Control of glucose metabolism by enzyme IIGlc of the phosphoenolpyruvate-dependent phosphotransferase system in Escherichia coli

The quantitative effects of variations in the amount of enzyme IIGlc of the phosphoenolpyruvate:glucose phosphotransferase system (PTS) on glucose metabolism in Escherichia coli were studied. The level of enzyme IIGlc could be adjusted in vivo to between 20 and 600% of the wild-type chromosomal level by using the expression vector pTSG11. On this plasmid, expression of the structural gene for enzyme IIGlc, ptsG, is controlled by the tac promoter. As expected, the control coefficient (i.e., the relative increase in pathway flux, divided by the relative increase in amount of enzyme) of enzyme IIGlc decreased in magnitude if a more extensive pathway was considered. Thus, at the wild-type level of enzyme IIGlc activity, the control coefficient of this enzyme on the growth rate on glucose and on the rate of glucose oxidation was low, while the control coefficient on uptake and phosphorylation of methyl alpha-glucopyranoside (an enzyme IIGlc-specific, nonmetabolizable glucose analog) was relatively high (0.55 to 0.65). The implications of our findings for PTS-mediated regulation, i.e., inhibition of growth on non-PTS compounds by glucose, are discussed.

The bacterial phosphoenolpyruvate:carbohydrate phosphotransferase system (PTS) is responsible for the translocation of carbohydrates across the cell membrane and their concomitant phosphorylation (22,30). Transfer of the phosphoryl group from phosphoenolpyruvate (PEP) to the carbohydrates is catalyzed by two general, cytoplasmic PTS proteins, enzyme I and HPr, and one of a number of carbohydrate-specific permeases, which are localized in the cytoplasmic membrane. The permeases generally consist of one (enzyme II) or two (enzymes II and III) subunits.
Escherichia coli contains two glucose-specific PTS permeases (4,7,38). One of these permeases has a broad substrate specificity (mannose, glucose, fructose, and glucosamine) and is designated mannose permease, because it is required for growth on mannose as the sole carbon source (Fig. 1). The second system, glucose permease, is specific for glucose and the nonmetabolizable glucose analog methyl a-glucopyranoside (cxMG). It consists of two subunits, the cytoplasmic phosphoryl-carrying protein 111GIc and the membrane-bound enzyme 11"GC, which catalyzes the actual glucose transport and phosphorylation (Fig. 1).
In addition to sugar uptake and phosphorylation, the PTS is involved in metabolic regulation (for reviews, see refer- ences 22, 30, and 33). Both the rate of synthesis of 3',5'cyclic AMP by adenylate cyclase and the transport of several compounds which are not substrates of the PTS are influenced by the PTS. Presently, a model is favored in which the phosphorylation state of IIIGlC is of prime importance for the regulatory functions of the PTS (Fig. 1). In the absence of a PTS substrate, the PTS proteins are thought to be phosphorylated, while during transport of a carbohydrate, such as glucose, the PTS proteins (including IIIG1c) could, at least partly, be dephosphorylated. Dephosphorylation of IIIGIc, by the addition of aMG to intact Salmonella typhimurium cells, has been reported by Nelson et al. (24). * Corresponding author.
Glucose transport catalyzed by the PTS and PTS-mediated regulation are mostly described in a qualitative way; the quantitative aspects of the PTS functions have been much less investigated. For instance, for a proper understanding of PTS-mediated regulation, it is important to know which enzyme (or enzymes) controls the fluxes of phosphoryl groups through the PTS pathways.
The control of an enzyme on the flux through a metabolic pathway can be determined by measuring the pathway flux as a function of varying amounts of the enzyme of interest. A useful model for a quantitative description of metabolic pathways has been put forward by Kacser and Burns (19) and Heinrich and Rapoport (13). They have developed a theoretical analysis of pathway fluxes and defined a control coefficient that expresses the degree to which individual steps influence the overall rate through a pathway.
In this report, we describe the effect of variations in the amount of IIGIc on the whole or on parts of the cellular metabolism of glucose in E. coli. We have been able to modulate the level of IIG'C to between 20 and 600% of the wild-type level by using an adjustable expression vector. The application of an expression vector in control analysis has been described previously by Walsh and Koshland (39). From our results, we conclude that at the wild-type level of IIGIc activity, a variation in the amount of this enzyme has no effect on the growth rate in glucose-containing medium or on oxidation of glucose. However, under the same circumstances, i.e., near wild-type activity, the control coefficient of IIGIc on uptake and phosphorylation of aMG is relatively high (0.55 to 0.65). Some consequences of these findings with respect to the model of the regulatory functions of the PTS are discussed.

MATERIALS AND METHODS
Bacterial strains and plasmids. The strains used are listed in Table 1 land). Plasmid pTSG11 (Fig. 2) contains the E. coli structural gene for 1IGIc, ptsG, under the control of the tac promoter (8).
Media and growth conditions. For all experiments, cells were grown overnight in 25 ml of liquid medium A [1 g of (NH4)2SO4, 10.5 g of K2HPO4, 4.5 g of KH2PO4, and 1 g of MgSO4 per liter of demineralized water] with 0.5% glycerol, harvested by centrifugation, resuspended in 250 ml of medium A with 0.2% glucose, and grown to exponential phase (optical density at 540 nm of 0.5 to 1.0). All cultures were grown at 37°C on a rqtary shaker and supplemented with thiamine (20 ,g/ml), ampicillin (50 ,ug/ml), and the indicated amount of isopropyl-,-D-thiogalactopyranoside (IPTG).
Chemicals. [U-14C]coMG (10.8 GBq/mmol) was obtained  Holmes and Quigley (16), and transformation of plasmid DNA was performed as described by Sambrook et al. (34). Preparation of P1 transducing lysates and transduction with bacteriophage P1 were performed as described by Arber (1). Excision of transposon TnJO from the chromosome was performed according to Bochner et al. (2).
Oxygen consumption and transport studies. Oxygen consumption was measured with a Clark-type electrode in medium A and expressed as nanoatoms of 0 consumed per minute per milligram (dry weight [DW]) at 25°C. Transport of labelled compounds was performed as described previously (28).
Preparation of cell extracts and enzyme assays. Cells were ruptured by passage through an Aminco French pressure cell at 1,100 kg/cm2, and cell extracts were prepared as described earlier (28). The cell extracts were centrifuged at 230,000 x g for 2 h at 4°C. The activity of II"GC, measured as PEPdependent phosphorylation of aMG, was determined in the 230,000 x g pellet essentially as described by Kundig and Roseman (20). As a source of the soluble components of the glucose PTS, a high-speed supernatant of strain PPA211 was used. The activities of enzyme I and HPr in strain PPA211/ pTSG11 were determined in the high-speed supernatant of cell extracts according to Waygood et al. (40). In this assay, phosphorylation of aMG is catalyzed by the mannose permease. For determination of enzyme I activity, a cell extract of S. typhimurium SB1690 was added as a source of the other components of the mannose PTS (HPr, IllMan, and 1IMan). HPr activity in PPA211/pTSG11 was determined by addition of a cell extract of S. typhimurium ptsH strain SB2226 (containing enzyme I, ,Man and 11Man) Immunochemical methods. Determination of the amount of IIGIc was performed with rocket immunoelectrophoresis as described by Scholte et al. (36). To determine the amount of VOL. 173, 1991 II" in membrane fractions of cell extracts, an enzymelinked immunosorbent assay (ELISA) was developed. To the first well of a microtiter plate (Maxisorp-F96; Nunc, Roskilde, Denmark), 50 p1 of a membrane preparation (250 jig of protein per ml) was applied and serially diluted twofold 11 times in water. The plates were dried for 2 h at 65°C and rinsed five times with PBST (140 mM NaCl, 3 mM KCI, 8 mM Na2HPO4, 1.5 mM KH2PO4, 0.1% [wt/vol] Tween 20, pH 7.5). Subsequently, 100 RI of an appropriate dilution of IIGlc-specific monoclonal antibody 5A5 (gift from B. Erni) in PBST was added to each well and incubated for 1 h at 42°C. Following five wash steps with PBST, the wells were incubated with 100 RI of 1:1,000 (in PBST) goat anti-mouse immunoglobulin G conjugated to horseradish peroxidase (Bio-Rad Laboratories, Richmond, Calif.) for 1 h at 42°C. The plates were rinsed 10 times with PBST and incubated for 15 to 30 min at room temperature with color reagent (100 pl per well) containing 2 mg of o-phenylenediamine per ml, 100 mM sodium phosphate (pH 6.0), and 0.015% H202. The reaction was subsequently stopped with 50 ,ul of 2 N H2SO4 per well, and absorption values were measured with a Titertec Multiscan Plus (EFLABoy, Helsinki, Finland) at 492 nm. The sigmoid-shaped dilution-absorption curve from each sample was compared with that of a wild-type sample; when the two curves were parallel, the ratio of dilutions at one absorption value gave the ratio of the amounts of IIcGi. The amount of lGlc is given relative to the amount present in a wild-type strain.
Protein and bacterial DW. Protein was determined by the method of Peterson (27), using bovine serum albumin as a standard. Bacterial DW was determined as described by Herbert et al. (15).
Phosphorylation via the PTS in toluene-treated cells. Phosphorylation was measured essentially as described previously (35), with the following modifications. After harvesting and washing, cells were resuspended in 1/50 growth volume of buffer B (5 mM MgCl2, 10 mM KF, 0.5 mM dithiothreitol, 3 mM KCN, 50 mM potassium phosphate, pH 7.5), resulting in approximately 5 mg of bacteria (DW) per ml. For each culture, the amount of toluene needed to obtain maximal glucose PTS activity was determined; generally, 2 to 2.5 pI of toluene per ml of cell suspension gave the best results. Phosphorylation was measured with a SLM Aminco DW2000 double-beam spectrophotometer (X1 = 350 nm; X2 = 375 nm).

RESULTS
Inducible expression of 11G1C. E. coli PPA211 and PPA234 and plasmid pTSG11 (Fig. 2) were used to quantify the control of the IIlc of the PTS on the bacterial glucose metabolism. On plasmid pTSG11, expression of the E. coli structural gene for ll0ic, ptsG, is dependent on the activity of the tac promoter. In the presence of the lacIq repressor (also encoded by pTSG11), expression from the tac promoter can be adjusted by varying the amount of inducer, IPTG.
Strain PPA211 does not contain any chromosomally encoded 11G1c, IlMan, or glucokinase. As a consequence, this strain is unable to take up or metabolize glucose. By transforming PPA211 with plasmid pTSG11, we were able to vary in vivo the amount of 11G1c, which then represents the sole glucose uptake system in this strain. To compare the amount of 11Glc synthesized from the plasmid with the amount of this enzyme in a wild-type strain, we constructed a ptsG+ derivative of strain PPA211, designated PPA234.
Strain PPA211 containing plasmid pTSG11 was grown on glucose in the presence of different concentrations of IPTG.
In the membrane fraction of these cells, the 11GIc activity was determined and appeared to increase linearly with the IPTG concentration (Fig. 3A). The maximum expression of 11Gic activity was reached at about 50 puM IPTG and corresponded to six to seven times the wild-type, chromosomal level (the values of the various experiments as determined for ptsG+ strain PPA234 are shown in Table 2). When the IPTG concentration was increased above 75 ,uM, the 116Ic activity decreased (Fig. 3A). In the absence of IPTG, the strain exhibited IIG'c activities that were about 20% of the wildtype level. This resulted possibly from read-through from other promoters present on the plasmid or from a low-level expression from the tac promoter despite the presence of the lacIq repressor, because strain PPA211 without the plasmid showed no 1I1IC activity (<1.0 nmol/mg/min). On plasmid pTSG11, the natural promoter of ptsG was exchanged for the synthetic tac promoter. Because normally glucose induces the expression of ptsG (31,38), we have investigated the effect of glucose on the expression of ptsG from pTSG11. When cells were grown on lactate, the relationship between IPTG concentration and 1I1lc activity was similar to that of glucose-grown cells. For example, ll0ic activity in membranes isolated from strain PPA211/pTSG11 grown on minimal medium A containing 0.5% lactate in the presence of 20 puM IPTG was 48 nmol/mg/min; in the case of glucose-grown cells, the activity was 46 nmol/mg/min (Fig.  3A). Therefore, it was concluded that the presence of glucose during growth did not influence the expression of ptsG from pTSG11.
It would have been possible that because of synthesis of large amounts of 11Gic, the enzyme was not or not properly inserted in the cytoplasmic membrane and was therefore inactive. For this reason, we determined the amount of 1101c immunologically. In an ELISA, the amount of I1Gic in cytoplasmic membranes was determined by comparing it with the amount of enzyme present in a ptsG+ strain. The relationship between the IPTG concentration present during growth and the amount of IGI0c in membrane preparations ( Fig. 3B) appeared to be comparable to the relationship between IPTG concentration and IIGic activity. Thus, at least up to 50 pLM IPTG, the ratio of the amount of 11Gic and its activity was constant, and it was therefore concluded that the lIGIc present in the cytoplasmic membranes was in its active form.
Since the expression of j1Gic from pTSG11 was, at low IPTG concentrations, in the physiological range, this plasmid could be used to determine the control by I1GIc of glucose metabolism in E. coli.
Control coefficient of IIG1C on growth rate and on oxidation of glucose. As mentioned previously, strain PPA211 containing pTSG11 is able to take up glucose only via IIGic expressed from the plasmid. When this strain was grown in batch cultures with glucose as the sole carbon and energy source, an increase of the 11"GC activity with respect to the wild-type level did not have an effect on the maximal growth rate (Fig. 4). Thus, under these excess-glucose conditions, IIGic had a low control coefficient on the growth rate in the physiological range of the enzyme's activity.
When the effect of the amount of 0IGIc on the oxidation rate of glucose was determined, similar results were obtained (Fig. 5). Thus, a slight variation of JIG"c activity compared with the wild-type level had no effect on the oxidation rate of glucose. In other words, at the wild-type level, IIG'c had a low control coefficient on the oxidation rate of glucose. In the case of both growth and glucose oxidation, small amounts of ITIGC, i.e., about 25% of the wild-type level, were able to support more than half the maximal growth rate and glucose oxidation rate that could be reached. This result suggests that under conditions in which glucose is amply available, 11"GC is present in great excess over the amount required for maximal growth or glucose oxidation.
Control coefficient of 1IG1c on the flux through the glucose PTS. Growth on and oxidation of glucose involve rather complex metabolic pathways. In contrast, the glucose PTS is a much shorter pathway, comprising only four enzymes (Fig.   1). We have determined the effect of the amount of 11I"C on a Cells of strain PPA234 containing pJF118EH were grown to exponential phase on glucose minimal medium. After harvesting, the cells were prepared for the different experiments as described in Materials and Methods. 11Ic activity in membranes and uptake of aMG by intact cells were determined  IIG'c activity than did the growth rate and the rate of glucose oxidation. The slight difference in the dependence of uptake and phosphorylation on the activity of IIGIC reflected the difference in the process studied. From these graphs, it was possible to calculate control coefficients as defined by Heinrich and Rapoport (13) and Kacser and Burns (19). The control coefficients (relative increase in pathway flux, divided by the relative increase in amount of enzyme) of IIGIc on aMG uptake by intact cells and on aMG phosphorylation by toluene-treated cells were 0.65 and 0.55, respectively.
It is known that the presence of IPG'c is required for complete induction of the general PTS proteins, enzyme I and HPr, during growth (5, 21, 31) on glucose, while the amount of IIGIc in the cell is more or less constant (5,36).
During our experiments, the amount of 1IGIc was varied, and it was therefore possible that the amounts of enzyme I and HPr also varied. In that case, the effects ascribed to IIGIc could, at least partially, be caused by variations in the amounts of enzyme I, HPr, or I0IGIc. Therefore we measured the activity of enzyme I and HPr and the amount of IlIGic in the same cells used for the studies with TIGIC. The HPr activity and the amount of IIIG'C did not vary much with the IIG'c activity but fluctuated around their respective wild-type values (Fig. 7). The enzyme I activity was also more or less constant except at very low TIGIC activities (Fig. 7). Therefore, at wild-type JIG"C activities, only an increase in the activity of 11GIc could account for an increase in flux through the glucose PTS.
Effect of IIGIc activity on inducer exclusion. We   " Cells were grown overnight on glycerol minimal medium and subsequently grown for 1 h on glucose minimal medium as described in Materials and Methods. Cells were then prepared for transport of [U-14C]glycerol (specific activity, 165 cpmtnmol) as described. aMG was added to a concentration of 10 mM 3 min before the uptake experiment was started.
glycerol uptake by the various E. coli strains in the presence and absence of axMG. The phosphorylation state of IIIGIc is of key importance for some regulatory functions of the PTS (33). Dephosphorylated IIIGIc inhibits, for instance, the enzyme glycerol kinase and in this way, indirectly, glycerol uptake by cells. Table 3 shows that addition of 10 mM aMG did not inhibit glycerol uptake in ptsG ptsM strain PPA211, which was expected because this mutant strain did not contain any functional IIGIc and was therefore unable to take up axMG. In contrast, glycerol uptake was inhibited 95% by addition of oxMG to the ptsG+ strain PPA234. Furthermore, when strain PPA211/pTSG11 was grown in the absence of IPTG (the amount of 11GIc then corresponded to about 20% of the wild-type level), glycerol uptake was inhibited more than 80% by addition of oaMG. This indicated that 20 to 30% of the wild-type activity of IIl"c is sufficient for almost complete inhibition of glycerol uptake by addition of aMG.

DISCUSSION
In this report, we describe the control by IIG0c of glucose metabolism in E. coli. Using an inducible expression vector, the effects of varying the amount of 11GIc on the growth rate in a glucose-containing medium, on the oxidation rate of glucose, and on the uptake and phosphorylation of aMG were determined.
Our results show that at the wild-type level of 1jGIc activity, this enzyme had a low control coefficient on the growth rate in batch cultures containing glucose. According to Jensen and Pedersen (18), the control of growth lies in the availability of building blocks for synthesis of cellular components. This implies that under most circumstances, the machinery polymerizing the building blocks to cellular structures is subsaturated. Because 11GIc did not control growth, apparently other steps in metabolism control the growth rate under these circumstances.
Dijkhuizen et al. (6) have determined that the lactose permease has a high control coefficient on the growth rate under lactose-limited conditions. Furthermore, Hunter and Kornberg (17) have shown that under glucose-limited conditions, the uptake capacity of glucose via the PTS limits growth. In these experiments, glucose-limited conditions were obtained by growing cells in chemostat cultures. During growth of the strains, which contain an intact PTS, in glucose-limited chemostat cultures, the glucose concentration in the culture fluid is usually lower than 10 ,uM (32). Furthermore, because the apparent Km of 1101c for glucose is about 10 ,uM (38), the enzyme is operating at rates below its V x under glucose-limited conditions. It is possible that II8 has a higher control coefficient on the growth rate under these conditions. Apart from the effects on growth rate, we have determined the control coefficients of 1101c on oxidation of glucose and on uptake and phosphorylation of the glucose analog otMG. Oxidation of glucose involves a large number of catabolic steps, while uptake and phosphorylation of aMG represent only the first two steps in the catabolism of glucose. These two reactions are catalyzed by the glucose PTS, which consists of four proteins (Fig. 1). Generally, when the length of a metabolic pathway (i.e., the number of enzymatic steps involved) is decreased, it is to be expected that the flux control of an enzyme in this pathway increases. This is observed in our experiments. IIG0c did not control the rate of oxidation of glucose (i.e., a large number of catabolic steps) but had a relatively high control coefficient (0.55 to 0.65) on uptake and phosphorylation of otMG (i.e., the first two steps of catabolism).
A few reports have appeared that deal with determination of the flux-controlling properties of PTS enzymes. However, in these studies, flux control is described in a more qualitative way. For instance, Ghosh et al. (10) used E. coli strains overproducing enzyme I and concluded that overproduction of enzyme I did not lead to increased uptake and phosphorylation of aMG. These results are in accordance with a high control coefficient of I1GIc on uptake and phosphorylation of aMG. The metabolic control theory states that the sum of the enzyme control coefficients on the flux through an investigated pathway equals 1. Thus, if the control coefficient of 11G1c on the flux through the glucose PTS is 0.55 to 0.65, there is still 0.35 to 0.45 of control left to distribute between enzyme I, HPr, and -11G1C* Because apparently enzyme I does not control the flux through the glucose PTS, this leaves only HPr and IIGIc. Experiments are in progress to determine the flux control exerted by these PTS enzymes.
From a series of reciprocal inhibition experiments, Scholte and Postma (35) concluded that in S. typhimurium, IIGIc and 1IMan competed for the pool of phosphorylated enzyme I and HPr and that consequently the flow of phosphoryl groups through enzyme I and HPr was the ratelimiting step in the PTS reactions. Although the authors did not quantify the control properties of enzyme I and HPr, it is possible that these proteins do control the flux through the glucose PTS to a certain extent. However, because the data presented in this report indicate that in E. coli the control coefficient of TIGIc on the flux through the glucose PTS is 0.55 to 0.65, it is not likely that the control coefficient of enzyme I or HPr approaches 1 (i.e., is rate limiting). We suggest that the control on the flux through the glucose PTS is distributed among enzyme I, HPr, IIGIc, and 11GIc, with the largest part on IlGIc. This distribution of control over several enzymes is generally found when flux-controlling properties of enzymes are determined (12).
According to Scholte and Postma (35), the rate-limiting properties of enzyme I and HPr in S. typhimurium were in agreement with the model of the PTS-mediated regulation, which predicts a net dephosphorylation of PTS components, in particular IIIGlc, in the presence of a PTS substrate (Fig.   1). In this model, dephosphorylated IIIG'c accounts for the inhibition of uptake of several non-PTS substrates (glycerol, lactose, maltose, and melibiose). The relatively high control of 11GIc on the flux through the glucose PTS, presented in this report, seems to be in contradiction with this model. A high control coefficient of j1GIc on the flux through the glucose PTS could mean that the supply of phosphoryl groups to 11G1c is faster than the rate at which IIG"c donates its phosphoryl groups to its sugar substrate. In other words, 111Ic would be phosphorylated even in the presence of a substrate of the glucose PTS. To investigate the effect of IIGIc activity on the phosphorylation state of IIIGIc, we have measured the inhibition of glycerol uptake by addition of aMG as a function of the 11Ic activity. It appeared that at a II"c activity comparable to the induced wild-type level, glycerol uptake was almost completely inhibited (>90%) by addition of atMG. Thus, in that case III'Gc must be dephosphorylated to such an extent that it is able to inhibit glycerol uptake.
To explain the data of oxMG uptake and inhibition of glycerol uptake, the kinetic characteristics of the enzymes should be taken into account. During aMG uptake, 11GIc had a high control coefficient on the flux through the glucose PTS. Therefore, if we assume Michaelis-Menten kinetics, l1Glc was operating at rates near its Vmax during cxMG uptake and phosphorylation. In other words, the concentration of phosphorylated IIIG'c must have been high relative to the Km of IIGIc for phosphorylated IIIGIc. This Km value has been reported to be 3.4 FM in S. typhimurium (23) and 5 puM in E. coli (11), while the intracellular concentration of IIIGlc is estimated to be 30 to 50 p.M (36). Thus, even if lllGc is only partly in the dephosphorylated state, the concentration of phosphorylated IIIGIC is higher than the Km value. The target molecule of IIGIc for inhibition of glycerol uptake is glycerol kinase (29). According to Heller et al. (14), the intracellular concentration of glycerol kinase is about 15 puM. With respect to the Ki of IIGIc for glycerol kinase, several values have appeared in the literature. Postma et al. (29) have reported a value of 50 ,uM for S. typhimurium (pH 7.5), while Novotny et al. (25) claimed that this K, is pH dependent in E. coli: 10 ,uM at pH 7.0 and 4 puM at pH 6.0. If we assume a low Ki (see also reference 33) and a stoichiometric interaction between dephosphorylated IIIGIc and glycerol kinase, then severe inhibition of glycerol kinase is possible when IIIGlc is only partly present in the dephosphorylated state.
Thus, although IIG'c controls the flux through the glucose PTS to a large extent, this is not necessarily in contradiction with the presently used model of the regulatory functions of the PTS. To elucidate regulation involving the PTS, experiments are now in progress to determine the flux-controlling properties of enzyme I, HPr, and IIIGIc. Finally, to understand the regulatory properties of IIIG'c, the characteristics of the interactions between this enzyme and its target proteins should be quantified.