Functional Linkage between the Glutaminase and Synthetase Domains of Carbamoyl-phosphate Synthetase

Mammalian carbamoyl-phosphate synthetase is part of carbamoyl-phosphate synthetase-aspartate carbamoyltransferase-dihydroorotase (CAD), a multifunctional protein that also catalyzes the second and third steps of pyrimidine biosynthesis. Carbamoyl phosphate synthesis requires the concerted action of the glutaminase (GLN) and carbamoyl-phosphate synthetase domains of CAD. There is a functional linkage between these domains such that glutamine hydrolysis on the GLN domain does not occur at a significant rate unless ATP and HCO3 −, the other substrates needed for carbamoyl phosphate synthesis, bind to the synthetase domain. The GLN domain consists of catalytic and attenuation subdomains. In the separately cloned GLN domain, the catalytic subdomain is down-regulated by interactions with the attenuation domain, a process thought to be part of the functional linkage. Replacement of Ser44 in the GLN attenuation domain with alanine increases thek cat/K m for glutamine hydrolysis 680-fold. The formation of a functional hybrid between the mammalian Ser44 GLN domain and the Escherichia coli carbamoyl-phosphate synthetase large subunit had little effect on glutamine hydrolysis. In contrast, ATP and HCO3 − did not stimulate the glutaminase activity, indicating that the interdomain linkage had been disrupted. In accord with this interpretation, the rate of glutamine hydrolysis and carbamoyl phosphate synthesis were no longer coordinated. Approximately 3 times more glutamine was hydrolyzed by the Ser44 → Ala mutant than that needed for carbamoyl phosphate synthesis. Ser44, the only attenuation subdomain residue that extends into the GLN active site, appears to be an integral component of the regulatory circuit that phases glutamine hydrolysis and carbamoyl phosphate synthesis.

Mammalian carbamoyl-phosphate synthetase is part of carbamoyl-phosphate synthetase-aspartate carbamoyltransferase-dihydroorotase (CAD), a multifunctional protein that also catalyzes the second and third steps of pyrimidine biosynthesis. Carbamoyl phosphate synthesis requires the concerted action of the glutaminase (GLN) and carbamoyl-phosphate synthetase domains of CAD. There is a functional linkage between these domains such that glutamine hydrolysis on the GLN domain does not occur at a significant rate unless ATP and HCO 3 ؊ , the other substrates needed for carbamoyl phosphate synthesis, bind to the synthetase domain. The GLN domain consists of catalytic and attenuation subdomains. In the separately cloned GLN domain, the catalytic subdomain is down-regulated by interactions with the attenuation domain, a process thought to be part of the functional linkage. Replacement of Ser 44 in the GLN attenuation domain with alanine increases the k cat /K m for glutamine hydrolysis 680-fold. The formation of a functional hybrid between the mammalian Ser 44 GLN domain and the Escherichia coli carbamoyl-phosphate synthetase large subunit had little effect on glutamine hydrolysis. In contrast, ATP and HCO 3 ؊ did not stimulate the glutaminase activity, indicating that the interdomain linkage had been disrupted. In accord with this interpretation, the rate of glutamine hydrolysis and carbamoyl phosphate synthesis were no longer coordinated. Approximately 3 times more glutamine was hydrolyzed by the Ser 44 3 Ala mutant than that needed for carbamoyl phosphate synthesis. Ser 44 , the only attenuation subdomain residue that extends into the GLN active site, appears to be an integral component of the regulatory circuit that phases glutamine hydrolysis and carbamoyl phosphate synthesis.
In mammals and most other species, the synthesis of carbamoyl phosphate in the de novo pyrimidine biosynthetic pathway occurs in a series of four partial reactions (1,2) that are catalyzed by distinct structural domains of carbamoyl-phosphate synthetase (CPSase, 1 EC 6.3.5.5). Mammalian carbamoyl-phosphate synthetase is part of a large multifunctional protein called CAD (3)(4)(5), which also has aspartate transcarbamoylase and dihydroorotase activities, enzymes that catalyze the second and third steps of the de novo pathway, respectively. The 243-kDa CAD polypeptide is organized into discrete structural domains each with a specific function (6 -9). The 38-kDa glutaminase (GLN) domain located on the amino end of the polypeptide (10, 11) catalyzes the hydrolysis of glutamine and transfers the ammonia to the adjacent 120-kDa synthetase (CPS) domain, where the remaining partial reactions take place. The CPS domain of CAD (10) and all known CPSases (12,13) consist of two homologous subdomains, CPS.A and CPS.B (10), that probably arose as a result of an ancestral gene duplication and fusion (12). Escherichia coli CPSase is a monofunctional protein (14,15) consisting of a 42-kDa GLN subunit and a 120-kDa CPS subunit. Despite differences in structural organization, the sequence and domain structure of the mammalian and bacterial proteins are very similar. There is extensive evidence that the two different ATP-dependent partial reactions, the activation of bicarbonate (Reaction 2) and the phosphorylation of carbamate (Reaction 4), are catalyzed by CPS.A and CPS.B, respectively (16 -22). The x-ray structure of the E. coli enzyme (23)(24)(25)(26) has been solved to a resolution of 1.8 Å. Remarkably, the active sites were found to be widely separated but connected by a narrow tunnel that passes through the interior of the molecule. The ammonia generated by hydrolysis of glutamine presumably diffuses through the tunnel to the active site of CPS.A, where it reacts with carboxy phosphate to form carbamate. The carbamate then diffuses through the tunnel to the active site of CPS.B, where carbamoyl phosphate is formed in the second ATP-dependent reaction.
The mechanism of glutamine hydrolysis by CPSase (27)(28)(29)(30)(31) and other trpG-type amidotransferases (32)(33)(34) is analogous to that of the thiol proteases. The reaction proceeds through a thioester intermediate, and there is evidence that a catalytic triad consisting of Cys 252 , His 336 , and Glu 338 in CAD promotes catalysis. The thioester was directly observed in the x-ray structure (24) of an E. coli CPSase mutant. The residues of the catalytic triad have the expected juxtaposition relative to the bound intermediate and in addition, Ser 47 (Ser 44 in CAD) was found to be close to the carbonyl carbon of the thioester carboxamide group. The authors suggested that this serine residue may participate in catalysis by positioning the carbonyl group for nucleophilic attack by the active site cysteine and by stabilizing the developing oxyanion in the transition state.
The partial reactions are coordinated by a reciprocal linkage between glutamine hydrolysis and carbamoyl phosphate synthesis. While the GLN and CPS domains of both E. coli and mammalian CPSase can function autonomously, many studies (1,(27)(28)(29)(30)(31)(35)(36)(37)(38) have demonstrated a functional linkage between the active sites that modulates their activities. Glutamine hydrolysis does not proceed at a significant rate in the absence of ATP and bicarbonate. This functional linkage minimizes the hydrolysis of glutamine when the other substrates needed for carbamoyl phosphate synthesis are limiting.
We have previously cloned and expressed the mammalian GLN domain (39) and showed that the purified recombinant protein forms a stable, stoichiometric complex with the E. coli CPS subunit. The hybrid molecule has kinetic parameters that are similar to those of the GLN domain of intact CAD, and the linkage between the GLN and CPS domains is fully functional (31,39). We have used this hybrid molecule to examine the role of Ser 44 and conclude that it is not a catalytic residue in the usual sense, but rather is a crucial element in the functional linkage that coordinates the reactions occurring on the GLN and CPS domains. Plasmids and Strains-The 7.1-kilobase plasmid pHGGLN52 (39) carries a 1.1-kilobase insert encoding the mammalian CAD GLN domain in a vector derived from pEK81 (40). The expression of the protein is under control of the pyrBI promoter. The E. coli host strain, EK1104 (40), lacks the pyrB and pyrI genes and has a leaky pyrF mutation. The high copy number plasmid PHN12 (41), which carries the carB gene that encodes the large subunit of E. coli CPSase, was kindly provided by Dr. Carol Lusty (The Public Health Research Institute of the City of New York, New York, NY) as was the E. coli strain L673, which is defective in the carA and carB genes, encoding both E. coli carbamoylphosphate synthetase subunits, as well as the Lon protease.
Cell Growth and Recombinant DNA Methods-Cells harboring the recombinant plasmids were routinely grown from a single colony in 2ϫ YT medium supplemented with 50 -100 mg/liter ampicillin. The expression of the mammalian GLN domain in EK1104 transformants was induced as described previously (39), while the E. coli CPS subunit is expressed constitutively (42) in L673 cells transformed with pHN12, a plasmid encoding the E. coli CPSase large subunit. Plasmids were isolated using the Bio101 RPM kit. Transformation and preparation of competent E. coli cells were carried out by the Hanahan procedure (43). Site-directed mutagenesis was carried out by PCR using the Stratagene QuickChange TM site-directed mutagenesis kit and PfuTurbo TM DNA polymerase, which replicates both plasmid strands with high fidelity. Oligonucleotide primers were obtained from Life Technologies, Inc. Custom Primers. Following PCR, the product was treated with DpnI endonuclease to digest the parental DNA template. The nicked vector DpnI-treated DNA was then transformed in E. coli Epicurian Coli XL1-Blue supercompetent cells. Colonies were grown and the DNA isolated and sequenced by the double-stranded dideoxy method at the Wayne State University Core Facility. The mutant plasmid was then transformed into the EK1104 strain. Restriction digests, ligations, and other DNA methods were carried out using standard protocols (44).
Protein Methods-CAD was isolated from an overproducing strain of Syrian hamster cells (BHK-128) as described previously (5,45). The E. coli CPSase large subunit was isolated from pHN12 transformants using the methods of Rubino et al. (42). The wild type CAD GLN domain and the mutant were isolated by the method described earlier (39). To form the hybrid CPSase (GLN-CPS), stoichiometric amounts of the wild type CAD GLN domain or its mutant and the E. coli CPSase large subunit were mixed together and incubated for 15 min (31,39). The complex was then concentrated using either a Speed-Vac at room temperature or by centrifugation using Millipore Ultrafree-MC 30,000 NMWL filter units at 4°C. Protein concentrations were determined by the Bradford dye binding method using bovine serum albumin as a standard (46). SDS-gel electrophoresis was carried out on 10% polyacrylamide gels (47).
Molecular Modeling-The structure of the CAD GLN-CPS domains was modeled with E. coli CPSase (Brookhaven Protein Data Bank; identification code 1JDB) serving as the tertiary template using the program Quanta version 4.0 (MSI). The alignment of CAD and E. coli CPSase sequences was carried out giving equal weight to secondary structure and sequence similarity. Undefined regions were regularized in two stages, 50 cycles of steepest descent, followed by 200 cycles of adopted basis set NR, prior to final energy minimization.
Enzyme Assays-The CPSase activity was assayed at 37°C using a radiometric procedure described previously (6, 48) using a 2 mM excess of MgCl 2 over the concentration of ATP in the assay. The GLNase activity (31) of the isolated CAD GLN domain and the mammalian-E. coli CPSase hybrid complex was measured by an assay that involved separating the reactant glutamine from the product glutamate by HPLC as described below. The assay buffer contained 25 mM HEPES, pH 7.4, 0.5 mM dithiothreitol, 0.05 mM EDTA, 25 mM KCl (with or without 10 mM ATP, 12 mM MgCl 2 , 15 mM NaHCO 3 ), and variable glutamine in a total volume of 0.35 ml. The reaction was initiated by adding 7-17 g of the GLN domain in 50 l, allowed to proceed for 1 h at 37°C, and then quenched with 100 l of 20% trichloroacetic acid. The samples stood on ice for 10 min prior to centrifugation for 5 min at 14,000 rpm to remove the precipitated protein. The supernatant (60 l) was then neutralized with 10.8 l of 1.2 M NaOH. For glutamine concentrations less than 10 mM, the samples were derivatized by directly adding 100 l of OPA reagent to the neutralized sample. At higher glutamine concentrations, 200 l of OPA reagent was added to 10 l of the neutralized protein solution. Exactly 90 s following the addition of OPA, a 100-l sample was injected onto the HPLC column. The activity is expressed as nanomoles/min/mg of the GLN domain, and the amounts assayed given in the legends represent micrograms of the GLN domain.
Gel Filtration-The formation of a complex between the isolated 38-kDa mammalian GLN domain and the mutant, with the 120-kDa E. coli CPSase synthetase subunit, was verified by gel filtration. The hybrid (45-50 g) in 0.1 M potassium phosphate, pH 7.6, 1 mM EDTA and 5% glycerol was applied to a 1.5 ϫ 63-cm Sephacryl S-300 column. The column was pre-equilibrated and eluted at 0.2 ml/min with the same buffer. Column fractions were analyzed by assaying CPSase and by SDS-gel electrophoresis. The GLN subunit consists of two subdomains, the catalytic (dark shading) and the attenuation subdomain (light shading). The diagram shows the location of the active site residues Cys 252 and His 336 , but Glu 338 located in close proximity to His 336 is partially obscured in this orientation and was omitted for clarity. Ser 44 , which is located within the attenuation subdomain, extends into the active site region of the catalytic subunit. A part of the interface between the GLN domain and part of the CPS domain (shown in black) is also visible. The structure was modeled based on the E. coli CPSase x-ray structure (23).

RESULTS
Mutation of Serine 44 -The mammalian GLN domain was modeled using the E. coli CPSase structure (23) as a template. As expected, given the strong sequence similarity of the mammalian and bacterial proteins, the configuration of active site residues closely resembles that observed for the E. coli enzyme. Ser 44 (corresponding to Ser 47 in the E. coli structure) is located within the catalytic site of the GLN domain. The side chain extends into the site ( Fig. 1) with its ␥-oxygen atom positioned within 4.2 Å of the Cys 252 sulfur atom and 4.9 Å from the ⑀ ring nitrogen of the catalytic histidine residue, His 336 .
Ser 44 was replaced with alanine by PCR-directed mutagenesis using the plasmid pHGGLN52 (39) as a template. High levels of the 38-kDa wild type and mutant domains were expressed when the plasmids were transformed into the E. coli strain EK1104. Both proteins were purified to homogeneity as described (39) previously. Gel filtration on a Sephacryl S-300 column showed that, as in the case of the wild type protein, a stoichiometric mixture of the mammalian Ser 44 3 Ala GLN domain and the E. coli CPSase CPS subunit formed a stable hybrid complex.
The Ser 44 3 Ala Mutation Activates the Isolated GLN Domain-The isolated GLN domain had very low catalytic activity (Table I), a consequence of a high K m (4 mM) and a very low k cat (0.020 s Ϫ1 ). The formation of a hybrid complex consisting of the mammalian GLN domain and the isolated E. coli CPS subunit restored the function of the GLN domain. The K m decreased 45-fold while the k cat increased 15-fold relative to the isolated domain. The steady state kinetic parameters of the hybrid are similar to those obtained for CAD.
Contrary to the results that would be expected if Ser 44 were a catalytic residue, the suppression of catalytic activity of the isolated domain was largely relieved by its replacement with alanine ( Fig. 2 and Table I). Compared with the isolated wild type domain, the K m was reduced 7-fold, while the k cat increased 21-fold. Thus, the mutation appreciably activated the isolated GLN domain, increasing the k cat /K m by a factor of 680.
The formation of the hybrid with the Ser 44 3 Ala GLN domain gave a species with kinetic parameters similar to those obtained for the wild type complex, although the k cat and V max values are about 1.7-fold higher. Compared with the isolated Ser 44 3 Ala GLN domain, the K m for glutamine was reduced another 6-fold in the mutant hybrid, but there was no appreciable change in k cat .
Thus, Ser 44 does not participate in glutamine hydrolysis in either the isolated domain or the hybrid in the absence of ATP and bicarbonate.
The Mutant Hybrid Protein Catalyzes the Synthesis of Carbamoyl Phosphate-The Ser 44 3 Ala hybrid protein can also catalyze the overall synthesis of carbamoyl phosphate. Saturation curves for the overall reaction (Fig. 3, A and B, and Table   II) show that the K m for both glutamine and ATP are very similar to the values determined for the wild type hybrid. However, the k cat values obtained from both glutamine (0.193 s Ϫ1 ) and ATP (0.164 s Ϫ1 ) are approximately 14-fold lower for the mutant protein.
ATP and Bicarbonate Do Not Activate the Glutaminase Reaction in the Mutant-The binding of ATP and bicarbonate to the CPS subunit stimulates the GLNase activity of the wild type hybrid and CAD 10-and 14-fold, respectively. The stimulation is the result of a corresponding increase in k cat , without any significant change in the affinity for glutamine. In the absence of ATP and bicarbonate (Fig. 4, Table I), the k cat for glutamine hydrolysis of the mutant hybrid (0.525 s Ϫ1 ) is only marginally increased compared with that of the wild type hybrid (0.309 s Ϫ1 ) and it does not significantly change in the presence of saturating concentrations of these substrates (0.552 s Ϫ1 ). Thus, the functional linkage that coordinates the reactions on the GLN and CPS domains is lost as a result of replacement of Ser 44 with alanine.
Coordination between the Activation of Bicarbonate and Glutamine Hydrolysis Is Lost in the Mutant-In CAD and in the wild type hybrid complex, the rate of glutamine hydrolysis is closely matched to the rate of carbamoyl phosphate synthesis. The maximum velocity for glutamine hydrolysis in the presence of saturating ATP and bicarbonate (Table I, k cat ϭ 3.11 s Ϫ1 ) parallels the overall rate of carbamoyl phosphate synthesis (Table II,   a replot (Fig. 3C) of the velocity data of carbamoyl phosphate synthesized (Fig. 3A) versus glutamine hydrolyzed (Fig. 4) measured at various concentrations of glutamine gives a slope of 1.15 Ϯ 0.04, in accord with the expected stoichiometry of the overall reaction at all concentrations of the substrate. In contrast, the rate of glutamine hydrolysis exceeds the rate of carbamoyl phosphate synthesized in the mutant hybrid. The k cat for the GLNase reaction measured at saturating ATP and bicarbonate is 0.552 s Ϫ1 , whereas the k cat for the CPSase reaction is only 0.193 s Ϫ1 . Similarly, the plot of glutamine hydrolysis versus carbamoyl phosphate synthesized had a slope of 2.9 Ϯ 0.1, indicating that the 1:1 stoichiometry observed for the reaction catalyzed by the wild type enzyme is not sustained in the mutant. This result would be expected if Ser 44 plays a major role in phasing the reactions occurring on the two domains. DISCUSSION Carbamoyl phosphate synthesis involves the concerted action of two domains that must act in synchrony. We are especially interested in the mechanism of glutamine hydrolysis and the reciprocal interactions between the GLN and CPS domains of the mammalian multifunctional protein CAD. Although we have expressed each of the functional domains, we cannot as yet obtain sufficient amounts of CPS needed for the sorts of studies described here. However, previous steady state and presteady state kinetic studies (31,39) showed that the hybrid consisting of the isolated mammalian GLN domain and the E. coli CPS subunit has catalytic parameters similar to CAD as well as a functional interdomain linkage. Thus, we have used this system to assess the role of Ser 44 in carbamoyl phosphate synthesis.
Whereas the hydrolysis of glutamine by CAD can be easily measured, the isolated GLN domain has barely detectable activity as a result of an appreciable increase in the K m for glutamine and decrease in k cat . However, the kinetic parameters are restored to the normal values found in CAD when a stoichiometric complex is formed by the non-covalent association of the isolated GLN domain and the CPS subunit of E. coli CPSase. Sequence comparisons (10,11) showed that the 40-kDa CPSase GLN domain consists of two distinct regions. The carboxyl half of the domain is homologous to the amidotransferase domain of all trpG or triad type amidotransferases (33,34,49), while the amino half of the domain is unique to the CPSases. Since amidotransferase domains of the other biosynthetic enzymes have an average molecular mass of 20 kDa and do not have a chain segment corresponding to the amino half of the CPS GLN domain, it was reasonable to assume that all of the residues required for glutamine binding and catalysis would be found in the carboxyl half of the CAD GLN domain. Support for this interpretation was obtained when we cloned and expressed (50) the two halves of the mammalian CPSase GLN domain and showed that they are autonomously folded subdomains with specific functions. The amino half has no catalytic activity but forms a stable complex with the CPS domain. The carboxyl half, the catalytic subdomain, binds glutamine with the same affinity and is more active (k cat ϭ 5.7 s Ϫ1 ) than the maximum glutaminase activity observed for intact CAD, even when assayed in the presence of saturating ATP and HCO 3 Ϫ (k cat ϭ 1.9 s Ϫ1 ). We therefore suggested that a major function of the amino half of the GLN domain, the attenuation subdomain, was to modulate the intrinsically high catalytic activity of the catalytic subdomain and that this region of the molecule was instrumental in relaying the interdomain signal between the GLN and CPS domains. In accord with this hypothesis, the x-ray structure of E. coli CPSase (23) subsequently showed that the attenuation subdomain makes exten- FIG. 3. Carbamoyl phosphate synthesis by the wild type and mutant hybrids. The carbamoyl-phosphate synthetase activity of the wild type (q) and mutant hybrids (E) was assayed by the radiometric method described under "Experimental Procedures." Panel A, CPSase activity of 7 g of the wild type and 12 g of the mutant hybrid measured versus glutamine concentration. Panel B, CPSase activity of 9.8 g of the wild type and 12 g of the mutant hybrid measured versus ATP concentration. Panel C, a replot of the data for the rate of glutamine hydrolysis in the presence of ATP and bicarbonate (see Fig. 4) against the rate of carbamoyl phosphate synthesis (A) measured at variable glutamine concentration. The slope for the wild type (q) and mutant (E) proteins represents the amount of glutamine hydrolyzed for each mole of carbamoyl phosphate synthesized. sive contacts with the CPS subunit in E. coli CPSase. The x-ray structure also showed that Ser 44 is the only residue in the attenuation subdomain that extends into the GLN active site, making it a prime candidate for participation in the functional linkage.
The activation of the GLN domain that occurs upon association with the CPSase subunit is almost entirely mimicked by the replacement of Ser 44 with alanine in the isolated domain. The k cat increases 20-fold to 0.43 s Ϫ1 compared with a value of 0.31 s Ϫ1 for the wild type hybrid complex. The high activity of the isolated Ser 44 3 Ala GLN domain confirms that this serine residue is not involved in catalysis. The K m also decreases to 0.6 mM in the isolated Ser 44 3 Ala domain, but is still 6-fold higher than that of the wild type hybrid complex. One could argue that this residue is important for binding glutamine, but, when the Ser 44 3 Ala GLN domain associates with the E. coli CPS subunit, the K m is virtually the same as the wild type hybrid complex. Thus, it seems clear that the suppression of activity in the isolated GLN domain is, to a large extent, the result of the serine residue. Since alanine is nearly isosteric with serine, the depression of activity is unlikely to be a consequence of steric interference, suggesting that Ser 44 may form a hydrogen bond with one of the active site residues that interferes with its normal function in catalysis.
When the wild type isolated domain combines with the CPS subunit, Ser 44 is likely to be displaced to its position located in the electron density maps of E. coli CPSase (23). Since the serine residue has been replaced in the mutant, little change would be expected when the Ser 44 3 Ala domain combines with the CPS subunit. Consistent with this interpretation, the kinetic parameters for glutamine hydrolysis by the Ser 44 3 Ala hybrid are similar to the values measured for the wild type hybrid complex, suggesting that Ser 44 is not required for glutamine hydrolysis by the hybrid in the absence of ATP and HCO 3 Ϫ . The hydrolysis of glutamine and the activation of bicarbonate are parallel reactions that must occur in phase to avoid the wasteful hydrolysis of glutamine or ATP that would otherwise occur. This requirement is especially important since the active sites are far from one another, about 45 Å in the E. coli structure (23), and ammonia must be delivered to the active site of CPS.A via a long interdomain tunnel. The coordination of these partial reactions requires that the GLNase activity be modulated so that it is not operating at or near its full catalytic potential unless the concentration of the other substrates needed for carbamoyl phosphate synthesis are saturating. A part of the interdomain functional linkage is the down-regulation of the glutaminase activity when ATP and bicarbonate are limiting. Steady state and presteady state kinetic studies of CAD (28,31) showed that, in the absence of ATP and bicarbonate, the thioester intermediate accumulates and the rate constant for the breakdown of the thioester intermediate is the same as the k cat for glutamine hydrolysis, indicating that it is the rate-limiting step. When ATP and bicarbonate are present, the breakdown of the thioester is accelerated, the intermediate cannot be detected, and the k cat for the hydrolysis of glutamine increases 14-fold. Since the substrate induced acceleration of catalysis is due to an increase in k cat without an apparent change on glutamine binding, it is likely that the juxtaposition of catalytic residues is suboptimal in the absence of ATP and bicarbonate. Similar results have been reported (29, 30) for E. coli CPSase. The functional linkage is abolished in the Ser 44 3 Ala hybrid. The addition of ATP and bicarbonate has no significant effect on either the K m or the k cat of the hybrid, suggesting that Ser 44 is essential for transmission of the allosteric signal that up-regulates the GLN domain. Since activation primarily involves an increase in the rate of breakdown of the thioester, it is possible that Ser 44 promotes its hydrolysis, but only when ATP and bicarbonate are bound to the CPS domain.
If this functional linkage is important in phasing the reactions occurring on the GLN and CPS domains, then the rate of glutamine hydrolysis need no longer match the rate of carbamoyl phosphate synthesis if the linkage is disrupted. The stoichiometry of carbamoyl phosphate synthesis is tightly controlled in the wild type hybrid, with 1 mol of glutamine hydrolyzed/mol of carbamoyl phosphate synthesized. In contrast, the mutant hybrid hydrolyzed 3 times more glutamine than that needed for carbamoyl phosphate synthesis, with the excess presumably leaking out of the complex.
We conclude that serine 44 in the GLN attenuation domain is not a catalytic residue in the usual sense but rather is an a The ATP saturation curves are sigmoidal and were fit to the Hill equation, so these represent apparent K m values. The Hill coefficient is 2.08 Ϯ 0.11 and 2.15 Ϯ 0.15 for the wild type and mutant proteins, respectively. The glutamine saturation curves were fit to the Michaelis-Menten equation.

FIG. 4. Glutaminase of the GLN-CPS hybrid in the presence and absence of saturating ATP and bicarbonate.
A hybrid complex of the mammalian GLN domain and the E. coli CPS subunit was formed as described under "Experimental Procedures." The hybrid with the wild type GLN domain (7 g of the GLN domain) was assayed in the presence of 10 mM ATP and 15 mM bicarbonate (q) and in the absence of these substrates (f). Similarly, the Ser 44 3 Ala hybrid complex (12 g of the GLN domain) was also assayed in the presence (E) and absence of ATP and bicarbonate (Ⅺ). The solid lines represent a least squares fit to the Michaelis-Menten equation. essential component in the regulatory linkage that phases glutamine hydrolysis and carbamoyl phosphate synthesis.