Bovine Cytosolic 5 (cid:1) -Nucleotidase Acts through the Formation of an Aspartate 52-Phosphoenzyme Intermediate*

Cytosolic 5 (cid:1) -nucleotidase/phosphotransferase (cN-II), specific for purine monophosphates and their deoxy-derivatives, acts through the formation of a phosphoenzyme intermediate. Phosphate may either be released leading to 5 (cid:1) -mononucleotide hydrolysis or be trans-ferred to an appropriate nucleoside acceptor, giving rise to a mononucleotide interconversion. Chemical re-agents specifically modifying aspartate and glutamate residues inhibit the enzyme, and this inhibition is par-tially prevented by cN-II substrates and physiological inhibitors. Peptide mapping experiments with the phosphoenzyme previously treated with tritiated

Hydrolysis of the phosphate esterified in 3Ј or 5Ј of mononucleotides is catalyzed by a family of nucleotidases whose members differ in terms of substrate specificity, cellular location, regulation, distribution, and amino acid sequence. A better knowledge of the structure and catalytic mechanism of all these proteins is necessary for the understanding of their origin, evolution, and physiological significance. A classification of the enzymes specific for purine 5Ј-mononucleotides has been proposed by Zimmerman on the basis of cellular location and substrate specificity (1). In vertebrates, there is a membranebound 5Ј-nucleotidase and two cytosolic 5Ј-nucleotidases that preferentially hydrolyses purine mononucleotides, one specific for AMP (cN-I) 1 and the other specific for IMP and GMP (cN-II). Moreover, other nucleotidases specific for 5Ј-and 3Ј-deoxynucleotides located in cytosol (dNT-1) or in mitochondria (dNT-2) have been described (2,3). Furthermore, 5Ј-nucleotidases specific for pyrimidine mononucleotides have been described in human erythrocytes (PN-I and PN-II) (4,5). Recently, PN-I has been found to be identical to p36, an ␣-interferon-induced protein playing a role in immune diseases (6). All these proteins expressing nucleotidase activity do not show significant sequence homologies to each other (7,8).
cN-II has been demonstrated to act also as a phosphotransferase, which is a consequence of its reaction mechanism, proceeding through the formation of a phosphoenzyme covalent intermediate (9,10). No phosphotransferase activity has been reported for the other nucleotidases described so far with the exception of PN-I and PN-II (11). cN-II is a widely expressed enzyme with a remarkable sequence conservation through evolution. Its activity is involved in the regulation of the availability of intracellular IMP (12). Because IMP is the precursor of all purine nucleotides, which are not only nucleic acid building blocks but also energy transducers, intracellular and extra cellular signals, and metabolic regulators, the regulation of its hydrolysis is of paramount importance for a number of cell functions. Recently, an increased activity of cN-II has been associated to a developmental neurological disorder and to Lesch-Nyhan syndrome (13,14).
In this work we provide evidence that cN-II becomes phosphorylated during the catalytic cycle on the first aspartate (Asp-52) occurring in a DMDYT motif conserved among the different species. This consensus sequence is also shared with other members of the 5Ј-nucleotidases family. In addition, a DXDX(T/V) motif has been described as being involved in the catalytic mechanism of a number of other phosphatases/phosphotransferases (15), suggesting a common mechanism of action for all these enzymes that do not present any significant sequence homology with cN-II.  (10).
Enzyme Preparation and Assay-Bovine recombinant cytosolic 5Јnucleotidase was prepared as previously described (16), and its concentration was determined by the method of Bradford (17) using bovine serum albumin as standard. Molar concentration of the enzyme was calculated using the molecular mass of its subunit (66 kDa). Phosphotransferase activity was assayed according to Tozzi et al. (18). Briefly, the reaction mixture in 50 mM Tris-HCl, pH 7.4, contained 1.4 mM 8-[ 14 C]inosine (specific radioactivity 6000 -8000 dpm/nmol), 20 mM MgCl 2 , 2 mM IMP, 4.5 mM ATP or BPG, 1 mM dithiothreitol. If not differently specified, cN-II was assayed at a final concentration of 3-6 nM (corresponding to 0.02-0.15 mU) in 50 l of reaction mixture.
Treatment of cN-II with Woodward's Reagent K-Freshly prepared stock solutions of WRK (0.2-22 mM) in ice cold 1 mM HCl were used. The inactivation reactions were performed at 25°C in 0.2 M MES buffer, pH 6.0. The reaction mixture contained: 1 M purified cN-II and WRK ranging from 0 to 400 M. At 0, 1, 2, 3 and 4 min, 10 l aliquots were withdrawn from the reaction mixture, diluted in 5 mM Tris-HCl, pH 7.4, and then assayed for phosphotransferase activity. Protection against inactivation was assessed with the following ligands added to the enzyme 5 min before the addition of WRK: 20 mM MgCl 2 , 2 mM IMP, 5 mM ATP, and 10 mM sodium phosphate. Thereafter, the incubation was continued for 4 min in the presence of the 400 M WRK before the activity assay. Stoichiometry of WRK incorporation was determined as follows: 20 M cN-II was incubated in the presence of different concentrations of WRK (0.2-1 mM) for 4 min in a final volume of 0.11 ml. Aliquots (10 l) were used for activity determination and 100 l were subjected to gel filtration on Bio-Spin-30 Columns (Bio-Rad) equilibrated with 10 mM Tris-HCl, pH 7.4. All enzyme samples treated with WRK had an absorption band with a max at 340 nm, which was not present in the untreated samples. The number of modified carboxyl groups was estimated by the increase in absorbance at 340 nm (⑀ ϭ 7000 M Ϫ1 cm Ϫ1 ) (19).
Chemical Entrapment of the Acyl Phosphate Intermediate-Bovine 5Ј-nucleotidase (190 g, 2.8 nmoles) was incubated in 150 l of reaction buffer containing no (control) or 5.3 mM IMP, 27 mM MgCl 2 , 6 mM ATP, 2.6 mM dithiothreitol, and 50 mM Tris-HCl, pH 7.4, at 0°C. The reaction was stopped after 30 s by the addition of 10% ice-cold trichloroacetic acid (750 l). After centrifugation (10 min at 12000 ϫ g and 4°C), the pellet was resuspended in 750 l of 5% ice-cold trichloroacetic acid and recentrifuged for 10 min at 4°C. The pellets were washed with 10 mM ice-cold HCl, centrifuged for 5 min at 4°C, and lyophilized. Protein samples obtained from the labeling experiment and control were independently resuspended in 65 l of Me 2 SO containing 25 mCi of [ 3 H]NaBH 4 (11.4 Ci/mmol) and incubated at room temperature. After 10 min, 10 mol of unlabeled NaBH 4 (solved in 35 l of Me 2 SO) were successively added and incubated at room temperature for other 10 min. One milliliter of 0.44 M ice-cold HClO 4 was added to both reaction mixtures, and samples were placed on ice for 30 min and then centrifuged for 20 min at 12000 ϫ g and 4°C. The resulting pellets were washed with 500 l of cold acetone and lyophilized. Preparation and Isolation of Tryptic Peptides-Protein samples were desalted and concentrated by a 12% mini SDS-polyacrylamide gel electrophoresis (20); gels were stained with Coomassie Blue R250 and destained for 3 h with 10% acetic acid. Protein bands were excised, and gel pieces were tritured, washed with water, reduced with dithiothreitol and carboxamidomethylated (21). Gel pieces were equilibrated in 0.4% NH 4 HCO 3 and finally digested in situ with trypsin in 0.4% NH 4 HCO 3 , pH 8.5, at 37°C for 18 h, (enzyme/substrate ratio 1:30 w/w). Peptides were extracted by sonication with 500 l of 0.4% NH 4 HCO 3 /acetonitrile 1:1 v/v, pH 8.5 (twice). Extracts were combined and lyophilized.
Peptide mixtures were fractionated by reverse-phase HPLC on a Vydac C 18 column 218TP52 (250 ϫ 2.1 mm), 5 m, 300 Å pore size (The Separation Group) by using a linear gradient from 5-60% of acetonitrile in 0.1% trifluoroacetic acid over 60 min at flow rate of 0.2 ml/min. Individual components were manually collected and dried in a Speed-

cN-II Forms an Asp-52-Phosphointermediate
vac centrifuge (Savant). Samples of the fractions were counted for radioactivity and analyzed by mass spectrometry.
Mass Spectrometry and Sequencing Analysis-Matrix assisted laser desorption ionization mass spectra were recorded using a Voyager DE MALDI-TOF mass spectrometer (Applied Biosystems); a mixture of analyte solution, ␣-cyano-4-hydroxy-cinnamic acid and bovine insulin was applied to the sample plate and dried in vacuo. Mass calibration was performed using the molecular ions from bovine insulin at 5734.54 Da and the matrix at 379.06 Da as internal standards. Raw data were analyzed by using computer software provided by the manufacturer and are reported as average masses. Assignments of the recorded mass values to individual peptides were performed on the basis of their molecular mass as previously described (21).
Sequence analysis was performed using a Procise 491 protein sequencer (Applied Biosystems) equipped with a 140C microgradient apparatus and a 785A UV detector (Applied Biosystems) for the automated identification of PTH-derivatives.
Deletion Mutant and Mutagenesis Studies-A mutant with a deletion of the first 42 amino acids at the N terminus (42⌬N) was obtained as follow: NcoI was used to eliminate a 195-base pair fragment at the 5Ј side of the cDNA for CN-II. One new construct was thus obtained and expressed the missing 65 amino acids (23 belonging to pET28c vector and 42 to the N terminus of cN-II). To prepare point mutants of Asp-52 and Asp-54, a slight modification of the protocol described by Ekici et al. PCR 2: no template, 100 pmol of P2R with the same cycling conditions. Primers were as follows: FnheI, 5Ј-CCGCTAGCATGACAACCTC-CTG-3Ј (from base Ϫ8 to base 14 of pET28c-5ЈN construct); P2R, 5Ј-CCCAAAAATGTGATCTCCAATA-3Ј (from base 1065 to 1044); D52-R1, 5Ј-TGTATAATCCATYNCAAACCCA-3Ј (from base 168 to 147); D54-R1, 5Ј-TGTATAYNCCATGTCAAACCCA-3Ј (from base 168 to 147).
Preparation of 32 P-Enzyme Covalent Intermediate-About 1.5 g of cN-II (wild type or mutants) were incubated 15 s on ice in 15 l of a solution of 20 mM MgCl 2 , 7.2 mM ATP, 1.5 mM [ 32 P]IMP, and 2 mM dithiothreitol. The reaction was stopped by adding 5 l of 8% SDS in 0.25 M Tris-HCl, pH 6.8; the samples were vortexed and loaded on a nitrocellulose membrane (0.2 m -Sartorius). The membrane was washed for 8 min with 100 ml of 2% SDS in 60 mM Tris-HCl, pH 7.4 (5 times), and radioactivity was measured with 5 ml of scintillation mixture.

5Ј-Nucleotidase Inactivation by Woodward's Reagent K-To
get information on the nature of the amino acid residues occurring in the active site of 5Ј-nucleotidase, different inhibitors were tested as effectors of the enzyme activity. Among these, carboxylate-directed compounds resulted to inhibit the enzyme. In fact, incubation of recombinant protein with Woodward's reagent K, at pH 6.0 resulted in a time-and concentrationdependent loss of activity (Fig. 1A). A plot of the observed rate constants (K obs ) versus reagent concentration over the range 5-400 M yielded a second order rate inactivation constant of 1.2 mM Ϫ1 m Ϫ1 (Fig. 1B). The effect on the chemical modification by substrates or inhibitors of the enzyme, was studied. Fig. 2 shows that among the different compounds tested, the allosteric inhibitor phosphate either in the absence or in the presence of Mg 2ϩ exerted the highest protection against enzyme inactivation. Similarly, IMP (the best cN-II substrate) also provided a good protection, which was increased by the presence of Mg 2ϩ . In contrast, the allosteric activator ATP in combination with Mg 2ϩ was unable to protect the enzyme in the absence of the substrate. However, addition of IMP and phosphate to this mixture resulted again in a substantial protection from chemical inactivation. Fig. 3 shows the stoichiometry of WRK incorporation into cN-II as a function of residual activity. The plot is monophasic, and the experimental data fit a straight line that intersects the abscissa at a point corresponding to the incorporation of 4.1 mol of WRK/mol of enzyme.
Identification of Phosphorylated Residue in cN-II-The experiments reported above indicated that carboxyl groups are involved in the catalytic mechanism of the enzyme. Because the hydrolysis of the monophosphate catalyzed by cN-II proceeds through the formation of a phosphoenzyme intermediate (10), we attempted to trap this derivative to identify the residue involved in catalysis. However, phosphoaspartyl and phosphoglutamyl residues are too labile to resist to the procedures commonly used to identify phosphorylated residues (enzymatic digestion of the protein, separation of the resulting peptides, and sequencing of the phosphorylated species). Therefore, we used the method of Degani and Boyer (23) in which the unstable carboxyl phosphate adduct is reduced with [ 3 H]NaBH 4 to a stable radiolabeled hydroxymethyl-containing derivative before further processing of the protein. Peptide mapping, radioactivity measurement, and mass spectrometric analysis of the fractions obtained allowed a direct identification of the phosphorylated residue.
The recombinant enzyme incubated with or without (control) inosine-monophosphate was treated with radioactive borohydride, denatured, and concentrated by a mini-SDS-polyacrylamide gel electrophoresis (20). Gel bands were reduced, alkylated, and digested in situ with trypsin as previously reported (21). The resulting digests were separated by reverse phase-HPLC as illustrated in Fig. 4 for the labeled protein (panel B) and control (panel C). A radioactive component was simulta- respectively. The latter was associated to the carboxamidomethylated peptide 48 -61; the former did not correspond to any predicted species with regard to its molecular mass. On the basis of the expected partial conversion of its parent carboxylate to hydroxymethyl group, the presence of a satellite species differing for Ϫ14 Da was indicative of the polypeptide where phosphorylation occurred during catalysis. Therefore, peptide 48 -61 apparently contained the active site residue of 5Ј-nucleotidase.
To ascertain the identity of this residue (discriminating be- FIG. 4. Reverse phase HPLC analysis of the tryptic digest obtained from the phosphorylated enzyme (B) and control (C) following reduction with [ 3 H]NaBH 4 . The hydrolysates were chromatographed on a narrow-bore column as described under "Experimental Procedures." Fractions of identical volumes were collected and measured for radioactivity. The absorbance profile of the control (not shown) was similar to that obtained for the phosphorylated species (A).

FIG. 5. MALDIMS analysis of the radioactive fraction specifically isolated from phosphorylated cN-II following reduction with [ 3 H]NaBH 4 and trypsin digestion.
Assignment of the recorded mass values to individual peptides was performed on the basis of their molecular mass and enzyme specificity. The nature of polypeptides was confirmed by automated Edman degradation. The satellite signal at 1716.9 m/z was associated to the modified peptide (48 -61) carboxamidomethyl resulting from the reduction of aspartic acid to homoserine.

cN-II Forms an Asp-52-Phosphointermediate
tween Asp-52 and Asp-54 present in the peptide sequence), fraction PL#29 was submitted to sequence analysis. A modified sequencer program allowed to collect a portion of PTH-derivative released at each cycle of Edman degradation successively measured for radioactivity. The results obtained are shown in Table I. Two different sequences were simultaneously detected, which confirmed that the peptides present in this radioactive fraction were those predicted by the MALDIMS experiments. The radioactivity recovered at each cycle of degradation indicated that the majority was observed in the position corresponding to Asp-52 (cycle 5). A significant amount of PTHderivative-Hse was found at this cycle, demonstrating the conversion of the carboxyl phosphate group of Asp-52 to the [ 3 H]hydroxymethyl group of Hse. A negligible amount of radioactivity was observed in the fraction from the cycle corresponding to Asp-54 (cycle 7). The absence of a satellite signal at Ϫ14 Da from peptide 292-303 and the low amount of radioactivity recovered at the cycle, where the only acid residue (Glu-298) present in the sequence occurred, ruled out the possibility that this peptide was the labeled one. Therefore, these experiments definitively demonstrated that the residue phosphorylated during the reactions catalyzed by 5Ј-nucleotidase is Asp-52.
Site-directed Mutagenesis of Cytosolic 5Ј-Nucleotidase-To substantiate the active-site labeling experiments, we constructed cN-II mutants in which we replaced separately Asp-52 and Asp-54 by either glutamate or alanine residues. In addition, we prepared a mutant where 42 amino acid residues at the N terminus were deleted, taking advantage of a restriction site specific for NcoI present in the nucleotide sequence. Extracts prepared from transformed cultures were tested for nucleotidase activity, as reported in Table II. Only the wild type enzyme was active; all mutants displayed an activity comparable with that observed in a control culture without a translatable insert. All of these extracts were chromatographed on a Talon ion metal-affinity chromatography column, yielding comparable amount of purified proteins as determined by SDSpolyacrylamide gel electrophoresis analysis (result not shown). Among the purified proteins, again only the wild type enzyme was active, and its specific activity indicated that by Talon chromatography a 10-fold purification was obtained (Table II). When the purified proteins were incubated in the presence of [ 32 P]IMP, only the wild type enzyme became radioactive, indicating that a certain amount of phosphointermediate was formed during catalysis; on the contrary, the mutant proteins were completely devoid of radioactivity (Table II).
Sequence Comparison-The result mentioned above demonstrated that Asp-52 is the amino acid residue of cN-II that is phosphorylated during catalysis; in addition, Asp-54 seems to assist in the formation of this covalent intermediate. Both

H-labeled 5Ј-nucleotidase following digestion with trypsin
The amount (picomoles) of PTH derivatives recovered at each cycle of degradation is given in parenthesis. Hse was quantified on the basis of the Thr response factor. The radioactivity measured for each cycle of the Edman degradation is also reported. CAM, carboxamidomethyl. Gly (206) Glu (178) Gly (115) Thr (81) Val (63) Leu (37) Arg (11) Radioactivity (cpm)

cN-II Forms an Asp-52-Phosphointermediate
residues are present in a DMDYT motif highly conserved among cN-IIs from different species. Although a BLAST search in different data banks revealed that cN-II do not present a significant sequence identity or homology with other proteins, a careful analysis revealed that other polypeptide species are characterized from this specific signature. Fig. 6 shows the alignment of the cN-II DMDYT motif with that occurring in other members of the 5Ј-nucleotidase family, such as PN-I (specific for pyrimidine mononucleotides), both cytosolic and mitochondrial 5Ј-3Ј-nucleotidases dNT-1 and dNT-2 (specific for deoxynucleotides), and cN-I (AMP-specific cytosolic 5Ј-nucleotidase). Membrane-bound 5Ј-nucleotidase do not contain this conserved region. Several other phosphatases and phosphotransferases have been reported to form a covalent phosphointermediate involving an aspartate residue (24,25). Recently, a new class of phosphatases or phosphotransferases have been described as containing a conserved DXDX(T/V) motif (15,26) (Fig. 6). Also for several members of this large family it was demonstrated that the first aspartate occurring in this region becomes phosphorylated during catalysis (15). Although these enzymes show minimal sequence homology to cN-II (less than 5%), they all contain the consensus sequence described in this manuscript. DISCUSSION The results reported in this work demonstrate that cN-II is inhibited by WRK in a dose-dependent manner, suggesting that this reagent modifies carboxylate residues important for proper enzymatic functioning. WRK is well established in terms of its ability to modify aspartates or glutamates (27,28). The selection of appropriate reaction conditions (pH 6.0 and very short inactivation times) ensured us an increased selectivity of WRK for acid residues and ruled out a possible reaction of this compound with other nucleophile amino acids (29 -31). The partial protection exerted by substrates and inhibitors of the enzyme is strongly indicative that some of the modified residues are located in the active site. It was previously observed that the allosteric inhibitor phosphate causes an increase in the K m for IMP, suggesting that the cN-II active site can result less accessibly in the presence of this anion. On the contrary, has been reported that the allosteric activator ATP, in the presence of Mg 2ϩ causes a moderate decrease in the K m for the substrate and an increase in the catalytic rate (32). We find here that phosphate is effective in protection against WRK inactivation, while ATP-Mg 2ϩ , which increase the accessibility of the active site, do not exert any protective effect. On the contrary, ATP in the absence of Mg 2ϩ indeed afforded a protection similar to that observed for IMP. ATP is also stabilizing the enzyme during storage and against thermal inactivation (12). These observations indicate that, in the absence of Mg 2ϩ , ATP can determine a protective effect possibly by an interaction with the active site.
In a previous paper we demonstrated that cN-II forms a phosphoenzyme intermediate during catalysis, which was extremely labile (10). On the basis of this observation and the enzyme sensitivity toward treatment with WRK, we deduced that an aspartate or glutamate residue was the phosphate cN-II Forms an Asp-52-Phosphointermediate acceptor. The identification of this amino acid has been obtained after reduction of the acyl phosphate intermediate with radiolabeled borohydride, and separation of its tryptic digest, radioactivity, and molecular mass measurements. Sequence analysis performed on the isolated peptide demonstrated that Asp-52 was the phosphorylated residue. Site-directed mutagenesis experiments confirmed the fundamental role of this residue for the enzyme activity; in fact, even a conservative substitution in this position totally abolished enzyme activity and prevented the formation of the phosphorylated enzyme. In addition, mutants on Asp-54 were inactive and not able to generate this intermediate. On this basis, our results demonstrate that cN-II belongs to the large class of phosphohydrolases described by Collet et al. (15) because it shows close to the N terminus a DXDX(T/V) motif in which the first Asp is phosphorylated during catalysis. Although not so structurally similar to cN-II to be related by a simple sequence analysis investigation, most of these enzymes present a phosphotransferase-phosphatase activity, a common catalytic dependence from Mg 2ϩ and the occurrence of a phosphointermediate. As shown in Fig. 6, all the proteins listed have the common motif flanked by almost conserved hydrophobic amino acids. A few months ago, the three-dimensional structure of the first member of this family, phosphoserine phosphatase from Methanococcus jannaschii, had been solved by x-ray crystallography (33). This enzyme presents a fold consisting of separate ␣/␤ and four-helix-bundle domains. A careful structural analysis revealed the simultaneous occurrence of four acid residues into its active site. Asp-11 is important in playing the role of phosphate acceptor; similarly, Asp-13, Asp-167, and Glu-20 seem to be essential for Mg 2ϩ coordination. Site-directed mutagenesis experiments revealed the importance of these residues in catalysis. The data reported in this paper on the stoichiometry of the cN-II inactivation by WRK demonstrate that, also in this case, four acid residues are modified with the same efficiency. Inactivation is protected by enzyme substrates and inhibitors, suggesting that modified residues are located into the active site. In addition, both Asp residues occurring in the DXDX(T/V) motif of cN-II (Asp-52 and Asp-54), similar to that of phosphoserine phosphatase (Asp-11 and Asp-13), are essential for catalysis. Therefore, these data are in strict analogy with that determined for phosphoserine phosphatase and suggest that also in cN-II the occurrence of different acid residues in the active site is important for Mg 2ϩ coordination and catalytic efficiency.
Furthermore, sequence comparison of cN-II with other nucleotidases demonstrates that also pyrimidine nucleotidase purified from human erythrocytes (PN-I) and 5Ј-3Ј-deoxynucleotidase located in the cytoplasm (dNT-1) and in its mitochondrial counterpart (dNT-2), contain a DXDX(T/V) motif. Also human cN-I shows a similar sequence even though located close to the C terminus. Pyrimidine nucleotidase PN-I has been shown to have a phosphotransferase activity (11), which implicates a reaction mechanism proceeding through a phosphorylated enzyme intermediate. In the case of dNT-1 and dNT-2, the reaction mechanism has not been described so far. The two enzymes present a high degree of identity in their structure and molecular and functional characteristics, suggesting that they act through the same mechanism. dNT-1 has been recently cloned and expressed in Escherichia coli and mammalian cells (8) and was found remarkably similar to PN-II described in human erythrocytes concluding that PN-II, which is a phosphatase/phosphotransferase (11), actually belongs to the class of the dNT's (8). Furthermore, substrate specificity of dNT-1 is remarkably similar to that displayed by the phosphotransferase with hydrolase activity acting on deoxynucleotides described by Tesoriere et al. (34). Human cN-I has been recently cloned (35), but its reaction mechanism has not been described so far. However its functional similarities with cN-II (such as the complete Mg 2ϩ dependence) suggest that it might have a similar reaction mechanism. Therefore, on the basis of these observations and the sequence alignment reported in Fig.  6 we can presumably assume that catalysis of all soluble nucleotidases proceed through a similar reaction mechanism involving the formation of a phosphorylated intermediate. Therefore, we conclude that cN-II is a first example of a group of eukaryotic cytosolic nucleotidases presenting a common catalytic machinery and conserved active site residues that resemble those occurring in other enzymes belonging to the superfamily of bacterial, eukaryotic and archeal phosphohydrolases (15,26,33).