Metabolic Engineering 13 (2011) 76–81

Contents lists available at ScienceDirect

Metabolic Engineering

1096-71

doi:10.1

n Corr

E-m

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

Manipulating redox and ATP balancing for improved productionof succinate in E. coli

Amarjeet Singh a, Keng Cher Soh b, Vassily Hatzimanikatis b, Ryan T. Gill a,n

a Department of Chemical and Biological Engineering, University of Colorado, UCB 424, Boulder, CO 80309, USAb Laboratory of Computational Systems Biotechnology, Ecole Polytechnique Federale de Lausanne (EPFL), CH 1015 Lausanne, Switzerland

a r t i c l e i n f o

Article history:

Received 19 October 2010

Accepted 21 October 2010Available online 30 October 2010

Keywords:

Escherichia coli

Redox ratio

NADH/NAD+

Succinate

Fumarate

Metabolic engineering

NZN111

AFP111

76/$ - see front matter & 2010 Elsevier Inc. A

016/j.ymben.2010.10.006

esponding author: Fax: +1 303 492 4341.

ail address: rtg@colorado.edu (R.T. Gill).

a b s t r a c t

Redox and energy balance plays a key role in determining microbial fitness. Efforts to redirect bacterial

metabolism often involve overexpression and deletion of genes surrounding key central metabolites, such

as pyruvate and acetyl-coA. In the case of metabolic engineering of Escherichia coli for succinate

production, efforts have mainly focused on the manipulation of key pyruvate metabolizing enzymes.

E. coli AFP111 strain lacking ldhA, pflB and ptsG encoded activities accumulates acetate and ethanol as well

as shows poor anaerobic growth on rich and minimal media. To address these issues, we first deleted

genes (adhE, ackA-pta) involved in byproduct formation downstream of acetyl-CoA followed by the

deletion of iclR and pdhR to activate the glyoxylate pathway. Based on data from these studies, we

hypothesized that the succinate productivity was limited by the insufficient ATP generation. Genome-

scale thermodynamics-based flux balance analysis indicated that overexpression of ATP-forming PEPCK

from Actinobacillus succinogenes in an ldhA, pflB and ptsG triple mutant strain could result in an increase in

biomass and succinate flux. Testing of this prediction confirmed that PEPCK overexpression resulted in a

60% increase in biomass and succinate formation in the ldhA, pflB, ptsG mutant strain.

& 2010 Elsevier Inc. All rights reserved.

1. Introduction

Escherichia coli performs mixed acid fermentation in theabsence of exogenous electron acceptors (Neidhardt and Curtiss,1996). The primary fermentation products are acetate, lactate andformate, while ethanol and succinate are formed in minor quan-tities (Gupta and Clark, 1989; Matjan et al., 1989). Pyruvate servesas an important branch point for carbon flux distribution (Arita,2004; Fell and Wagner, 2000). In wild-type strains of E. coli growinganaerobically, most of the pyruvate flux is handled by two primaryenzymes of fermentative metabolism—lactate dehydrogenase(ldhA) and pyruvate-formate lyase (pflB); accounting for theproduction of acetate, ethanol, formate and lactate. E. coli

NZN111 (DldhA, DpflB) was created to redirect the carbon fluxtowards the formation of succinic acid. This mutant, however, isincapable of growth on glucose in rich or minimal media underanaerobic conditions (Bunch et al., 1997; Gupta and Clark, 1989;Stols and Donnelly, 1997) and is known to accumulate high levels ofpyruvate and NADH (Vemuri et al., 2002a). Intracellular redoxratios (NADH/NAD+) as high as three times that of the wild-typeE. coli have been observed in NZN111 (Singh et al., 2009). Theinability to synthesize acetyl-coA and/or to regenerate NAD+ via

ll rights reserved.

pyruvate reduction are cited as the likely causes of the growthdefect of this strain (Stols and Donnelly, 1997; Stols et al., 1997).

A variety of metabolic engineering strategies including thosefocused toward the overexpression of pyruvate metabolizingenzymes have been pursued for improving succinate production inE. coli (Chatterjee et al., 2001; Millard et al., 1996; Sanchez et al., 2005;Stols and Donnelly, 1997; Vemuri et al., 2002a). Other strategiesinclude providing additional reducing power (Chatterjee et al., 2001;Hong and Lee, 2002; VanderWerf et al., 1997), as well as creativecombinations of aerobic and anaerobic metabolism (Sanchez et al.,2006; Vemuri et al., 2002a, b) to attain optimal succinate yields onglucose by activating the glyoxylate pathway and the reductive TCAcycle (Vemuri et al., 2002a). Evolutionary engineering strategies havealso been successful. By employing a dual metabolic engineering andevolutionary approach, Jantama et al. (2008) constructed an E. coli

strain capable of overproducing succinic acid on minimal media insingle batch fermentation. A spontaneous ptsG mutant of NZN111,with improved fitness and restored fermentation capability, wasreported by Donnelly et al. (1998) as E. coli AFP111. Inactivation ofptsG is thought to increase the PEP pool, which is then diverted to thereductive TCA cycle via the action of PEP-carboxylase (PPC)(Chatterjee et al., 2001; Lin et al., 2005; Vemuri et al., 2002a). Theredistribution of the carbon flux resulting from the ptsG mutation inAFP111, however, came at the cost of succinate yield. AFP111produces succinate, ethanol and acetate in a 1:0.5:0.5 ratios, pre-sumably to maintain redox cofactor balances (Chatterjee et al., 2001;Donnelly et al., 1998). The succinate yield and productivity was

A. Singh et al. / Metabolic Engineering 13 (2011) 76–81 77

increased by the overexpression of the anaplerotic pyruvate carbox-ylase (pyc) gene from Rhizobium etli in AFP111, although the formationof acetate and ethanol could not be completely eliminated (Vemuriet al., 2002a, b). In the studies reported herein, we observed thatdeletion of the acetate forming enzymes acetate kinase and phospho-transacetylase (ackA-pta) completely abolished growth in the initialmicroaerobic phase in microaerobic–anaerobic dual-phase fermenta-tion in our AFP111 equivalent strain AD32 (a pflB, ldhA, ptsG triplemutant). PTA mutants have widely been reported to not grow underanaerobic conditions and show reduced growth rates in aerobicconditions (Wolfe, 2005). Formation of acetate also results in ATPgeneration, which we suspected could be limiting anaerobic growth inthis strain as the formation of succinate via reductive TCA cycle doesnot generate any ATP.

PEP is converted to OAA as the first step in succinate production. InE. coli, this step is catalyzed by PEP-carboxylase (ppc), while inActinobacillus succinogenes, this step is catalyzed by the ATP-generatingPEP-carboxykinase (Kim et al., 2004; Laivenieks et al., 1997;VanderWerf et al., 1997). Prior studies have shown that PEP-carbox-ylase overexpression can significantly enhance succinate production inE. coli (Millard et al., 1996), while PEP-carboxykinase overexpressiononly enhances succinate production in an E. coli ppc mutant(Kim et al.,2004; Millard et al., 1996). These prior studies suggested that this isexplained by the roughly 100-fold lower Km towards bicarbonate of thePEP-carboxylase relative to PEP-carboxykinase enzyme. These resultsalso suggested that the PEP to OAA reaction can operate under thephysiological conditions in E. coli, thus promoting this strategy for theuse in succinate production strains. Here, we further assessed thefeasibility of this reaction using Thermodynamics-based flux balanceanalysis (TFBA) with the iAF1260 E. coli genome-scale metabolic model(Feist et al., 2007b). The model confirmed the reaction could proceed inthe ATP-generating direction at elevated extracellular CO2 concentra-tions. The model also predicted higher biomass and succinate yields asa result of increased ATP formation. Based on these data, weinvestigated the effect of PEPCK overexpression in E. coli strainsengineered for succinic acid production. Overexpression of PEPCKwas indeed observed to improve growth, glucose consumption, andsuccinic acid production in an ldhA, pflB and ptsG triple mutant strain.To further improve succinate yield, we deleted the ethanol formingalcohol dehydrogenase (adhE) enzyme and observed an increase insuccinate yield. However, byproduct formation could not be comple-tely eliminated as the deletion of ackA-pta completely abolishedmicroaerobic growth.

2. Materials and methods

2.1. Strains and plasmids

All mutations were made in E. coli BW25113 (D(araD-araB)567,DlacZ4787(::rrnB-3), lambda� , rph-1, D(rhaD�rhaB)568, hsdR514).Deletion strains were constructed following the method developedearlier (Datsenko and Wanner, 2000). The kanamycin resistancecassette was amplified from plasmid pKD13 by PCR using primerswith flanking homologous regions for the target gene. The purified PCRproduct was electroporated into host E. coli strain harboring l-Redrecombinase induced off the plasmid pKD46 using 10 mM arabinose.The resulting kanamycin resistant colonies were screened for thedesired gene knockout by the PCR amplification and subsequentsequencing. Primers for this confirmation step were designed to bind300–400 by upstream and downstream respectively of the target gene.Plasmid pCP20 carrying the FLP-recombinase was subsequently usedto excise the kanamcin selection marker from the knockouts strain. Allplasmids were cured by propagating the strains at 43 1C beforepreparing for the deletion of the next target gene. Strains and plasmidstocks were obtained from the E. coli Genetic Stock Center at Yale

University, New Haven, CT. Plasmid pAsPCK carrying the PEPCK genefrom A. succinogenes under the control of the inducible lac promoterwas provided by the Dr. C. Vielle’s laboratory at the Michigan StateUniversity (MSU), Lansing, MI. E.coli phosphoenol-carboxylase (ppc)was PCR amplified and cloned under the control of its natural promoterinto a low copy vector pSmart-LC-Kan from Lucigen Corp. The ASKAclone plasmid carrying the E.coli PPC under the control of inducible lac

promoter (JW3928) was also utilized for the overexpression studies(Kitagawa et al., 2005).

2.2. Microaerobic–anaerobic dual-phase fermentation

Single bacterial colonies were grown in a 15 ml plastic tubecontaining 5 ml of Luria-Bertani (LB) medium until it reached itsexponential growth at an OD600 nm of 0.4–0.6. The culture wasthen diluted to a new OD600 nm of 0.005–0.01 to be used as aninoculum for the fermentation studies in LB medium supplementedwith 50 mM of sugar source. Glucose, sorbitol and gluconate werepurchased from the Fisher scientific. 10 g/L NaHCO3 was added tothe media for the dual purpose of carbon dioxide source andbuffering agent. The cells were grown in the absence of airheadspace by filling the culturing tubes with the media to thetop to enable rapid anaerobiosis. The cultures were grown at 37 1Cand at an agitation rate of 200 rpm. Separate tubes were run foreach data point to avoid oxygen exposure during sampling.

2.3. Analytical techniques

Cell growth was monitored by measuring the optical density(OD) at 600 nm (UV–vis spectrophotometer, Shimadzu Corp.).To analyze the fermentation culture, 1 mL aliquots were centri-fuged at 13000 rpm for 2 min and the supernatant was filteredthrough a 0.2 mm syringe filter. Samples were analyzed using HPLCAgilent 1100 series Chemstation equipped with ICSep Coregel 64HHPLC column (Transgenomics, Omaha, NE). 50 ml of sample wasloaded into the column operated at 50 1C and ran isocratically with8 mM H2SO4 sat a flow rate of 0.6 ml/min. In the analysis, theconcentrations of organic acids such as succinate, lactate, acetateand formate were quantified from the signals obtained at 210 nmby the UV–vis detector (Eiteman and Chastain, 1997). The con-centrations of the metabolites were determined using a standardcurve constructed using HPLC grade reagents purchased fromSigma-Aldrich Co. Glucose concentrations were monitored by YSI2700 Select Biochemistry Analyzer.

2.4. Thermodynamics-based flux balance analysis using iAF1260

E. coli genome-scale metabolic model

The latest genome-scale E. coli metabolic model, iAF1260 (Feistet al., 2007a) was used to perform the analysis reported here. Thismodel was defined as the wild-type (WT), while the AD32 strainwas simulated by constraining the reactions for ldhA, pflB, ptsG tozero. Thermodynamics-based Flux Balance Analysis (TFBA) gen-erates thermodynamically feasible flux distributions in a givenmetabolic network as compared with conventional Flux BalanceAnalysis (FBA) which can produce distributions that violatethermodynamics (Henry et al., 2007). A minimal set of metaboliteswere allowed to be exported out of system (in this case: acetate,formate, malate, hydrogen gas, ethanol, lactate and succinate). Thein-silico media was set to glucose minimal media under anaerobicconditions. Model assumptions include (i) maintenance energyremains the same as per aerobic conditions at 8.39 mmol-ATP/gDW hr, (ii) biomass composition for anaerobic conditions simu-lated is the same as per aerobic condition, (iii) glucose uptake wasfixed at 10 mmol/gDW hr for ease of comparison of yield and

Table 1Redox balance and the theoretical succinate and ATP yields for the engineered succinic acid production strains of E. coli.

Strains Relevant deletions PEPCK Theoretical yield (mol / mol glucose) Experimental data (g/L h)

ATP generation Net NADH Growth rate Glucoseconsumption

Succinategeneration

AD12 ldhA, pflB No 0 �2a 0.074 0.012 0.008

AD32 ldhA, pflB, ptsG No 0.5 0b 0.091 0.067 0.056

AD32 ldhA, pflB, ptsG, ppc No 1 0 0.078 0.013 0

AD216 ldhA, pflB, ptsG, ppc Yes 2 �2c 0.093 0.107 0.083

AD483 ldhA, pflB, ptsG, ppc, ackA-pta Yes 2 �2c No growth

AD346 ldhA, pflB, ptsG, ppc, adhE Yes 2 �2c 0.09 0.104 0.08

AD568 ldhA, pflB, ptsG, ppc, adhE, ackA-pta Yes 2 �2c No growth

AD725 ldhA, pflB, ptsG adhE, ack-ptA pdhR, iclR pAsPCK Yes 1.4 0 No growth

a It is assumped that only the succinate pathway is active consistent with the reported data.b 1:0.5:05 ratio for succinate:acetate:ethanol is assumed consistent with the reported experimental data.c It is assumed that the presence of PEPCK would drive the carbon flux solely through the succinate branch.

A. Singh et al. / Metabolic Engineering 13 (2011) 76–8178

(iv) cytoplasmic pH was fixed at 7.2, with periplasmic andextracellular pH fixed at 7.

Fig. 1. Central carbon metabolism of E. coli. Target reactions for the improved

succinic acid production are highlighted in red. (For interpretation of the references

to colour in this figure legend, the reader is referred to the web version of this article).

3. Results and discussion

E. coli fitness is closely tied to its ability to balance reducingequivalents across various pathways while simultaneously gener-ating all required biosynthetic precursors. E. coli strains lacking ldhA

and pflB encoded activities suffer from an inability to synthesizesufficient acetyl-coA, accumulation of pyruvate, and formation ofundesired by-products acetate and ethanol, all of which lead toincomplete flux through the desired succinate pathway (Bunchet al., 1997; Matjan et al., 1989; Stols and Donnelly, 1997; Vemuriet al., 2002a). An additional mutation in the ptsG restores fermen-tative growth on glucose in complex media. This strain producessuccinate, acetate and ethanol in a molar ratio of 1:0.5:0.5. It isthought that pyruvate dehydrogenase complex maintains a lowlevel of activity under laboratory anaerobic conditions, thusenabling the conversion of pyruvate to acetyl-CoA. Ethanol is thenproduced via adhE with acetate arising from ackA-pta. With a ratio of1:0.5:0.5, this pathway is NADH balanced and nets 0.5 moles ofATP/mol of glucose consumed (Table 1). To maximize succinateyield, it is essential to remove byproduct formation; however, soleproduction of succinate through the reductive TCA pathway isneither NADH balanced nor does it lead to ATP generation (Fig. 1).Thus, removal of byproduct formation requires a complementarystrategy that yields ATP. Our strategy was to the use theATP-generating PEPCK enzyme in combination with deletions inethanol and acetate forming pathways to improve succinateproduction. Using TFBA, conversion of PEP to OAA and ATP byPEPCK enzyme was found to thermodynamically feasible at extra-cellular CO2 concentrations greater than 1.96 mM. Moreover, themodel also predicted a 15–20% increase in succinate productivityand yield. The fermentation media was thus supplemented with10 g/L sodium bicarbonate to favor the PEPCK reaction.

3.1. ATP production limits fermentative growth of E. coli AD32 (DldhA,

DpflB, DptsG)

Despite a positive overall ATP balance (Table 1), the deletion ofeither adhE or ackA-pta (or both; AD483) was observed to com-pletely abolish fermentative growth of AD32 (BW25113 ldhA� ,pflB� , ptsG�) (Fig. 2a). This result suggests that approximately0.5 mol ATP/mol glucose is required for maintenance under theseconditions and in this genetic background. To test this inference, wedeleted the ppc gene required for anaerobic succinate production.

This strain has the ability to be redox balanced as well as ATPpositive, which is not possible for the succinate production strainalone. As expected, we observed that the growth was maintained inthe Dppc strain albeit at a lower final cell density (Fig. 2a).

Fig. 2. (a) Effect of various gene deletions on growth profile on LB media

supplemented with 50 mM glucose in microaerobic–anaerobic conditions.

(b) Effect of the carbon sources of differing oxidation state on the growth of AD

12 (BW25113 ldhA� , pflB�) and AD32 (BW25113 ldhA� , pflB� , ptsG�) after 24 h on

LB media supplemented with 50 mM glucose.

Fig. 3. (a) Effect of pAsPCK on growth and fermentation profile in the microaerobic–

anaerobic dual-phase fermentation. (a) Open and closed squares represent anae-

robic growth profiles for AD32 (BW25113 ldhA� , pflB� , ptsG�) and AD216 (AD32

pAsPCK) in LB+10 g/L glucose, respectively. (b) Glucose consumption (squares),

succinate (triangles) and acetate (circles) production for AD32 (open symbols) and

AD216 (closed symbols).

A. Singh et al. / Metabolic Engineering 13 (2011) 76–81 79

To further explore the interrelationship between redox and ATPbalancing, we tested the growth of these mutants on sorbitol(Oxidation state¼�1), glucose (Ox state¼0) and gluconate(Ox state¼1). Biochemically, sorbitol metabolism results in3 mol/mol NADH compared with 2 mol/mol for glucose and1 mol/mol for gluconate, respectively. The substitution of glucosewith sorbitol lead to a 44% decrease in cell growth for AD32(BW25113 ldhA� , pflB� , ptsG�) as opposed to AD12 (BW25113ldhA� , pflB�) which grew to over four fold higher cell density(Fig. 2b). Since succinate appears to be the primary fermentativepathway in AD12, excess NADH would lead to redox balancing andhigher growth. The cause of lower cell growth of AD32 on sorbitol isunclear; we speculate that excess NADH would upset the fluxdistribution among the three end-products succinate, acetate andethanol. Higher cell density of AD12 in sorbitol also suggests that itsgrowth is not limited by acetyl-coA requirement for biosynthesis.

3.2. Effect of the overexpression of A. succinogenes PEPCK

Our observations suggest that at least 0.5 moles of ATP per moleof glucose is required for the growth of AD32, which is apparentlyaccomplished through conversion of 50% of the carbon to acetate

and ethanol as opposed to succinate. We reasoned that over-expression of an enzyme that could tie ATP production to succinateformation would increase the fitness of succinate productionstrains. PEP-carboxykinase (PEPCK) from A. succinogenes, whichcatalyzes the formation of oxaloacetate (OAA) from PEP along withthe generation of an ATP, has been previously cloned into anexpression plasmid (pAsPCK) and shown to complement the anE.coli K12 PEP-carboxylase (Dppc) mutant (Kim et al., 2004). Thetransformation of the pAsPCK in AD12 (BW25113 ldhA� , pflB�) andAD32 (BW25113 ldhA� , pflB� , ptsG�) did not significantly increasecell growth (data not shown), presumably due to high activity ofthe native phoephoenolpyruvate carboxylase (ppc), as previouslyreported (Kim et al., 2004; Millard et al., 1996). It has been shownthat the kinetic properties of the E. coli PPC are superior to those ofPEPCK with respect to bicarbonate. Specifically, E. coli PPC has a Km

of 0.19 and 0.1 towards PEP and bicarbonate anion, respectively,(Kai et al., 1999). In comparison, the Km towards bicarbonate forPEPCK is expected to be closer to that of PEP carboxykinases fromE. coli (13 mM) and Anaerobiospirillum succiniciproducens, whichhave been measured at about two orders of magnitude greater(17 mM) than E. coli PPC (Millard et al., 1996).

To address this issue, we deleted the native E. coli ppc andtransformed the pAsPCK plasmid into strain AD32 (designatedAD216). Indeed, we observed that overexpression of A. succinogenes

PEPCK in the absence of the native ppc gene increased growth and

Fig. 4. Growth of AD32 (BW25113 ldhA� , pflB� , ptsG�) derivatives harboring pAsPCK plasmid on LB media supplemented with 50 mM glucose after 24 h in microaerobic–

anaerobic conditions.

A. Singh et al. / Metabolic Engineering 13 (2011) 76–8180

succinate production in the microaerobic–anaerobic dual-phasefermentation (Fig. 3a). AD216 consumed 50% more glucose in72 h and produced 60% more succinate (5.97 g/L) titers comparedto AD32 (Fig. 3b) although the yield remained roughly the sameat 0.78 (70% of the theoretical maximum, note that this improve-ment is significant as AD32 has the same disrupted activity inldhA, pflB, and ptsG as previously described AFP111). Acetateconstituted 20% of the fermentation product. This observationsuggests that ATP limits cell growth but not the flux distribution inthis strain, as would be expected based on the redox and ATPbalance.

3.3. Efforts to increase theoretical yield of succinic acid production

We expected that diverting the flux towards succinate forma-tion via ATP-forming PEPCK would eliminate the need for the ATPproducing acetate pathway, and thus could be used to furtherincrease succinate formation. The effect of the presence of PEPCKwas thus studied in strains lacking ethanol and acetate pathways.As expected, the presence of PEPCK had no effect on the strainscarrying a native PPC, thus the studies were continued in a ppc

deletion background. The strain carrying the adhE deletion, AD346was able to grow anaerobically upon transformation with pAsPCK(Fig. 4). The growth of AD346 was similar to AD32, thus channelingthe carbon flux towards succinate formation could eliminate theneed for NADH oxidation via the ethanol pathway. However, thedeletion of acetate forming genes, ackA-pta in AD483 abolishedmicroaerobic and anaerobic growth even in the presence of pAsPCKplasmid (Fig. 4). As the derivatives of AD32 (BW25113 ldhA� , pflB� ,ptsG�) appear to have pyruvate to acetyl-coA conversion activity,we suspected that acetate formation would be essential in suchstrains in the absence of a functional glyoxylate pathway to avoidacetyl-coA accumulation.

A 71:29 ratio split in the carbon flux towards reductive TCAcycle and the glyoxylate pathway respectively would lead to redoxbalanced succinate production with a molar yield of 1.71 mol/molglucose (Vemuri et al., 2002a). Since the glyoxylate pathway istypically active during aerobic conditions, optimized dual-phaseaerobic/anaerobic fermentations have been successfully employed

to attain high yields of succinate formation at impressive produc-tivities (Sanchez et al., 2006; Vemuri et al., 2002b). With thesestudies in mind, we sought to investigate if the deletions of pdhR

and iclR would increase cellular fitness of our engineered strains.pdhR is the anaerobic repressor of the pyruvate dehydrogenasecomplex while iclR is the repressor of the glyoxylate pathwaygenes. The deletions of pdhR and iclR leads to the constitutiveexpression of the PDHc (Haydon et al., 1993) and the glyoxylatepathways genes (Sunnarborg et al., 1990), respectively. However,these two deletions had no effect on the fitness on the AD32 and itsderivatives, suggesting that neither the conversion of pyruvate toacetyl-CoA nor the glyoxylate pathway is limiting in our strains.

In summary, cellular fitness of the succinate production strainsappears to be limited by sub-optimal acetyl-coA metabolism.While the formation of acetyl-coA from pyruvate is essential forbiosynthesis, subsequent formation of acetate decreases the over-all yield. The engineered strains were observed to be sensitive toackA-pta mutation, underscoring the rigid control of acetyl-coAmetabolism. Future engineering efforts should focus on modulat-ing the pyruvate to acetyl-coA conversion to enable biosynthesis aswell as achieve the delicate split of the carbon flux between thereductive TCA and glyoxylate pathways.

Acknowledgments

KCS was supported by the Swiss National Science Foundation.VH was supported by funding from Ecole Polytechnique Federalede Lausanne (EPFL), SystemsX.ch and DuPont.

References

Arita, M., 2004. The metabolic world of Escherichia coli is not small. Proceedings ofthe National Academy of Sciences of the United States of America 101,1543–1547.

Bunch, P.K., MatJan, F., Lee, N., Clark, D.P., 1997. The IdhA gene encoding thefermentative lactate dehydrogenase of Escherichia coli. Microbiology-UK 143,187–195.

Chatterjee, R., Millard, C.S., Champion, K., Clark, D.P., Donnelly, M.I., 2001. Mutationof the ptsC gene results in increased production of succinate in fermentation ofglucose by Escherichia coli. Applied and Environmental Microbiology 67,148–154.

A. Singh et al. / Metabolic Engineering 13 (2011) 76–81 81

Datsenko, K.A., Wanner, B.L., 2000. One-step inactivation of chromosomal genes inEscherichia coli K-12 using PCR products. Proceedings of the national academy ofsciences of the United States of America 97, 6640–6645.

Donnelly, M.I., Millard, C.S., Clark, D.P., Chen, M.J., Rathke, J.W., 1998. A novelfermentation pathway in an Escherichia coli mutant producing succinic acid,acetic acid, and ethanol. Applied Biochemistry and Biotechnology 70-2,187–198.

Eiteman, M.A., Chastain, M.J., 1997. Optimization of the ion-exchange analysis oforganic acids from fermentation. Analytica Chimica Acta 338, 69–75.

Feist, A., Henry, C., Reed, J., Krummenacker, M., Joyce, A., Karp, P., Broadbelt, L.,Hatzimanikatis, V., Palsson, B., 2007a. A genome-scale metabolic reconstructionfor Escherichia coli K-12 MG1655 that accounts for 1260 ORFs and thermo-dynamic information. Molecular Systems Biology 3, 121.

Feist, A.M., Henry, C.S., Reed, J.L., Krummenacker, M., Joyce, A.R., Karp, P.D.,Broadbelt, L.J., Hatzimanikatis, V., Palsson, B.O., 2007b. A genome-scale meta-bolic reconstruction for Escherichia coli K-12 MG1655 that accounts for 1260ORFs and thermodynamic information. Molecular Systems Biology 3, 121.

Fell, D.A., Wagner, A., 2000. The small world of metabolism. Nature Biotechnology18, 1121–1122.

Gupta, S., Clark, D.P., 1989. Escherichia Coli derivatives lacking both alcohol-dehydrogenase and phosphotransacetylase grow anaerobically by lactatefermentation. Journal of Bacteriology 171, 3650–3655.

Haydon, D.J., Quail, M.A., Guest, J.R., 1993. A mutation causing constitutive synthesisof the pyruvate-dehydrogenase complex in Escherichia Coli is located within thePdhr gene. FEBS Letters 336, 43–47.

Henry, C., Broadbelt, L.J., Hatzimanikatis, V., 2007. Thermodynamics-based meta-bolic flux analysis. Biophysical Journal 92, 1792–1805.

Hong, S.H., Lee, S.Y., 2002. Importance of redox balance on the production of succinicacid by metabolically engineered Escherichia coli. Applied Microbiology andBiotechnology 58, 286–290.

Jantama, K., Haupt, M.J., Svoronos, S.A., Zhang, X.L., Moore, J.C., Shanmugam, K.T.,Ingram, L.O., 2008. Combining metabolic engineering and metabolic evolutionto develop nonrecombinant strains of Escherichia coli C that produce succinateand malate. Biotechnology and Bioengineering 99, 1140–1153.

Kai, Y., Matsumura, H., Inoue, T., Terada, K., Nagara, Y., Yoshinaga, T., Kihara, A.,Tsumura, K., Izui, K., 1999. Three-dimensional structure of phosphoenolpyr-uvate carboxylase: A proposed mechanism for allosteric inhibition. Proceedingsof the National Academy of Sciences of the United States of America 96,823–828.

Kim, P., Laivenieks, M., Vieille, C., Zeikus, J.G., 2004. Effect of overexpression ofActinobacillus succinogenes phosphoenolpyruvate carboxykinase on succinateproduction in Escherichia coli. Applied and Environmental Microbiology 70,1238–1241.

Kitagawa, M., Ara, T., Arifuzzaman, M., Ioka-Nakamichi, T., Inamoto, E., Toyonaga, H.,Mori, H., 2005. Complete set of ORF clones of Escherichia coli ASKA library

(A complete Set of E. coli K-12 ORF archive): unique resources for biologicalresearch. DNA Research 12, 291–299.

Laivenieks, M., Vieille, C., Zeikus, J.G., 1997. Cloning, sequencing and overexpressionof the Anaerobiospirillum succiniciproducens phosphoenolpyruvate carboxyki-nase (pckA) gene. Applied and Environmental Microbiology 63, 2273–2280.

Lin, H., Bennett, G.N., San, K.Y., 2005. Effect of carbon sources differing in oxidationstate and transport route on succinate production in metabolically engineeredEscherichia coli. Journal of Industrial Microbiology & Biotechnology 32, 87–93.

Matjan, F., Alam, K.Y., Clark, D.P., 1989. Mutants of Escherichia coli deficient in thefermentative lactate-dehydrogenase. Journal of Bacteriology 171, 342–348.

Millard, C.S., Chao, Y.P., Liao, J.C., Donnelly, M.I., 1996. Enhanced production ofsuccinic acid by overexpression of phosphoenolpyruvate carboxylase in Escher-ichia coli. Applied and Environmental Microbiology 62, 1808–1810.

Neidhardt, F.C., and Curtiss, R. (1996). Escherichia coli and Salmonella: cellular andmolecular biology, 2nd ed. (Washington, D.C., ASM Press).

Sanchez, A.M., Bennett, G.N., San, K.Y., 2005. Novel pathway engineering design ofthe anaerobic central metabolic pathway in Escherichia coli to increase succinateyield and productivity. Metabolic Engineering 7, 229–239.

Sanchez, A.M., Bennett, G.N., San, K.Y., 2006. Batch culture characterization andmetabolic flux analysis of succinate-producing Escherichia coli strains. MetabolicEngineering 8, 209–226.

Singh, A., Lynch, M.D., Gill, R.T., 2009. Genes restoring redox balance in fermenta-tion-deficient E. coli NZN111. Metabolic Engineering 11, 347–354.

Stols, L., Donnelly, M.I., 1997. Production of succinic acid through overexpression ofNAD(+)-dependent malic enzyme in an Escherichia coli mutant. Applied andEnvironmental Microbiology 63, 2695–2701.

Stols, L., Kulkarni, G., Harris, B.G., Donnelly, M.I., 1997. Expression of Ascaris suummalic enzyme in a mutant Escherichia coli allows production of succinic acidfrom glucose. Applied Biochemistry and Biotechnology 63-5, 153–158.

Sunnarborg, A., Klumpp, D., Chung, T., Laporte, D.C., 1990. Regulation of theglyoxylate bypass operon—cloning and characterization of Iclr. Journal ofBacteriology 172, 2642–2649.

VanderWerf, M.J., Guettler, M.V., Jain, M.K., Zeikus, J.G., 1997. Environmental andphysiological factors affecting the succinate product ratio during carbohydratefermentation by Actinobacillus sp. 130Z. Archives of Microbiology 167,332–342.

Vemuri, G.N., Eiteman, M.A., Altman, E., 2002a. Effects of growth mode and pyruvatecarboxylase on succinic acid production by metabolically engineered strains ofEscherichia coli. Applied and Environmental Microbiology 68, 1715–1727.

Vemuri, G.N., Eiteman, M.A., Altman, E., 2002b. Succinate production in dual-phaseEscherichia coli fermentations depends on the time of transition from aerobic toanaerobic conditions. Journal of Industrial Microbiology & Biotechnology 28,325–332.

Wolfe, A.J., 2005. The acetate switch. Microbiology and Molecular Biology Reviews69, 12–50.