ARTICLE IN PRESS

1096-7176/$ - se

doi:10.1016/j.ym

�CorrespondE-mail addr

1Current add

MA, USA.

Metabolic Engineering 9 (2007) 133–141

www.elsevier.com/locate/ymben

Recombinant protein production in an Escherichia coli reducedgenome strain

Shamik S. Sharmaa,1, Frederick R. Blattnerb, Sarah W. Harcumc,�

aDepartment of Chemical and Biomolecular Engineering, Clemson University, Clemson, SC 29634, USAbDepartment of Genetics, University of Wisconsin, Madison, WI 53706, USAcDepartment of Bioengineering, Clemson University, Clemson, SC 29634, USA

Received 11 July 2006; received in revised form 4 October 2006; accepted 11 October 2006

Available online 21 October 2006

Abstract

Recently, efforts have been made to improve the properties of Escherichia coli as a recombinant host by ‘genomic surgery’—deleting

large segments of the E. coli K12 MG1655 genome without scars. These excised segments included K-islands, which contain a high

proportion of transposons, insertion sequences, cryptic phage, damaged, and unknown-function genes. The resulting multiple-deletion

strain, designated E. coli MDS40, has a 14% (about 700 genes) smaller genome than the parent strain, E. coli MG1655. The multiple-

deletion and parent E. coli strains were cultured in fed-batch fermenters to high cell densities on minimal medium to simulate industrial

conditions for evaluating growth and recombinant protein production characteristics. Recombinant protein production and by-product

levels were quantified at different controlled growth rates. These results indicate that the multiple-deletion strain’s growth behavior and

recombinant protein productivity closely matched the parent stain. Thus, the multiple-deletion strain E. coli MDS40 provides a suitable

foundation for further genomic reduction.

r 2006 Elsevier Inc. All rights reserved.

Keywords: Reduced genome; Deletion; Growth rate; Acetate; Transposon; Insertion sequence

1. Introduction

Escherichia coli is one of the most studied and well-understood microorganisms. It naturally occurs in the gutof mammals, which is an anaerobic environment withconstantly changing nutrient conditions. E. coli is also acommonly used recombinant host for laboratory andindustrial recombinant protein production. As a recombi-nant host, E. coli is exposed to only a limited andcontrolled set of conditions. Specifically, in an industrialfermenter, an aerobic environment is usually desired andmaintained, the nutrient concentrations are maintainedwithin narrow ranges, and attachment to the vessel is notdesirable. Therefore, the genes required for survival in the

e front matter r 2006 Elsevier Inc. All rights reserved.

ben.2006.10.002

ing author. Fax: +1864 656 0567.

ess: harcum@clemson.edu (S.W. Harcum).

ress: New England Biolabs, 240 County Road, Ipswich,

gut may not be the same genes required for optimumrecombinant protein production. Further, the completegenome sequence of E. coli has revealed numerous genes ofunknown function and genetic material that has possiblybeen acquired from other organisms in the recent past(Blattner et al., 1997). In an effort to improve E. coli as arecombinant host, many researchers have deleted or addedsingle genes to the genome or modified plasmids tocomplement the existing genome (Andersen and Krum-men, 2002; Andersen et al., 2001; Baneyx and Mujacic,2004; Bessette et al., 1999; Cebolla et al., 2002; Chen et al.,2003; Chevalet et al., 2000; Chou et al., 1996; Diaz-Ricciet al., 1991; Pecota et al., 1997). These efforts have madeprogress, but have not truly addressed the global issue ofthe numerous genes in E. coli with unknown function orpotentially detrimental function. In this work, a multipledeletion strain—E. coli MDS40—was characterized withrespect to growth and recombinant protein production.E. coli MDS40 was derived from E. coli MG1655 by the

ARTICLE IN PRESSS.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141134

excision of many large nucleotide sequences correspondingto potentially non-essential genes, while retaining thebackbone E. coli genome.

A few research groups have created multiple deletionE. coli strains, although not always with the focus ofimproving recombinant protein production. These multipledeletion methods mainly used random techniques facili-tated by transposon libraries (Yu et al., 2002), specializedtransposons (Goryshin et al., 2003), and l phage homo-logous recombination (Hashimoto et al., 2005). One non-random or targeted deletion method, used sequentialplasmid-directed recombination events (Kolisnychenkoet al., 2002). Both the random and non-random multipledeletion techniques have generated strains with 8–30%smaller genomes. The random method-derived strains havebeen observed to have significantly lower growth rates,whereas the reported targeted multiple deletion strainshave been observed to have comparable growth rates to theparent strain on the most commonly used laboratorymedia, including a minimal medium. The targeted deletionmethod strains were created by excising sections of theE. coli MG1655 genome regarded as unnecessary ordetrimental, with the aim of improving its properties(Kolisnychenko et al., 2002; Posfai et al., 2006). Thetargeted genes were selected based on a comparison of thecomplete genome sequences of two E. coli strains (MG1655and O157:H7). This analysis revealed a ‘backbone’ ofgenetic information, which contained hundreds of strainspecific ‘islands’. For E. coli MG1655, a K12 strain, theseislands were designated K-islands. The backbone genome ishighly conserved and contained all essential genes. Incontrast, the ‘islands’ contains a disproportionate numberof genes with unknown function, toxin genes, transposableelements, and pseudogenes (Blattner et al., 1997; Kolisny-chenko et al., 2002). In the first targeted multiple deletionstrain described—E. coli MDS12—twelve sequential dele-tions of K-islands were made to E. coli MG1655, resultingin a multiple deletion strain with an 8% smaller genome.Twenty-eight further sequential deletions resulted in thestrain E. coli MDS40, which was used in this study. E. coli

MDS40 has a 14% smaller genome than E. coli MG1655 orroughly 700 fewer genes. Additionally, these deletionstrains have improved transformabilities and lower muta-tion rates (Posfai et al., 2006).

The focus of this study was to determine if E. coli

MDS40 was a robust and suitable host for recombinantprotein production. Fed-batch fermentations were used toobtain high cell densities under controlled growth condi-tions in minimal medium typical of industrial processes.Cell yield, byproduct accumulation, and recombinantprotein productivity were measured and compared to theparent strain, E. coli MG1655. Another aim was todetermine the suitability of E. coli MDS40 as the primarymultiple deletion strain for further targeted deletions.Additional deletions could, for example, target thearabinose utilization genes, such as the E. coli HB101strain, to allow recombinant protein expression using the

pBAD vector (Khlebnikov et al., 2002; Sorensen andMortensen, 2005).

2. Materials and methods

2.1. Strains and expression vector

E. coli MG1655 was obtained from American TypeCulture Collection (ATCC). The multiple deletion strainE. coli MDS40 was provided by Scarab Genomics(Madison, WI). E. coli MDS40 was created by 40successive targeted deletions from the E. coli MG1655genome i.e., 28 targeted successive deletions from E. coli

MDS12 (Kolisnychenko et al., 2002; Posfai et al., 2006).Chloramphenicol acetyltransferase (CAT) was used as themodel recombinant protein encoded by the plasmidpPROEXCAT (Invitrogen). The pPROEXCAT plasmidcontains a trc promoter that controls CAT expression viaIPTG-induction, a pBR322 origin of replication, and theb-lactamase gene for ampicillin resistance. E. coli MDS40was transformed with pPROEXCAT via the CaCl2 method(Sambrook et al., 1989). CAT was selected as the initialmodel recombinant protein for a few reasons. Namely,there is an extensive body of literature available for thisprotein’s expression in E. coli MG1655 and related strains.CAT levels in the cell can be easily quantified by anenzymatic assay (Rodriguez and Tait, 1983). Additionally,previous studies have observed that the unusual amino acidcomposition of CAT (11% phenylalanine) can triggermetabolic burden or stress responses in the host cell(Harcum and Bentley, 1993).

2.2. Shake flask cultivation

For the shake flask experiments, stock cultures stored at�80 1C, were thawed and 0.5ml was used to inoculate10ml of defined medium. The medium was as described inKorz et al. (1995), except the batch medium contained8 g/L KH2PO4, 0.4 g/L MgSO4 and 100 mg/ml ampicillin.Precultures grew overnight and were used to inoculate theshake flasks. The volume of preculture was adjusted, suchthat the shake flasks had equal initial cell densities. Shakeflasks were induced with 5mM IPTG, unless otherwiseindicated.

2.3. Fed-batch fermentation

Fermentations were conducted in a 2L Bioflo 110fermenter (New Brunswick Scientific, Edison, NJ). Definedbatch medium was used (Korz et al., 1995), except thebatch medium contained 5 g/L glucose, 8 g/L KH2PO4,0.4 g/L MgSO4 and 100 mg/ml ampicillin. The feed mediumcontained 481 g/L glucose, 4 g/L MgSO4, 40mg/L FeIIIci-trate, and 1.5� the trace metal concentration of the batchmedium. Stock cultures stored at �80 1C were thawed andused to inoculate 100ml of defined media. Precultures weregrown at 37 1C to a cell density (OD600) of approximately

ARTICLE IN PRESS

Fig. 1. Growth characteristics for E. coli MDS40 and MG1655 cultured in

shake flasks. (A) Cell densities for MDS40—induced (’) and uninduced

(&); E. coli MG1655—induced (m) and uninduced (n). (B) Specific CAT

activity for E. coli MDS40—induced (’) and uninduced (&); E. coli

MG1655—induced (m) and uninduced (n).

S.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141 135

0.8, and 75ml was used to inoculate 1.5 L of batch medium(5% v/v) in the fermenter. The pH and temperature duringthe fermentations were maintained at 6.75 and 37 1C,respectively. Air was sparged into the fermenter at aconstant rate and the dissolved oxygen (DO) was main-tained at 30% of saturation by varying the agitation rate.The batch phase of the fermentation was approximately9.5 h, at which time the glucose in the batch medium wascompletely consumed. Feeding was initiated at an expo-nential rate to control the growth rate (Akesson et al.,1999; DeLisa et al., 1999). Fermentations proceeded untilthe agitation rate reached its maximum value (1200 rpm)and the DO could no longer be maintained at 30%saturation. Fed-batch fermentations were induced withIPTG (1mM), unless otherwise indicated.

2.4. Analytical techniques

The glucose was measured using a OneTouch Basicblood glucose meter. Samples for dry cell weight and CATactivity were cooled on ice, and then centrifuged at 3500g

for 10min at 4 1C. The supernatant was saved (�20 1C) foracetate and formate analysis. The cell pellets for the dry cellweights were resuspended in water and re-centrifuged.Washed cell pellets were resuspended in a small volume ofwater, placed in a pre-weighed aluminum pan and dried at80 1C. For the CAT samples, the cell pellets wereresuspended in 50mM Tris buffer (pH 7.4) containing30 mM dithiothreitol (TDTT buffer), re-centrifuged, andthe washed pellets frozen at �20 1C for analysis later. Thefrozen cell pellets were thawed, resuspended in TDTTbuffer, sonicated on ice (Branson Sonifier 250, 3� 15 spulses) and centrifuged at 17,000g for 20min at 4 1C. Thecell lysate supernatant was collected and stored at �20 1C.CAT activity was measured using the kinetic assaydescribed by Rodriguez and Tait (1983), adapted to a 96-well plate format. Acetic acid and formate concentrationswere enzymatically determined (R-Biopharm), in a 96-wellplate format. Acetate and formate samples were centri-fuged at 17,000g for 18min, to remove particulates prior toanalysis.

3. Results and discussion

3.1. Shake flask cultivation

Growth characteristics and recombinant protein produc-tion for E. coli MDS40 were first investigated in parallelshake flasks in minimal media with E. coli MG1655 as thecontrol. Both uninduced and induced cultures wereassessed. The shake flask growth curves for uninducedand induced E. coli MG1655 and MDS40 are shown inFig. 1. The growth rates, as well as the final cell densitieswere comparable between the two strains. IPTG was addedto one pair of cultures at 12 h to induce CAT expression.Induction did not significantly impact the growth rate ofeither strain. The CAT expression profiles are also shown

in Fig. 1. The specific CAT activity increased rapidly afterinduction for both strains, remained level for a few hoursand decreased during the stationary phase. The observedmaximum specific CAT activity was comparable for E. coli

MDS40 and MG1655. The E. coli MG1655 and MDS40CAT expression profiles were similar to other reports, evenwith strain, plasmid and media differences (Cha et al.,2000; Gill et al., 1998; Haddadin and Harcum, 2005;Harcum and Bentley, 1999; Richins et al., 1997). Thus,E. coli MDS40 was capable of good growth and CATexpression under shake flask conditions.

3.2. Fed-batch fermentation

Since E. coli MDS40 was capable of growth rates andrecombinant protein production comparable to the parentstrain in shake flasks, the next step was to assess growthcharacteristics and recombinant protein expression underindustrial fermentation conditions. Fed-batch fermenta-tions were used to assess E. coli MDS40 at three different

ARTICLE IN PRESSS.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141136

the fed-batch growth rates: 0.15 h�1 (low), 0.25 h�1

(intermediate), and 0.5 h�1 (high), where the batch phasegrowth rates were not different. The parent strain, E. coli

MG1655, was only assessed at 0.25 h�1, as the behavior ofE. coli MG1655 and other related strains has been wellinvestigated with respect to recombinant protein produc-tivity at these controlled growth rates (Curless et al., 1990;Hellmuth et al., 1994; Sanden et al., 2003; Turner et al.,1994). The growth curves for E. coli MG1655 and MDS40are shown in Fig. 2. The growth rates in the batch phasewere approximately 0.45 h�1 for both strains. All batch

Fig. 2. Growth characteristics for E. coli MDS40 and MG1655 at controlled

IPTG (’), 5mM IPTG (.), and uninduced (&) at the low growth rate (0.15 h

at the intermediate growth rate (0.25 h�1). Cell densities for E. coli MG1655—

shown. (C) Cell densities for E. coli MDS40—induced with 1mM IPTG (E), 5

Specific CAT activity for E. coli MDS40 induced with 1mM IPTG (’) and 5m

MDS40 (K) and E. coli MG1655 (m) at the intermediate growth rate. (F) Sp

5mM IPTG (.) at the high growth rate.

phases lasted between 9.5 and 10 h, at which point theinitial glucose was completely consumed. The dry cellweights obtained for E. coli MG1655 and E. coli MDS40 atthe end of the batch phase were not significantly different(2.7270.71 and 2.1670.25 g/L, respectively). Thus, thestrains have equal substrate yield coefficients.The specific CAT activity profiles for both strains are

shown in Fig. 2. For all fermentations the specific CATactivity increased rapidly due to induction, leveled off, thendecreased for later fermentation times. The intermediatefed-batch growth rate fermentations for E. coli MDS40 and

growth rates: (A) Cell densities for E. coli MDS40—induced with 1mM�1). (B) Cell densities for E. coli MDS40—induced (K) and uninduced (J)

induced (m) and uninduced (n) at the intermediate growth rate are also

mM IPTG (.), and uninduced (B) at the high growth rate (0.5 h�1). (D)

M IPTG (.) at the low growth rate. (E) Specific CAT activity for E. coli

ecific CAT activity for E. coli MDS40 induced with 1mM IPTG (E) and

ARTICLE IN PRESS

Fig. 3. Acetate concentrations in the fermentation broth: (A) E. coli

MDS40—induced with 1mM IPTG (’), 5mM IPTG (.), and uninduced

(&) at the low growth rate (0.15 h�1). (B) E. coli MDS40 induced (K) and

uninduced (J) at the intermediate growth rate (0.25 h�1). E. coli

MG1655—induced (m) and uninduced (n). (C) E. coli MDS40—induced

with 1mM IPTG (E), 5mM IPTG (.), and uninduced (B) at the high

growth rate (0.5 h�1).

S.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141 137

MG1655 were conducted in duplicate and show thebiological variability in CAT expression for the well-controlled fed-batch fermentations. The multiple deletionstrain produced a comparable quantity of the recombinantprotein at intermediate fed-batch growth rate, althoughslightly less CAT when normalized to total protein. Asexpected, the specific CAT activity was lower for the highfed-batch growth rate culture. At the lower fed-batchgrowth rate, the specific CAT activity remained elevatedfor a longer time period. Both of these expressioncharacteristics are similar to what has been observed forE. coli MG1655 under similar conditions (Hoffman et al.,2002; Koo and Park, 1999; Levisauskas et al., 2003;Riesenberg et al., 1990) (Fig. 3).

Since the genome of the E. coli MDS40 strain is smaller,it was hypothesized that the controlled glucose feed ratemight result in higher biomass levels due to improvedsubstrate yield coefficients for E. coli MDS40 compared toE. coli MG1655. Instead, there were no measurable orsignificant difference in the biomass levels or substrateyields from glucose; however, we did observe slightlyhigher acetate levels during the batch phase for E. coli

MDS40. For the low and intermediate fed-batch growthrates, the accumulated acetate was consumed by E. coli

MDS40, and both strains, respectively, after the feedstarted. Additionally, acetate did not accumulate in the fedbatch phase for the low and intermediate fed-batchfermentations. However, for the high growth rate(0.5 h�1) fed-batch fermentations, acetate accumulated inthe batch phase and continued to accumulate in the fed-batch phase. Acetate accumulation is often observed incultures where the growth rate is greater than 0.3 h�1 forE. coli MG1655 (Lee, 1996; Swartz, 1996). Induction didnot significantly affect acetate accumulation for eitherstrain under the well-controlled growth conditions.

Acetate accumulation is considered an indicator ofoverflow metabolism or anaerobic growth conditions(termed mixed acid metabolism). Formate also accumu-lates during mixed acid conditions (Han et al., 1992; Lee,1996); thus the formate levels were measured for thefed-batch fermentations. The formate concentrationsfor all three fed-batch growth rates are shown in Fig. 4.Unlike acetate, which accumulated in the batch phase,and then was consumed in the fed batch phase, formatewas nearly undetectable in the batch phase. Formateaccumulated in the fed-batch phase for all three fed-batchgrowth rates. At the intermediate fed-batch growth ratethe two strains had nearly identical formate profiles. Theinduced cultures accumulated more formate than theuninduced cultures for both strains. For the low andhigh fed-batch growth rates, the formate profiles wereindependent of the induction strength. These formateprofiles indicate that mixed acid metabolism was not thesource of the accumulated acetate (Bylund et al., 2000;Castan et al., 2002; Xu et al., 1999). Thus, the slightlyhigher acetate levels observed for E. coli MDS40 in thebatch phase, where growth rates were higher, is likely due

to overflow metabolism. Very precise metabolism studiesare needed to determine if E. coli MDS40 reach overflowmetabolism more readily than E. coli MG1655. If E. coli

ARTICLE IN PRESS

Fig. 4. Formate concentrations in the fermentation broth: (A) E. coli

MDS40—induced (’) and uninduced (&) at the low growth rate

(0.15 h�1). (B) E. coli MDS40 induced (K) and uninduced (J) at the

intermediate growth rate (0.25 h�1). E. coli MG1655—induced (m) and

uninduced (n). (C) E. coli MDS40—induced (E) and uninduced (B) at

the high growth rate (0.5 h�1).

Fig. 5. Growth characteristics for E. coli MDS40 stressed by acetate in

shake flasks: (A) Cell densities for cultures with the acetate addition (K)

and no acetate addition (’). (B) Specific CAT activity for cultures with

the acetate addition (K) and no acetate addition (’). Error bars indicate

95% confidence intervals.

S.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141138

MDS40 reaches overflow metabolism conditions morereadily, it might be possible to redirect the excess carboninto recombinant protein by either fermenter control orgenetic modifications.

It has been reported that acetate accumulation lowerscell yields and recombinant protein productivity (Akessonet al., 1999; Diaz-Ricci et al., 1991; Farmer and Liao, 1997;Lee, 1996; Takahashi et al., 1999; Turner et al., 1994).Thus, the poorer recombinant protein productivity at thehigh fed-batch growth rate was hypothesized to be due tothe slightly higher acetate concentration. To test thishypothesis, acetate was added to duplicate E. coli MDS40shake flasks. Duplicate E. coli MDS40 shake flasks,without acetate addition, were also assessed to provide abaseline for this strain. Sodium acetate was added to theshake flasks at just under 1 OD. All four flasks wereinduced to produce CAT at approximately 7 h. The growthcurves for all four cultures are shown in Fig. 5. Asexpected, the growth rate of the cultures exposed toadditional acetate was lower. Unexpectedly, the specificCAT activities for the acetate stressed cultures were higherthan the control (unstressed) cultures as shown in Fig. 5.The shake flask experiment demonstrated that the presenceof acetate alone does not cause reduced recombinant

ARTICLE IN PRESSS.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141 139

protein productivity; however, it demonstrates that theconditions that favor the production of acetate by the cellsdoes lower recombinant protein productivity. Thus, it ispossible that the lower observed specific CAT activity inthe high fed-batch growth rate fermentation cultures couldhave been due to redirection of the carbon resources fromthe recombinant protein to acetate, due to overflowmetabolism, as seen with E. coli MG1655. The effect ofgrowth rate on recombinant protein expression has beenaddressed in literature for strains related to E. coli

MG1655, and the E. coli MDS40 strain behavior wasconsistent with respect to productivity and growth rates(Hoffman et al., 2002; Koo and Park, 1999; Levisauskaset al., 2003; Riesenberg et al., 1990).

Many expression systems are available to producerecombinant proteins in E. coli. Some expression systemshave extremely powerful promoters which have beenobserved to stress the cells or cause a metabolic burden(Glick, 1995; Haddadin and Harcum, 2005; Hoffman et al.,2002; Sanden et al., 2003; Wood and Peretti, 1990). Toassess the effect of strong promoters on E. coli MDS40, theeffect of changing the induction strength was investigated.It has been previously observed that very high inductionstrengths (45mM) can overwhelm E. coli JM105, anMG1655 derivative (Harcum et al., 1992). Since the lowand high fed-batch growth rate cultures provided a broaderrange of expression, these cultures were examined at both1mM (normal) and 5mM (high) induction strengths. Asexpected, the high induction strength resulted in higherspecific CAT activity (Fig. 2) for both the low and high fed-batch growth rate fermentations, although not proportion-ally higher. A non-proportional increase in recombinantprotein expression with respect to induction strength hasbeen observed in other E coli strains (Bentley et al., 1991;Ramirez and Bentley, 1995). Interestingly, the inductionstrength had less effect on recombinant protein productionfor the low fed-batch growth rate culture.

Although E. coli is a well studied and a commonly usedrecombinant host, the behavior of E. coli can still beunpredictable at times, especially while expressing recom-binant protein (Andersson et al., 1996; Baneyx, 1999;Baneyx and Mujacic, 2004; Gill et al., 2000; Oh and Liao,2000; Sanden et al., 2003; Swartz, 1996). One of the overallgoals of the multiple deletion strain project is to improvethe growth/yield properties and robustness of E. coli as arecombinant host by eliminating potentially non-essentialgenes. A short-term goal of this project and the focus ofthis study was to assess the capability of one multipledeletion strain (E. coli MDS40) under industrial condi-tions. Since conditions in a fermenter, especially high celldensities, are significantly different from those in shakeflasks, both of these conditions were examined (Anderssonet al., 1996; Gill et al., 2001; Humphrey, 1998; Lee, 1996;Yoon et al., 2003). Under the conditions assessed E. coli

MDS40 and its parent strain E. coli MG1655 had similarperformance characteristics. The similar behavior betweenthe two strains is encouraging as E. coli MDS40 was

derived from E. coli MG1655 by 40 large sequentialdeletions encompassing nearly 700 genes. These deletionsincluded all K-islands, numerous hypothetical operons, alltransposable genetic elements, many genes related to thesynthesis of flagella and fimbrae, and some lippopolysac-charide (LPS) genes. A disproportionate number of genesof unknown function were located within K-islands, thusthe consequences of removing these genes on growth andrecombinant protein production were unknown. An under-lying hypothesis of this project has been that if thetranscription and translation of these unknown genes doesnot occur, the number of endogenous proteins in E. coli

MDS40 should be less than E. coli MG1655. This lowernumber of endogenous proteins should have two positiveeffects on recombinant protein production: (1) reducedenergy requirements due to fewer proteins being synthe-sized, and (2) more efficient protein purification due tofewer endogenous/native proteins to remove. For example,in the case of flagella, the energy saving may be two-fold:(1) reduced energy needs due to eliminated synthesis of theflagella proteins, and (2) reduced energy needs associatedwith the flagella operation. An additional benefit of themultiple deletion strain includes removal of mobile geneticelements, such as insertion sequences and transposons.Deletion of these mobile elements provides some significantgenome and plasmid stabilization (Posfai et al., 2006),which could improve strain stability during long fermenta-tion runs with transient stresses (Cavin et al., 1999; Faureet al., 2004; Glick, 1995). The slightly higher acetateaccumulation in the batch phase observed for E. coli

MDS40 may indicate that overflow metabolism can occurmore easily in the deletion strain; however, more precisemetabolic studies are necessary to clarify this point.Mathematical models of E. coli have demonstrated higheryields of certain metabolites as a result of gene deletions(Burgard et al., 2003); however, these models did notaccount for the specific genes deleted to create E. coli

MDS40, due to lack of information regarding these genes.In all, the multiple deletion strain performs comparably toits parent strain under industrial conditions. This deletionstrain provides a sound basis for further deletions thattarget the specific needs of a particular industrial fermenta-tion. Additionally, further analysis is required to quantifythe energy and nutrient efficiencies due to reduction ofendogenous proteins.

4. Conclusion

This study was conducted to assess the ability of themultiple deletion strain—E. coli MDS40—to producerecombinant protein under industrial conditions. E. coli

MDS40 was generated by the targeted sequential deletion of40 regions of the E. coli MG1655 genome. The E. coli

MDS40 genome is 14% smaller than MG1655 withapproximately 700 fewer genes. The targeted genes includedall K-islands, flagella and fimbrae genes, and some LPSsynthesis genes. K-islands contain a disproportionate

ARTICLE IN PRESSS.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141140

number of genes with unknown function, hypotheticalgenes, transposons and insertion sequences. The multipledeletions did not affect the ability of E. coli MDS40 to growin defined minimal medium to high cell densities atcontrolled growth rates. The growth characteristics of theparental strain (E. coli MG1655) and E. coli MDS40 werenot significantly different. Recombinant protein productiv-ity for the E. coli MDS40 strain was also comparable to theparent strain. A complete analysis of the cell compositionwill provide a better understanding as to how the multipledeletions affect protein synthesis and cell energetics.

Acknowledgments

We would like to thank Dr. Gyorgy Posfai forconstructing the E. coli MDS40 strain, and Dr. JohnCampbell for reviewing the manuscript. Financial supportwas provided to Clemson University by Scarab Genomics,which was funded by NIH Grant GM35682 (FRB).

References

Akesson, M., Karlsson, E., Hagander, P., Axelsson, J., Tocaj, A., 1999.

On-line detection of acetate formation in Escherichia coli cultures using

dissolved oxygen responses to feed transients. Biotechnol. Bioeng. 64,

590–598.

Andersen, D.C., Krummen, L., 2002. Recombinant protein expression for

therapeutic applications. Curr. Opin. Biotechnol. 13, 117–123.

Andersen, D.C., Swartz, J., Ryll, T., Lin, N., Snedecor, B., 2001.

Metabolic oscillations in an Escherichia coli fermentation. Biotechnol.

Bioeng. 75, 212–218.

Andersson, L., Strandberg, L., Enfors, S.O., 1996. Cell segregation and

lysis have profound effects on the growth of Escherichia coli in high

density fed batch cultures. Biotechnol. Prog. 12, 190–195.

Baneyx, F., 1999. Recombinant protein expression in Escherichia coli.

Curr. Opin. Biotechnol. 10, 411–421.

Baneyx, F., Mujacic, M., 2004. Recombinant protein folding and

misfolding in Escherichia coli. Nat. Biotechnol. 22, 1399–1408.

Bentley, W.E., Davis, R.H., Kompala, D.S., 1991. Dynamics of induced

CAT expression in Escherichia coli. Biotechnol. Bioeng. 38, 749–760.

Bessette, P.H., Aslund, F., Beckwith, J., Georgiou, G., 1999. Efficient

folding of proteins with multiple disulfide bonds in the Escherichia coli

cytoplasm. Proc. Natl. Acad. Sci. USA 96, 13703–13708.

Blattner, F.R., Plunkett, G., Bloch, C.A., Perna, N.T., Burland, V., Riley,

M., ColladoVides, J., Glasner, J.D., Rode, C.K., Mayhew, G.F.,

Gregor, J., Davis, N.W., Kirkpatrick, H.A., Goeden, M.A., Rose,

D.J., Mau, B., Shao, Y., 1997. The complete genome sequence of

Escherichia coli K-12. Science 277, 1453–1474.

Burgard, A.P., Pharkya, P., Maranas, C.D., 2003. OptKnock: a bilevel

programming framework for identifying gene knockout strategies for

microbial strain optimization. Biotechnol. Bioeng. 84, 647–657.

Bylund, F., Castan, A., Mikkola, R., Veide, A., Larsson, G., 2000.

Influence of scale-up on the quality of recombinant human growth

hormone. Biotechnol. Bioeng. 69, 119–128.

Castan, A., Nasman, A., Enfors, S.O., 2002. Oxygen enriched air supply in

Escherichia coli processes: production of biomass and recombinant

human growth hormone. Enzyme Microb. Technol. 30, 847–854.

Cavin, J.F., Dartois, V., Labarre, C., Divies, C., 1999. Cloning of

branched chain amino acid biosynthesis genes and assays of p-

acetolactate synthase activities in Leuconostoc mesenteroides subsp.

cremoris. Res. Microbiol. 150, 189–198.

Cebolla, A., Royo, J.L., de Lorenzo, V., Santero, E., 2002. Improvement

of recombinant protein yield by a combination of transcriptional

amplification and stabilization of gene expression. Appl. Environ.

Microbiol. 68, 5034–5041.

Cha, H.J., Wu, C.F., Valdes, J.J., Rao, G., Bentley, W.E., 2000.

Observations of green fluorescent protein as a fusion partner in

genetically engineered Escherichia coli: monitoring protein expression

and solubility. Biotechnol. Bioeng. 67, 565–574.

Chen, Y., Song, J., Sui, S.-f., Wanga, D.-N., 2003. DnaK and DnaJ

facilitated the folding process and reduced inclusion body formation of

magnesium transporter CorA overexpressed in Escherichia coli.

Protein Expres. Purif. 32, 221–231.

Chevalet, L., Robert, A., Gueneau, F., Bonnefoy, J.Y., Nguyen, T.,

2000. Recombinant protein production driven by the tryptophan

promoter is tightly controlled in ICONE 200: a new genetically

engineered Escherichia coli mutant. Biotechnol. Bioeng. 69,

351–358.

Chou, C.-H., Bennett, G.N., San, K.-Y., 1996. Genetic manipulation of

stationary-phase genes to enhance recombinant protein production in

Escherichia coli. Biotechnol. Bioeng. 50, 636–642.

Curless, C., Pope, J., Tsai, L., 1990. Effect of preinduction specific growth

rate on recombinant alpha consensus interferon synthesis in Escher-

ichia coli. Biotechnol. Prog. 6, 149–152.

DeLisa, M.P., Li, J., Rao, G., Weigand, W.A., Bentley, W.E., 1999.

Monitoring GFP-operon fusion protein expression during high cell

density cultivation of Escherichia coli using an on-line optical sensor.

Biotechnol. Bioeng. 65, 54–64.

Diaz-Ricci, J.C., Regan, L., Bailey, J.E., 1991. Effect of alteration of the

acetic acid synthesis pathway on the fermentation pattern of

Escherichia coli. Biotechnol. Bioeng. 38, 1318–1324.

Farmer, W.R., Liao, J.C., 1997. Reduction of aerobic acetate production

by Escherichia coli. Appl. Environ. Microbiol. 63, 3205–3210.

Faure, D., Frederick, R., Wloch, D., Portier, P., Blot, M., Adams, J.,

2004. Genomic changes arising in long-term stab cultures of

Escherichia coli. J. Bacteriol. 186, 6437–6442.

Gill, R.T., Cha, H.J., Jain, A., Rao, G., Bentley, W.E., 1998. Generating

controlled reducing environments in aerobic recombinant Escherichia

coli fermentations: effects on cell growth, oxygen uptake, heat shock

protein expression, and in vivo CAT activity. Biotechnol. Bioeng. 59,

248–259.

Gill, R.T., Valdes, J.J., Bentley, W.E., 2000. A comparative study of

global stress gene regulation in response to overexpression of

recombinant protein in Escherichia coli. Metab. Eng. 2, 178–189.

Gill, R.T., Delisa, M.P., Valdes, J.J., Bentley, W.E., 2001. Genomic

analysis of high cell density recombinant Escherichia coli fermentation

and ‘‘cell conditioning’’ for improved recombinant protein yield.

Biotechnol. Bioeng. 72, 85–95.

Glick, B.R., 1995. Metabolic load and heterologous gene-expression.

Biotechnol. Adv. 13, 247–261.

Goryshin, I.Y., Naumann, T.A., Apodaca, J., Reznikoff, W.S., 2003.

Chromosomal deletion formation system based on Tn5 double

transposition: use for making minimal genomes and essential gene

analysis. Genome Res. 13, 644–653.

Haddadin, F.T., Harcum, S.W., 2005. Transcriptome profiles for high-

cell-density recombinant and wild type Escherichia coli. Biotechnol.

Bioeng. 90, 127–153.

Han, K., Lim, H.C., Hong, J., 1992. Acetic acid formation in Escherichia

coli fermentation. Biotechnol. Bioeng. 39, 663–671.

Harcum, S.W., Bentley, W.E., 1993. Response dynamics of 26-, 34-, 39-,

54-, and 80-kDa proteases in Induced cultures of recombinant

Escherichia coli. Biotechnol. Bioeng. 42, 675–685.

Harcum, S.W., Bentley, W.E., 1999. Heat-shock and stringent responses

have overlapping protease activity in Escherichia coli: Implications for

heterologous protein yield. Appl. Biochem. Biotechnol. 80, 23–37.

Harcum, S.W., Ramirez, D.M., Bentley, W.E., 1992. Optimal nutrient

feed policies for heterologous protein production. Appl. Biochem.

Biotechnol. 34, 161–173.

Hashimoto, M., Ichimura, T., Mizoguchi, H., Tanaka, K., Fujimitsu, K.,

Keyamura, K., Ote, T., Yamakawa, T., Yamazaki, Y., Mori, H.,

Katayama, T., Kato, J.-i., 2005. Cell size and nucleoid organization of

ARTICLE IN PRESSS.S. Sharma et al. / Metabolic Engineering 9 (2007) 133–141 141

engineered Escherichia coli cells with a reduced genome. Mol.

Microbiol. 55, 137–149.

Hellmuth, K., Korz, D.J., Sanders, E.A., Deckwer, W.D., 1994. Effect of

growth rate on stability and gene expression of recombinant plasmids

during continuous and high cell density cultivation of Escherichia coli.

J. Biotechnol. 32, 289–298.

Hoffman, F., Weber, J., Rinas, U., 2002. Metabolic adaptation of

Escherichia coli during temperature-induced recombinant protein

production: 1. Readjustment of metabolic enzyme synthesis. Biotech-

nol. Bioeng. 80, 313–319.

Humphrey, A., 1998. Shake flask to fermentor: what have we learned?

Biotechnol. Prog. 14, 3–7.

Khlebnikov, A., Skaug, T., Keasling, J.D., 2002. Modulation of gene

expression from the arabinose-inducible araBAD promoter. J. Ind.

Microbiol. Biotechnol. 29, 34–37.

Kolisnychenko, V., Plunkett, G., Herring, C.D., Feher, T., Posfai, J.,

Blattner, F.R., Posfai, G., 2002. Engineering a reduced Escherichia coli

genome. Genome Res. 12, 640–647.

Koo, T.Y., Park, T.H., 1999. Increased production of recombinant

protein by Escherichia coli deficient in acetic acid formation.

J. Microbiol. Biotechnol. 9, 789–793.

Korz, D.J., Rinas, U., Hellmuth, K., Sanders, E.A., Deckwer, W.D., 1995.

Simple fed-batch technique for high cell-density cultivation of

Escherichia coli. J. Biotechnol. 39, 59–65.

Lee, S.Y., 1996. High cell-density culture of Escherichia coli. Trends

Biotechnol. 14, 98–105.

Levisauskas, D., Galvanauskas, V., Henrich, S., Wilhelm, K., Volk, N.,

Lubbert, A., 2003. Model-based optimization of viral capsid protein

production in fed-batch culture of recombinant Escherichia coli.

Bioproc. Biosyst. Eng. 25, 255–262.

Oh, M.-K., Liao, J.C., 2000. DNA microarray detection of metabolic

responses to protein overproduction in Escherichia coli. Metab. Eng. 2,

201–209.

Pecota, D.C., Kim, C.S., Wu, K., Gerdes, K., Wood, T.K., 1997.

Combining the hok/sok, parDE, and pnd postsegregational killer

loci to enhance plasmid stability. Appl. Environ. Microbiol. 63,

1917–1924.

Posfai, G., Plunkett III, G., Feher, T., Frisch, D., Keil, G.M.,

Umenhoffer, K., Kolisnychenko, V., Stahl, B., Sharma, S.S., de

Arruda, M., Burland, V., Harcum, S.W., Blattner, F.R., 2006.

Emergent properties of reduced-genome Escherichia coli. Science 312,

1044–1046.

Ramirez, D.M., Bentley, W.E., 1995. Fed-Batch feeding and induction

policies that improve foreign protein synthesis and stability by

avoiding stress responses. Biotechnol. Bioeng. 47, 596–608.

Richins, R., Htay, T., Kallio, P., Chen, W., 1997. Elevated Fis expression

enhances recombinant protein production in Escherichia coli. Biotech-

nol. Bioeng. 56, 138–144.

Riesenberg, D., Menzel, K., Schulz, V., Schumann, K., Veith, G., Zuber,

G., Knorre, W.A., 1990. High cell-density fermentation of recombi-

nant Escherichia coli expressing human interferon. Appl. Microbiol.

Biotechnol. 34, 77–82.

Rodriguez, R.L., Tait, R.C., 1983. Recombinant DNA Techniques: An

Introduction. Addison-Wesley, Reading, MA.

Sambrook, J.F., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: A

Laboratory Manual. Cold Spring Harbor, New York.

Sanden, A.M., Prytz, I., Tubulekas, I., Forberg, C., Le, H., Hektor, A.,

Neubauer, P., Pragai, Z., Harwood, C., Ward, A., Picon, A.,

deMattos, J.T., Postma, P., Farewell, A., Nystrom, T., Reeh, S.,

Pedersen, S., Larsson, G., 2003. Limiting factors in Escherichia coli

fed-batch production of recombinant proteins. Biotechnol. Bioeng. 81,

158–166.

Sorensen, H.P., Mortensen, K.K., 2005. Advanced genetic strategies for

recombinant protein expression in Escherichia coli. J. Biotechnol. 115,

113–128.

Swartz, J.R., 1996. Escherichia coli recombinant DNA technology. In:

Neidhardt, F.C., Curtiss, R., Lin, E.C.C., Low, K.B., Magasanik, B.,

Reanikoff, W.S., Riley, M., Schaechter, M., Umbarger, H.E. (Eds.),

Escherichia coli and Salmonella, vol. 2. ASM Press, Washington, DC,

pp. 1693–1712.

Takahashi, C.M., Takahashi, D.F., Carvalhal, M.L.C., Alterthum, F.,

1999. Effects of acetate on the growth and fermentation performance

of Escherichia coli KO11. Appl. Biochem. Biotechnol. 81, 193–203.

Turner, C., Gregory, M.E., Turner, M.K., 1994. A study of the effect of

specific growth rate and acetate on recombinant protein production of

Escherichia coli JM107. Biotechnol. Lett. 16, 891–896.

Wood, T.K., Peretti, S.W., 1990. Depression of protein synthetic capacity

during cloned-gene expression in Escherichia coli. Biotechnol. Bioeng.

36, 865–878.

Xu, B., Jahic, M., Blomsten, G., Enfors, S.O., 1999. Glucose overflow

metabolism and mixed acid fermentation in aerobic large scale fed

batch processes with Escherichia coli. Appl. Microbiol. Biotechnol. 51,

564–571.

Yoon, S.H., Han, M.-J., Lee, S.Y., Jeong, K.J., Yoo, J.-S., 2003.

Combined transcriptome and proteome analysis of Escherichia coli

during high cell density culture. Biotechnol. Bioeng. 81, 753–767.

Yu, B.J., Sung, B.H., Koob, M.D., Lee, C.H., Lee, J.H., Lee, W.S., Kim,

M.S., Kim, S.C., 2002. Minimization of the Escherichia coli genome

using a Tn5-targeted Cre/loxP excision system. Nat. Biotechnol. 20,

1018–1023.