The 6-phosphogluconate dehydrogenase reaction in Escherichia coli.

This study is an attempt to relate in vivo use of the 6-phosphogluconate dehydrogenase reaction in Escherichia coli with the characteristics of the enzyme determined in vitro. 1) The enzyme was obtained pure by affinity chromatography and kinetically characterized; as already known, ATP and fructose-1,6-P2 were inhibitors. 2) A series of isogenic strains were made in which in vivo use of thereaction might differ, e.g. a wild type strain versus a mutant lacking 6-phosphogluconate dehydrase, as grown on gluconate; a phosphoglucose isomerase mutant grown on glucose or glycerol. 3) The in vivo rate of use of the 6-phosphogluconate dehydrogenase reaction was determined from measurements of growth rate and yield and from the specific activity of alanine after growth in 1-14C-labeled substrates. 4) The intracellular concentrations of 6-phosphogluconate, NADP+, fructose-1,6-P2, and ATP were measured for the strains in growth on several carbon sources. 5) The metabolite concentrations were used for assay of the enzyme in vitro. The results allow one to calculate how fast the reaction would function in vivo if ATP and fructose-1,6-P2 were its important effectors and if the in vitro assay conditions apply in vivo. The predicted in vivo rates ranged down to as low as one-tenth of the actual rates, and, accordingly, one cannot yet draw firm conclusions about how the reaction is actually controlled in vivo.

This study is an attempt to relate the in uiuo use of the 6-phosphogluconate dehydrogenase reaction in Escherichia coli with the characteristics of the enzyme determined in vitro. 1) The enzyme was obtained pure by affinity chromatography and kinetically characterized; as already known, ATP and fructose-1,6-Pz were inhibitors.
2) A series of isogenic strains were made in which in viva use of the reaction might differ, e.g. a wild type strain versus a mutant lacking 6-phosphogluconate dehydrase, as grown on gluconate; a phosphoglucase isomerase mutant grown on glucose or glycerol.
3) The in viva rate of use of the 6-phosphogluconate dehydrogenase reaction was determined from measurements of growth rate and yield and from the specific activity of alanine after growth in l-'4C-labeled substrates.
4) The intracellular concentrations of 6phosphogluconate, NADP', fructose-1,6-Pz, and ATP were measured for the strains in growth on several carbon sources. 5) The metabolite concentrations were used for assay of the enzyme in vitro. The results allow one to calculate how fast the reaction would function in uiuo if ATP and fructose-1,6-Pz were its important effecters and if the in vitro assay conditions apply in uivo. The predicted in viva rates ranged down to as low as one-tenth of the actual rates, and, accordingly, one cannot yet draw firm conclusions about how the reaction is actually controlled in z&o.
Neither the functions nor the control of the hexose monophosphate shunt are completely understood.
It can be thought of as a biosynthetic pathway leading to ribose-5-P, erythrose-4-P, and sedoheptulose-7-P, as a source of reduced NADP' for biosynthesis or as a cyclic pathway of carbohydrate metabolism forming glyceraldehyde-3-P.  Table II). Minimal medium was M63 (12) supplemented with 1 pg of thiamin-HCl, 0.2% carbon source unless indicated otherwise, and with histidine. The rich medium used for enzyme assays (Table II) 25'C unless otherwise specified. Glucose-6-P dehydrogenase was assayed like 6-phosphogluconate dehydrogenase (Tris buffer assay) but with 1 mM glucose-6-P as substrate.
Phosphoglucose isomerase was assayed as described (13), and 6phosphogluconate dehydrase was assayed by the procedure of Eisenberg and Dobrogosz (14). and affinity chromatography on Cibacron blue-Sepharose, following the report of Thompson et al. (16) that the yeast enzyme bound to such a column and could be eluted with NADPH.
Since a recent report describes an analogous purification (17), only limited details will be given. Cibacron blue FS-GA was coupled to Sepharose 4B-200 by the procedure of Rinderknecht et al. (18). One hundred forty-three units (in micromoles/min) of crude extract (specific activity, 0.6 units/mg of protein) were applied to a column (2.4 X 13 cm), and after washing the enzyme was eluted with 1 mM sodium 6phosphogluconate.
The pooled eluate (specific activity, 10.6) was rechromatographed on the same column using a linear gradient of 0 to 1 mM NADP'.
The enzyme eluted at 0.3 mM NADP' with a specific activity of 11.6. Metabolite Assays (Table III) After centrifugation a measured amount of the supernatant was neutralized with 1 eq of K&0:3, the KC104 was removed by centrifugation, and the supernatants were immediately assayed. The culture filtrates were also assayed; these values were not higher than the assay blank. Assays were fluorometric, and used a Farrand model A with a primary filter (Corning 5860) and a secondary filter (Corning 4303 and 3387, combined).
Specific assay methods were: B-phosphogluconate (20), glucose-6-P (19), NADP' (21), and fructose-1,6-P2 (22). Metabolite amounts were calculated back to internal concentrations using the values of an Am of 0.3, corresponding to 0.1 mg dry weight/ml (19), and of 2 ~1 of internal water/mg dry weight (5 Then specific activity of Ala = counts per min of Ala/total Ala = (1 -S)/(2(1 -S) + 5S/3), which simplifies to 1 -S = (5 x specific activity of Ala)/(3 -specific activity of Ala). For growth on [l-'4C]glucose or [l-'4C]glucose-6-P (specific activity, 1.0) of a wild type strain, the shunt gives unlabeled 3-carbon compounds at a rate of S/3, and the fractional rate of fructose-1,6-P2 formation from fructose-6-P is 1 -S/3. Then the specific activity of ). This may be solved for S by the quadratic formula. It should be noted that these estimates of S assume that alanine derives only from pyruvate, which in the case of the Entner-Doudoroff pathway comes equally from both halves of gluconate, and otherwise must derive equally from both triose phosphates.
For growth of the wild type strain on glucose equilibration of fructose-6-P and glucose-6-P is also assumed, and there is little if any metabolism by the Entner-Doudoroff pathway (13). The biosynthetic drain of metabolites from the shunt is not taken into account. Estimation of S without these assumptions is considerably more complex (l-3).

Enzyme
Kinetics-6-Phosphogluconate dehydrogenase was obtained pure by a new procedure using a strain with &fold normal amount of enzyme ("Experimental Procedures"). No activity was observed (1% would have been detected) with 1 mM glucose, glucose-6-P, fructose-6-P substituting for Gphosphogluconate or with 4 mM NAD' substituting for NADP+. In the latter characteristic 6-phosphogluconate dehydrogenase differs from glucose-6-P dehydrogenase of the same organism, which can use NAD+ with a K,,, loo-fold greater than that for NADP+ (25). Table I shows the effects of some metabolites on the reaction. In agreement with the results of others (6, 7) ATP and fructose-1,6-P* were inhibitory; ribulose-5-P and ADP also caused some inhibition. Fig. 1 shows that the ATP inhibition curve was sigmoidal with half-maximal inhibition at 3 mM. The fructose-1,6-P2 inhibition curve was hyperbolic with half-maximal inhibition at 0.025 mM. A similar fructose-1,6-Pz inhibition curve was Glucose-6-P 0 Fructose-6-P 0 Fructose-l-P 0 Fructose-l-P + fructose-6-P" 0 Fructose-1,6-P?  Reaction in E. cob also obtained with toluenized cells (Fig. l), which suggests that the partial inhibition is not an artifact of enzyme dissociation in dilute solution. Partial inhibition by fructose-1,6-P* is also a characteristic of the enzyme from Streptococcus faecalis (6).
The purpose of the present work was not to obtain a complete kinetic characterization of the enzyme. However, it was necessary to determine whether the rates depended greatly on the physical conditions of assay since we planned to later use the known in uiuo metabolite concentrations on the enzyme in vitro. Since the cells were grown in a medium containing 0.1 M phosphate buffer, and there was a report of internal and external concentrations of phosphate then being similar (27), we chose as a second assay system one containing 0.1 M potassium phosphate.
V,,,,, values were similar (see Table IV) in the two buffer systems (Tris and phosphate). Substrate affinities were similar for NADP+ (half-maximal rates at 1.7 and 2.8 X 10m5 M in Tris and phosphate, respectively), but differed considerably for 6-phosphogluconate (values of 1 X 10m5 and 1 X 10m4 M in Tris and phosphate, respectively). Inhibitions were less in the phosphate buffer, where 5 mM ATP inhibited only 20% (uersus 100% in Tris), and concentration of fructose-1,6-Pz about lo-fold higher was needed to give the same inhibition as in Tris. The pH uersus activity curves differed somewhat for the two buffer systems (not shown), but a pH of 7.5 was used in accord with recent determinations of intracellular pH (28,29). These limited data show that E. coli 6-phosphogluconate dehydrogenase shows fairly complex kinetics and that fructose-l&P2 and ATP are potential negative effecters in uiuo. The differences in rate of the enzyme reaction seen in the two assay systems presented a serious potential difficulty for experiments where in vitro and in vivo rates were to be correlated, since the true in uivo enzyme environment is unknown, and they were done using both assay systems.
Strains and Growth Characteristics- Fig.  2 shows the relevant pathways. A series of isogenic strains were prepared blocked at steps other than 6-phosphogluconate dehydrogenase (Table II). The growth of these strains in minimal medium on several carbon sources (glucose, glucose-6-P, gluconate, and glycerol) is also indicated. The key data are as follows. Strains containing both 6-phosphogluconate dehydrase (edd+), the first enzyme of the Entner-Doudoroff pathway, and 6-phosphogluconate dehydrogenase (grid+) grow about twice as fast on gluconate as strains containing only the latter enzyme. Likewise, growth on glucose-6-P or glucose is about twice as fast in a strain having phosphoglucose isomerase (pgi') as in a strain having only glucose-6-P dehydrogenase (zwfe). Strains lacking both phosphoglucose isomerase and either glucose-6-P dehydrogenase (DF1671DZl) or 6-phosphogluconate dehydrogenase (DF563) are completely unable to grow on glucose or glucose-6-P.
A strain lacking only glucose-6-P dehydrogenase (DF565) grows well on glucose. All the strains grow on glycerol. These growth characteristics are in accord with previous results (4) and with the scheme of Fig.  2, but the comparison has not previously been made in a single series of isogenic strains. Table II also       expected if the slower growth is caused by the inability of the cell to produce metabolic intermediates at normal rate. Levels of Metabolites during Growth-The next step was to determine levels of pertinent metabolic intermediates in the various strains in growth on several carbon sources (Table  III). We will not discuss each item, but make the following comments.
(i) There are several cases where certain metabolites were not found: for example, a pgi-mutant growing on glycerol contained no glucose-6-P or 6-phosphogluconate.
Such expected results confirm the assignment of genotypes and metabolic pathways.
(ii) In certain cases metabolite levels were quite high. An example is strain DF564, which lacks both phosphoglucose isomerase and 6-phosphogluconate dehydrase and where, in growth on glucose, the concentration of glucose-6-P was 18.5 mM (versus 1.25 mM in the wild type) and the 6-phosphogluconate concentration was 6.6 mM (versus 0.1 mu in the wild type). Such results were also expected and apparently reflect accumulation of a metabolite at a point where a major pathway is blocked but a second pathway remains open. They also show that the levels of certain metabolites are not tightly controlled, e.g. gluconate transport and phosphorylation activities do not act to maintain gluconate-6-P at wild type level. (iii) The concentration of NADP' did not vary much among strains and growth conditions. Its concentration probably would always be saturating for 6-phosphogluconate dehydrogenase.
(iv) 6-Phosphogluconate, the main substrate of interest in this work, would probably always (except for the special cases of zero concentration) saturate 6-phosphogluconate dehydrogenase, if one uses the enzyme characteristics as determined in Tris buffer. However, as assayed in phosphate buffer, this substrate would be below the K, in growth on glucose or glucose-6-P.
(v) The concentrations of fructose-1,6-Pz are in a range which would be expected to appreciably inhibit the enzyme, even as assayed in phosphate.
(vi) The ATP concentrations are relatively constant (approximately 4 mM for the few cases measured). This might be expected to give 50 to 80% inhibition if the enzyme in vivo were in an environment best matched by the Tris assay system, but if the phosphate assay system is a truer reflection of the cell, then ATP inhibition would be less. The Enzyme Activity Determined with in Vivo Substrate and Inhibitor Concentrations- Table  IV gives the predicted rate (as %V,,,,,) of 6-phosphogluconate dehydrogenase using the in viuo substrate and inhibitor concentrations from Table  III. For these experiments, the enzyme was a crude extract of the strain with a high level of 6-phosphogluconate dehydrogenase since the pure enzyme lost ATP sensitivity on storage. The experiments were done at the growth temperature (37°C) in both the Tris and phosphate assay systems. The results show the enzyme to function at a rate at least 50% of the V,,.  (Table VI). %V,,,., values are those obtained using a crude extract of the strain RW226/ pLC33-5 assayed at 37°C using the metabolite concentrations (NADP+, 6-phosphogluconate, glucose-6-P, fructose-1,6-P2, ATP, and ADP) determined for each of the five cases (Table III)  Alanine was isolated from cells grown in minimal medium with the indicated carbon sources (1 x lo5 cpm/pmol; relative specific activity = 1.0); see "Experimental Procedures." pgi-grid- To convert the %V msX values into predicted in viuo rates, one must fist know the amounts of enzyme in the several situations. Table IV also gives the enzyme levels ( Vmax) as assayed in the two buffer systems. The phosphate buffer gave slightly higher values than the Tris buffer, and the enzyme levels were similar in the several strains, as expected for a "constitutive" enzyme. The predicted in uivo rates would be the product of percent saturation and enzyme amount (see Table VI).
In Vivo

Use of '6Phosphogluconate
Dehydrogenase-When all substrate utilization is via the 6-phosphogluconate dehydrogenase reaction (e.g. growth on gluconate or glucose or glucose-6-P of a mutant lacking both 6-phosphogluconate dehydrase and phosphoglucose isomerase (edd-pgi-)), the in vivo use of the reaction may be determined as growth rate/ growth yield (i.e. micromoles of substrate used/unit cell/unit time). However, when glycolysis or the Entner-Doudoroff pathways are also available, only a fraction of total metabolism uses the shunt. We attempted to determine this fraction from the specific activity of a pyruvate derivative after growth on [ l-'4C]carbon source. Gluconate metabolism by the Entner-Doudoroff pathway or glucose metabolism by glycolysis would give radioactive pyruvate, while metabolism via the 6-phosphogluconate dehydrogenase reaction would give unlabeled products. Thus, one may derive (see "Experimental Procedures") an estimate of the fraction of metabolism by the two pathways from the specific activity of pyruvate. Alanine from protein is a convenient pyruvate derivative, and we have shown in earlier experiments that its radioactivity is in the carboxyl group when coming from l-labeled substrate by the Entner-Doudoroff pathway and not in the carboxyl group when derived by glycolysis (13). A similar protocol has recently been used to indicate pathways of fructose metabolism in pseudomonads (30). Table V shows the relative specific activities of alanine after growth on the l-labeled substrates. The expected values for a strain not using the shunt would be 0.5, and this expectation was fairly well met: DF563, grid-, and, presumably using the Entner-Doudoroff pathway, specific activity was 0.51 on gluconate; DF565, zwf, and, presumably using glycolysis, specific activity was 0.47 on glucose and 0.52 on glucose-6-P. The other control was a strain presumably using only the shunt (DF654, pgi-edd-): metabolites such as pyruvate should be unlabeled, which was the case for growth on llabeled gluconate (specific activity, 0.007); the value from cells grown on l-labeled glucose-6-P was 0.04, also low, but somewhat higher than expected (as found previously (13)).
The values for the strains where both the shunt and the Entner-Doudoroff pathway should be functioning (DF562) or both the shunt and the glycolytic pathway (DF567) were all approximately 0.43 in this experiment, as expected for minor use (approximately 20%) of the shunt in each case (see Table  VI). Table VI gives the rates of in uiuo use of the 6-phosphogluconate dehydrogenase reaction as determined from the total rate of substrate utilization corrected for fractional use of the shunt. It also shows (last column) the rates which would be predicted from the measured amount of enzyme were it to be functioning in vivo as it does in vitro with the actual concentrations of substrates and effecters. Predicted rates ranged between 16 and 98% of the estimated actual uses of the reaction.

DISCUSSION
There was generally poor agreement between the estimated rates of actual use of the 6-phosphogluconate dehydrogenase reaction in viva and the rates predicted from studies of the enzyme in vitro.
The amount of enzyme as assayed with saturating substrates and without inhibitors was somewhat lower than the estimated rates of in viuo use of the reaction, and lower yet using the in vivo substrate and inhibitor concentrations. Hence, the predicted rates ranged between 16 and 98% of the actual rates. There is considerable uncertainty in both sets of values, and the following problems may be cited.
(i) The physical state and environment of the enzyme in the cell is not known-there is even ignorance about the internal phosphate concentration-and any number of other ions might have importance. It is also not known whether the enzyme properties in dilute solution accurately reflect the in uiuo situation. We have not observed substantial kinetic differences using cells made permeable to substrates with toluene, but further studies of this type might be useful.
(ii) Even if the physical situation or ionic conditions were not critical, we may have not measured the right metabolites. There still might exist an unidentified activator. One modification of the experiments would be to include the reaction products ribulose-5-P, NADPH, and bicarbonate. An energy charge study would also be of interest.
(iii) Assay of metabolite concentrations is difficult. There may be degradation during harvest, or incomplete recoveries. Systematic errors of cell water content, etc., might not greatly affect conclusions about relative metabolite concentrations The 6Phosphogluconate Dehydrogenase Reaction in E. coli but could substantially affect the rates determined on the enzyme in vitro. Our measured metabolite levels are in general agreement with literature values (e.g. ATP measured with luciferase (31), fructose-1,6-Pz (32), NADP+ (5), and several metabolites measured by Lowry et al. (19)) but still might be too inaccurate for the intended purpose.
(iv) Another impediment to the evaluation of possible roles of ATP, fructose-1,6-P*, or other effecters is that there are no mutants in 6-phosphogluconate dehydrogenase altered allosterically but active catalytically. It is also not yet possible to change at will the level of the enzyme and observe the effect on metabolism.
Strains of these types probably could be obtained by appropriate mutagenesis and genetic construction.
(v) Determination of in vivo use of the Gphosphogluconate dehydrogenase reaction is only easy for the trivial case when all metabolism is by the shunt (Table VI, last two lines). For the other cases, where another and quantitatively more important pathway is also available, a correction was made according to the specific activity of a pyruvate derivative, alanine. Such calculations depend on many assumptions (see Refs. 1 and 2). In particular, biosynthetic use of metabolites from the shunt was not taken into account. Thus, if the 6phosphogluconate dehydrogenase reaction were the source of all ribose and aromatic amino acids, approximately 20% of all assimilation and 7% of metabolism'might use the shunt without affecting the specific activity of alanine; the in vivo estimates in the fist three cases of Table VI, therefore, might be low by approximately 30%. It should also be noted that it is not even certain that alanine derives exclusively from pyruvate (33).
Accordingly, although the present work has given useful data, it does not allow any strong conclusions to be drawn about control of metabolism in the hexose monophosphate shunt, let alone about the particular roles of fructose-1,6-Pz and ATP as inhibitors of the 6-phosphogluconate dehydrogenase reaction. Even for the single situation where there is strong indication that the shunt must be inhibited, namely, anaerobic growth (see Ref. 4), the substrates were present in adequate amount and the potential inhibitor, fructose-1,6-Pz, was in lower concentration than aerobically (see Table III).