Subunit-Subunit Interactions in Trimeric Arginase

The structure of the trimeric, manganese metalloenzyme, rat liver arginase, has been previously determined at 2.1-Å resolution (Kanyo, Z. F., Scolnick, L. R., Ash, D. E., and Christianson, D. W., (1996) Nature383, 554–557). A key feature of this structure is a novel S-shaped oligomerization motif at the carboxyl terminus of the protein that mediates ∼54% of the intermonomer contacts. Arg-308, located within this oligomerization motif, nucleates a series of intramonomer and intermonomer salt links. In contrast to the trimeric wild-type enzyme, the R308A, R308E, and R308K variants of arginase exist as monomeric species, as determined by gel filtration and analytical ultracentrifugation, indicating that mutation of Arg-308 shifts the equilibrium for trimer dissociation by at least a factor of 105. These monomeric arginase variants are catalytically active, with k cat/K m values that are 13–17% of the value for wild-type enzyme. The arginase variants are characterized by decreased temperature stability relative to the wild-type enzyme. Differential scanning calorimetry shows that the midpoint temperature for unfolding of the Arg-308 variants is in the range of 63.6–65.5 °C, while the corresponding value for the wild-type enzyme is 70 °C. The three-dimensional structure of the R308K variant has been determined at 3-Å resolution. At the high protein concentrations utilized in the crystallizations, this variant exists as a trimer, but weakened salt link interactions are observed for Lys-308.

the average individual excretes about 10 kg of urea per year (1) (Scheme 1).
Liver arginase, also designated type I arginase, has been isolated from human and rat liver, and both enzymes have been expressed in Escherichia coli (2,3). The physiological activator is Mn(II), and the fully active rat liver enzyme contains 2 Mn(II)/subunit (4,5). The crystal structure of type I arginase from rat liver has been determined at 2.1-Å resolution (4). The key features of this structure include a trimeric quaternary structure, a binuclear manganese cluster, and an S-shaped tail composed of 19 amino acids at the C terminus (Fig. 1). The S-shaped tail at the C terminus of each monomer is located at the subunit-subunit interface (4), and 54% of the intersubunit interactions (van der Waals, hydrogen bonds, and salt links) are mediated by this region of the protein (1,4).
In humans, type I arginase deficiency is a rare disorder that results in hyperargininemia characterized by episodic hyperammonemia and growth retardation, among other complications (6 -10). Hyperargininemia is a heterogeneous disease resulting from point mutations throughout the type I arginase gene (11)(12)(13)(14)(15)(16)(17). An R291X mutation has been identified that would lead to the production of a truncated protein lacking the final 32 amino acids at the C terminus of the 323 amino acid subunit. The hyperargininemia patient with this mutation has both low levels of arginase activity and low levels of immunocross-reactive material (1,11). This latter finding suggests that the R291X mutant may be characterized by decreased stability that results from loss of the S-shaped tail. This S-shaped tail appears to stabilize and maintain the native oligomeric state of the enzyme via a network of salt bridges nucleated by Arg-308. The Arg-308 residue forms an intramonomer salt bridge with Glu-262 and an intermonomer salt bridge with Asp-204 on the adjacent subunit (4). Mutations in this region may impact the oligomeric structure and stability of the enzyme as well as its kinetic properties.
The goal of the present study has been to utilize gel filtration, enzyme activity assays, EPR, analytical ultracentrifugation, differential scanning calorimetry, circular dichroism, and x-ray crystallography to elucidate the role of Arg-308, located at the S-shaped tail, in maintaining the overall oligomeric state and kinetic properties of rat liver arginase. 14 C]Arginine (specific activity 1.9 GBq mmol Ϫ1 ) was from PerkinElmer Life Sciences. isopropyl ␤-D-thiogalactopyranoside was from FisherBiotech. The Superdex-200 HiLoad 16/60 FPLC 1 chromatography column was from Amersham Pharmacia Biotech. Reactive Red-120 dye ligand chromatography media, Chelex-100, EDTA, L-arginine, L-ornithine, MnCl 2 , and all protein standards for gel filtration were from Sigma, except for transcription factor Rho, which was a gift from Dr. Barbara L. Stitt (Temple University). Synthetic oligonucleotide primers were prepared by Ransom Hill Bioscience. The QuikChange site-directed mutagenesis kit and E. coli BL-21(DE3) competent cells were from Stratagene. The DNA purification kit used was from Qiagen. Restriction enzymes were purchased from New England Biolabs. Centricon-30 microconcentrators were obtained from Amicon. Ecoscint scintillation solution was from National Diagnostics. Dialysis tubing was purchased from BioDesign of New York, Inc. All other reagents were of highest quality commercially available.

Materials-L-[guanidino-
Site-directed Mutagenesis-Using the QuikChange site-directed mutagenesis kit, mutations of Arg-308 to alanine, glutamate, or lysine were performed using the following oligonucleotides as mutagenic primers: 5Ј-GTCTTGTTTTGGAACGAAAGCTGAAGGTAATCATAAGCCA-G-3Ј and 5Ј-CTGGCTTATGATTACCTTCAGCTTTCGTTCCAAAACAA-GAC-3Ј as the forward and reverse primers, respectively, for the Arg-308 to alanine mutation. Wild-type 2 arginase cDNA in a pRSET-C T7-based expression vector (Invitrogen) was used as the DNA template for this site-directed mutagenesis procedure. The underlined codon, coding for arginine, was changed to GAA and TTC in the forward and reverse directions, respectively for the Arg-308 to glutamate mutation. For the Arg-308 to lysine mutation, the underlined codon was changed to AAA and TTT in the forward and reverse directions, respectively. The mutant constructs were used to transform BL-21(DE3) cells. Plasmid DNA was isolated for each mutant construct using a Qiagen kit and analyzed by restriction mapping with NdeI and PstI, enzymes known to have unique restriction sites located outside the arginase cDNA. Positive constructs for each of the R308A, R308E, and R308K mutants, which had been identified by restriction mapping, were sequenced at the DNA Sequencing Facility at the University of Pennsylvania to verify that only the specifically introduced mutations were present and that no other mutations had been introduced during polymerase chain reaction.
Overexpression and Purification of Mutant Enzymes-The constructs for the R308A, R308E, and R308K arginase variants were used to transform competent BL-21(DE3) cells. Expression of the mutant proteins by the BL-21(DE3) cells was induced by the addition of isopropyl ␤-D-thiogalactopyranoside (0.2 mg/ml final concentration) for 3 h after the cell culture had reached an A 600 of 0.8. The cells were then harvested by centrifugation at 5000 ϫ g for 35 min and the cell pellet was stored at Ϫ70°C overnight. The cells were then thawed and resuspended in 50 mM HEPES/KOH at pH 7.5, and the arginase variants were purified from the lysed cells by a modification of the method described by Cavalli et al. (2). The partially purified protein samples were chromatographed on a Sigma-Red dye ligand column (3.5 ϫ 25 cm), instead of the Amicon-Green dye ligand column described previously (2) and eluted with a 0 -0.3 M linear gradient of KCl in 50 mM HEPES/KOH at pH 7.5. The purity of the proteins was determined by SDS-polyacrylamide gel electrophoresis under reducing conditions. All arginase samples were stored at 4°C in ammonium sulfate at 85% saturation.
Enzyme Assay-The mutant R308A, R308E, and R308K enzymes and the recombinant wild-type enzyme were assayed for arginase activity by a modification of the method of Rü egg and Russell (18). EPR-The arginase samples, stored as ammonium sulfate suspensions, were centrifuged at 14,000 rpm in a microcentrifuge for 20 min at 4°C. The pellets were resuspended in 1 ml of 50 mM HEPES/KOH at pH 7.5 containing 10 mM MnCl 2 and then incubated at 60°C for 10 min. The samples were further centrifuged for 20 min at 4°C at 14,000 rpm in a microcentrifuge to remove any precipitate that might have formed during the heat activation step. The supernatants were dialyzed exhaustively in 50 mM HEPES/KOH, pH 7.5, at 4°C. The samples were then washed six times with Chelex-100-treated 50 mM HEPES/KOH at pH 7.5 in Centricon-30 microconcentrators (2 ml in each wash) and then concentrated to ϳ3 mg/ml each by centrifugation at 5000 ϫ g at 4°C for 30 min. The samples were then diluted 1:1 to a total volume of 400 l with 6% perchloric acid. Manganese standards were prepared in HEPES buffer and treated with perchloric acid as described above for the enzyme samples. EPR samples were placed in quartz tubes and spectra were recorded on a Bruker ER-200D-SRC spectrometer. All EPR spectra were recorded at 9.47 GHz, 100-kHz modulation frequency, 10-Gauss modulation amplitude, 1000-Gauss sweep width, 90-s sweep time, 0.164-s time constant, and a temperature of 300 K. The manganese content for each arginase sample was determined by a comparison of EPR spectral amplitudes for the enzyme samples with a standard curve generated by plotting spectral amplitudes versus manganese concentration for the solutions of known manganese concentration (5). The stoichiometry calculations are based on a subunit molecular mass of 35,000 Da.
Gel Filtration-The protein samples were prepared by exhaustive dialysis of the ammonium sulfate-suspended proteins in 50 mM HEPES/ KOH at pH 7.5. The samples were then concentrated to 12 mg/ml in Centricon-30 microconcentrators by centrifuging at 5000 ϫ g (4°C) for 30 min. The Superdex-200 column was equilibrated with 50 mM HEPES/KOH, 150 mM KCl at pH 7.5, at a flow rate of 1.5 ml/min. 100-l protein samples were chromatographed on the Superdex column using an Amersham Pharmacia Biotech FPLC system. Chromatography of the protein solutions was also performed in 50 mM HEPES/KOH, 150 mM KCl at pH 7.5 that contained 20 mM arginine and 20 mM MnCl 2 . The Superdex column was standardized using gel filtration standards from Sigma as well as transcription factor Rho. The molecular mass for each arginase sample was determined using a standard curve generated from the elution volumes for transcription factor Rho (282,000 Da), ␤-amylase (200,000 Da), glucose-6-phosphate dehydrogenase (110,000 raphy; CHES, 2-(cyclohexylamino)ethanesulfonic acid; DSC, differential scanning calorimetry. 2 Wild-type arginase refers to the recombinant rat liver enzyme expressed in and purified from Escherichia coli. Analytical Ultracentrifugation-Sedimentation equilibrium analytical ultracentrifugation of wild-type and mutant arginases was carried out using a Beckman XL-I analytical ultracentrifuge. Absorbance optics at 280 nm were used to measure the concentration of arginase. Enzyme was prepared for ultracentrifugation by exhaustive dialysis in 50 mM HEPES/KOH, pH 7.5, to ensure that the samples and blanks were of identical composition. Arginase concentrations for the analytical ultracentrifugation experiments ranged from 0.1 to 0.3 mg/ml. The centrifuge was operated at 20,000 rpm, 20°C, and data were collected over a 40-h period with 2-h sampling intervals. Equilibrium was reached when no further difference in the exponential distribution data was observed between two consecutive data acquisitions. Equilibrium data were analyzed with software written using IGOR (WaveMetrics, Lake Oswego, OR). A partial specific volume of 0.743, determined using the SEDNTERP program (19,20), was used in the calculation of the species concentration in solution. All calculations were based on the molecular composition of a single subunit.
Differential Scanning Calorimetry (DSC)-DSC data were measured with a Calorimetry Sciences Corp. NanoDSC instrument. Unfolding scans were taken at a rate of 1 degree/min, and unfolding was found to be irreversible based on the absence of transitions in reheating scans. Arginase concentrations were determined spectrophotometrically at 280 nm using a molar extinction coefficient of 24,600 M Ϫ1 calculated from the amino acid composition (21). DSC data were corrected for the instrumental base line by subtraction of water versus water scans. Final excess heat capacity scans were produced by subtracting linearpolynomial base lines using Calorimetry Sciences Corp. CpCalc software. Apparent unfolding enthalpy changes and midpoint temperatures were determined by integration of each unfolding peak.
CD-Circular dichroism experiments were conducted with a Jasco 710 spectropolarimeter. Prior to each experiment, arginase was bufferexchanged into 50 mM HEPES/KOH, pH 7.5. All experiments were conducted in the exchange buffer at 23°C, and data were collected in the range of 260 -200 nm. Data analyses were carried out using the Jasco system software.
Crystallography-R308K arginase was crystallized using conditions similar to those required for crystallization of the native enzyme (22). Crystals were nearly isomorphous with those of the native arginase trimer and belonged to space group P3 2 (a ϭ 87.9 Å, b ϭ 87.9 Å, c ϭ 112.1 Å; one trimer per asymmetric unit). Prior to data collection, crystals were gradually transferred to a cryoprotectant solution containing 30% glycerol and then flash-cooled with liquid nitrogen. X-ray diffraction data were collected using a ADSC Quantum-4 CCD detector at Brookhaven National Laboratory, beamline X12B, at the National Synchrotron Light Source (Upton, NY). Data integration and reduction were performed using the HKL suite of programs (23). With 69,842 measured reflections yielding 18,649 unique reflections, the completeness of data was 96% in the 20.0 -3.0-Å range with R sym ϭ 0.089. Initial phasing was achieved by molecular replacement using AMoRe (24,25) using the structure of the native arginase trimer (4) less the atoms of the variant side chain and the Mn 2ϩ ions as a search probe. Using intensity data in the 20 -3.5-Å shell with I Ͼ 3, the crossrotation search yielded a 26.3 peak with ␣ ϭ 117.17°, ␤ ϭ 168.52°␥ ϭ 337.74°. A subsequent translation search yielded a peak at 63.7 with fractional coordinates x ϭ 0.3333, y ϭ 0.3385, z ϭ 0.0000. Rigid body refinement lowered the initial crystallographic R factor from 0.438 to 0.411. Further refinement against 3.0-Å data was achieved using tor-sion angle molecular dynamics (T initial ϭ 3000 K) with energy minimization, using the maximum likelihood algorithm as implemented in CNS (26). Strict noncrystallographic symmetry constraints were employed in refinement. After initial rounds of refinement, electron density for the variant R308K side chain was clear and unambiguous. The Lys-308 side chain, 2 Mn 2ϩ ions, and 10 water molecules per monomer were modeled into electron density maps generated with Fourier coefficients 2ԽF o Խ Ϫ ԽF c Խ and ԽF o Խ Ϫ ԽF c Խ and phases calculated from the in-progress atomic model. Iterative rounds of refinement and rebuilding were performed with CNS and O, respectively (26,27). Individual B factors were refined, and a bulk solvent correction was applied. Refinement converged to a final crystallographic R factor of 0.263 (R free ϭ 0.296) with excellent stereochemistry; the root mean square deviations from ideal bond lengths and angles were 0.008 Å and 1.4°, respectively. The atomic coordinates of R308K arginase have been deposited in the Research Collaboratory for Structural Bioinformatics with accession code 1HQX.

RESULTS
Purification of Arginase Variants-The mutant arginases were purified to apparent homogeneity using the standard protocol for the wild-type enzyme. No modifications of the protocol were required.
Kinetic Constants-The kinetic constants for wild-type arginase and the R308A, R308E, and R308K mutant arginases are summarized in Table I. The wild-type protein had a K m of 1.1 mM for L-arginine, in good agreement with the values of 1-1.7 mM reported for rat liver arginase (2,5,28,29). The mutant arginases showed K m values for L-arginine that were about 2.5-fold higher compared with wild-type enzyme. The catalytic activities (as k cat ) of all the mutants ranged between 33 and 41% of that for the wild-type enzyme, and the k cat /K m values ranged between 13 and 17% of that for wild-type enzyme.
Manganese Stoichiometry-Because an intact binuclear Mn(II) center is essential for maximal catalytic activity (2,5), it is possible that the decreased activity of the R308A, R308E, and R308K mutants relative to wild-type enzyme resulted from altered manganese stoichiometries for the mutant enzymes. The Mn(II) stoichiometries for wild-type and the mutant arginases were determined by EPR spectroscopy using samples that were heat-activated in the presence of Mn(II) and then washed exhaustively to remove excess Mn(II). The results of the Mn(II) determinations are summarized in Table I. The wild-type enzyme had a full complement of 2 Mn(II)/subunit, in agreement with previously published stoichiometries (2,5). Similar Mn(II) stoichiometries were also determined for the R308A and R308E mutants. A slightly lower stoichiometry of ϳ1.3 Mn(II)/subunit was found with the R308K mutant.
Molecular Weight Determinations-The native molecular weights for wild-type and the mutant arginases (Table I) were determined by gel filtration utilizing a Superdex-200 column. The molecular weight for each arginase protein was calculated from a standard curve generated with proteins of known mo-  Fig. 2A. The elution profiles for wild-type and the mutant arginase R308A are shown in Fig. 2B. The wild-type protein eluted as a single peak corresponding to the trimeric molecular weight, while the R308A mutant eluted as a single peak with a molecular weight consistent with a monomeric species. The protein profiles for the R308E and R308K mutant arginases were the same as the R308A mutant shown in Fig.  2B. The monomeric state of the mutants did not change when chromatography was performed in the presence of added arginine and MnCl 2 (data not shown). Wild-type arginase was determined to have a molecular mass of about 94 -95,000 Da, which is consistent with previously reported molecular masses determined by gel filtration for the trimeric rat and human liver arginases (30,31). In contrast, the molecular masses determined for the mutant arginases were about 30,000 -34,000 Da, consistent with the monomeric molecular mass for rat liver arginase (31). Analytical Ultracentrifugation Analysis-Equilibrium sedimentation analytical ultracentrifugation data (Fig. 3) for wildtype arginase were best fit to a trimeric species only. Because no significant amount of monomer could be fit using the wildtype data, the monomer-trimer equilibrium constant for this interaction could not be determined directly. Allowing the fitting program to fit for molecular weight rather than for oligomeric state resulted in a mass of 104,500 Da, matching that of trimeric arginase (31). Mutant arginases R308A, R308E, and R308K were found only in monomeric form. Introduction of a trimeric species into the fitting analysis resulted in very large error values and poor fit parameters. Fitting these data for molecular mass gave results indicative of monomeric species, with values of 35,000 Ϯ 100 Da. Fig. 4 shows DSC data measuring the thermal stabilities of wild type and the mutant forms of arginase. Some pretransition base-line variability is seen between the different temperature scans. This variability is instrumental in origin and is related to the low protein concentrations (0.4 mg/ml) used. All three mutant forms of arginase unfold with very similar apparent unfolding enthalpy changes of 190 Ϯ 15 kcal/ mol of subunit and similar midpoint temperatures of 65.5°C (R308A), 64.1°C (R308E), and 63.6°C (R308K). In contrast, wild-type arginase unfolds at a higher temperature (70.0°C) and also with a much larger enthalpy change of 280 kcal/mol of subunit. The additional thermal stabilization observed for wild-type arginase is at least partly due to the intersubunit, trimerization bonding energy (32,33). These observed unfolding enthalpy changes agree with typical unfolding enthalpy changes observed for globular proteins (34), although, since unfolding of arginase was irreversible, the thermodynamic parameters may have contributions from side reactions such as aggregation in the unfolded form.

DSC-
CD-To evaluate the integrity of the secondary structure of the arginase mutants, the circular dichroic spectra of the wildtype enzyme and the arginase variants were compared. The CD spectra of wild-type and mutant arginases were nearly identi- cal over the wavelength range examined (data not shown). Minor differences in amplitude were observed, arising from slight variations in enzyme concentration, but the overall shapes of the curves were identical. These findings indicate that there are no gross structural changes, particularly in ␣-helical structure, occurring as a result of the mutagenesis, and this conclusion is consistent with the subsequently determined x-ray crystal structure of R308K arginase.
Crystallography-Overall, the refined structure of R308K arginase is very similar to that of the native enzyme; the root mean square deviation of 314 C␣ atoms in the monomer is 0.38 Å. Although the trimeric quaternary structure of this variant is destabilized by the R308K substitution, the trimer is sufficiently stable at the high protein concentrations (14 -18 mg/ml) employed for crystallization. The electron density map in Fig. 5 reveals clear and unambiguous density for the Lys-308 side chain, and the comparison of wild-type and R308K arginases in Fig. 6 reveals the structural basis for trimer destabilization, which is discussed further below.
Comparison of the active sites of trimeric wild-type and R308K arginases reveals no structural changes (data not shown). In R308K arginase, 2 Mn 2ϩ ions are bound at the base of the active site cleft (Mn 2ϩ -Mn 2ϩ separation ϭ 3.4 Å) in identical fashion observed in the wild-type enzyme (Mn 2ϩ -Mn 2ϩ separation ϭ 3.3 Å (4)). Although this Mn 2ϩ stoichiometry does not agree with that determined by EPR, 5 mM MnCl 2 was included in the crystallization solutions. This concentration of Mn 2ϩ is sufficient to saturate the Mn 2ϩ binding sites on the R308K enzyme. The nonprotein ligand, a bridging hydroxide ion, is not visible in the electron density map. This is probably a consequence of the moderate resolution of the map e.g. as found in the structure determination of H101N arginase (35). DISCUSSION The present analytical ultracentrifugation and gel filtration studies, as well as results from x-ray crystallography (4), have shown that both the native and recombinant forms of rat liver arginase exist as trimeric species. An unusual "S-shaped oligomerization motif," which begins at Phe-304 and extends to the C terminus of the 323-amino acid polypeptide chain, was identified in the crystal structure of the enzyme. Since ϳ54% of intermonomer contacts are mediated via this S-shaped tail (4), it has been suggested that this motif is critical for maintaining the trimeric form of arginase (1). The sequence of this motif is highly conserved between rat, human, mouse, and Xenopus arginases as shown in Table II, suggesting a similar role in stabilizing the oligomeric states of these enzymes. In a recent report, the role of the S-shaped tail in stabilizing the trimeric human arginase was examined (30). No loss of enzyme activity and no change in the state of oligomerization were observed for a human arginase variant that lacks the final 14 amino acids of this motif but retains Arg-308 as the C-terminal amino acid. Thus, Mora et al. (30) conclude that the S-shaped tail is not important for maintaining the structural integrity of arginase or for optimal activity. Examination of the crystal structure, however, indicates that Arg-308 plays a critical role in nucleating a series of intra-and intersubunit interactions. In addition, this residue is conserved in the rat, human, mouse, and Xenopus arginase sequences.
In contrast to the study of Mora et al. (30), the results presented here have shown that retention of the S-shaped tail and mutation of a single amino acid, Arg-308, results in a shift in the oligomeric state of arginase from a trimeric species to a monomeric species, as determined by analytical ultracentrifugation and gel filtration. With the present data, it is not possible to determine equilibrium constants for the dissociation of the wild-type trimer into monomers (since only trimer was observed under all experimental conditions) or for the dissociation of mutant trimers into monomers (since only monomeric species were observed under the experimental conditions). However, limits on K eq for the dissociation of wild-type trimers or mutant trimers to form monomers can be estimated from the data in Fig. 3. For the wild-type enzyme, assuming detection limits that would permit observation of 5% of the total protein as the monomeric form, an upper limit of 8 ϫ 10 Ϫ14 M 2 can be estimated for K eq . Similar considerations lead to a lower limit estimate of 4 ϫ 10 Ϫ8 M 2 for K eq in the dissociation of mutant trimers. Although these are rough estimates, it is clear that mutation of Arg-308 shifts the equilibrium constant for trimer dissociation by at least a factor of 10 5 .
The monomeric forms of arginase, R308A, R308E, and R308K, are all catalytically active. Previous studies using nonmutagenic techniques such as acid or EDTA treatment (36) and glutaraldehyde cross-linking to a column support (37)  relative to the wild-type enzyme and ϳ3-fold decreases in k cat (Table I). In addition, the mutant enzymes show decreased thermal stability, particularly at low protein concentrations (data not shown). The three monomeric mutant forms of arginase unfold with very similar unfolding enthalpy changes and temperature midpoints as determined by DSC (Fig. 4). However, the native trimeric form of arginase is much more stable with respect to unfolding. Direct comparison of the thermal stabilities of the monomeric mutant arginase molecules to the trimeric native molecule is somewhat complicated by the predicted concentration dependence in the case of the trimer. In fact, for reversibly associated oligomeric proteins in general, the unfolding thermodynamics include a translational entropy component corresponding to conversion from folded oligomer to unfolded monomers (38). The difference in folding stabilities of the monomeric and trimeric forms is consistent with very tight trimerization interactions of the native form (33). However, because arginase unfolding was irreversible, it is not possible to calculate a trimerization constant from the DSC data using an equilibrium theory.
The activity differences between monomeric mutants and trimeric wild type cannot be attributed to decreased binding of the essential Mn(II), since both the R308A and R308E mutants contain a full complement of Mn(II). In contrast, the R308K mutant consistently displayed manganese stoichiometries of 1.25 Mn(II)/subunit but catalytic activities that were comparable with the other two variants. The origin of this discrepancy is unclear at present, and further Mn(II) EPR studies may be required to establish the distribution of Mn(II) between mononuclear and binuclear sites in this variant.
Similarly, it is unlikely that the mutations result in any gross conformational changes in structure, since each of the variants has considerable catalytic activity, and the CD spectra for the variants are very similar to that for the wild-type enzyme. It is more likely that the mutations have resulted in small perturbations in active site structure that have altered the catalytic activity as well as stability of the mutant enzymes.
In this regard, it is interesting to note that the crystal structure for the arginase from Bacillus caldovelox does not show an S-shaped tail analogous to that found in rat liver arginase (39). This arginase, like many arginases from bacterial sources, is hexameric (a dimer of trimers) rather than trimeric, and the amino acid corresponding to Arg-308 in the rat enzyme is not conserved in the B. caldovelox enzyme. However, crystals of this enzyme grown in the presence of guanidinium hydrochloride have a guanidinium ion bound at the subunit-subunit interfaces in the trimer. Similarly, crystals grown in the presence of EDTA, to remove one of the catalytic metals, and arginine show two arginine binding sites. One molecule of arginine is bound at the active site, while the second is located at the subunit-subunit interface. In these structures, the guanidino group is stabilized by bidentate hydrogen bonding to Glu-256 (B. caldovelox numbering system) in one monomer and by a bifurcated hydrogen bond to Asp-199 of a second monomer. Both Glu-256 and Asp-199 are conserved in rat liver arginase (Glu-262 and Asp-204 in the rat numbering system) and form similar interactions with the guanidino moiety of Arg-308. Superposition of the structures for the rat liver enzyme and the B. caldovelox enzyme reveals that the guanidino group of Arg-308 in the rat enzyme superimposes with the guanidino group of the guanidinium ion or arginine in the B. caldovelox enzyme (39).
Comparison of crystal structures for the B. caldovelox enzyme determined in the presence or absence of guanidinium hydrochloride reveals that the binding of the guanidinium ion results in the rearrangement of a loop region composed of residues 12-20. Since these residues form part of the rim of the active site and their movement has a significant impact on other active site residues, it has been proposed that binding of the guanidinium ion or arginine to the external site at the monomer-monomer interface serves to activate the B. caldovelox enzyme. A similar role may be envisaged for Arg-308 in the rat liver enzyme, that of stabilizing the active conformation of the enzyme. Additional insights are provided by examination of the crystal structure of the mutant arginase R308K refined at 3.0 Å. In wild-type arginase, Arg-308 nucleates an alternating network of intra-and intermonomer salt links that stabilize trimer assembly. Specifically, Arg-308 donates intramolecular hydrogen bonds to Glu-262 and intermolecular hydrogen bonds to Asp-204. In R308K arginase, the intramolecular hydrogen bond with Glu-262 is broken, and the intermolecular hydrogen bond with Asp-204 is present but weakened by poor geometry. Possibly, the new intramolecular hydrogen bond between Lys-308 and the carbonyl oxygen of Arg-255 hinders the formation FIG. 6. Ribbon representation of arginase trimer association at one monomer-monomer interface (monomer A, blue; monomer B, yellow) in wild-type (a) and R308K (b) arginases. In wild-type arginase, Arg-308 nucleates an alternating network of intra-and intermonomer salt links that stabilize trimer assembly (dotted red lines). In R308K arginase, the intermonomer interactions with Asp-204 are weakened with poor geometry. This probably arises from the loss of the intramonomer interaction with Glu-262 and a new interaction with backbone carbonyl of Arg-255. This figure was generated with BOBSCRIPT and Raster 3D (40,41).

TABLE II
Amino acid sequence alignment for several arginases of an optimal intramolecular hydrogen bond with Asp-204. Therefore, the weakened intermolecular hydrogen bond between Lys-308 and Asp-204 accounts for the destabilization of trimer assembly in this variant. Comparison of the active sites of trimeric wild-type and R308K arginases reveals no significant structural changes. The origin of the 3-fold difference in activity of the Arg-308 mutants compared with that of wildtype remains unclear. Small (0.1-0.2 Å) movements of a catalytic residue would be difficult to discern at 3-Å resolution. Furthermore, greater conformational flexibility for the Arg-308 variants under assay (monomer) conditions might produce the observed effects on the kinetic constants for these proteins.
The modest changes in catalytic constants for the monomeric Arg-308 mutants relative to the wild-type enzyme are in stark contrast to the low levels of activity reported for the patient R291X mutation (1,11). This premature truncation deletes not only the S-shaped oligomerization motif, but also a portion of the final ␣-helix of the polypeptide chain, and is expected to produce a protein species with decreased stability. This conclusion is supported by the finding that patients with the R291X mutation have low levels of immuno-cross-reactive material, and by the observation that the R291X mutant is expressed in E. coli as an insoluble protein (data not shown).
In summary, we have established the role of Arg-308 in stabilizing the subunit-subunit interactions in trimeric rat liver arginase. The gel filtration data, in conjunction with the kinetic data, demonstrate that the mutant arginases are not only monomeric but that they are catalytically active. A single mutation at Arg-308 weakens the network of intermonomer salt linkages, leading to the generation of catalytically active monomers as demonstrated by gel filtration and ultracentrifugation experiments.