Determination of Flux through the Branch Point of Two Metabolic Cycles THE TRICARBOXYLIC ACID CYCLE AND THE GLYOXYLATE SHUNT*

The branch point of the tricarboxylic acid and gly- oxylate shunt has been characterized in the intact organism by a multidimensional approach. Theory and methodology have been developed to determine veloc- ities for the net flow of carbon through the major steps in acetate metabolism in Escherichia coli. Rates were assigned based on the 13C NMR spectrum of intracellular glutamate, measured rates of substrate incorpo- ration into end products, the constituent composition of E. coli, and a series of conservation equations which described the system at steady state. The in vivo fluxes through the branch point of the tricarboxylic acid and glyoxylate cycles were compared to rates calculated from the kinetic constants of the branch point enzymes and the intracellular concentrations of their substrates. conditions a

The branch point of the tricarboxylic acid and glyoxylate shunt has been characterized in the intact organism by a multidimensional approach. Theory and methodology have been developed to determine velocities for the net flow of carbon through the major steps in acetate metabolism in Escherichia coli. Rates were assigned based on the 13C NMR spectrum of intracellular glutamate, measured rates of substrate incorporation into end products, the constituent composition of E. coli, and a series of conservation equations which described the system at steady state. The in vivo fluxes through the branch point of the tricarboxylic acid and glyoxylate cycles were compared to rates calculated from the kinetic constants of the branch point enzymes and the intracellular concentrations of their substrates.
As the study of cellular metabolism progresses, it is becoming increasingly apparent that regulatory processes must be understood in the context of their cellular environment. Historically, our understanding of metabolic regulation has been developed through in vitro studies with purified enzymes. The results of these studies can be misleading since enzyme assays are generally performed under nonphysiological conditions such as low enzyme levels or high concentrations of substrates and putative regulators. Thus, a combination of in vivo and in vitro data is needed and ideally these should be applied to a critical regulatory step.
In bacteria, a key metabolic branch point exists between the tricarboxylic acid and glyoxylate cycles when cells are grown on acetate as the sole carbon source. The substrate at this branch point, isocitrate, can react with isocitrate dehydrogenase leading into the energy-producing steps of the tricarboxylic acid cycle, or isocitrate lyase which leads into the glyoxylate cycle (1) (Fig. 1). The latter cycle is essential to the bacteria during growth on acetate since without it both carbons would be burned to COz with no net formation of intermediates which are required for the synthesis of cellular materials. It was discovered some years ago by Holms and Bennett (2) that isocitrate dehydrogenase is regulated. Recently it has been established by Garnak and Reeves (3) and in our laboratory (4, 5 ) that this regulation is achieved by reversible phosphorylation. The kinase/phosphatase which acts on isocitric dehydrogenase has been purified and some * This research was supported in part by National Science Foundation Grant PCM7516410 and National Institutes of Health Grant AM09765. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertiement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. kinetic aspects of this regulation have been clarified (5, 6).
In this study, we have performed in vivo measurements to characterize the branch point between the tricarboxylic acid and glyoxylate cycles in Escherichia coli. Methodology was developed to measure flux, from the nutrient acetate through the major enzymatic steps which provide the system with cell constituents and energy. In the first part of the article, the theoretical approaches are described, and this is followed by the experiments which allowed us to determine values for each reaction. The measurements of velocities through the cycles allowed us to calculate the total rate of ATP production and the efficiency of ATP utilization by the cell.
Preparation of Cells-E. coli 23559 (Met-) cells were grown in MOPS minimal medium (7) containing 100 mM acetate and 0.2 mM methionine. Cells were grown to exponential phase on a rotary shaker at 37 "C. Prior to labeling, enzyme extraction, or small molecule extraction, the cells were harvested by centrifugation, washed, and resuspended in fresh growth medium containing 100 or 25 mM acetate. All growth experiments were performed with aerated cells growing at 37 "C. NMR spectra were obtained at 45.3 MHz using a Bruker 180 superconducting magnet interfaced to a Nicolet 1180 computer. Spectra were obtained using a 56" pulse width and a spectral width of 6024 Hz. Extracted metabolites, purified intracellular glutamate, or standards were dissolved in 98% D,O containing 10 mM sodium azide arr d 250 mM KC1 at pH 5.1. Chemical shifts are expressed relative to [99%-2-l3Cjacetate as the internal standard (6 = 25 ppm).
I3C-enriched glutamate was purified from cellular extracts by Dowex 50 chromatography. In this purification, the extract was labeled with carrier-free ~-(3,4-~H)glutamate prior to chromatography. The 3H-containing fractions were pooled and desalted on a Dowex 2 column. Glutamate was eluted from this column with 1 N acetic acid which was later removed by rotary evaporation.
Chromatography-High performance liquid chromatography was performed with a Waters Associates model 6000A The rate of acetate incorporation ( VCC) refers to the accumulation of acetate from the medium into total cell constituents. To measure this value, aliquots of labeled cells were removed at various times, added to 10 ml of unlabeled medium, and rapidly collected on 0.45pm Millipore filters. The filters were washed with an additional 10 ml of unlabeled medium and placed in scintillation fluid. Less than 3% of the accumulated [1,2-"Clacetate was lost during this wash procedure.
The measurement of rate of [1,2-"C]acetate incorporation into a saponified, chloroform-soluble fraction ( V%pn) was similar to the VCC measurement except the filters were immersed in chloroform:methanol:water (1:200.8) after washing. Filters were removed from the extraction fluid and total lipid was isolated by a modification of the quantitative procedure of Bligh and Dyer (8) as described (9). Lipids were saponified by refluxing at 80 "C for 2 h in 0.5 M methanolic KOH (10). Methanol was removed by rotary evaporation and the residue was acidified with prior to the re-extraction of fatty acids with chloroform. Chloroform was removed before the radioactivity was determined by scintillation counting. The rate of radiolabel incorporation into the saponified fraction was 70% of the rate into the nonsaponified fraction.
The rates of radioactive carbon efflux as extracellular end products ( VEK) were determined by scintillation counting after the medium was chromatographed by HPLC. At intervals, aliquots were removed from the labeling mix and the medium was rapidly separated from the cells by centrifugation through a 0.45-pm microfilter (Bioanalytical Systems, Inc.). Samples of bacterial medium were injected in 20pl portions and organic acids were separated on an HPX-87H column (see above). Extraction of Metabolic Intermediates-Intermediates were extracted by the formic acid method of Payne and Ames (11). Cells were collected by rapid filtration through Millipore 1.2-pm nitrocellulose filters. Immediately after filtration, the filters were immersed in 8 ml of cold 1.0 N formic acid. Four ml of water was used to wash the filter and was combined with the acid extracts. Debris was removed by centrifugation and the material was lyophilized to dryness. This residue was stored at -20 "C and was resuspended in water prior to I3C NMR spectroscopy, HPLC, or metabolite level measurements.

\ Cell Constituents
When NADP+ and NADPH levels were measured, cells were collected on 0.45-pm Nylon-66 filters (Ranin Instruments) and alkaline extracts were prepared to prevent the decomposition of NADPH (12).
Metabolite Level Measurements-Metabolite levels in extracts were determined fluorometrically (12,13). All measurements were made with a Spex Fluorolog which was equipped with a refrigerated RCA C31034 photomultiplier tube. Excitation and emission wavelengths were 352 and 474 nm, respectively, and the band pass was 10 nm.
Isocitrate Dehydrogenase Extraction and Assay-Isocitrate dehydrogenase activity was determined in cell extracts which were prepared by a modification of the method of Holms and Bennett (2). In this technique, 1 ml of 16 mM EDTA (pH 7.6) and 24 mM 0mercaptoethanol were added to 6 ml of culture in an aluminum sonication rosette. Essentially complete cell lysis was achieved by four 35-s bursts of ultrasonic vibration with a 25-9 pause between each burst. A Heat Systems model W-220 cell disruptor was used at a setting of 5 with a 1/2-inch probe. Diluted cell suspensions were cooled by immersing the rosette in an ice-water slurry. After cell lysis, extracts were assayed for isocitrate dehydrogenase activity at 37 "C by following NADP+ reduction spectrophotometrically at 340 nm. One-ml reaction mixtures contained 15 mM MOPS (pH 7.5), 5 mM MgC12, 200 pM NADP+, 250 p M threo-D.-(+)-isocitrate and 200 p1 of the sonicated extract. The maximal velocity was 22% greater than the measured velocity as calculated with the Michaelis constants for NADP+ and isocitrate. Isocitrate Lyase-Isocitrate lyase activity was assayed at 37 "C using the sensitive stop-time assay described by Roache et al. (14).  Table I.
The fluxes through the steps in Fig. 1 could be determined after making the measurements which are shown in Table II. The experimental observations in Table II were related to the theoretical equations in Table I by solving simultaneous equations and by considering the origin of various cellular materials. These relationships are shown in Table III. For example,    12). In addition to these equations, the ratio of the velocity through isocitrate dehydrogenase relative to the flux through the glyoxylate bypass was determined by examining the distribution of 13C atoms in the glutamate produced or by comparing the relative rates of 14C02 release from [l-"Cl-and [2-14C]acetate. The rationale for this analysis is shown in Fig. 2 and will be discussed in detail in the text.
The relationships used to calculate the fluxes to a common standard were: (a) suspensions of bacteria at 0.1 Asso contain 47 pg of cells dry weight/ml; (b) the total volume of cells were 2.5 ml/g of dry weight; and (c) since the periplasmic volume is 25% of the total cell volume (19), the rates and concentrations were calculated using 0.75 of the total cell volume because we are dealing with cytoplasmic processes. Values close to these have been used by others for calculating various properties of a cell (20). Since the same values were used for all calculations, the relative values of flux rates can be compared directly.

RESULTS
Isotopic Steady State-An isotopic steady state exists when the specific activity of cycle intermediates does not change with time. In this system, an isotopic steady state was indicated by three lines of evidence. First, intermediates were extracted at intervals following the addition of the carrierfree (1,2-W)acetate. These intermediates were separated on an HPLC column giving a typical profile as shown in Fig. 3. The peak that eluted at the same position as a citrate standard was pooled. Greater than 95% of the radioactivity in this peak was present as citrate as determined by thin layer chromatography in a chloroform:methanol:formic acid (80:20:1) solvent system. The radioactivity eluting at this peak at 1, 3, 6, and 12 min after addition of the radiolabeled acetate to the medium were 1173, 1367, 1051, and 1316 cpm, indicating that isotopic equilibrium had been reached in the cycle within 1 min.
A second indicator of isotopic steady state was the accumulation of radioactivity in COZ as shown in Figs. 4 and 7. The lack of a lag period indicates linear accumulation of CO, and rapid isotope equilibrium. Linear rates were also obtained for the fatty acid fraction and the amount of label in total cell constituents (Fig. 4). After 30 min, the plots became parabolic because cell growth became a significant factor.
Another line of evidence ccnsistent with the presence of an and COz are indicative of the relative contributions of the Krebs and glyoxylate cycIes in the formation of these compounds. This was the basis for determining the distribution of the flux through the branch point. The specific activity of the glutamate at a steady state was determined assuming equilibration of isotope between malate and fumarate. l o 2 enriched intracellular glutamate (Fig. 6). As will be discussed below, the splitting pattern of certain positions indicates that they are enriched with I3C to the same degree as the [99%-2-Wlacetate in the medium.
Measured Rates-The experimentally observed rates listed in Table I1 were obtained after isotopic steady state had been Derived Relationships-From the direct measurements of radioactive label and the theoretical relationships of Tables I1 and 111, some of the intracellular rates can be obtained.
The determination of Vsap0, establishes the rate VFA, which is the rate of acetyl coenzyme A incorporation into saponified fatty acids via the acetyl-coA carboxylase step (Equation 13).
The observed rates of VCO~, VCC and VEff given in Table I1 I 1200 I I  13C NMR-The enrichment pattern of glutamate and other metabolites has been useful in determining the relative contributions of pathways which lead to their synthesis from glucose (21, 22) and pyruvate or acetate (23)(24)(25). We have used this technique to study acetate metabolism in E. coli. metabolites from a culture which was grown aerobically in the presence of [99%-2-13C]acetate for 10 min. The most intense peaks in this spectrum had chemical shifts which correspond to the C-2, C-3, and C-4 of glutamate. The strong signal from glutamate is due to the high intracellular concentration of this metabolite (50 mM). Since glutamate had a strong signal and its labeling pattern reflects the isotopic distribution of a-ketoglutarate, we purified this compound from the extract (see "Experimental Procedures"). The spectrum of 13C-enriched glutamate is shown in Fig. 6.
The 13C-13C splitting of the C-3 and C-4 peaks indicates that intracellular glutamate was highly enriched by the [99%-2-'3C]acetate. This splitting pattern and the absence of a C-5 peak at 187 ppm (Inset A of Fig. 6) indicates that there were two predominant glutamate isotopomers, [2,3,4-13C]glutamate and [1,2,3,4-13C]glutamate. The complex splitting pattern of the C-2 peak arises from the superposition of the spectra from the two species.
The relative peak intensities of the purified intracellular glutamate were 0.1:1.01.1:1.1:0 for C-1 through C-5, respectively. The variations in the resonance intensity at the dfferent carbon positions is due to the nonstatistical distribution of the 13C label as well as differences in relaxation time and nuclear Overhauser enhancement. To determine the actual isotopomer ratio, natural abundance glutamate was run under identical conditions of pulse width, cycle time, pH, etc., to account for the differences in relaxation times and nuclear Overhauser enhancement at the different positions (Inset B of Fig. 6). Correcting for these differences, we found that the relative enrichment of the intracellular glutamate was 0.4:1.0:0.9:1.0:0 for C-1 through C-5, respectively. This indicates that approximately 40% of the glutamate was in the 1,2,3,4-I3C isotopomer form.
Determination of Flux at the Branch Point-The determination of the flux distribution at the branch point was complex because both cycles lead to common products. The logic of the determination was deduced from Fig. 2 as follows. The atoms of acetyl coenzyme A and oxalacetate are labeled a, b, and r, s, t, u, respectively. Next to each of the key compounds the predicted relative labeling is given for one turn of the cycle. For example, the formation of succinate, which is symmetric, scrambles the label between the two carboxyls and also between methylene carbons and therefore the specific activities become (b + s)/2 and (a + t)/2, respectively, for the succinate derived from a-ketoglutarate. The malate formed from the glyoxylate shunt is asymmetric initially but, if there is equilibration with fumarate, the C-1 and C-4 as well as the C-2 and C-3 of malate are scrambled. The oxalacetate formed at the end of the first cycle will be derived from two fourcarbon compounds produced by the glyoxylate cycle and the one four-carbon compound coming from the Krebs cycle. At steady state, the specific activity of every carbon atom will approach a fixed function of a and b depending on the relative contributions of each pathway. In Fig. 2, the predicted steady state specific activities are given for glutamate and CO, based on the "Krebs cycle only" and the "glyoxylate cycle only." Note that only C-1 of glutamate and CO, have specific activities which reflect the relative fluxes between pathways. The proportion of the flux through isocitrate dehydrogenase, fk, can be related to the specific activity of the C-1 position of glutamate, G,,,bs (Fig. 7) which gives a value of 0.72 for fk. If the flux through the branch point is calculated from the 14C02 data, no assumptions regarding the extent of isotope equilibration by fumarase are required. This is true because the 14C02 released by the Krebs cycle will represent a mixture of isotopes from the C-1 and C-4 of ~xalacetate.  acetate. Reactions were initiated by the addition of carrier-free label to cultures which were growing on 25 mM acetate. Con evolution was determined as described under "Experimental Procedures." fk value determined from the NMR data, the calculated fluxes through the shunt, isocitrate dehydrogenase and citrate synthase, were 24, 94, and 118 mM/min, respectively.
A confirmation of the value of Vt was obtained from the rate at which (1,2-I4C)acetate was incorporated into chloroform-insoluble, cell constituents (VCC -VS,,,). At steady state, the rate of flux through the glyoxylate bypass is equal to the total velocity of the reactions which deplete the cycle intermediate (Equation 11). Since the flow of intermediates through these pathways predominantly leads to the biosynthesis of chloroform-insoluble constituents, the flux through the bypass can be related to this experimental parameter (Equations 14 and 15). Since V G~~D H or VME plus VpEpc~ cannot exceed the flux through the glyoxylate bypass at steady state, limits can be set on the value of VL (Equation 16).

8, PPm
Using this relationship, the flux through the bypass was calculated to be between 26 and 43 mM/min. These limits agree well with the values determined independently for VL by the methods described previously. Other Steps in the Two Cycles-Having obtained the velocities through the branch point, it was possible to determine the flux through other steps in the two cycles using the conservation equations of Table I and by performing the appropriate additions and subtractions (Fig. 8). Velocities were assigned to the cycle-depleting reactions based upon composition data for E. coli and Equation 11. This equation demonstrates that at steady state the flux through the glyoxylate bypass (31 mM/min from the CO, data) must equal the sum of the fluxes which deplete cycle intermediates, or VCI~DH + VME + V~EPCK + VOAANH, = 31 mM/min. The ratio of the biosynthetic flux through these enzymes was calculated from the known weight percentages of cellular constituents which are derived from the different cycle-depleting reactions. For example, intermediates depleted through glutamate dehydrogenase will give rise to glutamate, glutamine, arginine, proline, and polyamines which comprise a total of 10.8% of the dry weight of E. coli (15,26). The flux through glutamate-oxalacetate transaminase gives rise to 15.6% of the dry weight and the gluconeogenic flux through the malic enzyme and phosphoenolpyruvate carboxykinase leads to the biosynthesis of 22.6% of the dry weight in protein and 16.6% in polysaccharide (27). Using these weight percentages, the ratio V, , , H: Vo, , H, : VpEpC, + VME was calculated at 1:1.4:3.6. The values for lipid, RNA, and DNA were not included in these calculations! Assuming the gluconeogenic flux is shared equally by the malic enzyme and phosphoenolypyruvate carboxykinase, the following velocities were assigned to the intermediate depleting reactions: VCluDH = 5 mM/min, VOAANH, = 7 mM/min, vM, = 9.5 mM/min, and VPEPCK = 9.5 mM/min. Using these values, velocities could be In this calculation, it was assumed that the biosynthesis of RNA and DNA do not significantly affect the ratio of the fluxes through the cycle-depleting steps. The weight percentage value for fatty acids was not included in the calculation because they are synthesized from carbon before it enters the cycle. External assigned to all of the major pathways associated with the early stages of acetate metabolism and this is summarized in Fig. 8. Upon closer examination, it can be seen that these assignments predict that the value of VCO, is 174 mM/min. Experimentally, we have determined that this value is 150 f 7 mM/min. The 16% discrepancy most likely arises from a slight error in our determination of the distribution of the flux at the branch point since this will have a significant influence on the velocities calculated for the other steps in the cycle. For example, using the larger fk value, which was calculated from the NMR experiment, the discrepancy was 32% between the measured and the calculated values. In Vitro Rates and the Concentrations of Intermediates and Enzymes-To further characterize the system, we have determined the intracellular levels of intermediates and enzymes  (Table IV). Some of these values have been determined before with different strains of E. coli and, in general, the values agree well (13). The intracellular concentrations of isocitrate lyase and isocitrate dehydrogenase were calculated from the purification data. Homogeneous preparations of lyase were obtained after a 22.6-fold purification indicating that the concentration of 45-kDa subunits is 200 p~ inside the cell. Similar calculations with purified isocitrate dehydrogenase indicated that the intracellular concentration of this enzyme was 40 p~. The total concentration of isocitrate was 160 p~ and the concentration of free isocitrate was calculated assuming hyperbolic binding to the lyase and the dehydrogenase and that the Michaelis constants were equal to the dissociation constants. These calculations indicated that approximately half of the total isocitrate was bound to enzyme. A similar situation probably exists for NADP+ but we could not calculate the levels of the free dinucleotide due to the prevalence of NADP+-linked dehydrogenases in E. coli.
Using the kinetic constants of the enzymes (Table IV), we calculated the fluxes through the branch point as an independent test for the rates measured in uiuo. Using the free isocitrate value, the calculated rate through the lyase was 39 mM/min. This agrees well with the flux of 31 mM/min observed in uiuo. Similar calculations were made for the dehydrogenase using the parameters in Table IV. When bacteria grow on acetate, much of the isocitrate dehydrogenase is in the phosphorylated form which is inactive. Under these conditions, the dehydrogenase has a maximal velocity of 138 mM/ min in the cell. The intracellular velocity through the dehydrogenase was calculated using a random bireactant rate equation with (Y = L 6 The calculated rate of 80 mM/min agrees exactly with the observed rate. This agreement is fortuitous since the calculations did not account for the amount of NADP+ sequestered by enzymes in the cell. Studies are currently under way to refine the calculations from in vitro data, but the data reported here are consistent with the i n vivo data.

DISCUSSION
The above experiments and theoretical approaches have allowed us to assign a value for the major enzymatic steps in both the glyoxylate cycle and the tricarboxylic acid cycle (Fig.  8). Since bacteria do not have compartments and numerous parameters were measured experimentally, it was possible to assign velocities for all steps by assuming only that (a) the major pathways of acetate metabolism have been correctly identified in Fig. 1 and (b) that the ratio of weight percentages of various cellular materials correctly estimates the ratio of fluxes which deplete cycle intermediates. The accuracy of these values can be evaluated in several cases since they were determined by more than one method. For example the rate of COz evolution was measured directly ( Table 11) and could be calculated by summing the fluxes of the major C0,-evolving steps (Equation 10). The discrepancy between the two values was 16% and this is probably a good indicator of the error in the values. The velocities assigned to the steps which deplete cycle intermediates probably have larger errors.
Isocitrate lyase and isocitrate dehydrogenase compete for isocitrate a t a major metabolic branch point. The conservation equations reveal that a change in the rate of intermediate flux through the lyase would necessitate an equal change in the rate at which intermediates are depleted for biosynthesis (Equation 11). Thus, at this branch point, the cell coordinates biosynthetic fluxes with the energy-producing flux through the oxidative steps of the Krebs cycle. The ratio of the flux calculated for the branch point is a critical parameter in the assignment of velocities to the other steps in the two cycles. The distribution of the flux at the branch point was determined by several methods: 13C NMR, the relative ratio of 14C02 evolution from [1-l4C]acetate and [2-14C]acetate, and calculations from kinetic constants. Additionally, limits were set for the flux through the bypass from the rate of carbon incorporation into chloroform-insoluble cell constituents ( VCC-VSapon). The values obtained from these different measurements were similar, indicating that we have correctly determined the ratio of the flux through the branch point.
The measurement of metabolite and enzyme levels at the branch point revealed that the concentration of isocitratebinding sites exceeded the total concentration of this intermediate. This suggests that a large proportion of isocitrate was bound to enzymes in vivo. A similar situation is believed to exist with fructose 1,6-diphosphate and oxalacetate in rat liver (28). The fluxes through the branchpoint were calculated from the kinetic constants of the enzymes, the total concentration of NADP+, and the calculated concentration of free isocitrate. These values were in agreement with the in vivo measurements, indicating that our i n vitro assays are performed under conditions which reasonably approximate intracellular conditions.
Besides the quantitative data, 13C NMR measurements provided additional information about the system. The proton-decoupled spectra of purified intracellular glutamate showed that there were two predominant isotopomer species present, [2,3,4-13C]glutamate and [1,2,3,4-'3C]glutamate, when cells were grown on [99%~-Z-'~Cjacetate (Fig. 6). These are the same two species which are predicted from the flow of carbon through the pathways outlined in Fig. 1 a t steady state. The lack of other isotopomer species is consistent with the supposition that the pathways in Fig. 1 are the quantitatively important reactions in acetate metabolism. For exam-ple, the absence of 13C enrichment in position 5 of glutamate suggests that there was no significant futile cycling through phosphoenolpyruvate carboxykinase, pyruvate kinase, and pyruvate dehydrogenase. The cycling of carbon through this pathway would scramble the label in Ac-CoA and be reflected by the appearance of 13C-nuclei in the C-5 position of glutamate. The lack of 13C-nuclei in this position is indicative of regulation at the level of pyruvate kinase or pyruvate dehydrogenase which promotes a gluconeogenic flux. This is in contrast to experiments in yeast where 13C measurements indicated significant futile cycling through these pathways (23). The presence of two predominant isotopomers of glutamate was also indicative of a minimal dilution of cycle intermediates by nonlabeled endogenous compounds or by the incorporation of nonlabeled CO,. Dilution by nonlabeled carbon would be expected to give rise to multiple isotopomers and this would result in a more complex splitting pattern than was observed in Fig. 6.
In the past, studies of microbial energetics have primarily been performed under anaerobic conditions since ATP yields could be precisely calculated from substrate catabolism (27,29). Under aerobic conditions, the molar production of ATP has been calculated from oxygen consumption data but this relationship is complex and can only be used as an approximation (30-32). Using the fluxes listed in Fig. 8, we were able to exactly calculate the intracellular rate of ATP production during aerobic growth on acetate. This value was 855 mM/ min. Since the intracellular concentration of ATP is approximately 2 mM (13), the half-life of this metabolite is less than 0.1 s under these conditions. From the calculated ATP-requiring pathways deduced by Stouthamer (27), 301 mM ATP/ min is needed for biosynthesis and transport during growth on acetate. This gives an overall efficiency of 35%, for the ATP synthesis-hydrolysis cycle. The uncoupling of energy production from growth may be due to inefficiencies which result from the high metabolic rate of bacteria relative to eukaryotic organisms (32). Extensive studies in eukaryotes indicate that the rates of carbon flow through the major metabolic pathways are substantially less than those found in E. coli (33-36).