Reactivation of Glutamine Synthetase from Escherichia coli after Auto-inactivation with L-Methionine-S-sulfoximine, ATP, and Mn2+*

Escherichia coli glutamine synthetase auto-inacti- vated with L-methionine-S-sulfoximine and ATP can be completely reactivated at pH 3.5-4.6 in 1 M KC1 and 0.4 M (NEt).&04. Both unadenylylated and adenylylated magneslum and manganese enzymes can be reacti- vated. Reactivation of fully inactivated enzyme is first 2 min at pH 4.1 and 37 "C) and coincides with the stoichiometric release of 0.95 k 0.05 eq each of L-methionine-S-sulfoximine phosphate and ADP and 2.0 f 0.2 eq of Mn2+ from each subunit. The rate of reactivation increases with decreasing pH and is proportional to the 3rd to 4th power of the hydrogen ion activity; the protonation of 3-4 carboxylic acid groups/ subunit therefore may be required to disrupt the enzyme complex. Reactivation rate also increases with increasing KC1 concentrations and temperature, with an Arrhenius activation energy of -26 kcal/mol, suggesting that some protein structural perturbation required to disrupt the complex. Upon neutralization of reactivation solutions, the ligands and metal ions re- combine with the enzyme resulting in its complete reinactivation. Thus, of glutamine synthetase at reversible, of

The glutamine synthetase of Escherichia coli is composed of 12 identical polypeptides arranged in 2 superimposed hexagonal rings (1). In addition to cumulative feedback inhibition, enzymatic activity of glutamine synthetase is modulated by covalent modification catalyzed by an adenylyltransferase that is part of an exquisitely regulated bicyclic cascade system (1, 2). Each subunit contains an active site (3), separate binding sites for several feedback inhibitors (1,4), and a specific tyrosyl residue that is the acceptor for the 5'-adenylate group (5). Also, each subunit has two essential sites for divalent metal ion binding; both sites must be occupied for f u l l expression of activity (3). The n~ site binds Mn2+ or M e with high affinity and has a structural and possibly a catalytic role (6-8). The n2 site has lower affinity for Mn2' and binds the nucleotide-metal ion complex with higher affinity than Mn2+ alone (3). Metal ions play a complex role in catalysis by glutamine synthetase: unadenylylated enzyme is active with * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed. magnesium but inactive with manganese in the biosynthetic reaction and is active with both metal ions in the y-glutamyltransferase reaction; adenylylated enzyme is active with manganese but inactive with magnesium in both the biosynthetic and the y-glutamyltransferase reactions (1).
Glutamine synthetases from E. coli and various other organisms undergo auto-inactivation in the presence of the Lglutamate analog, L-methionine-S-sulfoximine, metal ions, and ATP (9). Inactivation results from the extremely tight binding to the enzyme of the transition state analog L-methionine-S-sulfoximine phosphate and ADP, which are formed on the enzyme by transfer of the y-phosphoryl group of ATP to the imino nitrogen of L-methionine-S-sulfoximiie (10, 11).
This reaction is analogous to the first step of the biosynthetic reaction which involves the formation of an activated intermediate between L-glutamate and ATP, possibly y-glutamyl phosphate (9, 12, 13). Glutamine synthetase, however, does not display the same discrimination for metal ions in catalyzing the phosphorylation of L-methionine -5"sulfoximine as in catalysis of the biosynthetic reaction; the reaction occurs with both the manganese and magnesium forms of the adenylylated and the unadenylylated enzymes (14). Although it is known that all 12 subunits of glutamine synthetase are catalyticdy active (3), it is uncertain whether aU 12 subunits can catalyze the auto-inactivation reaction. Using enzyme with various extents of adenylylation, Weisbrod and Meister (14) found that fully inactivated glutamine synthetase from E. coli contained 9-11 mol of L-methionine-S-sulfoximine phosphate/ mol of dodecameric enzyme, and Rhee et al. (15) found that partially inactivated enzyme contained less than 1 eq of Lmethionine-S-sulfoximine phosphate bound/eq of activity lost. However, Hunt and Ginsburg (8) found that the inactivated enzyme contains 2 eq/subunit of Mn2+ or Mg2' which are bound with very high affiity ( K k > lo9 "I). Also, the stoichiometry for the reversible binding of L-methionine-Ssulfoximine to glutamine synthetase is 1 eq/subunit (16).
This paper describes conditions for the complete restoration of enzymatic activity to glutamine synthetase inactivated with ATP and L-methionine-S-sulfoxirnine. Reactivation experiments allowed characterization of several metal ion and ligand interactions with the enzyme and indicated the presence of inactive complex on all subunits of the dodecamer.

EXPERIMENTAL PROCEDURES
Materials-All aqueous solutions were made with distilled water that was deionized and filtered through a Millipore Milli Q2 reagent grade system. L-Methionine-SR-sulfoximine, Hepes,' Tris, and ATP

4272
Reactivation of Inactivated Ghtamine Synthetase were from Sigma. Poly(ethy1ene)imine cellulose was from Eastman, Chelex 100 was from Bio-Rad, and the tetrasodium salt of xylenol orange was obtained from Fisher. Separated S-and R-diasterioisomers of L-methionine-SR-sdfoximine were a kind gdt of Dr. Magnesium enzyme was prepared by treating purified manganese enzyme with magnesium-Titreplex (Merck) and dialyzing the enzyme against 3 X lo00 volumes of buffer containing 50 m~ MgCL and 100 mM KCl. Adenylylated enzyme was the g f t of R. J. Hohman and was prepared from unadenylylated glutamine synthetase by enzyme-catalyzed adenylylation (18). Assays of Enzyme Activity and Protein-Enzymatic assays for glutamine synthetase and its state of adenylylation were performed using the y-glutamyltransferase assay at pH 7.57 as described (19). Concentrations of active enzyme were calculated using a specific activity for pure enzyme of 107 units/mg. Protein concentrations were determined after correcting UV absorbance for light scattering using absorption coefficients of A% = 3.85 and A&$, = 7.33 + (rt/12)0.50 (20). Protein concentrations of inactive and other modified forms of glutamine synthetase were determined by the colorimetric assays of Bradford (21) and Lowry et al. (22) with purified glutamine synthetase as the protein standard.
Preparation of Inactive Enzyme Complex-Manganese or magnesium glutamine synthetase (1-10 mg/ml) was combined with 1-5 m~ L-methionine sulfoximine (either the commercially available Sand R-diasterioisomeric mixture or the resolved S-isomer) in 20 mM Hepes or 20 mM imidazole, pH 7.2,100 m~ KCl, and 5 m~ MnCL or 50 m~ MgClz. ATP (0.2-2 m~) was then added and the solution was incubated at 37 "C for 1-2 h and then overnight at room temperature. Unbound small molecules were removed by dialysis against 3 X lo00 volumes of buffer containing 100 mM KCI. Manganese enzyme retained ~0 . 5 % and magnesium enzyme 0.5-7% activity after this treatment. Storage for up to 4 months at 3-6 mg/ml led to no s i m c a n t return of activity and only a very slow release of tightly bound ligands (2-3%) as measured by release of [I4C]ADP. Inactive enzyme complex containing either [I4C]ADP or ~-[~~P]methionine-S-sulfoximine phosphate was prepared as described above with the addition of either [I4CJATP or [Y-~'P]ATP. Fully inactivated, unadenylylated enzyme had 11-12 eq of tightly bound ligands/dodecamer; fully inactivated adenylylated enzyme contained 9-10.5 eq of tightly bound ligands/ dodecamer.
Measurements of Rates of Inactivation or Reactiuation-Reactivation solutions contained 50 m~ acetic acid adjusted to various pH values with KOH and KC1 and (NH4)2S04 as indicated. For measurements of rates of inactivation or reactivation, incubations were performed with 50-200 pg/ml of enzyme which allowed direct transfer of 5-20-pl aliquots to the assay solution for convenient determination of enzymatic activity. Inactivation and reactivation were terminated upon transfer as judged by the linearity of the assay determined for 0.5-30 min. First order rate constants were calculated from semilog plots of the percentage of remaining active or inactive enzyme uersus time with 4-7 aliquots taken at timed intervals.
To measure the release of ['4C]ADP or ~-[~~P]methionine-S-sulfoximine phosphate from inactive complexes, the incubation mixtures were first diluted into y-glutamyltransferase assay solution; this procedure stops inactivation or reactivation immediately. Separation of protein-bound and unbound ligands was accomplished by passing the diluted enzyme ( 4 5 pg) through 0.45 p Millipore filters (HATF 01300 or Millex HA). Less than 2% of the protein passes through the filters when the ionic strength is kept high (>I00 m~ KC1). The filtrate was then analyzed for the desired ligands released. To measure Mn2+ release during reactivation, the reactivation mixture containing the precipitated enzyme was filtered directly through a 0.46 p Millipore filter. The precipitate was quantitatively retained by the fdter and separation of protein and free Mn2+ took -5 s.
Mnz+ Determinations-Free Mn2+ concentrations were measured using the dye-metal ion chelator xylenol orange (23). Stock solutions of 0.1-1 m~ xylenol orange* were prepared in 20 m~ Hepes/KOH-* We recommend storage of xylenol orange (see the Eastman Chemical catalog for chemical structure) stock solutions at -20 "C in the 100 mM KC1 and passed through small columns (0.6 X 4 cm) of the K' form of Chelex 100 (100-200 mesh) equilibrated in the same buffer. The preparation of the K+ form of Chelex 100 and standardization of the MnCb solution were as described previously (8). At pH 7.2 in 20 m~ Hepes/KOH-100 mM KC1 and 585 nm, E = 2.53 X lo4 M" cm" for xylenol orange, E = 4.44 X lo4 W 1 cm" for the Mn-xylenol orange complex, and A€ = 1.9 X lo4 M" cm" for xylenol orange binding Mn" to form a 1:l complex with K ' A = 1.6 X IO5 M" (23).
For spectrophotometric measurements of Mnz+, the 0.1 absorbance scale of a Cary model 15 spectrophotometer was used with a neutral density screen in the reference compartment and semimicrocuvettes of 10-mm path length in the sample compartment. Calibrations were performed keeping the pH and buffer composition constant. During reactivation experiments at pH 4.26 and 37 "C, deproteinized filtrates of 100 pl each (see above) were neutralized with 20 pl of 1 M Hepes/ KOH at pH 7.2, and 100 pl of each neutralized filtrate was added to 800 pl of 35. 2) added to the spectrophotometer cell confirmed the absorbance change due to the Mnz'-xylenol orange interaction. An aliquot of the dialyzed, inactive enzyme used for each reactivation experiment was added directly to a xylenol orange solution and the free Mn" initially present was determined to be 4 0 % of the protein-bound Mn".
p H Measurements-Radiometer PHM Type 26C pH meter equipped with a GK 2322 electrode (Radiometer) was adjusted with standard pH buffers at the temperatures indicated for pH measurements.
Scintillation Counting-All scintillation counting was done with 10 ml of Aquasol and 0.50-1.0-ml aqueous sample in a Beckman IS250 liquid scintillation counter.
Zdentification of ADP and L-Methionine-S-sulfoximzne Phosphate-Inactive enzyme complex containing ['4CJADP or ~-[~'P]methionine-S-sulfoximine phosphate was reactivated at pH 4.1 in 1 M LiC1. Precipitated protein was removed by filtering through a 0.45 p filter and 20 pl of the filtrate was spotted on a poly(ethy1ene)imke cellulose thin layer chromatogram. The spots were washed with anhydrous methanol and ascending chromatography was carried out with 0.5 M LiCl. Relative mobilities were: ATP, 0.10; ADP, 0.18; orthophosphate, 0.49; and L-methionine-S-sulfoximine phosphate, 0.40. Compounds were detected under UV light or by autoradiogra-Fluorescence Measurements-Protein tryptophanyl fluorescence was measured using a Hitachi Perkin Elmer MPF 2A Spectrofluorometer. Cuvettes were housed in a water jacketed cuvette holder at 25 "C. Excitation was at 300 nm and emission was at 340 nm with slits set at 6 nm. phy.

RESULTS
Stability of Inactive Complex-The complex formed when glutamine synthetase from mammalian brain or from E. coli is inactivated with ATP and L-methionine-S-sulfoximine is extremely stable. Previous efforts to reactivate the inactive enzyme complex of sheep brain glutamine synthetase by various methods (dialysis, gel filtration, ion exchange chromatography, urea treatment, or ethanol precipitation) were unsuccessful (24). Similarly, we have observed that the inactive complex of E. coli glutamine synthetase is not rapidly disrupted at neutral pH by any of the following treatments: incubation in buffers at pH 8-11; incubation with 2 M KC1,l dark, under which conditions the dye is indefinitely stable. Some decomposition of the dye was observed when stock xylenol orange solutions were left on the laboratory bench for more than a week. M LiCl, 10 m~ EDTA, or 20 m~ pyridine-2,6-dicarboxylic acid; heating to 70 "C with chelating agents present. The inactive enzyme complex is stable when stored in pH 7 buffer at 4 "C for >6 months. Meister (9) found that incubation of the inactive complex at 100 "C or in strong acid (<pH 2) or strong base (pH >12) could release L-methionine-S-sulfoximine phosphate and ADP from the inactive enzyme. We sought milder conditions for disruption of the inactive enzyme complex in order to rule out the possibility of covalent attachment of any of the components of the complex to the enzyme and in order to recover active enzyme. Reactivation of Inactivated Glutamine Synthetase-Complete reversal of inactivation with good recovery of the enzymatic activity of glutamine synthetase was obtained in these studies by incubating the inactive complex at 25 "C at pH 4.0 in the presence of 1 M KC1 and 0.4 M (NH4)2S04 (Table I).
Although quantitative yields of enzymatic activity can be obtained, the yields of activity vary from 60-100% primarily because of the susceptibility of glutamine synthetase to denaturation in acid and the great variability in the rate of reactivation (see below). Thin layer chromatography established that L-methionine-S-sulfoximine phosphate and ADP, which are the products released by heat or acid treatment (24), were released from the enzyme during reactivation. Equally good recoveries of activity were obtained with inactive complexes of unadenylylated glutamine synthetase containing either Mn2+ or Mg2' and with adenylylated enzyme containing Mn2+ (data not shown). Fig. 1 shows that the release of L-methionine-S-sulfoximine phosphate, ADP, and Mn2+ was in direct proportion to the extent of reactivation. From the specific activity of pure glutamine synthetase and the concentration of each ligand released from the enzyme throughout reactivation, it was calculated that 0.95 f 0.05 eq of each [14C]ADP and L-["P] methionine-S-sulfoxie phosphate and 2.0 f 0.2 eq of Mn2+ were removed per subunit of enzyme. Thus, dissociation of active-site ligands from an enzyme subunit restores full activity to that subunit. On the time scale of our experiments, there was a simultaneous release of Mn2+, L-methionine-Ssulfoximine phosphate, and ADP in the ratio of 2:l:l. Thus, the metal ions are integral components of the inactive enzyme complex in addition to the previously identified products of the auto-inactivation reaction, L-methionine-S-sulfoximine phosphate and ADP (10, 11, 24).
Following a short lag, the kinetics of reactivation of the inactive enzyme complex were first order ( Fig. 1, inset). Protein was collected by centrifugation at 4 "C for 15 min at 30,000 X g, washed once in the incubation buffer, collected as above, and redissolved in 20 m~ imidazole, 1 m~ MnC12, 100 mM KCl, pH 7.2. Assays and protein determinations were performed by routine methods. with L-methionine-S-sulfoximine and ATP as described under "Experimental Procedures." Four different preparations were made and in all cases the fully inactivated enzyme contained 11.2-11.7 eq of complex/dodecamer of active enzyme present initially. To measure release of radioactive ligand, samples from reactivation solutions were diluted directly into assay mixtures, and a portion of the assay solution was analyzed for free 32P or "C ligand by the Millipore filter assay and the remainder was analyzed for enzymatic activity. For Mn2+ release, samples were withdrawn for enzymatic assay and 20 s later samples were taken for Mn2+ assay. Correction for the differences in sample times were made by interpolation of the kinetic data for return of activity. The percent of Mn2+ released was based on the expected amount of Mn" present, 2 metal ions bound/subunit (8) activity of reactivated enzyme was linear when assayed from 0.5 to 30 min. Thus, any protein conformational changes that may occur are rapid upon dilution into the same assay solution. The correspondence between the kinetics of reactivation and ligand release suggest that the rate-limiting step in reactivation of inactivated glutamine synthetase is dissociation of the ligands bound at the active site.
Effects ofpH, Salt, and Temperature on the Rate of Reactivation- Fig. 2 shows that above pH 4, the first order rate constant of reactivation at 25 "C increased with the 3rd or 4th power of the hydrogen ion activity (A log k / A pH = 3.6 f 0.5). KCl, which increased the rate of reactivation, had no effect on the slope of the pH curves. The falling off of the reactivation rates below pH 4 is most likely due to the irreversible denaturation of glutamine synthetase in acid, since the recovery of enzyme activity was lower at pH t4 than at pH >4. If the linear portions of the pH curves in Fig. 2 are extrapolated to higher pH (which assumes that the rate-limiting step remains the same), a rate constant for reactivation at pH 7.0 of lo-" to min" can be estimated. Such a remarkably slow dissociation rate is consistent with the seemingly "irreversible" nature of the inactivation carried out at neutral Reactivation of the inactive complexes of unadenylylated manganese and magnesium enzymes and the adenylylated PH. manganese enzyme showed the same pH dependence. However, the unadenylylated manganese complex was more stable at acidic pH (reactivated more slowly) than the unadenylylated magnesium complex or the adenylylated manganese complex. This order of stability is consistent with the previously determined binding affities of Mn2+ and M e to glutamine synthetase (3, 25) and with the relative affinities of the unadenylylated and adenylylated enzymes for ADP. Mn2+ (3). It is likely that the steep pH dependence of reactivation reflects a requirement to protonate several ionized carboxylate groups in order to weaken metal ion binding in the complex.
Increasing concentrations of KC1 increased the rate of reactivation for both the manganese and the magnesium forms of the inactive complex (Figs. 2 and 3). The effect of KC1 on reactivation of unadenylylated manganese complex is shown in Table 11. Other neutral salts such as LiC1, NaC1, and NHX1 also accelerated the reactivation. Ammonium and potassium sulfates however were less effective than the respective chlorides (Table 11). Incubation of the inactive complex at 25 "C for 1 h in 50 m~ acetic acid (pH 3) with no additional salt, resulted in no reactivation and no dissociation of the complex.
The temperature dependence of the rate of reactivation of the inactive complex is shown in Fig. 3 ADP. MglGS. Below 55 "C, KC1 increased the rate of reactivation of inactive enzyme complexes equally at all temperatures. The break occurring above 55 "C in the Arrhenius curve for reactivation of unadenylylated [Mn. MSOXP -ADP.MnJ GS in the absence of KC1 (Fig. 3) was observed whether return of enzymatic activity (two experiments) or the release of ["C] ADP from the inactive complex (one experiment) was measured. KC1 (1 M) appeared to eliminate this thermal transition. A similar break in the curve for the reactivation of the magnesium complex in the absence of KC1 was not seen, indicating some difference in conformation between the manganese and the magnesium forms of the complex.
Reversible Binding of the Components of the Inactive Complex-When the pH of the reactivation solution was raised above pH 6, L-methionine-S-sulfoximine phosphate, ADP, and MnZ+ bound rapidly to the enzyme, resulting in complete reinactivation of the enzyme (Fig. 4). Reinactivation involved the rebinding of the free components of the inactive complex, since ['%]ADP released from the inactive complex during reactivation was freely exchangeable with exogenous ADP added during reinactivation (Fig. 4A) and chelation of the free Mn2+ with 1 m~ EDTA completely inhibited reinactivation. Reinactivation was rapid even when the concentration of glutamine synthetase subunits, ADP, and L-methionine-S-sulfoximine phosphate were diluted to 0.2 , UM in the presence of 1 mM MnZf (Fig. 4B). The kinetics of reinactivation with excess Mn2+ added appeared to be second order ( k -lo6 M" s-' at 25 "C), which suggests that binding of only one ligand is required for inactivation. This conclusion was confiied by removing ADP with charcoal prior to neutralization; inactivation proceeded rapidly with only L-methionine-S-sulfoximine phosphate and metal ions present (data not shown). When present in stoichiometric amounts with enzyme subunits, [l4C]ADP bound to the enzyme during reinactivation with apparently the same high affinity with which it binds in the original inactive complex. However, the addition of excess Mn. ADP (0.5 m~) to the reinactivation mixture partially blocked reinactivation, indicating that nucleotide has a complex role in formation of the inactive enzyme complex.
The data of Figs. 1 and 4 show clearly that the inactive complex consists of the products of the auto-inactivation reaction, L-methionine-S-sulfoximine phosphate and ADP, and 2 Mn2+ tightly bound to each enzyme subunit and that these ligands are released as free components during reactivation. The auto-inactivation reaction and conditions for dissociation (reactivation) and reassociation (reinactivation) of active site ligands in reversible binding processes are outlined in Scheme 1.
Auto-inactivation in the Presence of Limiting ATP or L -Methionine-S-sulfoximine-When manganese unadenylylated glutamine synthetase was incubated with limiting ATP or L-methionine-S-sulfoximine in the presence of excess Mn2+ and the other substrate, inactivation was proportional to the amount of the limiting substrate used (Fig. 5). This result is in agreement with the proportionality found between activity recovered and inactive complex dissociated during reactivation (Fig. 1). However, only 0.85-0.90 eq of reactant/subunit was incorporated for each equivalent of subunit activity lost, which suggests that formation of the inactive complex on one  (0, A). Inactivation reactions were initiated with ATP, although the order of addition had no effect when the magnesium enzyme was used.
subunit inactivates that subunit and in addition partly inhibits adjacent subunits. This ratio is close to that observed by Rhee et al. (15), who measured 0.83 eq of complex/subunit inactivated, but appears to contradict the results obtained with excess reactants present which gave 0.95 to 1.0 eq of complex/ enzyme subunit (Fig. 1).
Nucleotide-induced Heterogeneity of Inactiuation-Previous investigations have shown that the rate of auto-inactivation of glutamine synthetase (Scheme 1) decreases as the enzyme is inactivated (14, 15). The decrease in rate of inactivation is not simply a function of the extent of inactivation, since, when partidy inactivated enzyme was isolated and incubated again with ATP and L-methionine-S-sulfoximine, the fast and slow phases of inactivation had similar rates when starting with 100, 80, 50, or 20% active enzyme (data not shown). Moreover, the distribution between the rapidly inactivated and the slowly inactivated enzyme populations was influenced by ATP. Fig. 6A shows that when the unadenylylated enzyme was preincubated with ATP and Mn2+, -50% of the enzyme was inactivated in 1 min following addition of L-methionine-S-sulfoximine but the remaining 50% was inactivated at least 50 times more slowly. This effect was not dependent on the time of preincubation with ATP from 0.5-30 min. With the manganese adenylylated enzyme, the rate and extent of inactivation was not significantly influenced by the order of addition of ATP and L-methionine-S-sulfoximine. When the magnesium unadenylylated enzyme was used, preincubation with ATP was not required to induce heterogeneity in the rate of inactivation. At a low ratio of MgCL to ATP (2:l) >50% of the enzyme was inactivated slowly, whereas at a high ratio of MgC12 to ATP (20:l) only 5-10% of the enzyme was inactivated slowly (Fig. 6B). When a fmed concentration of 10 mM MgC12 was used, decreasing ATP concentrations from 5 m~ to 0.2 m~ led to progressively faster inactivation rates and more complete inactivation (-80% in 2 min at 0.2 m~ ATP).
The effects of ATP on the rates of inactivation are not reflected by the enhancement of enzyme tryptophanyl fluorescence observed upon binding of ligands. Manganese unadenylylated enzyme gave identical quantum yields when treated with ATP and L-methionine-S-sulfoximine whether ATP was added first or last. Similarly fluorescence of magnesium unadenylylated enzyme was the same when inactivated in the presence of 10 m~ MgC12 (-50% inactive) or 50 m~ MgCl2 (-95% inactive). Thus the fluorescence change occurs upon binding of ATP and L-methionine-S-sulfoximine and is not affected by the inactivation (phosphoryl transfer) reaction.

DISCUSSION
This study shows that E. coli glutamine synthetase inactivated with ATP and L-methionine-S-sulfoximine in the presence of Mn2+ or Mg+ can be reactivated in good yield under relatively mild conditions. Reactivation coincides with the stoichiometric release of 0.9-1 eq of each of L-methionine-Ssulfoximine phosphate and ADP and 2 eq of Mn2+/subunit from the inactive complex. On the time scale of the experiments (>0.1 min), all components of the complex appear to be released simultaneously; however, an ordered release of ligands in which the rate-limiting release of the first component is followed by rapid release of the other components cannot be ruled out from these data. Release of L-methionine-S-sulfoximine phosphate, ADP, and Mn2+ is a first order process. Reactivation follows exactly the same first order kinetics as dissociation of the complex, as would be expected if dissociation of the complex were the rate-limiting step in reactivation. Upon transfer of the enzyme from the reactivation mixture to assay solution there is no discernible lag in enzymatic activity (0.5-30 min). Either the enzyme is in a fully active conformation following dissociation of the complex at low pH or any conformational changes that occur are rapid compared to the rate of dissociation.
Reversibility of inactivation is clearly shown by the release of Mn2+ and the products of the auto-inactivation reaction from glutamine synthetase during reactivation at pH t4.6 and by the rapid rebinding of these components to the enzyme during reinactivation at pH 7 (Scheme 1). With the E. coli enzyme, the binding of stoichiometric amounts of L-methionine-S-sulfoximine phosphate alone to the manganese enzyme (0.2 p~ subunit) is sufficient to inactivate the enzyme, which indicates that this transition state analog binds with very high affinity. These results are in agreement with those of Meister et al. (9)(10)(11) who showed that, at high concentrations, chemically synthesized L-methionine-S-sulfoximine phosphate inactivates sheep brain glutamine synthetase and that the apparent rate of inactivation is increased by added ADP and Mg2' (10). The ability of very low concentrations of ADP to bind to the manganase unadenylylated enzyme in the presence of L-methionine-S-sulfoximine phosphate at pH 7.2 reflects enormous synergism in the binding of these compounds to the enzyme. Measurements of the exchange of [ 14C]ADP from the inactive complex in the presence of excess unlabeled ADP gave a maximum rate constant for dissociation of ADP at pH 7.2 and 25 "C of s-'. This rate constant is 5-6 orders of magnitude lower than that for ADP dissociation from manganese unadenylylated glutamine synthetase (26). Assuming no change in the rate of binding in the absence and presence of L-methionine-S-sulfoximine phosphate and using K A = 3.5 The fact that the rate of reactivation of inactive Complex is a simple function of pH, temperature, or KC1 concentration indicates that reactivation proceeds in a single well defined mechanism under a variety of conditions. The pH dependence of reactivation indicates that protonation of 3-4 acidic groups with pK, -3.5 is required to weaken binding of the components of the complex. The groups are probably carboxylate ions on the protein and may be involved in chelating one of both metal ions. Calorimetric evidence of Hunt et al. (27) suggests that carboxylic acid groups are at the nl Mn2+ ( M P ) binding sites of glutamine synthetase. Furthermore, the affiity constants for binding Mn" to nl and n2 sites are strongly dependent on pH (25,28) with 2 and 1 eq of protons released/ Mn2+ bound to nl and n2 sites at pH 7.2, respectively (28). our results also indicate that the metal ions play a key role in Stabilizing the inactive enzyme complex. Since both the manganese and magnesium inactive complexes of the unadenylylated enzyme have the same pH dependence of reactivation, the protonation of equal numbers of carboxylate groups are apparently involved in disruption of these complexes. Also, both adenylylated and unadenylylated enzymes are reactivated with the same pH dependence, despite the difference in the conformations of these enzyme forms induced by L-methionine-S-sulfoximine binding (29). Although the protonation of carboxylate groups within metal ion binding clusters could be the rate-limiting step in reactivation of the inactive enzyme complexes, protonation of one or more functional groups on the ligands or on the protein outside of the active site could also be involved.
It is likely that both high KC1 and high temperatures accelerate reactivation by disrupting protein conformations that favor the stability of the complex. The much smaller rate enhancement produced by K&04 (one-tenth that of KC1) may reflect the counteracting influence of sulfate which is known to promote "ordering" of proteins in solutions (30). The break in the Arrhenius curve for reactivation of manganese inactive complex without KC1 indicates the melting out of a particular protein conformation that contributes to the stability of the complex. Since no thermal transition was observed in the presence of 1 M KC1, it appears that high concentrations of KC1 eliminate this stabilizing conformation. The smaller effect of KC1 on the rate of reactivation of magnesium inactive complex suggests the absence of a similar conformation in the magnesium inactive complex, for which no break in the Arrhenius activation curve is seen. The above considerations indicate that favorable conformational changes of the protein make important contributions to the free energy of binding of the ligands in the inactive complex. The considerable resistance to dissociation and unfolding of the protein polypeptide chains upon formation of the inactive complex, described e l s e~h e r e ,~ show that the inactivating ligands markedly stabilize secondary, tertiary, and quaternary structures of the dodecamer.
The rate of auto-inactivation of glutamine synthetase decreases in a non-first order manner during inactivation ( Fig.  6 and Refs. 14 and 15). Rhee et al. (15) have suggested that the non-first order kinetics of inactivation of magnesium unadenylylated enzyme are caused by negative interactions between inactive and active subunits within the dodecamer. Spectrophotometric and fluorescence titrations (15, 29) have provided direct evidence for negative cooperativity in the reversible binding of L-methionine-S-sulfoximine to unadenylylated enzyme. Our findings that nucleotide-metal ion ratios and the order of addition of reactants drastically affect the rates and extents of inactivation (Fig. 6) indicates a more M. R. Maurizi and A. Ginsburg, submitted for publication. complex kinetic mechanism for the auto-inactivation reaction.
Also, reinactivation at neutral pH was inhibited by excess Mn -ADP (Fig. 4). Those data are consistent with a model in which ATP or metal ion-ATP binds to and partially stabilizes a form of glutamine synthetase that catalyzes the phosphorylation of L-methionine-S-sulfoximine either slowly or not at all. A ratelimiting interconversion between the resistant and susceptible conformers would result in non-first order kinetics of inactivation. Likewise, excess Mn. ADP may stabilize a form of the enzyme that has decreased ability to bind L-methionine-Ssulfoximine phosphate. It is probable that homologous subunit interactions produced by L-methionine-S-sdfoximine binding and the heterogeneity induced by nucleotide binding both contribute to the complex kinetics of the formation of the inactive enzyme complex.
Different apparent stoichiometries for inactive complex formation in the completely inactive enzyme were obtained by titration with limiting amounts of reactants (IO-ll/dodecamer) or by radioactive labeling with excess reactants (11-12/ dodecamer). These results imply that y-glutamyltransferase activity of all subunits is inhibited when 10-11 subunits of the dodecamer are inactivated, but phosphorylation of L-methionine-S-sulfoximine may occur on the remaining 1-2 subunits at a reduced rate. This possibility is not unreasonable since phosphorylation of L-methionine-S-sulfoximine is readily catalyzed by the magnesium adenylylated enzyme which is inactive in both the biosynthetic and the y-glutamyltransferase reactions. Despite the differences in stoichiometries obtained by the two methods, the amount of enzymatic activity is proportional to the amount of inactive complex present throughout the entire saturation range measured either by inactivation with limiting reactants (Fig. 5) or by dissociation of the inactive complex at low pH (Fig. 1). Enzyme heterogeneity and subunit interactions are often invoked to explain less than 1:l stoichiometry of ligand binding to subunits of oligomeric enzymes. However, in this case, such effects would have to occur equally at all extents of saturations, suggesting that an alternative explanation or additional considerations are required.
In summary, the very tight, synergistic binding of 4 components containing a transition state analog to the active site of each subunit of dodecameric glutamine synthetase has been demonstrated. The conditions for reversibly binding these ligands give some insight into the role that protein structure has during catalysis. Additionally, recent studies3 of the stability of the inactive enzyme complex have shown that both intra-and intersubunit bonding domains are strengthened by the presence of the inactivating ligands bound at the active site. Further studies should help characterize the physical interactions between liganded and unliganded subunits of glutamine synthetase and possibly clarify the effect of these interactions on catalysis by the enzyme.