Metabolic engineering of aerobic succinate production systems in Escherichia coli to improve process...

ARTICLE IN PRESS

1096-7176/$ - se

doi:10.1016/j.ym

�Correspondistry Engineeri

Houston, TX 7

E-mail addr

Metabolic Engineering 7 (2005) 116–127

www.elsevier.com/locate/ymben

Metabolic engineering of aerobic succinate production systems inEscherichia coli to improve process productivity and achieve the

maximum theoretical succinate yield

Henry Lina, George N. Bennettb, Ka-Yiu Sana,c,�

aDepartment of Bioengineering, 6100 Main Street, Rice University, Houston, TX 77005-1892, USAbDepartment of Biochemistry and Cell Biology, 6100 Main Street, Rice University, Houston, TX 77005-1892, USA

cDepartment of Chemical Engineering, 6100 Main Street, Rice University, Houston, TX 77005-1892, USA

Received 19 August 2004; accepted 29 October 2004

Available online 8 December 2004

Abstract

The potential to produce succinate aerobically in Escherichia coli would offer great advantages over anaerobic fermentation in

terms of faster biomass generation, carbon throughput, and product formation. Genetic manipulations were performed on two

aerobic succinate production systems to increase their succinate yield and productivity. One of the aerobic succinate production

systems developed earlier (Biotechnol, Bioeng., 2004, accepted) was constructed with five mutations (DsdhAB, Dicd, DiclR, DpoxB,

and D(ackA-pta)), which created a highly active glyoxylate cycle. In this study, a second production system was constructed with

four of the five above mutations (DsdhAB, DiclR, DpoxB, and D(ackA-pta)). This system has two routes in the aerobic central

metabolism for succinate production. One is the glyoxylate cycle and the other is the oxidative branch of the TCA cycle. Inactivation

of ptsG and overexpression of a mutant Sorghum pepc in these two production systems showed that the maximum theoretical

succinate yield of 1.0mol/mol glucose consumed could be achieved. Furthermore, the two-route production system with ptsG

inactivation and pepc overexpression demonstrated substantially higher succinate productivity than the previous system, a level

unsurpassed for aerobic succinate production. This optimized system showed remarkable potential for large-scale aerobic succinate

production and process optimization.

r 2004 Elsevier Inc. All rights reserved.

Keywords: Escherichia coli; Metabolic engineering; Succinate production; Aerobic fermentation; Gene inactivation; Process productivity; Theoretical

yield

1. Introduction

The valuable specialty chemical succinate and itsderivatives have extensive industrial applications. It canbe used as an additive and flavoring agent in foods, asupplement for pharmaceuticals, a surfactant, a deter-gent extender, a foaming agent, and an ion chelator(Zeikus et al., 1999). Currently, succinate is produced

e front matter r 2004 Elsevier Inc. All rights reserved.

ben.2004.10.003

ing author. Department of Bioengineering and Chem-

ng, Rice University, 6100 Main Street, MS-142,

7005-1892, USA. Fax:+ 713 348 5877.

ess: [email protected] (K.-Y. San).

through petrochemical processes that can be expensiveand can lead to pollution problems. Much effort hasshifted toward making biocatalysts a viable andimproved alternative for the production of succinate.The success of microbial fermentation coupled with theuse of renewable carbohydrates would significantlyimprove the economics of the succinate market (Schil-ling, 1995).

Various strains such as Anaerobiospirillum succinici-

producens, Actinobacillus succinogenes, and Escherichia

coli have been intensively studied for their potential asbiocatalysts in succinate fermentation. The obligateanaerobe A. succiniciproducens has shown high potential

www.elsevier.com/locate/ymben

ARTICLE IN PRESSH. Lin et al. / Metabolic Engineering 7 (2005) 116–127 117

for industrial scale succinate production because of itshigh conversion yield and productivity when fermentedwith whey (Lee et al., 2000). However, A. succinicipro-

ducens is not practical for commercial fermentationbecause it is unstable due to its tendency to degenerate,and requires environments absolutely free of oxygen forcultivation (Nghiem et al., 1999). E. coli has also beenextensively genetically engineered through the use ofrecombinant DNA technology in recent years togenerate strains which showed promise for succinatefermentation. E. coli naturally produces succinate as aminor fermentation product under anaerobic conditions(Clark, 1989). Under aerobic conditions, succinate is notproduced as a byproduct in E. coli and acetate is themain byproduct. Numerous metabolic engineeringstrategies to enhance succinate production in E. coli

have met with success. Strains in which enzymatic stepsinvolved in the succinate pathway were amplified andthe organism cultured under anaerobic conditionsyielded higher succinate production. An example of thiswas shown when phosphoenolpyruvate carboxylase(pepc) from E. coli. was overexpressed (Millard et al.,1996). Conversion of fumarate to succinate wasimproved by overexpressing native fumarate reductase(frd) in E. coli (Goldberg et al., 1983; Wang et al., 1998).Certain enzymes are not indigenous in E. coli, but canpotentially help increase succinate production. Byintroducing pyruvate carboxylase (pyc) from Rhizobium

etli into E. coli, succinate production was enhanced(Gokarn et al., 1998, 2000, 2001). Other metabolicengineering strategies also include inactivating compet-ing pathways of succinate. When malic enzyme wasoverexpressed in a host with inactivated pyruvateformate lyase (pfl) and lactate dehydrogenase (ldh)genes, succinate became the major fermentation product(Stols and Donnelly, 1997; Hong and Lee, 2001). Incultures of this pfl and ldh mutant strain, there is a largepyruvate accumulation. Overexpression of malic enzymein this mutant strain increased succinate productiondriven by the high pyruvate pool toward the direction ofmalate formation, which was subsequently converted tosuccinate. An inactive glucose phosphotransferase sys-tem (ptsG) in the same mutant strain (pfl� and ldh�) hadalso been shown to yield higher succinate production inE. coli and improve growth (Chatterjee et al., 2001).

The various genetic improvements described abovefor succinate production have all been done underanaerobic conditions utilizing the mixed-acid fermenta-tion pathways of E. coli. Unfortunately, anaerobicfermentation has inherent disadvantages that aredifficult to surmount. Anaerobic conditions often causepoor cell growth and slow carbon throughput, thereforegenerating low production rates. Succinate formation inmixed-acid fermentation is also hampered by thelimitations of NADH availability, since 2mol of NADHare required for every mole of succinate to be formed.

Strategies to overcome the anaerobic barrier haveincluded generating enough biomass under aerobicconditions, then switching to anaerobic conditions forsuccinate production. This was shown to be effectiveusing a ‘‘dual-phase’’ fermentation system, in whichinitial aerobic growth phase was started then followedby an anaerobic production phase (Vemuri et al.,2002a b).

Absolute aerobic production of succinate in E. coli

until now has not been feasibly engineered (Lin et al.,2004). Saccharomyces cerevisiae has increased succinateproduction when succinate dehydrogenase (sdh) isdisrupted to utilize the oxidative pathway of the TCAcycle for aerobic production (Arikawa et al., 1999). Thecapability to produce succinate under aerobic conditionswould mean an active oxidative phosphorylation forgenerating energy with O2 present as the electronacceptor. This would lead to higher biomass generation,faster carbon throughput and product formation. In thisstudy, we seek to develop succinate production systemsin E. coli that can function under absolute aerobicconditions. These systems would be robust and efficientwith high succinate yield capability, and productivity.

Under aerobic conditions, the production of succinateis not naturally possible since it is only an intermediateof the TCA cycle. It is formed by succinyl-CoAsynthetase, and then subsequently converted to fuma-rate by succinate dehydrogenase (SDH). Through theoxidation reaction by SDH, succinate provides electronsto the electron transport chain for oxidative phosphor-ylation. Because of this recycling process, succinate isnever detected in aerobic cultures of E. coli. Acetate isthe only major byproduct of E. coli under aerobicconditions. Previously, a novel aerobic succinate pro-duction system has been developed that can produce asubstantial amount of succinate under aerobic condi-tions (Lin et al., 2004). This aerobic succinate produc-tion system is based on the presence of five mutations(DsdhAB, Dicd, DpoxB, D(ackA-pta), and DiclR) thatcreate an active glyoxylate cycle in the host strain (Fig.2). This pentamutant strain, with its glyoxylate cycle,can produce succinate as a major product aerobically,but there is still substantial accumulation of pyruvateand TCA cycle C6 intermediates (citrate and isocitrate).Pathway modeling and simulation of aerobic metabo-lism shows that a maximum theoretical succinate yieldof 1.0mol/mol glucose consumed can be achieved.Because of the accumulation of pyruvate and TCAcycle C6 intermediates, the pentamutant strain washindered from achieving the maximum theoreticalsuccinate yield.

In this study, further genetic modifications are madeto the glyoxylate cycle system to improve succinateproduction and achieve the maximum theoreticalsuccinate yield. Another aerobic succinate productionplatform was also created to examine its efficiency in

ARTICLE IN PRESSH. Lin et al. / Metabolic Engineering 7 (2005) 116–127118

succinate production. This system is a mutant strainwith four mutations (DsdhAB, DpoxB, D(ackA-pta), andDiclR). These four mutations create two possible routesfor succinate production, the glyoxylate cycle and theoxidative branch of the TCA cycle (Fig. 3). These twoproduction systems both have the inherent capability toachieve the maximum theoretical succinate yield of1.0mol/mol. Therefore, the pathways of these twosystems were further optimized through genetic manip-ulation to achieve the maximum theoretical succinateyield. Inactivation of the glucose phosphotransferasegene ptsG and overexpression of a malate feedbackinhibition resistant Sorghum phosphoenolpyruvate car-boxylase (PEPC) in the two production systems showedthat the maximum theoretical succinate yield of 1.0mol/mol could indeed be achieved. Furthermore, the plat-form with two routes for succinate production (glyox-ylate cycle and oxidative branch of TCA cycle)demonstrated substantially higher succinate produc-tivity than the other platform when ptsG was inacti-vated and pepc was overexpressed. This productionsystem, in particular, has great potential for large-scaleaerobic production of succinate and further processoptimization.

Fig. 1. Genetic engineering of glycolysis, TCA cycle, and glyoxylate

bypass in the development of aerobic succinate production systems. 1

is icd knockout, 2 is sdhAB knockout, 5 is iclR knockout, 6 is poxB

knockout, 7 is ackA-pta knockout, and 9 is ptsG knockout.

2. Materials and methods

2.1. Strains

Mutations were created in the laboratory wildtypeGJT001, a spontaneous cadR mutant of MC4100(Tolentino et al., 1992). A library of mutant strainswas created during the process of constructing the finalmutant strains for aerobic succinate production. A list

Table 1

List of mutant strains and plasmids constructed and studied

Genotype

Strains

GJT001 Spontaneous cadR mutant of MC4100(ATC3569

HL51276k GJT001(DiclR, Dicd, DsdhAB, D(ackA-pta), Dpox

HL512769k GJT001(DiclR, Dicd, DsdhAB, D(ackA-pta), Dpox

HL2765k GJT001(DsdhAB, D(ackA-pta), DpoxB, DiclR::Km

HL27659k GJT001(DsdhAB, D(ackA-pta), DpoxB, DiclR, Dp

HL51276k(pKK313) HL51276k overexpressing S8D mutant Sorghum


HL2765k(pKK313) HL2765k overexpressing S8D mutant Sorghum p


HL51276k(pKK313C) Control strain for HL51276k(pKK313)




Plasmids

pKK313 S8D mutant Sorghum pepc, ApR

pKK313C Control vector of pKK313 with inactive Sorghum

of the final mutant strains that were studied is shown inTable 1. Knockouts were created in succinate dehydro-genase (sdhAB), pyruvate oxidase (poxB), acetatekinase-phosphotransacetylase (ackA-pta), isocitrate de-hydrogenase (icd), aceBAK operon repressor (iclR), andthe glucose phosphotransferase system (ptsG). Each oneof these mutations is designated by a number (1-Dicd, 2-DsdhAB, 5-DiclR, 6-DpoxB, 7-D(ackA-pta), 9-DptsG)used in the naming of the mutant strains (Fig. 1). Thekanamycin cassette was left in the final mutant strains to

Reference

5) Dlac(arg-lac)U169rpsL150relA1ptsF SmR Tolentino et al. (1992)

B::KmR) Lin et al. (2004)

B, DptsG::KmR) This studyR) This study

tsG::KmR) This study

pepc This study

pepc This study

epc This study

pepc This study

This study

This study

This study

This study

Wang et al. (1992)

pepc, ApR This study


provide selective pressure during the fermentationexperiments.

Plasmid pKK313, carrying the S8D mutation in theSorghum pepc gene (Wang et al., 1992), was used toproduce high levels of PEPC in the mutant strains(Table 1). Plasmid pKK313 confers ampicillin resis-tance. This altered Sorghum PEPC has a site-directedmutation that advantageously relieves malate feedbackinhibition from the enzyme (Wang et al., 1992). Acontrol plasmid was created from pKK313, designatedpKK313C, by excising a 2.4 kb fragment from pepc

using two SacI sites within the gene (Table 1). PlasmidpKK313C was then tested to show that it produced noPEPC activity.

2.2. Mutant construction

Mutations were created using the one-step inactiva-tion method of Datsenko and Wanner (Datsenko andWanner, 2000). This method first requires the construc-tion of the single mutations using the phage l Redrecombinase. P1 phage transduction was then used tocombine various mutations into one strain. Eachmutation had to be added to the strain one at a timebefore the introduction of the next mutation because thekanamycin cassette had to be removed at each stage toenable selection of the next mutation. PCR products ofthe kanamycin cassette gene flanked by FRT (FLPrecognition target) sites and homologous sequences tothe gene of interest were made using pKD4 (Datsenkoand Wanner, 2000) as the template. These PCRproducts were then transformed into the cells byelectroporation (Bio-Rad Gene Pulser) for insertionalinactivation of the gene of interest. These transformedcells carry the plasmid pKD46 (Datsenko and Wanner,2000) that expresses the l Red system (g, b, exo) forrecombination of the PCR product into the chromo-some. Once the kanamycin cassette is inserted, it can beremoved using the helper plasmid, pCP20 (Datsenkoand Wanner, 2000), that expresses FLP. The removal ofthe FRT-flanked kanamycin cassette leaves behind an84 base pair insertion cassette. At each stage ofmutation, experiments were performed to test theintermediate mutant strain for the effect on metaboliteproduction. Throughout the process of constructing theaerobic succinate production systems, a library ofdifferent mutant strains with varying types and numbersof mutations was created. All mutant strains were alsoverified with genomic PCR after construction to ensurethat the gene of interest had been disrupted.

2.3. Bioreactor culture medium and conditions

Aerobic batch reactor experiments were conductedfor all the mutant strains. The medium used is LB with2 g/L NaHCO3 and approximately 60mM of glucose.

The medium used for inoculum preparation is also LB,except glucose was not supplemented. NaHCO3 wasadded to the culture medium because it yielded bettercell growth and succinate production due to its pH-buffering capacity and its ability to supply CO2.Kanamycin was added to the medium at a concentrationof 50mg/L for strains not harboring plasmids. In strainsharboring pKK313 or pKK313C, ampicillin, carbeni-cillin, and oxacillin were added to the medium at aconcentration of 200mg/L each. Studies have shownthat the use of methicillin and ampicillin is effective as aselective pressure in the cultivation of recombinant E.

coli (Lee and Kim, 1996). Oxacillin is an analog ofmethicillin. The use of ampicillin, carbenicillin, andoxacillin in combination during the experiments en-forced that the plasmids were retained throughout theaerobic fermentation. IPTG was added at 1 mM to themedium to induce gene expression for plasmids pKK313and pKK313C.

The initial medium volume is 600ml in a 1.0-L NewBrunswick Scientific Bioflo 110 fermenter. A 1% (v/v)inoculum was used from an overnight culture grownfrom a single colony for 12 h. The pH was measuredusing a glass electrode and controlled at 7.0 using 1.5NHNO3 and 2N Na2CO3. The temperature was main-tained at 37 1C, and the agitation speed was constant at800 rpm. The inlet airflow used was 1.5 L/min. Thedissolved oxygen was monitored using a polarographicoxygen electrode (New Brunswick Scientific) and wasmaintained above 80% saturation throughout theexperiment. This dissolved oxygen level was to demon-strate that the succinate production systems wereworking under absolute aerobic conditions.

2.4. Analytical techniques

Optical density was measured at 600 nm with aspectrophotometer (Bausch & Lomb Spectronic 1001);the culture was diluted to the linear range with 0.15MNaCl. For analyzing the extracellular metabolites, 1mlof culture was centrifuged and the supernatant was thenfiltered through a 0.45-mm syringe filter for HPLCanalysis. The HPLC system (Shimadzu-10A Systems,Shimadzu, Columbia, MD) used was equipped with acation-exchange column (HPX-87H, BioRad Labs,Hercules, CA), a UV detector (Shimadzu SPD-10A)and a differential refractive index (RI) detector (Waters2410, Waters, Milford, MA). A 0.6mL/min mobilephase using 2.5mM H2SO4 solution was applied to thecolumn. The column was operated at 55 1C. Standardswere prepared for glucose, succinate, acetate, andpyruvate for both the RI detector and UV detector,and calibration curves were created. Glucose, succinate,and acetate were measured by the RI detector andpyruvate was measured by the UV detector at 210 nm.


3. Results and discussion

3.1. Comparison of two aerobic succinate production

systems

Previously, an aerobic succinate production platformwas developed, which could produce succinate exclu-sively through the glyoxylate cycle (Fig. 2) (Lin et al.,2004). Five mutations (DsdhAB, Dicd, DiclR, DpoxB,and D(ackA-pta)) (Fig. 1) were strategically created inthe glycolysis, TCA cycle, and glyoxylate bypassto create a pentamutant strain of E. coli, HL51276k(Table 1). These mutations together created an activeglyoxylate cycle, which was shown to produce asubstantial amount of succinate under aerobic condi-tions (Lin et al., 2004). This system was the firstplatform developed in E. coli to show the feasibility ofproducing succinate entirely under aerobic conditions.Strain HL51276k, although it produced a substantialamount of succinate also exhibited accumulation ofpyruvate and TCA cycle C6 intermediates, such ascitrate and isocitrate. Because of the accumulation, themaximum theoretical yield of one mole succinateproduced per mole glucose consumed could not beachieved and the productivity was hampered.

In the current study, another aerobic succinateproduction system was constructed in E. coli to improveupon the platform developed in strain HL51276k. Thissystem has four mutations created (DsdhAB, DiclR,DpoxB, and D(ackA-pta)) (Fig. 1). It is essentially the

Fig. 2. Aerobic succinate production platform, the glyoxylate cycle

(mutant strain HL51276k).

same as HL51276k, except icd is not inactivated. Thesefour mutations create the strain HL2765k (Table 1),which has two pathways opened for succinate produc-tion (Fig. 3). One pathway is the glyoxylate cycle leadingto succinate formation. The second pathway utilizes theoxidative arm of the TCA cycle leading to succinate.Both pathways branch from isocitrate to form succinate(Fig. 3).

Strains HL51276k and HL2765k were grown inbioreactors at 37 1C where the dissolved oxygen wasmaintained above 80% saturation throughout theexperiment. Their metabolite profiles were compared.Results showed that strain HL2765k had a highersuccinate production than HL51276k. At approximately48 h, the succinate concentration in the HL2765k culturewas 40mM compared to that of the HL51276k culture,which had 31mM succinate (Fig. 4a). Succinate molaryields at the highest concentration produced were 0.67for HL2765k and 0.65 for HL51276k (Table 2).HL2765k also had 65% higher volumetric succinateproductivity and 12% higher specific succinate produc-tivity than HL51276k (Table 2).

Strain HL2765k grew to a higher OD (14.27 OD) thanstrain HL51276k (9.21 OD). HL2765k also had a fasterbiomass generation rate (0.60 g/l h) than HL51276k(0.24 g/l h), because its glucose consumption rate isfaster than HL51276k. There was pyruvate accumula-tion in cultures of both strains, which was produced andthen consumed (Fig. 4c). HL2765k had a higherpyruvate accumulation in the beginning of fermentation

Fig. 3. Aerobic succinate production platform, the two-route system

with the glyoxylate cycle and the oxidative branch of the TCA cycle

(mutant strain HL2765k).

ARTICLE IN PRESS

0

10

20

30

40

50

60

70

80

0

5

10

15

20

25

30

35

0

10

20

30

40

50

60

0 5 10 15 20 25 30 35 40 45 50 55

Suc

cina

te (

mM

)

hr

0 5 10 15 20 25 30 35 40 45 50 55

hr

00

10

20

30

40

50

60

70

5 10 15 20 25 30 35 40 45 50 55

hr

0 5 10 15 20 25 30 35 40 45 50 55

hr

Glu

cose

(m

M)

Pyr

uvat

e (m

M)

Ace

tate

(m

M)

(a) (b)

(c) (d)

Fig. 4. Metabolite comparison in cultures of strains HL51276k, HL512769k, HL2765k, and HL27659k. (a) succinate production graph; (b) glucose

remaining graph; (c) pyruvate production graph; (d) acetate production graph. Solid square (’): HL27659k; solid triangle (m): HL2765k; open

square (&): HL512769k; open triangle (n): HL51276k. Cultivation medium is LB with 2 g/L NaHCO3 and approximately 60mM of glucose.

Table 2

Results of succinate yield and productivity of cultures of various

mutant strains in aerobic batch reactor experiments

Strain YS/G (mol/mol) Qp (g/l h) qp (mg/g h)

HL51276k 0.65 0.057 24.04

HL512769k 0.87 0.086 35.47

HL2765k 0.67 0.094 26.84

HL27659k 0.78 0.130 32.82

HL51276k(pKK313C) 0.61 0.048 27.54

HL51276k(pKK313) 1.09 0.140 44.26

HL512769k(pKK313C) 0.85 0.083 38.99

HL512769k(pKK313) 0.96 0.094 45.23

HL2765k(pKK313C) 0.71 0.113 28.33

HL2765k(pKK313) 0.75 0.111 35.54

HL27659k(pKK313C) 0.74 0.106 31.14

HL27659k(pKK313) 0.95 0.270 73.66

YS/G is the molar succinate yield at the end of fermentation (mole of

succinate produced per mole of glucose consumed); Qp is the average

volumetric succinate productivity at the end of fermentation (mass

concentration of succinate (g/l) over time (h)); qp is the average specific

succinate productivity at the end of fermentation (mass of succinate

(mg) per mass of biomass (g) over time (h)).

H. Lin et al. / Metabolic Engineering 7 (2005) 116–127 121

than HL51276k. This is because HL2765k has fasterglucose consumption and cell mass generation thanHL51276k. The pyruvate was taken up and consumedfaster by HL2765k than HL51276k (Fig. 4c). HL2765kalso produced acetate faster than HL51276k due to itsmore rapid glucose consumption rate (Fig. 4d). StrainHL2765k did not have any accumulation of TCA cycle

C6 intermediates, whereas strain HL51276k did (datanot shown). Lactate and ethanol were not detected inthe cultures of either strain. Comparison of these twostrains showed that HL2765k was more robust thanHL51276k. Strain HL2765k with two pathways engi-neered for succinate production has a faster succinateproductivity and glucose consumption rate thanHL51276k, which only utilizes the glyoxylate cycle forsuccinate production.

3.2. Effect of ptsG inactivation in the two aerobic

succinate production systems

The glucose phosphotransferase system (PTSG) wasstudied in the two strains HL2765k and HL51276k toexamine the possibility of reducing pyruvate and acetateaccumulation (Chou et al., 1994). By inactivating thephosphotransferase uptake system, pyruvate can nolonger be formed from phosphoenolpyruvate (PEP)through the transport of glucose. This genetic manip-ulation can potentially reduce pyruvate accumulation.Acetate formation occurs because of excess consump-tion of glucose that the cell is unable to utilize forbiomass synthesis or energy requirements, leading torepression of enzymes in the TCA cycle by glucose(Yang et al., 1999). The secretion of acetate leads to anuncoupled metabolism (Doelle et al., 1981). InactivatingptsG can slow glucose uptake and possibly allow a morebalanced glucose metabolism. Inactivation of ptsG hasbeen shown to increase succinate production in pyruvateformate lyase and lactate dehydrogenase mutant strains(Chatterjee et al., 2001; Chou et al., 1994). This effect is


probably due to more PEP being conserved andavailable for the succinate synthesis pathway, while alsogenerating a slower glucose uptake rate. ptsG wasinactivated in the strain HL2765k to form HL27659k(Table 1), and in the strain HL51276k to formHL512769k (hexamutant strain of E. coli) (Table 1).The number 9 represents the inactivation of ptsG

(Fig. 1).Strains HL27659k and HL512769k were grown

aerobically under the same batch reactor conditionsdescribed earlier for strains HL2765k and HL51276k.The results showed ptsG inactivation did improvesuccinate production. At approximately 48 h,HL27659k produced 49mM succinate compared toHL2765k, which produced 40mM, and HL512769kproduced 44mM succinate compared to HL51276k,which produced 31mM (Fig. 4a). At the highestsuccinate concentration produced, the molar yield wasalso higher when ptsG was inactivated (0.78 forHL27659k compared to 0.67 for HL2765k, and 0.87for HL512769k compared to 0.65 for HL51276k) (Table2). The succinate volumetric productivity and specificproductivity at the highest succinate concentration werealso higher from cultures of those strains with ptsG

inactivation. HL27659k had 38% higher succinatevolumetric productivity and 22% higher specific pro-ductivity than HL2765k (Table 2). HL512769k had 51%higher succinate volumetric productivity and 48%higher specific productivity than HL51276k (Table 2).The effects of ptsG inactivation improve succinateproduction more in HL51276k than in HL2765k;this is because there are more bottlenecks in theTCA pathways of HL51276k than in HL2765k(HL51276k has TCA cycle C6 accumulation andHL2765k does not).

ptsG inactivation caused cell growth to be lower dueto slower glucose consumption during the exponentialphase. HL27659k grew to an OD of 12.59 at the end ofits exponential phase compared to HL2765k, whichgrew to 14.27 OD. HL512769k grew to an OD of 8.31 atthe end of its exponential phase compared to HL51276k,which grew to an OD of 9.21. By the end of theexponential phase, the biomass generation rate of strainHL2765k was 0.60 g/l h compared to strain HL27659k,which was 0.27 g/l h. For strain HL51276k, the biomassgeneration rate at the end of the exponential phase was0.24 g/l h compared to strain HL512769k, which was0.13 g/l h.

Inactivation of ptsG did reduce pyruvate accumula-tion in cultures of strains HL2765k and HL51276k.Pyruvate accumulation in HL27659k only reached amaximum concentration of 48mM compared toHL2765k, which reached 72mM (Fig. 4c). StrainHL512769k produced a maximum pyruvate concentra-tion of 23mM, compared to HL51276k, which pro-duced 48mM (Fig. 4c). In strain HL512769k, there was

no pyruvate accumulation after 48 h. All the glucose wasconsumed by cultures of all four strains by the end the offermentation. Inactivation of ptsG also reduced acetateproduction. HL27659k had a lower acetate productionrate than HL2765k in the first 24 h. By the end of thefermentation though when all the glucose was consumedin the cultures, HL27659k had slightly higher acetateproduction than HL2765k (Fig. 4d). Acetate productionwas lower in HL512769k than in HL51276k throughoutthe fermentation (Fig. 4d). Strain HL512769k accumu-lated TCA cycle C6 intermediates and strain HL27659kdid not. Both lactate and ethanol were not detected inthe cultures of these strains. Results showed that ptsG

inactivation did improve succinate yield and productiv-ity in cultures of the two strains, HL2765k andHL51276k. Because of the ptsG inactivation, glucoseconsumption was slowed, which helped reduce pyruvateand acetate accumulation by providing a more balancedmetabolism.

3.3. Effect of pepc overexpression with ptsG inactivation

in the aerobic succinate production systems

ptsG inactivation has been shown to improve succi-nate yield and productivity. Yet in strains HL27659kand HL512769k, the maximum theoretical succinateyield of one mole produced per mole glucose consumedhas not been obtained. This indicates that the aerobicproduction systems can be further optimized. PEPCconverts PEP to OAA through a carboxylation reactionwith CO2 (Fig. 1). OAA is an important precursor forthe synthesis of succinate. Overexpression of PEPC in E.

coli has been shown to increase succinate production(Millard et al., 1996). Therefore, overexpression ofPEPC in the aerobic succinate production systemsshould improve succinate yield and productivity, andat the same time further reduce pyruvate and acetateaccumulation. A mutant PEPC from Sorghum wasoverexpressed on plasmid pKK313 (Table 1) in thefour strains HL51276k, HL512769k, HL2765k, andHL27659k. This mutant PEPC is feedback inhibitionresistant to malate (Wang et al., 1992). This mutation isadvantageous for the aerobic succinate productionsystems because even though succinate is not beingformed by the reduction of malate as in anaerobicconditions, malate is still present and required in theglyoxylate cycle as the precursor of OAA (Figs. 2 and 3).The control plasmid for pKK313 is pKK313C (Table 1),which was also transformed into the four mutantstrains. The mutant strains carrying the plasmids weregrown aerobically in bioreactors at 37 1C where thedissolved oxygen was maintained above 80% saturationthroughout the experiment. Mutant strains harboringpKK313 were compared in terms of their metaboliteproduction and succinate yield and productivity with thesame mutant strains harboring pKK313C.

ARTICLE IN PRESS

00 10 20 30 40 50 60

10

20

30

40

50

60

70

0

10

20

30

40

50

60

70

0

5

10

15

20

25

30

35

40

45

0

2

4

6

8

10

12

14

16

18

20

Suc

cina

te (

mM

)

Ace

tate

(m

M)

Glu

cose

(m

M)

Pyr

uvat

e (m

M)

hr

0 10 20 30 40 50 60

hr0 10 20 30 40 50 60

hr

0 10 20 30 40 50 60

hr(a) (b)

(c) (d)

Fig. 5. Metabolite comparison in cultures of strains HL51276k(pKK313) and HL51276k(pKK313C). (a) succinate production graph; (b) glucose

remaining graph; (c) pyruvate production graph; (d) acetate production graph. Solid diamond (E): HL51276k(pKK313); solid square (’):

HL51276k(pKK313C). Cultivation medium is LB with 2 g/L NaHCO3 and approximately 60mM of glucose.


Overexpression of the mutant Sorghum PEPC instrains HL51276k, HL512769k, HL2765k, andHL27659k was effective in increasing succinate produc-tion. The succinate production of cultures of strainHL51276k(pKK313) was 130% higher than those ofstrain HL51276k(pKK313C) at the end of the fermenta-tion when all the glucose had been consumed (Fig. 5a).Cultures of HL27659k(pKK313) had 37% highersuccinate production than its control strain HL27659k(pKK313C) (Fig. 8a). The increase in succinate produc-tion of cultures of strains HL512769k(pKK313) andHL2765k(pKK313) compared to their respective con-trols was not as substantial as that found forHL51276k(pKK313), although their succinate concen-trations were continuously higher than their controlsthroughout the production phase (Figs. 6a and 7a). Instrains with high levels of PEPC, the molar succinateyields for strains HL51276k(pKK313), HL512769k(pKK313) and HL27659k(pKK313) all reached themaximum theoretical value of one mole of succinateproduced per mole of glucose consumed (Table 2). Themolar succinate yield for strain HL2765k(pKK313)was 0.75 compared to 0.71 for its control strainHL2765(pKK313C) (Table 2). PEPC overexpression,therefore, was not as effective in increasing the succi-nate yield in strain HL2765k as in the other strains.The specific succinate productivity was higher in allthe strains overexpressing PEPC; strain HL51276k(pKK313) was 61% higher than its control; HL512769k(pKK313) was 16% higher, HL2765k(pKK313) was25% higher, and HL27659k(pKK313) was 137% higher

than respective their controls carrying pKK313C(Table 2). These results showed that high expressionof mutant Sorghum PEPC was very effective inimproving succinate yield in the mutant E. coli hoststrains and successfully optimized three of the aerobicproduction systems (HL51276k, HL512769k, andHL27659k) to produce the maximum theoreticalsuccinate yield of 1.0mol/mol glucose consumed.Although cultures of strains HL51276k(pKK313),HL512769k(pKK313), and HL27659k(pKK313) allachieved the maximum theoretical succinate yield,they still varied in their efficiency. Among the threestrains, HL27659k(pKK313) was the most efficient.Fermentation showed substantially higher volumetricsuccinate productivity (0.27 g/l h) and specific produc-tivity (73.66mg/g h) than the other two strains (Table2). The volumetric succinate productivity for culturesof HL27659k(pKK313) was 93% and 187% higherthan cultures of HL51276k(pKK313) and HL512769k(pKK313), respectively. As for specific succinateproductivity, cultures of HL27659k(pKK313) were66% and 63% higher than those of strainsHL51276k(pKK313) and HL512769k(pKK313), re-spectively. These results demonstrate that fermenta-tions of strain HL27659k(pKK313), with its 1.0mol/mol succinate yield, is a more efficient and robustaerobic succinate production system for large-scaleproduction than the other systems.

Overexpression of PEPC was also effective in redu-cing pyruvate production. Maximum pyruvate producedin cultures of strains HL51276k(pKK313) and

ARTICLE IN PRESS

02468

10121416182022

0

2

4

6

8

10

12

14

16

18

20

00 10 20 30 40 50 60 70 80

10

20

30

40

50

60

70

05

101520253035404550556065

Suc

cina

te (

mM

)

Ace

tate

(m

M)

Pyr

uvat

e (m

M)

Glu

cose

(m

M)

hr

0 10 20 30 40 50 60 70 80

hr0 10 20 30 40 50 60 70 80

hr

0 10 20 30 40 50 60 70 80

hr(a) (b)

(c) (d)




0

10

20

30

40

50

60

70

80

0

5

10

15

20

25

30

35

40

00 10 20 30 40 50

5101520253035404550

05

101520253035404550556065

Suc

cina

te (

mM

)

Ace

tate

(m

M)

Pyr

uvat

e (m

M)

Glu

cose

(m

M)

hr

0 10 20 30 40 50

hr0 10 20 30 40 50

hr

0 10 20 30 40 50

hr(a) (b)

(c) (d)




H. Lin et al. / Metabolic Engineering 7 (2005) 116–127124

HL2765k(pKK313) was lower than their controlscarrying pKK313C (Figs. 5c and 7c). In cultures ofstrains HL512769k(pKK313) and HL27659k(pKK313),pyruvate accumulation was virtually eliminated (Fig. 6cand Fig. 8c). These results demonstrate that over-

expression of pepc coupled with ptsG inactivation wasthe most effective in reducing pyruvate accumulation,thus providing more efficient carbon throughput.

Acetate production was reduced in cultures of themutant strains with the high levels of PEPC. Cultures of

ARTICLE IN PRESS

0

5

10

15

20

25

30

35

40

0

5

10

15

20

25

00 10 20 30 40 50

5101520253035404550556065

05

101520253035404550556065

Suc

cina

te (

mM

)

Ace

tate

(m

M)

Pyr

uvat

e (m

M)

Glu

cose

(m

M)

hr

0 10 20 30 40 50

hr0 10 20 30 40 50

hr

0 10 20 30 40 50

hr(a) (b)

(c) (d)





strain HL512769k(pKK313) had a 47% reduction,HL2765k(pKK313) had a 29% reduction, HL27659k(pKK313) had an 82% reduction compared to culturesof their respective control strains carrying pKK313C(Figs. 6d, 7d and 8d). Acetate production for strainHL51276k(pKK313) was lower than HL51276k(pKK313C) throughout production, but their finalacetate concentrations at the end of fermentation afterall the glucose was consumed were similar (Fig. 5d).Cultures of strain HL27659k(pKK313) exhibited thelowest acetate level produced among all the four strains(below 5mM) and also had little pyruvate accumulated(below 2mM). Its main product was succinate at 60mMat the end of fermentation when all the glucose wasconsumed, equivalent to approximately 1.0mol succi-nate/mol glucose. Lactate and ethanol were not detectedin the cultures of any of these mutant strains.

The culture conditions of the mutant strains devel-oped and studied for aerobic succinate production, thusfar, have not been optimized for production. Thecalculation of succinate yield and productivity (Table2) has included both the growth phase (biomassgeneration) and the production phase. These aerobicsystems are efficient and practical because they do notrequire separation of the growth phase from theproduction phase for succinate production as inconventional anaerobic succinate production systems(Vemuri et al., 2002a, b; Nghiem et al., 1999). Never-theless, the culture conditions of the aerobic succinateproduction systems can be further optimized to improveprocess productivity.

The reproducibility of the bioreactor experimentresults of the various mutant strains was demonstratedby repeating the experiments for a select few mutantstrains that are prominent for aerobic succinate produc-tion (HL51276k, HL51276k(pKK313), and HL27659k(pKK313)) (Fig. 9). The results of these duplicateexperiments showed that the trends of the metaboliteconcentrations were the same, therefore, reproducible.Analysis of metabolite samples was performed intriplicates in HPLC to ensure accuracy.

4. Conclusion

Although metabolic engineering of E. coli to enhancesuccinate production under anaerobic conditions hasgenerated many promising improvements, it is stillhampered by inherent anaerobic constraints, such ascofactor NADH requirement and slow cell growth. Thismakes application on an industrial scale difficult due tolow biomass generation and the poor physiological state ofthe cell. Aerobic conditions provide many advantages thatfavor implementation on an industrial scale due to fasterbiomass generation, carbon throughput and productformation. Aerobic succinate production systems werethus developed in this study to demonstrate promising andfeasible alternatives to anaerobic fermentation. Productionsystems were optimized to achieve the maximum theore-tical succinate yield of 1.0mol/mol of glucose.

Two aerobic succinate production platforms werepresented, HL51276k (Lin et al., 2004) and HL2765k

ARTICLE IN PRESS

00 5 10 15 20 25 30 35

0 5 10 15 20 25 30 35

40 45 50 55

10

20

30

40

50

60

70

0

10

20

30

40

50

60

0

10

20

30

40

50

60

70

0

10

20

30

40

50

60

70

hr

hr

0 5 10 15 20 25 30 35 40 45 50 55

hr

0 5 10 15 20 25

hr

gluc

ose

and

succ

inat

e (m

M)

gluc

ose

and

succ

inat

e (m

M)

gluc

ose

and

succ

inat

e (m

M)

pyru

vate

and

ace

tate

(m

M)

(a) (b)

(c) (d)

Fig. 9. Duplicate batch reactor experiments of select mutant strains to show result reproducibility. (a) open square (&): HL51276k glucose

consumption (data from Fig. 4); open triangle (n): HL51276k glucose consumption (repeated experiment); solid square (’): HL51276k succinate

production (data from Fig. 4); solid triangle (m): HL51276k succinate production (repeated experiment). (b) open square (&): HL51276k pyruvate

production (data from Fig. 4); open triangle (n): HL51276k pyruvate production (repeated experiment); solid square (’): HL51276k acetate

production (data from Fig. 4); solid triangle (m): HL51276k acetate production (repeated experiment). (c) open square (&): HL51276k(pKK313)

glucose consumption (data from Fig. 5); open triangle (n): HL51276k(pKK313) glucose consumption (repeated experiment); solid square (’):

HL51276k(pKK313) succinate production (data from Fig. 5); solid triangle (m): HL51276k(pKK313) succinate production (repeated experiment).

(d) open square (&): HL27659k(pKK313) glucose consumption (data from Fig. 8); open triangle (n): HL27659k(pKK313) glucose consumption

(repeated experiment); solid square (’):HL27659k(pKK313) succinate production (data from Fig. 8); solid triangle (m):HL27659k(pKK313)

succinate production (repeated experiment).

H. Lin et al. / Metabolic Engineering 7 (2005) 116–127126

(this study). Strain HL51276k’s central metabolism isthe glyoxylate cycle (Fig. 2), and strain HL2765k has theglyoxylate cycle and the oxidative branch of the TCAcycle for synthesizing succinate (Fig. 3). StrainHL2765k, with higher succinate productivity, producedmore succinate than strain HL51276k. It also had fasterbiomass generation and glucose consumption thanHL51276k. Strain HL2765k also did not accumulateany TCA cycle C6 intermediates, whereas strainHL51276k did.

With ptsG inactivated in strains HL512769k andHL27659k, succinate concentration and productivityincreased compared to fermentations of strainsHL51276k and HL2765k, respectively. Inactivation ofptsG also reduced pyruvate and acetate accumulation inthe two strains. Overexpression of the mutant Sorghum

pepc in strains HL51276k, HL512769k, HL2765k, andHL27659k promoted a significant increase in succinateyield and productivity. It also reduced pyruvate andacetate production in cultures of all four strains.Cultures of the three strains, HL51276k(pKK313),HL512769k(pKK313), and HL27659k(pKK313), achievedthe maximum theoretical succinate yield of 1.0mol/mol.Among these three production systems, strainHL27659k(pKK313) had substantially higher succinateproductivity than the other two strains. It also hadminimal pyruvate and acetate accumulation throughout

the fermentation. Strain HL27659k(pKK313) is the mostefficient aerobic succinate production system developedand its cultures demonstrate a succinate yield of 1.0mol/mol glucose. It is a promising system for large-scalesuccinate production and process optimization.

Acknowledgments

The authors would like to thank Dr. Jean Vidal forproviding the Sorghum pepc plasmid (pKK313). Theauthors would also like to thank Dr. Barry L. Wannerfor providing plasmids pKD4, pKD46, and pCP20 thatfacilitate the construction of mutations in E. coli. Thiswork was supported by grants from the NationalScience Foundation (BES-0222691 and BES-0000303).Henry Lin was supported by a training grant from theNational Science Foundation (DGE0114264).

References

Arikawa, Y., Kuroyanagi, T., Shimosaka, M., Muratsubaki, H.,

Enomoto, K., Kodaira, R., Okazaki, M., 1999. Effect of gene

disruptions of the TCA cycle on production of succinic acid in

Saccharomyces cerevisiae. J. Biosci. Bioeng. 87, 28–36.

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 in fermentation of glucose by Escherichia

coli. Appl. Environ. Microbiol. 67, 148–154.

Chou, C.-H., Bennett, G.N., San, K.-Y., 1994. Effect of modified

glucose uptake using genetic engineering techniques on high-level

recombinant protein production in Escherichia coli dense cultures.

Biotechnol. Bioeng. 44, 952–960.

Clark, D.P., 1989. The fermentation pathways of Escherichia coli.

FEMS Microbiol. Rev. 63, 223–234.

Datsenko, K.A., Wanner, B.L., 2000. One-step inactivation of

chromosomal genes in Escherichia coli K-12 using PCR products.

Proc. Natl. Acad. Sci. 97, 6640–6645.

Doelle, H.W., Ewings, K.N., Hollywood, N.W., 1981. Regulation of

glucose metabolism in bacterial systems. Adv. Biochem. Eng./

Biotechnol. 23, 1–36.

Gokarn, R.R., Eiteman, M.A., Altman, E., 1998. Expression of

pyruvate carboxylase enhances succinate production in Escherichia

coli without affecting glucose uptake. Biotech. Lett. 20, 795–798.

Gokarn, R.R., Eiteman, M.A., Altman, E., 2000. Metabolic analysis

of Escherichia coli in the presence and absence of the carboxylating

enzymes phosphoenolpyruvate carboxylase and pyruvate carbox-

ylase. 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 of Rhizobium etli pyruvate

carboxylase in Escherichia coli. Appl. Microbiol. Biotechnol. 56,

188–195.

Goldberg, I., Lonberg-Holm, K., Bagley, E.A., Stieglitz, B., 1983.

Improved conversion of fumarate to succinate by Escherichia coli

strains amplified for fumarate reductase. App. Environ. Microbiol.

45, 1838–1847.

Hong, S.H., Lee, S.Y., 2001. Metabolic flux analysis for succinic acid

production by recombinant Escherichia coli with amplified malic

enzyme acitvity. Biotechnol. Bioeng. 74 (2), 89–95.

Lee, D.-C., Kim, H.-S., 1996. Use of methicillin and ampicillin mixture

as a selective pressure in the cultivation of recombinant Escherichia

coli. Biotechnol. Tech. 10 (12), 937–940.

Lee, P.C., Lee, W.G., Kwon, S., Lee, S.Y., Chang, H.N., 2000. Batch

and continuous cultivation of Anaerobiospirillum succiniciproducens

for the production of succinic acid from whey. Appl. Microbiol.

Biotechnol. 54, 23–27.

Lin, H., Bennett, G.N., San, K.-Y., 2004. Genetic reconstruction of the

aerobic central metabolism in Escherichia coli for the absolute

aerobic production of succinate. Biotechnol. Bioeng., accepted

Millard, C.S., Chao, Y., Liao, J.C., Donnelly, M.I., 1996. Enhanced

production of succinic acid by overexpression of phosphoenolpyr-

uvate carboxylase in Escherichia coli. Appl. Environ. Microbiol. 62,

1808–1810.

Nghiem, N.P., Donnelly, M., Millard, C.S., Stols, L., February 1999.

US patent 5,869,301.

Schilling, L.B., 1995. Chemicals from alternative feedstocks in the

United States. FEMS Microbiol. Rev. 16, 101–110.

Stols, L., Donnelly, M.I., 1997. Production of succinic acid through

overexpression of NAD+-dependent malic enzyme in an Escher-

ichia coli mutant. Appl. Environ. Microbiol. 63, 2695–2701.

Tolentino, G.J., Meng, S.-Y., Bennett, G.N., San, K.-Y., 1992. A pH-

regulated promoter for the expression of recombinant proteins in

Escherichia coli. Biotech. Lett. 14, 157–162.

Vemuri, G.N., Eiteman, M.A., Altman, E., 2002a. Effects of growth

mode and pyruvate carboxylase on succinic acid production by

metabolically engineered strains of Escherichia coli. App. Environ.

Microbiol. 68, 1715–1727.

Vemuri, G.N., Eiteman, M.A., Altman, E., 2002b. Succinate produc-

tion in dual-phase Escherichia coli fermentations depends on the

time of transition from aerobic to anaerobic conditions. J. Ind.

Microbiol. Biotechnol. 28, 325–332.

Wang, X., Gong, C.S., Tsao, G.T., 1998. Bioconversion of fumaric

acid to succinic acid by recombinant Escherichia coli. Appl.

Biochem. Biotechnol. 70–72, 919–928.

Wang, Y.-H., Duff, S.M.G., Lepiniec, L., Cretin, C., Sarath, G.,

Condon, S.A., Vidal, J., Gadal, P., Chollet, R., 1992. Site-directed

mutagenesis of the phosphorylatable serine (Ser8) in C4 phosphoe-

nolpyruvate carboxylase from Sorghum. J. Biol. Chem. 267,

16759–16762.

Yang, Y.-T., San, K.-Y., Bennett, G.N., 1999. Redistribution of

metabolic fluxes in Escherichia coli with fermentative lactate

dehydrogenase overexpression and deletion. Metab. Eng. 1,

141–152.

Zeikus, J.G., Jain, M.K., Elankovan, P., 1999. Biotechnology of

succinic acid production and markets for derived industrial

products. Appl. Microbiol. Biotechnol. 51, 545–552.

Metabolic engineering of aerobic succinate production systems in Escherichia coli to improve process...

Documents

Transcript of Metabolic engineering of aerobic succinate production systems in Escherichia coli to improve process...