Cloning and Characterization of a New Member of the Nudix Hydrolases from Human and Mouse*

Proteins containing the Nudix box “G X 5 E X 7 REU X E- E X GU” (where U is usually Leu, Val, or Ile) are Nudix hydrolases, which catalyze the hydrolysis of a variety of nucleoside diphosphate derivatives. Here we report cloning and characterization of a human cDNA encoding a novel nudix hydrolase NUDT5 for the hydrolysis of ADP-sugars. The deduced amino acid sequence of NUDT5 contains 219 amino acids, including a conserved Nudix box sequence. The recombinant NUDT5 was expressed in Escherichia coli and purified to near homogeneity. At the optimal pH of 7, the purified recombinant NUDT5 catalyzed hydrolysis of two major substrates ADP-ribose and ADP-mannose with K m values of 32 and 83 m M , respectively; the V max for ADP-mannose was about 1.5 times that with ADP-ribose. The murine NUDT5 homolog was also cloned and characterized. mNudT5 has 81% amino acid identity to NUDT5 with catalytic activities similar to NUDT5 under the optimal pH of 9. Both NUDT5 the supplied in a total of 25 m l. After an initial denaturation at 94 °C for 3 30 cycles of denaturation at 94 °C for 30 s, annealing at 62 °C for 1 min, and extension at 72 °C for 1.5 min were performed, followed by a final extension at 72 °C for 7 min. The were by and scored for the presence or absence of the 230-bp product. In preliminary experiments, a single DNA band of expected size 230 bp was detected using mouse genomic DNA as template while no detected using as Two independent mapping band of expected Phylogenetic Analysis— Distance analysis performed using neighbor joining in the A multiple sequence alignment of “MutT” domain Pfam

acterized (for review, see Refs. 1 and 5). Besides dGTP and 8-oxo-dGTP as the major substrates for MutT protein, there is a wide spectrum of substrates with mostly nucleoside diphosphate derivatives for the subsequent characterized Nudix hydrolases, including dATP (6), Ap n A 1 (n ϭ 3, 4, 5, and 6) (7-12), NADH (7,13), GDP-mannose (14), ADP-ribose (7,15,16), and diphosphoinositol polyphosphates (17,18). Some Nudix hydrolases are highly specific to one substrate, but others can hydrolyze several different substrates. An example of the diversity of substrates is given by two enzymes, Aps1 from Schizosaccharomyces pombe and YOR163w from Saccharomyces cerevisiae, which catalyze the hydrolysis of two unrelated classes of substrates: diadenosine polyphosphates Ap 5 A and Ap 6 A, as well as diphosphoinositol polyphosphates (18). The three-dimensional solution structure of E. coli MutT (19,20) has revealed that the Nudix box has a loop-helix-loop motif, which is important for the binding of a divalent cation and may also contribute to the substrate binding.
ADP-ribosylation is a regulatory modification of protein in which an ADP-ribose moiety in ␤-NAD ϩ is transferred to the specific amino acid residue of an acceptor protein for cellular non-redox activities. Mono-ADP-ribosylation of protein by bacterial toxins leads to immediate cytotoxic effect (21). The mono-ADP-ribosylation reactions have also been detected in many eukaryotic organisms catalyzed by endogenous ADP-ribosyltransferases, which may have regulatory roles in cellular processes (for review, see Ref. 22). Poly-ADP-ribosylation of a number of nuclear proteins is carried out by poly(ADP-ribose) polymerases-1 and 2 in response to DNA strand breaks and resealing (23). ␤-NAD ϩ can also be converted to cyclic ADPribose, which is a second messenger in pancreatic islets for Ca 2ϩ mobilization in the endoplasmic reticulum to secrete insulin (24).
Free ADP-ribose is produced during the reverse processes of degrading protein-bound mono-or poly(ADP-ribose), or cyclic ADP-ribose. It is also the turnover product of ␤-NAD ϩ (25). Free ADP-ribose is a highly reactive molecule, which causes non-enzymatic mono-ADP-ribosylation of proteins. It has been speculated that free ADP-ribose could act on the targeting sites of poly(ADP-ribose) polymerases or bacterial toxins, which in either case will cause change of the activity of the modified proteins leading to intracellular damage (26). Despite lacking the knowledge about possible specific roles of free ADP-ribose in cellular processes, the level of ADP-ribose in the cell is * This work was supported by Tumor Immunology Training Grant 5-T32-CA009120 (to H. Y.) and National Institutes of Health Grant GM 32184 (to J. H. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  expected to be carefully maintained in order to minimize the potential detrimental effect of free ADP-ribose.
Free ADP-ribose is converted to AMP and ribose 5-phosphate by ADP-ribose pyrophosphatase. The activities of ADP-ribose pyrophosphatase are detected in microorganisms as well as in higher eukaryotes. So far, three genes encoding ADP-ribose pyrophosphatase activities have been identified and characterized. ORF186 from E. coli was characterized as a Nudix hydrolase with broad substrate specificity (7). It catalyzes the conversion of ADP-ribose to AMP and ribose 5-phosphate. It can also catalyze the hydrolysis of Ap 3 A and NADH with similar efficiencies. Recently, the product of the MJ1149 gene, another Nudix hydrolase from thermophilic Archaeon Methanococcus jannaschii, was characterized as a highly specific ADP-ribose pyrophosphatase (15). A bifunctional enzyme slr0787 from Synechocystis sp. was shown to have both ADP-ribose pyrophosphatase and NMN adenylyltransferase activity (16). Four other Nudix hydrolases, YSA1 of S. cerevisiae (ScYSA1), ORF209 of E. coli (EcORF209), YQKG from Bacillus subtilis (BsYQKG), and Hi0398 from Hemeophilus influenzae, were briefly described to have ADP-ribose pyrophosphatase activity (1, 7). 2 The functions of ADP-ribose pyrophosphatases were proposed to be the "housecleaning enzyme" that removes the highly reactive free ADP-ribose molecule as well as reutilize it after hydrolysis to form AMP and ribose 5-phosphate (1).
Although several biochemically distinct ADP-ribose pyrophosphatase activities have been identified from human erythrocytes, rat liver, and Artemia cysts (27)(28)(29)(30), no gene has been cloned and characterized. 3 In this work, we report cloning and characterization of a novel Nudix hydrolase NUDT5 from human and its murine homolog mNudT5. The major substrates for NUDT5 and mNudT5 are ADP-sugars with preference to ADP-ribose. The RNA transcripts of NUDT5 and mNudT5 are expressed in all tissues analyzed with preferential expression in liver. Furthermore, the genomic structure and chromosome localization for both NUDT5 and mNudT5 are also described.

EXPERIMENTAL PROCEDURES
Identification and Cloning of NUDT5 and mNudT5 cDNA-A cDNA fragment containing the NUDT5 coding region was identified in the Human Genome Sciences human cDNA sequence data base during searchs for fragments encoding protein sequence homologous to E. coli MutT using the program BLAST (31). A phage library containing mouse liver cDNA (Stratagene, La Jolla, CA) was screened using a probe containing the NUDT5 cDNA coding region obtained after PCR amplification. The hybridization and washing condition were according to the manufacturer protocols. Recombinant phages containing mNudT5 were isolated and DNA samples were prepared using protocols as described previously (32). The sequencing reactions were carried out using [␣-32 P]dATP and a SequiTherm Cycle Sequencing Kit (Epicentre Technologies, Madison, WI). The oligonucleotides used for sequencing were synthesized on a Beckman oligo1000 DNA synthesizer (Beckman Instruments). All oligonucleotides were deprotected in ammonium hydroxide and used without further purification.
Expression in E. coli and Purification of NUDT5 and mNudT5-The NUDT5 gene was cloned between the SalI and XbaI sites of the bacterial expression vector pQE9 (Qiagen, Chatsworth, CA) after PCR amplification. Transformants of E. coli M15/pREP4 with the pQE9/NUDT5 were grown at 37°C in 100 ml of LB medium with 100 g/ml ampicillin and 25 g/ml kanamycin. The plasmid pREP4 constitutively expresses the Lac repressor protein encoded by the lacI gene in order to reduce the basal level of expression (Qiagen). When the culture grew to A 600 of 0.7, a final concentration of 1 mM isopropyl-1-thio-␤-D-galactopyranoside was added to induce the expression of NUDT5 protein for 3 h. Bacterial lysate was prepared by French press (16,000 psi) in buffer A (50 mM sodium phosphate, pH 8.0, 300 mM NaCl) plus 0.5 mM phenylmethylsulfonyl fluoride. After clarification by centrifugation the lysate was mixed with 3 ml of Ni 2ϩ -NTA matrix (Qiagen) for 1 h at 4°C with gentle shaking. Then the mixture was poured into a column, washed with buffer A containing 0.1 M imidazole, and eluted with buffer A containing 1 M imidazole. Western blot analysis was performed using a primary antibody (RGS⅐His Antibody, Qiagen) and a secondary alkaline phosphatase-conjugated antibody (Sigma) to monitor the recombinant NUDT5 protein during purification. The fractions containing the recombinant NUDT5 proteins were dialyzed overnight in 1 liter of buffer B (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 50% glycerol) with two changes. A clear protein sample was obtained after centrifugation of the dialyzed sample, aliquots were stored at Ϫ80°C.
The mNudT5 gene was cloned between SphI and HindIII sites of the bacterial expression vector pQE30 (Qiagen) after PCR amplification. The expression condition and purification protocols were similar to those of NUDT5.
Electrospray Mass Spectrometry-A Perkin-Elmer Sciex (Thornhill, Canada) API III triple quadrupole mass spectrometer fitted with an Ion Spray TM source was tuned and calibrated as described previously (34). Positive ion protein spectra were produced by injection of the proteins dissolved in water onto a C 8 reverse phase column (Keystone Scientific BDS Hypersil C 8 , 100 ϫ 1 mm, 3-m particle size, 120 Å pore size, 20 l/injection) equilibrated in water/acetonitrile/trifluoroacetic acid (95/ 5/0.1 all by volume). The column was eluted (40 l/min) with a linearly increasing concentration of acetonitrile (2%/min) and the column eluant was passed in series through a UV detector (215 nm) and the Ionspray source. Data was recorded with the mass spectrometer scanning from m/z 400 -2000 (step size 0.3 Da, dwell time 1 ms, 5.61 s/scan, orifice voltage 90). The average of the spectra contributing to the peak in ion current was computed. Calculation of molecular mass from the series of multiply charged ions found in the protein was achieved with the MacSpec TM computer program (version 3.3, PE Sciex, Ontario, Canada). Calculation of theoretical protein average (chemical) molecular mass was achieved with the MacBiospec TM computer program (version 1.0.1, PE Sciex) based on the MacProMAss computer program of Lee and Vermuri (35).
Enzyme Assay-A slightly modified colorimetric procedure (15) was used to assay the hydrolysis of the substrates. A standard reaction mixture contained (in 50 l): 50 mM Tris, pH 7.0 (for NUDT5) or pH 9.0 (for mNudT5), 5 mM MgCl 2 , 1 mM dithiothreitol, 2 mM substrate, 1 unit of calf intestinal alkaline phosphatase, and 140 ng of NUDT5 or 400 ng of mNudT5. The reaction mixture was incubated at 37°C for 15 min, terminated by the addition 250 l of 20 mM EDTA. The inorganic orthophosphate produced was quantified by the colorimetric assay of Ames and Dubin (36).
For product identification by electrospray mass spectrometry, the calf intestinal alkaline phosphatase was omitted from the standard assay. Spectra of reaction mixtures were obtained by flow injection analysis of samples diluted (2/100) in water/acetonitrile/triethylamine (50/50/0.1, all by volume). Aliquots of the solution (10 -20 l) were injected into a stream of the same solvent entering the Ion Spray source TM (10 l/min) while the mass spectrometer was scanning in the negative ion mode from 200 to 1000 Da (0.3 Da step size, 5.47 s/scan, orifice voltage 60 -80).
Northern Blot Analysis-Mouse multiple tissue Northern (MTN TM ) blot containing 2 g of poly(A) ϩ RNA from various adult tissues (CLON-TECH) was hybridized with a 32 P-random primed 300-bp mNudT5 cDNA fragment containing exon 1-4 (Prime-It II kit, Stratagene). The hybridization and washing conditions were according to the manufacturer protocol "Multiple Tissue Expression Array User Manual." The blot was hybridized at 65°C overnight and washed in solution 1 (2 ϫ SSC and 1% SDS) 5 times at 65°C for 20 min and twice in solution 2 (0.1 ϫ SSC and 0.5% SDS) at 55°C for 20 min. The blot was stripped by incubation for 10 min in 0.5% SDS at 90 -100°C and reprobed with ␤-actin cDNA control probe.
The Human Multiple Tissue Expression (MTE TM ) Array (CLON-TECH, Palo Alto, CA) hybridization was carried out according to the manufacturer protocol "Multiple Tissue Expression Array User Manual" with a 32 P-random-primed 700-bp NUDT5 cDNA PCR probe containing the coding region (Prime-It II kit, Stratagene). The blot was washed under the same condition as for the MTN blot (see above).
Determination of the Exon-Intron Organization-A phage library containing mouse genomic DNA (strain 129 Svj, Stratagene) was screened using a NUDT5 cDNA probe containing the coding region. The recombinant phage was isolated and the DNA sample was prepared using protocols as described previously (32). The DNA was partially sequenced with primers designed on the basis of mNudT5 cDNA sequence to obtain the intron-exon organization. The 3Ј end of mNudT5 gene containing exons 7-9 was obtained from two overlapping PCR products from mouse genomic DNA (strain 129 Svj). The PCR products were cloned in pCR2.1-TOPO vector using TOPO TA cloning kit (Invitrogen, Carlsbad, CA) and partially sequenced. The sizes of introns were obtained from either sequencing of the genomic clone (introns 1, 3-5, and 7) or from PCR analysis (introns 2, 6, and 8).
To determine the exon-intron organization of the NUDT5 gene, 8 sets of PCR primers were designed to construct a contig of genomic DNA fragments covering the NUDT5 locus. Human genomic DNA (CLON-TECH) was used for PCR. PCR products were analyzed on agarose gel, cloned into the pCR2.1-TOPO vector using TOPO TA cloning kit (Invitrogen), and partially sequenced to obtain the exon-intron organization. The sequences of primers are available upon request.
Chromosomal Mapping of NUDT5 and mNudT5-Radiation hybrid mapping technique was used to map both mNudT5 and NUDT5.
The 100 cell lines of the T31 radiation hybrid panel, which carries fragments of the mouse genome on a hamster background (37), were used as templates for PCR amplification with mNudT5 primers 5Ј-CAAAGAACCCTGCACCATGA-3Ј, and 5Ј-TACCTACCTGGCTTTCA-CAC-3Ј derived from the sequence expanding the region of exon 4-intron 4 of the mNudT5 genomic DNA. Reactions were performed with 112.5 ng of hybrid clone DNA, 0.4 M each primer, 250 nM each dNTP, 0.625 units of TaKaRa Ex Taq polymerase (TaKaRa Shuzo Co., Japan) and the supplied buffer, in a total volume of 25 l. After an initial denaturation at 94°C for 3 min, 30 cycles of denaturation at 94°C for 30 s, annealing at 62°C for 1 min, and extension at 72°C for 1.5 min were performed, followed by a final extension at 72°C for 7 min. The PCR products were analyzed by gel electrophoresis in 1.2% agarose gels, stained with ethidium bromide, and scored for the presence or absence of the 230-bp product. In preliminary experiments, a single DNA band of expected size 230 bp was detected using mouse genomic DNA as template while no product was detected using hamster genomic DNA as template. Two independent reactions were performed with mNudT5 primers for the entire T31 panel. All data were submitted to the Jackson Laboratory Mouse Radiation Hybrid Data base for mapping analysis.
The 93 cell lines of the Gene Bridge 4 (GB4) radiation hybrid panel, which carries fragments of the human genome on a hamster background (38), were used as templates for PCR amplification with NUDT5 primers 5Ј-TCGTAAAATAAAAGCACAGAAC-3Ј and 5Ј-CTGGACTA-GAAAAGTAACTGAGCTGT-3Јderived from the 3Ј-untranslated region of the NUDT5 cDNA. PCR conditions were the same as described above for the mNudT5 gene. In preliminary experiments, a single DNA band of expected size 190 bp was detected using human genomic DNA as template while no product was detected using hamster genomic DNA as template. Two independent reactions were performed with NUDT5 primers for the entire GB4 panel. All data were submitted to the Whitehead Institute/MIT Center for Genome Research's Human Radiation Hybrid Data base for mapping analysis.
Phylogenetic Analysis-Distance analysis was performed using neighbor joining in the PAUP program (39). A multiple sequence alignment of the "MutT" domain was retrieved from the Pfam website (40). Representative sequences were selected from the Pfam alignment and five additional sequences were manually added (NUDT5, mNudT5, Tm, Pa, and Ap, see Fig. 10, legend).

Identification and Cloning of NUDT5 and mNudT5 Gene-
During the search for DNA repair genes in the Human Genome Sciences human cDNA sequence data base, a clone was identified, which encodes a protein of 28.5 and 21.8% amino acid identity to E. coli MutT and hMTH (41), respectively. The nucleotide sequence of the full-length cDNA (1129 base pairs) is shown in Fig. 1. It contains an open reading frame (ORF) of 660 nucleotides from the first translation initiation codon ATG to the termination codon TAA (nucleotides 120 -779). An inframe stop codon is located 39 base pairs upstream of the initiation codon. The complete ORF encodes a protein of 219 amino acids with a predicted molecular mass of 24357 Da and pI 4.8 (42). We confirmed the nucleotide sequence of 5Ј-nontranslated region with sequences obtained from the BLAST search (31). Of the six sequences from the GenBank that contained the 5Ј-nontranslated region (accession numbers: AA306176, AA306565, AA490510, W07480, W16891, and W92824), all of them contained the in-frame stop codon before the translation initiation codon ATG. The Nudix signature sequences of this ORF protein, together with E. coli MutT, hMTH, and other characterized Nudix hydrolases with unique substrate specificity are shown in Table I. The ORF protein contains the conserved amino acid residues in the Nudix signature sequence, which recently has been designated as NUDT5 on the website of the Human Gene Nomenclature Committee. One feature in NUDT5 Nudix sequence is Tyr-119, which is not a bulk amino acid like Ile, Leu, or Val as in the majority of other Nudix signature sequences.
A cDNA clone of a murine NUDT5 homolog (mNudT5) from mouse liver cDNA library was identified using NUDT5 cDNA sequence as a probe. It contains a 934-bp cDNA sequence with an open reading frame of 657-bp encoding a predicted protein product of 218 amino acids. The mNudT5 has 81% amino acid identity to the NUDT5 (Fig. 2) with a predicted molecular mass of 23986 Da and pI of 5.3 (42).
Expression of Both NUDT5 and mNudT5 in E. coli and Purification of the Recombinant Proteins-The initial attempt to characterize NUDT5 was to test whether the predicted protein product had the MutT activity. Complementation of E. coli mutT mutator phenotype was carried out using E. coli mutT strain expressing the cloned NUDT5 gene in the plasmid pKK388-1. The experiments were done simultaneously with both E. coli mutT and hMTH genes cloned in the same vector as positive controls. While E. coli mutT and hMTH totally complemented the E. coli mutT mutator phenotype, no complementation was observed with NUDT5 (data not shown). The dGTP hydrolase activity was also examined using the crude extract from E. coli mutT strain containing pKK388-1/NUDT5. No activity was detected (data not shown).
To further characterize the NUDT5 protein, we decided to purify the protein and to test its activity on potential substrates catalyzed not by MutT, but by other Nudix hydrolases. The hexahistidine-tagged recombinant proteins for both NUDT5 and mNudT5 were expressed in E. coli and purified to near homogeneity (Fig. 3). Both hexahistidine-tagged proteins have unusual tight binding to Ni 2ϩ -NTA resin as they eluted from the Ni 2ϩ -NTA column with 1 M imidazole at pH 8.0. The apparent molecular mass on the SDS-PAGE of both purified recombinant proteins were around 40 kDa, about 15 kDa bigger than the predicted molecular mass of both histidine-tagged proteins. Also the purified NUDT5 recombinant protein apparently has a doublet on the SDS-PAGE (Fig. 3).
Mass spectrometry was used as an alternative and more accurate method to investigate the apparent increased molecular mass and heterogeneity of both purified protein samples. LC/electrospray MS of the purified hexahistidine-tagged mNudT5 protein gave superimposable peaks of UV absorption and ion current eluting at around 28 min (approximately 20% acetonitrile). The mass spectrum yielded a molecular mass of 25429.4 Da (data not shown), which is close to the calculated mass of the protein (25425.9 Da). The same analysis on the NUDT5 revealed superimposable peaks of UV adsorption and ion current eluting in the same region of the chromatogram. The mass spectrum revealed microheterogeneity within the sample with three predominant and roughly equally abundant components with molecular masses of 26057.3, 26048.5, and 26017.2 Da. The difference in molecular masses between the lighter two components of 31.3 Da could be due to the inclusion of two oxygen atoms, possibly through oxidation of any two of the four Met and 10 His residues. The explanation for the difference of 8.5 Da between the heavier two forms is not obvious. This mass difference is incongruous with known protein covalent modifications. The observed molecular masses are to be compared with a calculated molecular mass of 25635.0 Da of the NUDT5, which is 382.2 Da lighter than the smallest of the observed forms. The explanation for this mass difference is obscure at the moment.
Substrate Specificity of the Recombinant Proteins-The first set of Nudix substrates tested were ADP-sugars including ADP-ribose and ADP-mannose. The reaction products were subjected to mass spectrometry analysis and the results from ADP-mannose hydrolysis by NUDT5 protein were shown in Fig. 4. The reaction products contain two new species of roughly equal amounts with the molecular mass correlated to AMP (346.1 Da) and mannose 5-phosphate (259.1 Da) suggesting that NUDT5 acts as a pyrophosphatase, which hydrolyzes ADP-mannose to generate products of AMP and mannose 5-phosphate. Similar results with AMP and ribose 5-phosphate peaks were obtained when the substrate ADP-ribose was used (data not shown).
Most of the characterized Nudix hydrolases require an alkaline pH and the presence of divalent ions to become fully active. The optimal pH was determined for both NUDT5 and mNudT5 with substrates ADP-ribose and ADP-mannose (Fig. 5). While mNudT5 has its optimal activity to both ADP-sugars at the expected alkaline pH around 9, the NUDT5 has an optimal FIG. 1. Nucleotide and deduced amino acid sequences of the human NUDT5 cDNA. The deduced amino acid sequence is shown in one-letter designation below the nucleotide sequence. The termination codon at the end of the ORF is represented by an asterisk. A Nudix signature sequence is boxed. The in-frame termination codon upstream of the initiation codon of the ORF is in bold.

Concensus
neutral pH around 7. The requirement for divalent ions was also studied. Both NUDT5 and mNudT5 require the presence of Mg 2ϩ to achieve optimal activity (1 mM Mg 2ϩ with NUDT5 and 5 mM Mg 2ϩ with mNudT5, data not shown). Similar catalytic activities were observed when Mg 2ϩ concentration was increased to 20 mM for both NUDT5 and mNudT5 (data not shown). Mn 2ϩ and Zn 2ϩ could partially substitute for Mg 2ϩ (data not shown). Other potential substrates for both NUDT5 and mNudT5 were tested and the results were shown in Fig. 6. Under the optimal pH 7 for NUDT5 with the presence of 5 mM Mg 2ϩ , NUDT5 prefers substrates containing ADP linked to sugar moieties, such as ADP-mannose, ADP-ribose, and ADP-glucose. The rate of hydrolysis decreased about 5-7-fold as the nucleoside in ADP-sugar changes from adenosine to guanosine or uridine. NUDT5 also displayed minor activities on NADH, as well as on Ap 2 A. A similar pattern of substrate specificity was observed with mNudT5 under its optimal alkaline pH (Fig. 6). Both NUDT5 and mNudT5 have no activities on ribo-and deoxyribonucleoside triphosphates (data not shown), which are the substrates for MutT (43) and another Nudix hydrolase, Orf17 (6).
The catalytic properties were studied on substrates ADPribose and ADP-mannose for both NUDT5 and mNudT5 (Table  II). NUDT5 has K m of 32 M for ADP-ribose and higher K m (83 M) for ADP-mannose. However, the V max of NUDT5 for ADPmannose is about 1.5 times that with ADP-ribose. Therefore, the overall catalytic efficiency (V max /K m ) of NUDT5 is higher for ADP-mannose than for ADP-ribose. mNudT5 has slightly different catalytic properties. For ADP-ribose mNudT5 has a K m similar to NUDT5, about 38 M. However, the K m for ADP-mannose for mNudT5 is 154 M, about four times as much as the K m for ADP-ribose. The V max of mNudT5 for ADPmannose is only about 1.2 times that with ADP-ribose. Therefore, the resulting overall catalytic efficiency (V max /K m ) of mNudT5 is higher for ADP-ribose than ADP-mannose. According to the standard practice of naming the enzyme after the substrate with the lowest K m , we suggest the name ADP-ribose pyrophosphatase for NUDT5 protein.
Tissue-specific Expression of NUDT5 mRNA-Northern blot of mouse poly(A) ϩ RNA isolated from various tissues was carried out using mNudT5 probe containing exons 1-4. It revealed a major transcript of 1.35 kb in all eight adult mouse tissues analyzed, with the most abundant expression in liver (Fig. 7). The expression of NUDT5 mRNA was analyzed using the Human Multiple Tissue Expression Array (Fig. 8). Abundant expression of NUDT5 mRNA was observed in liver, pituitary  (Table III, 6. Substrate specificity of NUDT5 and mNudT5. Purified NUDT5 or mNudT5 was incubated with 2 mM substrate at 37°C for 15 min in 50 mM Tris buffer (pH 7 for NUDT5 and pH 9 for mNudT5). The hydrolysis of the substrates was assayed using the colorimetric procedure described under "Experimental Procedures." Each result represents the mean Ϯ S.D. from three experiments.

TABLE II Kinetic parameters for NUDT5 and mNudT5
The colorimetric procedure described under "Experimental Procedures" was used with concentrations of 0.01 mM to 2 mM for all substrates. K m and V max were determined from a nonlinear regression analysis (47). A unit of enzyme hydrolyzes 1 mol of substrate/min. Each result represents the mean Ϯ S.D. from three experiments.  For determination of mNudT5 genomic structure one recombinant phage containing exons 1 to 6 of the mNudT5 gene was identified and partially sequenced to determine the exonintron boundaries. Exons 7 to 9 were determined by analyzing two overlapping PCR products covering partial intron 6 to the 3Ј-untranslated region. mNudT5 is approximately 13 kb in length and is divided into 9 exons (Table III, part b). All 5Ј donor and 3Ј acceptor splice sites conform to the consensus GT/AG rule (44). Both NUDT5 and mNudT5 have introns in the same position of their coding sequence.
Both NUDT5 and mNudT5 were mapped on their corresponding chromosomes using radiation hybrid mapping (37,38) as described under "Experimental Procedures." The radiation hybrid data placed mNudT5 on mouse chromosome 2 between markers D2Mit354 and D2Mit76 (Fig. 9). The LOD score was 13.6 between D2Mit354 and mNudT5, and 17.1 between mNudT5 and D2Mit76. According to genetic mapping data from the 1999 Chromosome Committee Reports available from the Jackson Laboratory website, the D2Mit354 and D2Mit76 are located at 0 and 2 centiMorgan (cM), respectively, on mouse chromosome 2. Based on these markers, the physical map position for mNudT5 corresponds to a locus at approximately 1 cM of chromosome 2 on the genetic map (Fig. 9). Murine vimentin (Vim) gene and leukemia viral oncogene homolog (Bmi1), which have been previously mapped by backcross analysis to 7 and 9 cM, respectively, of mouse chromosome 2 (1999 Chromosome Committee Reports available from the Jackson Laboratory website), are also indicated on Fig. 9.
The radiation hybrid data placed NUDT5 on human chromosome 10 between markers WI-4124 and WI-8819 (Fig. 9). The LOD score was 15 between WI-4124 and NUDT5, and 15 be-tween NUDT5 and WI-8819. Based on the integrated map of human chromosome 10 available from the Whitehead Institute, the physical map locations of WI-4124 and WI-8819 correspond to 28 -29 and 29 -32 cM, respectively (Fig. 9). Based on these markers, the physical map position for NUDT5 correspond to a locus at 29 -32 cM of chromosome 10 on the genetic map ( Fig.  9). Human VIM and BMI1, which have been previously mapped by radiation hybrid to human chromosome 10 as well (data were obtained from Gene Map'98 provided by the International radiation hybrid mapping consortium through the NCBI website), are approximately 21 centiRad and 43 centiRad from the NUDT5 locus (Fig. 9). Our data indicate that this portion of human chromosome 10 that contains NUDT5, VIM, and BMI1 is syntenic to the proximal portion of mouse chromosome 2 that contains mNudT5, Vim, and Bmi1.
Phylogenetic Analysis of ADP-ribose Hydrolases-The amino acid sequences of characterized and putative homologs of ADPribose pyrophosphatases and MutT were compared using multiple alignments from Pfam (40), additional manual alignments, and PAUP (39). As shown in Fig. 10, there are roughly three major branches; notably, ScYSA1, EcORF209, Hi0398, and BsYQKG are clustered near NUDT5 proteins on the phylogenetic tree. EcORF186 is remotely related to NUDT5 proteins. Mj1149 and slr0787 are clustered on a different branch with their putative sequence homologs from several bacteria and Archaea. DISCUSSION In this report, we describe the cDNA isolation and functional characterization of NUDT5, a Nudix hydrolase that catalyzes the hydrolysis of ADP-sugars to AMP and sugar 5-phosphate with preference for ADP-ribose. Initiated by sequence homol-  Several biochemically distinct ADP-ribose pyrophosphatase activities have been identified in mammalian tissues. In rat liver, three types of enzymes that hydrolyze ADP-ribose and other related substrates have been described (30). One type strictly requires the presence of Mn 2ϩ rather than Mg 2ϩ . The other two types, ADPRibase I and ADPRibase II, require the presence of Mg 2ϩ to be fully functional. The catalytic property of NUDT5 is closer to ADPRibase II. ADPRibase II has K m of 50 M for ADP-ribose with substrate specificity similar as NUDT5 (30). ADPRibase I, on the other hand, is highly specific for ADP-ribose with a low K m of 0.5 M.
NUDT5 is different from the three characterized Nudix hydrolases, ORF186 from E. coli, MJ1149 from M. jannaschii, and slr0787 from Synechocystis sp. Functionally, they have different substrate specificity. The slr0787 and the thermostable MJ1149 is highly specific for ADP-ribose and its closely related derivative, 2Ј-phospho-ADP-ribose (15,16), while ORF186 has similar catalytic activities on three substrates, ADP-ribose, NADH, and AP 3 A (7). For NUDT5, the major substrates are ADP-sugars including ADP-ribose, ADP-mannose, and ADPglucose. NUDT5 can hydrolyze NADH and Ap 2 A, but with 5-10-fold lower efficiency. Structurally, NUDT5, MJ1149, and ORF186 have diverged amino acid sequences, despite having the conserved Nudix box. These enzymes are remotely related on the phylogenetic tree, which suggests that they may belong to different subfamilies of ADP-ribose pyrophosphatases. Several ORFs (1, 7), ScYSA1, EcORF209, Hi0398, and BsYQKG, are clustered with NUDT5 on the phylogenetic tree, which suggest that they might be functional homologs of NUDT5 with similar catalytic properties. 2 Recently, a human "YSA1" homolog was mentioned to have NDP-sugar pyrophosphatase activity (5), which is similar to NUDT5 described in this paper. 3 The physiological function of NUDT5 is not fully understood. It has been proposed that the function of ADP-ribose pyrophosphatase is to remove the ADP-ribose, which is a potentially deleterious metabolite and to recycle it by hydrolyzing to AMP and ribose 5-phosphate (1,15). Perhaps the final understanding of the physiological function of ADP-ribose pyrophosphatase would be to study the mutant phenotype lacking the activity of ADP-ribose pyrophosphatase. Although the potential overlapping substrate spectra between Nudix hydrolases make this task difficult, cloning and characterization of NUDT5 certainly provide the first step toward understanding the role of ADP-ribose pyrophosphatase in vivo. NUDT5 is preferentially expressed in liver, pituitary gland, and placenta, which may indicate the importance of NUDT5 activity in these tissues. NUDT5 maps to human chromosome 10 between markers WI-4124 and WI-8819 at the end of the short arm of chromosome 10. Our mapping is consistent with data from GenMap'99 where the ESTs similar to ScYSA1 were mapped on chromosome 10 between markers: D10S189 and D10S191. According to this server, the gene mapped in the similar region (D10S189 and D10S191) on chromosome 10 is FIP2, coding for tumor necrosis factor. In close vicinity is the gene encoding phytanic acid hydrolase, which has been linked to a Refsum disease, a rare disorder of lipid metabolism. Screening through the Online Mendelian Inheritance in Man data base (OMIM) revealed that several other disorders have been mapped to the region, that at least partially overlapped with the localization of NUDT5, and have not been connected with any specific gene. They are HDR (hyperthyroidism, sensorineural deafness, and renal dysplasia), prostate adenocarconoma 1, DiGeorge syndrome, glaucoma1, and athabaskan severe combined immunodeficiency.
It is worth noting that during the cloning of the genomic NUDT5, we have identified a pseudogene (data not shown) that had all the attributes of a processed pseudogene (45). It extended from the first base of the cDNA to the polyadenylation site, contained no introns, and was lacking the ATG translation start codon. The pseudogene was flanked by 15-bp direct repeats: GAAAAGATGAGCCAT. Two frameshifts caused by 3 deletions and 4 insertions and an in-frame stop codon interrupted the reading frame of this pseudogene. The overall amino acid sequence identity between the pseudogene and the active gene is 86.7%. The sequence of this pseudogene is consistent with a reported nucleotide sequence (accession number Z95152) on human chromosome 6p21.1-21.33.
As an increasing number of complete genome sequences become available, more and more putative Nudix hydrolases are identified. So far, every characterized genome contains at least one putative Nudix protein (15). In E. coli, it has 10, among which 6 are characterized with different enzymatic activity (1). It becomes a challenge to annotate these putative Nudix proteins. Phylogenetic analysis may provide a powerful tool to cluster protein homologs together according to their unique conserved motifs. One could predict the function of the putative ORF if it is clustered with a characterized Nudix protein or one could identify a putative "novel Nudix protein" if it is clustered with the unknowns. The Nudix motif has been proven to be an excellent scaffold for enzymes that hydrolyze nucleoside diphosphate derivatives or related substrates.