Yeast epiarginase regulation, an enzyme-enzyme activity control: identification of residues of ornithine carbamoyltransferase and arginase responsible for enzyme catalytic and regulatory activities.

In the presence of ornithine and arginine, ornithine carbamoyltransferase (OTCase) and arginase form a one-to-one enzyme complex in which the activity of OTCase is inhibited whereas arginase remains catalytically active. The mechanism by which these nonallosteric enzymes form a stable complex triggered by the binding of their respective substrates raises the question of how such a cooperative association is induced. Analyses of mutations in both enzymes identify residues that are required for their association, some of them being important for catalysis. In arginase, two cysteines at the C terminus of the protein are crucial for its epiarginase function but not for its catalytic activity and trimeric structure. In OTCase, mutations of putative ornithine binding residues, Asp-182, Asn-184, Asn-185, Cys-289, and Glu-256 greatly reduced the affinity for ornithine and impaired the interaction with arginase. The four lysine residues located in the SMG loop, Lys-260, Lys-263, Lys-265, and Lys-268, also play an important role in mediating the sensitivity of OTCase to ornithine and to arginase and appear to be involved in transducing and enhancing the signal given by ornithine for the closure of the catalytic domain.

In Saccharomyces cerevisiae, the expression of arginine anabolic and catabolic genes is regulated at multiple levels in response to a wide variety of metabolic signals such as arginine and amino acid concentrations, nitrogen quality, and availability (reviewed in Ref. 1). In addition, two mechanisms are known to operate at the level of enzyme activity. Excess arginine causes feedback inhibition of the activity of the two first enzymes of the anabolic pathway, acetylglutamate synthase and acetylglutamate kinase (2,3). Another regulation known as the "epiarginase control" leads to the inhibition of the activity of the anabolic ornithine carbamoyltransferase (OTCase) 1 (EC 2.1.3.3) by the catabolic enzyme, arginase (EC 3.5.3.1). In the presence of ornithine and arginine, the respective substrates of the two enzymes (4), OTCase and arginase form a one-to-one enzyme complex in which the activity of OTCase is inhibited, whereas arginase remains catalytically active (5). The epiarginase regulation prevents the recycling by OTCase of ornithine produced by arginase and, because both enzymes are cytosolic in S. cerevisiae, is expected to avoid the operation of a futile urea cycle when yeast is growing on arginine as a nitrogen source (4). This control was found in yeast species showing a strong Pasteur effect and taxonomically related to Saccharomyces (6) and was also reported in Bacillus subtilis (7). The absence of the epiarginasic regulation in some species can be linked to a mitochondrial localization of OTCase, whereas arginase is cytosolic, as for example in Debaryomyces hansenii, Hansenula anomala, and Schizosaccharomyces pombe (8).
Direct interaction between OTCase and arginase, which are trimeric, was demonstrated by molecular sieving, a complex containing both enzymes being only observed in the presence of ornithine and arginine (4). Purification in the presence of these effectors (4) and equilibrium sedimentation and electron microscopy (9) identified an OTCase-arginase hexamer. The association of the two enzymes was rapidly reversible in vivo as well as in vitro under specific conditions, such as protein dilution, suggesting that no enzymatic modification was involved in this inhibitory mechanism (4,10,11). It was proposed that the ornithine regulatory receptor site on OTCase is distinct from the catalytic site, because by mutation, heating, or chemical alteration inhibition by arginase can be lost without impairing the catalytic activity of OTCase (5). However, the occurrence of a second ornithine binding site was not supported by the data of Hensley and co-workers (9) who proposed that the binding of ornithine to the active site promotes a conformational change needed for the complex formation with arginase.
Vissers et al. (6) have investigated the occurrence of the epiarginase control in 32 yeast species belonging to 12 different genera and obtained some correlation between the sensitivity of OTCases to excess ornithine and their sensitivity to arginase. Indeed, all OTCases inhibited by arginase present a marked inhibition by excess ornithine. However, the reverse is not true, and some arginase insensitive OTCases are inhibited by excess ornithine. To get deeper insight at the molecular level into the mechanism involved in the inhibition of OTCase by arginase, we created and analyzed in the S. cerevisiae OTCase a series of mutations leading to the loss of inhibition of its activity by arginase. We also identified residues in the arginase that are required for its regulatory function. EXPERIMENTAL (17-mer)3 -CYC1-lacZ) and Y187 (MAT␣,112, met Ϫ , gal4⌬, gal80⌬, URA3::GAL1-lacZ) (12) and was the recipient strain for experiments using the two-hybrid system. The yeast strain MaV99 (MATa, SPAL10::URA3), which contains an integrated GAL1-URA3 reporter gene activated by the GAL upstream activation sequence and derives from the strain MaV52 (MATa, ura3-52, leu2-3,112, trp1-901, his3-200, ade2-101, gal4⌬, gal80⌬, can1 R , cyh2 R , GAL1::HIS3@LYS2, GAL1::lacZ) (13), was the recipient strain for experiments using the reverse two-hybrid system. This strain lacks both GAL4 and GAL80 genes and contains integrated GAL1-lacZ and GAL1-HIS3 reporter genes. The yeast strain 02859d (ura3, leu2, trp1), isogenic to the wild type strain ͚1278b, was used to delete the CAR1 and ARG3 genes. Deletion of the CAR1 gene was achieved by transformating the strain 02859d with pWX4 plasmid as described by Dubois and Messenguy (14), leading to strain 02859d⌬car1. Deletion of the ARG3 gene was performed using the PCR strategy developed by Wach (15). The disruption cassette, containing the kanMX4 cassette with long flanking regions homologous to the ARG3 gene, was synthesized using PCR. This cassette confers to yeast a resistance to Geneticin (G418). The homologous regions of ARG3 gene were located upstream from the start codon and downstream from the stop codon. The PCR protocols for preparation of this cassette were identical to those described in Wach (15). Integration of the kanMX4 cassette at the target locus was confirmed using the PCR strategy as described by Wach (15). Strain 02859d⌬arg3-car1 (ura3, leu2, trp1, arg3::kanMX4, car1⌬) was used as recipient strain for transformation with plasmids overexpressing CAR1 or ARG3 genes. Strain 12S16c (ura3, leu2, arg3⌬) (a gift from M. Crabeel) was used as recipient strain for transformation with plasmids expressing wild type or mutated ARG3 genes or with plasmids expressing GBD-OTCase or GAD-OTCase. Strain 12T7cI (ura3, car1⌬) (14), isogenic to the wild type strain ͚1278b, was used as recipient strain for transformation with plasmids expressing GBD-arginase or GAD-arginase.
The strains used in the two-hybrid analysis were grown on synthetic medium containing 0.7% yeast nitrogen base without amino acids. This medium was supplemented with 2% glucose and all amino acids except those whose omission was required for plasmid selection. All the other strains were grown on minimal medium (M.am) containing 10 mM (NH 4 ) 2 SO 4 , 3% glucose or 2% galactose, vitamins, and mineral traces as described previously (16). The lithium acetate procedure was used to transform the recipient yeast strains (17) Escherichia coli strains XL1-Blue, XL1-Red, and HB101 from Stratagene were used for plasmid amplification, plasmid mutagenesis, and selection of plasmids containing a functional LEU2 gene and the GAD-ARG3 gene, respectively.

Construction of Plasmids Overexpressing OTCase and Arginase under the GAL10 Promoter
To overexpress OTCase and arginase, we amplified by PCR the full-length ARG3 and CAR1 genes, using as template the genomic DNA from the wild type strain ͚1278b and as primers the appropriate oligonucleotides extended with BamHI restriction sites. These fragments were inserted in the BamHI site of vector pYeF2 (pUC19, 2, URA3, GAL10 promoter) (18), yielding plasmids pDC2 (pGAL10-ARG3, 2, URA3) and pEJ24 (pGAL10-CAR1, 2, URA3). The two genes were sequenced to ensure that no mutation was introduced during the PCR procedure.

Construction of DNA-binding Domain and Activation Domain Fusions
The DNA binding domain of the Gal4 activator, Gal4-(1-147), is referred to as GBD, and its activation domain, Gal4-(768 -881), is referred to as GAD. The underlined GBD and GAD refer to the DNA sequences encoding these domains.
GBD-containing fusion genes were constructed in vector pAS2 (19). Transformants harboring the vector, or a derivative thereof, were selected by omitting tryptophan from the yeast nitrogen base medium (Difco, Detroit, MI). GAD-containing fusion genes were constructed in vector pACTII (19), transformants harboring the vector, or a derivative thereof, were selected by omitting leucine from the yeast nitrogen base medium.
GBD-ARG3 and GAD-ARG3 Fusions-To construct the GBD-ARG3 and GAD-ARG3 gene fusions, we PCR-amplified the ARG3 gene, using as template the genomic DNA from the wild type ͚1278b strain and as primers the appropriate oligonucleotides flanked by a BamHI site for the 5Ј-end and a XhoI site for the 3Ј-end. This DNA fragment was digested by BamHI and XhoI and inserted into the pAS2 vector, then digested by BamHI and SalI, leading to plasmid pGBD-ARG3, and inserted into the pACTII vector, and finally digested again by BamHI and XhoI, leading to plasmid pGAD-ARG3. In these GBD-ARG3 and GAD-ARG3 fusions, we determined the nucleotide sequence to ensure that the fusions were in-frame and that no mutation had been introduced during the PCR procedure.
GBD-CAR1 and GAD-CAR1 Fusions-To construct the GBD-CAR1 and GAD-CAR1 gene fusions, we amplified by PCR the CAR1 gene, using as template the genomic DNA from the wild type ͚1278b strain and as primers the appropriate oligonucleotides flanked by a BamHI site for the 5Ј-end and a XhoI site for the 3Ј-end. This DNA fragment was digested by BamHI and XhoI and inserted into the pAS2 vector, then digested by BamHI and SalI, leading to plasmid pGBD-CAR1, and inserted into the pACTII vector, then finally digested again by BamHI and XhoI, leading to plasmid pGAD-CAR1. In these GBD-CAR1 and GAD-CAR1 fusions, we determined the nucleotide sequence to ensure that the fusions were in-frame and that no mutation had been introduced during the PCR procedure.

Selection of Mutations in the ARG3 Gene by the "Reverse Two-hybrid" Assay
Plasmid pGAD-ARG3 was used to transform E. coli strain XL1-Red (Stratagene) to obtain mutations in the ARG3 gene, because this strain is highly mutagenic. Plasmids containing the mutated GAD-arg3 gene were extracted from a pool of bacterial transformants and used to transform yeast strain MaV99 containing plasmid GBD-CAR1. This strain is suitable for the use of the reverse two-hybrid assay, because it contains the URA3 reporter gene under the control of the UAS GAL sequences. The URA3 gene allowed the selection of mutated GAD-OTCases having lost their interaction with GBD-arginase, because only the strains that did not express the URA3 gene could grow on a medium containing 5-fluoroorotic acid plus 25 g of uracil. 15 candidates were obtained, and their DNA was extracted and used to transform E. coli strain HB101 (leu Ϫ , pro Ϫ , amp s ). This allowed the selection of plasmids containing a functional LEU2 gene and the GAD-ARG3 gene, and eliminated the pGBD-CAR1 plasmid.

Enzyme Assays
␤-Galactosidase activity was assayed as described by Miller (20). Protein contents were determined by the Folin method. Arginase activity was assayed as described previously (5). OTCase activity was determined by citrulline detection (21). Routinely the OTCase activities were measured in the presence of 50 mM Tris-HCl buffer, pH 8.0, 5 mM carbamoyl phosphate and at ornithine saturation. The reaction was carried out at 30°C for 30 min and was stopped by addition of HCl at a final concentration of 0.5 M. In the experiments of OTCase inhibition by arginase, 1 mM agmatine, an arginine analog, and increasing amounts of arginase (up to 1000 arginase units) were added in the incubation mixture. This enzyme concentration corresponded to an amount of wild type arginase 4-fold higher than the amount required for maximal inhibition of OTCase. Agmatine was used in the OTCase-arginase binding assays instead of arginine to avoid arginine degradation at high arginase concentration. The experiments shown in Fig. 5 (see below) were performed using the more sensitive assay of Prescott and Jones (22), for which standard curves for citrulline were generated in each experiment. Specific activities were expressed in micromoles of citrulline formed per hour and per milligram of protein.

Modeling of Yeast OTCase
Alignment of OTCase amino acid sequences of S. cerevisiae, S. pombe, human, Pyrococcus furiosus, E. coli, and Pseudomonas aeruginosa has been performed with the ClustalW program (23, 24). Based on this multiple sequence alignment, the structure of E. coli was selected as template to construct a three-dimensional model of S. cerevisiae OTCase. Assignment of the coordinates, construction of addi-tional loops (residues 45-50 and residues 160 -169 in the yeast sequence), and minimization of the structure were carried out using the Modeler4 program (25). The quality of the model has been checked using the Procheck program (26). The geometry of the final model complies with statistical criteria. About 98% of the residues fall in allowed regions (86.2% and 11.5% of phi/psi angles are, respectively, in the most favored or additional allowed regions) of the Ramachandran plot, and all bond lengths and valence angles corresponded to expected values. Graphical outputs were generated with Molscript (27). All calculations were carried out on an O2 SGI workstation operated with Irix 6.2. The coordinates of the model are available upon request to the authors.

Gel Filtration on the Superdex 200 Column
The gel filtration procedure was carried out at 4°C. A Superdex 200 column (1.6 ϫ 60 cm, Amersham Biosciences) was equilibrated with 50 mM Tris-HCl buffer, pH 8.0, containing either 15 mM ornithine or a concentration stated in the legends of Figs. 2 and 3C, and 1 mM agmatine. This column was pre-calibrated with serum albumin (molecular mass, 66 kDa), E. coli OTCase (molecular mass, 120 kDa), Enterococcus faecalis OTCase (molecular mass, 240 kDa), and P. furiosus OTCase (molecular mass, 420 kDa). Elution was carried out with the same buffer, at a flow rate of 0.5 ml/min, and 0.56-ml fractions were collected. Activities of OTCase and arginase were determined in each fraction as described.

DNA Manipulation and DNA Sequencing
Restriction reactions were performed as recommended by the enzyme supplier. Plasmid DNA was prepared using the alkaline lysis method (28). Double-stranded DNA was prepared using Qiagen columns. DNA was sequenced by the Thermosequenase radiolabeled terminator cycle sequencing kit (Amersham Biosciences). Site-directed in vitro mutagenesis was performed with the QuikChange site-directed mutagenesis kit from Stratagene.

Selection of a Mutated OTCase Resistant to Inhibition by
Arginase Using the Reverse Two-hybrid Assay-As a first attempt to select mutations in OTCase that impair its association with arginase, we took advantage of assays that had been developed in yeast (13,29). We first tested whether OTCase and arginase do interact in vivo using the two-hybrid assay (29). Therefore, we fused OTCase and arginase coding sequences to the DNA binding domain (GBD) and the activation domain (GAD) of the Gal4 protein, leading to the chimeric proteins GBD-OTCase, GAD-OTCase, GBD-arginase, and GAD-arginase (see "Experimental Procedures"), which were tested in vivo by the two-hybrid assay. The results indicated a significant interaction between GBD-OTCase and GAD-arginase and a stronger interaction between GBD-arginase and GAD-OTCase (data not shown). We used the reverse two-hybrid assay to select mutants impaired in the interaction between the two proteins (see "Experimental Procedures"). The plasmids of strains growing on 5-fluoroorotic acid were extracted, purified, and used to transform yeast strain 12S16c for OTCase activity determination, and strain HY (containing plasmid GBD-CAR1) was used to confirm the loss of interaction of these mutated OTCases with arginase. Only one candidate, named M3, retained sufficient OTCase activity to sustain the growth of a strain deleted of the ARG3 gene. This residual OTCase activity was resistant to the inhibition by arginase, even at ornithine concentrations higher than 100 mM (data not shown). The GAD-arg3-M3 gene was sequenced and found to harbor a single replacement Asn for Asp at position 182, a highly conserved residue present in all OTCases (Fig. 1). The mutation was introduced in the ARG3 gene yielding plasmid pFV48 (pGAL10-arg3D182N, 2, URA3). The substitution re-TABLE I Characteristics of the mutated OTCases OTCase activity was assayed at 30°C in extracts from strain 02859d⌬arg3-car1 (ura3, leu2, trp1, arg3::kanMX4, car1⌬) transformed with pYeF2 plasmid containing wild type ARG3 gene or the mutated arg3 genes after growth at 30°C on M.am plus 2% galactose plus 25 g of leucine plus g of tryptophan plus 25 g of arginine. The assays were performed in 50 mM Tris-HCl buffer, pH 8, and in the presence of 5 mM CP. The ornithine concentration used in the inhibition tests varied from 15 mM up to 2-fold the concentration of the apparent K m of mutated OTCases for ornithine. At least three independent transformants were tested, and the standard error was typically 10-15% of the mean. The OTCase specific activity is expressed in micromoles of citrulline formed/h/mg of protein. sulted in a slight reduction of OTCase specific activity, a strong decrease of the apparent affinity of the mutant enzyme for ornithine and a total loss of inhibition of its activity by arginase, even at high ornithine concentration (200 mM instead of 15 mM) (Table I and Fig. 2). Moreover, OTCase-D182N had lost the capacity to interact with arginase, as shown by molecular sieving (Fig. 3). In the presence of ornithine and arginine, wild type OTCase and arginase formed a complex of about 200 kDa, whereas, in contrast, OTCase-D182N and wild type arginase were unable to form such a complex, even at a concentration of ornithine equivalent to 2-fold the apparent K m value ( Fig. 3 and Table I).
Construction of Additional Mutations in OTCase by in Vitro Mutagenesis-At present more than 60 OTCase amino acid sequences are available (SwissProt search, 2002), and the crystal structure of OTCases from P. aeruginosa, E. coli, P. furio-sus, and human are known (30 -33). Each polypeptide chain from these trimeric enzymes folds into two domains, a carbamoyl phosphate binding domain and an L-ornithine binding domain (Fig. 4A). As shown in Fig. 1, alignment of the amino acid sequences of these OTCases and those of S. cerevisiae and S. pombe shows large regions of identity but also reveals several regions where the sequences diverge significantly. The region from residue 161 to 168 in particular is unique to the S. cerevisiae OTCase. However, deletion of amino acids E166/ V167/N168/K169 (⌬ in Fig. 1) had no effect on the OTCasearginase interaction (data not shown). Surprisingly, the mutation D182N rendering OTCase insensitive to arginase occurs in a conserved stretch (Gly-181, Asp-182, Asn-184, and Asn-185) in which some amino acids might interact with ornithine, ac- Among the crystal structures of OTCases available in the literature, the structure of E. coli (PDB code 1a1s) was retained to construct a model of yeast OTCase by homology modeling (Fig. 4, B and C, see "Experimental Procedures"). According to this model, Asp-182 in yeast OTCase could interact with residues of the SMG loop (Ser-253, Met-254, and Gly-255) involved in the closure of the ornithine and carbamoyl phosphate domains in the presence of the substrates. We chose to modify a series of residues in the vicinity of Asp-182, within and around the SMG loop, in the CP binding site, and in the ornithine binding site. The following substitutions were introduced: T68G, E123S, E123A, G181R, N184Q, N185Q, C191M/C194N/ F197V, E256Q or E256A, K260R or K260A, K263R or K263A, K265R or K265A, K268R or K268A, C289S, L290S or L290Q, and Q294P (see "Experimental Procedures"). These residues are marked in the S. cerevisiae OTCase sequence in Fig. 1  (white residues boxed in black). Residue Thr-68 is located in the STRTR sequence known to be involved in the binding of CP in E. coli OTCase, whereas Asn-184, Asn-185, Glu-256, Cys-289, and Leu-290 are in yeast OTCase residues that are liable to bind ornithine, in addition to Asp-182. Residue Glu-123 is located in the CP binding domain but opposite to the SMG loop and could therefore make contact with residues in this loop.
The data presented in Table I show the specific activities of the mutated OTCases determined at saturation for both substrates, their apparent K m for ornithine, and their sensitivity to arginase, determined as in Fig. 2. In none of the mutated OTCases was the apparent K m for CP significantly modified, except for the protein bearing the T68G substitution, which increased the apparent K m for CP from 0.1 to 2.4 mM. According to Western blot analysis, the amounts of the mutated OTCases were comparable to those of the wild type OTCase, indicating that the decrease in enzyme specific activities did not result from a lack of protein stability or reduced enzyme production (data not shown). Moreover, molecular sieving experiments showed that the mutations did not affect the trimeric structure of the different OTCases (data not shown).
The mutants fell into several categories. Some mutations did not affect any of the enzyme properties: E123S or E123A, K260R, and Q294P. The simultaneous replacements of  by the corresponding residues present in the S. pombe OTCase (MNV) reduced the enzyme activity, as did the T68G substitution. However, these two mutations did not modify the sensitivity of the enzyme to arginase (Table I). The replacement G181R completely abolished the enzyme activity but did not compromise its interaction with arginase, a 200-kDa complex being formed with wild type OTCase (data not shown).
In contrast, the substitutions N184Q, N185Q, E256A, K263R, K263A, and C289S led to an important modification of the affinity of the mutant OTCases for ornithine: the specific activities in crude cell extracts were reduced, and the enzymes were insensitive to arginase inhibition, even at high ornithine concentration (up to 10-fold the apparent K m value). The effects were comparable to those observed for the D182N mutation selected in vivo. The E256A substitution caused a reduction of the apparent affinity of the enzyme for ornithine and impaired its interaction with arginase, whereas the E256Q substitution only reduced the specific activity of the enzyme. The K260A, but not K260R, replacement had no effect on the interaction of OTCase with arginase when the test was performed with 15 mM ornithine, but the interaction was reduced at lower ornithine concentrations, as shown in Fig. 5B (compared with Fig.  5A with wild type OTCase). In contrast to the mutations described above, the substitution K268A, but not K268R, resulted in an enhanced affinity of the enzyme for ornithine. The OTCase-K268A was extremely sensitive to inhibition by excess of ornithine as compared with the wild type enzyme and exhibited a higher sensitivity to arginase as shown in Fig. 5D. Interestingly, the two mutations K265A (Fig. 5C) and L290S rendered the OTCase less sensitive to arginase without affecting the apparent K m of the mutated enzymes for ornithine. The substitution L290S was not expected to affect the catalytic properties of yeast OTCase, because its main chain only is involved in the direct stabilization of CP and ornithine (33), whereas the side chain of leucine is probably involved in inhibition by arginase. As indicated in Table I, most of the mutated OTCases had lost the capacity to interact with wild type arginase in molecular sieving experiments. Moreover, for some mutated enzymes, D182N, E256A, and K265A, the reduction of inhibition by arginase was correlated with the loss of interaction between the two proteins using the two-hybrid assay (data not shown). Fig. 4B shows those amino acids that appear to be important for the association of OTCase with arginase, whereas Fig. 4C shows those residues that were found not to be involved.

Construction of Mutations in Arginase by in Vitro
Mutagenesis-The comparison of the available arginase amino acid sequences (S. cerevisiae, S. pombe, Rattus norvegicus, and Bacillus caldovelox) revealed the presence in the S. cerevisiae enzyme of several short sequences that are absent or different in the arginases devoid of regulatory properties (Fig. 6). Among the known arginase sequences, R. norvegicus and B. caldovelox arginases were chosen for this comparison because their crystal structures had been determined (35,36). Regions of interest in the S. cerevisiae arginase were those between amino acids 78 and 86 (indicated by the arrow in Fig. 6), and more interestingly, the C-terminal end of the enzyme, which contains several cysteine and histidine residues that could constitute a zinc finger-like element (white residues in black boxes). The group of P. Hensley (37) has demonstrated that the binding of a weakly bound Mn 2ϩ ion confers catalytic activity, whereas the binding of a more tightly associated Zn 2ϩ ion confers substantial stability to the tertiary and quaternary structure of the enzyme, thus possibly playing a role in the formation of the OTCasearginase multienzyme complex. As a first attempt to localize in the arginase residues involved in its interaction with OTCase, we generated four mutants. The deletion of amino acids 78 -86 (pAM4) and the replacements of H309K/H312E (pME209), C321R/C326T (pME210), and H309K/H312E/C321R/C326T (pME222) in the C-terminal end of arginase were constructed by in vitro mutagenesis in plasmid pEJ24 expressing the CAR1 gene under the GAL10 promoter, as described under "Experimental Procedures." It must be emphasized that the amino acid replacements mentioned above introduce the cognate residues present in S. pombe arginase. The plasmids obtained were used to transform yeast strain 02859d⌬arg3-car1 (arg3::kanMX4, car1⌬, leu2, trp1, ura3) on M.am-glucose supplemented with 25 g of arginine, 25 g of leucine, and 25 g of tryptophan, per milliliter. Arginase specific activities were measured in the transformants bearing either the wild type or the mutated CAR1 genes, after growth on M.am-galactose supplemented with 25 g of arginine, 25 g of leucine, and 25 g of tryptophan, per milliliter. Deletion of the 9-amino acid stretch (78 -86) unique to the S. cerevisiae arginase had no effect on arginase activity and did not modify its inhibitory capacity on OTCase activity (data not shown). Replacing the two cysteines completely abolished the regulatory function of arginase, which had lost the capacity to bind and inhibit OTCase. In contrast, the substitution of the two histidine residues had only a minor effect (Table II and Fig. 7A). These mutations did not alter significantly the enzymatic properties of the enzymes; their catalytic activity and their affinity for arginine were unaffected (Table II). Moreover, the simultaneous substitution of the two cysteines and the two histidines only led to a 50% reduction of the activity without affecting the K m of the enzyme for arginine. Such a reduction could result from a slight effect on enzyme stability. The loss of interaction between the arginase-C231R-C326T and the wild type OTCase was confirmed using the two-hybrid assay (data not shown) and by molecular sieving (Fig. 7, B and C). Addition of mercaptoethanol in the inhibition test (data not shown) did not affect the regulatory capacity of wild type arginase. Moreover, addition of 1 mM zinc acetate in the molecular sieving experiment performed with wild type OTCase and arginase-C321R-C326T did not restore the formation of an OTCase-arginase complex (data not shown). As shown in Fig. 7C, the molecular mass of the mutated arginase was comparable to that of the wild type enzyme (about 114 kDa, shown in Fig. 3), indicating that the mutated enzyme was still a trimer. The two cysteine residues are thus important for the epiarginase function, whereas the binding of zinc to the enzyme, although necessary to maintain its quaternary structure (37), does not require the two cysteines. The arginase from Pichia sorbitophila (sequence provided by the Genolevures project) (38), which contains the two cysteines, is unable to inhibit S. cerevisiae OTCase (data not shown). Cys-321 and Cys-326 are thus necessary but not sufficient to confer a regulatory function to arginase. DISCUSSION For the yeast OTCase, as for most of OTCases, the catalysis of carbamoylation was predicted to follow a preferred ordered Bi-Bi binding mechanism with carbamoyl phosphate (CP) being the first substrate bound and ornithine the second one (enzyme-CP-ornithine) (39,40). In such a kinetic mechanism, a high ornithine concentration could favor a slower pathway of central ternary complex formation (enzyme-ornithine-CP) leading to inhibition of OTCase activity. In agreement with such a mechanism, it was shown that increasing the concentration of the preferred substrate (CP) overcame the inhibition of OTCase by ornithine and reduced its inhibition by arginase (4). Ornithine binding to the free enzyme could promote a conformational change in the enzyme, which could potentiate its association with arginase. The crucial role of ornithine in this regulatory process is emphasized by the fact that carbamoyl phosphate cannot replace ornithine in the complex with arginase, 2 and mutating Thr-68 to Gly in the STRTR carbamoyl phosphate binding domain, which had a dramatic effect on the catalytic activity of the enzyme, had no effect on its interaction with arginase (Table I).
By mutation analysis we identified in S. cerevisiae OTCase two regions required for the OTCase-arginase interaction. Residues Asp-182, Asn-184, Asn-185, Cys-289, and Leu-290 are crucial for the regulation by arginase and are probably involved in the binding of ornithine, based on the structure of the E. coli OTCase bound to the bisubstrate analog PALO (33,34). In the E. coli enzyme, Asp-231, Ser-235, and Met-236 are part of the flexible SMG loop, the main binding site for ornithine. In yeast OTCase, Glu-256 located in the SMG loop could also participate in ornithine binding, because its replacement strongly reduced the apparent affinity of OTCase for ornithine to the same extent as the replacement of residues Asp-182, Asn-184, Asn-185, and Cys-289. In contrast, the four lysine residues, Lys-260, Lys-263, Lys-265, and Lys-268 in the SMG loop were not reported to bind PALO in E. coli and human OTCases but appear to play an important role in the interaction between OTCase and arginase. Changing Lys-260, Lys-265, or Lys-268 to arginine had no effect, but their replacement by alanine modified the interaction with arginase attesting the importance of a basic residue at those positions. OTCase-K260A was only slightly less sensitive to arginase at low ornithine concentration, whereas OTCase-K265A had lost 50% of the inhibition capacity, but exhibited a normal K m for ornithine. In contrast, OTCase-K268A had a better affinity for ornithine than the wild type enzyme and, consequently, was much more sensitive to arginase at an ornithine concentration below 1 mM. Residue Lys-263 is of particular interest. When mutated to arginine,

TABLE II
Characteristics of the mutated arginases Arginase activity was assayed in 30°C in extracts from strain 02859d⌬arg3-car1 (ura3, leu2, trp1, arg3::kanMX4, car1⌬) transformed with pYeF2 plasmid containing wild type car1 gene or the mutated car1 genes after growth at 30°C on M.am plus 2% galactose plus 25 g of leucine plus 25 g of tryptophan plus 25 g of arginine. At least three independent transformants were tested and the standard error was typically 10-15% of the mean. The arginase specific activity is expressed in Micromoles of urea formed/h/mg of protein.  the amino acid present at that position in many other OTCases, the apparent K m of the yeast enzyme for ornithine dropped about 15-fold, rendering OTCase-K263R insensitive to arginase. Thus Lys-263 may constitute one of the key residues that differentiates S. cerevisiae OTCase from other OTCases for its inhibition by arginase. The four lysines, Lys-260, Lys-263, Lys-265, and Lys-268, appear to constitute contact points with arginase, when the latter enzyme binds arginine or agmatine, inducing a modification of the catalytic site conformation and reducing the catalytic activity of OTCase. In E. coli OTCase, the SMG loop undergoes a large conformational change upon substrate binding. It is possible that the yeast enzyme undergoes the transition from the open to the closed states with more ease than other OTCases and that the binding of arginase enhances this effect. Lysine 268 seems to restrain the transition, because K268A substitution results in an improved affinity for ornithine and an enhanced sensitivity to excess ornithine and to inhibition by arginase. Whether ornithine acts directly on these lysine residues remains questionable, but they could be involved in transducing and enhancing the signal given by ornithine for catalytic domain closure.
The regulatory function of arginase requires at least the residues Cys-321 and Cys-326. The group of Hensley (37) had proposed that these two cysteines form a putative "Zinc fingerlike" element with the residues His-309 and His-312 and that this structure could be involved in stabilizing subunit interactions and, thus, play a role in the formation of the OTCasearginase complex. However, modifying these four residues had no strong effect on the catalytic activity of arginase and did not modify the trimeric structure of the enzyme, suggesting that these amino acids are not crucial for the zinc-dependent enzyme stability. Moreover, few known arginases contain the two histidine and cysteine residues in their C-terminal end. Our data demonstrate that the two cysteine residues are necessary for the epiarginase function of the yeast arginase but that they are not sufficient, because the arginase from P. sorbitophila containing these two amino acids has no regulatory capacity. Based on sequence alignments and on crystal structures, no suitable templates are available to model the C-terminal part of the protein.
It is worth noting that the S. cerevisiae arginase contains two arginine binding sites (41) that might not be present in arginases of other species. Thus, determining the crystal structure of this unusual arginase and of the OTCase-arginase complex, with and without their effectors, will help to unravel this intricate mechanism of enzyme regulation by protein-protein association.