The Carbamoyl-phosphate Synthetase of Pyrococcus furiosus Is Enzymologically and Structurally a Carbamate Kinase*

The hyperthermophiles Pyrococcus furiosus and Pyrococcus abyssi make pyrimidines and arginine from carbamoyl phosphate (CP) synthesized by an enzyme that differs from other carbamoyl-phosphate synthetases and that resembles carbamate kinase (CK) in polypeptide mass, amino acid sequence, and oligomeric organization. This enzyme was reported to use ammonia, bicarbonate, and two ATP molecules as carbamoyl-phosphate synthetases to make CP and to exhibit bicarbonatedependent ATPase activity. We have reexamined these findings using the enzyme of P. furiosus expressed inEscherichia coli from the corresponding gene cloned in a plasmid. We show that the enzyme uses chemically made carbamate rather than ammonia and bicarbonate and catalyzes a reaction with the stoichiometry and equilibrium that are typical for CK. Furthermore, the enzyme catalyzes actively full reversion of the CK reaction and exhibits little bicarbonate-dependent ATPase. In addition, it cross-reacts with antibodies raised against CK fromEnterococcus faecium, and its three-dimensional structure, judged by x-ray crystallography of enzyme crystals, is very similar to that of CK. Thus, the enzyme is, in all respects other than its function in vivo, a CK. Because in other organisms the function of CK is to make ATP from ADP and CP derived from arginine catabolism, this is the first example of using CK for making rather than using CP. The reasons for this use and the adaptation of the enzyme to this new function are discussed.

In this reaction one ATP molecule is used/molecule synthesized of CP, and the true substrate that is phosphorylated is carbamate, which is generated chemically from bicarbonate and ammonia (1)(2)(3). Because of the unfavorable equilibrium of the reaction, CK is thought to function in vivo exclusively in the direction of ATP synthesis using the CP generated by catabolic ornithine transcarbamylase in the fermentative catabolism of arginine (4).
In contrast to CK, CPS synthesizes irreversibly the CP that is used in the biosynthesis of pyrimidines, arginine, and urea, according to the following reaction (5).
The reaction catalyzed by CPS differs from that catalyzed by CK not only in its irreversibility in the direction of CP synthesis, but also in the use of bicarbonate and ammonia as the true substrates and in the utilization of two ATP molecules/molecule made of CP. CK and CPS also differ structurally. CK is a homodimer of a polypeptide of approximately 33 kDa (6), whereas CPS is a 120-kDa polypeptide that is either associated or fused to another polypeptide of approximately 40 kDa (7). Alignment of the amino acid sequences of CK and CPS failed to reveal the existence of a statistically significant sequence identity between the two enzymes (8), whereas there is a high degree of sequence identity among different CKs (6) or different CPSs (9). No obvious structural similarities are found when the recently determined three-dimensional structures of CPS from Escherichia coli (10) and of CK from Enterococcus faecalis (11) are compared. The two proteins exhibit an open ␤-sheet ␣/␤ structure. Whereas CPS exhibits the fold found in biotin carboxylase and in other proteins that synthesize acylphosphate bonds, the CK fold appears not to be represented in structural data bases, although it is likely to be found in other enzymes of presently unknown structure that synthesize acylphosphates and exhibit sequence similarity with CK, such as acetylglutamate kinase, ␥-glutamyl kinase, and long chain fatty acyl-CoA synthetases (6).
Given the important differences between CPS and CK, the recent description of CK-like CPSs in the hyperthermophilic archea Pyrococcus abyssi (12) and Pyrococcus furiosus (13,14) is puzzling. In the extracts of these extremophiles that live at 100°C and, in the case of P. abyssi, at high pressure in the ocean bottom, the CK-like CPS was the only activity found to synthesize CP in reaction mixtures containing ATP, bicarbonate, and ammonia (12)(13)(14). The polypeptide mass, homodimeric nature, and amino acid sequence (reported only for P. furiosus) of these pyrococcal enzymes (12,14) are characteristic of CKs. However, similarly to CPSs, these enzymes were reported (12,14) to use two ATP molecules/molecule made of CP and to exhibit ATPase activity in the absence of ammonia, although the magnitude of the ATPase was greater, relative to the overall reaction, than in classical CPSs (15,16). In contrast with most CPSs, which use glutamine with preference to ammonia as the nitrogen source (7), the pyrococcal CPSs use exclusively ammonia (12)(13)(14) but this is also the case with the ureotelic CPSs (7). Given these puzzling characteristics of pyrococcal CPS, we decided to study it in detail as it might represent an intermediate step in the evolution of CP biosynthesis (14). Thus, we have cloned and hyperexpressed in E. coli the gene encoding the CPS from P. furiosus, and we have purified and crystallized the recombinant enzyme generated in E. coli. The large amounts of pure protein obtained in this way have permitted us to study the stoichiometry, reversibility, point of equilibrium, and the nature of the substrates in the reaction. Our results unequivocally show that the enzyme catalyzes the CK reaction. Furthermore, our initial results of x-ray studies on enzyme crystals also indicate that the structure of this enzyme resembles closely that of CK. Therefore, this appears to be the first example of a CK with an anabolic role that is reserved in other organisms for CPS (the synthesis of CP as a precursor of arginine and the pyrimidines).

EXPERIMENTAL PROCEDURES
Materials-Recombinant enterococcal CK was isolated from E. coli BL21 (DE3) cells (obtained from Novagen) transformed with the plasmid pCK41 exactly as described (6). The preparation and characterization of monoclonal antibodies mAbCK1, mAbCK2, which recognize epitopes within the C-terminal 14 residues of enterococcal CK, and mAbCK3, which recognizes an epitope toward the center of the polypeptide, have been reported already (6). Polyclonal monospecific antisera against enterococcal CK were prepared by immunization of rabbits with the purified recombinant enzyme following a standard immunization protocol (17). E. coli CPS was purified as described (18). Pyruvate kinase, lactate dehydrogenase (both from rabbit muscle, salt-free), and V8 staphylococcal protease were from Sigma. Ornithine transcarbamylase was purified partially, free of CK, from E. faecium according to Ref.
1. Hexokinase and glucose-6-phosphate dehydrogenase (both from yeast) were from Roche Molecular Biochemicals. Enzymes were freed from ammonium sulfate and were placed in the buffer used in the assays by centrifugal gel filtration through Sephadex G-50 (19). Buffer pH values were determined at 22°C. Goat anti-rabbit IgG or antimouse IgG conjugated with peroxidase were from Promega, ammonium carbamate was from Aldrich, dimethyl suberimidate was from Pierce, and polyethylene glycols were from Fluka or Hampton. Other reagents were of the highest quality available.
Polymerase Chain Reaction Cloning of the P. furiosus Gene for CPS-Genomic DNA from P. furiosus (a generous gift of Dr. F. E. Jenney, Jr., Dept. of Biochemistry, University of Georgia, Athens, GA) was used as a template for polymerase chain reaction amplification of the CPS gene using a high fidelity proofreading thermostable DNA polymerase (Deep Vent, New England Biolabs) and the primers 5Ј-GTGGTTTCCATGGG-TAAGAGGGTAGTGATTGC-3Ј and 5Ј-GCATTCGCTAAGCTGGGTCT-TCTAAAGTTCCTCAGG-3Ј. These primers were designed to amplify the entire open reading frame for the CPS (14) and to introduce a NcoI site at the initiator ATG and a BlpI site downstream of the stop codon. The polymerase chain reaction products were digested with NcoI and BlpI and inserted into the corresponding sites of plasmid pET-15b (Novagen) behind the T7 promoter using T4 DNA ligase (USB) followed by transformation of E. coli DH5␣. The CPS gene in the resulting plasmid called pCPS184 was sequenced using an ABII prism DNA Sequenator (Applied Biosystems).
Expression and Purification of Recombinant P. furiosus CPS-E. coli BL21(DE3) cells transformed with the plasmid pCPS184 were grown at 37°C in a shaking incubator in 3 liters of LB broth containing 0.1 mg/ml ampicillin until an A 600 of 0.5 was reached. After a 3-h induction with 1 mM isopropyl ␤-D-thiogalactoside, the cells (6 g) were harvested by centrifugation, resuspended in 40 ml of 50 mM Tris-HCl, pH 7.5, at 4°C, and disrupted by sonication. The recombinant CPS was detected in a low amount. Therefore, the pCPS184-transformed BL21(DE3) cells were transformed with the plasmid pSJS1240 (a plasmid that incorporates the spectinomycin resistance gene from pSJS974 (20) into the SphI site of pRI952 (21)) (pSJS1240 was provided by Dr. S. J. Sandler, Dept. of Microbiology, University of Massachusetts), which allows the expression of the rare E. coli tRNA codons for arginine (AGA and AGG) and isoleucine (ATA). A high degree of expression (about 15% of the protein) was observed in these cells after overnight growth in LB medium containing 0.1 mg/ml ampicillin and 0.05 mg/ml spectinomycin. Purification of the enzyme from 10 g of cells obtained in this way was carried out at 0 -4°C as described for P. furiosus cells (14) except for the omission of the final Sephadex G200, Superose P12, and Mono Q column steps and the replacement of Blue Sepharose by Affi-gel Blue (from Bio-Rad).
Crystallization of P. furiosus CPS-Crystallization conditions using the hanging drop vapor diffusion method were tested initially at 4 and 22°C with the sparse matrix sampling procedure (23) by mixing 1.5 l of protein solution (10 mg/ml in either 10 mM Tris-HCl, pH 7.5, or 50 mM Tris-HCl containing 20 mM ATP and MgCl 2 ) and 1.5 l of reservoir fluid. Crystals formed at 22°C in about a week under a large range of conditions. The larger crystals obtained in the absence of ATP were produced in 15% polyethylene glycol 8000, 0.25 M Li 2 SO 4 in 0.1 M Tris-HCl, pH 8.5. In the presence of MgATP, crystals reaching 0.7 mm were obtained in 1.3 M sodium citrate in the same buffer.
X-ray Crystallographic Studies-For diffraction studies, a harvesting solution was used consisting of, for crystals grown in the absence of MgATP, 30% polyethylene glycol 8000, 5% ethylene glycol, and 0.275 M Li 2 SO 4 in 0.1 M Tris-HCl, pH 8.5, whereas, for crystals grown in the presence of MgATP, 1.45 M Na citrate and 10% glycerol in 0.1 M Tris-HCl, pH 8.5, was used. Crystals of about 0.4 ϫ 0.3 ϫ 0.3 mm in size were examined with an Image Plate (MAR RESEACH) area detector mounted on a Rigaku rotating copper anode x-ray source ( ϭ 1.5418 Å) operating at 40 kV and 100 mA. All data were collected at 100 K temperature with crystals flash cooled using the Oxford Cryosystem. Data sets were processed using the DENZO and SCALEPACK programs (24). Molecular replacement was performed with the AMoRe program (25) using polyalanine models of the structures of enterococcal CK (11) and the N-terminal 315 residues of E. coli biotin carboxylase (Protein Data Bank, entry code: 1bcn) as search models. The rotation and translation functions were performed using amplitudes from 15 to 4.0 Å resolution. The values of the rotation and translation top solutions using the CK monomer as search model exceeded consistently the values of all other peaks: rotation function peak height/noise, 7.5/5.4 (free enzyme) and 12.6/8.4 (ATP-bound); translation, correlation coefficient/noise, 49.3/23.4 (free enzyme) and 28.5/14.2 (ATP-bound); R factor/noise, 49/55.8 (free enzyme) and 54/57.6 (ATP-bound). The solutions were evaluated graphically using the O program (26), resulting in the consistent packing of P. furiosus CPS dimers in the crystal without steric problems. A similar search done with biotin carboxylase as a search model gave no clear solutions above the noise level.
Other Assays-Indirect enzyme-linked immunosorbent assays in 96microwell plastic plates (from Costar) were carried out as described in Ref. 17 using phosphate-buffered saline solutions containing 2 g/ml protein antigens and 2% bovine serum albumin with 0.1% defatted dry milk, respectively, for coating and blocking the wells. Immunoperoxidase detection was carried out with o-phenylenediamine as a chromogenic substrate (27). CP was determined as P i (28) after alkaline hydrolysis (29) or as citrulline (30) by coupling with ornithine and ornithine transcarbamylase. ADP and ATP were measured spectrophotometrically at 340 nm with pyruvate kinase/lactate dehydrogenase and hexokinase/glucose-6-phosphate dehydrogenase, respectively (31). Protein was measured according to Bradford (32) using a commercial reagent from Bio-Rad and bovine serum albumin as a standard. SDS-PAGE was carried out according to Laemmli (33). Cross-linking with dimethyl suberimidate and SDS-PAGE of the covalent adducts was done according to Davies and Stark (34).

Expression and Characterization of Recombinant P. furiosus CPS-
The expected band of approximately 34 kDa (exact mass deduced from the gene sequence (14), 34.3 kDa) was detected by SDS-PAGE in low quantity (Ͻ5% protein) in extracts of E. coli cells carrying the pCPS184 plasmid but not in extracts of cells carrying the progenitor pET-15b plasmid without the in-sert ( Fig. 1). The low expression appeared to be because of inefficient translation caused by the frequent occurrence in the pyrococcal gene of the AGA, AGG, and ATA codons for arginine and isoleucine, three codons that are rarely used in E. coli and that appear 25 times in the P. furiosus gene on six occasions as three adjacent pairs (14). This interpretation was confirmed by the drastic increase in the production of the recombinant enzyme that was observed when the pCPS184-carrying E. coli cells were transformed with plasmid pSJS1240 (20,21), which encodes the tRNAs for these rare codons (Fig. 1). The enzyme produced by the doubly transformed E. coli cells was soluble and active and was purified to Ͼ95% homogeneity ( Fig. 1, right panel), exhibiting a specific activity in the assay of Legrain et al. (13) of 0.075 and 1 mmol CP ⅐ h Ϫ1 ⅐ mg protein Ϫ1 at 37 and 60°C, respectively. The activity is similar, although somewhat higher than that reported for the enzyme purified from P. furiosus cells (14). The N-terminal sequence was confirmed to be GKRVVIALG, and thus E. coli removes the initial methionine in a manner similar to that observed in P. furiosus (14). Limited digestion with V8 staphylococcal protease yields two complementary fragments of approximately 23 and 13 kDa (SDS-PAGE estimate, Fig. 1), of which the 23-kDa fragment exhibits the N-terminal sequence of the intact enzyme, and the 13 kDa fragment has the N-terminal sequence DGEIKGVEAV and thus corresponds to the fragment that begins in residue 203 of the reported amino acid sequence deduced from the gene sequence. In agreement with the specificity of V8 staphylococcal protease, residue 203 is preceded by a glutamate residue. The sequence-deduced mass of the fragment that begins in residue 203 and ends in the C terminus of the enzyme is 12.4 kDa, which is in excellent agreement with the electrophoretic estimate. All these data confirm the fidelity of the polymerase chain reaction cloning strategy used to generate the recombinant enzyme and the identity of the recombinant and naturally produced enzymes.
The pyrococcal enzyme was recognized in enzyme-linked immunosorbent assay tests and Western blots by polyclonal rabbit antiserum raised against the CK from E. faecium, although approximately 50-fold higher concentrations of the antiserum were necessary to get the same response as with E. faecium CK (Fig. 2). Of three monoclonal antibodies against E. faecium CK only mAbCK1, recognizing an epitope localized within the Cterminal end 14 residues of CK (6), cross-reacted with the pyrococcal enzyme (data not shown), although approximately 100-fold higher antibody concentrations were needed for an equal reaction as with enterococcal CK. The immunological cross-reactivity of the two enzymes reflects their relatedness; the differences in reactivity being compatible with the 49% sequence identity.
Stoichiometry of the Reaction and ATPase Activity- Table I compares the production of ADP, P i , and CP by the pyrococcal enzyme with those observed with typical CK and CPS in an assay mixture containing ATP, bicarbonate, and ammonia. Because these assays also included large amounts of ornithine and ornithine transcarbamylase to convert the CP produced to citrulline and phosphate, CK and CPS should yield in these assays, respectively, 1 and 2 mol of both P i and ADP/mole of citrulline produced. The results obtained with enterococcal CK and E. coli CPS agree, within experimental error, with these expectations. With the pyrococcal enzyme the results, at both 37 and 60°C (the ornithine transcarbamylase used for coupling is stable at 60°C) (see Ref. 1), are essentially the same as with CK: a ratio of approximately 1 is found between the amounts produced of P i , ADP, and citrulline.
Previously the CPSs purified from P. abyssi (12) and P. furiosus (14) were reported to exhibit an ATPase activity in the absence of ammonia and in the presence of bicarbonate, corresponding to half of the ATP consumption in the presence of ammonia. When we assayed this ATPase activity with the recombinant enzyme at 37 and 60°C, we did not detect such activity (Table I) unless the concentration of enzyme used was vastly increased (data not shown), corresponding to similarly low activity as with enterococcal CK (0.3% of full activity (1, 6)). In contrast, the bicarbonate-dependent ATPase activity of CPSs typically represents 10% of full activity (15,16).
Equilibrium of the Reaction-The equilibrium of the reaction catalyzed by CPS is fully displaced toward CP synthesis (see Ref. 5), whereas with CK the value of the K eq (1) predicts conversion of only a small fraction of the ATP to ADP at the concentrations of carbamate expected to be present in the assay mixtures used here (1). The results illustrated in Fig. 3 confirm for E. coli CPS and enterococcal CK these expectations. With the former the amount of CP (Fig. 3) produced increases linearly with the amount of enzyme even when a large fraction of the ATP is used up, whereas with enterococcal CK the production of CP rapidly flattens out with increasing amounts of the enzyme, despite the existence of high concentrations of ATP and large excesses of bicarbonate and ammonia assuring a constant concentration of carbamate. The results with the pyrococcal enzyme fully replicate those obtained with enterococcal CK, as expected if the reactions catalyzed by the pyrococcal and enterococcal enzymes are identical. Furthermore, the extent of the reaction with E. coli CPS monitored by the production of ADP (Fig. 4) was the same whether or not the CP formed was removed by coupling with ornithine transcarbamylase, whereas with both enterococcal CK and the pyrococcal enzyme, the addition of ornithine transcarbamylase greatly increased the production of ADP, as expected if the equilibrium were displaced in the forward reaction by the removal of the product CP.
Use of Carbamate as Substrate of the Pyrococcal Enzyme-CK phosphorylates carbamate (2, 3), whereas bicarbonate and ammonia are the true substrates of CPS (5). In agreement with this, Fig. 5 shows that the production of CP (determined as citrulline via coupling with ornithine transcarbamylase) by E. coli CPS is the same with fresh and aged mixtures of potassium carbonate and ammonium chloride, whereas with enterococcal CK substantially more citrulline is formed with the aged than with the fresh mixtures; for in the latter, at the moment of addition to the assay carbonate and ammonia have not yet equilibrated with carbamate (3). The results with the pyrococcal enzyme are similar to those with enterococcal CK, indicating that carbamate is also the substrate for the enzyme from P. furiosus. However, to observe differences with aged and fresh mixtures, the mixtures had to be diluted more in the case of the pyrococcal enzyme than with enterococcal CK. This observation suggests that the pyrococcal enzyme has a lower K m for carbamate than the CK from E. faecalis. 2 Further proof of the use of carbamate by the pyrococcal enzyme was obtained (Fig. 6) by comparing the production of CP when either ammonium carbamate or ammonium carbonate was abruptly added to mixtures at pH 9.5 and 10°C containing ATP and the enzyme. These conditions were used by Jones and Lipmann (2) to demonstrate the use of carbamate by E. faecalis CK, because at this pH and temperature the stability of carbamate is increased. Our results confirm (2) for enterococcal CK and demonstrate for the pyrococcal enzyme, which is also active and stable at pH 9.5 (data not shown), the production of more CP with ammonium carbamate than with the carbonate. Again, as in the experiments with the fresh and aged mixtures reported in the previous paragraph, lower concentrations of carbamate and carbonate had to be used with the pyrococcal than with the enterococcal enzyme to demonstrate the differences. In summary, the results with fresh and aged solutions of ammonium carbonate and the results using carbonate or carbamate concur by showing that the pyrococcal enzyme uses carbamate rather than bicarbonate and ammonia as the substrate of the reaction.
Phosphorylation of ADP by Carbamoyl Phosphate-The established function in vivo of CK is to synthesize ATP from ADP and CP. In agreement with this function, enterococcal CK catalyzes faster the phosphorylation of ADP than the synthesis of CP (1). In contrast, CPSs do not catalyze the full reversion of their reaction of CP synthesis (5,15,35), although they catalyze, as a partial reverse reaction occurring at a rate of only 20% of the full reaction (5,15), the synthesis of one molecule of ATP from one molecule of ADP and of CP. Table II confirms that enterococcal CK catalyzes faster the phosphorylation of ADP than the synthesis of CP. Under the conditions of the assays illustrated in the Table the rate of ATP formation by this enzyme is approximately 4-fold higher than the rate of CP synthesis. Table II also shows that the pyrococcal enzyme catalyzes actively, as expected for a CK, the formation of ATP from ADP and CP, although in this case the ratio between the forward and reverse reactions approximates unity, possibly reflecting the adaptation of this enzyme to the new function of making CP rather than using it.
Structural Similarity with Enteroccal CK Revealed by X-ray Studies-Crystals of the pyrococcal enzyme grown in the absence or presence of MgATP (Fig. 7) diffracted with a conventional x-ray source to at least 2.6 and 2.0 Å resolution, respectively, although spectra were collected at 2.9 Å (98% completeness; R merge , 8.4) without ATP and at 2.2 Å (95.6% completeness; R merge , 11.7) with MgATP. The space group is, for the crystal without substrates, tetragonal P4 with unit cell parameters a ϭ b ϭ 97.78 Å and c ϭ 135.42 Å. Packing density considerations (36) indicate that for a monomer mass of 34.3 kDa the unit cell could contain 16 monomers (V m , 2.35 Å 3 /Da; solvent content, 47%), corresponding to four monomers in the asymmetric unit. Cross-linking with dimethyl suberimidate (Fig. 8) confirms (14) that the enzyme is dimeric and reveals the formation of dimers of dimers. Upon treatment with dimethyl suberimidate a major band appears corresponding to 2 Preliminary estimates yield an approximate K m for carbamate of 7 M. This value is about 10-fold lower than the K m for carbamate of enterococcal CK (1). The concentrations of carbamate were calculated from the equilibrium with bicarbonate and ammonia given in Ref. 1.

TABLE I
Stoichiometry of the reactions catalyzed by E. coli CPS, enterococcal CK, and pyrococcal CK-like CPS; lack of substantial ATPase activity of CK and CK-like CPS Solutions containing 0.1 M Tris-HCl, pH 8.0 (assays at 37°C) or 9.0 (assays at 60°C), 5 mM ATP, 5 mM MgCl 2 , 40 mM NaHCO 3 , 6 mM L-ornithine, 25 units/ml ornithine transcarbamylase (free from ammonium sulfate), 0.2 M NH 4 Cl (or, when omitted, 0.2 M KCl), and one of the following: 33 g/ml E. coli CPS, 0.083 g/ml enterococcal CK, or 2.5 g/ml (37°C), or 0.45 g/ml (60°C) pyrococcal CK-like CPS were incubated at the indicated temperature. The reactions were initiated with the addition of the enzyme and were terminated 15 min later by the addition of an equivalent volume of 15% trichloroacetic acid. The precipitated protein was removed by centrifugation, and P i , citrulline, and ADP were determined in the supernatant. the dimer and a less prominent band with the mass of the tetramer is also seen, whereas the cross-linking of three monomers is detectable but less frequent. Thus, a dimer of dimers is likely to occupy the asymmetric unit of the crystal. Crosslinking of enterococcal CK under the same conditions (Fig. 8) confirms its dimeric character (1) and reveals lesser tendency to form dimers of dimers. In the presence of MgATP space group was orthorhombic P2 1 2 1 2 1 with unit cell parameters a ϭ 55.22 Å, b ϭ 90.92 Å, and c ϭ 132.93 Å and an estimated number of 8 monomers/unit cell (V m ϭ 2.53 Å 3 /Da; solvent content, 51%) or two monomers, possibly making a dimer, in the asymmetric unit. Given the existence of nearly 50% sequence identity between the pyrococcal enzyme and enterococcal CK (6), and because we have recently determined the three-dimensional structure of CK enterococcal by x-ray crystallography (11), the molecular replacement method (37) 4 Cl were preincubated 5 min at 37°C, and then the indicated enzyme concentrations were added and the incubation was continued for 15 min more. The reaction was terminated by alkalinization with 1/10 volume of 5 M NaOH and followed by a further 15-min incubation to allow decomposition of the CP. The solution was acidified with trichloroacetic acid (7% final concetration), and after centrifugation, P i was determined (28) in the supernatant. Given the stoichiometry of E. coli CPS, the amount of P i found with this enzyme was divided by two to obtain the P i derived from the alkaline decomposition of CP. coli CPS) at 0°C was added with immediate mixing within the bath. Exactly 30 s after this addition the incubation at 37°C was terminated by mixing in abruptly 50 l of 15% trichloroacetic acid at 0°C. Citrulline was measured in the deproteinized supernatant (30). The fresh ammonium carbonate solution was prepared by mixing, immediately prior to use, equal volumes of fresh solutions of 1 M K 2 CO 3 and 2 M NH 4 Cl prepared in ice-cold water and kept at 0°C. The aged solution was prepared by heating the mixture at 37°C for 15 min followed by cooling in a melting ice bath for 15 min. 10-fold dilutions of these solutions in ice-cold water were made immediately prior to use for addition to the incubations with CK-like CPS. 100% citrulline production was 0.036, 0.20, and 0.048 mM for CK-like CPS, enterococcal CK, and E. coli CPS, respectively. values of the rotation and translation solutions exceeding consistently the values of all other peaks, which strongly indicates that the correct orientation of the model was determined in the two cases (see "Experimental Procedures"). The solution corresponds to dimers with the same overall shape of the enterococcal CK dimer packed in the crystal without interference between different dimers (Fig. 9). In contrast, a similar study made with a polyalanine model of biotin carboxylase, which is the basic structure of the catalytic domains of the typical, high molecular weight CPS (10), failed to yield any unambiguous solution, indicating that the folding of the CPS from P. furiosus resembles much more CK than biotin carboxylase and, by extension, typical CPSs. In addition, the organization of the monomers in the dimer given by the solution obtained with the CK monomer does not resemble the subunit organization in the quaternary structure of biotin carboxylase (38). DISCUSSION The present results clearly show that the CP-synthesizing activity previously reported in P. furiosus (13,14) is due to an enzyme that uses chemically made carbamate and a single ATP molecule to synthesize CP reversibly. The reaction exhibits the characteristic equilibrium of the CK reaction, an equilibrium that does not favor at 37°C the accumulation of CP. The enzyme catalyzes with comparable efficiency the forward and FIG. 6. Comparison of carbamoyl phosphate synthesis from ammonium carbamate and ammonium carbonate. Each tube contained 0.1 M glycine, pH 9.5, 7 mM Mg 2 Cl, 8.8 mM ATP, and 0.02 or 2 mg/ml enterococcal CK or CK-like CPS, respectively. The open circles represent the tube to which either 136 mol/ml solid ammonium carbamate (enterococcal CK) or 13.3 mol/ml ammonium carbamate added as a 1.33 M freshly made solution at 0°C in assay buffer (CK-like CPS) were added at zero time. The solid circles represent the tubes to which solutions of potassium carbonate (134 or 13.4 mol/ml CK or CK-like CPS reaction medium, respectively) and ammonium chloride (268 or 26.8 mol/ml CK or CK-like CPS reaction medium, respectively) were added. The incubation was at 10°C. The total volume was 1 ml, of which samples were taken at the intervals noted for P i measurement after alkaline decomposition of CP with 0.35 N NaOH.

Rates of forward and reverse reactions for enterococcal CK and
pyrococcal CK-like CPS The assay system for the forward reaction consisted of 0.1 M Tris-HCl, pH 8.0, 50 mM KCl, 0.2 M NH 4 HCO 3 , 15 mM MgSO 4 , 5 mM ATP, 2.5 mM phosphoenolpyruvate, 5 mM ␤-mercaptoethanol, 0.3 mM NADH, 0.05 mg/ml pyruvate kinase, 0.025 mg/ml lactate dehydrogenase, and either 2.5 g/ml CK-like CPS or 0.03 g/ml CK. The assay mixture for the reverse reaction consisted of 0.1 M Tris-HCl, pH 8.0, 15 mM MgSO 4 , 0.2 mM (enterococcal CK) or 0.4 mM (CK-like CPS) ADP, 5 mM CP, 5.5 mM glucose, 1 mM NADP, 0.02 mg/ml hexokinase, 0.01 mg/ml glucose-6phosphate dehydrogenase, and either 0.005 g/ml CK or 2.5 g/ml CK-like CPS. Incubation temperature was 37°C and the change in the absorbance at 340 nm was determined. In the reverse reaction for CK-like CPS, the enzyme was found to catalyze phosphoryl group transfer from CP to glucose. Therefore, the reaction was carried out for 15 min in the absence of the coupling system, and the glucose, NADP, hexokinase, and glucose-6-phosphate dehydrogenase were added at the end of incubation in a small volume to measure the amount of accumulated ATP in a short period over which glucose phosphorylation by CK-like CPS was negligible.  reverse reactions, and it is very inefficient in phosphorylating bicarbonate instead of carbamate, judging from its very small bicarbonate-dependent ATPase activity. All these properties are shared by typical CKs, such as the enterococcal enzyme (1,2,6). Our results are in conflict with the previously reported stoichiometry of 2 mol of ADP released/mole of CP formed by the partially purified enzyme from P. furiosus, assayed at 37°C (14) or by the enzyme isolated from P. abyssi, assayed at 27°C (12). Because these earlier enzyme preparations exhibited much greater ATPase activity in the absence of ammonia than the highly purified recombinant P. furiosus enzyme used here, the discrepancy would be explained if there were contaminating ATPases in the previous preparations that might have led to overestimation of the ADP production associated with CP synthesis. This possibility should be rigorously excluded with enzyme preparations obtained from cultures of pyrococci. Alternatively, it might be speculated that the highly purified enzyme used here is an individual component of a multicomponent CPS that would exhibit the classical CPS stoichiometry of 2 mol of ADP/mole of CP and that would be present in the possibly less pure preparations previously obtained from pyrococci (12,14). However, this possibility is not supported by the similar specific activity and homodimeric nature of the recombinant and naturally produced enzymes and it also makes little biological sense, because the sole practical result of using such complex machinery would be to use an extra ATP molecule/ mole of citrulline made in the ornithine transcarbamylasecoupled reaction. The coupling with ornithine transcarbamy-lase is essential in pyrococci given the rapid decomposition of CP at high temperature (13).
The high degree of sequence identity and immunological cross-reactivity of the pyrococcal enzyme and the enterococcal CK confirm the similarity of the two enzymes. Furthermore, our initial structural data obtained by x-ray crystallography clearly show that the pyrococcal enzyme exhibits a three-dimensional structure and quaternary organization that are highly similar to those of enterococcal CK and that are very different from those of CPS. Although the CK structure (11) reveals the existence of two catalytic sites/enzyme dimer, the relative orientation of the sites and the absence of intramolecular tunnels joining them exclude the possibility of catalytic collaboration between the two sites that is required for the synthesis of CP from bicarbonate and ammonia in three steps (bicarbonate phosphorylation, carbamate formation, and carbamate phosphorylation) that characterizes the mechanism of CPS (10,39). In summary, all indicate that except for its extreme thermostability and low activity at normal temperatures, 3 the pyrococcal enzyme is endowed with the characteristics of classical CKs. The finding of similar enzymes in P. 3 However, the activity at 100°C of the P. furiosus enzyme may be similar to that of enterococcal CK at 37°C. Thus, the pyrococcal enzyme released at 95°C, under the conditions of the CK assay (1), approximately 250 mol P i ⅐min Ϫ1 ⅐mg protein Ϫ1 , whereas the enterococcal CK produces at 37°C in the same assay approximately 600 mol citrulline⅐min Ϫ1 ⅐mg protein Ϫ1 (1). abyssi (12) and Pyrococcus horikoshii (a gene has been found in this organism (GenBank TM accession number 3257702) encoding a putative polypeptide exhibiting 89% sequence identity with the enzyme from P. furiosus) strongly suggests that this enzyme constitutes a constant component of the catalytic machinery of the hyperthermophiles of the Pyrococcus genus.
As already indicated, the generally accepted function of CK is to synthesize ATP from ADP and the CP produced mainly in the catabolism of arginine by the arginine deiminase pathway (4). There are strong reasons to deny such a function for the present CK. No arginine deiminase activity was detected in the extracts of P. furiosus cells cultivated in a medium containing 0.2 g arginine⅐liter Ϫ1 (40), suggesting that the arginine deiminase pathway is not operative in this organism. In keeping with this, no arginine deiminase putative gene was identified in the entire genome of the related organism P. horikoshii (41). If such a gene had existed in P. horikoshii, it should have been identified given the constant sequence motifs that are characteristic of arginine deiminases (42). P. furiosus cells could be grown in a defined medium containing ornithine as an arginine precursor (40), and the enzymes of the biosynthetic pathway of arginine, anabolic ornithine transcarbamylase, argininosuccinate synthetase, and argininosuccinase were detected in this organism (43). Similarly, aspartate transcarbamylase, the enzyme that catalyzes the second step of pyrimidine biosynthesis, was detected in both P. furiosus (13) and P. abyssi and was characterized in the latter organism (44). Thus, CP has to be made in these microorganisms to be utilized by ornithine transcarbamylase and aspartate transcarbamylase in the biosynthesis of arginine and pyrimidines. The only CP-making activity detected in extracts from these microorganisms was due to the enzyme studied here (12,13). This enzyme appears to be coupled functionally and to form physical complexes with ornithine transcarbamylase (13) and aspartate transcarbamylase (45) for there is evidence of efficient channeling of the CP in the direction of citrulline and carbamoyl aspartate biosynthesis. The inexistence in these organisms of classical CPS is also supported by the lack of detection of a classical CPS gene in the full genome of P. horikoshii (41). Again, it would be unlikely that such a gene would have escaped detection because many constant regions of characteristic sequence exist in all classical CPSs (9), and, for example, classical CPS genes were detected (whereas no CK genes were detected) in the genomes of the other three archea that have been sequenced fully, Methanococcus jannaschii (46), Archaeoglobus fulgidus (47), and Methanobacterium thermoautotrophicum (48). Taken together, all these data strongly suggest that the CK studied here plays a new metabolic role: the biosynthesis of CP for anabolic purposes.
Such an extraordinary use of a CK appears to be rendered possible by the extreme living conditions of pyrococci. Whereas in the mesophilic world the chemical formation of carbamate (49) might be slower than required by the needs of CP, rendering essential the enzymatic formation of carbamate by CPS in the initial two steps of its reactional mechanism (39), the high temperature in hydrothermal vents assures rapid chemical formation of carbamate without the need for enzyme catalysis. Another function fulfilled by mesophilic CPS is the provision of a high local concentration of carbamate at the site of carbamate phosphorylation within the enzyme. This becomes possible by coupling the synthesis of carbamate with the cleavage of an extra ATP molecule. However, the enzymatic synthesis of high local concentrations of carbamate appears unnecessary in P. furiosus given the high affinity of the pyrococcal enzyme for carbamate, 2 particularly because the concentration of carbamate may be relatively high in the habitat of P. furiosus given the finding in hydrothermal vents of high concentrations of CO 2 (50) (the true reactant, rather than bicarbonate, in the chemical synthesis of carbamate (49)) and of 0.6 -1 mM ammonia possibly derived from sediments (51) similar to those from where P. furiosus was grown (52). In fact, the enzyme activity in P. furiosus appears more than enough to serve the needs of arginine and pyrimidine synthesis, even at suboptimal concentrations of carbamate. Thus, from the increase in the enzyme activity given in Ref. 13, when the assay temperature is raised from 60 to 90°C, the activity in the initial P. furiosus extract would be at 90°C approximately 10 mol ⅐ h Ϫ1 ⅐ mg protein Ϫ1 (14), a value that is 9-fold higher than the activity of CPS in E. coli extracts, assayed at 37°C (16). Another characteristic of the CPS reaction that is mimicked by the high temperature of the habitat of pyrococci is the irreversibility of the synthesis of CP, because the rapid decomposition of CP at 100°C (13) would cause the concentration of this product to be essentially nil, thus displacing strongly the reaction in the direction of CP synthesis. In summary, the extreme living conditions of the pyrococci may render unnecessary the enzymatic synthesis of carbamate by CPS with the associated expenditure of an extra ATP molecule, as a prelude to making CP, explaining the use of CK for CP synthesis in these organisms. It is of interest that CK appears to have become adapted in P. furiosus to its new anabolic function, because when compared with enterococcal CK, it exhibits greater apparent affinity for carbamate 2 and is less effective in the synthesis of ATP from ADP and CP, relative to CP synthesis. Other adaptations exhibited by the pyrococcal CK that deserve further study are the much lower specific activity at 37°C 3 and much higher thermal stability than classical CK. Detailed comparisons of the three-dimensional structures of the enterococcal and pyrococcal CKs and sitedirected mutagenesis of key residues in the two enzymes will be essential to ascertain the reasons for these differences. Experiments with these goals are currently in progress in our laboratory.