Magnesium starvation improves production of malonyl-CoA …homepages.rpi.edu/~koffam/papers2/2019_Tokuyama.pdf · 2019-08-28 · the pETM6 plasmid. Strain E. coli MG1655(DE3)/pETM6

Contents lists available at ScienceDirect

Metabolic Engineering

journal homepage: www.elsevier.com/locate/meteng

Magnesium starvation improves production of malonyl-CoA-derivedmetabolites in Escherichia coli

Kento Tokuyamaa, Yoshihiro Toyaa, Fumio Matsudaa, Brady F. Cressb, Mattheos A.G. Koffasb,c,Hiroshi Shimizua,⁎

a Department of Bioinformatic Engineering, Graduate School of Information Science and Technology, Osaka University, 1-5 Yamadaoka, Suita, Osaka 565-0871, JapanbDepartment of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, NY, USAc Department of Biological Sciences, Rensselaer Polytechnic Institute, Troy, NY, USA

A R T I C L E I N F O

Keywords:Nutrient starvationMetabolic profiling analysisMalonyl-CoA3-hydroxypropionic acidFlavonoids

A B S T R A C T

Starvation of essential nutrients, such as nitrogen, sulfur, magnesium, and phosphorus, leads cells into stationaryphase and potentially enhances target metabolite production because cells do not consume carbon for thebiomass synthesis. The overall metabolic behavior changes depend on the type of nutrient starvation inEscherichia coli. In the present study, we determined the optimum nutrient starvation type for producing mal-onyl-CoA-derived metabolites such as 3-hydroxypropionic acid (3HP) and naringenin in E. coli. For 3HP pro-duction, high production titer (2.3 or 2.0 mM) and high specific production rate (0.14 or 0.28mmol gCDW−1

h−1) was observed under sulfur or magnesium starvation, whereas almost no 3HP production was detectedunder nitrogen or phosphorus starvation. Metabolic profiling analysis revealed that the intracellular malonyl-CoA concentration was significantly increased under the 3HP producing conditions. This accumulation shouldcontribute to the 3HP production because malonyl-CoA is a precursor of 3HP. Strong positive correlation(r=0.95) between intracellular concentrations of ATP and malonyl-CoA indicates that the ATP level is im-portant for malonyl-CoA synthesis due to the ATP requirement by acetyl-CoA carboxylase. For naringeninproduction, magnesium starvation led to the highest production titer (144 ± 15 μM) and specific productivity(127 ± 21 μmol gCDW−1). These results demonstrated that magnesium starvation is a useful approach toimprove the metabolic state of strains engineered for the production of malonyl-CoA derivatives.

1. Introduction

Extensive efforts in metabolic engineering and synthetic biologyhave led to the development of microbial processes for heterologousmetabolites production. In recent years, the microbial process for theproduction of malonyl-CoA derivatives has been studied extensively(Xu et al., 2014). A target metabolite, 3-hydroxypropionic acid (3HP)can be produced from malonyl-CoA and has a wide variety of com-mercial applications (Werpy and Petersen, 2004; Tokuyama et al.,2014; Rathnasingh et al., 2012; Liu et al., 2013; Cheng et al., 2016;Jiang et al., 2009; Kumar et al., 2013; Kim et al., 2014; Borodina et al.,2015; Lan et al., 2015; Liu et al., 2015).

Plant-derived flavonoids have been widely used in functional foods,nutraceuticals, and cosmetics in global market (Panche et al., 2016),and some them are potential candidates as lead compounds for onsetdiseases, such as Alzheimer's disease and cancers (Baptista et al., 2014;Ibrahim et al., 2014). The microbial productions have attracted

increased attention due to lack of availability of plant sources for fla-vonoids production, a lot of metabolic engineering studies have beenreported to produce flavonoids in microorganisms (Xiu et al., 2017;Jones et al., 2016a; Xu et al., 2011; Leonard et al., 2007; Raman et al.,2014; Pandey et al., 2016). Naringenin is a potential immunomodulator(Zeng et al., 2018), and has neuro-protective effect on permanent is-chemic damage (Bai et al., 2014). The compound is synthesized from p-coumaric acid as a precursor using malonyl-CoA. In previous reports,malonyl-CoA availability was considered as a major bottleneck, becausemalonyl-CoA is used as a precursor of biomass components and itsconcentration is tightly controlled in the cells. Some previous studieshave focused on the optimization of de novo biosynthesis of malonyl-CoA using genetic engineering strategies (Zha et al., 2009; Xu et al.,2011; Johnson et al., 2017).

Bio-production during stationary phase is also an effective approachto enhance availability of intermediates related to biomass components,by decreasing the demand for cell growth (Toya et al., 2015). Essential

https://doi.org/10.1016/j.ymben.2018.12.002Received 13 August 2018; Received in revised form 4 December 2018; Accepted 5 December 2018

⁎ Corresponding author.E-mail address: shimizu@ist.osaka-u.ac.jp (H. Shimizu).

Metabolic Engineering 52 (2019) 215–223

Available online 06 December 20181096-7176/ © 2018 International Metabolic Engineering Society. Published by Elsevier Inc. All rights reserved.

T

http://www.sciencedirect.com/science/journal/10967176https://www.elsevier.com/locate/metenghttps://doi.org/10.1016/j.ymben.2018.12.002https://doi.org/10.1016/j.ymben.2018.12.002mailto:shimizu@ist.osaka-u.ac.jphttps://doi.org/10.1016/j.ymben.2018.12.002http://crossmark.crossref.org/dialog/?doi=10.1016/j.ymben.2018.12.002&domain=pdf

nutrient starvation is a cost-effective way to lead the cells into sta-tionary phase, and shift overall metabolic activity through regulatorymachinery, for adapting to the environmental perturbation (Chubukovand Sauer, 2014). There are various essential nutrients, such as ni-trogen, sulfur, magnesium, and phosphorus, for Escherichia coli. It hasbeen reported that the metabolic behavior changes depending on thetype of nutrient starvation in E. coli (Chubukov and Sauer, 2014). Liet al. (2016) successfully demonstrated that sulfur starvation was sui-table to produce acetyl-CoA derived mevalonate and tyrosine. Masudaet al. (2017) applied 13C-metabolic flux analysis on the mevalonateproducing E. coli, and showed that the sulfur starvation improvedacetyl-CoA availability compared to the growing cells. Since acetyl-CoAcarboxylase is the sole pathway of de novo synthesis of malonyl-CoA,nutrient starvation potentially improves malonyl-CoA availability cou-pled with the increased acetyl-CoA supply.

In the present study, we determined the optimum nutrient starva-tion condition for producing malonyl-CoA derivatives, such as 3HP andthe plant-derived flavonoid naringenin, in E. coli. The stationary phasecultures of 3HP-producing E. coli strain on M9 medium lacking ni-trogen, sulfur, magnesium, or phosphorus, showed significant differ-ences in 3HP productivities depending on the type of nutrient starva-tion. The cause of the differences in 3HP productivity following nutrientstarvation was investigated by metabolic profiling analysis of the cen-tral carbon metabolism.

2. Materials and methods

2.1. Strains and plasmids

All strains and plasmids used in this study are listed in Table 1. E.coli strain DH5α was used for plasmid construction and E. coli strainMG1655(DE3) was used as production host for 3HP. Malonyl-CoA re-ductase (MCR) gene was separated into N-terminal region and C-terminal region, to enhance its enzymatic activity (Liu et al., 2013), andintroduced into the host strain. The N-terminal region of MCR (mcrN,amino acids 1–549) and the C-terminal region (mcrC, amino acids550–1219) were derived from Chloroflexus aurantiacus. These gene re-gions were codon-optimized for expression in E. coli and synthesizedwith NdeI and XhoI sites in pMK-RQ by GeneArtR (Life technologies).T7 expression vector pETM6 was used to construct MCR expressionplasmid (Xu et al., 2012). McrN and mcrC were assembled in pETM6vector as a pseudo-operon following the protocol described by Xu et al.(2012). NdeI-mcrN-XhoI and NdeI-mcrC-XhoI fragments from pMK-RQwere cloned into the same restriction sites in pETM6 (pETM6-mcrN andpETM6-mcrC). The AvrII-mcrC-SalI fragment from pETM6-mcrC wasligated with SpeI/SalI-digested pETM6-mcrN (pETM6-mcrNC). Nar-ingenin producing E. coli (FlavoOpt) was previously constructed usingBL21star(DE3) that was transformed with a plasmid vector carrying the

naringenin biosynthetic genes (Jones et al., 2016b).

2.2. Batch culture protocol for 3HP production

Lennox medium (10 g L−1 tryptone, 5 g L−1 yeast extract, 5 g L−1

NaCl, 1 g L−1 glucose), standard M9 medium, and specific nutrient-freeM9 medium were used for culture experiments. Complete mediumcompositions are provided in Table S1 in the Supplemental materials.All cultures were supplemented with 0.08 g L−1 ampicillin to maintainthe pETM6 plasmid. Strain E. coli MG1655(DE3)/pETM6 mcrNC wasgrown aerobically at 37 °C overnight in a test tube containing 3mL ofLennox medium. The culture was diluted into 500mL baffled Erlen-meyer flasks containing 200mL of standard M9 medium with an initialoptical density at 600 nm (OD600) of 0.05 and cells were grown aero-bically at 37 °C in a shaking incubator at 150 rpm (BR-43FL, Taitec,Japan). For expression of MCR, 1.0mM isopropyl β-D-1-thiogalacto-pyranoside (IPTG), 0.04 g L−1 biotin and 20mM NaHCO3 were addedat mid log phase. After 20 h of cultivation, cells were collected bycentrifugation at 2900×g for 10min at 25 °C and inoculated into200mL baffled Erlenmeyer flasks containing 50mL of nitrogen-free,phosphorus-free, sulfur-free or magnesium-free M9 medium with aninitial OD600 of 2.0. Cells were cultured at 37 °C in a shaking incubatorat 150 rpm (BR-43FL, TAITEC, Saitama, Japan).

2.3. Batch culture protocol for naringenin production

Standard M9 medium and specific nutrient-limited M9 mediumwere used for culture experiments. Complete medium compositions areprovided in Table S2 in the Supplemental materials. All cultures weresupplemented with 0.08 g L−1 ampicillin to maintain the pFlavoopt

plasmid. The strain E. coli BL21star(DE3)/pFlavoopt was grown over-night at 37 °C in a 50-mL falcon tube containing 5mL of standard M9medium. After 16 h, the overnight cultures were serially diluted in 0.9%NaCl and then 50 μL of the diluted cells were inoculated in poly-propylene 48-well plates (5 mL, VWR) containing 5mL of standard M9medium or nutrient-limited M9 medium. The culture was inoculated atdilution ratios of 1:20, 1:30, 1:40, 1:60, 1:80, 1:120, 1:160, and 1:240from the overnight culture. The cells were allowed to grow aerobicallyat 37 °C for 6 h before induction with 1mM IPTG and 10mg/L biotin.As a substrate, 0.61mM p-coumaric acid was added at induction, and30 °C was used as the induction temperature.

2.4. Analytical methods

Cell optical density was measured at 600 nm by using UV-mini 1240(Shimadzu, Kyoto, Japan) for 3HP-producing E. coli, and using BioTekSynergy 4 microplate reader (BioTek, USA) for naringenin-producing E.coli. For determining cell dry weight (CDW) of 3HP-producing strain

Table 1E. coli strains and plasmids used in this study.

Strains or plasmids Description Source

StrainsDH5α F－, Φ80d lacZΔM15, Δ(lacZYA-argF)U169, deoR, recA1, endA1, hsdR17(rK－ mK+), phoA, supE44, λ－, thi－1, gyrA96,

relA1MG1655(DE3) F−, λ−, rph-1, λ (DE3[lacI lacUV5-T7 gene l indl sam7 nin5]) (Tokuyama et al., 2014)MG1655(DE3)/pETM6 mcrNC MG1655(DE3) transformed with pETM6 mcrNC This studyBL21star(DE3)/pFlavoOpt BL21star(DE3) transformed with pFlavoOpt (Jones et al., 2016b)PlasmidspETM6 lacI; expression vector; T7 promoter; T7 terminetor; ColE1-ori; Ampr (Xu et al., 2012)pMK-RQ mcrN Synthesized NdeI-mcrN-XhoI; ColE1-ori; Kanr This studypMK-RQ mcrC Synthesized NdeI-mcrC-XhoI; ColE1-ori; Kanr This studypETM6 mcrN pETM6 carrying artificially synthesized mcrN; Ampr This studypETM6 mcrC pETM6 carrying artificially synthesized mcrC; Ampr This studypETM6 mcrNC pETM6 carrying artificially synthesized mcrNC; Ampr This studypFlavoopt pETM6 carrying naringenin biosynthetic genes (At4CL, PhCHS,CmCHI); Ampr (Jones et al., 2016b)

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

216

under each starvation condition, a consistent amount of cells [OD600× culture volume (mL) =100] was harvested by 0.5-μm pore size filter(PTFE-type membrane, Advantec, Tokyo, Japan) at 20 h for nitrogenstarvation, 8 h for phosphorus or sulfur starvation, and 4 h for magne-sium starvation. These samples were dried at 60 °C until constantweight. For determining CDW of naringenin-producing strain, 40mL ofculture was collected by centrifugation at 3000×g for 10min at roomtemperature using a BD Falcon 50mL conical tube. After removing thesupernatant, the conical tube was dried at 60 °C until constant weight.Based on the subtraction of the initial weight of the filter or conicaltube, the conversion coefficients between OD600 and CDW were cal-culated.

Concentrations of glycerol, 3HP, succinate, lactate, acetate, formate,and ethanol present in the supernatant of the culture broths of 3HP-producing E. coli were determined by an HPLC system (HPLCProminence, Shimadzu) equipped with an Aminex HPX-87H column(Bio-Rad, Hercules, CA, U.S.A.), a UV/Vis detector (SPD-20A), and arefractive index detector (RID-10A). The column temperature was set to65 °C, and 1.5 mM H2SO4 was used as the mobile phase with a flow rateof 0.6 mLmin−1. The flow cell temperature of the refractive index de-tector was set to 35 °C. The supernatant of the culture broth was ob-tained by centrifugation at 21,500×g for 5min at 4 °C, and then filteredthrough a Millex HV 0.45-μm filter (Merck KGaA, Germany). The su-pernatant of naringenin-producing E. coli was analyzed for quantifica-tion of naringenin as described previously (Xiu et al., 2017).

2.5. Collection of intermediate metabolites

Culture broth of 3HP-producing E. coli was sampled rapidly andfiltered through a 0.5-μm pore size filter (PTFE-type membrane,Advantec, Tokyo, Japan). Cells on the filter were immediately im-mersed in 1.6 mL methanol (−80 °C) containing 5 μM cycloleucine asan internal standard for the quantitation by gas chromatography-massspectrometry (GC/MS) and kept at −80 °C until extraction. For thepreparation of isotope-labeled standards to quantify intracellular ab-solute metabolite concentration by liquid chromatography-tandemmass spectrometry (LC-MS/MS), 25mL culture broth of MG1655(DE3)strain grown on a standard M9 medium containing 4 g L− 1 [U-13C]glucose was collected at mid-log phase and stationary phase. Followingaddition of 1.6mL of chloroform (−30 °C) and 640 μL of Milli-Q water(4 °C) and vortexing for 1min, the mixture was centrifuged at 3700×gfor 20min at 4 °C. The aqueous layer was dispensed into sevenEppendorf tubes as 250 μL in each and dried using a SpeedVac SPD1010(Thermo Fisher Scientific, Waltham, MA, USA) at room temperature. Todetermine the concentrations of the 13C-labeled intermediates, thedried extract was dissolved in 50 μL of standard solution containingknown concentrations of unlabeled metabolites. For analysis ofMG1655(DE3)/pETM6 mcrNC, 10mL culture broth was collected at20 h for nitrogen starvation, 8 h for phosphorus or sulfur starvation, and4 h for magnesium starvation. Metabolites were extracted by additionof 1.6 mL of chloroform, 640 μL of Milli-Q water dissolving the twodried samples of the isotope labeled extract. Dried samples were ob-tained by the same procedure as above.

2.6. LC-MS/MS analysis

LC-MS/MS analysis (LC: Agilent 1100 series, Agilent Technologies,Santa Clara, CA, USA; MS/MS: API 2000, AB SCIEX, MA, USA) wasperformed under the following conditions (Nishino et al., 2015):column, ProteCol-P C18 HQ103 (2.1× 150mm, particle size of 3 µm);mobile phase, 10mM tributylamine/15mM acetic acid in water (A) andmethanol (B); flow rate: 0.2mLmin−1; gradient curve, 100% A:0% B at0min, 100% A:0% B at 8min, 10% A:90% B at 24min, 100% A:0% B at24.1 min, and 100% A:0% B at 30min; injection volume, 3 μL; columntemperature, 35 °C; mode of mass analysis, negative ion mode; nebu-lizer flow, 55 psi; dry gas flow rate, 10 Lmin−1; sheath gas flow rate,

11 Lmin−1; dry gas temperature, 300 °C; sheath gas temperature,380 °C; capillary voltage, 3.5 kV. The dried sample was dissolved in50 μL of Milli-Q water. The peak of each target metabolite was identi-fied by comparing its chromatographic behavior with that of an au-thentic standard. Peak area was determined using the software Analyst(version 1.6.2, AB SCIEX). Quantitation was performed based on the MSdata using the intracellular concentration of each metabolite (mmol L-cell−1) determined using the ratio of the 13C peak area to the 12C peakarea. The ratio of aqueous volume to cellular dry weight is 0.00177 L/gCDW for E. coli (Neidhardt et al., 1996).

2.7. GC/MS analysis

GC/MS analysis (GC/MS: GCMS-QP2010 Ultra, Shimadzu, Japan)was performed under the following conditions (Tsugawa et al., 2011):The column was a 30m×0.25mm i.d. fused silica capillary columncoated with 0.25 µm CP-SIL 8 CB low bleed/MS (Agilent). The frontinlet temperature was 230 °C. The helium gas flow rate through thecolumn was 1mLmin−1. The column temperature was held at 80 °C for2min isothermally and then raised by 15 °Cmin−1 to 330 °C and heldthere for 6min isothermally. The transfer line and ion-source tem-peratures were 250 °C and 200 °C, respectively. The dried extract wasdissolved in 50 μL methoxyamine hydrochloride (20mgmL-pyr-idine−1) and incubated at 30 °C for 90min. For trimethyl silylation,25 μL of N-methyl-N-(trimethylsilyl) trifluoroacetamide was added andincubated at 37 °C for 30min. After 2 h cooling and centrifugation for5min at 20,400×g, 1 μL of derivatized supernatants were injected withsplit injection ratio of 1:25.

2.8. Metabolome data analysis

Metabolite peak areas were obtained by analyzing LC-MS/MS orGC/MS data with each instrument's software. For the quantitation byLC-MS/MS, the intracellular concentration of each metabolite was de-termined using the ratio of the 13C peak area to the 12C peak area. Forthe quantitation by GC/MS, the intracellular concentration of eachmetabolite was determined with internal standard method using cy-cloleucine as an internal standard. The correlation analysis betweenmetabolites concentrations and culture profile was performed by R3.2.4. software (http://www.r-project.com) and R-studio.

3. Results

3.1. Comparison of 3HP production under different nutrient starvationconditions

3HP-producing E. coli of MG1655(DE3)/pETM6 mcrNC was con-structed by over-expression of codon optimized mcrNC gene derivedfrom C. aurantiacus using pETM6 vector in MG1655(DE3). The strainwas aerobically cultivated overnight in standard M9 medium. Cellpellets were obtained by centrifugation and resuspended in nitrogen-free or phosphorus-free or sulfur-free or magnesium-free M9 mediumwith an initial OD600 of 2.0. The strain continuously consumed glucosewith no cell growth in all starved conditions until glucose was com-pletely depleted (Fig. 1). The conversion coefficients between OD600and CDW for magnesium, sulfur, phosphorus, and nitrogen starvationwere 0.33 ± 0.02, 0.32 ± 0.00, 0.32 ± 0.00, and 0.33 ± 0.02gCDW L−1 OD600−1, respectively. No significant difference was ob-served among the conditions. The culture results are summarized inTable 2. High-level 3HP production by the strain was observed undersulfur starvation (titer and yield were 2.3 ± 0.1mM and 6.0 ± 0.1C-mol%) or magnesium starvation (titer and yield were 2.0 ± 0.0mMand 5.7 ± 0.1C-mol%), whereas almost no 3HP production was de-tected under nitrogen starvation (titer and yield were 0.1 ± 0.0mMand 0.2 ± 0.0C-mol%) or phosphorus starvation (not detected).Acetate was produced as a major byproduct in all the cultures. The

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

217

http://www.r-project.com

highest production of acetate was confirmed under phosphorus star-vation (15.6 ± 0.3mM). Ethanol, lactate, and succinate were not de-tected in all the cultures. Total carbon recovery was lowest under ni-trogen starvation (28.2 ± 0.8C-mol%). A part of consumed glucoseunder nitrogen starvation might be used for glycogen synthesis (Holmeand Palmstierna, 1956).

Specific glucose uptake rate was lowest under nitrogen starvation(0.28 ± 0.02mmol gCDW−1 h−1) and highest under magnesiumstarvation (2.1 ± 0.06mmol gCDW−1 h−1). This result is consistentwith the characteristics of wild type E. coli under nutrition starvationconditions (Chubukov and Sauer, 2014).

3.2. Metabolic profiling analysis of 3HP-producing E. coli

The metabolic states under the four different stationary phase cul-tivations were compared by metabolome-analysis with GC/MS and LC-MS/MS. The concentrations of 42 intracellular metabolites includingamino acids, sugar phosphates, and organic acids were determined(Fig. 2). Correlation analysis of all metabolite concentrations and thespecific rate was performed to evaluate intracellular metabolic stateand identify key factors for 3HP production (Fig. 3).

The specific 3HP production rate had a strong correlation(r > 0.85) with ATP, malonyl-CoA, intracellular 3HP, and glycolysismetabolites, such as glucose 6-phosphate, 3-phosphoglycerate, and

phosphoenolpyruvate (PEP). Because malonyl-CoA is a precursor of3HP, the higher accumulation of malonyl-CoA under magnesium orsulfur starvation condition contributed to the enhanced 3HP produc-tion. The high ATP concentration under magnesium or sulfur starvationcould contribute to the malonyl-CoA supply because malonyl-CoA issynthesized from acetyl-CoA and CO2 using ATP by acetyl-CoA car-boxylase. Strong positive correlation between ATP and malonyl-CoA(r=0.95), as well as between 3HP production rate and malonyl-CoA(r=0.84), also indicated that the ATP level is important to producemalonyl-CoA by acetyl-CoA carboxylase (Fig. 3A, C). Intracellular 3HPconcentrations also had a strong correlation with the 3HP productionrate (r=0.92, Fig. 3B), and was higher than their extracellular con-centrations under all starvation conditions. This result suggested thatthe transport of 3HP could be a limiting step for 3HP production duringstationary phase.

The specific glucose uptake rate had a strong correlation (r > 0.85)with PEP and methionine. PEP accumulation contributed to increase inthe specific 3HP production rate coupled with the specific glucose up-take rate, because PEP is a substrate for the phosphotransferase system(PTS) of glucose transport (Chubukov and Sauer, 2014) (Fig. 3D). Al-though it has been reported that a strong correlation exists betweenfructose-1,6-bisphosphate (FBP) and the glycolytic flux in growing E.coli cells (Kochanowski et al., 2013), no such correlation between FBPlevels and the specific glucose uptake rate (r < 0.01) was observed

Fig. 1. Time-course of OD600, glucose, 3HP,and acetate levels in the 3HP-producing E. coliunder the starvation conditions. (A) nitrogenstarvation, (B) phosphorus starvation, (C)sulfur starvation, and (D) magnesium starva-tion. Error bars represent standard deviation oftriplicate experiments. Almost all the errorbars in this figure are smaller than the symbols.

Table 2Summary of the culture results of 3-hydroxypropionic acid producing E. coli strain.

Nitrogen starvation Phosphorus starvation Sulfur starvation Magnesium starvation

Consumed glucose (mM) 17.8 ± 0.4 17.1 ± 0.1 19.1 ± 1.0 17.2 ± 0.13HP titer (mM) 0.1 ± 0.0 0.0 ± 0.0 2.3 ± 0.1 2.0 ± 0.0Glucose uptake rate (mmol gCDW-1 h-1) 0.28 ± 0.02 1.04 ± 0.04 0.73 ± 0.03 2.1 ± 0.063HP production rate (mmol gCDW-1 h-1) 0.00 ± 0.00 0.00 ± 0.00 0.14 ± 0.01 0.28 ± 0.01Acetate production rate (mmol gCDW-1 h-1) 0.22 ± 0.02 0.75 ± 0.03 0.54 ± 0.02 1.54 ± 0.063HP yield (C-mol%) 0.2 ± 0.0 0.0 ± 0.0 6.0 ± 0.1 5.7 ± 0.1Acetate yield (C-mol%) 22.9 ± 0.7 30.4 ± 0.4 25.0 ± 0.5 27.7 ± 0.6Biomass yield (C-mol%) 5.0 ± 0.2 17.5 ± 1.3 8.5 ± 0.3 18.2 ± 1.1Total carbon recovery (C-mol%) 28.2 ± 0.8 47.9 ± 1.7 39.5 ± 0.2 51.6 ± 0.9

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

218

Fig. 2. Metabolic profiling of central metabolic pathway. Each panel represents the absolute metabolites concentrations of MG1655(DE3)/pETM6 mcrNC undernitrogen starvation, phosphorus starvation, sulfur starvation and magnesium starvation. Error bars represent standard deviation of triplicate culture experiments.

Fig. 3. Scatter plot with correlation analysisresults. Red solid line indicates regression linewith strong correlation coefficient. (Red) ni-trogen starvation, (Orange) phosphorus star-vation, (Green) sulfur starvation and (Blue)magnesium starvation. (For interpretation ofthe references to color in this figure legend, thereader is referred to the web version of thisarticle.)

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

219

under the starvation conditions (Fig. 3E).The accumulation levels of almost all intracellular free amino acids

showed the same tendency depending on the nutrient starvation con-dition. Intracellular free amino acids under nitrogen starvation were thelowest of all starvation conditions. Sulfur starvation showed higheramount of accumulation of free amino acids except for methionine.These results suggested that E. coli cells consumed all of free aminoacids under nitrogen starvation, or methionine under sulfur starvation,as a source of nitrogen or sulfur, respectively.

3.3. Application of nutrient starvation for flavonoid production

To demonstrate the effect of sulfur or magnesium starvation on theother target metabolite derived from malonyl-CoA, the nutrient star-vation conditions were implemented for naringenin production usingan engineered E. coli strain constructed in previous study (Jones et al.,2016b). Naringenin was chosen for two reasons. First, naringenin isknown as a pharmacologically useful plant flavonoid molecule (Wanget al., 2016). Second, several reports have demonstrated that malonyl-CoA supply is a major bottleneck for the naringenin production(Leonard et al., 2008; Fowler et al., 2009; Wu et al., 2014). One mo-lecule of naringenin is synthesized from one molecule of p-coumaricacid using three molecules of malonyl-CoA.

The strain of BL21star(DE3)/pFlavoOpt has an IPTG-inducible ex-pression vector containing biosynthetic genes for naringenin produc-tion. The engineered strain optimized for the expression of the nar-ingenin biosynthetic genes, was cultivated using standard minimal M9medium and nutrient starved minimal M9 media. Because productivityof the flavonoid production is sensitive to the induction time point(Jones et al., 2016a), cells from stationary phase culture broth onstandard M9 medium were serially diluted and inoculated into a freshmedium, and IPTG was added at 6 h or 7 h. The culture results areshown in Fig. 4. The highest production titer of 144 ± 15 μM wasobserved under magnesium starvation with IPTG induction at mid-logphase, whereas the maximum titer was 117 ± 7 μM on standard M9medium. Maximal productivity was also higher under magnesiumstarvation culture (127 ± 21 μmol gCDW−1) than in standard condi-tion (41 ± 0.5 μmol gCDW−1). These results indicated that magne-sium starvation improved the carbon flux toward malonyl-CoA synth-esis, and led to increase in naringenin production. On the other hand,sulfur starvation was not suitable for naringenin production(23 ± 0.4 μM maximum titer), because the cell growth under sulfurstarvation condition was drastically decreased after the induction (seeFig. S1). The productivity under sulfur starvation was slightly improvedcompared with the standard M9 condition (61 ± 2 μmol gCDW−1).

To investigate whether the nutrient starvation approach is affectedby type of carbon sources, the naringenin production under the mag-nesium starvation was also evaluated with xylose or glycerol minimalmedium with IPTG induction after 6 h. Here again, maximum specific

productivities of naringenin were higher under magnesium starvationthan in the control condition (Fig. S2). These results demonstrated thatthe effect of magnesium starvation is independent of the carbon sourcetype, and is useful for the development of the microbial productionprocess.

4. Discussion

Essential nutrient starvation is one of the approaches for bio-pro-duction in stationary phase. Nitrogen starvation is widely used for in-hibiting cell growth and successfully enhancing productivity of varioustargets, such as poly(3-hydroxybutyrate) and butanol in stationaryphase (Wang and Lee, 1997; Al-Shorgani et al., 2012). However, theseproductivities in stationary phase might change with different nutrientstarvation because the metabolic state is dependent on the specificnutrient limitation. In the present study, we focused on the influence ofspecific nutrients starvations on metabolic activities and investigatedthe optimum nutrient starvation condition for producing malonyl-CoAderived metabolites in E. coli.

High-level 3HP production was observed under sulfur or magne-sium starvation, whereas almost no 3HP production was detected undernitrogen or phosphorus starvation. Increased ATP concentration shouldenhance the malonyl-CoA supply and lead to production of 3HP undersulfur or magnesium starvation. Culture experiments with the nar-ingenin-producing E. coli indicated that magnesium starvation leads tothe highest production titer (144 ± 15 μM) and specific productivity(127 ± 21 μmol gCDW−1). These results demonstrated that magne-sium starvation is a useful approach to activate metabolic state for theproduction of malonyl-CoA derivatives.

Specific substrate uptake rate is an important factor for 3HP pro-duction because the specific glucose uptake rate in stationary phasetends to be lower than in exponential growth phase (Chubukov andSauer, 2014). E. coli producing 3HP under nitrogen starvation showedthe lowest glucose uptake rate. This could be explained by accumula-tion of α-ketoglutarate (αKG), because αKG acts as an allosteric in-hibitor for PTS. The αKG concentration under nitrogen starvation(4.1 mM) was higher than the Ki value of PTS (1.3 mM). Previous studyreported that FBP concentration had a strong correlation with theglycolytic flux in growing E. coli cells (Kochanowski et al., 2013).However, no correlation between FBP concentration and the specificglucose uptake rate was observed in this study (r < 0.01, Fig. 3E). Theaccumulation of αKG under nitrogen starvation could have hamperedthis correlation. Nitrogen starvation is a challenge for improving theproduction rate because αKG strongly inhibits glucose uptake and ATPsupply in TCA cycle (Fig. 5A) (Doucette et al., 2011; Wright et al.,1967). In this study, the specific glucose uptake rate had a strong cor-relation with PEP (r=0.87). PEP is needed for glucose transport as asubstrate of PTS and the accumulation of PEP contributed to glucoseuptake into the cell. Increasing PEP accumulation would be effective for

Fig. 4. Naringenin production under nutrient starvationconditions. (A) Naringenin production versus cell densityat induction time point using standard M9 and nutrient-limited M9 (sulfur or magnesium starvation). (B) Maximalproductivity of naringenin per final culture volume at24 h. Cell density was calculated from OD600 using theexperimentally determined conversion coefficient of 0.8gCDW L−1 OD600−1. Error bars represent standard de-viation of triplicate experiments.

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

220

enhancing glucose uptake rate in stationary phase.Phosphorus starvation decreased the intracellular concentration of

malonyl-CoA and ATP. A previous study reported that phosphorusstarvation decreases flux of the TCA cycle with activation of tran-scriptional regulator arcA (Marzan and Shimizu, 2011). The decrease influx of TCA cycle led to shortage of ATP and malonyl-CoA (Fig. 5B).Sulfur starvation resulted in the highest ATP level and caused 3HPproduction increase due to increased supply of malonyl-CoA. E. coliuptakes sulfate as a source of sulfur and produces cysteine with con-sumption of two molecules of ATP and one molecule of NADPH persulfate molecule (Sekowska et al., 2000). Higher accumulation ofmalonyl-CoA should be achieved by reducing consumption of ATP viasulfur assimilation (Fig. 5C). Magnesium starvation improved glucoseuptake rate and specific 3HP production rate to the highest level in allof the cultures. Pyruvate kinase converts PEP to pyruvate, and needsmagnesium in the active site of the enzyme (Garfinkel and Garfinkel,1985; Kumar and Barth, 2010). Magnesium starvation decreased theactivity of pyruvate kinase and accumulated PEP, which led to in-creased specific glucose uptake (Fig. 5D). Magnesium starvation in-creased PEP levels compared with the other starvations. The sulfur ormagnesium starvation improved production of malonyl-CoA-derivedmetabolites. Because these two starvations increase ATP concentrationthrough different mechanisms (Fig. 5), the 3HP productivity might befurther enhanced under the double starvation of magnesium and sulfur.In a large scale process, cells cannot be collected by centrifugation andtransferred to a fresh medium, therefore, it should be led to the sta-tionary phase by reducing the concentration of an essential nutrient inthe initial medium. Since the cell growth stops when an essential nu-trient is depleted, it must be difficult to simultaneously deplete twoessential nutrients in industrial bio-production process. However, theevaluations under such combinatorial starvations would provide ussome interesting knowledge for the cellular system.

Strong positive correlation (r= 0.945) between intracellular con-centrations of ATP and malonyl-CoA indicates that the ATP level isimportant for malonyl-CoA synthesis. However, there is difference in3HP productivity among the conditions of sulfur or magnesium star-vation despite similar intracellular ATP (Fig. 3A, and C). This differencealso be related to the binding between Mg ion and ATP. In many en-zymes including acetyl-CoA carboxylase (Guchhait et al., 1974), ATPhas to bind with magnesium ion due to their activity. The proportion ofMgATP must decrease under magnesium starvation. Actually, it hasbeen reported that the proportion of MgATP was 88% under standardconditions, whereas it was 70% under magnesium-free conditions incytosol of sycamore cells (Gout et al., 2014). Although the proportion ofMgATP was not measured in this study, that under the magnesiumstarvation must be lower than under sulfur starvation. However, Fig. 3Ashows that the 3HP production under the magnesium starvation ishigher than that under the sulfur starvation. This suggests that there is

no shortage of MgATP under the magnesium starvation. Since the cellswere transfer to the magnesium-free medium after growing on normalM9 medium, a certain amount of Mg would exist in the cells under themagnesium starvation. A hypothesis can explain the reason why 3HPproduction under magnesium starvation was higher than that undersulfur starvation. The malonyl-CoA reductase required NADPH as co-factor. Chubukov and Sauer (2014) have proposed a hypothesis thatglycolytic flux was shuttled to the Entner-Doudoroff (ED) pathwayunder magnesium starvations. Because a NADPH is generated whenglucose is catabolized via the ED pathway, the activity of malonyl-CoAreductase would be activated due to the enhanced NADPH supply. Al-though the amount of NADPH under the magnesium starvation in Fig. 2did not increase, it would be balanced at this concentration becauseNADPH consumption also increased due to the increasing 3HP pro-duction rate.

This study evaluated the effect of nutritional starvation on 3HP-producing E. coli with overexpression of malonyl-CoA reductase; mag-nesium or sulfur starvation led to high-level 3HP production yield(5.7C-mol% and 6.0C-mol%). In a previous report, a higher yield of3HP was achieved at 18C-mol% by the batch cultivation of E. coli withoverexpression of acetyl-CoA carboxylase derived from C. glutamicum(Cheng et al., 2016). The additional overexpression of acetyl-CoA car-boxylase should improve 3HP production during stationary phase.Metabolic profiling analysis revealed that 3HP transport system is alimiting step because intracellular concentration of 3HP was higherthan that in extracellular. The 3HP production could be enhanced byactivation of the transporter for 3HP. Since little is known about thetransport system of 3HP in E. coli, it is important to identify thetransporter for 3HP.

The type of nutrient starvation should be selected for each targetmetabolite in bio-production with stationary phase cells. However, ef-fects of nutrient starvation for production of each target metabolitewere largely unclear. In this study, metabolic profiling analysis char-acterized the metabolic state during stationary phase and suggested anoptimum nutrient starvation condition for target metabolite produc-tion. For example, in production of malonyl-CoA derivatives, such as3HP, fatty-acids or flavonoids, it might be suitable to use sulfur ormagnesium starvation because of accumulation of high amount of in-tracellular malonyl-CoA. The naringenin-producing E. coli demon-strated that sulfur or magnesium starvation increased the specific pro-ductivity of naringenin (Fig. 4B). Furthermore, the production titerunder magnesium starvation condition was higher than of standardminimal medium without nutrient starvation, though magnesiumstarvation decreased the final cell density. For further development offermentation process of naringenin production, magnesium-limited fedbatch technique is a possible way to maintain the metabolic state forhigh production, and also improve cell densities (Willrodt et al., 2016).Production of acetyl-CoA derivatives, such as mevalonate or

Fig. 5. Metabolic state of 3HP-producing E. coli with malonyl-CoA reductase under several nutrient starvation conditions. Gray arrow indicates possible low-activereactions revealed by the metabolic profiling analysis.

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

221

isopropanol, might be increased using sulfur, magnesium, or phos-phorus starvation because of accumulation of high amount of in-tracellular acetyl-CoA. Amino acid production might be increased usingsulfur starvation because the highest accumulation of amino acids wasobserved under sulfur starvation. On the other hand, sulfur starvationmust not be the best strategy to improve the production of cysteine ormethionine because these amino acids are a source of sulfur for cellgrowth. Li et al. demonstrated the increased production of tyrosine andmevalonate in engineered E. coli upon sulfur starvation (Li et al., 2016).As shown in our results, evaluation of productivity of each metaboliteunder different nutrient starvation conditions is important for targetproduction.

CRediT authorship contribution statement

Kento Tokuyama: Conceptualization, Data curation, Writing -original draft, Writing - review & editing. Yoshihiro Toya:Conceptualization, Writing - original draft, Writing - review &editing. Fumio Matsuda: Conceptualization, Methodology, Writing -original draft, Writing - review & editing. Brady F. Cress:Conceptualization, Writing - original draft, Writing - review &editing. Mattheos A.G. Koffas: Conceptualization, Writing - originaldraft, Writing - review & editing. Hiroshi Shimizu: Conceptualization,Writing - original draft, Writing - review & editing.

Acknowledgements

This research was supported by a Grant-in-Aid for Japan Society forthe Promotion of Science (JSPS) Fellows (15J06350) and “Program forLeading Graduate Schools” of the Ministry of Education, Culture,Sports, Science and Technology, Japan. This work was partially sup-ported by a Japan Science and Technology Agency (JST)-Mirai ProgramGrant Number JPMJMI17EJ, Japan, and a Grant-in-Aid for ScientificResearch (B) No.16H04576, Japan. M.A.G. would like to acknowledgefunding from NSF CBET award number 1604547.

Appendix A. Supplementary material

Supplementary data associated with this article can be found in theonline version at doi:10.1016/j.ymben.2018.12.002.

References

Al-Shorgani, N.K.N., Ali, E., Kalil, M.S., Yusoff, W.M.W., 2012. Bioconversion of butyricacid to butanol by Clostridium saccharoperbutylacetonicum N1-4 (ATCC 13564) in alimited nutrient medium. Bioenergy Res. 5, 287–293. https://doi.org/10.1007/s12155-011-9126-6.

Bai, X., Zhang, X., Chen, L., Zhang, J., Zhang, L., Zhao, X., Zhao, T., Zhao, Y., 2014.Protective effect of naringenin in experimental ischemic stroke: down-regulatedNOD2, RIP2, NF-κB, MMP-9 and up-regulated claudin-5 expression. Neurochem. Res.39, 1405–1415. https://doi.org/10.1007/s11064-014-1326-y.

Baptista, F.I., Henriques, A.G., Silva, A.M., Wiltfang, J., da Cruz e Silva, O.A., 2014.Flavonoids as therapeutic compounds targeting key proteins involved in Alzheimer'sdisease. ACS Chem. Neurosci. 5, 83–92. https://doi.org/10.1021/cn400213r.

Borodina, I., Kildegaard, K.R., Jensen, N.B., Blicher, T.H., Maury, J., Sherstyk, S.,Schneider, K., Lamosa, P., Herrgård, M.J., Rosenstand, I., Öberg, F., Forster, J.,Nielsen, J., 2015. Establishing a synthetic pathway for high-level production of 3-hydroxypropionic acid in Saccharomyces cerevisiae via β-alanine. Metab. Eng. 27,57–64. https://doi.org/10.1016/j.ymben.2014.10.003.

Cheng, Z., Jiang, J., Wu, H., Li, Z., Ye, Q., 2016. Enhanced production of 3-hydro-xypropionic acid from glucose via malonyl-CoA pathway by engineered Escherichiacoli. Bioresour. Technol. 200, 897–904. https://doi.org/10.1016/j.biortech.2015.10.107.

Chubukov, V., Sauer, U., 2014. Environmental dependence of stationary-phase metabo-lism in Bacillus subtilis and Escherichia coli. Appl. Environ. Microbiol. 80, 2901–2909.https://doi.org/10.1128/AEM.00061-14.

Doucette, C.D., Schwab, D.J., Wingreen, N.S., Rabinowitz, J.D., 2011. α-Ketoglutaratecoordinates carbon and nitrogen utilization via enzyme I inhibition. Nat. Chem. Biol.7, 894–901. https://doi.org/10.1038/nchembio.685.

Fowler, Z.L., Gikandi, W.W., Koffas, M.A., 2009. Increased malonyl coenzyme A bio-synthesis by tuning the Escherichia coli metabolic network and its application toflavanone production. Appl. Environ. Microbiol. 75, 5831–5839. https://doi.org/10.

1128/AEM.00270-09.Garfinkel, L., Garfinkel, D., 1985. Magnesium regulation of the glycolytic pathway and

the enzymes involved. Magnesium 4, 60–72.Gout, E., Rébeillé, F., Douce, R., Bligny, R., 2014. Interplay of Mg2+, ADP, and ATP in

the cytosol and mitochondria: unravelling the role of Mg2+ in cell respiration. Proc.Natl. Acad. Sci. USA 111, E4560–E4567. https://doi.org/10.1073/pnas.1406251111.

Guchhait, R.B., Polakis, S.E., Hollis, D., Fenselau, C., Lane, M.D., 1974. Acetyl coenzymeA carboxylase system of Escherichia coli. site of carboxylation of biotin and enzymaticreactivity of 1′-N-(ureido)-carboxybiotin derivatives. J. Biol. Chem. 249, 6646–6656.

Holme, T., Palmstierna, H., 1956. Changes in glycogen and nitrogen-containing com-pounds in Escherichia coli B during growth in deficient media. Acta Chem. Scand. 10,578–586. https://doi.org/10.3891/acta.chem.scand.10-1553.

Ibrahim, A., Sobeh, M., Ismail, A., Alaa, A., Sheashaa, H., Sobh, M., Badria, F., 2014. Free-B-Ring flavonoids as potential lead compounds for colon cancer therapy. Mol. Clin.Oncol. 2, 581–585. https://doi.org/10.3892/mco.2014.278.

Jiang, X., Meng, X., Xian, M., 2009. Biosynthetic pathways for 3-hydroxypropionic acidproduction. Appl. Microbiol. Biotechnol. 82, 995–1003. https://doi.org/10.1007/s00253-009-1898-7.

Johnson, A.O., Gonzalez-Villanueva, M., Wong, L., Steinbüchel, A., Tee, K.L., Xu, P.,Wong, T.S., 2017. Design and application of genetically-encoded malonyl-CoA bio-sensors for metabolic engineering of microbial cell factories. Metab. Eng. 44,253–264. https://doi.org/10.1016/j.ymben.2017.10.011.

Jones, J.A., Collins, S.M., Vernacchio, V.R., Lachance, D.M., Koffas, M.A.G., 2016a.Optimization of naringenin and p-coumaric acid hydroxylation using the native E. colihydroxylase complex, HpaBC. Biotechnol. Prog. 32, 21–25. https://doi.org/10.1002/btpr.2185.

Jones, J.A., Vernacchio, V.R., Sinkoe, A.L., Collins, S.M., Ibrahim, M.H.A., Lachance,D.M., Hahn, J., Koffas, M.A.G., 2016b. Experimental and computational optimizationof an Escherichia coli co-culture for the efficient production of flavonoids. Metab. Eng.35, 55–63. https://doi.org/10.1016/j.ymben.2016.01.006.

Kim, K., Kim, S.K., Park, Y.C., Seo, J.H., 2014. Enhanced production of 3-hydro-xypropionic acid from glycerol by modulation of glycerol metabolism in recombinantEscherichia coli. Bioresour. Technol. 156, 170–175. https://doi.org/10.1016/j.biortech.2014.01.009.

Kochanowski, K., Volkmer, B., Gerosa, L., Haverkorn van Rijsewijk, B.R., Schmidt, A.,Heinemann, M., 2013. Functioning of a metabolic flux sensor in Escherichia coli. Proc.Natl. Acad. Sci. USA 110, 1130–1135. https://doi.org/10.1073/pnas.1202582110.

Kumar, S., Barth, A., 2010. Phosphoenolpyruvate and Mg2+ binding to pyruvate kinasemonitored by infrared spectroscopy. Biophys. J. 98, 1931–1940. https://doi.org/10.1016/j.bpj.2009.12.4335.

Kumar, V., Ashok, S., Park, S., 2013. Recent advances in biological production of 3-hy-droxypropionic acid. Biotechnol. Adv. 31, 945–961. https://doi.org/10.1016/j.biotechadv.2013.02.008.

Lan, E.I., Chuang, D.S., Shen, C.R., Lee, A.M., Ro, S.Y., Liao, J.C., 2015. Metabolic en-gineering of cyanobacteria for photosynthetic 3-hydroxypropionic acid productionfrom CO2 using Synechococcus elongatus PCC 7942. Metab. Eng. 31, 163–170. https://doi.org/10.1016/j.ymben.2015.08.002.

Leonard, E., Lim, K.H., Saw, P.N., Koffas, M.A.G., 2007. Engineering central metabolicpathways for high-level flavonoid production in Escherichia coli. Appl. Environ.Microbiol. 73, 3877–3886. https://doi.org/10.1128/AEM.00200-07.

Leonard, E., Yan, Y., Fowler, Z.L., Li, Z., Lim, C.G., Lim, K.H., Koffas, M.A., 2008. Strainimprovement of recombinant Escherichia coli for efficient production of plant flavo-noids. Mol. Pharm. 5, 257–265. https://doi.org/10.1021/mp7001472.

Li, S., Jendresen, C.B., Nielsen, A.T., 2016. Increasing production yield of tyrosine andmevalonate through growth limitation. Process Biochem. https://doi.org/10.1016/j.procbio.2016.09.007.

Liu, C., Ding, Y., Zhang, R., Liu, H., Xian, M., Zhao, G., 2015. Functional balance betweenenzymes in malonyl-CoA pathway for 3-hydroxypropionate biosynthesis. Metab. Eng.34, 104–111. https://doi.org/10.1016/j.ymben.2016.01.001.

Liu, C., Wang, Q., Xian, M., Ding, Y., Zhao, G., 2013. Dissection of malonyl-coenzyme Areductase of Chloroflexus aurantiacus results in enzyme activity improvement. PLoSOne 8, e75554. https://doi.org/10.1371/journal.pone.0075554.

Marzan, L.W., Shimizu, K., 2011. Metabolic regulation of Escherichia coli and its phoB andphoR genes knockout mutants under phosphate and nitrogen limitations as well as atacidic condition. Microb. Cell Fact. 10, 39. https://doi.org/10.1186/1475-2859-10-39.

Masuda, A., Toya, Y., Shimizu, H., 2017. Metabolic impact of nutrient starvation inmevalonate-producing Escherichia coli. Bioresour. Technol. 245, 1634–1640. https://doi.org/10.1016/j.biortech.2017.04.110.

Neidhardt, F.C., Curtiss III, R., Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B.,Reznikoff, W.S., Riley, M., Schaechter, M., Umbarger, H.E., 1996. Escherichia coliand Salmonella: cellular and molecular biology. Amer Society for Microbiology,Washington D.C.

Nishino, S., Okahashi, N., Matsuda, F., Shimizu, H., 2015. Absolute quantitation of gly-colytic intermediates reveals thermodynamic shifts in Saccharomyces cerevisiae strainslacking PFK1 or ZWF1 genes. J. Biosci. Bioeng. 120, 280–286. https://doi.org/10.1016/j.jbiosc.2015.01.012.

Panche, A.N., Diwan, A.D., Chandra, S.R., 2016. Flavonoids: an overview. J. Nutr. Sci. 5,e47. https://doi.org/10.1017/jns.2016.41.

Pandey, R.P., Parajuli, P., Koffas, M.A.G., Sohng, J.K., 2016. Microbial production ofnatural and non-natural flavonoids: pathway engineering, directed evolution andsystems/synthetic biology. Biotechnol. Adv. 34, 634–662. https://doi.org/10.1016/j.biotechadv.2016.02.012.

Raman, S., Rogers, J.K., Taylor, N.D., Church, G.M., 2014. Evolution-guided optimizationof biosynthetic pathways. Proc. Natl. Acad. Sci. USA 111, 17803–17808. https://doi.org/10.1073/pnas.1409523111.

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

222

https://doi.org/10.1016/j.ymben.2018.12.002https://doi.org/10.1007/s12155-011-9126-6https://doi.org/10.1007/s12155-011-9126-6https://doi.org/10.1007/s11064-014-1326-yhttps://doi.org/10.1021/cn400213rhttps://doi.org/10.1016/j.ymben.2014.10.003https://doi.org/10.1016/j.biortech.2015.10.107https://doi.org/10.1016/j.biortech.2015.10.107https://doi.org/10.1128/AEM.00061-14https://doi.org/10.1038/nchembio.685https://doi.org/10.1128/AEM.00270-09https://doi.org/10.1128/AEM.00270-09http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref9http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref9https://doi.org/10.1073/pnas.1406251111http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref11http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref11http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref11https://doi.org/10.3891/acta.chem.scand.10-1553https://doi.org/10.3892/mco.2014.278https://doi.org/10.1007/s00253-009-1898-7https://doi.org/10.1007/s00253-009-1898-7https://doi.org/10.1016/j.ymben.2017.10.011https://doi.org/10.1002/btpr.2185https://doi.org/10.1002/btpr.2185https://doi.org/10.1016/j.ymben.2016.01.006https://doi.org/10.1016/j.biortech.2014.01.009https://doi.org/10.1016/j.biortech.2014.01.009https://doi.org/10.1073/pnas.1202582110https://doi.org/10.1016/j.bpj.2009.12.4335https://doi.org/10.1016/j.bpj.2009.12.4335https://doi.org/10.1016/j.biotechadv.2013.02.008https://doi.org/10.1016/j.biotechadv.2013.02.008https://doi.org/10.1016/j.ymben.2015.08.002https://doi.org/10.1016/j.ymben.2015.08.002https://doi.org/10.1128/AEM.00200-07https://doi.org/10.1021/mp7001472https://doi.org/10.1016/j.procbio.2016.09.007https://doi.org/10.1016/j.procbio.2016.09.007https://doi.org/10.1016/j.ymben.2016.01.001https://doi.org/10.1371/journal.pone.0075554https://doi.org/10.1186/1475-2859-10-39https://doi.org/10.1186/1475-2859-10-39https://doi.org/10.1016/j.biortech.2017.04.110https://doi.org/10.1016/j.biortech.2017.04.110http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref30http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref30http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref30http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref30https://doi.org/10.1016/j.jbiosc.2015.01.012https://doi.org/10.1016/j.jbiosc.2015.01.012https://doi.org/10.1017/jns.2016.41https://doi.org/10.1016/j.biotechadv.2016.02.012https://doi.org/10.1016/j.biotechadv.2016.02.012https://doi.org/10.1073/pnas.1409523111https://doi.org/10.1073/pnas.1409523111

Rathnasingh, C., Raj, S.M., Lee, Y., Catherine, C., Ashok, S., Park, S., 2012. Production of3-hydroxypropionic acid via malonyl-CoA pathway using recombinant Escherichia colistrains. J. Biotechnol. 157, 633–640. https://doi.org/10.1016/j.jbiotec.2011.06.008.

Sekowska, A., Kung, H.F., Danchin, A., 2000. Sulfur metabolism in Escherichia coli andrelated bacteria: facts and fiction. J. Mol. Microbiol. Biotechnol. 2, 145–177.

Tokuyama, K., Ohno, S., Yoshikawa, K., Hirasawa, T., Tanaka, S., Furusawa, C., Shimizu,H., 2014. Increased 3-hydroxypropionic acid production from glycerol, by mod-ification of central metabolism in Escherichia coli. Microb. Cell Fact. 13, 64. https://doi.org/10.1186/1475-2859-13-64.

Toya, Y., Hirasawa, T., Ishikawa, S., Chumsakul, O., Morimoto, T., Liu, S., Masuda, K.,Kageyama, Y., Ozaki, K., Ogasawara, N., Shimizu, H., 2015. Enhanced dipicolinicacid production during the stationary phase in Bacillus subtilis by blocking acetoinsynthesis. Biosci. Biotechnol. Biochem. 8451, 1–8. https://doi.org/10.1080/09168451.2015.1060843.

Tsugawa, H., Bamba, T., Shinohara, M., Nishiumi, S., Yoshida, M., Fukusaki, E., 2011.Practical non-targeted gas chromatography/mass spectrometry-based metabolomicsplatform for metabolic phenotype analysis. J. Biosci. Bioeng. 112, 292–298. https://doi.org/10.1016/j.jbiosc.2011.05.001.

Wang, J., Guleria, S., Koffas, M.A., Yan, Y., 2016. Microbial production of value-addednutraceuticals. Curr. Opin. Biotechnol. 37, 97–104. https://doi.org/10.1016/j.copbio.2015.11.003.

Wang, F., Lee, S.Y., 1997. Poly(3-hydroxybutyrate) production with high productivityand high polymer content by a fed-batch culture of Alcaligenes latus under nitrogenlimitation. Appl. Environ. Microbiol. 63, 3703–3706. https://doi.org/10.1023/a:1012222625201.

Werpy, T., Petersen, G., 2004. Top value added chemicals from biomass. vol. I: results ofscreening for potential candidates from sugars and synthesis gas. U.S. Dep. Energy.https://doi.org/10.2172/926125.

Willrodt, C., Hoschek, A., Bühler, B., Schmid, A., Julsing, M.K., 2016. Decoupling

production from growth by magnesium sulfate limitation boosts de novo limoneneproduction. Biotechnol. Bioeng. 113, 1305–1314.

Wright, J.A., Maeba, P., Sanwal, B.D., 1967. Allosteric regulation of the activity of citratesynthetase of by α-ketoglutarate. Biochem. Biophys. Res. Commun. 29, 34–38.https://doi.org/10.1016/0006-291X(67)90536-0.

Wu, J., Yu, O., Du, G., Zhou, J., Chen, J., 2014. Fine-Tuning of the Fatty Acid Pathway bySynthetic Antisense RNA for Enhanced (2S)-Naringenin Production from L-Tyrosinein Escherichia coli. Appl. Environ. Microbiol. 80, 7283–7292. https://doi.org/10.1128/AEM.02411-14.

Xiu, Y., Jang, S., Jones, J.A., Zill, N.A., Linhardt, R.J., Yuan, Q., Jung, G.Y., Koffas,M.A.G., 2017. Naringenin-responsive riboswitch-based fluorescent biosensor modulefor Escherichia coli co-cultures. Biotechnol. Bioeng. 114, 2235–2244. https://doi.org/10.1002/bit.26340.

Xu, P., Li, L., Zhang, F., Stephanopoulos, G., Koffas, M.A., 2014. Improving fatty acidsproduction by engineering dynamic pathway regulation and metabolic control. Proc.Natl. Acad. Sci. USA 111, 11299–11304. https://doi.org/10.1073/pnas.1406401111.

Xu, P., Ranganathan, S., Fowler, Z.L., Maranas, C.D., Koffas, M.A.G., 2011. Genome-scalemetabolic network modeling results in minimal interventions that cooperatively forcecarbon flux towards malonyl-CoA. Metab. Eng. 13, 578–587. https://doi.org/10.1016/j.ymben.2011.06.008.

Xu, P., Vansiri, A., Bhan, N., Koffas, M.A.G., 2012. ePathBrick: a synthetic biology plat-form for engineering metabolic pathways in E. coli. ACS Synth. Biol. 1, 256–266.https://doi.org/10.1021/sb300016b.

Zeng, W., Jin, L., Zhang, F., Zhang, C., Liang, W., 2018. Naringenin as a potential im-munomodulator in therapeutics. Pharmacol. Res. 135, 122–126. https://doi.org/10.1016/j.phrs.2018.08.002.

Zha, W., Rubin-Pitel, S.B., Shao, Z., Zhao, H., 2009. Improving cellular malonyl-CoA levelin Escherichia coli via metabolic engineering. Metab. Eng. 11, 192–198. https://doi.org/10.1016/j.ymben.2009.01.005.

K. Tokuyama et al. Metabolic Engineering 52 (2019) 215–223

223

https://doi.org/10.1016/j.jbiotec.2011.06.008http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref36http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref36https://doi.org/10.1186/1475-2859-13-64https://doi.org/10.1186/1475-2859-13-64https://doi.org/10.1080/09168451.2015.1060843https://doi.org/10.1080/09168451.2015.1060843https://doi.org/10.1016/j.jbiosc.2011.05.001https://doi.org/10.1016/j.jbiosc.2011.05.001https://doi.org/10.1016/j.copbio.2015.11.003https://doi.org/10.1016/j.copbio.2015.11.003https://doi.org/10.1023/a:1012222625201https://doi.org/10.1023/a:1012222625201https://doi.org/10.2172/926125http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref43http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref43http://refhub.elsevier.com/S1096-7176(18)30325-2/sbref43https://doi.org/10.1016/0006-291X(67)90536-0https://doi.org/10.1128/AEM.02411-14https://doi.org/10.1128/AEM.02411-14https://doi.org/10.1002/bit.26340https://doi.org/10.1002/bit.26340https://doi.org/10.1073/pnas.1406401111https://doi.org/10.1016/j.ymben.2011.06.008https://doi.org/10.1016/j.ymben.2011.06.008https://doi.org/10.1021/sb300016bhttps://doi.org/10.1016/j.phrs.2018.08.002https://doi.org/10.1016/j.phrs.2018.08.002https://doi.org/10.1016/j.ymben.2009.01.005https://doi.org/10.1016/j.ymben.2009.01.005

Magnesium starvation improves production of malonyl-CoA-derived metabolites in Escherichia coliIntroductionMaterials and methodsStrains and plasmidsBatch culture protocol for 3HP productionBatch culture protocol for naringenin productionAnalytical methodsCollection of intermediate metabolitesLC-MS/MS analysisGC/MS analysisMetabolome data analysis

ResultsComparison of 3HP production under different nutrient starvation conditionsMetabolic profiling analysis of 3HP-producing E. coliApplication of nutrient starvation for flavonoid production

Discussionmk:H1_16AcknowledgementsSupplementary materialReferences