Substrate Binding and Reaction Intermediates of Glutamine Synthetase (Escherichia coli W) as Studied by Isotope Exchanges*

SUMMARY Substrate concentration effects on isotopic exchange rates at equilibrium have been measured with Escherichia coli glutamate synthetase (adenylated form) with azP, W, l*O, and ISN. A new test for possible compulsory substrate-binding orders is presented, involving measurement of equilibrium exchange rates while increasing concentration of all substrates. This test shows random substrate-binding patterns for the enzyme. Inhibition of some equilibrium exchange rates while various pairs of substrates were increased in concentration appears to result from competitive rather than compulsory binding order effects. The relative rates of equilibrium exchanges were (glutamate & glutamine) The inequalities show that interconversion of bound substrates is not the only rate-limiting step, and allow de-ductions about relative association-dissociation rates of various substrates. Glutamine synthetase from E. coli do& not catalyze any detectable all


SUMMARY
. Their work has resulted in an understanding of the main features of the enzyme's structure and of its intricate control by a variety of metabolic products derived in part from glutamine.
Information about substratebinding patterns and about possible intermediates in covalent interconversion are obvious requirements for satisfactory understanding of the enzyme's action.
This paper reports the results of investigations into these questions by measurement of partial reactions and equilibrium exchange rates with isotopic probes. Since the initial experiments by Doudoroff et al.
(2) with SUcrose phosphorylase, the demonstration of pertinent partial reactions has become increasingly recognized as an important means of revealing covalent intermediates in enzyme catalyses.
Such searches with E. coli glutamine synthetase have not been reported in any depth. With respect to elucidation of substratebinding orders, over the past decade measurements of isotopic exchange rates at chemical equilibrium in multisubstrate enzyme systems has developed as an important approach (for recent examples, see References 3 to 8). Although the theoretical treatments for such systems (8-12) have assumed the applicability of the usual Michaelis-Menten saturation kinetics, the prominent diagnostic features for compulsory binding orders from equilibrium exchange rates at various substrate levels are not dependent upon hyperbolic saturation kinetics.
For example, increase in the level of a substrate which binds last in a compulsory sequence necessarily at first stimulates then inhibits equilibrium exchange rates of substrates binding earlier irrespective of whether hyperbolic (Michaelis-Menten), cooperative, or other relations exist between initial velocity and substrate concentrations.
Relative initial rates of exchange between substrate pools can reveal mechanistic features as well. For example, if covalent interconversion of enzyme-bound substrates is definitively slower than substrate binding and release, all exchanges between or among various substrates must be equal. Although the kinetic and control patterns for glutamine synthetase are doubtless intricately complex, with adequate care in selection of conditions the above approaches sufKce to reveal important features of the enzyme mechanism. 25,1972 F. C. Wedler and P. D. Boyer 9% rich glutamate-glycerol medium 2 hours into stationary phase. Minor modifications of the purification were that the first ammonium sulfate precipitation of the enzyme was carried out at pH 4.60, not pH 4.40. Also, in Step 7 complete recovery of all activity required repeated extractions with buffer of the pH 5.15 precipitate.

Issue of February
The purified enzyme showed a single major (>99%) band upon disc gel electrophoresis. After dialysis to remove ammonium sulfate, the enzyme was stored at pH 7 in 0.01 M imidazole, 0.01 M MnCl?, buffer at 4'. Enzyme used in experiments to probe for partial reactions and intermediates was chromatographed on a column of Sephadex G-50 (bead), reprecipitated as in Step 7, redissolved, and dialyzed against the pH 7 imidazole-MnClz buffer. Then 5 ml of the dialyzed solution containing 50 mg of enzyme were then passed through a bed of Dowex l-chloride resin, 1 x 5 cm, pH 7.0. This is referred to in the text as "Dowex-treated" enzyme. A similar technique was used previously (14) to show the absolute requirement of submitochondrial particles for ADP in the Hz0 * Pi exchange during oxidative phosphorylation.
The observed specific activity of the purified enzyme agreed reasonably well with published values (13), as determined by biosynthetic or transferase assays. The degree of adenylylation was determined by ultraviolet absorption spectra, comparing AQeO and AN,, and by differential kinetic assays with MnCls and MgClz (15). The average number of AMP moieties per 12 subunits was found to be 10, the enzyme thus being designated as Elo. L-[14C]Glutamic acid was a Schwarz product, purified by DEAE-cellulose chromatography (see "Methods"), and was shown to be free of glutamine and pyrrolidone carboxylate by paper chromatography with l-butanol-acetic acid-water (4: 1: 1). n-[*4C]Glutamine was produced from L-[l'C]glutamate by a biosynthetic reaction with glutamine synthetase, NHI, and ATP. Purification was carried out by DEAE-cellulose chromatography.
[r4C]Pyrrolidone carboxylate was produced by the reaction of [14C]glutamate, 100 pmoles (0.010 PCi per pmole), in aqueous medium, pH 4.0, in the presence of 200 pmoles of phosphate for 48 hours at 100" (16). Separation of product from reactants was by paper chromatography and indicated a yield of 35%. L-Glutamate and L-glut,amine were Schwarz products, recrystallized from ethanol-water.
Nucleotides were from P-L Biochemicals.
All other compounds were reagent grade. Deionized water, twice glass distilled, was used for all solutions.
Methods-Separation of all substrates' in a reaction mixture was obtained by column chromatography with DEAE-cellulose (formate), usually 1 x 20 cm. Typically, a l.O-ml reaction mixture was diluted to 3 ml, applied to the column, and followed by 2 ml of distilled water. Sequential elution was accomplished by a linear gradient of pH 3.65 formate, formed from 20 ml of water and 20 ml of 0.8 M formate.
The order of elution was (ammonia), glutamine, glutamate, phosphate, ADP, then ATP. Amino acid peaks were routinely located by ninhydrin spray tests of 5-~1 portions of each fraction (usually 1.0 ml each) spotted on filter paper strips.
Phosphate was located by 32P 1 In this paper, the designation substrates refers to NHI, glutamate, ATP, glutamine, Pi, and ADP. Reactants refer to the first three; products to the latter three. The designation reaction components includes the six substrates plus metal ions, salts, buffer, etc.

tracer. Nucleotides
were located by A2e0. Where exchanges had occurred, radioactivity could be used as an additional check on location and separation of peaks.
Selection of reaction conditions was based on reported characteristics of the enzyme plus some additional experimental evaluations.
Imidazole buffer was avoided since it apparently can suppress enzyme activity as can L-histidine.2 Most studies were made at pH 6.50 because the adenylylated enzyme has maximal activity at this pH in the presence of manganese ion.
The equilibrium constant for the reaction at pH 6.50, 37", p = 0.25 M, was determined by carrying out the reverse reaction with 32P-labeled phosphate.
Since the micromoles of ATP formed as measured by 32P incorporation also equals the NH3 and glutamate formed as well as the glutamine, Pi, and ADP depleted, the calculation of the apparent Keq, where M-ATP and I!-ADP are the metal complexes, was relatively simple.
Triplicate determinations yielded an average value of 460 rt 30, falling within the range of published values at pH 6 and 7 (17). Pi and ATP were routinely separated by extraction of the neutral phosphomolybdate complex from acid solution into isobutyl alcohol-benzene (1: 1) or 4-methyl-2-pentanone; ATP remained in aqueous phase.
Levels of Mn+f and Mg++ were selected to insure optimal formation of enzyme-manganese and nucleotide-magnesium complexes but with limited formation of enzyme-magnesium and nucleotide-manganese complexes.
The reasons for this were that Mg++ has been noted to inhibit noncompetitively the Mn++dependent catalytic activity of adenylylated enzyme subunits2 and that equilibrium calculations assume (see equation above) complete formation of metal (manganese or magnesium) complexes. Manganese ion in excess of 1 mM at pH 6.5 formed colloidal precipitates in reaction mixtures. Thus Mg++ was added, equivalent to the ATP + ADP concentration.
Because of the large association constants for Mg+f-and Mn++-nucleotides (18), the level of free Mg++ was quite low. Because glutamine synthetase binds Mn++ some lOOO-fold more tightly than Mg+f at pH 6.50 (19) essentially no inhibitory enzyme-magnesium complex was formed.
Further evidence of the enzyme's lack of sensitivity to Mg ++ at pH 6.50 was the observation that addition of a slight excess of Mgff above total nucleotide at pH 6.5 did not appreciably alter either exchange activity at chemical equilibrium or biosynthetic (initial velocity) activity. In a typical procedure for measurement of isotopic exchange reactions, substrates, buffer, and metal ions were added at concentrations at or close to equilibrium values. The pH was adjusted, if necessary, the enzyme added, and the mixture incubated long enough to assure establishment of equilibrium.
The exchange reactions were then started by addition of very low amounts of highly labeled compounds, so that equilibrium was not perturbed.
The reactions were stopped after an appropriate interval of time, usually in the range of 10 to 30 min, by addition of 1 N HCl to bring the pH to 4.0 to 4.5, followed by freezing until ready for chromatographic separation of the substrates. The following procedure was used to allow variation in levels of substrate pairs or modifiers without altering pH, ionic strength, or other crucial parameters.
Two stock solutions were prepared.

986
E. coli Gluta,mine Xynthetase Mechanism Vol. 247,No. 4 Solution A contained all reaction components or modifiers at maximal concentration, and Solution B contained identical levels of components except those to be varied.
Typically, the ionic strength was well above the maximal total variat'ion in concentration of any component levels. After adjustment of the pH of the two solutions, Solutions A and B were then mixed in various proportions to a total volume of 1.0 ml. Preparation of samples for analysis of 15N in NH3 and gluta-mine was carried out by procedures outlined by San Pietro (20). The glutamine sample from a DEAE-cellulose column was diluted with nonenriched carrier.
To a lo-ml sample was added 5 ml of 40% KOH, and the solution was heated in a micro-Kjeldahl apparatus, rapidly flushed with Nz. The liberated ammonia was distilled, along with about 9Oy, of the water, into 5 ml of rapidly stirred 0.07 M HzS04. Either NH3 or glutamine could be treated directly in this manner.
In the experiments to test for possible r5NH3-labeled enzyme, the enzyme from the Sephadex column was treated similarly with KOH.
After distillation, the (NH&S04 solution was reduced in volume on a rotary evaporator to about 1 ml, then placed in one arm of a small Y tube. The other arm of the Y tube contained 1 to 2 ml of strongly alkaline hypobromite solution. Both solutions were then frozen, the ice degassed thoroughly at <5 /.L mercury pressure, then both were thawed carefully and mixed in vacua. This reaction converted NH3 to Nz gas, which was then analyzed in a high resolution Nuclide mass spectrometer for the 29 : 28 mass ratio (' sN14N : 14N14N).
The levels of radioactivity present in substrate pools were determined by liquid scintillation counting with Bray's solution (21) in a Packard Tri-Garb instrument.
Calculations of the total amount of substrate exchange were made using the relationship where F is the fraction exchange and X and Y are the amounts of exchanging components in micromoles.
Checks were made for ATPase or glutaminase activities, which could interfere with equilibrium exchange measurements if the hydrolytic rates were significant, relative to exchange rates. Glutaminase activity with the purified, 11owex-treated enzyme (20 units) was negligible: when 5 InM [14C]glutamine (0.02 /&i per pmole) was incubated with MnC12, KCl, and buffer at pH 6.5, 37" no appearance of 1% label in glutamic acid could be det)ected at a time when >50% glutamate would have been produced if Pi and ADP had been present.
ATPase activity under similar conditions, however, was appreciable but not, prohibitive.
Release of 32Pi from [Y-~~P]ATP with only ATP present occurred at about 0.17; of the net, init)ial rate of the complet,e reaction.
Thus inherent .%TPase activity of the enzyme could interfere in prolonged incuba,tions at equilibrium. Exchange reactions were usually carried out t'o only 10 to 20% of approach to isotopic equilibrium, which avoided any appreciable error resulting from ATP hydrolysis.

Partial
E'zchange Reactions--An isotopic exchange indicative of a llartial reaction is a quite sensitive l)robe for formation Of covalent intermediates from a given substrate. The intermediate may involve either an enzyme-functional group or a second substrate to give a noncovalently bound moiety. Such examples may be represented as3 In either case the AB * B isotopic exchange is indicative of intermediate formation, but in the former case involving E-A the A-B it B exchange occurs independent of added X. Table I presents the results of tests for possible phosphoryl enzyme or adenosine diphosphoryl enzyme intermediates, as probed by [14C]ADP ti ATP and "Pi 2 ATP exchanges.
In these experiments, sufficient radioactive la.bel was present to have allowed detection of any exchange occurring even at a rate only low4 that of the rate with all components present. That neither exchange occurs above this lower limit suggests that neither phosphoryl enzyme nor adenosine diphosphoryl enzyme moieties occur as kinetically significant, covalently distinct intermediates on the reaction pathway. Separate additions of NHs-free glutamate (recrystallized, Dowex-Kf treated) or of NH3 gave no observable stimulation of these exchange activities.
Hence, no enzyme-bound phosphorylated or adenosine diphosphoryl derivatives of these compounds are likely as reaction intermediates, most significantly not y-glutamyl phosphate.
Results of probes for amide enzyme or glutamyl enzyme intermediates are presented in Table II. Once again, neither predictable exchange, glutamate ti glutamine or NH3 $ glutamine, is observable within the limits of detection.
Added phosphate produces no stimulatory effects, arguing against bound glutamyl phosphate formation from t'he glutamine side of the reaction. Such formation would likely give detectable NH3 ti GluNHz exchange in presence of Pi. Intermediates-The lack of exchanges resulting from partial reaction systems, although quite indicative, cannot absolutely rule out the formation of covalent intermediates. Thus tests were made for direct isolation or detection of such possible intermediates. The search for phosphoryl enzyme involved incubation of enzyme with an equilibrium reaction mixture containing 32Pi, which was then quenched by extraction of the protein into phenol.
This procedure (22) (25) also revealed no additional spots, arguing for no amino acids present without free a-amino groups.
An investigation of the requirements for transferase activity of the enzyme revealed that omission of any or all reactants other than glutamine and NHsOH essentially abolished all activity. However, ADP alone was able to stimulate activity to 130/, of maximal, ADP plus phosphate to 18v/,. Phosphate acted antagonistically toward arsenate-stimulated activity.
That glutamine, NH20H, and either Pi or Asi or both in the absence of ADP resulted in no formation of detectable y-glutamyl hydroxamate also argues against an enzyme-bound activated glutamic acid unless ADP is present to complete the reaction system.
The ability of E. coli W glutamine synthetase to utilize pyrrohdone carboxylate was tested as follows.
About 1 pmole of 14C-labeled pyrrolidone carboxylate was incubated with MnClz, NH3, ATP, and enzyme at pH 6.5, 37", and this reaction com-Relative Exchange Rates-Comparison of the relative rates of exchange for ADP e ATP and Pi ti ATP (Table I) reveals that they are essentially equal. These rates, however, are not equal to the rates of glutamate F? glutamine and NH% + glutamine exchange (Table II), which are themselves unequal.
Arranged in decreasing order of velocity the exchange rates are: glutamine synthetase at pH 6.5, 37". The lack of detectable partial reactions or isolable intermediates points to a concerted mechanism for covalent interconversion and argues against any "ping-pang" mechanism. If this is so, the relative rates of exchange between substrate pools allow one to exclude certain orders of binding.
In a compulsory binding sequence, the most rapid exchange must occur between the last reactant to bind and the first product to dissociate; all other exchanges must be slower or equal in rate.
With three reactants and three products, the glutamine synthetase reaction offers various reaction component pairs for which the levels may be varied at equilibrium and between which isotopic exchanges may be observed, to give information about possible orders.
For example, one may vary the structurally similar pairs of ATP and ADP, glutamate and glutamine, Pi and ATP, and NH3 and glutamine.
Less similar pairs which are variable include NH3 and Pi, NH, and ADP, glutamate and Pi, glutamate and ADP, and ATP and glutamine.
With dissimilar pairs, competitive inhibition effects may result which might be avoided by varying similar pairs. The possible exchanges one may observe include glutamate ti glutamine, NH3 F? glutamine, ADP ti ATP, and Pi & ATP.
In addition, 180 is exchanged among the substrates and this additional isotopic probe can provide valuable information about the reaction mechanism at several different levels of interpretation, including substratebinding order, relative rates of association-dissociation, and stereospecific handling of normally asymmetric groups such as carboxylate or phosphate bound to the enzyme surface (26). Possible I80 exchange measurements include glutamate + glutamine, glutamate * Pi, glutamate F? ATP, glutamine G ATP, Pi ti ATP, and Pi ti glutamine.
Only some selections among these possibilities are reported in the paper which follows.
The effects of varying the levels of glutamate and glutamine in constant ratio upon the rates of glutamate F? glutamine and Pi * ATP exchange are presented in Fig. 1.4. The glutamate e glutamine exchange rises to a maximum, but with some slight inhibitory or discontinuous behavior. The data points of Fig.  1A were found to be reproducible in multiple experiments with less than 5% error.
An error of 20%, especially in the third data point at 1.25 mM glutamate would be necessary for the curve to appear hyperbolic.
The nonhyperbolic behavior may arise from one of several possible effects. Anticooperative substrate binding may occur,4 or high levels of glutamate and glutamine may induce nonproductive binding modes. Alternatively, substrates may bind to noncatalytic sites and induce thereby some weak activation or inhibition effects. As noted in Fig. IA, with increase in the glutamate and glutamine concentrations, the Pi F? ATP exchange is first stimulated, then suppressed strongly.
This effect could result from a compulsory binding order (glutamate may bind after ATP, or glutamine may bind after Pi in the reverse direction, or both may be true) or from a direct competition of glutamate or glutamine for a Pi-or ATP-binding site. Fig. 1B presents the effects of increasing ATP and Pi together: both exchanges rise smoothly to a maximum.
The effects of varying ADP and ATP as a pair are presented in Fig. 2A, and quite similar behavior is observed. These kinetic data allow exclusion of compulsory binding orders with ATP, ADP, or Pi as the last substrate to bind. Also none of these substrates act as negative modifiers at control sites separate from the catalytic site. ADP has been observed to compete directly for the ATPbinding locus in the catalytic site (28) Fig. 2A because both ATP and ADP were varied together. Increased levels of NH3 and glutamine produce the striking effects shown in Fig. 2B: both exchanges are initially stimulated, then suppressed strongly.
One may interpret the effects on the glutamate * glutamine exchange as either compulsory binding of NH, after glutamate, or as direct competition of NH, for a glutamate-or glutamine-binding locus. The interpretation of the suppression of the Pi d ATP exchange rate is more complex. It may be attributed to either compulsory binding of NH3 after ATP, glutamine after Pi, or both. Alternatively, or in addition, NH3, or glutamine, or both, may compete directly for an ATPor Pi-binding locus.
The results thus far presented show that nucleotides and phosphate do not appear to bind after amino acid or NH3 substrates in a compulsory manner, nor do they appear to compete with the latter for binding sites, but that the reverse may be true. The amino acids and NH3 do exert inhibitory effects upon the exchanges between ATP and Pi, and NH8 can exert similar effects on the glutamate e glutamine exchange.
To help discern among possibilities, an additional isotopic probe, the ATP F? ADP exchange was measured. Fig. 3A shows the effect of increased ATP and Pi upon the ATP * ADP exchange: the rate rises smoothly to a maximum value.
Thus Pi either dissociates after or randomly relative to ADP.
Since the conclusion from Fig. 2B was that ADP could not bind after Pi, ADP and Pi dissociate randomly relative to each other.

exchange.
Since ADP and Pi d' associate randomly relative t'o each other, the possible explanations for this phenomenon include either preferential but not compulsory binding of glutamate after ATP, or a competition of glutamate for an ATP-binding site. Also possible is an action of glutamate as a weak negative modifier.
The effect of variation in glutamate and ADP as a pair, Fig.  3C, produces more complete suppression of ATP * ADP exchange, but under these conditions (low fixed level of ATP) likely ADP competition for the ATP-binding site occurs, possibly added to the effects observed with glutamate in Fig. 3B.
Effect of Increasing Concentrations of all Substrates on Exchange Rates-The results given above, although eliminating some possibly compulsory binding orders, show inhibitions characteristic of certain compulsory and perhaps preferential orders, or of direct or indirect inhibitory effects, especially with the amino acid and ammonia substrates.
Further clarification was thus desirable.
One of the most definitive and direct experiments to substantiate or negate some of the above alternatives involves varying the levels of all substrates simultaneously in constant ratio. If either noncompetitive effects or compulsory binding orders occur, inhibition of the appropriate exchanges should still be observed when the concentration of all substrates is increased. However, if only competitive effects between substrates are the causes for inhibition when substrate pairs are increased, such competitive inhibitory effects should be absent when concentrations of all substrates are increased with constant substrate ratios. Fig. 4 presents the results of varying all substrate levels simultaneously in constant ratio at chemical equilibrium, as probed by the glutamate + glutamine and Pi * ATP exchanges. Both exchanges rise smoothly to a maximum.
The concentration range for each substrate extends beyond that which gave inhibitory effects reported in Figs. lA, 2B, 3B, and 3C. In addition, the maximum concentrations were considerably above the reported Km values. The results rule out compulsory binding orders or noncompetitive effects as responsible for the previously observed inhibitions. simultaneously and in constant ratio. The l.O-ml reaction at pH 6.50, 37" with 1.00 relative concentration contained (in micromoles) : 2 NHJ, 2 glutamate, 1 ATP, 20 glutamine, 20 Pi, and 4 ADP, plus buffer, salts, and enzyme as in Fig. 1. One prominent feature of the present data is the demonstration of random binding and release of substrates.
The simple experiment reported in Fig. 4 appears to represent a new and powerful approach to test for compulsory binding orders.
In this experiment, concentrations of all substrates were increased by the same ratio.
ilny competitions between substrate for binding sites thus remain unaltered.
The absence of decline in exchange rates as concentration of all substrates is increased eliminates compulsory substrate-binding orders or a noncompetitive inhibition by a particular substrate at a control site.
The application of this approach with glutamine synthetase was particularly useful because the inhibitions noted with some exchanges, as concentrations of reactant pairs were increased while maintaining equilibrium (Figs. IA and ZB), are consistent with patterns expected when compulsory or partially compulsory binding occur. These inhibitions must have other explanations, such as competitive displacement of a substrate whose concentration is not being increased.
In the cell, control of and catalysis by glutamine synthetase probably occur near chemical equilibrium, responding to relatively small perturbations in substrate or modifier levels. The competitive inhibitory effects observed for glutamate, glutamine, and NH3 may reflect important additional control mechanisms for this already complex enzyme system.
Some conclusions derived from data at chemical equilibrium may not be readily observable under initial velocity conditions. The criticisms by Dalziel (29) of some kinetic derivations by Fromm (30) appear to provide an example: partially compulsory binding order mechanisms may be observable and predictable over limited ranges of substrate concentration under initial velocity conditions, whereas for derivations involving exchanges at chemical equilibrium (9, 10) such binding orders are more clearly predicted and demonstrable in both theory and practice.
A second important feature of the results given in this paper concerns the relative rates of the various steps involved in catalysis and exchange.
A random binding order does not imply that various substrate exchange rates must be equal, as in the "rapid equilibrium-random" mechanism, which represents only one possibility.
From the relative exchange rates observed, namely that (glutamate Z+ glutamine) > (NH8 F? glutamine) > (Pi * ATP) = (SDP * ATP), one can deduce the relative rates of substrate association-dissociation, depicted as Steps 1 to 6 in The latter possibility of Statement c includes the situation where the rate of ATP < 4DP or Pi, and is more probable than the first alternative, where a fortuitous equality of ADP and Pi binding and release is required.
That NH3 * glutamine is slower than glutamate ti glutamine exchange confirms a point made earlier (Fig. 4), namely that the observed inhibitory effect of NH, upon the glutamate F? glutamine exchange (Fig. 2B) cannot be due to compulsory bindof NH, after glutamate.
If this were true NH3 e glutamine would necessarily be at least as rapid as glutamate 8 glutamine.
The observations that all substrate interchanges have rates within an order of magnitude of each other under the conditions tested may reflect a contribution of the rate of interconversion of the quaternary reactant-product complexes (Fig. 5) as well as substrate dissociation steps to limitation of observed over-all catalytic rates.
An additional important feature of the present data concerns the absence of detectable partial reactions; all reactants or products apparently must be present before any exchange between substrate moieties can occur. Similar behavior has been noted previously for the ADP F? ATP and Pi G ATP exchanges with glutamine synthetase from peas and from brain (25,27,31) but the tests reported here were more sensitive.
For example, in our experiments, the ADP * ATP exchange in presence of ATP and glutamate but in the absence of ammonia was less than lop4 of of that observed at equilibrium with the same concentrations of ADP and ATP in the presence of ammonia.
The absence of ADP F! ATP exchange, even in the presence of added glutamate and also the lack of observable NH3 F? glutamine exchange in the presence of Pi are of particular importance. Both these exchanges would be expected if y-glutamyl phosphate formed as a catalytic intermediate in an independent step. To distinguish definitely between a reaction sequence involving y-glutamyl phosphate as a catalytic intermediate as contrasted to a concerted mechanism does not appear possible from presently available data.
Meister has reviewed findings, principally from his laboratory, felt to favor y-glutamyl phosphate participation (32) with the ovine brain enzyme. But none of the results conclusively establishes this mechanistically attractive possibility.
For example, Krishnaswamy et al. (24) suggested that the lack of an ADP & ATP exchange in presence of glutamate resulted because both ADP and y-glutamyl phosphate remained firmly bound to the enzyme.
Such a possibility is difhcult to accept for E. coli glutamine synthetase in view of the random binding of substrates demonstrated by the present studies, Absence of dissociation of ADP formed from ATP in presence of glutamate would mean that ADP cannot leave the enzyme unless RCONH2 is bound. This would suggest a compulsory binding order in the reverse relation, in which ADP binds after RCONH2.
Considerations analogous to the ADP or! ATP exchange apply to the absence of an NH3 ti glutamine exchange in presence of Pi. Lack of dissociation of NH3 must be postulated if y-glutamyl phosphate is formed from glutamine and Pi. This would suggest a compulsory binding order in the forward reaction with ATP adding before NH3. Such compulsory binding order is contrary to our present findings.
Although lack of dissociation of NH3 and of ADP seems un- This result could, however, mean that the rate of association and reaction of hydroxylamine with enzyme having bound ATP and glutamate is more rapid than the rate of glutamate dissociation.
Another observation is the binding and migration of glutamate with the enzyme in presence of ATP.
This could reflect allosteric effects of ATP or a compulsory binding order with the brain enzyme.
Similarly the apparent formation of and specific binding to the enzyme of ADP and Pi from ATP in the presence of glutamate could result from the weak inherent ATPase activity of the enzyme, rather than from formation of y-glutamyl phosphate. Other observations include pyrrolidone carboxylate formation upon heat denaturation or ethanol quenching of reaction mixtures containing enzyme, n4g ++, ATP, and glutamine, as well as y-glutamyl hydroxamate formation from glutamine and NHzOH in presence of ADP and Asi. Bound reactants might yield pyrrolidone carboxylate as the enzyme active site is disrupted by protein denaturation. A catalytic site that binds ATP and glutamate for a concerted reaction with NH, might also produce y-glutamyl phosphate very slowly as a side reaction upon substitution of NHSOH for NHP. Also, if NHzOH and Asi can substitute for NH3 and Pi, or the y-phosphoryl of ATP, or both, transferase reactions might simply involve a rapid reversal and substitution of NHzOH for NH3 in the activated complex shown in Fig. 5 rather than "trapping" of an activated covalent glutamate intermediate.
Finally, differences in the E. coli and ovine brain enzymes may be such that the latter can stabilize a y-glutamyl phosphate and allow for its formation as a discrete substance.
It is perhaps significant to note here that the E. coli enzyme does not produce either directly or via y-glutamyl phosphate any detectable pyrrolidone carboxylate with the conditions of Meister (32) under which the ovine brain enzyme does so. In this sense the enzymes are obviously quite different.
Convincing evidence for y-glutamyl phosphate as a catalytic intermediate would be demonstration of a steady state level during catalysis, with rates of formation and disappearance that establish its kinetic competence as an intermediate.
Such an approach has formidable experimental difficulties. There remains another possibility not considered by Meister (32) for y-glutamyl phosphate participation. This is that very pronounced substrate synergism (34) exists, so that y-glutamyl phosphate formation requires the presence, but not covalent participation of, all reactants, or, conversely, of all products. An absolute requirement of bound NH8 for the y-glutamyl phosphate formation from ATP and glutamate, or a similar requirement of ADP with glutamine and Pi as substrates does not seem likely; where quantitation of the effects of other substrates has been possible, as in the observed substrate synergism with succinyl-Cob synthetase (33) and with phosphoribosyl pyrophosphate synthetase (7) only stimulatory but not absolute requirements have been observed.
In addition, no such absolute requirement of one substrate for a partial reaction of another substrate with an enzyme has to our knowledge been reported.
For example, with phosphoryl-transferring enzymes, some have been shown to involve formation of phosphoryl enzyme intermediates, while others appear to catalyze direct transfer of the phosphoryl group between substrates without phosphoryl enzyme formation.
As discussed elsewhere (35), in all instances where phosphoryl enzyme formation has been demonstrated or definitively indicated, the expected partial reactions have been detected by isotopic exchanges.
A concerted mechanism for enzymic reactions analogous to those of glutamine synthetase was suggested some time ago by Buchanan and Hartman (36).
A concerted reaction and. the associated substrate binding and release steps are indicated for E.
coli glutamine synthetase in Fig. 5. Obviously, one way in which an enzyme might favor a concerted reaction is by binding the substrates in appropriate juxtaposition.
For example, approach of the lone pair of electrons from the ammonia nitrogen to the y-car-boxy1 carbon of glutamate should enhance the ability of one car-boxy1 oxygen to form a partial bond with the y-phosphoryl phosphate of ATP, thus enhancing the leaving group capacity of ADP. The partially formed bond between the carboxyl and phosphoryl moieties serves to activate both, allowing ADP or NH3 to leave or attack, depending upon the over-all direction of the reaction under consideration.
Although the reaction profile may actually involve and reflect participation of more than one transition state form, not just the single form indicated in Fig. 5, such intimate details of the chemical interconversion are difficult to establish by present approaches.
The designation "concerted" appears useful, however, when discrete covalent intermediates are not detectable or isolable participants.