Specificity determining residues in ammonia- and glutamine-dependent carbamoyl phosphate synthetases.

Carbamoyl phosphate synthetases (CPSs) utilize either glutamine or ammonia for the ATP-dependent generation of carbamoyl phosphate. In glutamine-utilizing CPSs (e.g. the single Escherichia coli CPS and mammalian CPS II), the hydrolysis of glutamine to yield ammonia is catalyzed at a triad-type glutamine amidotransferase domain. Non-glutamine-utilizing CPSs (e.g. rat and human CPS I), lacking the catalytic cysteine residue, can generate carbamoyl phosphate only in the presence of free ammonia. Frog CPS I (fCPS I), unlike mammalian CPS Is, retains most of the glutamine amidotransferase residues conserved in glutamine-utilizing CPSs, including an intact catalytic triad, and could therefore be expected to use glutamine. Our work with native fCPS I provides the first demonstration of the inability of this enzyme to bind/utilize glutamine. To determine why fCPS I is unable to utilize glutamine, we compared sequences of glutamine-using and non-glutamine-using CPSs to identify residues that are present or conservatively substituted in all glutamine-utilizing CPSs but absent in fCPS I. We constructed the site-directed mutants Q273E, L270K, Q273E/N240S, and Q273E/L270K in E. coli CPS and have determined that simultaneous occurrence of the two substitutions, Gln-->Glu and Leu-->Lys, found in the frog CPS I glutamine amidotransferase domain are sufficient to eliminate glutamine utilization by the E. coli enzyme.

Formation of carbamoyl phosphate (CP) 1 is the first step in the synthesis of arginine and pyrimidines and, in higher organisms, is also critical in nitric oxide formation and in ammonia detoxification via the urea cycle. Carbamoyl phosphate synthetases (CPSs) are made up of a two-domain glutamine amidotransferase (GAT) component and a four-domain synthetase (SYN) component (1)(2)(3)(4)(5). The N-terminal domain of the GAT component is required for interaction with the SYN component (6,7), and the C-terminal domain has a triad-type GAT structure (5,8). In glutamine-utilizing CPSs (e.g. the single Escherichia coli CPS and mammalian CPS II), the hydrolysis of glutamine to yield ammonia is catalyzed at the GAT domain, where the cysteine of the Cys-His-Glu triad carries out a nucleophilic attack on the amide carbonyl of glutamine (9 -12). The ammonia thus liberated is subsequently channeled to the SYN component where it, together with bicarbonate, is used to generate CP with an energy expenditure of two ATP molecules per CP synthesized. The SYN component consists of two ATPutilizing domains (13), one allosteric interaction domain (14 -17), and one domain for interaction with the GAT component (6,7). In the absence of other CPS substrates, the GAT domain is able to catalyze a glutaminase reaction, and in the absence of glutamine, the SYN component is able to synthesize CP from NH 3 (18 -20). However, each component can sense occupancy of the other, and when all substrates are present, the glutaminase activity is increased 600-fold, and the ATPase activity is increased 15-fold. The final rates of the partial reactions involved in glutamine-dependent CP synthesis are matched so that the cleavage of glutamine is coupled to the cleavage of ATP, and there is no futile cycling of substrates (21,22).
Non-glutamine-utilizing CPSs (e.g. mammalian CPS I) apparently share the two-domain GAT plus the four-domain SYN structure (1,3,23) but are only able to generate CP in the presence of free ammonia (24). In the case of the rat and human CPS I, this inability to utilize glutamine is explained by the absence of the critical triad cysteine residue (25,26). Interestingly, however, frog CPS I retains an intact catalytic triad (27). It is also noteworthy that, in all cases studied, CPS I has an ϳ100-fold enhancement of NH 3 binding relative to that of glutamine-utilizing CPSs. The CPS I K m for NH 3 of ϳ40 M (28) allows that enzyme to function efficiently in the detoxification of NH 3 , whereas glutamine-utilizing CPSs (including CPS IIs from ureotelic organisms) are unable to perform this critical physiological role.
The solved triad GAT structures for E. coli CPS (5) and GMP synthetase (8,29) reveal a nearly identical topology of ␤-sheet flanked by ␣-helices and a ␤-ribbon. Surprisingly, both structures also display a fairly unformed glutamine specificity pocket (8). Apart from the essential triad cysteine 269, the only E. coli CPS side chain groups that have been identified as interacting with glutamine are serine 47 and glutamine 273. S47A and Q273A have 13-and 20-fold increased K m values, respectively, but unchanged rates of glutamine utilization (22). It has been assumed that conformational changes must occur to yield the known tight specificity of triad GATs for glutamine, but no crystals have yet been obtained with alternate conformations (8). In the present work, we have utilized site-directed mutagenesis to further probe the molecular basis for the glutamine specificity inherent in E. coli CPS, for the unexpected inability of frog CPS I to utilize glutamine, and for the dramatic differences in ammonia binding among CPSs. We have targeted for analysis those residues that are conserved in the GAT domains of glutamine-utilizing CPSs but that are not present in frog CPS I. purchased from PerkinElmer Life Sciences. Preformulated bacterial growth media (L-Broth, Terrific Broth II) were purchased from Qbiogene.
Strains and Plasmids-XLI-Blue E. coli cells were purchased from Stratagene and used for transformation and propagation of plasmid DNA. E. coli strain L673, which lacks both CPS subunits (GAT and SYN components) and is defective in the Lon protease, was kindly provided by Dr. Carol Lusty (New York Public Health Research Institute (6)). The plasmid pUCABI (9606 base pairs), encoding both the GAT and SYN components of E. coli CPS (carA and carB, respectively) as well as ornithine transcarbamoylase (argI), was a generous gift of Dr. Mendel Tuchman (Washington Children's Hospital); in construction of this plasmid, expression of CPS was placed under control of the isopropyl-1-thio-␤-D-galactopyranoside-inducible trc promoter, the carA translational start site was changed from TTG to ATG, and a Met 3 Val substitution was made at amino acid position two (30). For the present study, pUCABI was modified such that it would carry only the carAB genes. The argI (ornithine transcarbamoylase) gene was excised from pUCABI by digestion of the plasmid DNA with BamHI and KpnI. The large fragment was gel-purified, treated with mung bean nuclease to create blunt ends, and recircularized to create pUCAB (8594 base pairs).
Recombinant DNA Methods-Bacterial transformations and recombinant DNA techniques were carried out as described in Sambrook et al. (31). Restriction enzymes, nucleases, and ligase were obtained from New England Biolabs. Recombinant Pfu DNA polymerase was from Stratagene. Site-directed mutants were generated using the QuikChange method (Stratagene). Each mutagenesis cassette was sequenced to verify that no undesired changes were incorporated into the nucleotide sequence. Oligonucleotide primers for mutagenesis and sequencing were synthesized at the Tufts University Core Lab Facility. Mutagenesis primers corresponding to the coding strand sequence (with mutated residues in bold) were as follows: Q273E 5Ј-GTCTCGG-TCATGAGCTCCTGGCGCTGG-3Ј; Q273E/L270K ,5Ј-GGTATTCGGC-ATCTGTAAAGGTCATGAGCTCCTGG-3Ј; Q273E/N240S, 5Ј-CATCT-TCCTCTCCTCGGGTCCTGGCGACC-3Ј; L270K, 5Ј-GGTATTCGGCA-TCTGTAAAGGTCATCAGCTCCTGG-3Ј. CPS Purification-Bullfrogs, Rana catesbiana, were obtained from the Lemberger Co. (Wisconsin). Native frog liver CPS I (fCPS I) was prepared as described previously via a protocol including cetyltrimethylammonium bromide and acetone precipitation steps and Affi-Gel blue (Bio-Rad) chromatography (32).
The purification protocol for recombinant wild type E. coli CPS (eCPS) and for its site-directed mutants was adapted from those previously described for native (33) and recombinant (19) E. coli CPS. E. coli strain L673 transformed with pUCAB was grown to stationary phase at 37°C in Terrific Broth containing 100 mg/liter ampicillin, then diluted 100-fold in the same medium and grown to an optical density of about 0.6 at 600 nm. The culture was induced for 4 h with 1 mM isopropyl-1thio-␤-D-galactopyranoside, harvested by centrifugation at 5,000 ϫ g for 10 min, and resuspended in 0.2 M potassium phosphate, 1 mM EDTA, 1 mM dithiothreitol, and 0.2 mM phenylmethylsulfonyl fluoride, pH 7.6. The cells were disrupted by sonication (six 30-s pulses, with cooling intervals between pules), and cell debris was removed by centrifugation at 16,000 ϫ g for 20 min. Cleared lysate (Ϸ100 ml) was applied to a Hi-Prep™ 16/10 DEAE column (Ä KTA fast protein liquid chromatography, Amersham Biosciences, Inc.) equilibrated in 0.1 M potassium phosphate and 1 mM EDTA, pH 7.6 (Buffer A). Bound protein was eluted from the column with a 0 -55% discontinuous gradient of 1 M KCl (0 -10% in 80 ml, 10 -55% in 120 ml) in Buffer A. The eCPS peak typically eluted at 250 mM KCl. eCPS-containing fractions were pooled and concentrated by the addition of solid ammonium sulfate to 75%. The protein was precipitated by centrifugation at 10,000 ϫ g for 25 min, and the pellet was resuspended in Buffer A in a total volume of 5 ml for application to a Hi-Load™ 16/60 Superdex 200 column. eCPS was eluted from this column in Buffer A and stored for further use.
The purity of all protein preparations was at least 95%, as assessed by Coomassie Blue staining of SDS-polyacrylamide (4 -15% linear gradient) gels (34). Protein concentration was determined by the dye binding assay of Bradford (35) or by A 280 (0.685/mg of eCPS) (19). Western blot analysis was carried out as previously described (36), except that the primary antibody used in the present studies was rabbit antipeptide polyclonal antibody prepared by Sigma Genosys against a peptide corresponding to residues 112-130 of the 42-kDa GAT subunit of eCPS.
Enzyme Assays and Data Analysis-CP synthesis was determined in a two-step assay by coupling the CPS reaction to that of ornithine transcarbamoylase and then quantitating the resulting citrulline. The reaction mixtures contained 50 mM HEPES, 100 mM KCl, 10 mM ATP, 20 mM MgCl 2 , 20 mM NaHCO 3 , 1 mM dithiothreitol, 5 mM ornithine, 0.2 units of ornithine transcarbamoylase, and either 300 mM NH 4 Cl or 10 mM glutamine and were initiated by the addition of CPS (37°C, pH 7.6). After varying times of incubation, citrulline was determined colorimetrically with diacetylmonoxime, as previously described (37).
ATPase activities were determined in a pyruvate kinase/lactate dehydrogenase coupled assay, essentially as previously described (18). The reaction mixtures contained variable ATP, 50 mM HEPES, 100 mM KCl, 20 mM MgSO 4 , 40 mM NaHCO 3 , 1 mM sodium phosphoenolpyruvate , 0.2 mM NADH, 18 units of pyruvate kinase and 24 units of lactate dehydrogenase (25°C, pH 7.6). 20 mM ornithine was included where indicated. After initiation by the addition of CPS, the reaction was monitored continuously at 340 nm, and the amount of ADP formed was calculated from the loss of NADH. To determine ammonium-dependent ATPase activity, 300 mM NH 4 Cl was included in the reaction mixture, and to determine glutamine-dependent ATPase activity, 10 mM glutamine was included. To determine bicarbonate-dependent ATPase activity, no ammonia source was added to the assay mixture.
Glutamine hydrolysis was determined as previously described (11) by coupling glutamate formation to the glutamate dehydrogenase-catalyzed reduction of 3-acetyl pyridine dinucleotide (⑀ 363 ϭ 8.3 mM Ϫ1 cm Ϫ1 ). The reaction mixtures contained variable amounts of glutamine, 100 mM HEPES, 10 mM ATP, 20 mM MgCl 2 , 40 mM NaHCO 3 , 100 mM KCl, 10 mM ornithine, 1 mM 3-acetylpyridine nucleotide, and 20 units of glutamate dehydrogenase (25°C, pH 7.6). After initiation by the addition of CPS, the reaction was monitored continuously at 363 nm, and the amount of glutamate formed was calculated from the formation of reduced 3-acetylpyridine nucleotide. For all routine assays involving fCPS I, 5 mM N-acetylglutamate was included in the reaction mix.
Kinetic data were collected on a Beckman DU 640 spectrophotometer and fit by nonlinear regression (Beckman Enzyme Mechanism software) to the equation v ϭ V max S/(K m ϩ S), where v is the initial velocity, V max is the maximal velocity, S is the substrate concentration, and K m is the Michaelis-Menten constant.
Glutamine Binding-Binding of glutamine to CPS was determined in a radiometric assay as previously described (19,38). 100 g of CPS was allowed to interact for 1 min at 25°C with 120 M [ 3 H]glutamine (0.39 Ci/mmol) in 143 mM potassium phosphate plus 4.8 mM EDTA, pH 6.8. Enzyme-bound [ 3 H]glutamine was separated from free [ 3 H]glutamine by centrifugation through NICK ® spin columns (Amersham Biosciences, Inc.). The column effluent was added to 10 ml of ScintiSafe Plus 50% (Fisher), and counts were recorded in a Beckman Model LS3801 scintillation counter.

Inability of Frog CPS I (fCPS I) to Utilize Glutamine-Al-
though early studies on rat and human CPS I established clearly that they were unable to utilize glutamine (24), fCPS I was simply assumed to also not utilize glutamine since its other determined properties were identical to those of the rat and human enzymes. However, the relatively recent determination of the fCPS I sequence revealed that it was the only CPS I that retained the entire GAT catalytic triad and raised the question of whether it might utilize glutamine as an aminating substrate (27). To answer this question, we have purified native fCPS I and determined its ability to utilize glutamine in synthesizing CP. The purified frog enzyme was fully active when free ammonia was the aminating substrate. In the presence of 30 mM NH 4 Cl, the specific activity of the enzyme was 0.8 mol of CP/min/mg. However, in the absence of ammonia and presence of glutamine (up to 135 mM), fCPS I had no detectable activity. We also tested glutamine in the absence of the allosteric activator N-acetylglutamate (39) and again found no detectable activity.
Because fCPS I proved unable to catalyze the overall synthesis of CP from glutamine, there could be an impairment of glutamine recognition/utilization by the CPS I GAT domain, and/or glutamine hydrolysis might have, by some other mechanism, become uncoupled from CP synthesis. To address the first possibility, we undertook experiments to examine both the glutamine-hydrolyzing and glutamine binding abilities of the native enzyme. As a positive control for these studies, we used the glutamine-utilizing eCPS. eCPS, like other glutamine-utilizing CPSs, is able to catalyze the hydrolysis of glutamine at the GAT domain independently of the overall CP synthesis reaction. However, the rate of this uncoupled partial reaction is extremely slow, with a k cat value of 0.25 min Ϫ1 (22). When both the GAT and SYN active sites are occupied (with glutamine and with ATP/bicarbonate, respectively), the k cat value for the coupled glutaminase increases to 2.9 s Ϫ1 (22). Our glutaminase assay conditions yielded a k cat for glutamate formation by eCPS very similar to that previously reported (1.5 s Ϫ1 in the presence of ATP/bicarbonate). However, despite varying assay conditions (10 -30 mM glutamine, with and without N-acetylglutamate, with and without ATP/bicarbonate), fCPS I did not display any detectable glutamine hydrolysis activity. We also carried out glutamine binding studies on fCPS I. As presented below, the binding study protocol yielded significant [ 3 H]glutamine binding for wild type eCPS and for several of its mutants. However, with fCPS I, no [ 3 H]glutamine binding was detected. Thus, despite the striking presence of an intact catalytic triad, fCPS I is unable to bind and utilize glutamine.
Identification of Residues Potentially Critical for GAT Domain Activity-Our initial studies of the native fCPS I together with its known sequence established that changes in non-triad residues must be responsible for its inability to utilize glutamine. Although it is possible that alterations in other portions of the fCPS I structure might alter the GAT structure and thereby be responsible for the inability to utilize glutamine, in this work we have focused on the GAT domain itself. To identify candidate residues that may be critical for the interaction of CPSs with glutamine, we analyzed an amino acid sequence alignment of the GAT domain of 23 CPSs. Of the aligned CPSs, 20 (including the single E. coli CPS, eCPS) can utilize glutamine, whereas CPS I from rat, frog, and human cannot utilize glutamine. Of the 200 residues within the eCPS GAT domain, 29 are invariant in the other 20 glutamine-utilizing CPSs (Fig.  1), including the 3 triad residues and 6 additional residues that are invariant in all triad GATs (8). We also included in the Fig.  1 analysis leucine 270, which is conservatively substituted by methionine in 3 of the glutamine-utilizing CPSs. Comparison of the sequences of glutamine-using and non-glutamine-using CPSs (Fig. 1) revealed only four residues that are present or conservatively replaced in all glutamine-utilizing CPSs and absent from fCPS I. These residues, numbered according to their positions in eCPS, are asparagine 240, leucine 270, glutamine 273, and lysine 285. We have excluded lysine 285 from at least the initial analysis since the fCPS I substitution of arginine is a very conservative one. However the other fCPS I substitutions might potentially cause significant alterations in the GAT active site, such as replacement of asparagine by serine at position 240, replacement of leucine by lysine at position 270, and replacement of glutamine by glutamate at position 273 (Fig. 2). A potential role for these residues in glutamine interaction is also consistent with their occurrence in the immediate vicinity of the glutamine binding pocket in the x-ray crystal structure of eCPS (12) (Fig. 3). A study that was published while the present work was under way showed that substituting alanine for glutamine 273 resulted in moderate impairment of glutamine binding, further suggesting that this residue could be critical for glutamine utilization (22). However, the published work did not include the glutamate substitution at residue 273 that was identified as being of potential significance in the present analysis. In the present study, we have constructed the mutants Q273E, L270K, Q273E/L270K, and Q273E/N240S in eCPS to mimic the potentially significant substitutions we observed in fCPS I and to allow determination of the effects of these changes on the glutamine binding/utilizing ability of CPS.
Preliminary Screening of eCPS Mutants-Initial analysis of CP synthesis activity was made using cleared lysates of wild type eCPS and its four mutants Q273E, L270K, Q273E/L270K, and Q273E/N240S (Table I). For all of the mutants, ammoniadependent CP synthesis activities were very similar to that of the wild type eCPS. This ability of the mutant enzymes to function with ammonia indicates that their structural integrity is maintained and strongly suggests that the mutations have not caused global misfolding. However, when CP synthesis assays were performed using glutamine as the nitrogen donor, the L270K/Q273E mutant synthesized CP at least 25-fold less well than the wild type enzyme. The two mutations appeared to exert a synergistic effect since the L270K mutant synthesized CP only 5-fold less well than wild type eCPS, whereas CP synthesis by Q273E was equivalent to that of the wild type enzyme. Because the glutamine-utilizing behavior of the Q273E/N240S mutant was also equivalent to that of the wild type enzyme, the effects of the N240S mutation were not analyzed further. This substitution is quite conservative in terms of side chain size and polarity, so its elimination in the preliminary screening was not surprising.  (Table II). To ensure specificity of binding, glutamine was supplied at a concentration close to its K m value for eCPS (120 M), and the incubation was for only 1 min (19). After removal of free [ 3 H]glutamine by centrifugal desalting (38,40), wild type eCPS exhibited robust binding (0.147 mol of [ 3 H]Gln/mol of CPS). However, both the Gln 3 Glu and Leu 3 Lys substitutions made in eCPS to mirror fCPS I residues significantly altered the ability of eCPS to bind this substrate. The Q273E eCPS mutant was Ͼ10-fold impaired in its glutamine binding ability as compared with the wild type enzyme, whereas in L270K as well as in the double mutant Q273E/L270K, glutamine binding was almost entirely abolished. The behavior of these mutants suggests that these two substitutions in the fCPS I GAT domain may contribute significantly to this enzyme's inability to use glutamine as a nitrogen donor despite its retention of the catalytic triad.
Glutamine Hydrolysis Activity of Wild Type and Mutant eCPSs-The glutamine-hydrolyzing activities of purified wild type and mutant eCPSs were determined as previously described (11) by following the glutamate dehydrogenase-mediated reduction of 3-acetylpyridine dinucleotide. As discussed above, wild type eCPS carries out the uncoupled glutaminase reaction at an extremely slow rate that would make difficult the accurate determination of even slower mutant rates. However, the glutaminase activity is increased 600-fold when both the GAT and SYN active sites are occupied (with glutamine and with ATP/bicarbonate, respectively). Therefore, rates of glutamine hydrolysis for wild type and mutant eCPS were determined under substrate-loaded conditions. Even under these assay conditions, the Q273E/L270K mutant exhibited undetectable glutamine turnover compared with 1.5 s Ϫ1 for wild type eCPS. The efficiency of glutamine hydrolysis by the Q273E mutant was impaired approximately 7-fold (1.3 s Ϫ1 mM Ϫ1 ), whereas that of the L270K mutant was more than 3 orders of magnitude worse (0.005 s Ϫ1 mM Ϫ1 ) than the wild type enzyme (9.81 s Ϫ1 m Ϫ1 ). Whereas Q273E exhibited a wild type like K m for glutamine (0.34 versus 0.15 mM), L270K had a more than 80-fold elevated K m for glutamine (13.21 mM).
CP Synthesis by Wild Type and Mutant eCPSs-Purified wild type eCPS and the three mutant enzymes Q273E, L270K, Q273E/L270K were examined for their abilities to synthesize CP when supplied with either glutamine or ammonia as a nitrogen source. When glutamine was supplied as the nitrogen donor (Table III), wild type eCPS and the Q273E mutant displayed equally robust function, synthesizing 7.28 and 7.44 mol of CP/mg of CPS, respectively, in 10 min, whereas the other single mutant, L270K, was significantly impaired (1.03 mol of CP/mg of CPS). The double mutant Q273E/L270K, like the fCPS I it was designed to emulate, was unable to utilize glutamine for synthesis of CP. All four enzymes were well able to generate CP when ammonia was supplied as a substrate and exhibited comparable rates of product formation ranging from 10.17 to 12.46 mol of CP/mg of CPS (Table III).
Parameters for Glutamine, Ammonia, ATP, and Allosteric Effector Interaction for Wild Type and Mutant eCPSs-We have utilized three assays to further define the effects of the mutations in the GAT domain. Both the GAT and SYN components are involved in glutamine-dependent CP formation. Only the SYN component is involved in ammonia-dependent CP formation (20). And only the first of the two SYN ATP sites is involved in the bicarbonate-dependent ATPase assay, which reflects ATP cleavage to form the intermediate carboxyphosphate. In the absence of an ammonia source, carboxyphosphate breaks down, creating a futile cycle that occurs at only 1% the rate of the coupled ATP utilization (41). As described under "Experimental Procedures," all three reactions were monitored by following the time course of ATP hydrolysis. Ornithine (the co-substrate with CP for the subsequent reaction of the arginine biosynthetic pathway) is an allosteric activator that binds at the C-terminal allosteric domain of eCPS (4,15). Ornithine primarily acts to decrease the K m at the second of the two ATP sites, although it also has a slight effect on the K m for the first ATP (21). We have carried out the three assays in the absence and presence of ornithine to further dissect the effects of the mutations.
Glutamine Utilization during CP Synthesis by Q273E and L270K-Compared with wild type eCPS, Q273E exhibited only a slight perturbation in its glutamine K m (0.36 versus 0.16 mM for eCPS), whereas L270K had an ϳ25-fold increased K m for this substrate (3.93 mM, Table IV). Both Q273E and L270K exhibited depressed turnover rates, ϳ4and 10-fold slower than the wild type enzyme, respectively (Table IV). L270K displayed even greater deterioration in efficiency of glutamine utilization, with k cat /K m more than 2 orders of magnitude worse than the value for wild type eCPS (0.27 versus 57.59 mM Ϫ1 s Ϫ1 ). As expected, given its previous characterization as an effector of K m,ATP , the addition of ornithine had no noteworthy effect on these parameters for the utilization of glutamine (Table IV).
Q273E/L270K had no detectable glutamine-dependent CP synthetic activity. The double mutant did display bicarbonatedependent ATPase activity, reflecting uncoupled nonproductive turnover at the first ATP site. However, the addition of glutamine at concentrations up to 20 mM did not cause significant changes in the rate of ATP hydrolysis, with k cat values of 0.70 and 0.78 s Ϫ1 in the presence and absence of glutamine, respectively (Table V). Interestingly, the bicarbonate-dependent ATPase activity for Q273E/L270K was much greater than wild type. Values for k cat in the absence and presence of ornithine were 0.062 and 0.073 s Ϫ1 for wild type and 0.78 and 2.11 s Ϫ1 for the double mutant (Table V). It thus appears that Q273E/L270K has undergone a relaxation of the structural constraints normally limiting ATP hydrolysis in the absence of an ammonia source. In contrast both L270K and Q273E display essentially wild type coupling behavior.
Ammonia Utilization during CP Synthesis by Q273E, L270K, and Q273E/L270K-Ammonia-dependent ATPase assays were performed to assess the functioning of the SYN component of the eCPS mutants. Because the mutations were all in the GAT component and the ammonia-dependent reaction occurs only on the SYN component, any effects on this activity should reflect conformational changes in the SYN component caused by interaction with the GAT mutants. The primary effect of the mutations was on the K m for ammonia, whereas the turnover numbers were similar to those of wild type eCPS (Table IV). Although uncharged NH 3 is known to be the actual CPS substrate (28), the data are presented in terms of NH 4 ϩ since it is the level of NH 4 Cl that is varied during the experiment; under the conditions of the experiment, NH 3 represents about 4% of the total NH 4 ϩ added to the solution (28). Surprisingly, Q273E displayed an ϳ6-fold increase in ammonium K m (615 versus 96.1 mM for eCPS). However, L270K had a slightly lower ammonium K m (64 mM) than wild type eCPS, and Q273E/L270K had a 4-fold lower ammonium K m (26 mM). An enhanced interaction with ammonia in the mutants is consistent with the very low fCPS I K m for ammonia. As expected, given its previous characterization as an effector of K m,ATP , the addition of ornithine had no noteworthy effect on the utilization of ammonia (Table IV).
ATP and Allosteric Effector Interaction in Q273E, L270K, and Q273/L270K-ATP utilization was probed in three assays as follows. Glutamine-dependent ATPase activity reflects cleavage at both ATP sites of the SYN component coupled to glutamine cleavage at the GAT domain, ammonia-dependent ATPase activity reflects cleavage at both sites of the SYN component, and bicarbonate-dependent ATPase activity reflects uncoupled nonproductive cleavage at only the first of the two SYN ATP sites. There were no striking effects of the mutations on any of the K m values for ATP. For Q273E in all three assays, the K m for ATP was similar to that for wild type eCPS (Table V). For L270K, the K m,ATP was decreased 4-fold in theglutamine-dependentassay,increased3-foldintheammoniadependent assay, and unchanged in the bicarbonate-dependent assay. As discussed above, the glutamine-dependent assay for Q273E/L270K reflected only the basal uncoupled ATP hydrolysis activity. The Q273E/L270K K m for ATP was increased 3-fold in the ammonia-dependent assay with values of 0.18 and 0.06 mM, respectively, for the double mutant and the wild type enzymes.
When glutamine-dependent CP synthesis was determined in the presence of the positive allosteric effector ornithine, wild type eCPS exhibited the expected 9-fold decrease in its K m for ATP (0.053 versus 0.444 mM) and a corresponding 9-fold increase in efficiency (Table V). L270K showed a similar ornithine response, with a 10-fold decrease in its K m for ATP, whereas Q273E showed only a 1.7-fold decrease in its K m for ATP (0.278 versus 0.465 mM).
In addition to serving as a substrate, ammonia is also a positive allosteric effector for eCPS. Like ornithine, ammonia acts to decrease the K m for ATP (21). However, the details of this interaction are not defined. Neither the allosterically active molecular species (ammonia and/or ammonium) nor the localization of the binding site are known. In contrast to its abnormally small response to ornithine, Q273E showed a wild type-like decrease in the K m for ATP when ammonia was present (Table V). For wild type eCPS, the K m for ATP in the ammonia-dependent ATPase reaction was 0.062 mM (versus 0.444 mM in the glutamine-dependent reaction), and for Q273E, the K m values were 0.092 and 0.465 mM. As an additional probe for the Q273E allosteric response, we tested the effect of UMP, which serves as a feedback inhibitor for eCPS. In the presence of 0.1 mM UMP, the Q273E mutant and eCPS exhibited similar decreases in glutamine-dependent ATPasespecific activity (8 and 3%, respectively, of the activity observed in the absence of UMP).

TABLE I CP synthesis by WT and mutant CPS IIs
CP synthesis from either glutamine or ammonia was determined using cleared lysates. eCPS concentrations were normalized by densitometric analysis of a Western blot. Assays were carried out at 37°C for 10 min as described under "Experimental Procedures." eCPS generated 0.11 and 0.14 mol of citrulline in the glutamine-and ammonia-dependent assays, respectively. These values were used to represent 100% activity. Corresponding activity values were calculated for each of the four mutants. H]Glutamine binding studies for purified WT eCPS and mutant CPSs were performed with non-saturating glutamine. Binding was for 1 min at 25°C using 100 g of CPS in 143 mM potassium phosphate buffer, pH 6.8, 4.8 mM EDTA, and 120 M [ 3 H]glutamine (100 l total volume). Enzyme-bound radioactivity was recovered in the effluent after centrifugation through pre-packed spin columns and was added to 10 ml of scintillation fluid to determine cpm. In addition to the no enzyme control, the SYN component of eCPS (120 kDa) was included as a control for non-specific binding of [  WT and mutant CPSs eCPS, Q273E, L270K, Q273E/L270K, and fCPS I were incubated for the indicated times at 37°C, pH 7.6, in 50 mM HEPES, 50 mM NaHCO 3 , 20 mM MgCl 2 , 100 mM KCl, 10 mM ATP in the presence of either NH 4 Cl (30 mM) or glutamine (10 mM). N-Acetylglutamate (5 mM) was added to the reaction mix for fCPS I. As described under "Experimental Procedures," 5 mM ornithine and 0.2 units of ornithine transcarbamoylase were included in all reaction mixes to convert the CP produced to citrulline, which could subsequently be colorimetrically quantitated. Note that all CP values were determined to be zero at 0 min. One major goal of the present work was to determine why frog CPS I is unable to utilize glutamine even though it retains the GAT catalytic triad and most of the GAT residues conserved in glutamine-utilizing CPSs. As a prelude to addressing this question, we used native fCPS I to demonstrate for the first time that the enzyme is unable to use glutamine, in keeping with the earlier findings for CPS I from rat and human tissue (24). It thus appears that the ability of CPS to bind ammonia tightly, as has been observed only for CPS I, is consistently accompanied by the inability to utilize glutamine. We have determined that simultaneous occurrence of two substitutions found in the fCPS I GAT domain, Q273E and L270K, are sufficient to eliminate glutamine utilization by E. coli CPS. The Q273E/L270K mutant showed negligible glutamine binding, no detectable glutamine hydrolysis, and no detectable glutaminedependent CP synthesis. In striking contrast, the double mutant had an unchanged k cat for ammonia-dependent CP synthesis relative to that for wild type eCPS and displayed tighter binding of ammonia.
Of the two single mutations, L270K had a much greater impact on the ability of eCPS to utilize glutamine with negligible glutamine binding under the selective binding study conditions, 25-fold elevated K m for glutamine, and 10-fold slower rate of carrying out glutamine-dependent CP synthesis (Table  IV). The Q273E mutant showed more moderate effects, with a 10-fold reduction in glutamine binding under selective binding conditions, a 2-fold higher K m for glutamine, and a 4-fold slower rate of carrying out glutamine-dependent CP synthesis (Table IV). Interestingly, the sole effect of the more chargeconservative Q273A mutation was to increase the K m for glu-tamine by 20-fold (22). For wild type eCPS, Q273A, and Q273E, respectively, the K m values for glutamine were 0.16, 2.1, and 0.24 mM, and the k cat values were 9.1, 9.9, and 1.9 s Ϫ1 . Like Q273E/L270K, both Q273E and L270K displayed wild type rates for ammonia-dependent CP synthesis, indicating no major disruption of the SYN component in any of the mutants. However, the synergistic effect of the double mutation on glutamine usage, with Q273E/L270K completely incapable of utilizing glutamine, whereas the two component mutations were much less crippled, suggests a significant disruption of GAT active site structure when both changes occur simultaneously in the glutamine binding pocket.
As a first step toward assessing the structural effects of the present mutations, we have modeled these changes in silico using the wild type eCPS structure as a scaffold (Swiss-Model, Glaxo Wellcome; Fig. 2) (42,43). In this model, substitution of the leucine residue at position 270 with the longer lysine side chain results in a more constrictive architecture imposed on the glutamine binding pocket. It appears that glutamine could be partially occluded, the thioester intermediate could be positioned in a manner unfavorable for efficient hydrolysis, and/or the channel connecting the GAT active site to the SYN component could be blocked. Based on x-ray crystallography data, Huang and Raushel (44) suggest the involvement of 10 residues, Ser-35, Met-36, Asp-45, Lys-202, Gly-293, Ala-309, Asn-311, His-353, Pro-358, Gly-359, in formation of this channel. Site-directed mutagenesis analysis of these residues primarily implicated Gly-359, with kinetic parameters for the G359Y and G359F proteins indicating impaired glutamine utilization. The spatial clustering of the three residues targeted for mutagenesis in our study is distinct from the previously identified Kinetic parameters for utilization of glutamine and ammonia by WT and mutant CPSs ATPase assays were performed in a 500-l volume at 25°C, pH 7.6, in 50 mM HEPES, 40 mM NaHCO 3 , 20 mM MgSO 4 , 100 mM KCl, 5 mM ATP and in the absence and presence of the allosteric activator ornithine (20 mM). The rate of ADP formation was monitored in the presence of glutamine (0.08 -10 mM; glutamine-dependent ATPase) and NH 4 Cl (0.04 -540 mM; ammonia-dependent ATPase) as described under "Experimental Procedures." Glutamine-dependent activity was not detected for Q273E/L270K (NA). S.E. of the kinetic parameters was determined (GraFit, version 5.01) from nonlinear regression curve fitting and was within Ϯ10%. channel residues and may impact glutamine utilization at the point of entry of glutamine into the channel as opposed to constricting the diameter of the passage farther down (Fig. 4).
Precluding entry of glutamine would be an efficient mechanism for non-glutamine utilizing CPS Is to adopt and could explain why fCPS I remains unable to utilize this substrate despite retaining the appropriate catalytic machinery.
In wild type eCPS, the amide nitrogen of Gln-273 appears to form a hydrogen bond with the ␣-carboxylate of glutamine and thereby helps to correctly position both the substrate and the intermediates derived from it (12,22). The Q273E mutation would alter this interaction since it replaces the hydrophilic amide with a negatively charged oxygen and might well result in a more extended alteration of the wild type pattern of hydrogen bonds and salt bridges since the model predicts that the orientation of the mutant side chain is very different from that of the wild type.
These mutants also provide some insight into the molecular basis for the varying ammonia selectivity that occurs in CPSs. In mammals, the hepatic [NH 3 ϩ NH 4 ϩ ] is maintained at about 0.5 mM, and coma and death result when this level is exceeded (45). CPS I from various sources has an ammonium K m of 1-2 mM (equivalent to 40 M ammonia) that allows it to effectively detoxify excess ammonia, but the enzyme is completely unable to utilize glutamine. In contrast, eCPS has a K m for ammonium of 96 mM that would preclude serving in ammonia detoxification but can efficiently utilize glutamine, with a K m of 0.2 mM. In addition to mirroring CPS I in its inability to utilize glutamine, the Q273E/L270K double mutant of eCPS has a 4-fold decrease in ammonium K m and a corresponding 4-fold increase in its efficiency for utilizing ammonium. Similarly, the G359Y and G359F mutants designed to block the ammonia channel (44) displayed 8 -11-fold decreases in the ammonium K m . Clearly, additional structural differences in eCPS and CPS I must contribute to their much larger difference in ammonium K m . However, this movement of the double mutant toward a smaller K m is more significant than the actual measured value since the double mutant must "offset" the 6-fold increase in ammonium K m observed for the Q273E single mutant.
Because ammonia-dependent CP synthesis occurs entirely on the SYN component, and since this activity is altered in Q273E and Q273E/L270K, structural effects of these mutations must be communicated across the GAT domain interface to the SYN component. This cross-subunit communication is also ev-idenced in the alteration of uncoupled ATP hydrolysis that is localized to the first ATP site of the SYN component. Q273E/ L270K shows a 10-fold rate increase, suggesting that there has been a relaxation of the structural constraints normally limiting ATP hydrolysis in the absence of an ammonia source.
These studies have also provided the first suggestion of communication between the GAT and allosteric domains. In wild type eCPS, ornithine (4,15) and UMP (16) bind at the allosteric domain, and this occupancy is communicated to alter utilization of ATP, primarily at the second ATP-utilizing domain and slightly at the first ATP-utilizing domain (21). It has also been well established for wild type eCPS that there is two-way communication of occupancy of the GAT and first ATP-utilizing domains. However, Q273E has a greatly decreased response to the activator ornithine while retaining responsiveness to the inhibitor UMP. Therefore, a structural change resulting from the Q273E mutation must be communicated to the allosteric domain itself and/or to the second ATP-utilizing domain. Q273E/L270K displays enhanced responsiveness to ornithine at the first ATP-utilizing domain, as evidenced by a 30-fold increase in the uncoupled ATPase activity when ornithine is present. It thus appears that the structural effects of the Q273E mutation are communicated throughout all or most of the SYN component. Whereas the single Q273E mutation serves as a long range effector of binding site linkages but has only a small effect on glutamine utilization, the single L270K mutation shows wild type communication among domains but provides a significant loss of glutamine utilization. A detailed elucidation of how these two mutations combine to yield a eCPS mutant incapable of utilizing glutamine will most likely require solution of CPS I crystal structures as well as structures of eCPS mutants incorporating CPS I features.