway using a metabolic toggle switch, Metab. Eng., 23, 175e184 (2014).
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
13
C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli Keisuke Wada, Yoshihiro Toya, Satomi Banno, Katsunori Yoshikawa, Fumio Matsuda, and Hiroshi Shimizu* Department of Bioinformatic Engineering, Graduate School of Information Science and Technology, Osaka University, 1-5 Yamadaoka, Suita, Osaka 565-0871, Japan Received 15 June 2016; accepted 1 August 2016 Available online xxx
Mevalonate (MVA) is used to produce various useful products such as drugs, cosmetics and food additives. An MVAproducing strain of Escherichia coli (engineered) was constructed by introducing mvaES genes from Enterococcus faecalis. The engineered strain produced 1.84 mmol/gDCW/h yielding 22% (C-mol/C-mol) of MVA from glucose in the aerobic exponential growth phase. The mass balance analysis revealed that the MVA yield of the engineered strain was close to the upper limit at the biomass yield. Since MVA is synthesized from acetyl-CoA using NADPH as a cofactor, the production of MVA affects central metabolism in terms of carbon utilization and NADPH requirements. The reason for this highly efficient MVA production was investigated based on 13C-metabolic flux analysis. The estimated flux distributions revealed that the fluxes of acetate formation and the TCA cycle in the engineered strain were lower than those in the control strain. Although the oxidative pentose phosphate pathway is considered as the NADPH generating pathway in E. coli, no difference of the flux was observed between the control and engineered strains. The production/consumption balance of NADPH suggested that additional requirement of NADPH for MVA synthesis was obtained from the transhydrogenase reaction in the engineered strain. Comparison between the measured flux distribution and the ideal values for MVA production proposes a strategy for further engineering to improve the MVA production in E. coli. Ó 2016, The Society for Biotechnology, Japan. All rights reserved. [Key words:
13
C-metabolic flux analysis; Mevalonate; Escherichia coli; [1-13C] glucose; Flux distribution; Redox balance; Flux balance analysis]
Mevalonate (MVA) is a valuable precursor for the synthesis of drugs, cosmetics, and food additives (1e4), and can be produced by fermentation using microorganisms. MVA is synthesized from acetyl-CoA (AcCoA) in central carbon metabolism, and is subsequently used for the synthesis of isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) as essential building blocks of isoprenoids (5). High MVA-producing organisms have been obtained by screening techniques (6,7). It has been reported that Saccharomycopsis fibuligera ADK8107 produces 19 g L1 of MVA in 12 days (7). Since this yeast strain consumes MVA for isoprene synthesis, loss of the MVA product is natural for their survival and maintenance (8,9). Bacteria used for MVA production, such as Escherichia coli, do not possess the biosynthetic pathway via MVA, which have been investigated to avoid MVA consumption (9). E. coli produces IPP and DMAPP using an alternative pathway from pyruvate (Pyr) and glyceraldehyde 3-phosphate (GAP) (10). The heterologous enzymes of the mevalonate pathway from Streptococcus pneumoniae, Staphylococcus aureus, and Enterococcus faecalis have been studied for the efficient production of MVA and its derivatives in E. coli (1,4,9,11). Tabata and Hashimoto (9) achieved 47 g L1 of MVA
production for 2 days under fed-batch culture by introducing mvaES genes from E. faecalis as these genes function most effectively in E. coli. Since MVA is synthesized from AcCoA using NADPH as a cofactor (12), MVA production does affect central metabolism in terms of NADPH requirements and carbon utilization. It is important to optimize the flux distribution in central carbon metabolism for further improvement of MVA production because NADPH is generated by some reactions in these metabolic pathways. However, currently, little is known about the flux distribution in the central carbon metabolism of the MVA-producing strains. In the present study, an MVA producing E. coli (engineered) strain was also constructed by introduction of mvaES genes from E. faecalis. The effect of MVA production on the central carbon metabolism was investigated by 13C-metabolic flux analysis (13CMFA). 13C-MFA is a method to experimentally estimate the flux distribution in pathways based on mass balance constraints and isotope labeling measurements (13,14). The flux distribution suggested that additional requirements of NADPH for MVA synthesis are obtained from the transhydrogenase reaction. The flux distributions provide a strategy for rational engineering to improve the production of MVA.
* Corresponding author. Tel.: þ81 6 6879 7446; fax: þ81 6 6879 4359. E-mail addresses: [email protected] (K. Wada), ytoya@ist. osaka-u.ac.jp (Y. Toya), [email protected] (S. Banno), yoshikawa@ ist.osaka-u.ac.jp (K. Yoshikawa), [email protected] (F. Matsuda), [email protected] (H. Shimizu).
MATERIALS AND METHODS Bacterial strains and culture condition All E. coli strains were grown as pre-pre-cultures overnight at 37 C in Lennox medium (10 g L1 tryptone, 5 g L1 yeast extract, 5 g L1 NaCl, 1 g L1 glucose). Pre-pre-cultures were transferred to
1389-1723/$ e see front matter Ó 2016, The Society for Biotechnology, Japan. All rights reserved. http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
WADA ET AL.
J. BIOSCI. BIOENG.,
Construction of the plasmid and strain The E. coli strains used in this study are derived from MG1655(DE3) (15). The plasmid pCOLADuet-1 (Merck KGaA, Darmstadt, Germany) was used as the vector. Acetoacetyl-CoA synthase/3hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase gene (mvaE) and HMG-CoA synthase mvaS genes derived from E. faecalis were synthesized by GeneArt (Thermo Fisher Scientific, MA, USA) according to E. coli codon usage. These genes were trimmed with NdeI and XhoI, and then subcloned into pCOLADuet-1 to generate pCOLADuet-1/mvaES (mvaE and mvaS expression plasmid, T7 promoter, KmR). E. coli MG1655(DE3) was transformed with pCOLADuet-1 (control) and pCOLADuet-1/mvaES (engineered), respectively. Analyses of dry cell weight and extracellular metabolites Cell growth was monitored by measuring OD600 using the UVmini-1240 (Shimadzu, Kyoto, Japan). Dry cell weight was calculated using a conversion coefficient of 0.3 gDCW L-1 OD-1 600 . The concentrations of glucose and organic acids (lactate, formate, acetate, succinate and MVA) in the supernatant of the culture broth were determined by an HPLC system (Shimadzu) equipped with an Aminex HPX-87X column (BioRad, Hercules, CA, USA), UV/vis detector (SPD-20A), and a refractive index detector (RID-10A). The column temperature was set to 65 C, and 1.5 mM H2SO4 was used as the mobile phase with a flow rate of 0.5 ml min1. The supernatant of the culture broth was obtained by centrifugation at 18,000 g for 5 min at 4 C, and then filtered through a Millex HV 0.45-mm filter (Merck KGaA). Flux prediction by simulation analysis Constraint-based Flux Balance Analysis (FBA) was performed by using the core metabolic model of E. coli K-12, named as ecoli_core_model obtained from BiGG Models (16). All calculations were performed using a solver for linear programming, the GNU Linear Programming Kit in Matlab and the COBRA Toolbox (17). The MVA synthesis pathway was added to this model. The glucose uptake rate was set to the experimental value of 7.74 and 8.21 mmol/gDCW/h for control and engineered strains, respectively. In both strains, the non-growth-associated ATP maintenance requirement rate was set to the default value (8.39 mmol/gDCW/h), which was estimated using experimental data from continuous cultures with glucose as a carbon source (18). To illustrate the relationship between the biomass yield and the MVA yield, the maximum growth rate was calculated using an objective function as a maximization of biomass production. The maximum and minimum fluxes of MVA production were calculated using an objective function as the maximization or minimization of MVA production with a constraint of each fixed growth rate from zero to the maximum value. The MVA production rate and the growth rate were converted to carbon yields using coefficient 6 C-mol/mol-MVA and 63.1 C-mol/gDCW, respectively. The flux distribution with theoretical MVA production yield was calculated by FBA using an objective function as a maximization of MVA production. To minimize the Euclidian norm of internal fluxes, the optional parameter minNorm of optimizeCbModel function was set to ‘one’. The glucose uptake rate was set to the experimentally measured value (8.21 mmol/gDCW/h) of the engineered strain.
Control 10
OD600
20 1
15 10
0.1
5 0.01
0
2 OD600
4
6 8 10 Time (h) Glucose
12
0
MDV5e6h ¼ where MDVxh is
MDV6h OD6h MDV5h OD5h OD6h OD5h
(1)
13
C-entrichment at x h, and ODxh is the cell concentration at x h.
13
13 C-metabolic flux analysis C-MFA was performed using the OpenMebius software (20), which is based on the elementary metabolite units framework (21) in Matlab (MathWorks Inc., Natick, MA, USA). The building block and cofactor requirements for E. coli biomass formation was referred to in a previous study (22). The metabolic pathway model of E. coli that was used in this study has been described previously (19). The following MVA synthesis pathway was added. The amino acid fragments were chosen according to the previous study (23). Metabolic fluxes were estimated by minimizing the residual sum of squares between the experimentally measured and model predicted 13C-enrichment using the fmincon optimization solver in the Matlab toolbox. The standard deviation (SD) of 13C-enrichment was set to 0.01. The 95% confidence intervals were calculated using a grid search method as described in previous studies (24).
Malic enzyme activity measurement The crude cell extract was prepared from M9 batch cultures that were harvested at 2 h after the addition of IPTG. Cells were washed twice with 0.9% NaCl, and disrupted by sonication. The activity of two types of malic enzymes (SfcA and MaeB) was measured as described in a previous study (25). Enzyme activities were measured spectrophotometrically at 30 C by monitoring NADPH production at 340 nm with the UVmini-1240 system (Shimadzu). The reaction was started by addition of 10 mL of crude cell extract. The protein concentration in the supernatant was determined by using the Bradford method.
RESULTS AND DISCUSSION Growth characteristics of the MVA-producing strain The control and MVA-producing (engineered) strains were constructed by transforming E. coli MG1655(DE3) with pCOLADuet-1 and pCOLADuet-1/mvaES harboring acetoacetyl-CoA synthase/HMGCoA reductase gene (mvaE) and HMG-CoA synthase gene (mvaS) from E. faecalis, respectively. The codon usage of mvaES genes and its expression vector were modified from the previous report (9). The control and engineered strains were grown in batch cultures on glucose under aerobic conditions. IPTG was added at 4 h after inoculation for inducing the expression of mvaES. Fig. 1 shows the cell growth and metabolite production of the two strains. Growth parameters are summarized in Table 1. These specific rates were calculated at the exponential phase between 4 and 6.5 h. Both
B
25 Concentration (mM)
A
13 Analysis of 13C-enrichment of proteinogenic amino acids C-enrichment of proteinogenic amino acids was measured by GCeMS. Broth culture (7.5 ml) was taken from the flask, and centrifuged at 7000 g, for 5 min at 4 C. The cell pellet was then washed twice with 0.9% NaCl, and hydrolyzed in 6 mol/L HCl at 105 C for 18 h. The resulting proteinogenic acids were derivatized with N-(tert-butyldimethylsilyl)N-methyl-trifluoroacetamide containing tert-butyldimethylchlorosilane in acetonitrile at 105 C for 1 h, and then analyzed by a GCeMS [Agilent 7890A GC and 5975C Mass Selective Detector (Agilent Technologies, Santa Clara, USA)] equipped with a DB5MSþDG column (Agilent Technologies). The analytical conditions used are described elsewhere (19). The data obtained from GCeMS were corrected by reduction of the natural abundance ratio of C, H, O, N, and Si isotopes. Cumulative 13C-enrichment from 5 to 6 h after cultivation was calculated using the following equation:
Engineered 10
25 20
1
15 10
0.1
5 0.01
Succinate
0
2
4
6 8 Time (h) Acetate
10
12
Concentration (mM)
50 ml of M9 medium (17 g L1 Na2HPO4$12H2O, 3.1 g L1 KH2PO4, 1 g L1 NH4Cl, 0.5 g L1 NaCl, 0.25 g L1 MgSO4$7H2O, 15 mg L1 CaCl2$2H2O, 8.1 mg L1 FeCl3, 1 mg L1 MnSO4$4H2O, 1.7 mg L1 ZnCl2, 0.43 mg L1 CuCl2$2H2O, 0.6 mg L1 CoCl2$6H2O, 0.6 mg L1 Na2MoO4$2H2O, and 4 g L1 glucose) to form pre-cultures with an initial optical density of 0.05 at 600 nm (OD600). Pre-culture was performed for reducing the risk of interfusion of rich culture medium. After 24 h, pre-cultures were inoculated in 50 ml of M9 medium containing [1-13C] glucose as the main culture with an initial OD600 of 0.05. [1-13C] glucose was purchased from Cambridge Isotope Laboratories (Andover, MA, USA). Kanamycin (0.03 g L1) and isopropylthiogalactoside (IPTG, 24 mg L1) was added whenever necessary. Dry cell mass was determined by filtration, washing of cells, and drying at 70 C until a constant biomass of an equivalent of 25 mL culture volume at an OD600 of 1.0 was obtained.
OD600
2
0
MVA
FIG. 1. Growth characteristics of control (A) and engineered (B) strains in batch cultures grown on glucose under aerobic conditions. Closed squares, biomass; circles, glucose; triangles, acetate; open squares, succinate; diamonds, MVA. Data shown are mean SD (n ¼ 3).
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
VOL. xx, 2016
ANALYSIS OF MVA-PRODUCING E. COLI
3
TABLE 1. Growth characteristics. Strain
Growth rate (1/h)
Specific rate (mmol/gDCW/h) Acetate production
Succinate production
Mevalonate production
Mevalonate
7.74 0.17 8.21 0.22
4.41 0.08 0.42 0.24
0.20 0.00 0.08 0.04
N.D. 1.84 0.07
N.D. 0.22 0.00
0.43 0.01 0.48 0.01
Control Engineered
Yield (C-mol/C-mol)
Glucose uptake
The growth characteristics of the control and engineered strains during 4e6.5 h. The data are presented as means SD (n ¼ 3). N.D. represents not detected.
strains converted glucose to biomass, acetate, succinate, and MVA. No other by-products were detected by HPLC. The growth rate and glucose uptake rate in the engineered strain were slightly faster than in the control strain. Acetate production rate in the engineered strain was reduced to 10% of the control strain. While no MVA production was observed for the control strain, the engineered strain produced MVA with the productivity at 1.84 mmol/gDCW/h; corresponding to the yield at 22% MVA Cmol/Glucose C-mol.
Evaluation of MVA production based on mass balance analysis One molecule of MVA is synthesized from three molecules of AcCoA (carbon backbone) using two molecules of NADPH as a cofactor. Since AcCoA and NADPH are also required for cell growth, the capacity of the MVA production yield should be restricted by the biomass yield. The yields of MVA and biomass of the strains were evaluated in terms of mass balances. To reveal the relationship between biomass yield and MVA yield, the maximum and minimum rates of MVA production at each growth rate were calculated using an E. coli core metabolic pathway model. The growth rate and the MVA production rate were normalized to the yield of carbon atoms. The gray area in Fig. 2 indicates the feasible MVA yield at each biomass yield; thus, the measured data must be located in this area. The open square symbol represents a metabolic state in which the cells can most efficiently use glucose for cell growth with no MVA production. In contrast, the open triangle symbol represents a state in which the cells can most efficiently convert glucose to MVA without growth. The measured carbon yields of biomass and MVA were 22 (Cmol/C-mol) and 62 (C-mol/C-mol) in the engineered strain, respectively, and were 0 (C-mol/C-mol) and 58 (C-mol/C-mol) in the control strain, respectively. Open and closed circles in Fig. 2 represent the MVA production of the engineered and control strains, respectively. It is confirmed that both data are located
MVA yield (MVA C-mol/Substrate C-mol)
100
Control Engineered
80
Theoretical maximum yield of biomass Theoretical maximum yield of MVA
60 40
(62, 22) 20 (58, 0) 0 0
20 40 60 80 100 Biomass yield (Biomass C-mol/Substrate C-mol)
FIG. 2. Feasible range of the MVA yield against the biomass yield. The gray area indicates the feasible MVA production yield. Open and closed circles indicate the measured metabolic states of engineered and control strains. Open square and open triangle represent metabolic states in which the cells can most efficiently use glucose for cell growth or MVA production.
within the feasible area in Fig. 2. The MVA yield of the engineered strain was close to the upper limit at the biomass yield. Comparison of 13C-flux distributions between control and engineered strains The metabolic flux distributions of control and engineered strains were estimated by 13C-MFA. The cells were cultured using [1-13C] glucose as sole carbon source. The 13C-enrichments of proteinogenic amino acids were measured by GCeMS. In this study, the proteinogenic amino acids produced at the exponential phase between 5 and 6 h were used for the flux estimations. The malic enzyme reaction was not considered in the 13CMFA due to the low specific activity (<0.01 mmol/gDCW/h) by the enzyme assay. The measured and estimated 13C-enrichments are shown in Table S1. The threshold of chi-squared test for the goodness of fit was 88 (the number of degrees of freedom of the reaction model was 20 and the number of independent data was 88). Both control and engineered strains passed the chi-squared test as the values in these strains were 42 and 20, respectively. This suggests that the estimated flux distributions can statistically explain the experimentally measured 13C-enrichment of amino acids. The absolute flux values (mmol/gDCW/h) of the best-fit estimation in each reaction are shown on the pathway map in Fig. 3. The best-fit values of metabolic fluxes were estimated by minimizing the residual sum of squares between the experimentally measured and model-predicted 13C-enrichment. For example, the glucose (7.7 mmol/gDCW/h) was metabolized through glycolysis and the pentose phosphate (PP) pathway with fluxes of 5.0 and 2.7 mmol/gDCW/h in the control strain, respectively. The flux distribution of the control strain was similar to the previously reported results of wild type E. coli (26). Whereas no significant differences were observed in the fluxes of glycolysis, PP pathway, and the ED pathway between control and engineered strains, there were differences in the fluxes in the pathways downstream of AcCoA. Although the acetate formation flux was 4.4 mmol/gDCW/h in the control strain, the flux decreased to 0.4 mmol/gDCW/h and the MVA formation flux increased to 1.8 mmol/gDCW/h in the engineered strain. The TCA cycle flux of the engineered strain was lower than that of the control strain. It seems that AcCoA was diverted to the MVA synthesis pathway in the engineered strain due to overexpression of enzymes in the pathway. NADPH generation in the entire central carbon metabolism does affect MVA production since MVA synthesis requires NADPH as a cofactor. NADPH is produced by glucose-6-phosphate dehydrogenase, phosphogluconate dehydrogenase, isocitrate dehydrogenase, and NADPþ-dependent malic enzyme, and is consumed for cell growth and MVA synthesis. Although the oxidative PP pathway is considered as the NADPH-generating pathway in E. coli, the flux was not significantly different between the control and engineered strains. The isocitrate dehydrogenase flux of the engineered strain was lower than that of the control strain since the AcCoA was diverted from the TCA cycle to MVA synthesis. No enzymatic activity of the NADPþ-dependent malic enzyme was confirmed in both strains. Because the imbalance between NADPH production and consumption rates is maintained by transhydrogenase reactions to achieve homeostasis in cells (27), the flux of the transhydrogenase reaction was calculated based on the mass balance. Fig. 4 shows the NADPH balance in the control and engineered
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
4
WADA ET AL.
J. BIOSCI. BIOENG.,
FIG. 3. (A) Metabolic flux distributions in the central metabolic pathways of the control and engineered strains during exponential growth phase. (B) Simulated metabolic flux distribution with maximal theoretical MVA yield. The numbers on the arrows indicate absolute flux values (mmol/gDCW/h). Asterisks indicate biomass formation.
strains calculated from flux distributions. The total NADPH production/consumption rate in the engineered strain was 1.5 times higher than that in the control strain. In other words, the turnover rate of NADPH in the engineered strain was higher than that of the control strain. This difference was related to the high requirement of NADPH for MVA synthesis in the engineered strain. The balance of production/consumption NADPH rate suggested that the additional requirement of NADPH for MVA synthesis was obtained from the transhydrogenase reaction in the engineered strain. He et al. reported the flux responses to the overproduction of fatty acids in E. coli (26). The fatty acids are synthesized from AcCoA using NADPH as a cofactor, as with MVA. The observed decrease in acetate production and the increase in transhydrogenase flux
converting NADH to NADPH in the engineered strain in the present study were in accordance with the flux responses in the fatty acidproducing strain. On the other hand, the growth reduction, the increasing of the PP pathway flux, and the decreasing of the anaplerotic pathway flux in the fatty acid-producing strain were not observed in the MVA-producing strain. It has been reported that the overproduction of fatty acids in E. coli requires significantly higher maintenance energy than in the control strain because the overproduction of free fatty acids induces cellular membrane stress (28). In the case of MVA production, the growth reduction did not occur due to the non-toxicity of MVA. Since the role of anaplerotic reactions in the TCA cycle is to supply the precursors of biomass components, the small difference in the flux between the control
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
VOL. xx, 2016
15
ANALYSIS OF MVA-PRODUCING E. COLI
Control
Engineered
mmol/gDCW/h
12 9
5
suggested that the additional requirements of NADPH for MVA synthesis were obtained from the transhydrogenase reaction. Furthermore, comparison of the experimentally obtained and ideal fluxes could be useful for planning a strategy for metabolic modification for future engineering towards increasing the MVA production. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.jbiosc.2016.08.001.
6 ACKNOWLEDGMENTS
3 This work was partially supported by the Grant in Aid for Scientific Research (A) No. 24246134 and by the Grant in Aid for Scientific Research (B) No. 16H04576.
0
References
Citrate synthase
Cell synthesis
G6P dehydrogenase
MVA synthesis
6PG dehydrogenase
Transhydrogenase
FIG. 4. Estimated NADPH balance in control and engineered strains using the best-fit metabolic flux values. Transhydrogenases function as equilibrators of NADPH balance. Error-bars indicate 95% confidence intervals of each flux.
and engineered strains is consistent. Furthermore, a fatty acid (C16:0) requires 8 AcCoA and 14 NADPH molecules, whereas an MVA requires 3 AcCoA and 2 NADPH (26). MVA production requires much less NADPH than fatty acid production; thus, the MVAproducing strain can cover the NADPH needs for MVA synthesis by increasing the transhydrogenase flux without increasing PP pathway flux. Prediction of engineering for further MVA production The potential for improvement of MVA production by further engineering was investigated by comparing current and ideal flux distributions. The ideal flux distribution with theoretical maximum MVA production yield was calculated by flux balance analysis using an objective function as a maximization of MVA production (Fig. 3B). The glucose uptake rate was set to the experimentally measured value (8.2 mmol/gDCW/h) of the engineered strain. This metabolic state corresponds to the position of the open triangles in Fig. 2. The uptake glucose is catabolized only through glycolysis and is used for MVA synthesis without competitive pathways such as acetate formation and the TCA cycle in this ideal metabolic state. Repression of the citrate synthase (CS) or the PP pathway would be required to improve MVA production because excess fluxes were observed in these pathways by 13CMFA. However, deletion of CS is undesirable for bioproduction due to gene essentiality. A switching system of CS expression might be effective. Soma et al. (29) already developed a switch for CS expression during the exponential growth phase to efficiently produce isopropanol in E. coli, and succeeded in increasing productivity. This system may be useful for effective MVA production by achieving precise control of the CS flux. The fed-batch process development is also suitable to achieve high titer production of MVA using this strain. The cells enter the stationary phase by CS flux repression. It is expected that a high yield and titer production of MVA can be achieved by maintaining this state for a long period, since the carbons are not consumed for the biomass synthesis. In the present study, we evaluated the metabolic state of an MVA-producing E. coli strain based on 13C-MFA. The metabolic flux
1. Martin, V. J., Pitera, D. J., Withers, S. T., Newman, J. D., and Keasling, J. D.: Engineering a mevalonate pathway in Escherichia coli for production of terpenoids, Nat. Biotechnol., 21, 796e802 (2003). 2. Haratake, A., Ikenaga, K., Katoh, N., Uchiwa, H., Hirano, S., and Yasuno, H.: Topical mevalonic acid stimulates de novo cholesterol synthesis and epidermal permeability barrier homeostasis in aged mice, J. Invest. Dermatol., 114, 247e252 (2000). 3. Wang, C., Yoon, S. H., Shah, A. A., Chung, Y. R., Kim, J. Y., Choi, E. S., Keasling, J. D., and Kim, S. W.: Farnesol production from Escherichia coli by harnessing the exogenous mevalonate pathway, Biotechnol. Bioeng., 107, 421e429 (2010). 4. Yoon, S. H., Lee, S. H., Das, A., Ryu, H. K., Jang, H. J., Kim, J. Y., Oh, D. K., Keasling, J. D., and Kim, S. W.: Combinatorial expression of bacterial whole mevalonate pathway for the production of beta-carotene in E. coli, J. Biotechnol., 140, 218e226 (2009). 5. Miziorko, H. M.: Enzymes of the mevalonate pathway of isoprenoid biosynthesis, Arch. Biochem. Biophys., 505, 131e143 (2011). 6. Tamura, G., Ando, K., Kodama, K., and Arima, K.: Production of mevalonic acid by fermentation, Appl. Microbiol., 16, 965e972 (1968). 7. Kuzuyama, T., Hemmi, H., and Takahashi, S.: Mevalonate pathway in bacteria and archaea, pp. 493e516, in: Mander, L. and Liu, H.-W. (Eds.), Comprehensive natural products II, vol. 1. Elsevier, Oxford (2010). 8. Kuzuyama, T., Dairi, T., Yamashita, H., Shoji, Y., and Seto, H.: Heterologous mevalonate production in Streptomyces lividans TK23, Biosci. Biotechnol. Biochem., 68, 931e934 (2004). 9. Tabata, K. and Hashimoto, S.: Production of mevalonate by a metabolicallyengineered Escherichia coli, Biotechnol. Lett., 26, 1487e1491 (2004). 10. Kuzuyama, T.: Mevalonate and nonmevalonate pathways for the biosynthesis of isoprene units, Biosci. Biotechnol. Biochem., 66, 1619e1627 (2002). 11. Zheng, Y., Liu, Q., Li, L., Qin, W., Yang, J., Zhang, H., Jiang, X., Cheng, T., Liu, W., Xu, X., and Xian, M.: Metabolic engineering of Escherichia coli for highspecificity production of isoprenol and prenol as next generation of biofuels, Biotechnol. Biofuels, 6, 57 (2013). 12. Hedl, M., Sutherlin, A., Wilding, E. I., Mazzulla, M., McDevitt, D., Lane, P., Burgner, J. W., 2nd, Lehnbeuter, K. R., Stauffacher, C. V., Gwynn, M. N., and Rodwell, V. W.: Enterococcus faecalis acetoacetyl-coenzyme A thiolase/3hydroxy-3-methylglutaryl-coenzyme A reductase, a dual-function protein of isopentenyl diphosphate biosynthesis, J. Bacteriol., 184, 2116e2122 (2002). 13. Wiechert, W.: 13C metabolic flux analysis, Metab. Eng., 3, 195e206 (2001). 14. Wittmann, C.: Fluxome analysis using GC-MS, Microb. Cell Fact., 6, 6 (2007). 15. Tokuyama, K., Ohno, S., Yoshikawa, K., Hirasawa, T., Tanaka, S., Furusawa, C., and Shimizu, H.: Increased 3-hydroxypropionic acid production from glycerol, by modification of central metabolism in Escherichia coli, Microb. Cell Fact., 13, 64 (2014). 16. King, Z. A., Lu, J., Dräger, A., Miller, P., Federowicz, S., Lerman, J. A., Ebrahim, A., Palsson, B.Ø., and Lewis, N. E.: BiGG models: a platform for integrating, standardizing and sharing genome-scale models, Nucleic Acids Res., 44, D515eD522 (2015). 17. Becker, S. A., Feist, A. M., Mo, M. L., Hannum, G., Palsson, B.Ø., and Herrgard, M. J.: Quantitative prediction of cellular metabolism with constraintbased models: the COBRA Toolbox, Nat. Protoc., 2, 727e738 (2007). 18. Feist, A. M., Henry, C. S., Reed, J. L., Krummenacker, M., Joyce, A. R., Karp, P. D., Broadbelt, L. J., Hatzimanikatis, V., and Palsson, B.Ø.: A genomescale metabolic reconstruction for Escherichia coli K-12 MG1655 that accounts for 1260 ORFs and thermodynamic information, Mol. Syst. Biol., 3, 121 (2007). 19. Okahashi, N., Kajihata, S., Furusawa, C., and Shimizu, H.: Reliable metabolic flux estimation in Escherichia coli central carbon metabolism using intracellular free amino acids, Metabolites, 4, 408e420 (2014).
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001
6
WADA ET AL.
20. Kajihata, S., Furusawa, C., Matsuda, F., and Shimizu, H.: OpenMebius: an open source software for isotopically nonstationary 13C-based metabolic flux analysis, Biomed. Res. Int., 2014, 627014 (2014). 21. Antoniewicz, M. R., Kelleher, J. K., and Stephanopoulos, G.: Elementary metabolite units (EMU): a novel framework for modeling isotopic distributions, Metab. Eng., 9, 68e86 (2007). 22. Ingraham, J. L., Maaløe, O., and Neidhardt, F. C.: Growth of the bacterial cell pp. 128. Sinauer Associates Inc., Sunderland, MA, USA (1983). 23. Antoniewicz, M. R., Kelleher, J. K., and Stephanopoulos, G.: Accurate assessment of amino acid mass isotopomer distributions for metabolic flux analysis, Anal. Chem., 79, 7554e7559 (2007). 24. Antoniewicz, M. R., Kelleher, J. K., and Stephanopoulos, G.: Determination of confidence intervals of metabolic fluxes estimated from stable isotope measurements, Metab. Eng., 8, 324e337 (2006). 25. Bologna, F. P., Andreo, C. S., and Drincovich, M. F.: Escherichia coli malic enzymes: two isoforms with substantial differences in kinetic properties, metabolic regulation, and structure, J. Bacteriol., 189, 5937e5946 (2007).
J. BIOSCI. BIOENG., 26. He, L., Xiao, Y., Gebreselassie, N., Zhang, F., Antoniewiez, M. R., Tang, Y. J., and Peng, L.: Central metabolic responses to the overproduction of fatty acids in Escherichia coli based on 13C-metabolic flux analysis, Biotechnol. Bioeng., 111, 575e585 (2014). 27. Sauer, U., Canonaco, F., Heri, S., Perrenoud, A., and Fischer, E.: The soluble and membrane-bound transhydrogenases UdhA and PntAB have divergent functions in NADPH metabolism of Escherichia coli, J. Biol. Chem., 279, 6613e6619 (2004). 28. Youngquist, J. T., Lennen, R. M., Ranatunga, D. R., Bothfeld, W. H., Marner, W. D., 2nd, and Pfleger, B. F.: Kinetic modeling of free fatty acid production in Escherichia coli based on continuous cultivation of a plasmid free strain, Biotechnol. Bioeng., 109, 1518e1527 (2012). 29. Soma, Y., Tsuruno, K., Wada, M., Yokota, A., and Hanai, T.: Metabolic flux redirection from a central metabolic pathway toward a synthetic pathway using a metabolic toggle switch, Metab. Eng., 23, 175e184 (2014).
Please cite this article in press as: Wada, K., et al., 13C-metabolic flux analysis for mevalonate-producing strain of Escherichia coli, J. Biosci. Bioeng., (2016), http://dx.doi.org/10.1016/j.jbiosc.2016.08.001