Vol. 56, No. 5APPLIED AND ENVIRONMENTAL MICROBIOLOGY, May 1990, p. 1296-13020099-2240/90/051296-07$02.00/0Copyright C) 1990, American Society for Microbiology

Improved Expression of Human Interleukin-2 in High-Cell-DensityFermentor Cultures of Escherichia coli K-12 by a

Phosphotransacetylase MutantKEITH A. BAUER, ARIE BEN-BASSAT, MIKE DAWSON, VIRGINIA T. DE LA PUENTE,

AND JUSTIN 0. NEWAY*

Department of Fermentation Research and Development, Cetus Corporation, 1400 Fifty-third Street,Emeryville, California 94608

Received 13 November 1989/Accepted 12 February 1990

A fluoroacetate-resistant mutant of Escherichia coli K-12 (MM-294) accumulated less acetate in the mediumduring growth to high cell density in fermentor cultures and was shown to be defective in its phosphotrans-acetylase activity. The mutant had an improved ability to continue growing during induction of interleukin-2(IL-2) synthesis, and in fermentor cultures it gave a higher level of specific IL-2 accumulation than its parentduring expression under control of the temperature-sensitive PL promoter. In flask cultures at lower celldensity, the mutant again produced less acetate than the parent, although both showed a much lower level ofacetate accumulation than that seen in fermentors at high cell density. Both showed a higher specific expressionlevel of IL-2 in flask cultures, and there was a greater difference between the mutant and its parent in the finalextent of specific IL-2 accumulation in fermentor cultures compared with flask cultures. Thus, the concentra-tion of acetate in the medium, which was much higher in fermentor cultures (.300 mM after 5 h of induction)than in flask cultures ('3 mM) of the parent organism, was a significant factor in limiting expression of theheterologous protein product, IL-2. The acetate kinase-phosphotransacetylase pathway was therefore a majorsource of acetate formation in these cultures. Blocking this pathway improved accumulation of IL-2 and did notslow growth.

Organic acids accumulate in the culture medium duringaerobic growth of Escherichia coli on glucose (17). The mostabundant organic acid is often acetic, and its concentrationcan build up to levels that are inhibitory to growth (1, 13).In a previous study (10), we showed that intracellularaccumulation of interleukin-2 (IL-2), under control of thetemperature-sensitive bacteriophage lambda PL promoter,was inversely correlated with cell density and acetate accu-mulation in fermentor cultures. These observations providedcircumstantial evidence that acetate was at least partiallyresponsible for the cessation of product accumulation duringexpression of heterologous genes in E. coli and suggestedthat higher levels of IL-2 accumulation could be expected ifacetate formation could be blocked.Two enzymatic pathways for acetate formation in E. coli

have been identified. Acetate can be derived directly frompyruvate by pyruvate oxidase, but the activity of this en-zyme in E. coli is thought to be too low to account for theamount of acetate produced (4). Acetate can also be derivedfrom acetylcoenzyme A (CoA) by the acetate kinase-phos-photransacetylase (ACK-PTA) pathway (3). In E. coli B,PTA is activated by pyruvate and inhibited by NADH (14).The genes coding for PTA and ACK have been mapped andform an operon in E. coli and Salmonella typhimurium whichis induced as much as twofold under anaerobic conditions(8). Mutants defective in both activities can be isolated byselection for fluoroacetate resistance (3).

In the work presented here, we examined the role of theACK-PTA pathway in the formation of acetate during intra-cellular accumulation of IL-2. We isolated 52 fluoroacetate-resistant mutants from E. coli MM294-1 and partially char-

* Corresponding author.

acterized them. Some of these mutants had reduced acetate-forming ability, as low as 10 to 20% of parental levels in flaskcultures, with near-normal aerobic growth rates on glucose.We describe here the characterization of one of thesemutants and its performance as a host strain for PL promot-er-driven accumulation of IL-2.

MATERIALS AND METHODSStrains. All strains used in this study were derived from E.

coli K-12 and are listed in Table 1. Phage stocks used fortesting c1857 function were the gift of D. Gelfand. HW21 wasthe gift of H. C. Wong. The poxB strains were the gift ofJ. E. Cronan and the pta-ack deletion strains were the gift ofG. F.-L. Ames.Media. N8-2 medium consisted of NH4Cl (10 mM),

KH2PO4 (21.9 mM), Na2HPO4 (28.1 mM), K2SO4 (9 mM),MgSO4 (0.2 mM), MnSO4 (3 ,M), ZnSO4 (3 ,uM), andCuSO4 (0.1 ,uM) in deionized water. The medium wassterilized by autoclaving, after which the following sterileadditions were made: glucose (2 g/liter); thiamine hydrochlo-ride (10 mg/liter); FeSO4 (10 ,uM).Minimal agar plates were of similar composition to defined

flask medium (see below), except that the trace metalssolution was 1 ml/liter, MgSO4 was at 1 mM, and agar wasadded to 1.5% (wt/vol). Alternate carbon sources (galactoseor sodium pyruvate) were also used at 5 g/liter. Ampicillin(50 mg/liter) and tetracycline (15 mg/liter) were used asindicated. Sodium monofluoroacetate (Tull Chemical Co.,Inc., Oxford, Ala.) was added to 25 mM as indicated.R2-4 agar plates were standard rich-medium plates de-

scribed elsewhere (10).Isolation plates were composed of Na2HPO4 (5.68 g/liter),

KH2PO4 (3.54 g/liter), trisodium citrate. 2H20 (0.44 g/liter),a solution containing ZnSO4 (30 mM), MnSO4 (30 mM), and

1296

on May 12, 2018 by guest

http://aem.asm

.org/D

ownloaded from

PTA- MUTANT OF E. COLI SHOWING IMPROVED IL-2 EXPRESSION 1297

TABLE 1. Strains used

Strain Relevant characteristics Source

DG116 cI857+, derived from CMCCaMM294-1

HW21 Thi+ derivative of DG116 CMCCHW21(pFC54.t) TSb copy number (16), PL- This work

IL-2 (15) in HW21KB100 Thi+ derivative of MDO0O This workKBlOO(pFC54.t) TS copy number (16), PL- This work

IL-2 (15) in KB100MDO0O Pta- This workMDOOL Pta-, cI857+ (lysogen) This workMDO5OL(pFC54.t) TS copy number (16), PL- This work

IL-2 (15) in MDO5OLMH6 TnS inserted into poxB J. E. Cronan

(PoxB-)MM294-1 endAI hsdRJ7 supE thi, CMCC

derived from MM294 (11)MM294-1 Gal- strain for P1 D. Gelfand

galE: :TnJO transductionTA3514 pta G. F.-L. AmesTA3515 ack G. F.-L. AmesTA3516 A(ack, pta, hisP-hisQ) G. F.-L. AmesTA3521 Ack+ Pta+ G. F.-L. AmesYYC201 TnlO inserted near poxB J. E. Cronan

(PoxB+)a CMCC, Cetus Master Culture Collection.b TS, Temperature sensitive.

CuSO4 (30 mM) at 0.2 ml/liter; (NH4)2SO4 (2.0 g/liter),MgSO4 - 7H20 (102 mg/liter), and agar (1.5%, wt/vol). Theliquid was sterilized by autoclaving, and the following sterileadditions were made: FeSO4 (200 mM) in 2.5 mM H2SO4,0.36 ml/liter, CaCl2 (1.0 M), 0.1 ml/liter, thiamine hydrochlo-ride (1.0%), 2.0 ml/liter. Pyruvate, lactate, acetate, or glu-cose was added at 0.5% (wt/vol) as carbon source. Sodiummonofluoroacetate was added at a concentration of 25 mMalong with lactate or pyruvate to select for putative mutantswith reduced ability to produce acetate.

Defined flask medium consisted of trisodium citrate (4mM), (NH4)2SO4 (10 mM), KH2PO4 (50 mM), and tracemetals solution (2) (4 ml/liter). The pH of the medium was

adjusted to 6.8 with NaOH, and the solution was autoclaved.Glucose (5 g/liter), thiamine hydrochloride (20 mg/liter), andMgSO4 (3 mM) were added to the cooled medium fromsterile concentrated stock solutions.For fermentor medium, the medium of Bauer and Shiloach

(2) with some modifications was used as described previ-ously (10), except that 20 mg of thiamine hydrochloride perliter was added after autoclaving. The 14-liter Chemapfermentors as well as inoculation and operating conditionsused in this study were as described previously (10).

Assays. Assays for acetate, cell growth, and IL-2 inisolated refractile bodies, as well as the various conversionfactors used, were as described previously (10).

(i) Preparation of cell extracts. Cell extracts for assays ofPTA and ACK activities were prepared as follows: 100 ml ofa defined flask medium culture in exponential growth wascentrifuged at 3,000 x g for 10 min at 5°C and then washedtwice in 30 ml of 10 mM sodium phosphate buffer, pH 7.5 (atroom temperature), which included 10mM MgCl2 and 1 mMEDTA. The cells were suspended in the same buffer to a finaltotal protein concentration of 2 to 2.4 mg/ml (based onoptical density at 680 nm [OD680] converted to proteincontent) and sonicated on ice to completion (approximately5 min, microscopic observation). Sonic extracts were cen-

trifuged at 28,000 x g for 60 min at 5°C and stored at -20°Cprior to use.

(ii) PTA activity. PTA activity was measured by themethod of Brown et al. (3). In this assay, the conversion ofacetylphosphate to acetyl-CoA is coupled to NADH forma-tion through malate dehydrogenase and citrate synthase.The reaction mixture contained the following components in1 ml: Tris hydrochloride (pH 8.0 at room temperature), 100,umol; MgCl2, 5 ,umol; NAD+, 0.5 pumol; CoA, 0.5 ,umol;malate, 5 ,umol; acetylphosphate, 10 ,umol; malate dehydro-genase, 27.5 U; and citrate synthase, 6.75 U. One unit ofPTA activity was defined as the amount of enzyme requiredto allow the formation of 1 ,umol of NADH per min.

(iii) ACK activity. ACK activity was measured by themethod of Fox and Roseman (6). In this assay, ADP forma-tion from ATP and acetate is coupled to NADH consump-tion through pyruvate kinase and lactate dehydrogenase.The reaction mixture contained the following components in1 ml: PIPES [piperazine-N,N'-bis(2-ethanesulfonic acid)]-KOH (pH 7.0 at room temperature; Sigma Chemical Co.), 50,umol; glycerol, 100 mg; MgCl2, 5 ,umol; dithiothreitol, 1,umol; phosphoenolpyruvate, 7 ,umol; potassium acetate, 300,umol; ATP, 30 ,umol; NADH, 1 ,umol; pyruvate kinase, 50U; and lactate dehydrogenase, 10 U. One unit of ACKactivity was defined as the amount of enzyme required toallow the oxidation of 1 ,umol of NADH per min.For both enzyme assays, the reactions were started by

addition of extract and followed by measuring the change inA340 over time at room temperature, using a Hewlett-Packard HP8452A spectrophotometer equipped with en-zyme kinetics software. PTA activity was shown to becompletely dependent on the presence of both primarysubstrates, acetylphosphate and CoA. ACK activity wasdependent on the presence of ATP and was greatly reducedbut not eliminated when acetate was omitted. This wasprobably due to the presence of other ATPases in the crudeextracts. Boiling the extracts eliminated all activity in bothassays. Both assays were used in the linear range of enzymeactivity.

Genetic techniques. (i) Selection of low-acetate-producingstrains. The ACK-PTA pathway is bi-directional and allowsentry of acetate into central metabolism via acetyl-CoA.Mutants with low acetate-producing ability were thereforeselected by isolating strains resistant to the tricarboxylicacid cycle inhibitor fluoroacetate (3, 9). E. coli MM294-1 wasgrown in nutrient broth to a cell density of approximately 2.0OD680 (approximately 3 x 109 cells per ml). Cells were thenplated at approximately 2 x 107, 2 x 108, and 2 x 109 cellsper plate on isolation plates containing either lactate plusfluoroacetate or pyruvate plus fluoroacetate and incubated at37°C for 3 days. A total of 52 individual colonies were pickedfrom these plates, streaked for isolation onto the sameselective medium, and incubated at 37°C. Because there wasstill perceptible growth of carry-over wild-type cells on someplates, the 52 isolates were again picked from single colo-nies, streaked for a second time onto fresh plates containingthe same selective medium, and incubated at 37°C. Singlecolonies were again picked after 48 h, suspended in 1 ml ofsaline, and streaked for a third time onto fresh selectivemedium. The same isolates were also streaked onto mediumcontaining acetate as sole carbon source to test for differ-ences in ability to grow on acetate.

Strains were evaluated for acetate production and growthrate by inoculating 125-ml flasks containing 25 ml of pre-warmed N8-2 medium at an initial cell density of 0.05 OD680.Cultures were incubated at 37°C on a New Brunswick rotary

VOL. 56, 1990

on May 12, 2018 by guest

http://aem.asm

.org/D

ownloaded from

APPL. ENVIRON. MICROBIOL.

shaker at 200 rpm, and density readings were taken approx-imately every 0.5 h. At approximately 1.2 OD680, smallculture samples were taken for acetate analysis. Many of the52 isolates showed greatly reduced acetate accumulation,some as little as 10% of parental levels. Strain MD050 wereselected for further experimentation because it consistentlygave the lowest levels of acetate accumulation and had agrowth rate essentially indistinguishable from that of itsparent (approximately 0.6 h-').

(ii) P1 transduction. Genetic elements coding for markerssuch as c1857, Thi+, and poxB::Tn5 were transferred by P1transduction, essentially as described by Miller (12). Toselect for c1857, recipient strains were first made Gal-, usinga lysate grown on MM294:galE::TnJO. This intermediatewas then transduced back to Gal' by using a lysate grown onDG116 (a strain containing cI857 in the chromosome). Col-onies were screened for cotransduction of the tightly linkedc1857 marker by testing for lambda resistance at 30 and 41°C.To select for the Thi+ phenotype, strains were plated onminimal glucose agar plates without thiamine. The resultingcolonies were restreaked for purity and then grown inglucose-containing defined flask medium lacking thiamine toconfirm that their growth rate in this medium was identical tothat of the parent strain in thiamine-containing medium.

(iii) poxB-containing strains. To create a Pta- poxB doublemutant, the poxB::Tn5 mutation from MH6 was transducedinto KB100. To confirm that the poxB mutation had beentransferred, a lysate was grown on the putative doublemutant and used to transduce YYC201 to kanamycin resis-tance. The resulting transductants were shown to be poxB onpyruvate tetrazolium plates, as described by Chang andCronan (5).

RESULTS

Characterization of MD050. (i) PTA activity. Selection forfluoroacetate resistance results in mutations in at least twogenes, pta and ack (3, 9). MD050 was a presumptive Pta-mutant based on its reduced acetate secretion levels andslow growth rate on acetate-containing plates. To confirmthis, PTA and ACK activities were compared inMDO5OL(pFC54.t) and HW21(pFC54.t) grown at 30°C indefined medium. The two host strains were isogenic, exceptfor the lesion recovered by fluoroacetate selection; theywere both derived from the same MM294-1 parent. The onlyother difference between them was that HW21 was Thi+,allowing it to grow without the addition of thiamine to themedium. This change made no discernible difference toeither acetate formation or IL-2 expression in HW21 whencompared with its parent MM294-1 (data not shown). Alldefined media used here included thiamine.Three sets of extracts were prepared on separate days

from shake flask cultures grown at 30°C (i.e., repressed IL-2expression). The results (Table 2) clearly showed thatMDOOL had greatly reduced PTA activity. No PTA activitywas detected in the mutant in the first set of assays (detec-tion limit, <1% of wild type), and small amounts (1.5% or

less of wild type) were detected in the subsequent two tests.The low PTA activity in the second extract of MDO5OL(Table 2, set 2) was dependent on the presence ofacetylphosphate in the assay mixture, but was still presentwhen CoA was omitted. Furthermore, doubling the amountof CoA in the reaction mix inhibited the residual PTAactivity in MD050L extracts. In contrast, the PTA activity inHW21 extracts had the expected dependence on both CoAand acetylphosphate. A comparison of ACK and PTA levels

TABLE 2. Specific enzyme activities of PTA and ACK

Activity (U/mg of protein)Strain

PTA ACK

Set 1HW21(pFC54.t) 1.2 0.45MDO5OL(pFC54.t) <0.01 1.2

Set 2HW21(pFC54.t) 0.48 0.26MD05OL(pFC54.t) 0.007 0.40

Set 3HW21(pFC54.t) 1.5 0.14MD05OL(pFC54.t) 0.002 0.19

in samples sonicated for different lengths of time (data notshown) showed that release of ACK activity was maximalunder the sonication conditions used to generate the assayresults shown in Table 2. Further sonication resulted in asmall increase in the PTA activity seen in HW21 extracts,but the minute amount of activity seen in MDOOL extractswas lost altogether. The observed enzyme activity differ-ences were therefore not due to differences in the extent ofcell breakage between the two strains, and the small amountof apparent PTA activity in MDOSOL was probably due to anactivity other than normal PTA.ACK activity in the mutant strain was higher than that in

the PTA-competent strain by 2.7-fold in the first assay,1.5-fold in the second, and 1.3-fold in the third assay.Enhanced ACK activity in pta mutants has not been re-ported in the literature; in fact, in one report, mutations inpta were said to have no effect on ACK activity (3).

(ii) Acetate accumulation. To determine the extent ofacetate accumulation in flask cultures, HW21 (Pta+) andKB100 (Pta-) were grown for 15 generations in glucose-containing defined medium at 30°C to achieve balancedgrowth and to eliminate carry-over of undefined nutrientsfrom the inoculum. Growth and acetate accumulation rateswere measured over an OD680 range of 0.02 to 0.3 in whichgrowth was exponential and the supply of nutrients (includ-ing oxygen) was not limiting. Data from these experimentsare shown in Table 3. The Pta- mutant and nonmutantstrains both had very similar specific growth rates of 0.59 to0.60 h-1. The specific acetate production rate of KB100 in

TABLE 3. Growth rate and acetate production

Specific growth Specific acetate

Expt and straina rate (h-') production(mmollg Idry wt]300C 370C per h)

Expt 1 (flaskcultures)

HW21 0.60 4.62HW21(pFC54.t) 0.50 3.55KB100 0.59 0.59

Expt 2 (flaskcultures)

TA3521 0.69 6.07TA3514 (pta) 0.55 1.38TA3515 (ack) 0.49 2.74TA3516 [A(ack- 0.56 1.40

pta-hisQ-hisP)]a Medium and conditions were as described in the text.

1298 BAUER ET AL.

on May 12, 2018 by guest

http://aem.asm

.org/D

ownloaded from

PTA- MUTANT OF E. COLI SHOWING IMPROVED IL-2 EXPRESSION 1299

C:

0-v

co

a)

(00.

Q-0

a-)

100.0o.

10.0

3.0 L1Fr

.- I-I-13.05 16 17 18 19

Culture Age (Hours)FIG. 1. Mock inductions in fermentor cultures: growth (closed

symbols) and acetate production (open symbols) during mock in-ductions of the Pta- mutant MD050 (triangles) and a Pta nonmutantstrain, HW21 (circles). In each case, the temperature was shiftedfrom 30 to 37°C when the OD680 reached 20.

these flask cultures was approximately sevenfold lower thanthat of HW21.To explore acetate accumulation by the Pta- mutant in

fermentor cultures, mock inductions of strains MD050 andHW21 were performed by shifting the temperature from 30to 37°C at an OD680 of 20. These two strains did not containplasmid and thus did not express IL-2. The results are shownin Fig. 1. The Pta- mutant and nonmutant strains had similargrowth kinetics at both temperatures, while specific acetateproduction in the mutant was lower by a factor of approxi-mately twofold. Specific acetate production rates were lowerin fermentors than in flasks for both strains, with thedifference being more pronounced in HW21 than in MD050.This may have been due to simple inhibition of the reversibleACK-PTA pathway, caused by the higher levels of acetate inthe medium of fermentor cultures of HW21. There was littlechange in specific acetate production rates in the parentstrain before and after mock induction. This was in contrastto the increase of 16-fold in specific acetate productivity thatoccurs when PTA-competent cells contain a plasmid allow-ing IL-2 expression (10).

Acetate accumulation in flasks was also measured in an

isogenic set of E. coli K-12 strains which included a differentpta mutant (TA3514) and a pta-ack double deletion mutant(TA3516). These strains were grown in defined medium plushistidine (100 ,ug/ml to satisfy an auxotrophic requirement) at37°C for 20 h prior to analyzing the supernatant for acetate.Both the deletion strain and our Pta- strain (MD050) were

found to produce lower amounts of acetate compared withtheir respective nonmutant parents (Table 3).

Finally, we studied the effect of a mutation in pyruvateoxidase (poxB) on acetate accumulation in our Pta- mutant.A TnS insertion in the pyruvate oxidase gene was transferredinto KB100 by P1 transduction. Acetate accumulation and

growth rate in this double mutant were measured in glucose-containing defined medium at 37°C and were found to beindistinguishable from those in its parent.

Effect of the Pta- mutation on IL-2 production. The effectsof the Pta- mutation on acetate and IL-2 accumulation wereexamined in flask and fermentor cultures, after lysogenizingKB100 (Table 1) to carry the c1857 gene in the chromosome,and transforming both it and HW21 with a plasmid (pFC54.t)which allowed intracellular accumulation of IL-2 undercontrol of the temperature-sensitive PL promoter.

(i) Shake flasks. Cultures, 100 ml, of KB100(pFC54.t) andHW21(pFC54.t) (Table 1), growing exponentially in glucose-containing defined medium, were added to 400-ml portionsof prewarmed (38°C) fresh medium in Fernbach flasks andshaken at 38°C for up to 5 h. Samples for high-pressure liquidchromatography analysis of IL-2 were taken at 3 and 5 h ofinduction. Growth kinetics were similar in the two strains,with the mutant growing slightly more in the later hours ofinduction, while specific acetate accumulation was calcu-lated to be approximately sevenfold higher in the parentstrain than in the mutant. IL-2 expression (as a percentage oftotal intracellular protein) after 3 and 5 h was 16 and 19%,respectively, for the HW21 strain and 18 and 21%, respec-tively, for the KB100 strain. Therefore, the Pta- mutationresulting in lower acetate accumulation in KB100 did notnegatively affect either growth or IL-2 accumulation andmay have enhanced them slightly.

(ii) Fermentor cultures. MDO5OL(pFC54.t) was grown in afermentor as described above and induced by shifting thetemperature from 30 to 37°C at cell densities of approxi-mately 20 and 40 OD680 in two separate runs. For compari-son, HW21(pFC54.t) was grown similarly, with induction ofIL-2 expression by temperature shift at 20 and 40 OD680. Thekinetics of growth, acetate production, and IL-2 expressionin these experiments are presented in Fig. 2. While bothstrains had a similar 2.5-fold increase in OD680 in the first 3h of induction in flasks, the Pta+ strain stopped growing aftera cell density increase of only 1.6-fold in the 20-OD680fermentor induction and 1.5-fold in the 40-OD680 fermentorinduction. In contrast, the Pta- mutant grew substantiallymore during induction at both cell densities. Specific acetateproduction during the 5-h induction period was seven- toninefold less for the mutant than for the wild type at thedifferent cell densities tested.

Specific IL-2 accumulation in the transformed Pta- mu-

tant was 23% lower in fermentor cultures than observed inflask cultures (17 versus 21% in flasks), while in the wild typeit was 36% lower than that seen in flask cultures (14 versus

19% in flasks). The data for specific IL-2 accumulation in thePTA-competent strain at the induction cell densities tested inthese studies were consistent with our previous observations(10) in which IL-2 expression was less affected by the celldensity at induction in the 20- to 40-OD680 range. Figure 2Dshows that the Pta- mutant accumulated more total IL-2 inthe fermentor as a result of the increase in specific IL-2accumulation combined with continued growth during induc-tion.

Effect of Pta- mutation on ethanol, lactate, and pyruvateaccumulation. In fermentor cultures of the PTA-competentstrain HW21(pFC54.t) grown in glucose-containing definedmedium, D-lactate and pyruvate were not produced in largeamounts. During the 5-h induction period, the concentrationof D-lactate increased from approximately 1 to approxi-mately 4 mM, and that of pyruvate increased from approx-imately 0.1 to approximately 7 mM. There was a higher levelof accumulation of these acids during growth and induction

VOL. 56, 1990

on May 12, 2018 by guest

http://aem.asm

.org/D

ownloaded from

1300 BAUER ET AL.

0

ol

I.-

J

(0

-5

C\

0

04-

CD

a)

1-5

0)

I--,

0

F-j

0 1 2 3 4 5 0 1 2 3 4 5

Time After Induction (h) Time After Induction (h)FIG. 2. Growth (A), acetate production (B), specific IL-2 production (C), and total IL-2 production (D) in fermentor cultures: Pta- mutant

MD05OL(pFC54.t) (filled symbols) and a Pta nonmutant strain, HW21(pFC54.t) (open symbols) during induction at OD680s of 20 (triangles)and 40 (circles). Both strains carried a plasmid which allowed expression of IL-2 when the temperature was shifted from 30 to 37°C at the timeof induction. The expression level of IL-2 is reported as the weight percentage of total intracellular protein.

when MDO5OL(pFC54.t) was used. In these fermentor cul-tures, D-lactate accumulated to 9 mM by the start of induc-tion and subsequently rose to 15 mM at 3 h of inductionbefore falling to 5 mM by the end of the 5-h period. Pyruvateaccumulation increased from 1.5 to 8.4 mM during the same

period. Ethanol was not produced in amounts greater that 1mM in either culture before or during induction. Clearly, thelowered amount of acetate produced in the Pta- mutant was

not completely compensated for by the slight increases inD-lactate and pyruvate production, particularly when thedifferences in cell density are taken into account.

DISCUSSION

The data presented here provide evidence that MD050 is aPta- mutant. At least two groups (3, 9) have reported thatthe method used to select MD050 yields two classes ofmutants, pta and ack. Both classes of mutants are reportedto have near-normal aerobic growth rates on glucose, buthave a greatly reduced growth rate on acetate, particularlythe pta mutants. Brown et al. (3) showed that pta mutantshave greatly reduced acetate production, while ack mutantsaccumulate near-normal levels of acetate. This could be

100.0

0co0

00

0

I._

Q._

~o

00.

al)

I I

C)

350.0

300.0

Q-

C 250.00

L-J

C 200.0

0 150.0

- 100.0ci

E

APPL. ENVIRON. MICROBIOL.

on May 12, 2018 by guest

http://aem.asm

.org/D

ownloaded from

PTA- MUTANT OF E. COLI SHOWING IMPROVED IL-2 EXPRESSION 1301

because acetylphosphate, which is expected to be accumu-lated by ack mutants, is unstable at physiological pH andtemperature and may decompose to acetate. The lack ofgrowth of MD050 on acetate-containing plates and its re-duced acetate accumulation implied that it was a Pta-mutant rather than an Ack- mutant. The absence of PTAactivity in MD050 extracts provided further proof. Whentransformed for IL-2 production under control of the tem-perature-sensitive PL promoter, the Pta- mutant continuedto grow during induction to a greater extent than its parent.It produced less acetate and greater amounts of IL-2 in termsof both specific and total IL-2 accumulation. These proper-ties made MD050 superior to its parent, MM294-1, as a hostorganism for expression of IL-2 and possibly other heterol-ogous or homologous protein products.We believe that MD050 was nearly devoid of PTA activ-

ity, despite the fact that it still produced some acetate. Twolines of evidence support this view: first, only trace amountsof PTA activity could be detected in crude extracts; andsecond, strains carrying a known deletion of PTA continuedto produced acetate. However, the precise location andnature of the genetic lesion in MD050 have not been deter-mined by mapping or other more rigorous means, and wecannot completely rule out the possibility that some of theremaining acetate production in MD050 was due to residualPTA activity. Other possible sources of acetate productioninclude the enzyme pyruvate oxidase, which catalyzes theoxidative decarboxylation of pyruvate, and spontaneoushydrolysis of acetyl-CoA. Our results indicate that pyruvateoxidase was not responsible for the remaining acetate pro-duced by MD050.The 27% improvement in specific IL-2 accumulation

achieved in this study by lowering acetate accumulation,using the Pta- mutant, was slightly better than the 17%improvement seen in our previous study, which used me-dium perfusion to keep acetate concentrations low (10). Ifthis difference is significant, then it may result from the factthat, in the perfusion cultures, although its concentrationwas kept below 100 mM, acetate may still have beensomewhat inhibitory to maximal IL-2 accumulation. It couldalso be explained if the blockage in the Pta- mutant used inthe present study led to improved utilization of acetyl-CoAthrough the tricarboxylic acid cycle and oxidative phosphor-ylation. This may have allowed the formation of moreintermediates and ATP for IL-2 synthesis than would beexpected from waste acetate formation. Comparisons of theflux rates through the tricarboxylic acid cycle and throughoxidative phosphorylation before and after induction of IL-2synthesis will provide additional information in this regard.The reason E. coli makes acetate during aerobic growth on

glucose is a matter for speculation. It is clear that the bulk ofthe acetate accumulating in the medium during IL-2 expres-sion came directly from acetyl-CoA via the PTA-ACKpathway (i.e., it appeared not to be made directly frompyruvate). Presuming that pyruvate formate lyase was notactive under these conditions, this probably means thatpyruvate dehydrogenase is sufficiently active to supply boththe tricarboxylic acid cycle (for oxidative phosphorylationand amino acid biosynthetic needs) as well as acetate secre-tion. Our data show that reducing the acetate formation rateby as much as sevenfold by mutation of pta had no measur-able effects on growth rate. We did, however, observe thatgreater amounts of D-lactate and pyruvate were produced,but not in sufficient amounts to make up the difference inacetate accumulation. Perhaps the production of these acidsreflects an imbalance between the cell's glycolytic capacity

and its respiratory capacity (7). The excess glycolytic capac-ity is shunted into acetate or, failing that, into D-lactate andpyruvate. Conditions that result in increased organic acidproduction may be those in which there is reduced respira-tory flux, exacerbating the imbalance between glycolysis andrespiration. High-level IL-2 expression may be one suchcondition. It may inhibit respiratory flux directly, by inter-fering with some component of the respiratory machinery, orindirectly, perhaps by excessive shunting of carbon throughthe glyoxylate bypass to provide sufficient intermediates foramino acid biosynthesis, but at the expense of energyproduction. Measurements of the rate of bulk protein syn-thesis during IL-2 accumulation along with metabolic fluxrates through glycolysis, the tricarboxylic acid cycle, and theglyoxylate bypass would help develop these ideas further.The accumulation of organic acids in E. coli cultures has

practical significance. For the production of cloned proteinson a large scale, it is desirable to achieve a high cell densityto maximize volumetric productivity. However, the con-comitant high acetate concentration was found to be apossible cause of reduced IL-2 expression per cell (10). ThePta- mutant used in this study reduced acetate formationand improved IL-2 expression, even at high cell density. Ifthe nutritional and physical conditions in the flasks wereoptimal, then the 21% IL-2 accumulation level seen in flasksmay represent an upper limit for IL-2 accumulation infermentor cultures when all acetate is removed or whenacetate production is completely blocked. Parameters suchas temperature, pH, dissolved oxygen concentration, specialnutrient conditions, and the kinetic details of the tempera-ture shift itself, all of which can be controlled in a fermentor,may also play a role in defining the maximum level of IL-2accumulation. Further improvements in IL-2 accumulationin fermentor cultures may also be possible when the effectsof IL-2 expression on central metabolism are more fullyunderstood.

ACKNOWLEDGMENTS

We thank Heather MacDonald for help with the high-pressureliquid chromatography assay for IL-2 and Hing C. Wong for adviceregarding the selection of Thi+ strains.

LITERATURE CITED1. Allen, B. R., and G. W. Luli. 1987. A gradient-feed process for

obtaining high cell densities for recombinant E. coli. BioPhar-macology 0:38-41.

2. Bauer, S., and J. Shiloach. 1974. Maximal exponential growthrate and yield of E. coli obtainable in a bench-scale fermentor.Biotechnol. Bioeng. 16:933-941.

3. Brown, T. D. K., and M. M. Jones-Mortimer, and H. L.Kornberg. 1977. The enzymatic interconversion of acetate andacetyl-coenzyme A in Escherichia coli. J. Gen. Microbiol.102:327-336.

4. Chang, Y.-Y., and J. E. Cronan. 1982. Mapping nonselectablegenes of Escherichia coli by using transposon TnlO: location ofa gene affecting pyruvate oxidase. J. Bacteriol. 151:1279-1289.

5. Chang, Y.-Y., and J. E. Cronan. 1983. The genetic and biochem-ical analysis of Escherichia coli strains having a mutation in thestructural gene (poxB) for pyruvate oxidase. J. Bacteriol. 154:756-762.

6. Fox, D. K., and S. Roseman. 1986. Isolation and characteriza-tion of homogenous acetate kinase from Salmonella typhimu-rium and Escherichia coli. J. Biol. Chem. 261:13487-13497.

7. Holms, W. H. 1986. The central metabolic pathways of Esche-richia coli: relationship between flux and control at a branchpoint, efficiency of conversion to biomass, and excretion ofacetate. Curr. Top. Cell. Regul. 28:69-105.

8. Kwan, H. S., H. W. Chui, and K. K. Wong. 1988. ack::Mu dl-8

VOL. 56, 1990

on May 12, 2018 by guest

http://aem.asm

.org/D

ownloaded from

1302 BAUER ET AL. APPL. ENVIRON. MICROBIOL.

(Apr lac) operon fusions of Salmonella typhimurium LT2. Mol.Gen. Genet. 211:183-185.

9. LeVine, S. M., F. Ardeshir, and G. F. Ames. 1980. Isolation andcharacterization of acetate kinase and phosphotransacetylasemutants of Escherichia coli and Salmonella typhimurium. J.Bacteriol. 143:1081-1085.

10. MacDonald, H. L., and J. 0. Neway. 1990. Effects of mediumquality on the expression of human interleukin-2 at high celldensity in fermentor cultures of Escherichia coli K-12. Appl.Environ. Microbiol. 56:640-645.

11. Meselson, M., and R. Yuan. 1968. DNA restriction enzyme fromE. coli. Nature (London) 217:1110-1114.

12. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

13. Shimizu, N., S. Fukuzona, K. Fujumori, N. Nishimura, and Y.Odawara. 1988. Fed-batch cultures of recombinant Escherichia

coli with inhibitory substance concentration monitoring. J.Ferment. Technol. 66:187-192.

14. Suzuki, T. 1969. Phosphotransacetylase of Escherichia coli B,activation by pyruvate and inhibition by NADH and certainnucleotides. Biochim. Biophys. Acta 191:559-569.

15. Wang, A. M., A. A. Creasey, M. B. Ladner, L. S. Lin, J.Strickler, J. N. Van Arsdell, R. Yamamoto, and D. F. Mark.1985. Molecular cloning of the complementary DNA for humantumor necrosis factor. Science 228:149-154.

16. Wong, E. M., M. A. Muesing, and B. Polisky. 1982. Tempera-ture-sensitive copy number mutants of ColEl are located in anuntranslated region of the plasmid genome. Proc. Natl. Acad.Sci. USA 79:3570-3574.

17. Zabriskie, D. W., and E. J. Arcuri. 1986. Factors influencingproductivity of fermentations employing recombinant organ-isms. Enzyme Microb. Technol. 8:706-717.

on May 12, 2018 by guest

http://aem.asm

.org/D

ownloaded from