Metabolic Fluxes, Pools, and EnzymeMeasurements Suggest a TighterCoupling of Energetics and BiosyntheticReactions Associated with ReducedPyruvate Kinase Flux

Akshay Goel,1 Jinwoon Lee,2 Michael M. Domach,2 Mohammad M. Ataai1

1Department of Chemical Engineering and Center for Biotechnology andBioengineering, University of Pittsburgh, 300 Technology Drive, Pittsburgh,Pennsylvania 15219; telephone: 412-383-9744; fax: 412-383-9710;e-mail: ataai@engrng.pitt.edu2Department of Chemical Engineering, Carnegie Mellon University,Pittsburgh, Pennsylvania

Received 1 February 1997; accepted 1 May 1998

Abstract: In this study, it is found that, for Bacillus sub-tilis, citrate–glucose cometabolism leads to zero acid pro-duction over a wide range of growth rates and nearlytheoretical carbon yield. Experimental results are pre-sented that point to pyruvate kinase (PYK) as a site ofcitrate-mediated glycolytic flux attenuation. First, themeasured fluxes show that, compared with culturesgrown on glucose, the PYK flux drops by more than ten-fold when citrate is added. Second, relative to culturesmetabolizing glucose, the phosphoenolpyruvate (PEP)pool elevates substantially, whereas the pyruvate pooldrops, when citrate is present. Finally, our modeling re-sults indicate that maximizing carbon yield correspondsto nearly eliminating pyruvate kinase (PYK) flux and thatthe pyruvate supplied by the PEP-consuming glucosetransport system can supply the biosynthetic require-ments. A literature review suggests some mechanismsfor how PYK attenuation by citrate addition can occur. Atthis juncture, we hypothesize that direct PYK inhibitionoccurs which, in turn, also leads to phosphofructokinaseinhibition via the elevated PEP pool. These two inhibitionevents combine to throttle glycolytic flux; minimize acidformation; and substantially increase cellular, product,and energetic yields. © 1999 John Wiley & Sons, Inc. Biotech-nol Bioeng 64: 129–134, 1999.Keywords: Bacillus subtilis; metabolic engineering; acidformation; pyruvate kinase mutant

INTRODUCTION

Poor coordination between glucose consumption and pre-cursor synthesis in the TCA cycle is viewed by some mi-crobial physiologists as responsible for carbon overflow(acid production) and excess ATP production (Jensen andPederson, 1990; Marr, 1991). From a technological stand-point, the acid production and excess ATP synthesis poten-tially arising from this mismatch adversely affects culture

yield and process stability. Acid production, in addition tolimiting the maximal cell concentration, correlates with re-duced protein production and compromised fed-batch reac-tor stability, thus necessitating the use of substrate feedingcontrol. For example, the specific growth rate of manyEscherichia coliand Bacillus strains in dense, fed-batchculture must be kept below a low critical growth rate whenglucose is a growth medium constituent because of the ac-cumulation of acetate (Fieschko and Ritch, 1986). Thus, itwould be desirable to employ metabolic engineering toeliminate acid formation and attain yields closer to theoret-ical values, simplify process control schemes, and increaseculture productivity.

We found that, whenBacillus subtilis is supplementedwith citrate and glucose (6 mol glucose/mol citrate), acidproduction over a wide range of growth rates is nonexistent,glucose consumption decreases per cell, and nearly theoret-ical carbon yield is attained without a reduction in maximalgrowth rate (Goel et al., 1995). Table I illustrates that citrateaddition leads to a substantial enhancement of carbon yieldand complete elimination of acid formation (Goel et al.,1995). The theoretical values have been obtained from Leeet al. (1997). To determine whether the citrate effect resultsfrom altered flux regulation or citrate simply replacing theglucose used to form TCA-cycle amphibolites, citrate wasreplaced with glutamine. The results (Goel et al., 1995)showed that acid formation remained significant in the pres-ence of glutamine. Moreover, experiments in which the glu-cose feed was supplemented with a complex mixture ofnutrient broth and amino acids also exhibited significantacid formation (Lee et al., 1997).

Citrate addition also increased the productivity of proteinexpression in recombinantB. subtilischemostat cultures bytenfold, perhaps due to the reduction in formation of inhibi-tory acidic byproducts (Vierheller et al., 1995). Correspond-

Correspondence to:M. Ataai; e-mail: ataai@engrng.pitt.eduContract grant sponsor: NSFContract grant numbers: BCS-92222557, BCS-92077614

© 1999 John Wiley & Sons, Inc. CCC 0006-3592/99/020129-06

ing results are not currently attainable forE. coli because itdoes not readily transport citrate (Hirato et al., 1984).

The attainment of high carbon yield has both physiologi-cal and technical significance; hence, these results wereprobed further in order to deduce what regulatory shiftsaccompany citrate addition. The metabolic fluxes at a givengrowth rate when citrate was present or absent were ob-tained by using metabolite balances and measurements(Goel et al., 1995; Lee et al., 1997). These results are shownin Figure 1 for glucose and glucose-plus-citrate-grown cul-tures. The fluxes determined from the data acquired fromcultures (Fig. 1) revealed that the flux of the pyruvate kinase(PYK)-catalyzed reaction (PEP to PYR) is diminished for

cells grown in glucose plus citrate (0.21 mmol/g cells? h)versus glucose-only medium (6.12 mmol/g cells? h).

In addition, optimal flux distributions in pathways wereascertained by using linear programming (Lee et al., 1997),akin to the analysis of fat synthesis in adipose tissue (Felland Small, 1986). These distributions provide the fluxes thatcan satisfy objectives such as minimal carbon use subject tothe constraints of biosynthetic requirements, minimal fluxthrough a particular enzyme, etc. Figure 2 shows the opti-mal fluxes for glucose-only and glucose plus citrate. Theresults (Fig. 2) reveal that maximizing carbon yield, forcells fed either glucose or glucose plus citrate, correspondedto nearly eliminating PYK flux, indicating that the pyruvatesupplied by the PEP-consuming glucose transport systemcan supply the biosynthetic requirements. The optimal re-sults for both glucose-only and glucose-plus-citrate metabo-lism (Fig. 2) indicate that the network is intrinsically ca-pable of simultaneously supplying the minimal carbon skel-eton and biosynthetic ATP requirement with a carbon yieldsubstantially greater than that exhibited for glucose only(Table I). Thus, the success of citrate–glucose metabolismprobably is not attributable to alleviating inherent stoichio-metric constraints imposed by the routes available for glucoseuse (for examples of such constraints, see Lee et al. [1997]).Rather, citrate addition induces regulatory changes that per-mit a facsimile of the optimal use of glucose to be achieved.

Figure 1. Network showing central pathways. The flux assignments(mmol/g cells? h) are those obtained for glucose-only (the first number)and glucose-plus-citrate-fed cultures (the number given in parentheses)grown in the chemostat at dilution rate of 0.4 h−1 (Table I). Reactionsproducing ATP, NADH, NADPH, and FADH2 are shown (see symbolkey). Note that the conversion of PEP to PYR is occurring by the reactioncatalyzed by pyruvate kinase as well as by the glucose transport system(PTS). Thus, one of the fluxes for the conversion of PEP to PYR is thesame as for glucose to G6P. In addition, because the flux values are theleast square solution of the flux equations, the input to and output from ametabolic pool are not exactly equal. Abbreviations: AcCoA, acetyl coen-zyme A; CIT, citrate; E4P, erythrose-4-phosphate; F6P, fructose-6-phosphate; G6P, glucose-6-phosphate; KG,a-ketoglutarate; OAA, oxaloa-cetate; PEP, phosphoenol pyruvate; 3PG, 3-phosphoglycerate; MAL, ma-late; PYR, pyruvate; organic acids, in general, acids derived from pyruvateexcept acetate (but in our case was only the secreted pyruvate as the otherswere insignificant); R5P, ribose-5-phosphate; Ribulose 5P, ribulose-5-phosphate; Succ, succinate; TP, triose phosphate; Xylulose 5P, xylulose-5-phosphate.

Figure 2. The flux assignments are those that maximize carbon yield forB. subtilismetabolizing glucose-only (the first number) and glucose pluscitrate (given in the parentheses). The measured fluxes for glucose/citrate(Fig. 1) correspond closely to the optimal assignments given in this figure.

130 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 64, NO. 2, JULY 20, 1999

In this study, some of the intracellular changes associatedwith the addition of citrate in the feed are examined. Theresults (e.g., high PEP and low pyruvate concentrations aswell as lower activities of glycolytic enzymes) are consis-tent with the attenuated PYK flux as the basis for success ofcitrate in reducing acid formation and promoting high cel-lular and energetic yields. The PYK flux attenuation mayresult from a potential mechanism involving Ca2+-coupledcitrate uptake and the sensitivity of PYK activity to Ca2+.

EXPERIMENTAL METHODOLOGY

Bacterial Strains and Medium

Bacillus subtilis (Marburg strain) was obtained from theAmerican Type Culture Collection (ATCC No. 6051). Oneliter of media contained: 1 g NH4Cl; 0.04 g tryptophan; 1.0g KH2PO4; 2.72 g K2HPO4; 0.284 g Na2SO4; 0.17 gNaNO3; 0.15 g KCl; 25 mg MgCl2 ? 6H2O; 2.16 mgFeCl3 ? 6H2O; 15 mg MnCl2 ? 4H2O; and 22 mgCaCl2 ? 6H2O. The concentrations of glucose and citrateused in each experiment are noted in what follows.

Reactor and Measurement of Cell Mass, Glucose,and Extracellular Organic Acids Concentrations

Fermentation vessels with 2-L capacity (Applikon, Austin,TX) were used. An on-line data acquisition system (Beitleand Ataai, 1991) collected data for pH, agitation, air flowrate, CO2 evolution rate, and oxygen uptake rate. The airflow rate was set at 2 L/min and pH was constrained be-tween 6.7 and 6.9. Optical density was measured off-lineusing a Lambda-6 UV/Vis Perkin Elmer spectrophotometer(Perkin Elmer, Norwalk, CT; from our calibrations 1 OD660

4 0.35 g cell dry weight/L). The glucose concentration wasmeasured enzymatically using a Sigma (St. Louis, MO) Di-agnostic Kit. To measure the concentrations of organic acids(lactate, acetate, acetoin, formate, pyruvate, succinate, andcitrate), a high-performance liquid chromatography (HPLC)gradient module (Bio-Rad), equipped with a UV/Vis moni-tor was employed (Goel et al., 1995).

Intracellular Enzyme and Metabolite Assays

Steady-state cell samples for enzyme assays were taken bypumping a fixed volume (100 mL) rapidly (within 10 s) intoa conical flask and chilling the sample using liquid nitrogen.The sample preparation was as reported by Fisher and Ma-gasanik (1984). The enzymatic assays were performed forphosphoglucose isomerase (PGI; Noltmann, 1966), pyru-vate kinase (PYK; Malcovati and Valentini, 1982), malicenzyme (ME; Hsu and Lardy, 1969), pyruvate carboxylase(PC; Young et al., 1969), and isocitrate dehydrogenase(ICDH; Sigma Diagnostic Kit 153-UV). Pyruvate dehydro-genase complex (PDC) was measured by estimating thelevels of one of its components, dihydrolopoyldehydroge-nase (Reed and Willms, 1966). For metabolite analyses, asample was withdrawn via a syringe and quickly (10 s)filtered followed by immersion of the filter into liquid ni-trogen. The metabolites were extracted from the filteredcells by immersing the filter in formic acid (Fisher andMagasanik, 1984). A Perkin-Elmer fluorospectrometer wasused for the assays as given by Lowry and Passonneau(1972).

RESULTS

As noted earlier, metabolite balances and linear-program-ming models suggest that pyruvate kinase flux is highlyattenuated. Thus, activity of selected enzymes and concen-trations of several metabolites were measured. The synthe-sis levels of intracellular enzymes of the dual-feed (glucoseplus citrate) and single-feed (glucose alone) cultures arecompared in Figure 3. Malic enzyme increased considerablyin the dual-feed cultures, suggesting that reaction of malatewith pyruvate may have taken place. Malic enzyme wasselected for assay because linear-programming analysis(Lee et al., 1997) indicated that, for glucose-plus-citrate-grown cells, slightly higher (ca. 3%) cell yield could poten-tially be attained if the reaction of malate to pyruvate (cata-lyzed by malic enzyme) was included in the pathway, asshown in Figure 2. Although malic enzyme activity is notappreciable in bacteria when glucose is present, its presencein the dual-feed scenario is consistent with the consumptionof citrate. Note that, in glucose-only cultures, pyruvate car-boxylase flux (PYR to OAA) provided for the biosynthetic

Table I. Yield and acid production ofB. subtilischemostat cultures.a

Mediumglucose(g/L)

Mediumcitrate(g/L)

Residualglucose(g/L)

Residualcitrate(g/L)

C yieldmeasured % Acid % CO2

C yieldtheoreticalb

YATP

theoreticalb

2 0 0 NA 0.31 41c 26 0.80 27.22 0.64 0 0.1 0.63 0 27 0.75 23.12 0.32 0 0 0.73 0 27 0.75 23.1

aIn all experiments the dilution rate was 0.4 h−1. C yield is defined as grams of carbon in cells per total grams of carbon consumed.YATP is the ATPyield coefficient (g cell/mol ATP). % Acid and CO2 are based on grams of carbon in acids or CO2 per total grams of carbon consumed.

b Theoretical data are from Lee et al. (1997).c The acids were acetate and pyruvate at concentrations of 0.1 and 0.7 g/L, respectively.

GOEL ET AL.: ASSOCIATION OF REDUCED PYK FLUX AND HIGH YIELDS 131

fluxes of the TCA cycle, whereas, in glucose-plus-citratecultures, citrate carbon could also be used for this purpose.However, if citrate transport rate is higher than its consump-tion, it could lead to a rise in the pool of TCA-cycle me-tabolites, including malate, which may in turn induce/derepress the malic enzyme. Malic enzyme will allow forconsumption of citrate transported in excess of its use insynthesis of amino acids and other precursors derived froma-ketoglutarate and oxaloacetate.

Unlike malic enzyme, the activities of the other enzymeswere lower in the cultures with citrate. The lower enzymesynthesis level of the other enzymes is consistent with thelower fluxes displayed by the dual-feed cultures. PYK ac-tivity in the cell extract decreased much less than its in vivoflux (see Fig. 1), suggesting that repression may not fullyaccount for the reduction in PYK flux in the cultures pro-vided with citrate.

To investigate the enzyme activity results further, intra-

cellular metabolite pools were measured. A comparison ofintracellular metabolic concentrations is provided in Figure4. The measurements of the PEP and other pools were theaverage of two readings and the variations were in the rangeof ±20%. The PEP pool is elevated sixfold (0.07 vs. 0.01mM) in cells provided with citrate. Note that PEP pool islow in bacteria; the reported value forE. coli is 0.03 mM(Lowry et al., 1971). Thus, although PEP measurements areconsistent with PYK flux reduction, they may be prone touncertainty. Therefore, the larger pyruvate pool was alsomeasured. Pyruvate was barely detectable in the dual-feedcultures (ca. 0.1 vs. 1.1 mM in Fig. 4). The depressed py-ruvate pool is consistent with reduced acetate formation(Table I) and reduced PYK flux (Fig. 1). Taken together,PEP elevation and pyruvate depression suggest that a bottle-neck was created in the PYK-catalyzed reaction under dual-feed conditions.

The glucose-6-phosphate pool was also elevated fourfoldin cultures fed glucose and citrate. Because the isomerasethat converts glucose-6-phosphate to fructose-6-phosphateis reversible, it is conceivable that, in the dual-feed cultures,

Figure 3. Specific activity of enzymes in cells grown in continuouscultures of glucose and glucose-plus-citrate-fed cultures at a dilution rate of0.4 h−1. Glucose was 2 g/L and citrate was 0.64 g/L. Enzymes are: ICDH(isocitrate dehydrogenase catalyzing citrate to isocitrate); PDC (pyruvatedehydrogenase complex, catalyzing pyruvate to acetyl CoA); PG (phos-phoglucoisomerase, catalyzing glucose 6P to fructose 6P); PC (pyruvatecarboxylase, catalyzing pyruvate to oxaloacetate); and PYK (pyruvate ki-nase, catalyzing PEP to pyruvate).

Figure 4. Intracellular metabolite concentrations. PYR, CIT, G6P, andPEP represent pyruvate, citrate, glucose 6P, and phosphoenolpyruvate,respectively. For the glucose-only feed, PEP is not detectable. The value of0.01 is the detection limit and thus represents the upper limit of the PEPconcentration.

132 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 64, NO. 2, JULY 20, 1999

phosphofructokinase (PFK) activity becomes a local bottle-neck, thereby causing upstream metabolites such as glu-cose-6-phosphate to increase. Because PEP is a strong in-hibitor of PFK in bacteria, bottleneck formation of PFK canconceivably arise from reduced PYK activity. Finally, thecitrate pool is lower in the glucose-plus-citrate cultures thanin glucose-only cultures; the lower pool corresponds withthe lower value of the TCA cycle flux.

The PYK flux attenuation may result from a potentialmechanism involving Ca2+-coupled citrate uptake and thesensitivity of PYK activity to Ca2+. Such inhibition has beenobserved for other pyruvate kinases (Genin and Andreev,1975; Katchmar and Boyer, 1953), butB. subtilisPYK hasnot been examined. Thus, crude extractof B. subtiliswasincubated with different Ca2+ concentrations. It was foundthat the PYK activity decreased with increasing calciumconcentration (Fig. 5), and the activity was completely abol-ished at a Ca2+ concentration of 30 mM. Higher Ca2+ con-centration than the free intracellular concentration of freeCa2+ was used because of the presence of Mg2+ in the assaysolution and the fact that raw cell extract, which could bindsome Ca2+, was used as the source of enzyme. The purpose

of this experiment was only to determine if, like other pro-caryotes,B. subtilisPYK is also inhibited by Ca2+.

DISCUSSION

Citrate addition was previously found to reduce glucose use,drastically increase cell yield, and eliminate acid formationin B. subtilis(Goel et al., 1995). In addition, the productivityof recombinant cultures was greatly enhanced (Vierheller etal., 1995). Metabolite balances and linear-programminganalyses indicated that malic enzyme utilization and sup-pression of PYK flux are among the intracellular eventsprompted by citrate uptake.

Because the regulatory features of PYK and PFK fromB.subtilis have not been examined thus far, it is difficult tospeculate, with a high degree of certainty, on how citratereduces the glycolytic flux. However, as discussed next, thepool (elevated PEP and glucose-6-phosphate and decreasedpyruvate) and enzyme activity measurements are consistentwith PYK flux attenuation.

Unlike mammalian PFK I which is inhibited by citrate,the activity of PFKs from different bacteria, includingE.coli (Blangy et al., 1968) andBacillus stearothermophilis(Evans and Hudson, 1979), are not affected by citrate, butare inhibited by PEP. Kinetic analysis ofB. stearother-mophilis, the only Bacillus species whose PFK has beenstudied in detail, has shown allosteric activation by ADPand inhibition by PEP (Evans and Hudson, 1979; Hengart-ner and Harris, 1975). This allosteric behavior was similarto that found forE. coli PFK (Auzat et al., 1994; Lau andFersht, 1987; Rypniewski and Evans, 1989). PEP wasshown to inhibit PFKs by either blocking the fructose-6-phosphate site (Schirmer and Evans, 1990), or dissociatingthe subunit interface that forms this site (Xu et al., 1969).

We believe that the interaction between citrate and diva-lent metal ions might play a critical role in reduction ofPYK activity. One possibility is that citrate might alter,through chelation, the relative abundance of metals in bothintracellular and extracellular environments and its conse-quences on glycolytic enzyme activities can feasibly ac-count for the observed decrease in glycolytic flux in glu-cose-plus-citrate-fed cultures. Another route for intracellu-lar metal speciation and its relative availability is via citratetransport, which, inB. subtilis,is coupled with the uptake ofMg2+ and Ca2+ (Willecke et al., 1973) with a stronger pref-erence for Mg2+ than Ca2+. Although Mg2+ is required forpyruvate kinase I (PYK I) activity, the major PYK inE. coli,Ca2+ substitution for Mg2+ leads to an inactive enzymecomplex (Boiteux et al., 1983). The proposed metal effect(Ca2+ replacement of Mg2+ and inhibition of PYK), or theintracellular metal speciation and their relative availability,should be viewed as hypotheses that require further explo-ration to establish their validity.

From a more general perspective, the present results alsosuggest the possibility of generic metabolic engineering,which could be used to transform microbes such asE. colithat do not readily transport citrate: namely, direct restric-

Figure 5. Effect of Ca2+ on PYK activity. Because the assay is performedin the presence of 10 mM Mg2+, a significantly higher concentration ofCa2+ than the typical intracellular concentration of free calcium (in micro-molar range) is used in this experiment.

GOEL ET AL.: ASSOCIATION OF REDUCED PYK FLUX AND HIGH YIELDS 133

tion of glycolysis at PYK by isozyme deletion and indirectlyupstream at PFK to lower acid production potential. Creat-ing a bottleneck at PYK could cause some phosphorylatedmetabolite pools to rise, similar to the observed increase inthe PEP and glucose-6-phosphate pools. However, the over-flow (leakage) of these metabolites could be avoided be-cause the strategy of phosphorylating metabolites, whichcells use to suppress leakage, is exploited. In addition, theomnipresent inhibition of PFK by PEP in procaryotes isexploited to further depress glycolysis and acetate produc-tion. This second potential site of flux regulation also pro-vides a “safety valve” for checking triose pool increases inresponse to the PYK bottleneck. Furthermore, the elevationof glucose-6-phosphate suggests that precursor supply forNADPH production by the HMP pathway is not threatened,which could conceivably result if the PTS system weredownregulated (Chou et al. 1994).

References

Auzat I, Bras GL, Branny P, Theunissen F, Garel J. 1994. The role ofGlu187 in the regulation of phosphofructokinase by phosphoenolpyru-vate. J Mol Biol 235:68–72.

Beitle RR, Ataai MM. 1991. A unique cultivation monitoring and controlsystem. Biotechnol Technol 5:77–82.

Blangy D, Buc H, Monod J. 1968. Kinetics of the allosteric interactions ofphosphofructokinase fromEscherichia coli.J Mol Biol 31:13–35.

Boiteux A, Markus M, Plesser T, Hess B, Malcovati M. 1983. Interactionof pyruvate kinase fromEscherichia coli with fructose 1,6-biphosphate and calcium ions. Biochem J 211:631–640.

Chou CH, Bennett GN, San KY. 1994. Effect of modifying glucose uptakeusing genetic engineering techniques on high level recombinant pro-tein production inEscherichia colidense culture. Biotechnol Bioeng44:952–969.

Evans PR, Hudson PJ. 1979. Structure and control of phosphofructokinasefrom Bacillus stearothermophilus.Nature 279:500–504.

Fell DA, Small JR. 1986. Fat synthesis in adipose tissue: an examinationof stoichiometric constraints. Biochem J 238:781–786.

Fieschko, J. and Ritch T. 1986. Production of humana-consensus inter-feron in recombinantEscherichia coli.Chem Eng Commun 45:229–240.

Fisher SH, Magasanik B. 1984. Synthesis of oxaloacetate inBacillus sub-tilis mutants lacking thea-ketoglutarate dehydrogenase enzyme com-plex. J Bacteriol 158:55–62.

Genin MS, Andreev VS. 1975. Identification of the inhibition type ofhexokinase and pyruvate kinase activity by Ca2+. Biokhimiya 40:885–889.

Goel A, Lee J-W, Domach MM, Ataai MM. 1995. Suppressed acid for-mation by cofeeding glucose and citrate inBacillus cultures: emer-

gence of pyruvate kinase as a potential metabolic engineering site.Biotechnol Prog 11:380–386.

Hengartner H, Harris JI. 1975. Purification by affinity chromatography,properties and crystallization of phosphofructokinase fromThermo-philic microorganisms. FEBS Lett 55:282– 285.

Hirato T, Shinagawa M, Ishiguro N, Sato G. 1984. Polypeptide involved intheEscherichia coliplasmid-mediated citrate transport system. J Bac-teriol 160:421–426.

Hsu RY, Lardy HA. 1969. Malic enzymes. Meth Enzymol 13:230–235.Jensen KF, Pederson S. 1990. Metabolic growth rate control inEscherichia

coli may be a consequence of subsaturation of the macromolecularapparatus with substrates and catalytic components. Microbial Rev54:89–100.

Kachmar JF, Boyer PD. 1953. Kinetic analysis of enzyme reactions. II. Thepotassium activation and calcium inhibition of pyruvic phosphoferase.J Biol Chem 200:669–682.

Lau FT, Fersht AR. 1987. Conversion of allosteric inhibition to activationin phosphofructokinase by protein engineering. Nature 326:811–812.

Lee J, Goel A, Ataai MM, Domach MM. 1997. Optimality of co-feedingof carbon sources for maximizing cellular and energetic yields using aconstrained network analysis. Appl Environ Microbiol 63:710–718.

Lowry OH, Carter J, Ward JB, Glaser L. 1971. The effect of C and Nsources on the level of metabolic intermediates inE. coli. J Biol Chem126:6511–6521.

Lowry OH, Passonneau JV. 1972. A flexible system of enzymatic analysis.New York: Academic Press.

Malcovati M, Valentini G. 1982. AMP and fructose 6-P activated pyruvatekinase fromEscherichia coli.Meth Enzymol 90:170–179.

Marr AG. 1991. Growth rate ofEscherichia coli.Microbial Rev 55:316–333.

Noltmann EA. 1966. Phosphogluco isomerase. Meth Enzymol 19:557–565.

Pertierra AG, Cooper RA. 1977. Pyruvate formation during the catabolismof simple sugars byEscherichia coli:studies with pyruvate kinase-negative mutants. J Bacteriol 129:1208–1214.

Reed LJ, Willms JA. 1966. Purification and resolution of the pyruvatedehydrogenase complex (Escherichia coli). Meth Enzymol 9:247–265.

Rypniewski WR, Evans PR. 1989. Crystal structure of unliganded phos-phofructokinase fromEscherichia coli.J Mol Biol 207:805–821.

Schirmer T, Evans PR. 1990. Structural basis of the allosteric behaviour ofphosphofructokinase. Nature 343:140–145.

Vierheller C, Goel A, Domach D, Ataai MM. 1995. Sustained and consti-tutive high levels of protein production in continuous cultures ofB.subtilis.Biotechol Bioeng 47:520–524.

Willecke K, Gries E, Oehr P. 1973. Coupled transport of citrate and mag-nesium inBacillus subtilis.J Biol Chem 248:807–814.

Willecke K, Pardee AB. 1971. Inducible transport of citrate in a Gram-positive bacterium,Bacillus subtilis.J Biol Chem 246:1032–1040.

Xu J, Sekis M, Denda K, Yoshida M. 1991. Molecular cloning of phos-phofructokinase I gene from thermophilic bacteriumThermus ther-tophilus.Biochem Biophys Res Commun 176:1313–1318.

Young MR, Tolbert B, Utter MF. 1969. Pyruvate carboxylase fromSac-charomyces cerevisiae.Meth Enzymol 13:250–258.

134 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 64, NO. 2, JULY 20, 1999