Regulation of an Escherichia coli/Mammalian Chimeric Carbamoyl-phosphate Synthetase*

Carbamoyl-phosphate synthetase (CPSase) consists of a 120-kDa synthetase domain (CPS) that makes carbamoyl phosphate from ATP, bicarbonate, and ammonia usually produced by a separate glutaminase domain. CPS is composed of two subdomains, CPS.A and CPS.B. Although CPS.A and CPS.B have specialized functions in intact CPSase, the separately cloned subdomains can catalyze carbamoyl phosphate synthesis. This report describes the construction of a 58-kDa chimeric CPSase composed of Escherichia coli CPS.A catalytic subdomains and the mammalian regulatory subdomain. The catalytic parameters are similar to those of the E. coli enzyme, but the activity is regulated by the mammalian effectors and protein kinase A phosphorylation. The chimera has a single site that binds phosphoribosyl 5′-pyrophosphate (PRPP) with a dissociation constant of 25 μm. The dissociation constant for UTP of 0.23 mm was inferred from its effect on PRPP binding. Thus, the regulatory subdomain is an exchangeable ligand binding module that can control both CPS.A and CPS.B domains, and the pathway for allosteric signal transmission is identical in E. coli and mammalian CPSase. A deletion mutant that truncates the polypeptide within a postulated regulatory sequence is as active as the parent chimera but is insensitive to effectors. PRPP and UTP bind to the mutant, suggesting that the carboxyl half of the subdomain is essential for transmitting the allosteric signal but not for ligand binding.

, that catalyze carbamoyl phosphate formation from bicarbonate, ATP, and ammonia (4, 6 -13). The overall reaction involves the concerted action of CPS.A and CPS.B, each of which catalyzes two different ATP-dependent partial reactions (3,16). The E. coli enzyme is allosterically regulated (1,14) by metabolites from both pyrimidine and arginine biosynthetic pathways. UMP is a feedback inhibitor, while ornithine, IMP, and NH 3 activate the enzyme. The elegant three-dimensional structure of E. coli carbamoyl-phosphate synthetase has recently been solved (15) to 2.8-Å resolution.
While the CPS.A and CPS.B domains have specialized functions when fused together in the intact molecule (13, 38 -40), we have made the surprising discovery (41) that when cloned and expressed separately, each of the domains is functionally equivalent. CPS.A alone can catalyze both ATP-dependent partial reactions and ammonia-dependent carbamoyl-phosphate synthesis. The same results were obtained for the isolated CPS.B domain, except in this case, unlike the CPS.A domain, the activity is inhibited by UTP and activated by PRPP. The molecules are dimeric (Fig. 2), and while the monomers can each catalyze both partial reactions, the dimer (42) is required to catalyze the overall biosynthetic reaction. The monomers in the homodimer probably have the same function as the two fused domains, CPS.A and CPS.B, in the intact molecule.
Allosteric effectors bind to the B3 subdomain at the extreme carboxyl end of the synthetase domain in E. coli CPSase (39, * This work was supported by U.S. Public Health Grant GM47399. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 47, 59 -62) and mammalian mitochondrial CPSase I (63,64). In mammalian CAD, the serine residue phosphorylated by protein kinase A is also located (37) within the B3 subdomain. The biochemical studies were confirmed in the case of E. coli CPSase by locating the UMP site, tentatively, and ornithine binding sites, with certainty, in the x-ray structure (15). Moreover, kinetic studies of the E. coli enzyme (66) showed that CPS.B is the major target of allosteric regulation.
We have found (65) that allosteric effectors also bind to the B3 region of CAD by constructing an interesting chimeric protein in which the E. coli CPSase regulatory domain (B3) was replaced by the B3 subdomain of CAD. The chimera was inhibited by UTP but not by the E. coli effector UMP and was activated by PRPP, a metabolite that has no effect on the E. coli enzyme. In this report, we describe the construction and the function of a chimeric CPSase consisting of E. coli CPSase A1 and A2 subdomains fused to the mammalian CPSase B3 subdomain.

EXPERIMENTAL PROCEDURES
Plasmids and Strains-The 7.3-kilobase pair plasmid pHL2, containing the coding sequences for a chimeric molecule consisting of the complete E. coli CPSase domain with only the B3 regulatory domain replaced with the corresponding region of CAD, was constructed previously (65) and is under the control of the carB promoter. The E. coli strain RC50, defective in the carA and carB genes encoding the small and large subunits of E. coli CPSase, respectively, was kindly provided by Dr. Carol Lusty (Public Health Research Institute of the City of New York) (46).
Cell Growth and Recombinant DNA Methods-The recombinant proteins, which are expressed constitutively under the control of the carB promoter, were isolated from 200-ml cultures of transformed RC50 cells grown to stationary phase (46). The cells were resuspended in 2 ml of 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 5% glycerol buffer and broken by sonication for 2 min at 4°C. The sonicate was centrifuged at 10,000 ϫ g for 20 min at 4°C. Transformation was carried out by the Hanahan procedure (67). Restriction digests, ligations, and other DNA procedures were carried out using standard protocols (68).
Construction of the Recombinant Plasmids-The plasmid pHE-A12 e B3 m encoding the E. coli CPS A1 and A2 and mammalian B3 subdomains ( Fig. 1), was constructed 2 by reacting the plasmid, pHL2, Previous studies showed that the isolated CPS subunits (41) and each of the separately cloned (58) domains, A2 and B2, catalyze ammonia-dependent carbamoyl phosphate synthesis. We previously constructed (65) a chimeric molecule (CPS.A e B1 e B2 e B3 m ) consisting of the entire E. coli CPSase with the B3 domain at the carboxyl end replaced with the corresponding segment of CAD. In this study, the mammalian B3 regulatory subdomain (dark shaded bar) is directly fused to the E. coli CPSase A1-A2 subdomains to produce a chimeric molecule (A1 e A2 e B3 m ) only half the size of the native subunit. The chimeric deletion mutant (A1 e A2 e B3 m ⌬) lacks 59 of 163 residues from the carboxyl end of the E. coli/mammalian chimera. with NspV (site in the E. coli sequence, position 2,730) and StuI (site in the CAD sequence, position 3,892). The 4.3-kilobase pair fragment was then religated using T4 DNA ligase following treatment with the Klenow fragment of DNA polymerase. The 4.2-kilobase pair recombinant, pNS-A12 e B3 m ⌬, was constructed in order to study the consequences of deleting the carboxyl end of the CAD regulatory domain in the chimera. The plasmid, pHE-A12 e B3 m was reacted with SalI (site in CAD, position 4,205) and EcoRI (site in the vector), and the ends were made flush using the Klenow fragment prior to religation.
Protein Methods and Enzyme Assays-Protein concentrations were determined using the micro-BCA protein assay reagent kit from Pierce and by scanning Coomassie Blue-stained polyacrylamide gels. SDS-gel eletrophoresis was carried out on 10% polyacrylamide gels using the Laemmli procedure (69). The gels were scanned using an HP Scan Jet 4C, UN-SCAN-IT gel Automated Digitizing System, and the concentration of the isolated proteins was determined by measuring the ratio of peak density to total density in a given lane and the total amount of protein applied to the gel. The background density was subtracted, and all measurements were made within the linear range of the densitometer. The recombinant protein was cross-linked using dimethyl suberimidate following the procedure (42) used for intact CAD. The reaction mixture consisted of 0.03 mg of cross-linking reagent and 10 -20 g of protein in a total volume of 0.5 ml of 0.1 M N-ethylmorpholinacetic acid, pH 8.5. The reaction, which was carried out at room temperature, was initiated by the addition of the cross-linking reagent and quenched by the addition of 1 M glycine to a final concentration of 0.1 M. The extent of cross-linking was determined by SDS-polyacrylamide gel electrophoresis.
Carbamoyl-phosphate synthetase activity was assayed using a radiometric procedure (20,22) as modified by Carrey et al. (35) in which carbamoyl phosphate was trapped as acid-stable carbamoyl aspartate. The ATP concentration ranged between 0 and 5 mM, and 0.75 mM ATP, 10 mM MgCl 2 , 20 mM NH 4 Cl, and 12 mM NaHCO 3 were used for all effector response curves. The protocol for phosphorylation was described previously by Carrey et al. (36) with slight modification (65). The kinetic data were fit by least squares analysis (Micromath Scientist) to the Michaelis-Menten equation, the Hill equation, or the expres- Preparation of [ 32 P]PRPP-The procedure of Khorana (70) was used to synthesize [ 32 P]phosphoribosyl-5Ј-pyrophosphate. Radiolabeled PRPP (18,000 cpm/nmol) is enzymatically synthesized from ribose-5Јphosphate and ATP (500 Ci of [␥-32 P]ATP; Amersham Pharmacia Biotech) in a reaction catalyzed by PRPP synthetase, which was kindly provided by Robert Switzer (University of Illinois Urbana-Champaign). The specific radioactivity was determined by quantitating PRPP using an assay (18) that measures the amount of [ 14 C]CO 2 produced from PRPP and [ 14 C-COOH]orotidylate in the coupled reactions catalyzed by orotidylate PRPP transferase and OMP decarboxylase. The final concentrations in a 1-ml assay mixture were 0.1 M Tris-HCl, pH 7.5, 2 mM MgCl 2 , 100 M [ 14 C]orotate, and 1-20 nmol of PRPP. The reaction was initiated with 1-10 units of PRPP transferase-OMP decarboxylase mixed enzyme (Sigma). Following incubation for 1 h at room temperature, the reaction was quenched with 0.2 ml of 4 M perchloric acid. The [ 14 C]carbon dioxide generated in the reaction was trapped in 0.2 ml of a 1:1 (v/v) ethylene glycol/ethanolamine solution and counted in a liquid scintillation counter.
PRPP Binding Assay-the binding of radiolabeled PRPP to protein was measured using the spin column procedure (72) that separates free-ligand and protein-bound ligand. The protein sample was incubated with [ 32 P]PRPP for 15 min at 37°C. The 150-l reaction was then transferred to a 0.7 ϫ 2.8-cm bed Nick spin column (Amersham Pharmacia Biotech). The protein-bound ligand was eluted by centrifugation at 1,000 ϫ g for 4 min. Under these conditions, the volume of the eluant varied from 145 to 155 l, and the recovery of the protein was 98 -100%. The amount of bound PRPP was determined by liquid scintillation counting. The validity of the method was confirmed by measuring the binding of PRPP to CAD by continuous flow dialysis 3 and, for the chimeric protein, by equilibrium dialysis. For equilibrium dialysis, a 100-l sample of the chimeric protein (0.61 mg/ml) was dialyzed against 0.5 ml of PRPP, at the indicated concentrations, in 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 5% glycerol. After 24 h of equilibration, the PRPP concentration within the cell was determined by the PRPP assay described above. The same binding parameters were obtained by flow dialysis for CAD and by equilibrium dialysis for the chimeric protein as were obtained by the microcolumn method. Although the fraction of total ligand bound to the protein is low in the microcolumn method, the data are nevertheless reasonably accurate, since the column effectively separates free and bound ligand. The background counts in the eluted fraction obtained by applying the same amount of PRPP to the microcolumn in the absence of protein were quite low. For example, at 50 M PRPP, the average value of the background for six experiments was 46 Ϯ 6 cpm, compared with a typical value of 1,615 cpm in the presence of protein. At all PRPP concentrations, the signal:background ratio was at least 10.
The dissociation constant for UTP was determined indirectly by measuring the PRPP binding in the presence of several different UTP concentrations. The dissociation constant was obtained using the ex-

Construction, Expression, and Purification of the Chimeric
Protein-We previously constructed (65) a plasmid, pHL2, which codes for a chimeric protein in which the CPS B3 subdomain of the E. coli synthetase subunit was replaced by the corresponding subdomain of CAD (Fig. 1). A new plasmid, pHE-A12 e B3 m , that encodes the E. coli CPS A1 and A2 (residues 1-359) fused to the mammalian B3 subdomain (residues 1,298 -1,461) was constructed by deleting the E. coli CPS A3, B1, and B2 subdomains from the chimeric plasmid, pHL2 (65). When pHE-A12 e B3 m was transformed into the E. coli strain RC50, a uridine auxotroph that lacks the E. coli carA and carB genes, SDS-gel electrophoresis of cell extracts revealed a 58-kDa protein, not present in extracts of untransformed cells. The recombinant protein constituted 10% of the total soluble protein and could be readily isolated by chromatography on a fast flow 15 ϫ 3-cm Q-Sepharose column equilibrated with 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 5% glycerol. The protein CPS A1 e A2 e B3 m , which eluted in nearly homogeneous form (Ͼ90% pure) was identified by CPSase assays and SDS-gel electrophoresis. The oligomeric structure of the chimeric protein was determined by chemical cross-linking (data not shown) with dimethyl suberimidate. As the crosslinking reaction proceeded, the 58-kDa monomer was replaced by a species with a molecular mass of 120 kDa, suggesting that the chimeric protein is dimeric.
Catalytic Activity and Regulation of the Chimeric Protein-E. coli CPSase is more efficient than the mammalian enzyme. The K m for ATP is 5-fold lower and the k cat is 10-fold higher (Table I) than the corresponding values obtained for CAD. The purified chimeric protein also efficiently catalyzed ammonia-dependent carbamoyl phosphate synthesis. ATP saturation curves (Fig. 3A) of the chimeric enzyme yielded steady state kinetic parameters ( Table I) that were very similar to those obtained for E. coli CPSase. While the chimeric protein was insensitive to UMP, both mammalian effectors modulated its activity. There was little effect on K m , but the k cat was decreased 2-fold by 3 mM UTP and increased 1.5-fold by 200 M PRPP. Protein kinase A phosphorylation has no effect on the activity of the E. coli CPSase enzyme but was found (Fig. 3B and Table I) to significantly alter the kinetics of the chimeric protein. The phosphorylated chimera has approximately the same K m and k cat as the unphosphorylated protein in the absence of allosteric ligands, but UTP inhibition is appreciably reduced. The response to saturating PRPP was not affected by phosphorylation. Thus, the catalytic activity of the chimera closely resembles that of the E. coli enzyme, while regulation is characteristic of the mammalian protein.
The extent of activation and inhibition of the chimeric protein ( Fig. 4 and Table II) as a function of the concentration of the allosteric ligands was also very similar to that observed for CAD. The apparent dissociation constant K d for UTP was found to be 0.74 mM compared with a value of 0.64 mM obtained for CAD. Moreover, saturating UTP inhibits the chimera to nearly the same extent as CAD (72 and 91%, respectively). In the case of PRPP, the maximum activation is similar, although in this case the affinity of the activator for the chimeric protein is 6-fold lower than the value obtained for CAD.
The effect of phosphorylation of the B3 subdomain of the chimeric protein on allosteric regulation is also similar to that observed for CAD. UTP inhibition is effectively abolished. The maximum activation observed with saturating PRPP is not appreciably altered by phosphorylation, but the apparent affinity of the chimera for the activator was decreased. Phosphorylation results in a 4-fold increase in the concentration of the PRPP required for half-maximum activation. It is interesting that the PRPP response curve appears to be sigmoidal. This phenomenon is not observed in CAD and may reflect cooperative interactions between two PRPP binding sites in the chimeric dimer.
Binding of Allosteric Ligands to the Chimeric Protein-The binding of PRPP to the chimeric protein was determined by a microcolumn method and by equilibrium dialysis as described  under "Experimental Procedures." The Scatchard plot (Fig. 5A) of the binding data was linear and gave a dissociation constant ( Table III) that was about 5-fold higher than the value determined for the CAD complex (K d ϭ 5.1 M). The curve extrapolated to 1 mol of PRPP/mol of the chimeric protein monomer at saturation.
The dissociation constant for UTP is too high to measure using the same method. However, the displacement of bound PRPP by the addition of UTP indicates that the nucleotide effector also binds to the chimeric protein. In the presence of 5 mM UTP (Fig. 5A and Table III), the K d for PRPP increases about 5-fold. Similar results were obtained with CAD. Inexplicably, the number of PRPP binding sites appears to increase to 1.4 in the presence of UTP. While it is possible that UTP inhibition induces a conformation in which a weak binding site is exposed, it is more likely that this observation reflects inaccuracies of the determination. The dissociation constant for UTP determined by measuring PRPP binding as a function of the nucleotide concentration (Fig. 5B) was 0.23 mM. High concentrations of UTP completely displace bound PRPP.
Construction and Expression of the Deletion Mutant-Rubio and associates (60) identified a region within the regulatory domain that they propose to be important for allosteric regulation. In CAD, the sequence extends from residue 1,347 to 1,421. To further investigate the function of this region, a deletion mutant was constructed (Fig. 1) that produced a truncated polypeptide. The segment encoding residue 1,402 to the last residue, 1,461, in the chimera was excised from the plasmid pHE-A12 e B3 m by digestion with SalI and EcoRI. The resulting plasmid, pNS-A12 e B3 m ⌬, produced a polypeptide of the expected molecular mass, 51 kDa, when transformed into the E. coli strain RC50. The truncated protein was purified using the same procedure that was used for the full-length chimeric protein, but in this case there were small amounts of several contaminants in the purified preparation. The concentration of the recombinant protein was estimated by densitometry of a Coomassie-stained SDS gel.
Kinetics and Effector Binding of the Deletion Mutant-The deletion mutant was somewhat more active than the fulllength chimeric protein (Table I) but had reduced sensitivity to allosteric effectors. The ATP saturation curves (Fig. 6) obtained in the absence of effectors and in the presence of 200 M PRPP were indistinguishable. At low concentrations of ATP, 3 mM UTP had no significant effect on the activity, although slight inhibition was observed at ATP concentrations above 1 mM.
PRPP binds (Fig. 7 and Table III) to the deletion mutant with a dissociation constant of 79 M, 3-fold higher than the value obtained for the parent chimeric protein. Again, 5 mM UTP reduced PRPP binding, although in this case the nucleotide produced only a 2-fold increase in the K d , suggesting that the nucleotide also binds to the deletion mutant but with appreciably lower affinity. Thus, the carboxyl half of the regulatory domain is essential for the transmission of the allosteric signal but not for ligand binding. DISCUSSION While the catalytic mechanism of CPSase appears to be universal (3), the allosteric effectors of the enzyme from different pathways and organisms (73)(74)(75)(76)(77)(78) is highly diverse. In each case, however, the locus of regulation is the B3 subdomain at the carboxyl end of the CPS domain or subunit. We previously found (65) that the catalytic activity of E. coli CPSase could be placed under control of the mammalian allosteric effectors by exchanging the E. coli regulatory domain for the corresponding region of CAD. The chimeric protein reported here consists of two E. coli CPS.A subdomains, A1 and A2. The 31-kDa A2 subdomain (58) is one of two catalytic subdomains that alone can catalyze ammonia-dependent synthesis of carbamoyl phosphate at a rate 10 times faster than either CPS.A or the intact synthetase subunit. The 9-kDa A1 subdomain is an attenuator that modulates the intrinsically high activity of A2, and thus the catalytic activity of the fused A1-A2 domain is comparable with that of the parent E. coli CPS subunit (58). The CPS.A and CPS.B subdomains of both mammalian and E. coli CPSase as well as the B2 catalytic subdomain of the E. coli enzyme are all dimers. Moreover, only the dimer (42) can catalyzes the overall synthesis of carbamoyl phosphate. Therefore, the dimeric structure of the chimeric protein (Fig. 2), determined by chemical cross-linking, is probably also required for the CPSase activity of the chimera. The functional studies (58) of the isolated components suggest that the A3 subdomain is not required for catalytic activity, and thus it is not surprising that the K m and k cat of the chimeric protein are quite similar to the values obtained for the native E. coli CPSase.
A careful kinetic and thermodynamic analysis (66,67) of the E. coli enzyme suggests that allosteric effectors primarily alter the dissociation of ADP from the binary ADP-CPS.B complex probably by altering protein dynamics rather than by inducing a static perturbation of the substrate binding domain. This interpretation could account for the observation that allosteric effectors alter the K m and not the k cat of the native enzyme. The effect of allosteric ligands on the chimeric protein differs in that the major effect is on the k cat , not the K m . Similar observations were made (41) in studies of the isolated CPS.A and CPS.B domains. Whether this difference reflects changes in protein dynamics or is simply a consequence of the greater substrate inhibition exhibited by the isolated domains is not known. One possibly significant difference, assuming that carbamoyl phosphate synthesis requires the concerted action of the two monomers within the dimer (41,42), is that the effectors bind to both , where [X] 1/2 is the concentration of allosteric effector that produced half-maximum activation or inhibition, [S] is the substrate concentration, K d and K s are the apparent dissociation constants of the allosteric ligand and substrate, respectively, and ␣ is the ratio of the K m (no ligand)/K m (saturating ligand) as described previously (65).
c The curve was fit to an isotherm that incorporates cooperativity. The Hill coefficient was found to be 1.90 Ϯ 0.41. All other data in Fig. 4 were fit to an ordinary binding isotherm. monomers, and thus both partial reactions may be modulated.
Regardless of the details of allosteric mechanism, the response of the chimeric protein to the mammalian allosteric effectors has some interesting implications. First, CPS.B is not unique in its sensitivity to allosteric effectors. There has been no differentiation in the structure or the function of this do-main that has rendered it subject to control. The regulation of the chimera clearly shows that the CPS.A domain, like CPS.B, can be allosterically controlled. The binding of allosteric ligands to the regulatory domain triggers an allosteric signal that is transmitted as efficiently to CPS.A as to CPS.B. The allosteric regulation of CPS.A is especially interesting in view of the finding (15) that the tertiary structure of the B3 regulatory domain is distinctly different from that of the A3 subdomain present in the CPS.A domain of the parent molecule.
A plausible explanation for these observations is that modulation of catalytic activity results from a global change in the tertiary structure of the regulatory domain that alters its interactions with the catalytic subdomain. In this view, the regulatory domain has differentiated in different CPSase molecules to allow binding of specific allosteric ligands depending on the metabolic role of the enzyme in the cell. All activators are likely to induce the same change in regulatory domain conformation and subunit interactions so as to promote catalysis. While the effect of inhibitors and activators on the regulatory domain structure should be quite different, the conformation induced by all inhibitors would be expected to be the same or similar.
Second, the observation that the E. coli catalytic subdomains are regulated by the mammalian effectors in the chimera indi- FIG. 5. Effector binding to the chimeric protein. A, the binding of phosphoribosyl-5Ј-pyrophosphate to 12 g of the chimeric protein was measured by the microcolumn method (q) and by equilibrium dialysis (f) as described under "Experimental Procedures." In the Scatchard plot depicted above, b represents mol of bound PRPP/mol of protein, and [PRPP] represents the concentration of free PRPP. The effect of 5 mM UTP on the binding of PRPP (E) was also measured. The binding parameters obtained by a least squares fit of this Scatchard plot are given in Table III   The binding of PRPP to 9.8 g of the deletion mutant was measured in the presence (E) and absence (q) of 5 mM UTP. Axis labels are the same as in Fig. 5. The binding parameters are given in Table III. cates that the mechanism for allosteric regulation is virtually identical in mammals and E. coli and thus originated at an early stage in the evolution of the enzyme. Although the sequences of the regulatory domain of E. coli and mammalian CPSase are quite dissimilar (21% sequence identity), the domain from both species must have a similar tertiary fold and must function in a similar fashion.
E. coli CPSase, which lacks the target serine residue, is not phosphorylated by cAMP-dependent protein kinase. Replacement of the isoleucine at this position with serine 4 did not confer sensitivity to phosphorylation. However, when the entire E. coli regulatory domain was replaced with the corresponding region of mammalian CPS.B, phosphorylation had much the same effect on the E. coli catalytic domains as on CAD. This observation is consistent with the idea that the conformational changes induced by allosteric effectors and phosphorylation occur entirely within the regulatory domain and that the response is mediated by a change in its juxtaposition relative to the catalytic sites.
Previous mutagenesis studies (62) showed that deletion of 91 or 119 residues, representing about two-thirds of the regulatory domain, abolished allosteric regulation of E. coli CPSase by UMP but not ornithine. In the deletion mutant we constructed, 59 of 163 residues of the regulatory domain in the chimera were deleted. This site was selected because it bisects the sequence proposed by Cervera et al. (60) to be crucial for allosteric regulation. As predicted by their hypothesis, the deletion mutant lost sensitivity to both PRPP and UTP. Nevertheless, both effectors still bind to the mutant, although the dissociation constants are significantly increased. It is reasonable to expect that UTP would bind to the mammalian regulatory domain in much the same way as UMP binds to its bacterial counterpart. Unfortunately, the UMP site has not, as yet, been located (15) in the E. coli CPSase structure, but inorganic phosphate was found to bind to the regulatory domain (B3 or D). Since Thr 976 , one of the phosphate ligands, had been shown to be essential for UMP inhibition, the suggestion was made (15) that the inorganic phosphate may be binding to the nucleotide site. Three of four phosphate binding residues are conserved in CAD and are present near the carboxyl end of the deletion mutant. In the mammalian B3 subdomain, a tryptophan replaces Lys 992 , the fourth phosphate ligand in the E. coli enzyme. This substitution (60), considered to be a hallmark that distinguishes CPSase molecules that bind UTP from those that are regulated by UMP, could account in part for the lower affinity of the mammalian protein for nucleotides. The observation that the deletion mutant binds UTP, albeit with significantly lower affinity, suggests that in addition to these three interactions, many of the other residues that bind UTP must be located within the amino half of the regulatory domain. Although the K d is 3-fold higher in the deletion mutant, the PRPP binding site must also be located in this region. This conclusion is consistent with Davidson's studies of CAD gene deletions (79) that suggested that PRPP is closer to the amino end of the regulatory domain than the UTP binding site.
The specific binding of PRPP and UTP to the deletion mutant, accompanied by impaired allosteric regulation, suggests that the carboxyl end of the regulatory domain is involved in transmission of the allosteric signal. The x-ray structure of E. coli CPSase showed that the A2 and B2 subdomains have an identical tertiary fold. Thus, assuming that the mammalian regulatory subdomain, B3, has a homologous structure to the corresponding region of the E. coli CPSase, the structure of E. coli CPSase B2 and B3 subdomains (Fig. 8A) is a plausible representation of the A2 and B3 subdomains of the chimera. The structure shows that there are extensive interactions between the subdomains (Fig. 8A), most of which are lost in the deletion mutant (Fig. 8B). Thus, this region of the regulatory domain is a good candidate for mediating interdomain interactions involved in allosteric regulation. In contrast, the residues tentatively implicated in nucleotide binding are present in both the chimeric protein and the deletion mutant.
We conclude that the regulatory domain is an exchangeable ligand binding module that has differentiated in different species and pathways to bind a diverse complement of allosteric effectors. The allosteric ligands and phosphorylation induce conformational changes within the regulatory domain that alter its interactions with the catalytic subdomains. Although CPS.B is normally the regulatory locus, CPS.A can also be 4 V. Rubio, personal communication.
FIG. 8. Model structures of the chimeric protein and the deletion mutant. A, the B2-B3 subdomains of E. coli CPSase (residues 687-1,073 in the E. coli numbering system), shown here, are homologous to the A2-B3 subdomains of the chimeric protein. The structure was drawn using the E. coli CPSase coordinates deposited (15) in the Protein Data Bank (Brookhaven National Laboratory) using the program RasMol (R. Sayle, RasWin Molecular Graphics, version 2.6). The model shows ADP bound to the catalytic subdomain in yellow and Lys 992 , a residue in B3 postulated to be involved in binding the nucleotide effector, in white. B, this diagram represents the structure of the deletion mutant. The structure is the same as in A except that residues 1,020 -1,073 in the B3 subdomain, corresponding to residues missing in the deletion mutant, have been omitted. placed under allosteric control. Thus, CPS.A and CPS.B are functionally equivalent. The nucleotide and PRPP binding sites are likely to be located in the amino half of the domain, while the carboxyl half is essential for mediating interdomain interactions.