Structural Requirements for the Activation of Escherichia coli CTP Synthase by the Allosteric Effector GTP Are Stringent, but Requirements for Inhibition Are Lax*

Cytidine 5′-triphosphate synthase catalyzes the ATP-dependent formation of CTP from UTP using either NH3 or l-glutamine (Gln) as the source of nitrogen. GTP acts as an allosteric effector promoting Gln hydrolysis but inhibiting Gln-dependent CTP formation at concentrations of >0.15 mm and NH3-dependent CTP formation at all concentrations. A structure-activity study using a variety of GTP and guanosine analogues revealed that only a few GTP analogues were capable of activating Gln-dependent CTP formation to varying degrees: GTP ≈ 6-thio-GTP > ITP ≈ guanosine 5′-tetraphosphate > O6-methyl-GTP > 2′-deoxy-GTP. No activation was observed with guanosine, GMP, GDP, 2′,3′-dideoxy-GTP, acycloguanosine, and acycloguanosine monophosphate, indicating that the 5′-triphosphate, 2′-OH, and 3′-OH are required for full activation. The 2-NH2 group plays an important role in binding recognition, whereas substituents at the 6-position play an important role in activation. The presence of a 6-NH2 group obviates activation, consistent with the inability of ATP to substitute for GTP. Nucleotide and nucleoside analogues of GTP and guanosine, respectively, all inhibited NH3- and Gln-dependent CTP formation (often in a cooperative manner) to a similar extent (IC50 ≈ 0.2-0.5 mm). This inhibition appeared to be due solely to the purine base and was relatively insensitive to the identity of the purine with the exception of inosine, ITP, and adenosine (IC50 ≈ 4-12 mm). 8-Oxoguanosine was the best inhibitor identified (IC50 = 80 μm). Our findings suggest that modifying 2-aminopurine or 2-aminopurine riboside may serve as an effective strategy for developing cytidine 5′-triphosphate synthase inhibitors.

Hence, the precise interactions between CTPS and GTP remain unknown, thereby precluding detailed modeling studies.
Because of the absence of detailed crystallographic structural information on the effects of the interaction of GTP with CTPS, we have used a series of guanosine and GTP analogues to determine the structural features that are required for activation and inhibition of E. coli CTPS. We show that the 5Ј-triphosphate, 2Ј-OH, and 3Ј-OH are required for activation but not inhibition. The 2-NH 2 group of GTP plays an important role in binding recognition, whereas substituents at the 6-position play an important role in activation. The presence of a 6-NH 2 group obviates activation, consistent with the inability of ATP to activate the glutaminase activity of CTPS. CTPS is a recognized target for the development of antineoplastic (29 -33), antiviral (30, 34 -36), and antiprotozoal (37)(38)(39)(40) agents. Our structureactivity study suggests that inhibitors targeting GTP-depend-ent activation of CTPS need only be derivatives of 2-aminopurine and that the ribose and 5Ј-triphosphate groups are not required for effective inhibition.
Expression and Purification of CTPS-Recombinant wildtype CTPS was overexpressed and purified from E. coli strain BL21(DE3) cells containing the pET15b-CTPS1 plasmid as described previously (11). This plasmid encodes wild-type E. coli CTPS bearing an N-terminal hexahistidine tag SCHEME 1. Reactions catalyzed by CTPS.  (28)) play a role in GTP-dependent activation of Gln-dependent CTP synthesis. Residues between Val 428 and Thr 438 (both magenta) are not visible in the x-ray crystal structure (3). Arg 468 , Arg 470 , and Glu 68 also appear to interact with the purine moiety of GTP (3). Since GTP-dependent activation of Gln-dependent CTP formation requires the presence of a carbonyl group at the 6-position of GTP, the enzyme must undergo a change in conformation from that shown so that the protein interacts with the 6-position of the purine ring. In addition, a change in conformation may be required so that the 104 -110 loop (25,26) and Arg 429 (not visible) (27) can interact with GTP.

R470
(MGSSHHHHHHSSGLVPR2GSHMLEM 1 . . . CTPS). The soluble histidine-tagged CTPS was purified using metal ion affinity chromatography, and the histidine tag was subsequently removed using thrombin-catalyzed cleavage (new Nterminus, GSHMLEM 1 . . . , where M 1 represents the first residue of the wild-type enzyme). The results of purification and cleavage procedures were routinely monitored using SDS-PAGE. Typically, enzyme preparations were Ն98% pure. Enzyme Assays and Protein Determinations-CTPS activity was determined at 37°C using a continuous spectrophotometric assay by following the rate of increase in absorbance at 291 nm due to the conversion of UTP to CTP (⌬⑀ ϭ 1338 M Ϫ1 cm Ϫ1 ) (12). The standard assay mixture consisted of Hepes buffer (70 mM, pH 8.0) containing EGTA (0.5 mM), MgCl 2 (10 mM), CTPS (20 -80 g/ml), and saturating concentrations of UTP (1 mM) and ATP (1 mM) in a total volume of 1.0 ml. Enzyme and nucleotides were preincubated together for 2.5 min at 37°C, and then the reaction was initiated by the addition of either NH 4 Cl or Gln. GtetraP, 2Ј-dGTP, ddGTP, ACVMP, 2-NH 2 -PRTP, 6-thio-GTP, 2-NH 2 -ATP, N 7 -Me-GTP, 8-oxo-GTP, 7-deaza-GP, and O 6 -Me-GTP were kept on ice prior to addition to the assay solution. The concentrations of nucleosides and nucleotides were estimated spectrophotometrically using their molar extinction coefficients (see Table 1S).
The effects of varying concentrations of guanosine (0 -0.   Table 1) were examined using a fixed Gln concentration of 10 mM. The activation data were analyzed using the kinetic mechanism shown in Scheme 2. The initial velocity expression corresponding to this mechanism is given in Equation 2, where k o and k act are the rate constants for the formation of product as defined in Scheme 2, K A is the dissociation constant for the E⅐GTP complex, and K i n is the apparent dissociation constant for the E⅐(GTP) n complex. Pre-vious studies had indicated that the E⅐(GTP) n ⅐Gln complex is nonproductive (22). To simplify the kinetic analysis, inhibition by GTP was examined only at saturating concentrations of Gln (i.e. [Gln] ϭ 10 mM Ͼ Ͼ K m (assumed to approximate K S ) ϭ 0.32 mM (11)). Under these maximal velocity conditions, Equation 2 simplifies to Equation 3, and the initial velocity data for the inhibition of Gln-dependent CTP formation by GTP were fit to this latter equation.
The effect of fixed concentrations of guanosine (0.1 and 0.2 mM) on the GTP-dependent activation of Gln-dependent CTP formation was examined for GTP concentrations ranging between 0 and 0.5 mM.
The IC 50 values for the inhibition of both NH 3 -dependent CTP formation and Gln-dependent CTP formation by the GTP analogues were determined by fitting the observed relative velocities to Equation 1. For NH 3 -dependent CTP formation, assays were conducted with the concentration of NH 4 Cl fixed at 150 mM. For the Gln-dependent CTP formation, the concentrations of Gln and GTP were fixed at 10 mM and 0.15 mM, respectively. The concentrations of all purines and of purine nucleosides and nucleotides used in these assays ranged between 0 and 0.6 mM with the exception of inosine (0 -8 mM), , and 6-thioguanine (0 -0.8 mM). Due to the limited solubility of 6-thioguanine in aqueous solution, assays with this compound were conducted in solutions containing Me 2 SO (20%, v/v). The concentrations of CTP used in inhibition studies ranged between 0.1 and 0.6 mM.
The effect of guanosine (0, 0.15, 0.30, and 0.50 mM) on the glutaminase activity of CTPS (2.6 -2.8 g/ml) was assayed using a high pressure liquid chromatography-based assay as described previously (44) and using a saturating concentration of Gln (10 mM) in the presence of GTP (0.15 mM) to estimate the value of k cat .
The ability of 2-NH 2 -ATP (2 mM) to replace ATP as a substrate was examined in the presence of UTP (5 mM to ensure tetramerization of CTPS), NH 4 Cl (150 mM), and CTPS (20 g/ml) using the standard assay conditions. Protein concentrations were determined using the Bio-Rad protein assay with bovine serum albumin standards.
Data Analysis-Kinetic parameters were determined by fitting initial rate kinetic data to the appropriate equations by nonlinear regression analysis using the program KaleidaGraph

RESULTS AND DISCUSSION
Both purine (ATP and GTP) and pyrimidine (UTP and CTP) nucleotides are ligands for CTPS, and the ability of CTPS to utilize these different nucleotides for unique roles implies that the enzyme is able to discriminate effectively among them at the molecular level. For example, the enzyme has two overlapping binding sites for UTP and CTP (9). This arrangement permits the spontaneous development of resistance to drugs such as cyclopentenylcytosine and release from feedback inhibition by CTP (45-48) through clustered CTPS gene mutations alter-ing residues in the CTP-binding site while retaining the ability of CTPS to bind UTP and catalyze CTP formation. Consequently, designing drugs targeted at CTPS based solely on the structure of CTP or UTP may not be an effective strategy. An alternative strategy is to block GTP-dependent activation.
We reported previously that GTP functions as both a positive and negative allosteric effector of E. coli CTPS (22). At concentrations less than 0.15 mM, GTP activates Gln-dependent CTP formation by CTPS; however, at concentrations greater than 0.15 mM, Gln-dependent CTP formation is inhibited. In the absence of crystal structures of CTPS with bound GTP, it is difficult to identify potential interactions between CTPS and GTP that might contribute to this behavior (3,8,9,23). Therefore, we undertook a structure-activity study using GTP/ guanosine analogues (Fig. 2) to delineate the molecular features of GTP that give rise to activation and inhibition of CTPS activity.
Structural Requirements for Activation of Gln-dependent CTP Formation-The ability of a variety of GTP analogues to activate Gln-dependent CTP formation catalyzed by CTPS was examined (Table 1 and Fig. 3). For all activators identified, the rate of Gln-dependent CTP formation obtained a maximum value with respect to effector concentration and then decreased as elevated concentrations of the effector inhibited catalysis.

B. Ribose and 5'-triphosphate modifications
Arrows indicate changes in the structure relative to GTP/guanosine. The curves shown in Fig. 3 are direct fits to Equation 3 that describes the kinetic mechanism shown in Scheme 2 (22). For this kinetic mechanism, a reduction of the apparent K i value causes a decrease in the concentration of activator that yields maximal activation and a reduction of the absolute value of the initial velocity obtained at maximal activation. A decrease in the value of the apparent k act or an increase in the apparent K A causes a reduction in the absolute value of the initial velocity obtained at the point of maximum activation but does not alter the concentration of effector that gives maximum activation. For example, Fig. 4 shows the effect of guanosine on the activation and inhibition of CTPS by GTP. Examination of Table 1 reveals that there is a reduction in the values of the apparent k act and the apparent K i with increasing concentration of guanosine. On the other hand, the value of the apparent K A is not altered significantly, and the value of n is only slightly reduced.
Although this kinetic mechanism provides a satisfactory model for the observed activation, it is likely that the true kinetic mechanism is more complex. For example, the values of K i and k act are apparent values that are altered upon introduction of an inhibitor, such as guanosine (Table 1). This mechanism was employed because it was assumed that the effector A (or inhibitor) binds at two sites: one giving rise to inhibition and one giving rise to activation. One alternative kinetic mechanism that can be envisioned involves binding of GTP/guanosine analogues only at the GTP site (Scheme 3). This mechanism seems reasonable, since it assumes that n Ϫ 1 molecules of A bind to EA or EAS (i.e. after the initial molecule of activator A has bound) to form EA n or ESA n . Both kinetic mechanisms fit the v i /[E] T data well (R 2 Ϸ 0.999) but do give different values for K i . For example, fitting the curve for GTP-dependent activation and inhibition of Gln-dependent CTP formation gives K i ϭ 0.25 Ϯ 0.01 mM and K i ϭ 0.41 Ϯ 0.01 mM using Equations 3 and 4, respectively. Interestingly, for all GTP/guanosine analogues that activated Gln-dependent CTP formation, the K i values determined by fitting the v i /[E] T data with Equation 3 agree well with the corresponding observed IC 50 values, although this agreement may be fortuitous. It is important to note that both kinetic mechanisms are written such that the observed initial velocity is the rate of CTP production and not Gln production.   Table 1. The enzyme concentration was 26 g/ml.    Table 1. Inset, the activation curve for ITP reveals that this nucleotide is able to activate CTPS but only at higher concentrations relative to those required for GTP-dependent activation. For all activation experiments, the enzyme concentration was 26 g/ml. ͓E͔ T ϭ Our observation that guanosine, GMP, and GDP did not activate CTPS is in agreement with similar observations reported by Long and Pardee (12) and confirms that the 5Ј-triphosphate moiety is required for activation. Interestingly, GtetraP activated CTPS but exhibited a 3.5-fold drop in maximal activity relative to GTP (Fig. 3) due to a 2.6-fold reduction in the value of k act , a 2.3-fold increase in the value of K A , and a 1.8-fold increase in the value of K i , relative to that observed for GTP ( Table 1). The increase in the value of K i also results in a slight increase in the concentration of activator that yields maximal activation. Hence, the activation of CTPS appears to be optimal when a 5Ј-triphosphate moiety is present. 2Ј-dGTP was also able to activate Gln-dependent CTP formation but only to ϳ14% of the maximal activity observed with GTP. The absence of the 2Ј-OH causes a 7.1-fold reduction in the value of k act relative to GTP and a 2.6-fold increase in the value of K A . Thus, the 2Ј-OH is required for both activation and binding. The absence of both the 2Ј-OH and 3Ј-OH functions (i.e. ddGTP) obviated activation.
The role of the 2-NH 2 group in activation was investigated using ITP. This nucleotide exhibited a 2.5-fold drop in maximal activity relative to GTP. Interestingly, this decrease in activation efficiency arises from a minor reduction in the value of k act (2.0-fold) and large increases in the values of K A (36-fold) and K i (16-fold). Thus, the 2-NH 2 group plays only a minor role in activation but a major role in binding as is indicated by the shift of the point of maximal activation from ϳ0.2 mM in the case of GTP to ϳ3 mM for ITP (Fig. 3). It is important to note that even with apparently poor binding (i.e. elevated K A ), reasonable activation can be achieved. On the other hand, analogues such as O 6 -Me-GTP and 2Ј-dGTP have K A values consistent with reasonably good binding (i.e. similar to that observed for GtetraP) but exhibit poor activation. This suggests that the major roles for the 2Ј-OH and 6-O groups are in activation rather than binding. Replacement of the 2-NH 2 by oxygen (i.e. XTP) obviated activation, suggesting that a hydrogen bond donor is required at the 2-position of the purine for activation. However, it is also possible that the geometric and electronic perturbation arising from the substitution could alter the hydrophobicity and electron distribution of the purine ring, pK a values of the endocyclic nitrogens, or the hydrogen bonding capacity of other groups. For example, the site for ionization changes from N-1 for guanosine (pK a ϭ 9.2) to N-3 for xanthosine (pK a ϭ 5.7) (49).
The role of the carbonyl at C-6 of GTP was investigated using 2-NH 2 -PRTP, 6-thio-GTP, O 6 -Me-GTP, S 6 -Me-GTP, and 2-NH 2 -ATP. The inability of 2-NH 2 -PRTP to activate CTPS highlights the absolute requirement for the presence of a functional group on C-6. Replacement of the oxygen by sulfur (i.e. 6-thio-GTP) did not significantly alter the activation kinetics relative to GTP ( Table 1). The presence of a methoxy substituent on C-6 (i.e. O 6 -Me-GTP) reduced the binding affinity (increased K A ) and activation (k act ) ϳ1.6and 3.8-fold relative to GTP, respectively, but had minimal effect on K i . CTPS was not activated when either a CH 3 S-(i.e. S 6 -Me-GTP) or NH 2 -(i.e. 2-NH 2 -ATP) substituent was present at the 6-position of GTP. Thus, functional groups at the 6-position of the purine ring play an important role in activation either through direct interactions with the protein or through modulation of the electronic properties of the purine ring. For example, the pK a value of 9.2 for deprotonation of N-1 of guanosine (49) is reduced to 4.4 for 2-aminoadenosine (50), which could change N-1 from a hydrogen bond donor to a hydrogen bond acceptor. Only a limited increase in steric bulk appears to be tolerated at the 6-position.
N 7 -Me-GTP did not activate Gln-dependent CTP formation, indicating that CTPS may not tolerate additional steric bulk and/or a positive charge at the 7-position. Alternatively, a hydrogen bond acceptor is required at the 7-position. This latter possibility is consistent with the inability of 7-deaza-GTP to activate CTPS.
Since ATP (7 mM) cannot replace GTP to activate Gln-dependent CTP formation, and because GTP (Յ3.0 mM) cannot substitute for ATP as a substrate, molecular recognition of the guanine and adenine moieties of GTP and ATP by CTPS is highly specific. Our results indicate that this specificity arises because binding at the GTP-binding site requires the presence of a 2-amino function on the purine (absent in ATP) and because activation is obviated by the presence of an amino group at the 6-position (present in ATP). The requirement for a functional group that is a hydrogen bond acceptor of limited size (i.e. ϭO, ϭS, and -OCH 3 but not -SCH 3 ) but apparently not a hydrogen bond donor (i.e. -NH 2 ) at the 6-position for activation and the steric/charge sensitivity at the 7-position are interesting when considered in the context of the model of CTPS with GTP bound at the putative GTP-binding site (Fig.  1). In the model developed by Baldwin and co-workers (3), the carbonyl group at C-6 of GTP is protruding into the bulk solvent, and there appear to be no interactions between the protein and groups located at the 1-, 6-, or 7-positions of GTP. Site-directed mutagenesis, kinetic studies (5,24,25,27,28), and structural studies (3,9,23) support the identification of the GTP-binding site reported by Baldwin and co-workers (3) and also suggest that major conformational changes must occur upon GTP binding. Our results reveal that interactions between the protein and all functional groups on GTP, and especially groups at the 6-and 7-positions, must occur for efficient activation of Gln-dependent CTP formation. This is additional evidence that major conformational changes must occur upon GTP binding.
Structural Requirements for Inhibition of Gln-dependent CTP Formation-To examine the structural features of GTP that lead to inhibition of Gln-dependent CTP formation, we determined the IC 50 values for GTP/guanosine analogues under assay conditions where CTPS was fully activated by GTP (i.e. [GTP] ϭ 0.15 mM) ( Table 2). This approach allowed us to assess the relative inhibitory properties of the analogues in the absence of a detailed kinetic mechanism accounting for the observed inhibition and permitted us to establish an empirical measure of binding affinity. Interestingly, for those GTP ana-logues that activate CTPS, the values of K i determined from fitting the activation curves ( Fig. 3 and Table 1) are similar to the IC 50 values determined in the presence of the maximum activating concentration of GTP (Table 2).
With respect to inhibition of Gln-dependent CTP formation, guanosine, GMP, GDP, and GtetraP all exhibited approximately the same IC 50 values, indicating that inhibition, for the most part, may be attributed to the nucleoside. Consequently, guanosine analogues were employed in many of our inhibition studies rather than the more expensive 5Ј-triphosphates. Comparison of the IC 50 values for 2Ј-dG, 3Ј-dG, ACV, ACVMP, 2Ј-dGTP, and ddGTP with the IC 50 value for guanosine reveals that the 2Ј-OH and 3Ј-OH groups do not contribute significantly to inhibition, suggesting that the ribose moiety is not required for inhibition. Unfortunately, guanine was not sufficiently soluble under the assay conditions, so its ability to inhibit Gln-dependent CTP formation could not be determined. However, 6-thioguanine was soluble (using 20% Me 2 SO as a co-solvent) and inhibited CTPS with an IC 50 value that was only 1.7-fold greater than that observed for 6-thio-G. Inhibition of Gln-dependent CTP formation, therefore, arises primarily from interactions between CTPS and the purine base.
Modifications of the purine ring produced the most pronounced effects on the observed inhibition. The IC 50 values for inosine and ITP were ϳ16-fold greater than the value obtained for guanosine, and the IC 50 value for adenosine was elevated 34-fold relative to guanosine. These results reinforce our conclusion that the 2-amino group plays a major role in binding. The presence of a carbonyl function at the 2-position (i.e. xanthosine and XTP) gave only slightly improved inhibition relative to guanosine and GTP. Substitution by an amino group (i.e. 2-NH 2 -A) or sulfur (i.e. 6-thio-G) at the 6-position, removal of the oxygen substituent at C-6 (i.e. 2-NH 2 -PR), methylation of N-7 (i.e. N 7 -Me-G), methylation of O-6 (i.e. O 6 -Me-G), or removal of N-7 (i.e. 7-deaza-G) also improved inhibition relative to guanosine, but only slightly. Interestingly, although the absence of O-6 (i.e. 2-NH 2 -PR) did not affect inhibition, this group was absolutely required for activation. The most pronounced inhibition was observed with 8-oxo-G, which had an IC 50 value ϳ4-fold less than that of guanosine and 5-fold less than that of the feedback inhibitor CTP.
The dependence of v i /v o on the concentration of the GTP/ guanosine analogues at a saturating concentration of Gln (10 mM) exhibited varying degrees of sigmoid behavior, depending on the analogue examined (see the supplemental materials). The reason for this difference in behavior is not clear and does not appear to be related to the IC 50 value. For example, the v i /v o values exhibited a weak sigmoid dependence on the concentration of guanosine relative to that exhibited by O 6 -Me-G at various fixed concentrations of Gln. The corresponding IC 50 values for guanosine and O 6 -Me-G (Table 3) did not depend on Gln concentration, consistent with noncompetitive multisite inhibition with respect to Gln (22,42,43). Such inhibition could arise through cooperative binding of the GTP/guanosine analogues at GTP-binding sites within the active CTPS tetramer (see below).
Thus, although the structural requirements for activation of Gln-dependent CTP formation were quite stringent, numerous modifications of the purine ring produced only minor changes in the observed inhibition. The only exception to this generalization is the strict requirement for an amino group at the 2-position for effective binding.
Structural Requirements for Inhibition of NH 3 -dependent CTP Formation-To examine the molecular features of GTP responsible for inhibition of NH 3 -dependent CTP formation, we determined the apparent IC 50 values of various GTP/ guanosine analogues (Table 2) to obtain an empirical measure of their ability to inhibit CTPS. Overall, the IC 50 values determined for the inhibition of NH 3 -dependent CTP formation by the GTP/guanosine analogues were only slightly greater than the corresponding IC 50 values for the inhibition of Gln-dependent CTP formation (with the exception of GMP, GDP, guanosine, and inosine). In addition, the dependence of the observed inhibition on the structure of the GTP/guanosine analogues followed a trend similar to that observed for the inhibition of Gln-dependent CTP formation with only a few exceptions. For example, ddGTP, ACVMP, 6-thio-G, and 7-deaza-G did not inhibit NH 3 -dependent CTP formation as well as guanosine but did inhibit Gln-dependent CTP formation slightly better than guanosine. The reason for these differences is not clear; however, the changes in IC 50 values relative to that observed for guanosine are minor (Յ1.5-fold). Overall, the similarity between the IC 50 values for the inhibition of Gln-and NH 3 -dependent CTP formation suggests that the GTP/ guanosine analogues inhibit CTP formation by binding at the same site.
For all of the GTP/guanosine analogues, the dependence of the observed relative velocities (v i /v o ) on the analogue concentration was sigmoid for the NH 3 -dependent CTP formation with the exception of GtetraP and the weakly binding analogues adenosine, inosine, and ITP (see the supplemental materials). The IC 50 values for guanosine, GMP, GDP, ITP, ACV, and GTP (Table 4) did not appear to depend on the NH 3 concentration, consistent with noncompetitive multisite inhibition with respect to NH 3 (22,42,43). Again, such inhibition could arise through cooperative binding of the GTP/guanosine analogues at GTP-binding sites within the active CTPS tetramer (see below).
The Inhibitory Guanine Nucleoside-binding Site-Where do the GTP/guanosine analogues bind to effect inhibition? The most obvious site is the putative GTP-binding site identified by Baldwin and co-workers (3). It is also possible that molecular recognition within the ATP-binding site is not stringent, and analogues of GTP or guanosine might inhibit NH 3 -and Glndependent CTP formation by binding at the ATP-binding site. However, this latter possibility seems unlikely, since adenosine is a very poor inhibitor with an IC 50 value ϳ40-fold greater than the values obtained for the GTP/guanosine analogues. In addition, we have confirmed the findings of Long and Pardee (12) that GTP (3 mM) cannot replace ATP as a substrate for the synthase reaction. These observations suggest that GTP does not bind to any significant extent at the ATP-binding site and that we are not probing interactions at the ATP-binding site. Conversely, GTP-binding studies conducted in the presence of ATP have shown that ATP does not bind at the GTP-binding site (10). This conclusion is also in accord with the inability of ATP to activate the glutaminase activity of CTPS and the observation that ATP does not inhibit activation by GTP when present at concentrations 10-fold greater than GTP (22). Thus, CTPS appears to exhibit exquisite specificity for the purine nucleotides ATP and GTP. Since the only structural differences between GTP and ATP are at the 2-and 6-positions, it is not surprising that molecular recognition of O-6 and the 2-amino  group is critical for binding discrimination. Although both groups are required for activation, it appears that the 2-amino group is more important for binding than for activation. On the other hand, the O-6 is required for efficient activation but plays only a minor role in binding. The x-ray crystal structure of E. coli CTPS with bound ADP and CTP reveals that specificity for the adenine ring of ATP is provided by main chain hydrogen bonds from the amide of Val 241 and the carbonyl of Lys 239 to the N-1 and N-6 atoms, respectively (9). The C-2 atom of the adenine ring appears to be exposed to solvent. However, we found that 2-NH 2 -ATP (2 mM) did not serve as a substrate for NH 3 -dependent CTP formation catalyzed by CTPS. Discrimination against purine nucleotides bearing a 2-amino substituent (e.g. 2-NH 2 -ATP or GTP) may arise because of the residues (Ala 182 , Ala 183 , Asp 240 , Val 241 , and Asp 242 ) that pack near the C-2 atom. On the other hand, in models (3), the specific recognition of the 2-amino function of GTP appears to arise through hydrogen bonding interactions with the side chains of Glu 68 and Arg 468 (Fig. 1).
High concentrations of GTP (22) or guanosine (Fig. 5) do not inhibit the glutaminase activity. Hence, the concentrations of guanosine or GTP that give rise to inhibition do not displace GTP that is activating the enzyme. This suggests that the GTP/ guanosine analogues may be binding at an allosteric site and is in accord with our observations that the inhibition of NH 3 -dependent CTP formation conforms to a noncompetitive, multisite inhibition model and that the inhibition of both Gln-and NH 3 -dependent CTP formation is cooperative (n Ϸ 2-3). From the activation plots (Fig. 3), it appears that there are two sites with different affinities: a low affinity site with K i (or IC 50 ) Ϸ 0.2-0.5 mM, and a high affinity site with K A Ϸ 0.08 mM. Hence, the allosteric site may be either a second, low affinity guanosine nucleoside/tide binding site on each monomer or a GTP site in an adjacent subunit. The parsimonious explanation would be that inhibition arises due to cooperative effects of GTP/ guanosine analogues binding at the GTP activation sites of adjacent subunits. Indeed, this conclusion is in accord with the observations of Levitzki and Koshland (10), who reported the values of the four intrinsic dissociation constants for GTP binding to the CTPS tetramer (i.e. in the presence of ␤,␥-CH 2 -ATP and UTP) as 0.06, 0.12, 0.47, and 0.32 mM. The first two dissociation constants agree well with the K A value of 0.08 mM that we measured for the GTP-dependent activation of Glndependent CTP formation, and the second two dissociation constants agree well with the K i value of 0.28 mM observed under the same conditions.
With the exception of GtetraP, adenosine, and ITP, plots of v i /v o versus inhibitor concentration (see the supplemental materials) for the inhibition of NH 3 -dependent CTP formation exhibit a pronounced sigmoid character with n Ϸ 2-3. This implies marked cooperativity of inhibitor binding. On the other hand, with the exception of 2-NH 2 -PR, 7-deaza-G, 8-oxo-G, and O 6 -Me-G, plots of v i /v o versus inhibitor concentration for the inhibition of Gln-dependent CTP formation exhibit only slight sigmoid character with less cooperativity (n Ϸ 2). Since the IC 50 values for the inhibition of Gln-dependent CTP formation were determined in the presence of 0.15 mM GTP, the enzyme was in its fully active conformation for efficient catalysis of Gln hydrolysis. Although the presence or absence of sigmoid inhibition behavior is difficult to rationalize from the present data, it is possible that binding of GTP/guanosine analogues at the allosteric site in the presence of GTP may not induce such a dramatic change in conformation as occurs when GTP/guanosine analogues inhibit NH 3 -dependent CTP formation. Hence, less cooperativity is observed for the inhibition of Gln-dependent CTP formation relative to that observed for the inhibition of NH 3 -dependent CTP formation.
Mechanism of Inhibition-Previously, we suggested that GTP could inhibit NH 3 -dependent CTP formation by blocking the entry of NH 3 into the enzyme (22). Although this may still occur, it appears that the inhibition phenomenon is more complex. The similarity of the IC 50 values for the inhibition of NH 3dependent and Gln-dependent CTP formation by GTP/ guanosine analogues implies that inhibition may be occurring via a common mechanism. Previously, we showed that concentrations of GTP up to 0.5 mM do not inhibit the glutaminase activity (22). In addition, concentrations of guanosine up to 0.5 mM do not inhibit the glutaminase activity in the presence of GTP (0.15 mM) or the intrinsic glutaminase in the absence of GTP (Fig. 5), suggesting that nucleoside inhibitors do not displace activating GTP (at least at concentrations of Յ0.5 mM). This does not preclude binding of the inhibitors at additional GTP-binding sites. Interestingly, k cat for Gln-dependent CTP formation is approximately half the k cat value for NH 3 -dependent CTP formation (25,26). This difference could arise because allosteric activation by GTP bound at the two high affinity sites only gives rise to catalysis of Gln hydrolysis in two subunits of the tetramer, whereas all four subunits are capable of catalyzing NH 3 -dependent CTP formation. This scenario seems plausible, considering that affinity labeling by 6-diazo-5-oxonorleucine of only half of the subunits of the CTPS tetramer abolishes all Gln-dependent CTP formation (51). Inhibition of Gln-dependent CTP formation could, therefore, arise from binding of GTP/ guanosine at the remaining two low affinity sites.