The Nudix Hydrolase Ndx1 from Thermus thermophilus HB8 Is a Diadenosine Hexaphosphate Hydrolase with a Novel Activity*

The ndx1 gene, which encodes a Nudix protein, was cloned from the extremely thermophilic bacterium Thermus thermophilus HB8. This gene encodes a 126-amino acid protein that includes the characteristic Nudix motif conserved among Nudix proteins. Ndx1 was overexpressed in Escherichia coli and purified. Ndx1 was stable up to 95 °C and at extreme pH. Size exclusion chromatography indicated that Ndx1 was monomeric in solution. Ndx1 specifically hydrolyzed (di)adenosine polyphosphates but not ATP or diadenosine triphosphate, and it always generated ATP as the product. Diadenosine hexaphosphate (Ap6A), the most preferred substrate, was hydrolyzed to produce two ATP molecules, which is a novel hydrolysis mode for Ap6A, with a Km of 1.4 μm and a kcat of 4.1 s–1. These results indicate that Ndx1 is a (di)adenosine polyphosphate hydrolase. Ndx1 activity required the presence of the divalent cations Mn2+, Mg2+, Zn2+, and Co2+, whereas Ca2+, Ni2+, and Cu2+ were not able to activate Ndx1. Fluoride ion inhibited Ndx1 activity via a non-competitive mechanism. Optimal activity for Ap6A was observed at around pH 8.0 and about 70 °C. We found two important residues with pKa values of 6.1 and 9.6 in the free enzyme and pKa values of 7.9 and 10.0 in the substrate-enzyme complex. Kinetic studies of proteins with amino acid substitutions suggested that Glu-46 and Glu-50 were conserved residues in the Nudix motif and were involved in catalysis. Trp-26 was likely involved in enzyme-substrate interactions based on fluorescence measurements. Based on these results, the mechanism of substrate recognition and catalysis are discussed.

The Nudix hydrolase family comprises enzymes that catalyze a reaction where the substrate is a nucleoside diphosphate linked to another moiety, X (1). This family is characterized by the Nudix motif GX 5 EX 7 REUXEEXGU, where X is any amino acid, and U is one of the bulky hydrophobic amino acids, Ile, Leu, or Val (1). This motif forms a loop-helix-loop structure that is involved in substrate binding and catalysis (2,3). These enzymes are found in all kingdoms. It has been proposed that the function of Nudix proteins is housecleaning to eliminate potentially toxic nucleotide metabolites from the cell and to regulate the concentrations of nucleoside diphosphate derivatives (1).
A recent Pfam (4) search of the data bases for the Nudix signature sequence has revealed about 1100 open reading frames from more than 250 species ranging from viruses to humans, and about 70 of the gene products have been identified. These products hydrolyze nucleoside diphosphate derivatives such as (deoxy)nucleoside triphosphate (5), nucleotide sugar (6 -8), dinucleotide polyphosphate (9 -11), NADH (12), and coenzyme A (13). One example is the MutT protein, which degrades 8-oxo-deoxyguanine triphosphate to prevent mutations caused by oxidation of guanine nucleotides (14). Furthermore, diphosphoinositol polyphosphate (DIPP) 1 (15) and phosphoribosyl pyrophosphate have also been reported as substrates for Nudix hydrolases (16). Although preferred substrates have been identified for some Nudix enzymes, the functions and molecular mechanisms, including substrate recognition, remain to be elucidated.
Thermus thermophilus HB8 is a Gram-negative bacterium that grows at temperatures above 75°C (17). It is the most thermophilic eubacterium for which a gene manipulation system has been established (18 -20). Proteins from this bacterium are stable against heat and are, thus, suitable for physicochemical studies, including x-ray crystallography. We selected T. thermophilus HB8 for the systematic study of the structures and functions of all proteins from a single organism in a project named the Whole Cell Project (21,22). Interestingly, Deinococcus radiodurans, which is very closely related to T. thermophilus, possesses twenty-three Nudix genes (23). This bacterium is characterized by extraordinary resistance to ionizing radiation; it is thought that some of its Nudix proteins may be associated with novel DNA repair pathways (24). However, the diversity of substrates and functions of Nudix proteins in vivo remain unclear. Such unique features make proteins of this family good targets for structural and functional proteomics.
Therefore, we aimed to investigate the molecular mechanism and physiological functions of Nudix proteins from T. thermophilus HB8. In this work, we describe the overexpression and purification of T. thermophilus HB8 Ndx1 protein. The enzymatic activity and the biochemical properties of Ndx1 are also described.
Overexpression of the ndx1 Gene-Preliminary sequence data for the T. thermophilus HB8 ndx1 gene, which contains the Nudix motif, was provided by the T. thermophilus HB8 genome project (21,22). Using this information, two primers for amplification of the target gene were synthesized, and PCR (polymerase chain reaction) was carried out using these primer and LA Taq polymerase. The primer sequences were 5Ј-ATATCATATGGAGCTAGGGGCCGGGGGCGTGGTCTT-3Ј and 5Ј-ATATAGATCTTTATTAAAGCGGTAGACGCTCAAGGG-3Ј, and the underlining indicates NdeI and BglII sites. The amplified gene fragment was ligated into pET-11b (Novagen) using NdeI and BamHI sites following TA cloning and sequencing. E. coli BL21(DE3) cells transformed by the resulting plasmid were grown at 37°C to 5 ϫ 10 8 cell/ml on 1.5 liters of LB medium containing ampicillin. The cells were then incubated for 4 h in the presence of isopropyl-␤-D-thiogalactopyranoside, harvested by centrifugation, and stored at Ϫ20°C.
Purification of Ndx1-All purification steps described below were carried out at room temperature. Frozen cells (3 g) were suspended in 30 ml of lysis buffer (50 mM Tris-HCl, 0.1 mM EDTA, 10% (w/v) glycerol, and 20% (w/v) saccharose (pH 8.5)) and disrupted by ultrasonication on ice. After Brij-58 was added to a final concentration of 0.2% (w/v), the cell lysate was stirred for 1 h at 4°C. Next, the lysate was incubated at 70°C for 10 min and centrifuged (about 30,000 ϫ g) for 60 min. Then the supernatant was dialyzed against buffer I (50 mM Tris-HCl, 0.1 mM EDTA, and 10% (w/v) glycerol (pH 8.5)) for 16 h. The dialysate was applied to a Toyopearl-SuperQ column (bed volume 12 ml) that had been equilibrated with buffer I. Proteins were eluted with a linear gradient of NaCl from 0 to 1.0 M in a total volume of 60 ml of buffer I. Solid ammonium sulfate was added to the fractions containing the Ndx1 protein to a final concentration of 15% saturation. The protein solution was then applied to a Toyopearl-Phenyl 650M column (bed volume 10 ml) previously equilibrated with buffer I containing 15% saturated ammonium sulfate. The proteins were eluted with a linear gradient of ammonium sulfate from 15 to 0% saturation in a total volume of 50 ml of buffer I. Fractions containing the Ndx1 protein were collected and concentrated using a Vivaspin (5000 cut off) concentrator. The concentrated solution was applied to a Superdex 75 10/30 column (bed volume 24 ml) previously equilibrated with buffer II (50 mM Tris-HCl and 100 mM KCl (pH 7.5)) and eluted with the same buffer using ⌬KTA explorer (Amersham Biosciences). The fraction containing the Ndx1 protein was concentrated and stored at 4°C. At each chromatography, the elution profile was assessed by SDS-PAGE containing 12% (w/v) acrylamide. The N-terminal sequence of the purified Ndx1 was analyzed on an Applied Biosystems 473A protein sequencer.
CD Spectrometry-Circular dichroism (CD) measurements were carried out with a Jasco spectropolarimeter, model J-720W. CD spectra of 5 M Ndx1 were measured in a 1-mm cell in the far-UV region between 200 and 250 nm. Measurements were performed after incubation at 25°C in 50 mM potassium phosphate (pH 7.5) and 100 mM KCl. Thermostability was assessed by measuring CD values at 222 nm at a 1°C/min rate. CD data were converted to the mean residue ellipticity, [], in deg cm 2 dmol Ϫ1 .
Size Exclusion Chromatography-The oligomeric state of Ndx1 in solution was assessed by size exclusion chromatography on Superdex 75 HR 10/30. A sample contained 50 mM Tris-HCl (pH 7.5), 100 mM KCl, and 0.1 M Ndx1, which were the same conditions used to assay Ndx1 activity. The protein was eluted with 50 mM Tris-HCl and 100 mM KCl (pH 7.5) with a flow rate of 0.5 ml/min by the ⌬KTA system. The molecular weight of Ndx1 was estimated using molecular weight marker proteins (Sigma).
Hydrolase Activity Assay-The activity of Ndx1 was measured by quantifying the amounts of substrates and products with an ion-pair reversed-phase high performance chromatography according to a slightly modified method of Samizo et al. (25). Reaction mixtures (100 l) containing 50 mM Tris-HCl, 100 mM KCl, 5 mM MgCl 2 , substrate, and Ndx1 were incubated at 25°C. The reaction was stopped by adding 100 l of 100 mM EDTA, and the protein was removed by ultrafiltration using a membrane filter. The 100-l aliquot of the filtrate was applied to a reversed-phase column (CAPCELL PAK C18, 4.6 ϫ 75 mm), which was equilibrated with 20 mM sodium phosphate (pH 7.0), 5 mM tetran-butyl ammonium phosphate, and 10% methanol. Elution was performed by a gradient of 10 -50% methanol. Nucleotides were detected at 260 nm, and their identifications were based on their retention times. Their concentrations were calculated by the integration of their respective peak areas. Concentrations of substrates were varied between 0.2 and 16 M. Initial velocity was calculated from product concentration and plotted against substrate concentration. These were fitted to the Michaelis-Menten equation and Hanes-Woolf plot, and the kinetic constant was calculated using the software Igor Pro 3.14 (Wave Metrics).
Fitting Plot of pH-dependent Assay-Assuming Scheme 1 for the Ndx1 reaction, the following equations could be generated (26). In short, there are two residues that are related to catalysis by Ndx1.
The pK a values of one residue are pK e1 and pK es1 at free enzyme and complex, respectively. Similarly, those of the other residue are pK e2 and pK es2 . These pK a values were calculated by fitting the data to Equations 1-3 using the software Igor Pro 3.14 (Wave Metrics) Fluorescence Measurements-The fluorescence emission of Ndx1 was measured with a Hitachi spectrofluorometer, model F-4500. All measurements were taken with an excitation wavelength of 295 nm in a 5 ϫ 5-mm quartz cuvette at 25°C. The fluorescence titration was carried out by measuring the emission intensity at 328 nm. The reaction mixture (200 M) contained 50 mM Tris-HCl, 100 mM KCl, 5 mM MgCl 2 , 0.5 M Ndx1, and various concentrations of ATP (pH 7.5).
Site-directed Mutagenesis-Pairs of oligonucleotides about 30 bases in length with melting temperatures of about 67°C and containing the desired substitutions were designed. The ndx1 gene includes NruI and PshAI sites at the 60th and 155th positions, respectively. For W26A substitution, the region between the NruI site and BglII site at the 3Ј terminus of the gene was amplified by PCR. For R45K, E46Q, E49Q, and E50Q substitution, the regions between the 5Ј terminal NdeI site and PshAI site were amplified. The amplified DNA fragments were cloned into the pT7Blue vector by TA cloning. E. coli DH5␣ were transformed with the constructed plasmid DNA and cultured. Substitution at the desired positions was confirmed by sequencing. Then the wild-type fragments in the expression vector were replaced by the confirmed fragments containing the mutations.

RESULTS
Preparation of Ndx1-Using information from the T. thermophilus HB8 genome project, we identified eight open reading frames containing Nudix motifs and named them ndx1 to ndx8. The ndx1 gene product (DDBJ/EMBL/GenBank TM accession number AB125632; project code 1331) comprises 126 amino acids, has a molecular mass of 14.2 kDa, and has a theoretical pI of 4.8. When a BLAST search was carried out using the Ndx1 sequence as a query, Ndx1 was the most similar (about 25% identity) to Ap 4 A hydrolases from Caenorhabditis elegans, hu-SCHEME 1 man, and pig ( Fig. 1A). However, the sequence similarity was restricted to the surroundings of the Nudix motif and was not enough to determine whether Ndx1 is an Ap 4 A hydrolase or not.
Thus, Ndx1 was overexpressed in E. coli BL21(DE3) under the control of an isopropyl-␤-D-thiogalactopyranoside -inducible T7 promoter. The induced band at ϳ15 kDa (corresponding to the size of Ndx1) was observed in the soluble fraction (Fig. 2), and we purified the protein to homogeneity utilizing three column chromatography steps (see details under "Experimental Procedures"). The sequence of the N-terminal nine residues of the overexpressed protein agreed with the residues predicted from the ndx1 sequence, confirming that the purified protein was Ndx1. Approximately 25 mg of Ndx1 was obtained from 3 g of cells.
Physicochemical Properties-Size exclusion chromatography was performed to investigate the oligomeric state of Ndx1. The elution profile of Ndx1 showed a single peak (Fig. 2B). The apparent molecular mass corresponding to the peak was estimated to be 17 kDa, which was similar to the 14.2-kDa mass calculated from the sequence. This indicates that Ndx1 exists in a monomeric state in solution.
The far-UV CD spectrum of the purified Ndx1 showed negative double maxima at 209 and 220 nm (Fig. 3A), characteristic of an ␣-helical structure. The stability of Ndx1 to temperature and pH was examined based on the mean residue ellipticity at 222 nm ([] 222 ). Ndx1 was stable up to 95°C at pH 7.5 (Fig. 3B) and stable in a wide range of pH at 25°C (Fig. 3C).
Enzymatic Activity-Most Nudix proteins examined to date are nucleotide pyrophosphatases that hydrolyze a nucleoside diphosphate linked to another moiety (1). Therefore, enzymatic activity of Ndx1 was examined for a wide range of nucleotides known to be substrates of other Nudix proteins. Ndx1 was inactive toward the following nucleotides when assayed at 50 M: 5Ј-(deoxy)nucleoside triphosphates, 5Ј-nucleoside diphosphates, nucleoside diphosphate sugars, NADH, NAD ϩ , CoA, and acetyl CoA. Significant activity was found toward dinucleotide polyphosphates and nucleotide polyphosphates. The respective substrates yielded products as follows: Ap 6 A, 2ATP; Ap 5 A, ATP and ADP; Ap 4 A, ATP and AMP; p 4 A, ATP and inorganic orthophosphate (Table I). In all cases ATP was generated as a product. ATP and Ap 3 A were not hydrolyzed by Ndx1. These data indicate that Ndx1 protein has ATP-gener-ating (di)nucleotide polyphosphate hydrolase activity. Table I shows the steady-state kinetic constants for each active substrate assuming Michaelis-Menten type reactions. Whereas the K m values for these substrates were all about 1 M, the catalytic constants (k cat ) varied among the tested substrates (Table I). Based on the catalytic efficiencies (k cat /K m ), the highest activity was observed for Ap 6 A followed by p 4 A, Ap 5 A, and Ap 4 A ( Table I). The substrate preference was Ap 6 A Ͼ p 4 A Ͼ Ͼ Ap 5 A Ͼ Ap 4 A for polyphosphates and adenosine Ͼ guanosine for the base. Therefore, we conclude that Ndx1 is an ATP-generating Ap 6 A hydrolase. Among the known enzymes that specifically hydrolyze Ap 6 A, Ndx1 is the only enzyme that symmetrically hydrolyzes Ap 6 A.
Divalent metal ions were essential for Ndx1 activity, which is a common property of Nudix proteins (27). The effect of several divalent metal ions (each 5 mM) on Ndx1 activity for Ap 6 A was investigated. Significant activity was observed in the presence of Mn 2ϩ , Mg 2ϩ , and Zn 2ϩ with the apparent rate constant of k app values of 8.6, 4.2, and 4.1 s Ϫ1 , respectively. In contrast, there was low activity in the presence of Co 2ϩ (k app , 0.51 s Ϫ1 ), and there was no activity in the presence of Ca 2ϩ , Ni 2ϩ , and Cu 2ϩ . Among monovalent ions, fluoride ion showed strong inhibition of the Ndx1 activity. Dependence of the inhibitory effect on fluoride ion concentration revealed that the inhibition was in a non-competitive manner with a K i of 424 M toward free enzyme and a K i of 80 M toward complex.
Ndx1 exhibited higher activity at higher pH and little activity at lower pH. This observation is typical for the Nudix class of enzymes, which usually have optimal pH values in the alkaline range. The presence of two pK a values (7.9 and 10.0) was found in the plot of k cat against pH (Fig. 4A). The optimal pH for Ndx1 activity was about 8, judged by the catalytic efficiency k cat /K m (Fig. 4B). The plot of the k app against temperature was bell-shaped; the optimal temperature for Ndx1 activity was 70°C. This higher activity at higher temperature reflects a common property of enzymes from T. thermophilus.
Protein-Substrate Interaction-The K m values of Ndx1 for several (di)adenosine polyphosphates were about 1 M. Because these results suggest that the affinity for these substrates is almost the same, we hypothesized that Ndx1 recognized a common moiety of the substrates at the initial binding phase. Although several tertiary structures of Nudix proteins have been reported (2, 3, 28, 29), detailed investigations of substrate binding mechanisms have not often been performed by biochemical methods. The adenosine phosphate moiety is commonly contained in Ndx1 substrates, and Ndx1 has no activity toward ATP. Therefore, we measured the intrinsic fluorescence spectrum of Ndx1 upon binding to ATP to investigate the interaction between the protein and the adenosine phosphate moiety. The emission spectra were measured using an excitation wavelength of 295 nm, which excited only tryptophan residues. When ATP was added to Ndx1, the fluorescence intensity at around 328 nm decreased (Fig. 5A, bold solid line). As the ATP concentration increased, the fluorescence intensity gradually decreased. These results suggest that the decrease in fluorescence intensity reflects the protein-substrate interaction. When dATP was used as a ligand, the fluorescence intensity also decreased but not equivalently to ATP. When the change in fluorescence intensity was plotted against ligand concentration, K d was determined to be 13 M for ATP and 36 M for dATP based on bimolecular binding reaction (Fig. 5B). The affinity of Ndx1 for these nucleotides was also investigated by examining the inhibition of ATP and dATP on Ap 6 A hydrolysis by Ndx1 (not data shown). These results showed that K i was 13 M for ATP and 41 M for dATP. When Mg 2ϩ was omitted from the reaction mixture, no fluorescence change was observed (Fig. 5A, dotted line). Also, when GTP was used, the fluorescence intensity did not decrease, as is predicted by the low affinity of Ndx1 toward diguanosine polyphosphates.
Among the four Trp residues of Ndx1, Trp-26 was conserved in the N-terminal half of the Nudix motif in the sequence of dinucleotide polyphosphate hydrolases (Fig. 1B). This raised the possibility that the observed fluorescence changes could be ascribed to Trp-26. This hypothesis was confirmed by the observation that the mutant W26A Ndx1, in which Trp-26 was replaced by alanine, showed no decrease in fluorescence intensity upon the addition of ATP (Fig. 5C).
Catalytic Residues-From the dependence of Ndx1 hydrolysis activity on pH, pK a values of 7.9 and 10.0 were obtained from the k cat plot (Fig. 4A), and pK a values of 6.1 and 9.6 were obtained from the plot of k cat /K m (Fig. 4B). These pK a changes demonstrated that when free Ndx1 bound to the substrate, the pK a of one residue changed from 6.1 to 7.9, and the pK a of a second residue changed from 9.6 to 10.0. It is thought that two residues whose pK a values change play an important role in Ndx1 hydrolysis.
In Ndx1, the conserved Nudix motif is located between Gly-31 and Val-53. In this motif, Glu-37, Arg-45, Glu-46, Glu-49, and Glu-50 are highly conserved across the sequences of Nudix proteins from plants, animals, and bacteria (1). When the glutamic acid residues (Glu-46, Glu-49, and Glu-50) in the Nudix motif were replaced by glutamine, the E46Q and E50Q mutants showed a 2.2 ϫ 10 4 -fold reduction and a 1.3 ϫ 10 5 -fold reduction in k cat , respectively (Table II). In contrast, the E49Q mutation had very little effect on activity (Table II). Moreover, mutation of the conserved Arg-45 residue to lysine only slightly reduced the activity (Table II). DISCUSSION Many Nudix genes have been found in organisms whose genomes have been sequenced. The human genome project has identified at least 18 Nudix genes, some of which are 90% alike in amino acid sequence (30). In contrast, the sequences of bacterial Nudix proteins have weaker similarity to each other within a species and among species. For example, the E. coli genome contains 11 Nudix genes, whose products do not highly resemble each other (about 20% identity or less) and have different substrate specificities. The genome of D. radiodurans contains at least 23 Nudix genes, which is the largest number of Nudix genes found in any species (23). The products of these genes show diversity in size as well as in sequence; they include larger proteins (about 40 -60 kDa) than the average Nudix protein (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). The T. thermophilus HB8 genome project has revealed eight Nudix genes, designated as ndx1 to ndx8. These gene products are not highly similar to each other (about 20% identity) except for the region around the Nudix motif. Although T. thermophilus is very closely related to D. radiodurans, only 2 of its Nudix proteins show a relatively high sequence similarity (54 and 40%) to Nudix proteins of D. radiodurans. The low sequence similarity found in the Nudix proteins in respective bacteria may suggest that these proteins have diverse functions.
In this study we have demonstrated that Ndx1 has the ability to hydrolyze diadenosine polyphosphates and showed the highest activity when Ap 6 A was used as a substrate. Proteins known to hydrolyze Ap 6 A include Schizosaccharomyces pombe Aps1 (31), Saccharomyces cerevisiae YOR163w (32), and several human proteins, hAps1, hAps2, hDIPP-1, hDIPP-2␣, and hDIPP-2␤ (30,33). Ndx1 has two features not found in these Ap 6 A hydrolases. The first is that Ndx1 cleaves Ap 6 A at a single cleavage site, the bond between P3 and P4, leading to the production of two ATP molecules. In contrast, the other Ap 6 A hydrolases cleave their substrates at multiple sites. YOR163w and Aps1 produce ADP ϩ p 4 A as major products and AMP ϩ p 5 A and 2 ATP as minor products (31,32). For hAps1, hAps2, hDIPP-1, hDIPP-2␣, and hDIPP-2␤, the major and minor products are AMP ϩ p 5 A and ADP ϩ p 4 A, respectively (30,33). The second feature that distinguishes Ndx1 from others is that Ndx1 produced ATP from several substrates. For example, Ndx1 produced ATP not only from Ap 6 A but also from Ap 5 A, Ap 4 A, and p 4 A. ATP was not hydrolyzed by Ndx1. This contrasts with studies of other Ap 6 A hydrolases; from the same substrates, YOR163w always produces p 4 A (YOR163w also produces ATP and P i from p 4 A, but with low efficiency (32)), Aps1 always produces ADP (31), and hAps1, hAps2, hDIPP-1, hDIPP-2␣, and hDIPP-2␤ always produce AMP. In short, Ndx1 hydrolyzes (di)adenosine polyphosphates in a strict manner, which always generates ATP. Generation of ATP as a product is also a characteristic of Ap 4 A hydrolases (9 -11). Interestingly, Ndx1 shows high sequence similarity to Ap 4 A hydrolases (about 30% identity), especially animal-type enzymes (C. elegans and Homo sapiens, etc.) compared with Ap 6 A hydrolases (about 25% identity or less) (Fig. 1A). This implies a phylogenetic relationship between Ndx1 Ap 6 A hydrolase and Ap 4 A hydrolases.
The features described above may be due to highly specific substrate recognition by Ndx1. The lower value of K m for Ap 4 A than Gp 4 G indicates that Ndx1 prefers adenine to guanine as a base moiety. In addition, Ndx1 had lower affinity for dATP than for ATP, suggesting involvement of the sugar moiety in substrate recognition. ATP was not hydrolyzed by Ndx1, but its K d was 13 M, based on the intrinsic fluorescence change. ADP did not decrease the fluorescence intensity, indicating no binding to the enzyme. In addition, Ndx1 showed almost the same K m values (about 1 M) toward diadenosine polyphosphate (Ap n A) (n ϭ 4ϳ6). These data strongly suggest that the ATP moiety of Ap n A (n ϭ 4ϳ6) significantly contributes to the affinity of these substrates for Ndx1. However, the k cat values were different with each substrate, and its value was highest for Ap 6 A. We hypothesize that the ATP moiety contributes mainly at the initial binding phase.
Fluorescence measurements using ATP also provided some information about the recognition site of Ndx1 for the ATP moiety. The substitution of Trp-26, a highly conserved residue, for Ala resulted in no fluorescence change. The reaction kinetics of the W26A mutant showed a lower rate (k cat ϭ 1.3 s Ϫ1 ) and lower affinity (K m ϭ 19 M) than the wild-type protein (Table  II). In the crystal structure of Ap 4 A hydrolase from Lupinus angustifolius (29), the tryptophan residue corresponding to Trp-26 of Ndx1 is located at the substrate-binding site. These observations imply that Trp-26 of Ndx1 interacts with the substrate, possibly via stacking of the adenine base moiety of the substrate.
In addition, it should be noted that at extreme pH (11.5), minor products (p 4 A ϩ ADP) were observed in addition to the major products (2 ATP) (data not shown). This supports the notion that the residue whose the pK a value changed from 9.6 to 10.0 when free Ndx1 bound to the substrate is related to binding the substrate. This residue is probably Lys or Arg and   (34). This feature may be related to a substrate, which includes some dissociative groups in its structure. Another characteristic of Ndx1 reaction is that the k cat value for Ap n A (n ϭ 4ϳ6) was related to the length of the phosphate FIG. 4. Nudix activity depending on pH. The Ndx1 activity assay was performed at several pH levels, and the kinetic parameters were calculated using the Michaelis-Menten equation. The buffers used were: 50 mM HEPES (pH 6.74 -7.05), 50 mM Tris-HCl (pH 7.5-8.8), and 50 mM Gly-NaOH (pH 9.06 -10.66), each of which contained 100 mM KCl. A, k cat , plotted relative to pH. This plot was fitted to Equation 1 (see "Experimental Procedures"), and the theoretical curve was drawn. B, k cat /K m , plotted relative to pH. This plot was fitted as in A using Equation 3. C, K m , plotted relative to pH. This plot was fitted as in A using Equation 5. group of the substrate. Comparison of the k cat for Ap 5 A and Ap 6 A suggests that P6 is important for catalysis. In addition to the ATP moiety containing the P1-P3 region, a contact of Ndx1 at around the P6 site and/or the terminal adenosine may be required for stability of the substrate in the transition state. As mentioned above, Ndx1 has a higher sequence similarity to Ap 4 A hydrolases than Ap 6 A hydrolases and generates ATP as a product in a similar manner as Ap 4 A hydrolases. Nevertheless, Ap 6 A was the preferred substrate for Ndx1. This interesting contradiction may be resolved by elucidation of the tertiary structure of Ndx1 complexed with the substrate (or its analogue).
As for the catalytic step, we determined that the Glu-46 and Glu-50 residues contributed to catalysis from the assay of several mutants. This result is in agreement with previously reported mutant experiments about other enzymes (35)(36)(37). The reaction mechanism of Nudix proteins was proposed from the study of tertiary structure (2,3,38) and NMR (39 -41). This mechanism suggests that a glutamic acid residue is coordinated by Mg 2ϩ , and a water molecule coordinating to this Mg 2ϩ directly attacks the substrate (2,41).
The catalytic residue may be the residue whose pK a changed from 6.1 to 7.9 because Ndx1 has no activity in acid pH (Fig.  4B). It is possible that this residue is His, judged by the pK a . One candidate is His-32 in the Nudix motif region, although this residue is not completely conserved among the Ap 4 A hydrolase (Fig. 1B). A distal histidine in associated with the true catalytic residue could be a candidate. Protonation of such a histidine would affect the catalysis via a hydrogen bond network. Alternatively, Glu-46 or Glu-50 could be a candidate for the residue showing the pK a change if the pK a is shifted to neutral pH range in a microenvironment of the active site.
Another possibility is that the hydration of Mg 2ϩ has the pK a changed from 6.1 to 7.9. It is suggested that Mg 2ϩ is coordinated by the glutamic acid residue, which is most likely to be Glu-46 or Glu-50 for Ndx1. Mn 2ϩ , Zn 2ϩ , Mg 2ϩ , and Co 2ϩ activated Ndx1, but Cu 2ϩ , Ca 2ϩ , and Ni 2ϩ were unable to activate Ndx1. Although Ndx1 activity was the highest in the presence of Mn 2ϩ , it is reasonable to believe that Ndx1 is activated by Mg 2ϩ in vivo since Mn 2ϩ is a minor component in the cell (below 0.5 M) (42). In the proposed hydrolysis mechanism of E. coli MutT (2) and L. angustifolius Ap 4 A pyrophosphatase (29), H 2 O bound to the Mg 2ϩ ion functions as the attacking nucleophile. Therefore, the degree of the hydration of Mg 2ϩ could directly influence the catalytic reaction. Because the pK a value for the metal ion hydrate is different from each metal, the kinetic data for other metal ions may help to verify this hypothesis.
To date, there has not been distinct evidence that a water molecule coordinating to Mg 2ϩ directly attacks the substrate. However, this may be supported by the observation that Nudix hydrolase activity is inhibited by F Ϫ ion (43). Ndx1 activity was also noncompetitively inhibited by F Ϫ . Monovalent anions such as fluoride have been used extensively to probe water or hy-droxide binding to a metal center in the active site of metalloproteases (44) and phosphatases (45). Therefore, noncompetitive inhibition by F Ϫ is consistent with the presence of a metalcentered water/hydroxide that can function as the reactive nucleophile in the hydrolysis reaction by Ndx1.
It has been reported that Ap 6 A is synthesized by acyl-CoA synthetase (46). Diadenosine polyphosphates have been found in E. coli (47), erythrocytes (48), platelets (49), and mammalian tears (50,51) and are supposed to function as signaling molecules (1). At present there is no evidence that diadenosine polyphosphates function as a signaling molecule in prokaryotes including T. thermophilus. However, some kinases are known to be inhibited by diadenosine polyphosphates including Ap 6 A (52, 53). Because diadenosine polyphosphates are very stable (these hardly hydrolyze by incubation at 100°C for 20 min (50,51) and can accumulate in the cell, Ndx1 may function as a housecleaning enzyme to prevent accumulation of those compounds. While considering whether such a molecule exists in T. thermophilus and whether it is meaningful that Ndx1 symmetrically hydrolyzes Ap 6 A, we are studying the physiological function of Ndx1 in relationship to Ap 6 A.
As mentioned earlier, proteins from T. thermophilus are stable against heat and extreme pH. This property makes these proteins suitable for structural and functional analysis. Actually, the high stability of Ndx1 enabled us to examine the dependence of the activity on pH. Now crystallization of Ndx1 is under way to determine the tertiary structure. We have already succeeded in crystallization of another Nudix protein (Ndx4) from T. thermophilus (54). Further details concerning the mechanisms of hydrolysis and substrate specificity of Ndx1 will be uncovered by structural analysis of Ndx1.