Metabolic Engineering 12 (2010) 499–509

Contents lists available at ScienceDirect

Metabolic Engineering

1096-71

doi:10.1

n Corr

E-m1 Cu

Chile.

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

Metabolic impact of the level of aeration during cell growth on anaerobicsuccinate production by an engineered Escherichia coli strain

Irene Martınez a,1, George N. Bennett b, Ka-Yiu San a,n

a Department of Bioengineering, Rice University, Houston, TX, USAb Department of Biochemistry and Cell Biology, Rice University, Houston, TX, USA

a r t i c l e i n f o

Article history:

Received 9 May 2010

Received in revised form

7 September 2010

Accepted 21 September 2010Available online 29 September 2010

Keywords:

Succinate

Aeration effect

E. coli

Transition

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

016/j.ymben.2010.09.002

esponding author.

ail address: ksan@rice.edu (K.-Y. San).

rrent address: Pontificia Universidad Catolic

a b s t r a c t

The metabolic impact of two different aeration conditions during the growth phase on anaerobic

succinate production by the high succinate producer Escherichia coli SBS550MG (pHL413) was

investigated. Gene expression profiles, metabolites concentrations and metabolic fluxes were analyzed.

Different oxygen levels are known to induce or repress transcription, synthesis of different enzymes, or

both, affecting cell metabolism and thus product yield and productivity. The succinate yield was 1.55

and 1.25 mol succinate/mol glucose, and the productivity was 1.3 and 0.9 g L�1 h�1) for the low

aeration experiment and high aeration experiment, respectively. Changes in the level of aeration during

the cells growth phase significantly modified gene expression profiles and metabolic fluxes in this

system. Pyruvate was accumulated during the anaerobic phase in the high aeration experiment, which

could be explained by a lower pflAB expression during the transition time and a lower flux towards

acetyl-CoA during the anaerobic phase compared to the low aeration case. The higher PflAB flux and the

higher expression of genes related to the glyoxylate shunt (aceA, aceB, acnA, acnB) during the transition

time, anaerobic phase, or both, improved succinate yield in the low aeration case, allowing the system

to attain the maximum theoretical succinate yield for E. coli SBS550MG (pHL413).

& 2010 Elsevier Inc. All rights reserved.

1. Introduction

Facultative organisms are able to grow under aerobic andanaerobic conditions by changing their cell physiology andmetabolic pathways to adapt to the new environment. Changesin cell metabolism are controlled by sensing and regulatorysystems that sense oxygen levels and transmit a signal to modifygene expression accordingly (Sawers, 1999). Escherichia coli hasseveral sensing mechanisms, among them the Fnr and Arcregulator systems have been studied (Becker et al., 1996; Shalel-Levanon et al., 2005a, 2005b; Spiro and Guest, 1990). In anaerobicconditions E. coli metabolism generates several fermentationproducts such as acetate, lactate, formate, ethanol and succinatein various yields. Efforts to genetically modify E. coli and carefullycontrol growth conditions to generate high production of a singleproduct have been subjects of many investigations in appliedmicrobiology.

Succinate is widely used in industry as additive in food andpharmaceuticals, as well as a precursor of biodegradable poly-mers, surfactants, synthetic resins, among other uses (Hong and

ll rights reserved.

a de Valparaıso, Valparaıso,

Lee, 2002; Lee et al., 2004; Zeikus et al., 1999). E. coli does notnaturally accumulate succinate; its natural production is ratherlow (maximum theoretical yield of 1 mol succinate per mol ofglucose). In the past few years, several engineered E. coli strainshave been created to produce succinate from glucose at variousyields. Some approaches include the deletion of genes encodingenzymes involved in competing pathways such as LdhA (Jantamaet al., 2008b; Mat-Jan et al., 1989; Sanchez et al., 2005b), AdhE(Jantama et al., 2008b; Sanchez et al., 2005b), PflAB (Jantamaet al., 2008b) and AckA-Pta (Jantama et al., 2008b; Sanchez et al.,2005b); the overexpression of enzymes that channel the carbontowards succinate through the formation of oxaloacetate, such asPpc and Pyc (Gokarn et al., 1998, 2000; Lin et al., 2005; Millardet al., 1996; Vemuri et al., 2002a; Wang et al., 2006), or throughmalate formation by the malic enzyme (Hong and Lee, 2001;Kwon et al., 2007; Stols and Donnelly, 1997; Stols et al., 1997). AnE. coli strain with a deletion in the isocitrate lyase repressor gene(iclR) (Sanchez et al., 2005b) and an Aspergillus niger strainoverexpressing the isocitrate lyase gene (icl) have also beenconstructed to potentially increase the flux through the glyoxylateshunt and therefore increase succinate production. However, iclR

deletion in E. coli did not show significant difference in succinateproduction and icl overexpression in A. niger produced an increasein fumarate instead of succinate (Meijer et al., 2009). On the otherhand, the deletion of transporter related genes, such as theglucose-specific permease of the phosphotransferase system, Ptsg

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509500

(Chatterjee et al., 2001; Donnelly et al., 1998) and the formatetransporter, FocA (Jantama et al., 2008a) have been performed toincrease succinate production.

The strain used in this work was SBS550MG (pHL413), createdby Sanchez and collaborators (Sanchez et al., 2005b). This strainhas the following genes deleted adhE, ldhA, iclR and ackA-pta andoverexpresses the gene pyc encoding for pyruvate carboxylase fromLactococcus lactis. This strain has shown to produce succinate athigh yields, 1.6 mol succinate/mol glucose in a dual-phase system;where the first phase was aerobic and the cells were grown inshake flasks followed by an anaerobic phase where succinate wasproduced in shake flasks or batch bioreactor (Sanchez et al., 2005b,2006). This yield value is also the theoretical maximum for thisstrain (Cox et al., 2006; Sanchez et al., 2006). In this study wedeveloped the entire process for succinate production by E. coli

strain SBS550MG (pHL413) in a bioreactor; where the strain wasgrown aerobically and then the condition was switched toanaerobic in the same bioreactor for succinate production. Thecentrifugation step was then eliminated saving time and resources,which is desirable for scale up purposes.

The physiological state of the cells is critical in a dual-phasesystem where the enzymes produced during the first phase(aerobic-growth phase) will influence the performance of the cellsin the anaerobic-production phase. In the present study, weanalyzed the metabolic effect of two different aeration conditionsduring the growth phase on metabolite and gene expressionprofiles, as well as the metabolic fluxes, during the productionphase to better understand the differences in cell metabolism thatoriginate differences in product distribution and, elucidatepotential bottle necks and possible strain improvements.

2. Material and methods

2.1. Strain

The E. coli strain SBS550MG (pHL413) was used for theexperiments. This strain contains the following gene deletions:adhE ldhA iclR ackpta::Cm (Sanchez et al., 2005b) and harbors thepHL413 plasmid containing the pyruvate carboxylase gene (pyc)from L. Lactis in pTrc99A, ApR (Lin et al., 2004).

2.2. Batch cultures

The fermenter system used was a 1.0 L Bioflo 110 fermenter (NewBrunswick Scientific) equipped with a glass pH electrode and apolarographic dissolved oxygen (DO) electrode Ingold InPro 6000(Mettler Toledo) to monitor and control pH and dissolvedoxygen conditions, respectively. The medium used for the aerobicphase contained 20 g/L tryptone, 10 g/L yeast extract, 0.9 g/LK2HPO4�3H2O, 1.1 g/L KH2PO4, 3 g/L (NH4)2SO4, 0.5 g/L MgSO4�

7 H2O, 0.25 g/L CaCl2�H2O, 200 mg/L ampicillin, 30 mL/L antifoamsigma #204, 1 mg/L thiamine, 1 mg/L biotin, 2 g/L glucose. Another2 g/L of glucose were added when the initial glucose was depleted.No pH control was applied during aerobic growth. After theadditional glucose was depleted and the pH increased to 7.4–7.5,the air flow was stopped and a 0.2 L/min CO2 flow was established.At this time 20 g/L glucose were added to the culture. Sampleswere taken at different times and their metabolite concentrationswere determined by HPLC.

2.3. Extracellular metabolites analysis

Extracellular metabolites such as glucose, succinate, formateand pyruvate among others were analyzed by HPLC as describer

earlier (Sanchez et al., 2005b). In brief, the HPLC system(Shimadzu-10A System, Shimadzu, Columbia, MD, USA) wasequipped with a cation-exchange column (HPX-87H, BioradLabs, Hercules, CA, USA), a differential refractive index (RI)detector (Waters 2410, Waters, Milford, MA, USA) and anultraviolet (UV) detector (Shimadzu SPD-10A). The mobile phaseused was a 2.5 mM H2SO4 solution at a 0.6 mL/min flow rate. Thecolumn was operated at 55 1C.

2.4. Metabolic fluxes analysis

The metabolic network considered for the analysis, wasadapted from the network described by Sanchez and collaborators(Sanchez et al., 2006). In brief, 17 fluxes were considered, v1–v17,including glycolysis, glyoxylate shunt, anaerobic fermentationreactions, heterologous conversion of pyruvate into oxaloacetate,catalyzed by the L. lactis pyruvate carboxylase (Pyc), and malate,succinate and pyruvate excretion. The redox balance for NADHwas also included assuming no NADH accumulation (Sanchezet al., 2006). E. coli strain SBS550MG (pHL413) did not showgrowth during the anaerobic phase, thus the flux to biomass (v2)was assumed to be zero. In the present work, the high aerationexperiment showed pyruvate accumulation; then pyruvateexcretion was also included in the metabolic flux analysis. Thestoichiometric matrix including the above fluxes and the NADHbalance was based on the pseudo-steady-state hypothesis (PSSH)for the intracellular intermediate metabolites and the law of massconservation (Stephanopoulos et al., 1998). The description of themetabolic flux determination is presented in Appendix A.

2.5. Gene expression analysis

2.5.1. RNA preparation

Total RNA was isolated using the RNeasy mini kit (Qiagen,Valencia, CA, USA) according to the manufacturer’s protocol.Samples that were not extracted immediately were treated withProtect reagent from Qiagen (Valencia, CA, USA), frozen usingliquid nitrogen and stored at �80 1C until extraction. The isolatedRNA was treated with DNaseI (Promega, Madison, WI, USA) andRNase Inhibitor (Promega, Madison, WI, USA) according tomanufacturer protocol. The reaction was incubated at roomtemperature for 15 min. The RNA was then extracted withPhenol/CIA extract solution (CIA¼24:1 chloroform:isoamyl alco-hol) once and then with chloroform once. The RNA was mixedwith 1/10 volume of 3 M NaCl and 2 volumes of ethanol andincubated at �80 1C for 20 min. Then, the mixture was centri-fuged and washed with 75% ethanol (Shalel-Levanon et al.,2005a). The precipitated RNA was resuspended in RNase-freewater (Sigma, St. Louis, MO, USA). The concentration of RNA wasquantified by measuring the absorbance at 260 nm and applyingthe formula, concentration (mg/mL)¼A260�40�dilution factor.

2.5.2. cDNA synthesis and quantitative PCR amplification

The cDNA was synthesized using the Promega ReverseTranscriptase System (Promega, Madison, WI, USA) and the RNAextracted as described above. The reaction was carried out in aRoboCycler Gradient 96 (Stratagene, La Jolla, CA, USA). The cDNAwas synthesized in a total reaction mixture volume of 60 mLcontaining 1 mg of RNA as template. The reaction mixture wasincubated for 10 min at room temperature for primer extension,30 min at 50 1C for reverse transcription, and then 5 min at 95 1Cand 5 min at 6 1C for inactivation of the reverse transcriptase.Non-amplification controls were prepared by not adding reversetranscriptase to the mixture. The synthesized cDNA was then

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509 501

diluted 10-fold with nuclease-free water and stored at �20 1Cuntil further use.

Quantitative real-time polymerase chain reaction (PCR) wasperformed in a ABI Prism 7000 Sequence Detection System(Applied Biosystems, Foster City, CA, USA) using the SYBR GreenPCR Master Mix (Applied Biosystems, Foster City, CA, USA). Theincrease in fluorescence is directly related to the formation of PCRproduct, the SYBR Green dye binds to the double-stranded DNAformed during the amplification. Forward and reverse primerswere designed for each gene to generate a PCR product of15072 bp. The list of selected genes for the study and therespective primer sets for the qRT-PCR are shown in Table 1.Reactions were carried out in 96-well plate using a mixture of3 mL diluted cDNA prepared above as template, 2 mL mixedprimers (1.25 pmol/mL), 10 mL SYBR Green PCR Master Mix, and5 mL nuclease-free water. The reaction for each gene wasperformed in quadruplicates and a non-amplification controlwas used.

The reaction mixture was incubated 2 min at 50 1C, 10 min at95 1C for Taq activation, followed by 40 cycles of 15 s at 95 1C(denaturation) and 1 min at 60 1C (annealing/extension).

The comparative CT method, according to manufacturer’sprotocol, was used for relative quantification of gene expression(ABI Prism 7700 Sequence Detection System User Bulletin #2,Applied Biosystems). The rrsA gene encoding 16S ribosomal RNAwas used as a reference gene to standardize the amount oftemplate added to a reaction. The difference between the CT of thegene of interest and the CT of the reference gene (DCT) wascalculated for all the genes in Table 1. The DDCT was calculated bytaking the difference between the DCT of a gene in the lowaeration experiment and the DCT of the same gene in the highaeration condition experiment. The comparative expression levelwas calculated using the formula 2�DDCT as previously described(Shalel-Levanon et al., 2005b). This method was used to comparethe expression levels of the same gene at different time points anddifferent experimental conditions, but cannot be used to comparethe expression levels of different genes.

3. Results and discussion

Succinate production experiments were conducted in a dual-phase system in bioreactors where the first phase was aerobic forbiomass accumulation followed by an anaerobic phase forsuccinate production. The metabolic behavior of the succinateproducer E. coli SBS550MG (pHL413) strain was evaluated undertwo different aeration conditions for the aerobic phase. The pH,DO, extracellular metabolites, metabolic fluxes and gene expres-sion profiles were analyzed. The results and discussion aredescribed below.

3.1. DO and pH profiles

The bioreactor experiments were operated at two differentaeration conditions during the aerobic-growth phase. The highaeration condition comprised an agitation speed of 800 rpm andan air flow of 2.5 L/min, whereas the bioreactor under lowaeration was agitated at 500 rpm and had an air flow of 1.5 L/min.When the 2 g/L glucose present initially were consumed, glucosewas added to 2 g/L final concentration. After this batch of glucosewas depleted and the pH increased to 7.4–7.5, the air flow wasstopped and the system was switched to the production phase,where a CO2 flow (0.2 L/min) was established. At this time, 20 g/Lglucose were added to the culture. The different aerationconditions led to different dissolved oxygen profiles (Fig. 1a).The dissolved oxygen (DO) level decreased considerably faster in

the low aeration case, leading to a microaerobic condition thatwas maintained for about 3 h before the system was switched tocompletely anaerobic. The high aeration experiment neverreached a microaerobic condition before switching to theanaerobic phase. This microaerobic period may have contributedto differences in cell metabolism, reflected as differences inmetabolic profiles, gene expression profiles and final productyield and productivity described below.

The pH was not controlled during the aerobic phase. Instead,the pH was used as an indicator of glucose consumption todetermine glucose feeding times (Ishizaki and Vonktaveesuk,1996). The pH profiles in both experiments were comparable untilabout 7 h into the aerobic phase (Fig. 1b). The pH decreased asglucose was consumed and then rose again as the acids producedby glucose metabolism were utilized by the cells in the absence ofglucose. Once glucose was exhausted and the pH reached 7.4–7.5,glucose was added to 2 g/L final concentration. Glucose wasrapidly consumed and the pH recovered rapidly in the highaeration case, probably due to a higher DO and a higher cellconcentration (Fig. 1b). In the low aeration condition, the pHdecreased to about 6.6 before the acid products were uptaken bythe cells and the pH increased. In both experiments, after the pHreached 7.4–7.5 again, a second batch of glucose was added(to a final concentration of 20 g/L), the pH control was activated(pH 7.0), the air flow was stopped, and a CO2 flow (0.2 L/min) wasestablished to create anaerobic conditions. In the low aerationexperiment the switching to the anaerobic phase was delayed forabout 1 h, the time required for the pH to increase back to 7.4.

The cell growth during the aerobic phase was comparable in thelow and high aeration cases until about 7.5 h in the aerobic phase.After this time, the DO decreased to a very low level in the low aeratedbioreactor, creating a microaerobic environment where the cells grewat a lower rate compared to the high aerated bioreactor probably dueto oxygen limitation (Fig. 2a). At the end of the growth phase, the cellconcentration reached 18.5 and 16.5 OD600 for the high and lowaeration experiments, respectively. Therefore, a higher aerationresulted in a higher cell growth. No cell growth was observed in theanaerobic phase in both cases. This result agrees with Sanchez andcollaborators study where no cell growth was detected during theproduction phase using the same strain (Sanchez et al., 2006).

3.2. Comparison of metabolites profiles

The extracellular metabolite profiles during the anaerobicproduction phase are shown in Fig. 2b and c. More glucose wasconsumed in the low aeration case, 79.6 versus 67.6 mmol in thehigh aeration case. The final succinate titer was 166.9 and115.4 mM for the low and high aeration experiments, respec-tively. This resulted in a yield of 1.55 and 1.25 mol succinate/molglucose, respectively; and a productivity of 1.3 and 0.9 g L�1 h�1,respectively (Fig. 2b). Therefore, the level of aeration during theaerobic phase significantly influences the anaerobic succinateproduction in E. coli SBS550MG (pHL413).

Succinate was the main fermentation product in both aerationcases; although, a small amount of malate was also produced, 3.7and 11.2 mM in the low and high aeration experiment, respec-tively (Fig. 2c). The acetate produced was also higher in the highaeration case (Fig. 2c). Almost no pyruvate accumulation wasobserved in the low aeration case, whereas the pyruvateaccumulation in the high aeration experiment reached 27.0 mM(Fig. 2c). Then, the formation of fermentation side-productsmalate, acetate and pyruvate, was higher in the high aerationexperiment. Pyruvate accumulation has been also observed inE. coli W3110 after a downshift of oxygen level from aerobicto anaerobic at high glucose concentrations (Soini et al., 2008).

Table 1List of genes studied, enzyme reaction and corresponding primers used for qRT-PCR.

Gene name Enzyme name Reaction Primers

GlycolysispykA Pyruvate kinase A PEP+ADP2Pyruvate+ATP pykAa: 50-actgacgctgtgatgctgtc-30

pykF Pyruvate kinase F pykAb: 50-ccacattgtcgaactgaacg-30

pykFa: 50-accgccattgaaggtaacaa-30

pykFb: 50-tgttcgcaaccaaagatcag-30

aceE Pyruvate dehydrogenase multienzyme

complex

Pyruvate+NAD+CoA-Acetyl CoA+CO2+NADH aceEa: 50-acgtaccggctgacgactac-30

aceF aceEb: 50-cttatcgatttcgccacgtt-30

lpd aceFa: 50-aagtgaccgaaatcctggtg-30

aceFb: 50-cactttgtcacccacgttca-30

lpda: 50-cgaaagagaaaggcatcagc-30

lpdb: 50-tagtaccgacaatcgcacca-30

TCA cycle and glyoxylate shuntacnA Aconitase Citrate2Isocitrate acnAa: 50-aaaacgtacgcctggaaatg-30

acnB Aconitase B acnAb: 50-cacactgctttgccgagata-30

acnBa: 50-tggtgatgcagtctttctgc-30

acnBb: 50-catacggttcagccaggagt-30

aceA Isocitrate lyase Isocitrate2Succinate+Glyoxylate aceAa: 50-atctgatcacctccgattgc-30

aceAb: 50-caccagaccaggtcagcata-30

aceB Malate synthase A AcetylCoA+H2O+Glyoxylate2Malate+CoA aceBa: 50-ttctgactgagctggtgacg-30

aceBb: 50-gcgaattttccaatcagcat-30

fumA Fumarase A Malate2Fumarate+H2O fumAa: 50-ggggatctggatcgaaaaac-30

fumB Fumarase B fumAb: 50-agaaacgggatactgcgaca-30

fumC Fumarase C fumBa: 50-gacattcgcgttatccgtct-30

fumBb: 50-gaatgtactggcctgggttg-30

fumCa: 50-tgatccacaatttcctgcaa-30

fumCb: 50-gtgttaagcgcagtcaccag-30

frdA Fumarate reductase A Fumarate+MQH2+2H+-Succinate+MQ frdAa: 50-cgataagaccggcttccata-30

frdB Fumarate reductase B frdAb: 50-ccttccatcatgttcattgct-30

frdC Fumarate reductase C frdBa: 50-gctaacttcccgattgaacg-30

frdD Fumarate reductase D frdBb: 50-tggtgatacttcgccatctg-30

frdCa: 50-acccggttatcgtgatcatt-30

frdCb: 50-ccgcccagagacttttgata-30

frdDa: 50-tgggtattctgctgccact-30

frdDb: 50-cgtggtgcatacggtgtaaa-30

mdh Malate dehydrogenase Malate+NAD2OAA+NADPH mdha: 50-ggcgttagttttaccgagca-30

mdhb: 50-cacgaaccagagacagacca-30

gltA Citrate synthase OAA+AcetylCoA2Citrate gltAa: 50-atgattctttccgcctgatg-30

gltAb: 50-tgttttccagctccatagcc-30

Anapleroticpps Phosphoenolpyruvate synthase Pyruvate+ATP+H2O2PEP+AMP+Pi ppsa: 50-tatcgtgtgcaccagggtta-30

ppsb: 50-tgcggaagtgataaacacca-30

ppc Phosphoenolpyruvate carboxylase PEP+CO22OAA+Pi+H+ ppca: 50-gacgtactggctgtccacct-30

ppcb: 50-atcaggccacgataccagtc-30

pyc Pyruvate carboxylase from L. lactis Pyruvate+CO2+ATP2OAA+ADP+Pi pyca: 50-cggagctcttgaatttgctc-30

pycb: 50-aaggcaccgatagcttctga-30

Anaerobic fermentationpflA Pyruvate formate-lyase activating

enzyme (PFL-activating enzyme)

Activates PflB pflAa: 50-tacgatccggtgattgatga-30

pflAb: 50-tcacatttttgttcgccaga-30

pflB Pyruvate formate-lyase (inactive) Pyruvate+CoA2Acetyl CoA+Formate pflBa: 50-gatctggaaaacggcgtaaa-30

pflBb: 50-gatagcttcctgagcgttgg-30

pflD Formate acyltransferase 2 Pyruvate+CoA2Acetyl CoA+Formate pflDa: 50-cccttactggctgctgaaag-30

pflDb: 50-atctgcccgttgatgaaatc-30

yfiD Glycyl radical protein similar to the

pyruvate formate-lyase subunit, stress-

induced

Pyruvate+CoA2Acetyl CoA+Formate yfiDa: 50-aactctttctggctgctgga-30

yfiDb: 50-acgcgaacttctggtttcac-30

RegulatorsarcA Transcriptional dual regulator arcAa: 50-tgttttcgaagcgacagatg-30

arcAb: 50-gaacatcaacgcaacattcg-30

arcB Sensory histidine kinase, subunit of the

dual transcriptional regulator ArcAB

arcBa: 50-ctggaatggatgatgtgctg-30

arcBb: 50-gcatgggaatatcgagcaat-30

fnr Transcriptional regulator fnra: 50-tgatcctgctgttgtcgaag-30

fnrb: 50-tcaggcccagatagttaccg-30

House-keeping generrsA 16S ribosomal RNA House-keeping gene rrsAa: 50-tgtagcggtgaaatgcgtag-30

rrsAb: 50-cctccaagtcgacatcgttt-30

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509502

High aeration

D E

6.26.46.66.8

77.27.47.6

A

B

C

D

E

F

0

20

40

60

80

100

120

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Time (h)

High aeration Low aeration High aeration

Time (h)

High aeration Low aeration

pH

A B

C FdO

2 (%

)

Fig. 1. Culture parameters and sampling of bioreactor experimentas under different conditions of aeration of strain SBS550MG (pHL413): (a) dissolved oxygen profile and

(b) pH profile for high and low aeration bioreactor experiments. Samples were analyzed by qRT-PCR at three time points for each experiment. The arrows indicate the times

where samples were analyzed by qRT-PCR. A and D: 5 h into the aerobic phase; B and E: transition time from aerobic to anaerobic phase; and C and F: 3 h into the anerobic

phase. A–C samples correspond to the high aeration experiment; and D–F samples correspond to the low aeration experiment.

0

5

10

15

20

25

30

Con

cent

rati

on (

mM

)

Time (h)

Acetate-High aeration Acetate-Low aerationPyruvate-High aeration Pyruvate-Low aerationMalate-High aeration Malate-Low aeration

0

4

8

12

16

20

OD

(60

0 nm

)

Time (h)High aeration Low aeration

020406080

100120140160180

0 6 8 10 12 14 16

0 2 4 6 8 10

2 40 2 4 6 8 10 12 14 16

Con

cent

rati

on (

mM

)

Time (h)Glucose-High aeration Glucose-Low aeration

Succinate-High aeration Succinate-Low aeration

Fig. 2. Growth and metabolic profiles of cultures of strain SBS550MG (pHL413): (a) cell growth profile during aerobic phase, (b) glucose consumption and succinate

production profiles during the anaerobic phase, and (c) acetate, pyruvate and malate profiles during anaerobic phase for high and low aeration conditions during the

aerobic-growth phase on bioreactor experiments.

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509 503

In our experiments very low DO concentrations (microaerobicconditions) were reached in the low aeration experiment3 h before the conditions were switched to the production phase(Fig. 1a). When the DO level decreased close to zero in the lowaeration case, glucose was not in high concentration, butrather low, no pyruvate was accumulated in that case, andwhen glucose was added (at the same time that CO2 flow started)the anaerobic enzymatic machinery was ready and no pyruvateaccumulation was observed. Sanchez and collaborators showedthat pyruvate was accumulated in SBS550MG cultures where noPYC was overexpressed leading to a lower succinate production(Sanchez et al., 2005b). The level of pyc mRNA in the low aerationcase at the time of transition was significantly higher (2.2-fold)than in the high aeration case (Fig. 4c) which may have

contributed to no pyruvate accumulation. Pyruvate accumulationand lower succinate production have also been observed in E. coli

mutants lacking LdhA and PflAB (pyruvate utilization enzymes)(Lin et al., 2005; Vemuri et al., 2002a) implying a limitation inthe capacity of E. coli to direct all the carbon flow towardssuccinate production. Vemuri and collaborators compared succi-nate production using an E. coli strain containing deletions onpflAB and ldhA, and a mutation on ptsG, and the same strainoverexpressing the pyc gene from Rhizobium etli (Vemuri et al.,2002b). They concluded that the strain overexpressing pyc

showed higher succinate yield, supporting the idea that Pyc isan important factor in succinate yield, likely because Pyc directsthe carboxylation of pyruvate to oxaloacetate decreasing pyruvateaccumulation.

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509504

The high aerated bioreactor showed a productivity of 0.92compared to 1.32 g L�1 h�1 for the low aeration experiment.Therefore, the higher cell concentration reached in the highaerated case did not have a positive effect on the volumetricproductivity. In conclusion, the low aeration experiment showeda higher succinate yield, titer and productivity.

3.3. Metabolic fluxes

Metabolic fluxes were estimated as described in Materials andmethods and Appendix A and the results are shown in Fig. 3.

The metabolic fluxes obtained for the low aeration case werecomparable to the estimated fluxes reported by Sanchez andcollaborators using the same E. coli strain but where the cells weregrown aerobically in shake flasks and then transferred to abioreactor for the anaerobic production phase (Sanchez et al.,2006), instead of performing both steps in the same bioreactor

AceA

Isocitrat

OAA

Malate

Succinate

NAD+NADH

glyoxylate

Acetyl-CoAAceB

Fumarate

NADH

NAD+

AceA

Citrate

CO2

CO2

Pp

Succinate

ν5

ν11

ν12

ν13

ν14

ν1

ν11

ν11ν11

Malateν17

120115

8493

117103

155124

312

ν11

Mdh

Fig. 3. Metabolic flux distribution in succinate producing strain SBS550MG (pHL413).

high aeration cases during the growth-aerobic phase. The values in the box show the m

and the high aeration experiment, respectively. The fluxes are expressed in mmol/(

Abbreviations: GAP: glyceraldehyde 3-phosphate; DHAP: dihydroxyacetone-phosphate.

Pta: phosphate acetyl transferase; AckA: acetate kinase; AdhE: alcohol dehydrogena

Pyc: pyruvate carboxylase from L. lactis; Ppc: phosphoenolpyruvate carboxylase.

(this work). Therefore, to carry out the growth phase ofSBS550MG (pHL413) in a bioreactor with low aeration resultedequivalent to the growth of the cells in aerobic shake flasks, bothproviding cells in the right metabolic state for efficient anaerobicsuccinate production. The growth of the cells in a bioreactorinstead of shake flasks eliminates the centrifugation step,simplifying process scalability for industrial applications.

The traditional fermentation pathway in wild type E. coli

produces 2 moles of NADH per mole of metabolized glucose; andsuccinate production requires 2 moles of NADH per mole ofsuccinate produced through this pathway. Then, the succinatemaximum theoretical yield for this pathway is one mol of succinateper mol of glucose due to NADH availability limitation. When theglyoxylate shunt is activated, a new succinate production pathwayis available with a lower NADH requirement, only 0.5 mole ofNADH per mole of succinate produced. However, the productionof succinate through the glyoxylate shunt is limited by carbonavailability. A compromise between both pathways gives a

e

NADH

Pyruvate

PflAB

GAP

Glucose

G6P

Pi + NAD+

Pyk

PEP

Acetyl-CoA

LactateLdhA

PtaAcetyl-P

AdhE2 NADH

2 NAD+

Ethanol

Formate

AckAAcetate

NAD+NADH

Pyc

c

Formate

ν1

ν3

ν4

biomassν2

ν6

ν7

ν8

ν9ν10

ν15

1

Pyruvateν16

Fdh

100100

00

101100

201201

15

8053

6211

7100

0

3721

1942

029

Flux distribution was determined for the anaerobic-production phase for low and

etabolic fluxes, the first and second rows correspond to the flux in the low aeration

gDCWnh), normalized by the glucose uptake rate and expressed in percentage.

Pyk: pyruvate kinase; PflAB: pyruvate formate lyase; LdhA: lactate dehydrogenase;

se; AceA: isocitrate lyase; AceB: malate synthase; Mdh: malate dehydrogenase;

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509 505

maximal theoretical succinate yield of 1.6 mole succinate/moleof glucose, where about 30% of the oxaloacetate is directedto the glyoxylate shunt and ca. 70% is fermented through malate(Cox et al., 2006; Sanchez et al., 2006). Sanchez and collaboratorshave reported this succinate yield experimentally using E. coli

strain SBS550MG (pHL413) cultured aerobically in shake flasksand producing succinate anaerobically in a bioreactor (Sanchezet al., 2006). A theoretical study performed by Cox andcollaborators has shown that succinate yield is highly sensitiveto the oxaloacetate split ratio between the traditional fermenta-tion pathway and the glyoxylate shunt (Cox et al., 2006). In ourstudy, the oxaloacetate split ratio was different for both growthconditions, where the low aeration experiment showed asuccinate yield and oxaloacetate split ratio comparable to thatobtained by Sanchez and collaborators, as mentioned earlier(Sanchez et al., 2006); and the high aeration experiment showeda lower flux of oxaloacetate directed to the glyoxylate shunt(ca. 18% compared to ca. 30%) which led to a significantly lowersuccinate yield (1.25 compared to 1.55 mole succinate/moleglucose). This result reaffirms the importance of the oxaloacetatesplit ratio on succinate yield.

The PEP-PYR node flux to oxaloacetate (v5) was similar in bothexperiments (the difference was only ca. 4%); although, the fluxthrough the PflAB (v7) was significantly lower in the high aerationcase leading to pyruvate accumulation and a lower flux to the

0

1

2

3

4

5

5h aerobic phase Transition 3h anaerobic phase

Rel

ativ

e ge

ne e

xpre

ssio

n

(low

aer

atio

n/hi

gh a

erat

ion)

Rel

ativ

e ge

ne e

xpre

ssio

n

(low

aer

atio

n/hi

gh a

erat

ion) Glycolysis genes

Anaplerotic genes

0

1

2

3

5h aerobic phase Transit

Regulator gen

0

1

2

3

4

pykA pykF aceE aceF lpd

5h aerobic phase Transition 3h anaerobic phase

Rel

ativ

e ge

ne e

xpre

ssio

n (l

ow a

erat

ion/

high

aer

atio

n)

pyc pps ppc

arcA arcB

Fig. 4. Effect of the level of aeration during the growth (aerobic) phase on ge

glyoxylate shunt, decreasing succinate yield. An E. coli strain withdeletions in ldhA and pflB also showed to accumulate pyruvate(Singh et al., 2009). In both studies, the pyruvate accumulationwas probably due to a limited capacity of the cells to channel allthe pyruvate to succinate production and a potential redoximbalance (NADH/NAD+) within the cells resulting in lowsuccinate yield and growth limitation in both cases (Singh et al.,2009). Growth limitation has also been observed in othermicroorganisms when a limited availability of reducing power inthe form of NADPH is present (Panagiotou et al., 2009).

In conclusion, the metabolic flux distribution and product yieldduring the anaerobic phase were significantly affected by the levelof aeration during the growth phase.

3.4. Comparison of gene expression profiles

The difference in gene expression was considered significantif the expression of a particular gene under a specific conditionwas at least twice the mRNA abundance of the same gene in adifferent condition. Samples for qRT-PCR were taken at 5 h intothe aerobic phase (A and D in Fig. 1), at the transition time fromaerobic to anaerobic (B and E in Fig. 1), and 3 h into the anaerobicphase (C and F in Fig. 1) to compare gene expression profiles atdifferent stages in the process. Fig. 4 shows the results for the gene

02468

10121416

5h aerobic phase Transition 3h anaerobic phase

Anaerobic genes

0

2

4

6

8

5h aerobic phase Transition 3h anaerobic phase

Rel

ativ

e ge

ne e

xpre

ssio

n (l

ow a

erat

ion/

high

aer

atio

n) TCA cycle and glyoxylate shunt genes

Rel

ativ

e ge

ne e

xpre

ssio

n (l

ow a

erat

ion/

high

aer

atio

n)

ion 3h anaerobic phase

es

pflA

acnA acnB aceA aceB fumA fumB fumC mdh gltA

fnr

pflB pflD yfiD frdA frdB frdC frdD

ne expression profiles in E. coli SBS550MG (pHL413) in batch bioreactor.

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509506

expression profiles of the genes listed in Table 1. The relative level ofexpression was determined using the high aeration case as thereference.

3.5. Glycolytic genes

The genes pykF and pykA encode for pyruvate kinaseisoenzymes; and PykF is the main contributor to the pyruvatekinase activity in E. coli (Garrido-Pertierra and Cooper, 1983;Ponce et al., 1995). Quantitative RT-PCR revealed that pykF wasup-regulated (2.6 times) during the transition time in the lowaeration case compared to the high aeration experiment (Fig. 4a).Peng and Shimizu showed a modest increase (o2-fold) on pykF

expression in wild type E. coli (K12) during microaerobiccondition (Peng and Shimizu, 2003).

Pyruvate is mainly metabolized to acetyl-CoA via the NAD-dependent pyruvate dehydrogenase complex (Pdh) under aerobicconditions and through the pyruvate-formate-lyase (PflAB) duringanaerobiosis (Kim et al., 2008). The Pdh complex contains threeenzymatic components encoded by aceE, aceF and lpd co-expressed from a single operon from the Ppdh promoter.However, lpd could also be expressed from an independentpromoter (Plpd) which is anaerobically repressed by ArcA (Kanget al., 2005). Pdh transcription is repressed by PdhR (present alsoin the PDH operon), activated by pyruvate and has a site for Fnrregulation (Cassey et al., 1998; Kang et al., 2005; Quail et al.,1994). In addition to transcriptional regulation, Pdh is repressedat the enzyme level by high NADH/NAD+ levels and inhibited byacetyl-CoA (Cassey et al., 1998). In our experiments, the Pdh genesaceE, aceF and lpd were down-regulated in the aerobic phase(A and D in Fig. 1) in the low aeration case compared to the highaeration case, probably due to the relatively lower oxygenavailability (Fig. 4a). On the other hand, the genes aceE and lpd

were overexpressed in the anaerobic phase in the low aerationexperiment compared to the high aeration case, 2.1 and 3.8 times,respectively (samples correspond to C and F in Fig. 1). Theexpression of the anaerobic genes pflB and pflA, involved in theconversion of pyruvate to acetyl-CoA under anaerobic conditions,are discussed below.

3.6. TCA cycle and glyoxylate shunt genes

The genes acnA and acnB encode for aconitate hydratases(aconitases), and catalyze the conversion of citrate into isocitrate.This reaction showed a higher metabolic flux for the low aerationcase (Fig. 3), in agreement with the higher gene expressionobserved in the anaerobic phase (Fig. 4b).

The glyoxylate shunt genes aceA and aceB were up-regulatedduring the transition and anaerobic phases which is consistentwith a higher flux through the glyoxylate shunt and highersuccinate yields. In the high aeration case, where pyruvate wasaccumulated, a significantly lower isocitrate lyase expression(aceA) was observed (Fig. 4b); this result is in agreement withVemuri and collaborators observation that pyruvate was accu-mulated when no isocitrate lyase activity was present (Vemuriet al., 2002a).

The malate dehydrogenase gene (mdh), responsible for thereversible oxidation of malate to oxaloacetate, was highly up-regulated during the anaerobic-production phase (6.9 fold)potentially contributing to the higher succinate production inthe low aeration case (Fig. 4b). Mdh overexpression has also beenshown to increase succinate production in a pflB ldhA doublemutant E. coli strain (Wang et al., 2009).

3.7. Anaplerotic genes

The phosphoenolpyruvate synthase gene, pps, was up-regu-lated during the transition and anaerobic phases; whereas thephosphoenolpyruvate carboxylase gene, ppc, showed to be up-regulated only during the transition time in the low aeration case(Fig. 4c). In the past, the overexpression of ppc has been shown toincrease succinate production (Millard et al., 1996), possibly byincreasing the oxaloacetate pool. Pps converts pyruvate into PEP,Ppc can then carboxylate PEP to generate oxaloacetate which canbe finally transformed into succinate. The overexpression of themalate feedback inhibition-resistant S. vulgare phosphoenolpyr-uvate carboxylase, encoded by ppc, has also been shown toincrease succinate production (Lin et al., 2004).

The strain used for this study, SBS550MG (pHL413), over-expresses pyruvate carboxylase, Pyc, from L. lactis (Sanchez et al.,2005b). This enzyme catalyzes the carboxylation of pyruvate togenerate oxaloacetate. The overexpression of this pyc gene or the pyc

gene from R. etli (Gokarn et al., 2001) increases the oxaloacetate pooland has been shown to consequently increase succinate productionand yield (Gokarn et al., 2001; Sanchez et al., 2005a). In addition topps and ppc, the pyc gene expression was also up-regulated duringthe transition time in the low aeration case; these enzymes mayhave contributed to the slightly higher carbon flux towardsoxaloacetate observed in the low aeration case (Figs. 3 and 4c).

3.8. Anaerobic genes

In our experiment at the transition time the low aeratedbioreactor was in a microaerobic environment for a few hours,which could have contributed to the up-regulation of fermenta-tive genes and therefore to the preparation of the enzymaticmachinery for the upcoming anaerobic environment.

The pyruvate formate lyase genes, genes pflA and pflB were up-regulated during the microaerobic phase (Fig. 4d). Their up-regulation may have led to the higher flux towards the acetyl-CoApool (Fig. 3) and to relieve the pyruvate node, eliminating thepyruvate accumulation observed in the high aeration experiment.Peng and Shimizu (2003) studied the effect of microaerobicconditions on protein expression and enzyme activity in wild typeE. coli K12 grown in LB medium supplemented with glucose. Theyfound differences in protein synthesis levels when the DOcondition was changed from aerobic to microaerobic. Thefermentative genes such as pflB and adhE were up-regulated inmicroaerobic conditions compared to aerobic growth (Peng andShimizu, 2003). An increase in PflAB protein expression of 11.2-fold, based on two-dimensional electrophoresis, when the condi-tion changes from aerobic to microaerobic has been reported(Peng and Shimizu, 2003). Alexeeva and collaborators reportedthat PflAB synthesis was maximal in microaerobic conditions andthat the ArcAB regulatory two-component system was moreactive under this condition, activating pflB gene expression(Alexeeva et al., 2000). In our experiment, at the transition timethe low aerated bioreactor was in a microaerobic environment fora few hours. At this time the pflB gene expression was up-regulated in the low aeration case, microaerobic condition; thisresult is in agreement with the reports mentioned above(Alexeeva et al., 2000; Peng and Shimizu, 2003).

PflAB activity is not only regulated by gene transcription but isalso regulated by the presence of oxygen. PflAB is highly sensitiveto oxygen, is usually inactive during aerobic conditions and activeduring anaerobiosis. YfiD has been suggested to activate PflABduring microaerobic conditions by replacing the glycyl radicaldomain in PflAB (Wagner et al., 2001; Wyborn et al., 2002).Transcription studies have shown maximum yfiD expression in

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509 507

cultures under microaerobic conditions (Marshall et al., 2001) andthat yfiD expression could also been induced by low pH (Wybornet al., 2002). In our experiments, the yfiD gene was highlyexpressed (10.8-fold) at the transition time in the low aerationcase, where the cells had been under microaerobic conditions forabout 3 h. This result agrees with the transcription study byMarshall and collaborators (Marshall et al., 2001). YfiD has beenshown to contribute to PflAB flux in other mutant strains, e.g. anarcA mutant (Zhu et al., 2007). A higher yfiD expression during thetransition time in our low aeration experiment may havecontributed to the increase in PflAB flux and therefore to avoidpyruvate accumulation in SBS550MG (pHL413).

3.9. Regulators

The Arc two-component signal transduction system is knownto regulate the expression of numerous genes in response to theredox state of the cell (Georgellis et al., 1997; Liu and De Wulf,2004). Alexeeva and collaborators analyzed the importance of thepresence of the ArcAB system in E. coli redox regulation atdifferent oxygen levels; and they concluded that this system ismore important in microaerobic conditions to maintain the redoxbalance (Alexeeva et al., 2000, 2003). In our work, the geneencoding for ArcB (arcB), the sensor kinase component of the Arctwo-component signal transduction system, was up-regulated inthe aerobic phase; and no significant difference in gene expres-sion was observed in the arcA and arcB genes at the transitiontime where the bioreactor with low aeration was under amicroaerobic condition. At low oxygen levels the transmembranesensor kinase ArcB autophosphorylates; and subsequently phos-phorylates the response regulator component ArcA. The systemcontinues to be active during anaerobic growth (Jeon et al., 2001;Lee et al., 2001; Shalel-Levanon et al., 2005a). Arc-P binds to thepromoter region regulating transcription of numerous genes.Thus, the action of this system is controlled by phosphorylationinduced by the cell redox state.

In addition to the ArcAB regulatory system, E. coli also containsa Fnr global regulator which is known to activate expression ofgenes involved in anaerobic growth and repression of aerobicgenes (Liu and De Wulf, 2004; Shalel-Levanon et al., 2005a). In thepresent study, no significant difference in fnr gene expression wasfound in both experiments, low and high aeration during theaerobic growth phase, at the analyzed time points.

Glucose consumed

Biomass

Lactate

Acetate

Ethanol

GAP

G6P

PEP=PYR

Acetyl-CoA

Formate

Succinate

OAA

Excreted succinate

Residual formate

Malate

Excreted pyruvate

Excreted malate

NADH

26666666666666666666666666666666666666664

37777777777777777777777777777777777777775

¼

1 0 0 0 0 0 0 0 0

0 1 0 0 0 0 0 0 0

0 0 0 0 0 1 0 0 0

0 0 0 0 0 0 0 0 1

0 0 0 0 0 0 0 0 0

0 0 2 �1 0 0 0 0 0

1 �1 �1 0 0 0 0 0 0

0 0 0 1 �1 �1 �1 0 0

0 0 0 0 0 0 1 0 �1

0 0 0 0 0 0 1 �1 0

0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0

0 0 0 1 0 �1 0 0 0

26666666666666666666666666666666666666664

4. Conclusion

The low aeration experiment showed better succinate produc-tion with respect to yield, titer and volumetric productivity. Thiswas most likely due to the period of microaerobiosis that the cellsexperienced before switching to a completely anaerobic environ-ment. This microaerobic environment may have facilitated thecells to adjust their enzymatic machinery by expressing theproteins that were required in the anaerobic-production phase,making them a better biocatalyst for succinate production.

The main differences in gene expression levels were detectedduring the late aerobic growth period. The high aerationexperiment showed pyruvate accumulation, which correlatedwith a lower pflAB expression during the transition time and alower flux towards acetyl-CoA during the anaerobic phasecompared to the low aeration case. The increase in glyoxylateshunt related genes expression (aceA, aceB, acnA, acnB) during thetransition time, anaerobic phase, or both, improved succinateyield in the low aeration case. Therefore, the increase in metabolicflux towards acetyl-CoA and the increased activity of theglyoxylate shunt were crucial for the accomplishment of themaximum theoretical succinate yield for the E. coli SBS550MG(pHL413) strain.

The metabolic behavior of a cell is highly influenced by theoperation conditions of the process. The genetic manipulationthemselves are not enough to make a strain a high producer orattain the maximum theoretical yield. Appropriate processconditions are crucial for the metabolic engineered strains toreach their maximum potential, as they may allow the modifiedmetabolic routes to behave as predicted.

Appendix A. Metabolic flux analysis

The metabolic fluxes were determined considering the stoi-choimetry of the system as follows (adapted from Sanchez et al.,2006):

rðtÞ ¼ A vðtÞ ðA1Þ

where r(t) is the matrix containing the net reaction ratesmeasured (18�1), A is the stoichiometric coefficient matrix(18�17) and v(t) contains the metabolic fluxes (17�1) to bedetermined as indicated in Fig. 3. Then,

0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

1 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

0 0 0 0 0 0 �1 0

�1 �2 0 0 0 0 0 0

0 0 0 0 0 �1 0 0

0 1 0 1 �1 0 0 0

0 �1 �1 0 0 0 0 0

0 0 0 0 1 0 0 0

0 0 0 0 0 1 0 0

0 1 1 �1 0 0 0 �1

0 0 0 0 0 0 1 0

0 0 0 0 0 0 0 1

�2 0 �1 �1 0 0 0 0

37777777777777777777777777777777777777775

v1

v2

v3

v4

v

v6

v7

v8

v9

v10

v11

v12

v13

v14

v15

v16

v17

266666666666666666666666666666666666664

377777777777777777777777777777777777775

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509508

The system is overdetermined and it was solved using theleast-square fit as follows (Stephanopoulos et al., 1998):

vðtÞ ¼ AT A� ��1

AT rðtÞ ðA2Þ

The metabolic fluxes determined as described above areshown in Fig. 3.

References

Alexeeva, S., de Kort, B., Sawers, G., Hellingwerf, K.J., de Mattos, M.J., 2000. Effectsof limited aeration and of the ArcAB system on intermediary pyruvatecatabolism in Escherichia coli. J. Bacteriol. 182, 4934–4940.

Alexeeva, S., Hellingwerf, K.J., Teixeira de Mattos, M.J., 2003. Requirement of ArcAfor redox regulation in Escherichia coli under microaerobic but not anaerobic oraerobic conditions. J. Bacteriol. 185, 204–209.

Becker, S., Holighaus, G., Gabrielczyk, T., Unden, G., 1996. O2 as the regulatorysignal for FNR-dependent gene regulation in Escherichia coli. J. Bacteriol. 178,4515–4521.

Cassey, B., Guest, J.R., Attwood, M.M., 1998. Environmental control of pyruvatedehydrogenase complex expression in Escherichia coli. FEMS Microbiol. Lett.159, 325–329.

Cox, S.J., Shalel Levanon, S., Sanchez, A., Lin, H., Peercy, B., Bennett, G.N., San, K.Y.,2006. Development of a metabolic network design and optimization frame-work incorporating implementation constraints: a succinate production casestudy. Metab. Eng. 8, 46–57.

Chatterjee, R., Millard, C.S., Champion, K., Clark, D.P., Donnelly, M.I., 2001.Mutation of the ptsG gene results in increased production of succinate infermentation of glucose by Escherichia coli. Appl. Environ. Microbiol. 67,148–154.

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. Appl. Biochem. Biotechnol. 70-72, 187–198.

Garrido-Pertierra, A., Cooper, R.A., 1983. Evidence for two distinct pyruvate kinasegenes in Escherichia coli K-12. FEBS Lett. 162, 420–422.

Georgellis, D., Lynch, A.S., Lin, E.C., 1997. In vitro phosphorylation study of the arctwo-component signal transduction system of Escherichia coli. J. Bacteriol. 179,5429–5435.

Gokarn, R.R., Altman, E., Eiteman, M.A., 1998. Expression of pyruvate carboxylaseenhances succinate production in Escherichia coli without affecting glucoseuptake. Biotechnol. Lett. 20, 795–798.

Gokarn, R.R., Eiteman, M.A., Altman, E., 2000. Metabolic analysis of Escherichia coliin the presence and absence of the carboxylating enzymes phosphoenolpyr-uvate carboxylase and pyruvate carboxylase. Appl. Environ. Microbiol. 66,1844–1850.

Gokarn, R.R., Evans, J.D., Walker, J.R., Martin, S.A., Eiteman, M.A., Altman, E., 2001.The physiological effects and metabolic alterations caused by the expression ofRhizobium etli pyruvate carboxylase in Escherichia coli. Appl. Microbiol.Biotechnol. 56, 188–195.

Hong, S.H., Lee, S.Y., 2001. Metabolic flux analysis for succinic acid production byrecombinant Escherichia coli with amplified malic enzyme activity. Biotechnol.Bioeng. 74, 89–95.

Hong, S.H., Lee, S.Y., 2002. Importance of redox balance on the production ofsuccinic acid by metabolically engineered Escherichia coli. Appl. Microbiol.Biotechnol. 58, 286–290.

Ishizaki, A., Vonktaveesuk, P., 1996. Optimization of substrate feed for continuousproduction of lactic acid by Lactococcus lactis IO-1. Biotechnol. Lett. 18,1113–1118.

Jantama, K., Haupt, M.J., Svoronos, S.A., Zhang, X., Moore, J.C., Shanmugam, K.T.,Ingram, L.O., 2008a. Combining metabolic engineering and metabolic evolu-tion to develop nonrecombinant strains of Escherichia coli C that producesuccinate and malate. Biotechnol. Bioeng. 99, 1140–1153.

Jantama, K., Zhang, X., Moore, J.C., Shanmugam, K.T., Svoronos, S.A., Ingram, L.O.,2008b. Eliminating side products and increasing succinate yields in engineeredstrains of Escherichia coli C. Biotechnol. Bioeng. 101, 881–893.

Jeon, Y., Lee, Y.S., Han, J.S., Kim, J.B., Hwang, D.S., 2001. Multimerization ofphosphorylated and non-phosphorylated ArcA is necessary for the responseregulator function of the Arc two-component signal transduction system.J. Biol. Chem. 276, 40873–40879.

Kang, Y., Weber, K.D., Qiu, Y., Kiley, P.J., Blattner, F.R., 2005. Genome-wideexpression analysis indicates that FNR of Escherichia coli K-12 regulates a largenumber of genes of unknown function. J. Bacteriol. 187, 1135–1160.

Kim, Y., Ingram, L.O., Shanmugam, K.T., 2008. Dihydrolipoamide dehydrogenasemutation alters the NADH sensitivity of pyruvate dehydrogenase complex ofEscherichia coli K-12. J. Bacteriol. 190, 3851–3858.

Kwon, Y.D., Kwon, O.H., Lee, H.S., Kim, P., 2007. The effect of NADP-dependentmalic enzyme expression and anaerobic C4 metabolism in Escherichiacoli compared with other anaplerotic enzymes. J. Appl. Microbiol. 103,2340–2345.

Lee, S.Y., Hong, S.H., Lee, S.H., Park, S.J., 2004. Fermentative production ofchemicals that can be used for polymer synthesis. Macromol. Biosci. 4,157–164.

Lee, Y.S., Han, J.S., Jeon, Y., Hwang, D.S., 2001. The arc two-component signaltransduction system inhibits in vitro Escherichia coli chromosomal initiation.J. Biol. Chem. 276, 9917–9923.

Lin, H., San, K.Y., Bennett, G.N., 2005. Effect of Sorghum vulgare phosphoenolpyr-uvate carboxylase and Lactococcus lactis pyruvate carboxylase coexpression onsuccinate production in mutant strains of Escherichia coli. Appl. Microbiol.Biotechnol. 67, 515–523.

Lin, H., Vadali, R.V., Bennett, G.N., San, K.Y., 2004. Increasing the acetyl-CoA pool inthe presence of overexpressed phosphoenolpyruvate carboxylase or pyruvatecarboxylase enhances succinate production in Escherichia coli. Biotechnol.Prog. 20, 1599–1604.

Liu, X., De Wulf, P., 2004. Probing the ArcA-P modulon of Escherichia coli by wholegenome transcriptional analysis and sequence recognition profiling. J. Biol.Chem. 279, 12588–12597.

Marshall, F.A., Messenger, S.L., Wyborn, N.R., Guest, J.R., Wing, H., Busby, S.J.,Green, J., 2001. A novel promoter architecture for microaerobic activation bythe anaerobic transcription factor FNR. Mol. Microbiol. 39, 747–753.

Mat-Jan, F., Alam, K.Y., Clark, D.P., 1989. Mutants of Escherichia coli deficient in thefermentative lactate dehydrogenase. J. Bacteriol. 171, 342–348.

Meijer, S., Otero, J., Olivares, R., Andersen, M.R., Olsson, L., Nielsen, J., 2009.Overexpression of isocitrate lyase—glyoxylate bypass influence on metabo-lism in Aspergillus niger. Metab. Eng. 11, 107–116.

Millard, C.S., Chao, Y.P., Liao, J.C., Donnelly, M.I., 1996. Enhanced production ofsuccinic acid by overexpression of phosphoenolpyruvate carboxylase inEscherichia coli. Appl. Environ. Microbiol. 62, 1808–1810.

Panagiotou, G., Grotkjaer, T., Hofmann, G., Bapat, P.M., Olsson, L., 2009. Over-expression of a novel endogenous NADH kinase in Aspergillus nidulansenhances growth. Metab. Eng. 11, 31–39.

Peng, L., Shimizu, K., 2003. Global metabolic regulation analysis for Escherichia coliK12 based on protein expression by 2-dimensional electrophoresisand enzyme activity measurement. Appl. Microbiol. Biotechnol. 61,163–178.

Ponce, E., Flores, N., Martinez, A., Valle, F., Bolivar, F., 1995. Cloning of thetwo pyruvate kinase isoenzyme structural genes from Escherichia coli: therelative roles of these enzymes in pyruvate biosynthesis. J. Bacteriol. 177,5719–5722.

Quail, M.A., Haydon, D.J., Guest, J.R., 1994. The pdhR-aceEF-lpd operon ofEscherichia coli expresses the pyruvate dehydrogenase complex. Mol. Micro-biol. 12, 95–104.

Sanchez, A.M., Bennett, G.N., San, K.Y., 2005a. Efficient succinic acid productionfrom glucose through overexpression of pyruvate carboxylase in an Escherichiacoli alcohol dehydrogenase and lactate dehydrogenase mutant. Biotechnol.Prog. 21, 358–365.

Sanchez, A.M., Bennett, G.N., San, KY., 2005b. Novel pathway engineering design ofthe anaerobic central metabolic pathway in Escherichia coli to increasesuccinate yield and productivity. Metab. Eng. 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. Metab.Eng. 8, 209–226.

Sawers, G., 1999. The aerobic/anaerobic interface. Curr. Opin. Microbiol. 2,181–187.

Shalel-Levanon, S., San, K.Y., Bennett, G.N., 2005a. Effect of ArcA and FNR on theexpression of genes related to the oxygen regulation and the glycolysispathway in Escherichia coli under microaerobic growth conditions. Biotechnol.Bioeng. 92, 147–159.

Shalel-Levanon, S., San, K.Y., Bennett, G.N., 2005b. Effect of oxygen, and ArcA andFNR regulators on the expression of genes related to the electron transferchain and the TCA cycle in Escherichia coli. Metab. Eng. 7, 364–374.

Singh, A., Lynch, M.D., Gill, R.T., 2009. Genes restoring redox balance infermentation-deficient E. coli NZN111. Metab. Eng. 11, 347–354.

Soini, J., Falschlehner, C., Liedert, C., Bernhardt, J., Vuoristo, J., Neubauer, P., 2008.Norvaline is accumulated after a down-shift of oxygen in Escherichia coliW3110. Microb. Cell Fact. 7, 30.

Spiro, S., Guest, J.R., 1990. FNR and its role in oxygen-regulated gene expression inEscherichia coli. FEMS Microbiol. Rev. 6, 399–428.

Stephanopoulos, G.N., Aristidou, A.A., Nielsen, J., 1998. Metabolic Engineering:Principles and Methodologies. Academic Press, San Diego 327 pp.

Stols, L., Donnelly, M.I., 1997. Production of succinic acid through overexpressionof NAD(+)-dependent malic enzyme in an Escherichia coli mutant. Appl.Environ. Microbiol. 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. Appl. Biochem. Biotechnol. 63–65, 153–158.

Vemuri, G.N., Eiteman, M.A., Altman, E., 2002a. Effects of growth mode andpyruvate carboxylase on succinic acid production by metabolically engineeredstrains of Escherichia coli. Appl. Environ. Microbiol. 68, 1715–1727.

Vemuri, G.N., Eiteman, M.A., Altman, E., 2002b. Succinate production indual-phase Escherichia coli fermentations depends on the time of transitionfrom aerobic to anaerobic conditions. J. Ind. Microbiol. Biotechnol. 28,325–332.

Wagner, A.F., Schultz, S., Bomke, J., Pils, T., Lehmann, W.D., Knappe, J., 2001. YfiD ofEscherichia coli and Y06I of bacteriophage T4 as autonomous glycyl radicalcofactors reconstituting the catalytic center of oxygen-fragmented pyruvateformate-lyase. Biochem. Biophys. Res. Commun. 285, 456–462.

Wang, Q., Wu, C., Chen, T., Chen, X., Zhao, X., 2006. Expression of galactosepermease and pyruvate carboxylase in Escherichia coli ptsG mutant increases

I. Martınez et al. / Metabolic Engineering 12 (2010) 499–509 509

the growth rate and succinate yield under anaerobic conditions. Biotechnol.Lett. 28, 89–93.

Wang, W., Li, Z., Xie, J., Ye, Q., 2009. Production of succinate by a pflB ldhA doublemutant of Escherichia coli overexpressing malate dehydrogenase. BioprocessBiosyst. Eng. 32, 737–745.

Wyborn, N.R., Messenger, S.L., Henderson, R.A., Sawers, G., Roberts, R.E., Attwood,M.M., Green, J., 2002. Expression of the Escherichia coli yfiD gene responds to

intracellular pH and reduces the accumulation of acidic metabolic endproducts. Microbiology 148, 1015–1026.

Zeikus, J.G., Jain, M.K., Elankovan, P., 1999. Biotechnology of succinic acid production andmarkets for derived industrial products. Appl. Microbiol. Biotechnol. 51, 545–552.

Zhu, J., Shalel-Levanon, S., Bennett, G., San, K.Y., 2007. The YfiD protein contributesto the pyruvate formate-lyase flux in an Escherichia coli arcA mutant strain.Biotechnol. Bioeng. 97, 138–143.