You are on page 1of 11

Physiology and metabolic fluxes of wild-type and riboflavin-producing Bacillus subtilis.

U Sauer, V Hatzimanikatis, H P Hohmann, M Manneberg, A P van Loon and J E Bailey Appl. Environ. Microbiol. 1996, 62(10):3687. Downloaded from http://aem.asm.org/ on April 21, 2013 by National Dairy Research Institute

Updated information and services can be found at: http://aem.asm.org/content/62/10/3687 These include:
CONTENT ALERTS

Receive: RSS Feeds, eTOCs, free email alerts (when new articles cite this article), more»

Information about commercial reprint orders: http://journals.asm.org/site/misc/reprints.xhtml To subscribe to to another ASM Journal go to: http://journals.asm.org/site/subscriptions/

3 ADOLPHUS P. The nonpathogenic bacterium Bacillus subtilis has become a model organism not only for the genus Bacillus but also for grampositive bacteria in general. American Society for Microbiology Vol. MATERIALS AND METHODS Bacterial strains. up- * Corresponding author. subtilis rib operon. 3687 .1 VASSILY HATZIMANIKATIS. a commercially important additive used in the feed and food industries. 46. the total amino acid composition of the wild type was experimentally determined and the complete building block requirements for biomass formation were derived. pRF93 is a modified version of the B. riboflavin-producing strain. Eidgeno ¨ssische Technische Hochschule Zu ¨rich. An in-depth understanding of bioenergetics. subtilis. Similar to the two modified rib operons pRF69 and pRF89 (36).1 HANS-PETER HOHMANN.00ϩ0 Copyright ᭧ 1996. the bioenergetics of growth and product formation in glucose-limited chemostat cultivations of wild-type and riboflavin-producing B. Maintenance coefficients of 0. The experimental data were further used in a detailed stoichiometric framework of metabolite (mass) balances to estimate the metabolic flux distribution. the maximum molar growth yield of 82 to 85 g (cell dry weight)/mol of glucose was found to be almost identical in both strains. riboflavin formation completely ceased at specific growth rates below 0. this flux is indicated to be increased in the strain engineered for riboflavin formation.to fivefold more than in Escherichia coli. No. In particular. which serve as prosthetic groups for many oxidoreductases (3. Flux estimates are compared with the data available in the literature. 1996. 20). their ability to secrete large quantities of protein directly into the medium has rendered them very attractive for commercial applications. G. the murein sacculus was found to constitute approximately 9% of B. was chosen as a model product to investigate the effects of product formation on energy conversion in B. surprisingly little information is available on the bioenergetic aspects of its growth and product formation (42). Glucose catabolism at low growth rates with reduced biomass yields was supported mainly by the tricarboxylic acid cycle. subtilis 1012 (leuA8 metB5) (40) were transformed by congression.org/ on April 21. 2013 by National Dairy Research Institute Continuous cultivation in a glucose-limited chemostat was used to determine the growth parameters of wildtype Bacillus subtilis and of a recombinant.3 F. various competitive fermentation processes have already been developed. revealing a coupling of product formation and growth under strict substrate-limited conditions. spo1-15. which is present at map position 135Њ of the RB50 genome. However. usually involving the fungi Eremothecium ashbyii and Ashbya gossypii (14). Switzerland. maintenance metabolism is of both scientific and industrial interest (39. three.2 AND JAMES E. subtilis generally yields final product concentrations exceeding 15 g/liter (36). particularly the maintenance demand and its impact on product formation. The genus Bacillus includes a variety of industrially important species that are commonly used as hosts in the fermentation industries (1. subtilis genetics and biochemistry. Estimation of intracellular metabolic fluxes by a refined mass balance approach revealed a substantial.66 mmol of glucose g؊1 h؊1 were determined for the wild-type and recombinant strains. Since many industrial bioprocesses are carried out in the fed-batch mode. Phone: 41/1/633 3170. key variables for a rational strain development. In particular. interrupting the nonessential bpr gene. However. subtilis biomass. The overall amino acid composition was determined for the wild type. Oct. 3687–3696 0099-2240/96/$04. CH-4002 Basel. respectively. may be necessary for the design of an optimal fermentation process (46). The resulting leucine prototrophic colonies were screened for sporulation-defective phenotypes. growth rate-dependent flux through the oxidative branch of the pentose phosphate pathway. but higher animals require it in their diet. allowing conclusions on biosynthetic requirements for growth. pRF93 contains two strong. or by a mixed process. Hoffmann-La Roche Ltd. p. which at later stages is characterized by very low growth rates. Vitamin and Fine Chemical Division. ETH Ho ¨nggerberg.asm. Mailing address: Institut fu ¨r Biotechnologie. with chromosomal DNA from the riboflavin-producing strain RB50 pur60 Agr-11 Dcr-15 Msr-46 spo0A ribC (36). wholly by microbial synthesis. In fermentations for penicillin production. On a commercial scale. subtilis strains were studied. by following standard protocols (21). Its biological function is as a precursor for the formation of the flavin nucleotides. Most prominently. CH-8093 Zu ¨rich. In this study. The phage lysate was obtained from the riboflavin-producing strain RB50 pur60 Agr-11 Dcr-15 Msr-46 spo0A ribC:: [pRF69]::[pRF93] (36). constitutive phage promoters. This is especially noticeable at lower growth rates. Competent cells of B.2 and Pharmaceutical Research-Gene Technologies. where the effects of maintenance metabolism are more pronounced relative to those of growth metabolism. 35).2 MICHAEL MANNEBERG. CH-8093 Zu ¨rich. A faster process involving the bacterium B.1 and Section Biotechnology.45 and 0. Riboflavin (vitamin B2). 62.. for example. despite the extensive knowledge of B. M. The modified rib operon pRF93 and the deregulating ribC mutation were transferred to one of the spo leuϩ colonies via PBS1 phage transduction. For molecular characterization of B. 47). about 70% of the carbon source is utilized for maintenance (22). BAILEY1* Institute of Biotechnology. Switzerland Received 5 March 1996/Accepted 24 July 1996 Downloaded from http://aem. and the problems inherent in this standard approach are discussed. subtilis. This water-soluble. 10 Physiology and Metabolic Fluxes of Wild-Type and Riboflavin-Producing Bacillus subtilis UWE SAUER.15 h؊1. Furthermore. A nonlinear relationship between the specific riboflavin production rate and the dilution rate was observed. Fax: 41/1/633 1051. Industrial production of riboflavin is achieved by complete chemical synthesis. VAN LOON. yellow vitamin is synthesized from the purine building block GTP and ribulose-5-phosphate by plants and microorganisms via seven enzymatic reactions.APPLIED AND ENVIRONMENTAL MICROBIOLOGY.

at a temperature of 37ЊC.4-ml volume of the resulting solution was neutralized with 1 ml of 0. ␯ is the vector of 21 unknown metabolic fluxes to be determined. Hewlett Packard. MICROBIOL. Cysteine and cystine residues were determined by quantification of their oxidation product following HCl hydrolysis in the presence of 0. Switzerland.3688 SAUER ET AL. the max corr following equation was used to determine Yglc and mcorr values (44): q ϭ ␮ Y max corr ϩ mcorr ϩ aq p in which a is the amount of glucose required for the formation of 1 mol of product and qp is the specific product formation rate. which were harvested by centrifugation. Therefore. and it also acts as a repressor of rib gene expression (8). ensuring dissolved-oxygen levels above 25%. Brea. formate. and the A444 was measured. The fermentation volume was kept constant by a weight-controlled pump. The samples were then hydrolyzed at 110ЊC for 24 or 72 h and subsequently dried under vacuum in a desiccator. 0. the reaction network consists of 23 metabolites and 21 reactions with unknown fluxes. 1 were determined by a least-squares fit of the data.02% 3-(2-aminoethyl)indole (32).6 g.55 g. pyruvate. Bacterial cell pellets were dried in a vacuum oven over P2O5 at 80ЊC for 24 h. extracellular protein. temperature. 0. The pH was controlled at 6. subtilis wild type and PRF93. ethanol. and ␯12). The solution to equation 1 was determined by a constrained least-squares approach with the objective of minimizing the sum of the squares of residuals from the metabolite mass balances. as well as the carbon dioxide evolution rate (qCO2). and riboflavin. Network reactions in the central metabolism were constructed primarily from the established biochemistry in the literature (19. 0. (NH4)SO4. The only constraint in the least-squares problem was a demand for nonnegative fluxes in the irreversible reactions (␯1.) as described by Bergmeyer (5) with either kits supplied by the manufacturer or chemicals and enzymes purchased from Sigma. in which the sum of the fluxes to and from any particular intermediary metabolite equals zero. and m is the maintenance coefficient (37. Denmark). and 2. respectively. max Correction of the PRF93 Yglc for riboflavin formation results in a negligible increase. was used for intracellular metabolite pools to generate the following linear mass balance equation for calculation of the flux distribution: S⅐␯ϭb (1) stream of ribA (ORF3) and ribT (ORF5). A (pseudo)-steady-state approximation. 4 g. MnCl2 ⅐ 4H2O.) on a Carbowax MD-10 column (Macherey-Nagel. typical Bacillus fermentation by-products. was considered to be converted to CO2. K2HPO4. The deregulating mutation in the ribC gene of PB50 was cotransduced with pRF93 and the tetracycline resistance marker. Sensitivity analysis on variations in biomass composition (relative protein.2 ml of 1 M NaOH. Italy). Amino acid composition analysis. Continuous cultivations in aerobic. 0. The gluconeogenic enzyme phosphoenolpyruvate carboxykinase was reported to be not active under the strict glucose-limited conditions used and was therefore not considered (12). and succinate as well as high-pressure liquid chromatography analysis of amino acids.06 g. The residues were dissolved in a diluent buffer (pH 2. which can be ex- . 3 mol of formate per mol of riboflavin. 3. 42). Elution profiles were compared with the profiles of a standard amino acid solution and of cysteic acid. MgSO4 ⅐ 7H2O. Formate formed during riboflavin biosynthesis. and dried at 85ЊC for 24 h to a constant weight (Ϯ3% variation). encoded at map position 146Њ. 2013 by National Dairy Research Institute In this representation. Precultures and batch cultures were grown in baffled shake flasks with either Luria-Bertani complex medium or Spizizen’s minimal salts medium supplemented with 20 mM glucose (21). Milan. Steadystate data for both strains were obtained from several (11 and 4. RNA. and succinate concentrations were determined on a Synchron CX5CE automated enzyme analysis system (Beckman.05 during the fermentation. is the flavin kinase of B. Downstream of pRF93. Munich.2). CoCl2 ⅐ 6H2O.06 g. FeCl3. The amino acids were separated on a cation exchanger (CK10F. Du ¨ren. Cell dry weight was determined from at least eight parallel 10-ml cell suspensions. RibC. subtilis 1012 and the recombinant. Fermentation broth samples were analyzed for cell dry weight.2 g. 0. 1.2-␮m-pore-size filter. Analytical techniques. In this study. ZnCl2. Bru ¨el & Kjær. B. When corrections were made to account for the glucose required for product formation. Amino acid analysis was performed by a modification of the method of Spackman et al. and the biosynthetic pathway to riboflavin was constructed as described by Perkins and Pero (35) and Bacher (3). formate.2% sodium azide (28. The maximum molar growth yield for max oxygen (YO ) was calculated and found to be virtually identi2 cal for both strains (Fig.asm. methionine sulfoxide. except the off-gas data for the wild-type. as shown in Fig.35 M sodium citrate (pH 3. Calif. and foam probes. and b is the vector of 23 known fluxes from measurable products and substrates for each of the 23 metabolite balances. Samples weighing approximately 4 to 5 mg were placed in a 4-ml long-neck glass vial with 1. respectively) independent cultivations. pyruvate.org/ on April 21. A bioreaction network with only branch point-associated metabolites was assembled for the calculation of intracellular fluxes. subtilis 1012 metB5 spo0A and 1012 metB5 spo0A ribC::[pRF93] were provided with a wild-type metB5 gene and designated B. Germany. and methionine sulfone standards (Sigma). RESULTS Determination of growth parameters. The medium was acidified to a pH between 2 and 3 by addition of 14 ml of H2SO4 (95 to 97%) and sterilized by being passed through a 0. optical density. A 0. equipped with pH. Mitsubishi Kasei Corp. and the vial was sealed by melting while still under vacuum. ␮ is the specific growth rate. exhibit linear relationships with the specific growth rate (assuming ␮ ϭ D at steady state).1 g. Pa. detected in the fermentations. riboflavin-producing strain PRF93. which were derived from 1 cultivation only. A constant air flow of 2 liters/min was achieved by a mass flow controller. Uxbridge. was used for the determination of extracellular protein in the supernatant. The experimental data on the continuous-growth physiology of both strains are in agreement with the chemostat model (37): steady-state specific glucose (qglc) and oxygen (qO2) consumption rates.1 M potassium phosphate buffer (pH 6. The elution was monitored by recording the A440 and A570 of the ninhydrin derivatives.0). dissolved-oxygen. Samples were taken only from chemostats in steady state. APPL. 29). 3427. 0. Selected samples were also subjected to enzymatic analysis of ethanol. Y max is the maximum molar growth yield. washed with distilled water. ␯7.02% phenol. 0. and cell wall content) and in production and consumption rates (within the confidence intervals) of the estimated fluxes was performed to analyze error propagation. For riboflavin measurements. United Kingdom) or an emission monitor (no. Concentrations of oxygen and carbon dioxide in the off-gas were determined with either a quadrupole mass spectrometer (Fisons. and aliquots were withdrawn for analysis. acetoin. Essentially identical maximum molar growth yields on glumax cose (Yglc ) of 82 and 83 g (dry weight) of cells per mol of glucose consumed were obtained for both strains as the inverse of the slope calculated by linear regression. 2 g. 1 g. 1B and E. The following equations were used for the calculation of the growth parameters in Table 1: q ϭ ␮ ϩ m Y max (2) where q is the specific rate of substrate consumption with respect to product formation. S is the stoichiometric bioreaction matrix (23 ϫ 21) based on the metabolic map.8 ml of culture broth was mixed with 0. ENVIRON. Elemental biomass composition analysis was performed on lyophilized cells with an EA 1108 elemental analyzer (Carlo Erba Instruments.16 to 0. Toulouse. glucose-limited chemostats were conducted at a working volume of 3 liters in a 6-liter bioreactor (LH-Inceltech. Acetate. the solution was frozen in dry ice and evaporated at below 1. 39). Table 1). Glucose.) and eluted with a step gradient from 0. This demonstrates an unchanged growth phenotype (Fig. which were defined as at least 5 volume changes after adjusting to a new dilution rate and stable optical density as well as oxygen uptake and carbon dioxide production rate for minimally 1 volume change. The trace element solution (1:10 dilution) was added after the filtration to prevent the formation of precipitates. in order to eliminate a potential cycle in the pyruvate shunt. Mass balances for CO2 and NAD(P)H were also included in the model. CuCl2 ⅐ 2H2O. Avondale. After addition of 1 ml of 6 M HCl in 0. because formate could not be Downloaded from http://aem. Experimental values for the maintenance coefficient (m). 1A and D. ␯10. 0. continuous cultivation in glucose-limited chemostats was used to determine the growth parameters of wild-type B.17 g. France). The recovery of tryptophan was determined after hydrolysis with 4 M methanesulfonic acid in the presence of 0. Nærum. Buchs.7-mm-thick walls. Antifoam (P2000) was present only during the initial batch cultivation but was not added to the chemostat medium to avoid any possible interference with bacterial membrane functions.24). subtilis.500 rpm. Flux balance model. Germany) with butyrate as an internal standard.3-butanediol were measured by gas chromatography (5890E. and Na2MoO4 ⅐ 2H2O. and the agitation speed was set to values between 600 and 1.043 g. Inclusion of the malic enzyme reaction does not affect any of the other net fluxes and was therefore not considered. ␯11. (43) on a Kontron Liquimat III amino acid analyzer. The linear regression lines in Fig. glucose.21 to 4. a tetracycline resistance gene was inserted for selection. Table 1).3 kPa for 3 min. Growth conditions and media. and 10 ml of trace element solution with the following composition (per liter of distilled water): CaCl2 ⅐ 2H2O.60 Ϯ 0. A kit from Bio-Rad. The medium contained (per liter of distilled water) glucose.

05–0. relative to the wild type.6 ng 25 (11) 32 (26) a b 83 0. (17). The increased maintenance metabolism for the recombinant strain. with a maximum at D ϭ 0. riboflavin production became highly unstable when D was below 0.6 4.6 Ϯ 0.07 0. 2013 by National Dairy Research Institute FIG. respectively (Table 1). subtilis 1012 (A to C) and PRF93 (D to F) during aerobic glucose-limited growth. pressed as the substrate consumption rate (mglc or mO2) or the CO2 evolution rate (mCO2) required to fulfill the non-growthassociated demand. at 0.35 91 Ϯ 2 84 Ϯ 4 97 Ϯ 3 100 Ϯ 6 58 Ϯ 13 53 Ϯ 6 0. Comparison of growth and maintenance parameters for Bacillus species Valuea for: B. However. e Corrected for product formation. No metabolic overflow products were formed under the conditions investigated.4 Ϯ 0. The estimated mglc values were relatively high.6 Ϯ 0. on the basis of the minimum detection level of 1 mg/liter.65 Ϯ 0.1 Ϯ 0.3 2. . A carbon balance of 100% Ϯ 3% was obtained at all reported growth rates. d Maximal molar growth yields expressed as grams (cellular dry weight) per mole of glucose or oxygen. Following this batch cultivation.09 0.2 and 0.14 Ϯ 0.4 21 The 95% confidence intervals based on the data shown in Fig. Data reported by Frankena et al.75 maxd Y glc 82 Ϯ 2 83 Ϯ 2 max corre Yglc 85 Ϯ 2 max YO 38 Ϯ 3 42 Ϯ3 2 mglcf 0.05–0. Effect of dilution rate on the rates of glucose (A and D) and oxygen (B and E) consumption and carbon dioxide evolution (C and F) by B. were determined by extrapolating the least-squares linear fit of the data to ␮ ϭ 0. Correcting for product formation has only a minor effect on this parameter.org/ on April 21. The specific riboflavin production rate increased up to D ϭ 0.6 mO2f 3. independent of the chemostat generation.66 for the wild-type and overproducing strains. because riboflavin productivity did not decrease more than 20% at constant D to this point. these cultures TABLE 1.45 and 0.05–0.45 hϪ1.08 0. Riboflavin formation in chemostat cultivation.5 1.0 Ϯ 0.23 Ϯ 0.11 mCO2f 3. Numbers in parentheses represent the experiments with off-gas analysis.6 3. 1996 PHYSIOLOGY AND METABOLIC FLUXES IN BACILLUS SUBTILIS 3689 Downloaded from http://aem. CO2. 1 are given. Therefore. constant decrease as a function of total volume changes (generations) in the chemostat.15 hϪ1. g The number of experiments used to calculate the parameters.12 0.10 corrf mglc 0. The substrate carbon was quantitatively recovered in biomass. 2A). (41). between 0. D. c Data reported by Santana et al.4 Ϯ 1. A similarly high maintenance coefficient value was calculated by the method of Stouthamer and van Verseveld (46) for an overproducing strain in fed-batch cultivations (data not shown). licheniformisb Nonprotease producing Protease producing ␮ range (hϪ1) 0.75 0. The product concentration in the medium was not only a function of D but also subject to a slow. and riboflavin. Stable production could be restored by subjecting these cultures to an intermediate batch cultivation with a 5% inoculation volume. and there was no residual glucose in the medium. subtilis Parameter 1012 (wild type) PRF93 168 (wild type)c B.0 Ϯ 0.66 Ϯ 0.3 hϪ1 (Fig. The highest riboflavin concentrations in glucose-limited chemostat culture were found at dilution rate. only data points from cultures with fewer than 50 generations were considered.45 hϪ1.24 Ϯ 0. 2B. 62. 1.11 0. was also reflected in the mCO2 value of this strain (50% increase). when an essentially constant maximal level was reached.45 Ϯ 0. as shown in Fig.asm.3 1. f Maintenance coefficients expressed as micromoles per gram (cellular dry weight) per hour.16 Ϯ 0.VOL.7 Ϯ 0.

because their peptidoglycan fraction was included in our analysis of total biomass. (iii) the precursors for odd-numbered TABLE 2. coli. APPL. e Corrected for hydrolysis losses. subtilis and Escherichia coli wild-type strains shown in Table 2. These chemostat cultures were phenotypically indistinguishable from PRF93 and produced normal riboflavin levels again. MICROBIOL.15 hϪ1 were cultivated again in a chemostat. The presence of the PRF operon in all analyzed single colonies was also indicated by the unchanged tetracycline resistance of these cells.910 95 60f 83 Data from reference 33. quantified through its diaminopimelate content. Determined as cystic acid.1 for fatty acids was calculated from Table 1 of de Mendoza et al. As lipopolysaccharides do not occur in the gram-positive bacilli. f Not corrected for hydrolysis losses. as well as known B. thus excluding a possible mutation mechanism. g Determined indirectly. we determined the total amino acid composition of logarithmically growing B. subtilis. (33). More than 50% of the bacterial biomass is made up of amino acids. subtilis biomass and that the relative contribution to total biomass is therefore three to five times higher than in E. Determination of monomeric cellular composition. Because diaminopimelate is used solely for cell wall synthesis. and the determination of glucosamine was subject to significant hydrolysis losses. with the known cell wall stoichiometry (2). subtilis physiology.asm.15 hϪ1). Absolute riboflavin concentration (A) and specific riboflavin production rate (B) in steady states of continuous culture of B. coli. To obtain the specific precursor requirements for B. However. The lipid composition of B. (ii) valine was assumed to be the precursor for the branched C16 fatty acids. we have quantified the detailed requirements for biomass building blocks in B. 2013 by National Dairy Research Institute FIG. coli. the cell wall requirements for alanine and glutamate are relatively small in comparison with their protein fraction. the value of 95 ␮mol/g of cellular dry weight can be used directly. coli B reported by Neidhardt et al. From the diaminopimelate data. Several differences in the composition are evident when E. we therefore conclude that the murein sacculus constitutes approximately 9% of B. subtilis are compared. d Glutamine and glutamate.org/ on April 21. is also reflected in increased levels of the other monomeric cell wall components: alanine. In particular. On the basis of the determined amino acid composition. The total amino acid composition determined for E. subtilis (Table 3). coli W3110 (total) B. to calculate the precursor requirements for cell wall formation. subtilis PRF93. subtilis Amt of amino acid (␮mol/g [dry wt] of cells) in: Amino acid E. subtilis were therefore introduced: (i) an average acetyl coenzyme A requirement of 6.260 19 32f 45 594 198 426 60c 639 495 86 288 368 342 118 183 175 239e 239e 0f 125 335 4. coli and B. Single colonies obtained from chemostat cultures that had lost riboflavin production at D lower than 0. which makes up a significant fraction of the dry weight and is clearly distinct from E. these requirements were not included. coli B/ra (in protein only) E. Asparagine and aspartate. subtilis 1012 (total) Ala Arg Asxb Cys Glxd Gly His Ile Leu Lys Met Phe Pro Ser Thr Trp Tyr Val Total Diaminopimelate Glucosamine Glycerol a b c 488 281 458 87 500 582 90 276 428 326 146 176 210 205 241 54 131 402 5. Amino acid composition of E. subtilis differs significantly from that of E. . Significant variations are seen mainly for alanine and glutamate.081 28g 28g 129g 527 273 521 90c 580 578 93 272 420 304 136 173 207 254e 275e 19f 143 395 5. coli and B. ENVIRON. and glucosamine. the following specific changes for lipid precursors of B. since mainly branched fatty acids and phosphatidylglycerol are found (11. exhibited a stable but lower riboflavin production phenotype in chemostat cultivation (at D higher than 0. On the basis of the determined molar amounts reported for E. 2. coli W3110 is very similar to the commonly used values for the amino acid composition of E. glutamate.3690 SAUER ET AL. coli (33). the nonproteinogenic amino acid. Downloaded from http://aem. 24). The higher relative portion of cell wall in B. subtilis biomass. diaminopimelate. Dashed lines indicate unstable riboflavin production. Peptidoglycan requirements were adjusted to the determined value of 95 ␮mol/g of cellular dry weight. yields useful information for precursor quantification of the gram-positive cell wall. (11).

coli and B. Experimental data on biomass and product formation rates.071 1. see the legend to Fig.413 262 50 103 95 917 Ϫ1.100 6.05 hϪ1 should be regarded as qualitative only.060 2.163 300 3.132 1.419 855 154 118 16. the most significant influence on the fluxes computed is exerted by the constraint that NADPH formation is strictly coupled to the biosynthetic NADPH requirements. NADPH was assumed to be produced only to fulfill biosynthetic requirements.5 to 7%. Such stoichiometric mass-balancing analyses depend upon precise knowledge of biochemical pathways. 23. and CO2 requirements for B. Furthermore.787 18.060 2. the oxidative branch of the PPP is not required any more.274 Ϫ3. coli values (33) were used (e. When specific growth rate and biomass yield (Yglc) increased. 3-phosphoglycerate.1 hϪ1. 3). As a result. and (iv) a phospholipid composition of 25% phosphatidylethanolamine and 75% phosphatidylglycerol was used (11). 46 to 48%. ␯15.485 Ϫ3. cofactor.013 361 129 1.928d 1. 4).488 10. were used in combination with the monomeric cellular composition of B. At dilution rates (D) of Յ0. When no specific information was available for Bacillus spp.873 16. coli in Table 4. . 1 and 2 were used to estimate the production and consumption rates.302 1. There is an uncertainty in the flux results obtained by this kind of metabolic stoichiometric analysis. subtilis biomass composition is compared with that of E. were specified by the detailed requirements for phosphoribosyl diphosphate. and estimation of the intracellular carbon flux distribution from determination of extracellular products has become a standard technique (10. 3.082 Ϫ2.488 16.833 2. Building block and cofactor requirements for the formation of 1 g of E.366 Ϫ200 Ϫ103 Ϫ76 Ϫ3. 49–51).540 1. were obtained for the riboflavin-producing strain. 18.923 Ϫ2.353 711 3. and N. and GTP.353 711 3. This causes the PPP to operate only to satisfy the cellular requirements for NADPH and pentose precursors. 1.547 18. 3. were not within the working range of a chemostat. and the NADH flux was set to equal the oxygen uptake. glycogen. Information on metabolic fluxes is of the utmost importance for rational manipulation of microbial metabolism. as well as the coupled CO2 formation. negative values indicate formation.071 1.058 Downloaded from http://aem. In the particular case reported here. the experimentally determined elemental composition of B.. However. H.088 1. 1996 PHYSIOLOGY AND METABOLIC FLUXES IN BACILLUS SUBTILIS TABLE 3.496 719d 2. because the corresponding physiological data.440 335 285 368 1. Estimation of metabolic fluxes.743 Ϫ1. DNA. an increased oxidative pentose phosphate pathway (PPP) flux (␯1) was observed for the wild-type strain at the expense of the tricarboxylic acid (TCA) cycle flux (␯14.058 For definitions of abbreviations.923 Ϫ2. because it depends on a priori assumptions regarding unknown features of the metabolic reaction network. Murein fraction of alanine and glutamate subtracted and included in peptidoglycan.079 1. GDP. branched fatty acids were assumed to be isoleucine and leucine in equal molar amounts (24). the one-carbon (C1) requirements were recalculated.274 Ϫ1. subtilis 1012 and PRF93 harvested at various dilution rates revealed no growth rate or strain-specific variations and was as follows: C.071 6. as well as on the CO2 evolution rate and O2 and glucose consumption rates. and ␯16) (Fig.132 1. the reported E. metabolic demands for growth. 62. The stoichiometric mass balance analysis employed provided an interpretation of the experimentally observed physiology by a quantitative description of the flux distribution within the defined bioreaction network and has several noteworthy characteristics. Flux values at D Յ 0. and polyamines). subtilis 1012 3691 Precursor G6P F6P R5P E4P T3P PGA Requirement (␮mol/g [dry wt] of cells)a PEP Pyr AcCoA OaA OGA CO2 ATP NADH NADPH Protein RNA DNA Lipids Peptidoglycan Glycogen C1 units Polyamines Total a b b 86 630 100 190 154 308 194 794 368 50 65 76 616 2.215 15.574 190 95 1.060 2.asm. and the actual values used for the flux analysis calculations are given in a separate column.org/ on April 21. subtilis used in analysisc G6P F6P R5P PRPP IMP GDP GTP E4P T3P PGA PEP Pyr AcCoA OGA OaA CO2 ATP NADH NADPH a b c 205 71 898 154 190 816 154 190 398 190 25 203 308 194 726 711 3.923 Ϫ3. a noticeably higher PPP flux and lower TCA cycle flux were estimated for any given D. obtained from extrapolating the relationships shown in Fig. d Corrected from reference 33. R5P and PGA requirements expressed as specific precursors corrected for cofactors. subtilis given in Table 4.277 760 76 180 16. 12 to 13%. IMP. This behavior is caused mainly by the NADPH requirements for biosynthesis of the purine precursor. see the legend to Fig. The biosynthetic requirements for ribose-5-phosphate.VOL. Specific precursor. Values of the least-squares fits of the data from the chemostat experiments shown in Fig. The calculated B. Data from reference 33. 2013 by National Dairy Research Institute 154 190 816 308 194 1. subtilis biomass Precursor metabolitea Amt required (␮mol/g [dry wt] of cells) E.873 For definitions of abbreviations. Similar patterns of PPP and TCA cycle utilization. Although supplying NADPH TABLE 4.132 1. subtilis 1012 B.225 308 194 1.g. as functions of dilution rate. 6. and NAD(P)H. coli B/rb B. RNA. 38. and optimization principles to predict the metabolic flux distribution within a defined network (Fig.811 Ϫ368 Ϫ50 Ϫ568 Ϫ76 95 59 1.

and TCA cycle reactions are shaded in grey. and such behavior has not yet been described in the literature. lysine-producing Corynebacterium glutamicum that the actual NADPH formation exceeded the biosynthetic requirements by 21. If no further assumptions regarding the biological function of the PPP under these conditions are made. metabolize glucose entirely through the PPP.4-(1H. it was shown for continuously growing. would require a net recycling of fructose-6-phosphate into the PPP. Abbreviations: G6P.org/ on April 21. which exceeds the biomass requirements for NADPH and pentose formation under many conditions (19. The reactions illustrated by dashed arrows have not been included. PPP. these mutants must have mechanisms to reoxidize NADPH. MICROBIOL. phosphoribosyl diphosphate. triose-3-phosphate.3692 SAUER ET AL. respectively. indicating the robustness of the solution for a transfer of reducing equiva- . 2013 by National Dairy Research Institute FIG. and C4 represent 5-amino-6-(ribitylamino)-2. PGA. In calculations performed with this assumption. oxoglutarate. and precursors is generally believed to be the primary biological function of the oxidative branch of the PPP. phosphoenolpyruvate. the inclusion of this constraint may force the flux calculations to yield a biased solution in which all of the substrate carbon that is not accounted for in biomass and products will be converted to CO2 in the TCA cycle. acetyl coenzyme A. which could be explained biolog- ically by a transhydrogenase activity (19). 6. 3. 5) are very similar to the solution shown in Fig. The abbreviations A. The biochemical reaction network considered. 4. E4P. reaching relative fluxes much higher than unity at the expense of the TCA cycle for all values of D (data not shown). the estimated PPP flux is very sensitive to the respiratory quotient. Thus. whereas at values higher than unity. Glycolytic. by using labelling studies combined with metabolite balancing. ENVIRON. B. A PPP flux higher than unity. glucose-6-phosphate. AcCoA. Ribu5P. and 3. we assumed interchangeability of reducing equivalents. the estimated oxidative PPP flux is significantly increased. Pyr. ribulose-5phosphate. for example. respiratory quotient values at or below unity result in increased flux through the oxidative branch of the PPP.1% (30).3H)-pyrimidinedione. pyruvate. At low D. other than the anabolic utilization of NADPH. coli. oxaloacetate. PEP. F6P. Recently. X5P. OGA. T3P. 3-phosphoglyceric acid.4-dihydroxy-2-butanone. 33). xylulose-5-phosphate. Shaded arrows indicate the withdrawal of precursors for biosynthesis. In an attempt to obtain an unbiased estimate of this flux pattern. however. the same flux is estimated to be close to zero. PRPP. ribose-5-phosphate. it has not been established that its operation is exclusively governed by this demand (15. What may be the best current estimate was obtained by allowing a realistic 20% exchange of reducing equivalents from NADPH to NADH in the optimization calculations (Table 5). Phosphoglucoisomerase-negative mutants of E.asm. Hence. for example. based on the experimental result of Marx et al. Furthermore. PPP flux is generally reported to be in the range of 20 to 30% of the total glucose uptake in bacteria.7-dimethyl-8-ribityllumazine. as demonstrated by their ability to grow on glucose as the sole carbon source (16). OaA. fructose-6phosphate. (30). S7P. seduheptulose-7-phosphate. The flux estimates obtained with this assumption for the PPP and TCA cycle (Fig. 53). Downloaded from http://aem. APPL. NADH and NADPH. erythrose-4-phosphate. The actual flux distribution is most probably somewhere between the two above-described extremes. R5P.

coli.76.54 1. which consumes 10 to 20% of the glucose input (17).51 0.75 1.31.05 0. the dilution rate at which the PPP flux decreases to zero is not significantly affected.40. 0.52. 0. 0.07 0.00. 0.28 0. 0.98 0.08 0.95. 0.09 0. subtilis.68. In contrast to the behavior of the facultative aerobic B.38.17.00 0.12 0. 0. whereas in our experiments. 0. all of the energy not used for growth is included in the maintenance term (m).08 0. Thus.04 0. The relative differences between the strains are conserved in both solutions.11 0. as shown in Table 1. 0. 2013 by National Dairy Research Institute TABLE 5. DISCUSSION Growth parameters. 0.13. the metabolic burden associated with substantial extracellular protein formation. possibly for cell wall turnover (45).68.62.22.asm.01 0. subtilis (41) is almost identical to our findings.66.08 Ϫ0. Metabolic uncoupling in B.29.00.73.24. 0.00. 1.08. subtilis devotes a relatively high portion of its energy supply to maintenance metabolism. 1. For E. 1.85.61 0.04. 0.13 0.74 1.37 0. 0.95 Ϫ0. A significant (47%) increase in the maintenance coefficient was found for the riboflavin-producing strain PRF93 despite an essentially wild-type molar growth yield. In particular.51 0. are very sensitive to maintenance requirements. 0. 0. Ϫ0.24.11 0.10 0.03 0.38 0. respectively. 0.org/ on April 21. 0. 1.98 1.07.23 0. 0. such as the commonly used fed-batch cultivation.04 0. 0. 1. 0.10. even when up to 100% higher glucose concentrations than those reported were used. factors that are by defini- Downloaded from http://aem.60 1. 0.01 0. 0.64.02. 0.77 0.04.79. licheniformis (17) and E.23. B.30 throughout the range of D investigated.81. Industrial-scale fermentations under conditions of very slow growth. 0.04. 0. 62.72 1.02.70.14 0. However.01 Fluxes correspond to the labels used in Fig.09 0. the mglc determined for wild-type B.07.24 and 0.54.98 1. coli (48.56 1.42 0. 0.30 0.07 0.21 1.26 0.3 hϪ1 0. 0. Values for the wild-type strain are followed by values for the riboflavinproducing strain. metabolic overflow in glucose-limited chemostat cultures is related to the low maximum O2 utilization rate of 15 mmol of O2 g (dry weight) of cellsϪ1 hϪ1 (51). 0.00.04 0.05 0. The glucose required for maintenance in both strains is almost completely burned to CO2.45 determined for wild-type B.48 0. Ϫ0. 0.02 0. 3. 0.61 0.68. 0.68. The sensitivity analysis of the experimentally determined biomass composition and production and consumption rates indicated that within the confidence intervals. This was demonstrated by the lack of metabolic overflow products detected in the culture supernatant. 0. 0.17 0.06 1.52 0. subtilis is a clear illustration of the influence of the intrinsic physiological properties of a microorganism on productivity in bioreactors.04.10.10 0.10.61 0.07.12. coli (51).51 0.VOL. assuming that NADPH formation is strictly coupled to biosynthetic requirements. 0. 0.10.05 0.00.26 0. 0.05.24 and 0.55. 0.03.19.55 0.82.97.03. licheniformis (Table 1). 0.86.04 0.71 0.20 0. for various dilution rates in continuous culture of wild-type and riboflavin-producing B.46.5 mmol of O2 g (dry weight) of cellsϪ1 hϪ1.1 hϪ1 0. 0. 0. potentially could further increase metabolic overflow.57 0.68. Metabolic flux estimation for the TCA cycle (A) and the PPP (B) of wild-type and riboflavin-producing B. 1.60 0.45.30 0. the maximum molar growth yield reported for batch cultivation of B.46.02 0. 0.65.58. subtilis strain did not exhibit uncoupled growth at high growth rates in the glucose-limited chemostat. subtilis strains Fluxa Value of flux for D value of b: 0. 0. 0.00. 0.07. subtilis is high compared with those for B.89 1.00. the highest O2 utilization rate was determined to be 22. This observation is further strengthened by the lower maximal biomass yield on glucose and oxygen when compared with B. 0.11. Therefore.17 1. 1996 PHYSIOLOGY AND METABOLIC FLUXES IN BACILLUS SUBTILIS 3693 FIG. 0.34.04.7 hϪ1 ␯1 ␯2 ␯3 ␯4 ␯5 ␯6 ␯7 ␯8 ␯9 ␯10 ␯11 ␯12 ␯13 ␯14 ␯15 ␯16 ␯17 ␯18 ␯19 ␯20 ␯21 ␯22 a b 0.00 0.12 0. 0. This finding indicates that B.24 0.46 0. Like B. 0. Ϫ0.5 hϪ1 0. The maintenance coefficient for glucose of 0. 1. 0. 0.28 0.00. 0.6 hϪ1 (6).61 0. 17) and E.30. The most pronounced difference was found for the PPP flux at higher D.17 to 0.44.02.61 0.42. 0. 0. normalized to glucose uptake. It should be recognized that in the relationship between growth yield and growth rate (equation 2). This has significant implications for biotechnological applications. 0.30. 0. 4. 0. because the molar mCO2 values observed are approximately 6 times the mglc values expected for complete conversion of glucose to CO2. 0. 0.55.51 0. 0. 0.31 0. 1. 0. our B.3 hϪ1 (50 to 100% of ␮max) is presumably necessary to produce the higher rates of ATP turnover required to sustain higher biosynthesis (growth) rates. 51) of 0. 0. 0. 0.00. 0. 0. subtilis.59.36 0.59.10.25. 1.46 0. Estimated fluxes.77 1.22.35 0. 0.05 0. licheniformis (6. lents from NADPH to NADH in the physiological range. these input parameters had negligible effects on the flux estimates.05 0.03. 0. The estimated flux through the anaplerotic reaction (␯11) remained between 0.46 0.13. licheniformis is capable of O2 utilization rates above 20 mmol of O2 g (dry weight) of cellsϪ1 hϪ1 but has a comparatively low ␮max of 0.59.07. Our physiological data are in good agreement with those obtained by other investigators for various Bacillus species.47 0. 0. Additionally. which is estimated to reach values of around 50% instead of 30%. 0. 1.04. 0.02 0.05 0.94. 0. licheniformis at D Ն 0.01 0.73 0.02. since the relative importance of maintenance requirements increases at lower growth rates (46). 0. 0.55.14 to 0. 0.86 0. .78.04.

subtilis. The specific riboflavin production rate is apparently positively correlated with the growth rate under strict glucose-limited conditions and is thus coupled to growth. The first and to our knowledge the only other report on metabolic flux patterns in B. 27. (18). strain PRF93 is not able to stably maintain riboflavin production. However. However.3694 SAUER ET AL. and not a decreased flux through the purine biosynthesis pathway. 4 and 5) can be explained by several differences between the mass balance equations used (18) compared with our system: (i) the energetic requirements for glucose uptake were considered in terms of ATP equivalents instead of the phosphoenolpyruvatedependent phosphotransferase system. In the latter function.org/ on April 21. 31). tion not maintenance processes. The resulting energy limitation reduces the phosphorylation potential within the cells. Nevertheless. (ii) isocitrate dehydrogenase was defined as the major source of NADPH. the relative differ- Downloaded from http://aem. ENVIRON. The most likely explanation for this effect is a metabolic perturbation. 47). for example. are not yet fully understood. since the riboflavin biosynthetic enzyme levels in our strain did not significantly exceed 1% of the total cellular protein (data not shown). it is tempting to speculate that the limited availability of the GTP precursor.3 to 4. This completely different flux pattern (compare with Fig. caused by riboflavin formation. From our data.7% of the supplied glucose (on a molar basis). such as coupling of NADPH formation to biosynthetic requirements. operation of futile cycles. The formation of riboflavin per se cannot account for the increased maintenance coefficient in the overproducing strain. the uncertainties arising from assumptions about imprecisely understood biological functions clearly illustrate the problems inherent in a purely stoichiometric flux analysis. since 50 to 100 mg of riboflavin per liter. driving higher maintenance at essentially unchanged growth yields (13. our PPP flux estimates would be too high. corresponding to 2. 34. subtilis is a minor source of NADPH. The nature of this significant increase in maintenance requirements will be the subject of further investigations. 39. If. 5. the PPP in B. form part of the parameter m. it is interesting that elevated temperatures may perturb a normal metabolic balance. The flux estimates presented here should therefore be viewed as approximations that are consistent with the data. The nature of this instability is not clear. it is also possible that the defective bpr gene in PRF93 contributes to the increased maintenance requirements. APPL. 2013 by National Dairy Research Institute . we conclude that at higher D there is a substantial flux (30 to 50%) through the PPP in B. Under continuous-culture conditions that were comparable to our experiments. coli (25. The apparent reduction of PPP utilization at low D in both strains is strongly influenced by the assumed coupling of NADPH production with biosynthetic requirements in combination with lower biomass yields and is therefore likely to be underestimated. Thus. as was shown for E. such as potential spilling mechanisms.3 to 1. For D between 0. under similar conditions. FIG. assuming up to 20% transfer of NADPH reducing equivalents to NADH. coli (9). Riboflavin formation. a TCA cycle flux in the range of 1. (iii) no O2 data were used. Flux analysis. was produced under the strict carbon-limited conditions used.1 and 0. Pleiotropic effects of the ribC mutation cannot be excluded. similar results showing relatively lower oxidative PPP flux under conditions of slow growth were obtained from [14C]glucose and [18O]glucose experiments with E.15 hϪ1. In this connection. Furthermore. Such an analysis is predicated on the assumption (among others) that all of the reactions involving the metabolites considered are known in vivo under the conditions used. 52). the normal response to decreasing guanine nucleotide levels is the initiation of sporulation (26). potentially in the purine nucleotide availability. as was observed in the present study following the production of a relatively small amount of riboflavin. This is illustrated by the minor change in the product-corrected corr mglc for PRF93 in Table 1. MICROBIOL. as deduced from a linear regression of substrate uptake rate against growth rate (37. the feeding profile should be designed in such a way as to maintain a minimal specific growth rate above 0.asm. In bacilli.5 (normalized to glucose uptake) and a PPP flux close to zero were estimated for three different steady states (D ϭ 0. Estimation of metabolic fluxes based on stoichiometric models is without doubt a valuable and easy-to-use tool. These results also suggest that for desirable production in a glucose-limited fedbatch process.15 hϪ1. there will be considerable uncertainty in some of the flux estimates derived from a strictly stoichiometric mass balance analysis. bacterial cells are known to enter a growth rate domain that is dominated by maintenance energy demands (7). The macromolecular burden of protein synthesis (4) can be excluded from having a substantial impact on the maintenance parameter.3. the presence of the ribC mutation in the riboflavin-producing strain might contribute to increased maintenance metabolism through effects other than regulation of the levels of the riboflavin biosynthetic enzymes. so that NAD(P)H balances depended only on glucose uptake and biomass formation. and slippage in membrane energy transduction. and (iv) a different biomass composition was used. which is prevented by a spo0A mutation in our strain. 0. and 0. As many reactions in the central metabolism.5 hϪ1). subtilis was based on a similar stoichiometric approach by Goel et al. Alternatively. since this protein functions not only as a repressor of rib gene expression but also as a flavin kinase. subtilis. While the absolute flux distribution may still be uncertain to some degree. this enzyme converts riboflavin to the coenzymes flavin mononucleotide and flavin adenine dinucleotide (8).15. would limit riboflavin formation. causing the oxidative PPP to be utilized only when this reaction could not fulfill the NADPH demand. Metabolic flux estimation for the TCA cycle (A) and the PPP (B) of wild-type and riboflavin-producing B.

W. Perkins. 26. and C. H. and cobalamin. 1996.. and cell cycle behavior. The central metabolic pathways of Escherichia coli: relationship between flux and control at a branch point. function. Bergmeyer. Microbiol.). and W. W. 10.). Bioeng. 29. vol. vol. 13. 1994. 42. I. 1986. G. Lopez. Gottschalk. Arbige.). 9. Diesterhaft. Ott. wish to enhance the awareness of its limitations and to avoid assumptions about metabolic functions whenever possible. Eifert. D. 1993. 36. 1995. Metabolic pathway rates and culture fluorescence in batch fermentations of Clostridium acetobutylicum. Hoch. 8. Losick (ed.. Chem. Ingraham. H. p. Losick (ed. Bulthuis. Roels. Hempfling. Fountoulakis. 24. Bacher. Application of balancing methods in modeling the penicillin fermentation. Cell. Bacterial metabolism. E. 25. Eggeling. Fraenkel. and J. Acta 301:53–70. Neidhardt. M. 1–48. Biochim. J.. M. 30. 37. F. Pero.. Mainzer. Chem. A. p. Csonka. Biosynthesis of riboflavin. Bioeng. A. A. We thank Dan R. R. J. 1965. Chem. Springer-Verlag. Microbiol. 33. physiology. de Graaf. A. A. vol. Sonenshein. Ferrance. Heijnen. van Dijl. In A. Sloma. 1979. and H. John Wiley & Sons. and R. 1990. Switzerland) in obtaining the CHN composition of our cells is gratefully acknowledged. 45. Microbial biochemistry. Rev. In G.). and M. Bulthuis. Physiology of the bacterial cell: a molecular approach. Appl. A. W. 38. A comparison between aerobic growth of Bacillus licheniformis in continuous culture and partial-recycling fermentor. Adv. 1987. REFERENCES 1. Genet. and D..-J. 1967.org/ on April 21. M. 22. Biotechnol. 3:153–167. Hancock. Biophys. 1985. C. p. and molecular genetics. W. Harwood. J. ACKNOWLEDGMENTS This work was supported by the Swiss Priority Program in Biotechnology (SPP Biotechnology). C. Rittenberg. transketolase. R. 20.C. and D. 1989. 41. 115–157.. Fermentation of Bacillus. Neidhardt. 43. In A. Jones. A. Cutting. Ann Arbor. van Verseveld. 5:149–155. H. 231:349–353. Hohmann. 381–410. Reed and H. 2nd ed. Trends Biotechnol. In J.. Stouthamer. Biochem. Washington. Santana. R. Biotechnol. Cronan. Oxidation of cysteine and methionine residues during acid hydrolysis of proteins in the presence of sodium azide. Role of pyruvate carboxylase. Biophys. Host-vector interactions in Escherichia coli.. C. efficiency of conversion to biomass. L. Weinheim. J. W.. Washington. Lasko for critical reading of the manuscript.C. and D. 40. Glaser. 1995. Mu ¨ller (ed. Losick (ed. and H. Analysis of metabolic fluxes in batch and continuous cultures of Bacillus subtilis. Vertes. Anal.. Stouthamer. J. Mass. Acta 587:238–252. and turnover. Ingraham. D. 21. A.. with contributions to the discussion on maintenance energy demand. Despite the tenor of the preceding discussion. p. 7. Schaechter. American Society for Microbiology. Magasanik. G. Spackman. J. VCH Publishers. 215–259. A. W. Riboflavin overproducing strains of bacteria. Proc. Melenhofer (ed. Manneberg. 1973. D. Florent. 1990. Washington. M. Bacillus subtilis and other gram-positive bacteria: biochemistry.C... A. Iso. S. American Society for Microbiology. Mol. 1973. phosphoenolpyruvate carboxykinase. 142–150. D. F. Germany. and R. L. M. pentose phosphate pathway. 1996 PHYSIOLOGY AND METABOLIC FLUXES IN BACILLUS SUBTILIS 3695 ences between the strains are probably realistic estimates. the contribution of the pentose cycle. Metabolic and energetic aspects of the growth of Bacillus stearothermophilus in glucose-limited and glucosesufficient chemostat culture. Microbiol. Stouthamer. 1. A. 629–653. H.and anteiso-fatty acids in bacteria: biosynthesis.). and excretion of acetate. I. Fountoulakis. VCH Verlagsgesellschaft.asm. CRC Press. Bailey. 23. 1991. 163:224–231. Manneberg. Model. In A. Application of a metabolic balancing technique to the analysis of microbial fermentation data. 1993. Sunderland. The help of EAWAG (Zu ¨rich. Arch. de Hollander. and E. folic acid. physiology.. B. and molecular genetics. The decrease of guanine nucleotides initiates sporulation of Bacillus subtilis. 32. Bacteriol. G. The effect of temperature on the molar growth yield and maintenance requirement of Escherichia coli W during aerobic growth in continuous culture. A. B. and S. L. A. J. The riboflavin regulatory gene ribC has significant homology to flavin kinase and FAD synthase.). 1979. American Society for Microbiology. P. Chemistry and biochemistry of flavoenzymes.. 1979. H. Bacillus subtilis F0F1 ATPase: DNA sequence of the atp operon and characterization of atp mutants. H. 1988. Washington. Biochemistry 6:69–79. Holms. Arch. and C.).. Coquard. Microbial energetics should be considered in manipulating metabolism for biotechnological purposes. J. Pirt. Eng. A. Biotechnol. Energetics of bacterial growth: balance of anabolic and catabolic reactions. and H. Bioeng.. Ataai. Longin. 319–334. 871–895. The estimated flux differences between the strains are rather small but nevertheless are significant. and M. and R. J. European patent application 0-405-370-A1. 28:69– 105. J. J. Arbige. Washington. 19. Ann Arbor Publishers. Freese. Shibaat. and H. van Loon. Grau. Inc. physiology. 1986. H. 44. Chemistry and biology of peptides. Biochem. Ecol. H.. 1991. Proteins. 46. In F. American Society for Microbiology. and taxonomic significance.). J. Losick (ed. A. P. Vinopal. 1986. Pennock. J. D. and C. and molecular genetics. 49:111–129. Schultz. Goel. A. Reardon. Cell wall structure.-P. van Verseveld. J. R. Variations in its activity in Escherichia coli with growth conditions. Microbiol. Quayle (ed.. Gen. P. A. and R. J. Fraenkel. Kunst. When nutrient limitation places bacteria in the domains of slow growth: metabolic. and S. 1993.. Tempest. Biochemistry 6:2227–2247. rather. 1993. G. 252:3382–3391. Harwood. Deerfield Beach.. p. Stein. Downloaded from http://aem. 5. Lahm. and R. Boca Raton. 126:251–256. Russell. Crabb. Biol. p. Escherichia coli and Salmonella typhimurium: cellular and molecular biology. D.. 15. Bacillus subtilis and other gram-positive bacteria: biochemistry. M. S. Sahm. D. FEBS Lett. 1987. 2. Hoch. 10:247–256. T. and malic enzyme during growth and sporulation of Bacillus subtilis. 55:288–302.VOL. and E. Sonenshein. Sonenshein. American Society for Microbiology. In F. Rev. J. 176:6802–6811. 21. Regul. K. 59:48–62. 150:452–459. L. D. Hoch. T. U. A continuous culture study of the bioenergetic aspects of growth and production of exocellular protease in Bacillus licheniformis. New York. 1990. 1967. B. Bacillus subtilis and other gram-positive bacteria: biochemistry.. 3. Antonie Leeuwenhoek 60:275– 292. Bacteriol. R..-W. L. H. 67:359–363. and A. 224:122–127. 1993. B. Rognstad. The search for correlation between theoretical and experimental growth yields. Stouthamer. 1993. In A. A. M. 21:2175–2201.. Rev. synthesis. M. 152:499–507. M. Huccas. Stouthamer. Vitamins. Scheper. Jr. 2013 by National Dairy Research Institute . van Verseveld. Inc. C. Fla. Danchin. R. B. Biochem. since only a small difference exists between the two strains in terms of Yglc and product formation under the conditions used. Hoch. Harwood. vol. Molecular biological methods for Bacillus. H. American Society for Microbiology. J. J. A. 1985. Losick (ed. E. Marks. E. W. G. Pero. 1991. W. The labeling of pentose phosphate from glucose-14C and estimation of the rates of transaldolase. In J. 1972. 18. Annu. Umbarger (ed. Chesbro. J.). J. Sonenshein. IV.. Pathways of NADPH formation in Escherichia coli.C. Fla.-W. Biotechnology. Chichester. p. Fraenkel. S. Moore. E. 248:6062–6070. J. H. de Mendoza. and A. J. Biochim. H. M. 1991. England. 31. T. E. biotin. and ribose phosphate synthesis. J. 4. 74:103–119. A.. p. Bacillus subtilis and its relatives: molecular biological workhorses. Frankena. K. Utilization of energy for growth and maintenance in continuous and batch cultures of microorganisms. Soc.. and R. J. Measurement of the activity of the hexose monophosphate pathway of glucose metabolism with the use of [18O] glucose. FEMS Microbiol. W. and A. 16. we do not advocate avoiding mass balance-based analysis but. 42:686–696. A. Microbiol. Prog. Biotechnol. and R. M. Anal. Biotechnol. Mich. Moore. S. physiology. 6. and molecular genetics. Biol. Methods of enzymatic analysis. G. C. L. 1976. 34. 11. Wiechert. Bailey. Farmer. 27. 30: 1190–1206. 1977. p.. J. M.-H. A. Archibald. 170:117–122. J.. and T. R. Microbiol. London Ser. Marx. and D. and J. Cook. D. Bacillus subtilis and other gram-positive bacteria: biochemistry. physiology. Jeong. Lahm. G. H. D. D. 1976. and EntnerDoudoroff pathway. and M. D. L. 411–421. C. Katz. Stouthamer. M. 1958. 22:169–176.C. and P. Top. 1992. Determination of the fluxes in the central metabolism of Corynebacterium glutamicum by nuclear magnetic resonance spectroscopy combined with metabolite balancing. 27:69–100. Kaneda. Sinauer Associates. Biotechnol. The maintenance energy of bacteria in growing cultures. Washington. Submitted for publication. T. and R. Baltimore. and H.. Automatic recording apparatus for use in the chromatography of amino acids. B. J.. M. Low. L.). University Park Press. Ando. Schaechter.. B. Sonenshein. 1987. Biosynthesis of flavins. 1973. J. Bettenhausen. and M. A. Quantification of cysteine residues following oxidation to cysteic acid in the presence of sodium azide. 62. R. Perkins. M. and J.. N. Effects of growth temperature on yield and maintenance during glucose-limited continuous culture of Escherichia coli. and molecular genetics. 17. J. Saito. B. morphologic. Curr. M. 4. Biosynthesis and function of membrane lipids. L. J. G. 14. Hoch. 35. and G. 39. S. 1979. Bacillus subtilis and other gram-positive bacteria: biochemistry. Koningstein. Freese. 12. Carbohydrate metabolism in bacteria. Rehm (ed.C. V. M. 1993. Glycolysis. H. 1995. Mapping of genes determining nonpermissiveness and host specific restriction to bacteriophages in Bacillus subtilis Marburg. 48:29–52. Trends Biotechnol.. L. 28. Anal. Ionescu.

Tempest. and B. A.3696 SAUER ET AL. Environ. B.. D. Sikdar. M.. Flux determination in cellular bioreaction network: application to lysine fermentation. O. M. 1984. J. Ingraham. Press. Inc. and O. C. Lett. 50. 1994.). 47.. M. 51. 52. Inc. Wood. Growth yield and energy distribution.org/ on April 21.C. Wallace. 49. scientific. and G. Neijssel..). Bio/Technology 12:994–998. In S. M. B. H. Microbiol. Holms. O. Academic Press. Varma. CRC Downloaded from http://aem. Varma. Maintenance coefficients and rates of turnover of cell material in Escherichia coli ML308 at different growth temperatures. T. Fla. 1990. R. and P. D. 1986. Palsson. 37:317–320. 38:459–486.. Schaechter. Neidhardt. Fla. 2013 by National Dairy Research Institute . 1994. American Society for Microbiology. Stoichiometric flux balance models quantitatively predict growth and metabolic by-product secretion in wildtype Escherichia coli W3110. Metabolic flux balancing: basic concepts. In F. Todd (ed. 1985.. Microbiol. and H. L. D. Frontiers in bioprocessing. MICROBIOL. Annu. The status of YATP and maintenance energy as biological interpretable phenomena. Tempest. Bier. W. p. Magasanik. A. The pentose phosphate pathway. and W.. Escherichia coli and Salmonella typhimurium: cellular and molecular biology. E. Stephanopoulos. Boca Raton. D. 205–219. ENVIRON. Rev. Orlando. Appl. Palsson.. Umbarger (ed. W. FEMS Microbiol. J. 48. and O.asm. 60:3724–3731. and practical use. Vallino. 53. Washington. Neijssel. Low. and B. p. K. K. 1987. 797–806. APPL. J.