INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Escherichia coli excretes acetate as a major by-product of its aerobic metabolism (3). Acetate is produced from acetyl coenzyme A (CoA) via acetyl phosphate by the Pta-AckA (Pta, phosphotransacetylase; AckA, acetate kinase) pathway, which is reversible and constitutively expressed (6). The excreted acetate is reused via acetyl CoA synthetase (ACS) or the Pta-AckA pathway after the depletion of favored carbon sources such as glucose. ACS or the Pta-AckA pathway, showing higher or lower affinities for the acetate, respectively, activates acetate to acetyl CoA (24).

Aerobic acetate production depends on the growth rate. When the growth was limited by restricting the feeding rate of the carbon source, acetate excretion diminished and, under a critical growth rate (µ_c), was stopped in chemostat cultures of E. coli and Aerobacter aerogenes (13, 36, 42). The specific oxygen uptake rates increased linearly with increased growth rate under µ_c in both strains (36, 42). At growth rates higher than µ_c, specific oxygen uptake rates did not increase further and acetate production started in E. coli (36). Similar findings for batch cultures with various carbon sources that supported different growth rates were reported (3). Based on the effect of the relationship between acetate production and specific oxygen uptake rate upon growth rates and the fact that acetate biosynthesis is accompanied by equimolar substrate-level ATP generation, acetate excretion is regarded as an overflow metabolism, which provides additional energy when the respiration capacity is saturated (3).

In the context of metabolic flux analysis, aerobic acetate production has been understood as an effect caused by the imbalance between the uptake of the substrate and the demand for biosynthesis and energy production (18). From stoichiometric flux analysis of the central metabolic pathways in E. coli, Holms proposed that the ratio of flux to energy supply does not limit growth and that consequently acetate excretion functions as a safety valve to balance excessive inputs to the fluxes of precursor metabolites for biomass synthesis (19). The fact that E. coli grown on glycerol or fructose does not excrete acetate was explained to be an effect of the restricted uptake of substrates. The maximal uptake rate of these substrates was not supposed to reach the threshold required to trigger acetate excretion (19). Though the analysis adequately demonstrated the existence of an imbalance between substrate uptake and the demands for anabolism, the exact cause of such an imbalance has not been explained.

A possible role of acetate metabolism is for the production of acetyl phosphate, which was recently found to be a phosphate donor for many signal transduction response regulators including CheY, PhoB, NR1, and OmpR (14, 27, 28, 33). It has been shown that the intracellular level of acetyl phosphate varies over a wide range, depending on the presence of a pta or ackA mutation, the phase of growth, and temperature (27, 35). Moreover, the pta mutant has been reported to be impaired in its ability to survive during glucose starvation, while the ability of the ackA mutant to survive remained the same as that of the parent strain (30). Therefore, providing an appropriate level of acetyl phosphate may be an important function of the acetate pathway in E. coli.

In this work, the growth phenotypes of a pta mutant were investigated to understand the physiological roles of acetate metabolism. Several unexpected growth characteristics, such as a large decrease of the growth rate and the excretion of pyruvate and D-lactate, were found during previous investigations of pta mutants (12, 13, 21). Because the altered acetate metabolism must cause severe metabolic flux reorientation, which should result in the accumulation of by-products, the profiles of by-product accumulation, the consumption pattern of amino acids in complex medium, and the growth rates in cultures with a variety of carbon sources were examined in greater detail. Metabolic flux analysis, based on the calculated amounts of precursor metabolites required for the synthesis of biomass, revealed that pyruvate is always excessively generated because of the characteristics of the phosphoenolpyruvate (PEP)-dependent phosphotransferase system (PTS). Therefore, it is proposed that acetate production in E. coli is an unavoidable consequence of substrate transportation via PTS and that the growth defect of the pta mutant is due to the unbalanced flux of acetyl CoA. Because the introduction of phbCAB genes redirecting acetyl CoA flux to poly--hydroxybutyrate (PHB) was found to improve the growth of the pta mutant, any method of alleviating the perturbed acetyl CoA flux would be expected to restore the normal growth of the pta mutant.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. The bacterial strains and plasmids used in this study are listed in Table 1. To construct pta mutants, a pta-1::Tn10-lacZ fusion was introduced into W3110 or RR1 by P1 transduction with lysates of CP993 (39). Plasmid pSK2665 harboring phbCAB genes from Alcaligenes eutrophus was provided by A. Steinbüchel (38). Plasmid pKTPHB was derived from pKT230 with the 5.2-kb BamHI-EcoRI fragment containing phbCAB genes from pSK2665.

Culture media and growth conditions. Physiological characterizations were carried out by using M9 minimal medium (containing 6 g of Na₂HPO₄, 3 g of KH₂PO₄, 1 g of NH₄Cl, and 0.5 g of NaCl liter¹ with 2 g of each carbon source liter¹, tryptone broth (TB; containing 10 g of tryptone liter¹ and M9 minimal salts), or Luria-Bertani broth (LB; 5 g of yeast extract, 10 g of tryptone, and 10 g of NaCl liter¹). For the comparison of growth rates, amino acid consumption, and starvation survival, strains were cultivated at 37°C in 500-ml baffled flasks shaken at 200 rpm. Batch cultivations were carried out at 37°C in a 2.5-liter fermentor (Korea Fermentor Co., Inchon, Korea) which contained 1 liter of M9 minimal medium with 2 g of glucose liter¹. The air flow rate was fixed at 1 liter min¹, and dissolved oxygen was maintained above 20% of its saturation level by manually adjusting the agitation speeds in the range of 500 to 900 rpm. pH was kept constant at 7.0 by the addition of 2 M NaOH. Tetracycline (13 µg ml¹) or kanamycin (50 µg ml¹) was added as required. All the cultivations were repeated at least twice.

Analytical methods. Optical density (OD) was measured at 600 nm (Ultrospec 2000 UV/visible spectrophotometer; Pharmacia Biotech Ltd, Uppsala, Sweden), and dry cell weight was determined gravimetrically after the culture broth was centrifuged at 6,000 × g, washed with distilled water, and dried overnight at 105°C. One OD unit was found to be equivalent to 0.39 ± 0.05 g (dry cell weight) · liter¹. The glucose concentration in the medium was measured with a glucose analyzer (model 2300; YSI Co., Yellow Springs, Ohio). The concentrations of acetate, D-lactate, and pyruvate in the sample were assayed with enzymatic test kits (Boehringer, Mannheim, Germany). Concentrations of amino acids in the samples were determined with an amino acid analyzer (AminoQuant 1090; Hewlett-Packard, Avondale, Penn.). The synthesis of by-products was monitored by performing in vivo nuclear magnetic resonance (NMR) scans of whole cultures as described by Alam et al. (1, 26). Proton NMR spectra of culture broth samples were obtained with a Varian UNITY spectrometer operating at 500 MHz (Korea Basic Science Institute, Taejon, Korea). The water peak was suppressed, the field was locked onto the solvent D₂O, and the H₂O peak at 4.65 ppm was used as the internal reference. Dimethyl sulfone (100 mM) was used as an internal standard (3.12 ppm) for the quantification of the fermentation products. PHB levels were determined by gas chromatography (Hewlett-Packard) with benzoic acid as an internal standard (5).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The E. coli W3110 pta mutant accumulates pyruvate, D-lactate, and L-glutamate instead of acetate. The growth phenotypes of an E. coli W3100 pta mutant were compared with those of wild-type strain W3100 to study the physiological changes due to the pta mutation. In glucose minimal medium, the pta mutant JP231 grew more slowly than its parental strain, W3110. The maximum specific growth rates were 0.49 and 0.60 h¹ for JP231 and W3110, respectively (Fig. 1). While the biomass yields of both strains on glucose were the same, 0.077 g (dry weight) · mM¹, the specific glucose consumption rate of the pta mutant, 6.06 mM · (g [dry weight] · h)¹, differed from that of wild type, 7.87 mM · (g [dry weight] · h)¹. Introduction of a pta mutation did not completely eliminate acetate production during the growth phase. The acetate yield of the pta mutant on glucose, 0.11 mM · (g [dry weight])¹, was reduced to one-fourth that of the wild type, 0.45 mM · (g [dry weight])¹. Pyruvate oxidase (PoxB) presents a possible route for the generation of acetate in the pta mutant (8), because acetyl phosphate may be converted to acetate and inorganic phosphate through the action of either AckA or nonenzymatic degradation (6). However, a further mutation in the poxB gene did not result in the complete elimination of acetate (40). As previously reported (12, 21), pyruvate and D-lactate were excreted during the exponential growth of JP231, while in its parental strain (W3110) the excretion of pyruvate and D-lactate was quantitatively insignificant. Once glucose was exhausted, the pta mutant reused the excreted acetate mainly through ACS (24), pyruvate via pyruvate dehydrogenase (16) or pyruvate oxidase (8), and D-lactate via the NAD-independent, membrane-bound D-lactate dehydrogenase (D-LDH) (23).

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Growth and by-product excretion profiles of W3110 (a) and JP231 (b). Strains were grown in an aerated fermentor containing M9 minimal medium supplemented with 2 g of glucose liter1. Symbols: , dry cell weight; , glucose; , acetate; , pyruvate; , D-lactate.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Glutamate accumulation in a pta mutant grown on M9 glucose minimal medium. Profiles of growth (a) and glutamate accumulation (b) of W3110 (open symbols) and JP231 (solid symbols) are represented.

The pta mutant shows a different consumption profile of amino acids. The growth defect of the pta mutant on glucose and the accumulation of glutamate led us to check whether the consumption profiles of amino acids by the JP231 mutant, on TB containing M9 salts, differed from those of W3110. It is worth pointing out that the growth defect was more pronounced on TB containing M9 salts than on glucose minimal medium. On TB, the growth rate of JP231 was only 45% that of the wild-type strain, whereas it reached 78% of the control on glucose. Since amino acids serve as the primary carbon source, the levels of sequential consumption of amino acids by both W3110 and JP231 growing on TB containing M9 salts were analyzed (Fig. 3). The consumption profiles of amino acids in the wild-type strain were consistent with those from a previous report (34). During the early exponential growth phase, L-serine was consumed exclusively; this was followed by the consumption of L-aspartate (Fig. 3a). In this phase, acetate accumulated and was later reused when consumption moved to L-glutamate, L-threonine, L-alanine, and glycine. In the stationary phase, L-arginine was the only amino acid to be consumed. Glycine was accumulated when the cells were consuming L-serine, L-threonine, and L-alanine, indicating an imbalance between the assimilation of such amino acids and their related catabolic pathways.

View larger version (26K): [in this window] [in a new window]   FIG. 3.   Profiles of amino acid consumption patterns in W3110 (a) and JP231 (b) grown on TB. Symbols: , OD at 600 nm; , L-serine; , L-aspartate; , L-glutamate; , L-alanine; , glycine; , L-threonine; , L-arginine.

The pta mutant grows slowly on glucose, TB, or pyruvate but normally on glycerol or succinate. Maximum growth rates of JP231 on various carbon sources were compared with those of W3110, because the pta mutant had shown different growth defects in glucose and TB. JP231 grew more slowly than W3110 not only in glucose and TB containing M9 salts but also on fructose and pyruvate (Table 2) (17). The specific growth rates of the pta mutant on fructose and pyruvate were 75 and 45%, respectively, of those of the wild-type strain. However, on glycerol or succinate, the growth rates of the pta mutant were comparable to those of the wild type. Metabolism of substrates on which the pta mutant showed the growth defect is directly linked to pyruvate either by PTS (glucose and fructose), catabolism of amino acids (TB), or by pyruvate itself. Because PTS using PEP as the phosphate donor links the uptake of glucose or fructose with the stoichiometric coproduction of pyruvate, oversupplied pyruvate should be excreted as acetate. Therefore, the fluxes of pyruvate and consequently acetyl CoA in the pta mutant grown on PTS carbon sources would be expected to be severely perturbed, as is manifested by the excretion of pyruvate, D-lactate, and L-glutamate. In addition to the excretion of unusual by-products, the perturbation of acetyl CoA and possibly pyruvate fluxes in the pta mutant, in some way, causes the decrease in the growth rate.

Defective growth of the pta mutant is recovered by phbCAB introduction. In an attempt to check whether acetyl CoA flux is responsible for the decreased growth rate of the pta mutant, PHB synthesis genes (phbCAB) from A. eutrophus (38) were introduced into W3110 and JP231. Initially, the effects of introducing PHB synthesis pathway genes upon the distribution of acetyl CoA flux were examined by analyzing the by-products. In the wild-type strain, the introduction of phbCAB genes had no effect on the by-product pattern, in that acetate was excreted as the only product in M9 medium containing glucose (Fig. 4a). In contrast, the introduction of phbCAB genes repressed the excretion of the by-products, the characteristic of the pta mutant: excretion of pyruvate and D-lactate was diminished (Fig. 1b and 4b), and the maximum level of glutamate was lowered to one-sixth that associated with the pta mutant (data not shown). Moreover, the maximum level of acetate was also reduced to three-fourths that of the control mutant strain. The PHB content in the pta mutant harboring phbCAB was approximately twice those of the respective control strains. On LB supplemented with 20 g of glucose liter¹, the PHB content of JP231 harboring pKT230 reached 24.5% of dry cell weight, while that of W3110 harboring the same plasmid was only 14.7%. This result indicates that the perturbed acetyl CoA flux was indeed redirected to PHB synthesis.

View larger version (17K): [in this window] [in a new window]   FIG. 4.   Growth and byproduct excretion profiles of recombinant strain W3110 (a) and JP231 (b) harboring phbCAB genes from A. eutrophus. Strains were grown in an aerated fermentor containing M9 minimal medium supplemented with 2 g of glucose per liter. Symbols: , dry cell weight; , glucose; , acetate; , pyruvate; , D-lactate.

1

Starvation survival defect of the pta mutant is also recovered by the introduction of phbCAB. One of the most prominent phenotypes of the pta mutant is its poor starvation survival, which was proposed to be a result of its inability to provide acetyl phosphate (30). Encouraged by the finding that the introduction of phbCAB genes repressed the defective growth of the pta mutant, we examined the effect of the introduction of phbCAB genes on the starvation survival of the pta mutant. W3110, JGP231, and their recombinant strains harboring pKTPHB were cultivated in M9 minimal medium with 0.2 g of glucose liter¹. After the depletion of glucose, incubation was continued for 9 days and viable cells were counted as colonies plated on LB plates after appropriate dilution (30). As shown in Fig. 5, the viability of JP231 harboring phbCAB genes was indistinguishable from that of the wild-type strain while the pta mutant showed poor survival. This result indicates that the defective starvation survival of the pta mutant could also be recovered by redirecting acetyl CoA flux to PHB synthesis, demonstrating that managing the acetyl CoA flux rather than providing acetyl phosphate is important for the survival of E. coli under starvation conditions.

View larger version (18K): [in this window] [in a new window]   FIG. 5.   Starvation survival of W3110, JP231, and their recombinant strains harboring phbCAB genes from A. eutrophus. Strains were grown in shaking flasks containing M9 minimal medium supplemented with 0.2 g of glucose liter1. After growth was arrested, incubation was continued for 9 days under the same condition. Viable cells were counted as colonies in LB plates after appropriate dilutions. One hundred percent viability corresponds to the number of viable cells counted 1 h after entering starvation. Symbols: , W3110; , W3110(pKTPHB); , JP231; , JP231(pKTPHB).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we showed that the pta mutation resulted in a growth defect and the excretion of unusual by-products. In order to illustrate further the metabolic basis of phenotypic changes in the pta mutant, the stoichiometric method of flux analysis by Holms (19) was adopted with a slight modification in that the TCA cycle does not function in the cyclic mode. Components of the TCA cycle function almost exclusively to provide three precursor metabolites, namely, oxaloacetate (OAA), -ketoglutarate, and succinyl CoA (29). The TCA cycle does not function as an energy-generating cycle. Moreover, in a glucose medium, the expressions of the enzymes in the TCA cycle are repressed, and, especially, -ketoglutarate dehydrogenase activity was absent (2, 31, 41). In Fig. 6a, which shows a flux distribution of the precursor metabolites required for the synthesis of 1 g of biomass, the flux to PEP is insufficient and the insufficiency is resolved hypothetically by postulating a reverse flux through PEP synthetase, indicating the oversupply of pyruvate. A more likely way of supplying sufficient PEP flux is the uptake of additional glucose and its conversion to acetate, which is illustrated in Fig. 6b. In this case, it was assumed that 15% of the input glucose was excreted as acetate, reflecting the typical acetate yield of E. coli cultured on glucose. Now no hypothetical reverse flux from pyruvate to PEP is required. However, glucose transportation via PTS and the action of pyruvate kinase generate pyruvate (12.864 mmol) in excess of the required amount for precursor metabolites downstream of pyruvate, namely, acetyl CoA, -ketoglutarate, and pyruvate (7.659 mmol). Oversupplied pyruvate and consequently acetyl CoA cannot be oxidized completely to CO₂ by the TCA cycle due to the repression of the TCA cycle enzymes, particularly of -ketoglutarate dehydrogenase (2, 31, 41). Therefore, the pyruvate flux should be balanced by opting for acetate excretion (12). In the pta mutant, acetyl CoA flux is perturbed in that no apparent pathway(s) is available for treating the acetate CoA flux normally secreted by the wild-type strain.

View larger version (25K): [in this window] [in a new window]   FIG. 6.   Stoichiometric calculation of metabolic flux distribution in E. coli. The flux distribution was analyzed by Holms' method (16) with a slight modification in that the TCA cycle was assumed not to work in a cyclic mode but to function only for the generation of OAA, and -ketoglutarate (2, 27, 36). Numbers in the middle of arrows indicate the fluxes through the specific metabolic step in the direction of the arrow. Italic numbers under precursor metabolites indicate the fluxes incorporated into biomass. A negative value means that the flux is in the direction opposite to the arrow. (a) Flux distribution in an ideal E. coli cell in which the central metabolism is working only for generating the exact amount of the required precursor metabolites and NADPH to make 1 g of biomass (25). The flux through a specific metabolic step was calculated by adding the amounts of precursor metabolites downstream of the metabolic step to make 1 g of biomass. In the resulting flux distribution, the amount of NADPH produced (4.319 mmol) was much smaller than that required to make 1 g of biomass (18.225 mmol). The shortage was covered by the uptake of additional (1.159 mmol) glucose and metabolizing it through the pentose phosphate pathway. (b) Flux distribution in wild-type E. coli in which the central metabolism works to supply the required amounts of precursor metabolites, NADPH, and ATP for 1 g of biomass and the production of acetate, which corresponds to 15% of input carbon. Numbers in parentheses indicate the fluxes for the pta mutant. The fluxes were calculated by adding the fluxes presented in panel a and the additional flux needed to produce acetate (wild type) or acetate, pyruvate, D-lactate, and glutamate (pta mutant). Abbreviations: AcCoA, acetyl CoA; -KG, -ketoglutarate; CIT, citrate; ERP, erythrose 4-phosphate; F6P, fructose 6-phosphate; G6P, glucose 6-phosphate; IsoCIT, isocitrate; 3PG, 3-phosphoglycerate; PP, pentose phosphate; PYR, pyruvate; TP, triose phosphate; SucCoA, succinyl CoA.

In response to this metabolic flux perturbation, the pta mutant was found to excrete pyruvate, D-lactate, and L-glutamate. The flux values in parentheses in Fig. 6b illustrate the theoretical flux distribution in the pta mutant, showing that the reduced acetate flux was redirected to the excretion of pyruvate, D-lactate, and L-glutamate. Excretion of pyruvate is a well-known characteristic of a mutant with a defective Pta-AckA pathway (12, 21). The accumulated acetyl CoA inhibits the activity of the pyruvate dehydrogenase complex, which causes the accumulation of pyruvate. Excretion of D-lactate may be the result of an elevated expression of D-LDH as well as pyruvate accumulation. In this regard, it is notable that the expression of ldhA, encoding fermentative D-LDH, increased and became independent of the medium pH by the mutations in the pta gene (7). Another possible response of a pta mutant to solve the problem involves the redirection of the acetyl CoA flux into the TCA cycle. Increased acetyl CoA flux into the TCA cycle requires an equivalent increase of PEP flux to OAA. The increased carbon flux must be excreted as by-products such as intermediates of the TCA cycle or amino acids if the TCA cycle does not function in a cyclic mode. As expected, the pta mutant excreted glutamate in an aerobic cultivation on glucose minimal medium or TB. Atkinson and Ninfa's recent finding that higher activity of glutamine synthetase increased the flux from -ketoglutarate to glutamate and resulted in a decreased level of acetyl phosphate (4) is consistent with glutamate accumulation in the pta mutant. The excretion of glutamate confirms that -ketoglutarate dehydrogenase is repressed during aerobic growth on glucose or amino acids (2, 31, 41). The flux to pyruvate, D-lactate, and glutamate in the pta mutant was equivalent to the reduction in acetate flux compared to that for the wild-type strain (Fig. 1b and 6b).

Why does the pta mutant grow slower than the wild-type strain if flux balance is accomplished by excreting pyruvate, D-lactate, and glutamate instead of acetate? Since acetate metabolism has been proposed to function as a secondary energy-yielding pathway, when the TCA cycle or respiration capacity or both are saturated (3), reduced energy generation due to the pta mutation may be responsible for the growth defect. However, the amount of ATP generated by the pta mutant was calculated as 73.425 mM · g (dry weight)¹, which reached 90.5% of the total energy generated by the wild type (81.125 mM · g [dry weight]¹). In this calculation, it was assumed that all NADH generated was converted to ATP and that the P/O ratio was 1.5. Moreover, at higher growth rates, sufficient energy (ATP) is available from the Embden-Meyerhof-Parnas (EMP) pathway to minimize the energy-producing role of the TCA cycle (22). Therefore, a reduction in ATP generation cannot explain the large growth defect of the pta mutant, although the possibility of its partial contribution cannot be excluded. It is more plausible that the retarded growth on several carbon sources is due to the perturbation in the fluxes of acetyl CoA and pyruvate, which in turn would affect the substrate transport rate. In the pta mutant grown on glucose, the accumulation of pyruvate will lower the PEP/pyruvate ratio, which results in a lowered glucose uptake rate (25, 32). Moreover, accumulated acetyl CoA activates PEP carboxylase (Ppc), a PEP-consuming enzyme, which will decrease available PEP for glucose transport (20). Glutamate excretion in the pta mutant reflects the increased flux through Ppc. The slow uptake of substrate was also observed in the cultivation of the pta mutant on TB, which is demonstrated by the lags between the assimilation of amino acids and the disappearance of glycine accumulation. The retardation of the amino acid assimilation of the pta mutant is comparable to that of nuo mutants; a nuo mutant, defective in membrane-bound NADH dehydrogenase I, consumed L-glutamate slowly and glycine, L-threonine, and L-alanine poorly at the entry into stationary phase (34). The slow or poor consumption of the amino acids was proposed to be a result of a higher NADH/NAD⁺ ratio, inhibiting both TCA cycle enzymes and enzymes functioning in the assimilation of amino acids (34). The pta mutation would decrease substrate uptake, and this in turn would lower the rate of growth.

If the growth defect of the pta mutant is caused by an unbalanced flux of acetyl CoA, the introduction of any pathway that can convert acetyl CoA to cell components or other metabolic end products should restore the growth of the pta mutant. As expected, when phbCAB, the PHB synthesis genes from A. eutrophus, were introduced, the growth defect of the pta mutant was restored (Table 3 and Fig. 5). The PHB content in the pta mutant was two times higher than that of the wild type, which demonstrated a redirection of the acetyl CoA flux normally destined for acetate synthesis. The difference between the PHB synthetic fluxes from acetyl CoA of the pta mutant (6.23 mmol) and the wild-type strain (3.31 mmol) was 2.92 mmol or 86% of the reduction of acetate flux (3.38 mmol), which shows that the increased flux directed to PHB synthesis actually contributes to the management of acetyl CoA flux. Diaz-Ricci et al. reported similarly that the expression of pyruvate decarboxylase and alcohol dehydrogenase genes from Zymomonas mobilis overcame the metabolic stress caused by the plasmid, enhancing growth and glucose uptake rates of a pta mutant to the observed values for the plasmid-free pta mutant (12). Particularly notable is that the starvation survival of the pta mutant was recovered by phbCAB introduction, which suggested that the poor survival of the pta mutant was due to problems of acetyl CoA flux rather than its inability to synthesize acetyl phosphate.

Our conclusion is that aerobic acetate production in E. coli is due to the oversupply of pyruvate from PTS, which strongly suggests that the production of acetate is a function of the balance of metabolic flux and not of the central metabolic rate or the growth rate itself. If the production of acetate is a matter of balance, why is acetate accumulation eliminated at lower growth rates? Two explanations are possible. First, the elimination of acetate excretion at lower growth rates may be due to the change in the glucose uptake system. Transport of substrate through the non-PTS system does not require a massive conversion of PEP to pyruvate, which may eliminate the pyruvate oversupply and in turn the excretion of acetate. In this context, it is worth noting that the non-PTS glucose uptake systems such as the maltose or galactose transport systems were found to be derepressed at lower growth rates in continuous culture (10, 11). Recently, Chen et al. showed that when E. coli exclusively transports glucose by the galactose transport system, a non-PTS Na symport system, the aerobic production of acetate was totally eliminated (9), which supports our contention that aerobic acetate production depends on the nature of the glucose uptake system. Second, the elimination of acetate excretion at lower growth rates may be the effect of derepression of TCA cycle enzymes, which permits a further metabolism of acetyl CoA. As TCA cycle enzymes are derepressed at lower growth rates in continuous culture, oversupplied pyruvate and acetyl CoA may be metabolized to CO₂ via the TCA cycle and acetate excretion may be eliminated. Further studies to examine the effects of changes of the glucose transport system or derepression of TCA cycle enzymes on acetate metabolism are in progress.

Another intriguing question arises here: why is the PTS chosen as the principal substrate uptake system if acetate excretion is due to PTS? Does PTS offer a competitive advantage in terms of outgrowth competition? Because the Mgl system coupled with LamB glycoporine appears to support slower growth (15), it is tempting to suggest that the PTS system is used to achieve maximum growth rate at the expense of the growth yield due to acetate excretion. It is worth noting that a mutant with a higher glucose transport capacity evolved as one of the major populations during a long-term culture of E. coli in glucose-limited continuous culture (37). Furthermore, excreted acetate may be reused during the stationary growth phase when the principal substrates are exhausted (24), allowing cells to adapt to and survive starvation conditions.