Lysine 129 of CD38 (ADP-ribosyl cyclase/cyclic ADP-ribose hydrolase) participates in the binding of ATP to inhibit the cyclic ADP-ribose hydrolase.

CD38 catalyzes not only the formation of cyclic ADP-ribose (cADPR) from NAD+ but also the hydrolysis of cADPR to ADP-ribose (ADPR), and ATP inhibits the hydrolysis (Takasawa, S., Tohgo, A., Noguchi, N., Koguma, T., Nata, K., Sugimoto, T., Yonekura, H., and Okamoto, H. (1993) J. Biol. Chem. 268, 26052-26054). In the present study, using purified recombinant CD38, we showed that the cADPR hydrolase activity of CD38 was inhibited by ATP in a competitive manner with cADPR. To identify the binding site for ATP and/or cADPR, we labeled the purified CD38 with FSBA. Sequence analysis of the lysylendopeptidase-digested fragment of the labeled CD38 indicated that the FSBA-labeled residue was Lys-129. We introduced site-directed mutations to change the Lys-129 of CD38 to Ala and to Arg. Neither mutant was labeled with FSBA nor catalyzed the hydrolysis of cADPR to ADPR. Furthermore, the mutants did not bind cADPR, whereas they still used NAD+ as a substrate to form cADPR and ADPR. These results indicate that Lys-129 of CD38 participates in cADPR binding and that ATP competes with cADPR for the binding site, resulting in the inhibition of the cADPR hydrolase activity of CD38.

quences of Aplysia ADP-ribosyl cyclases (14,15) showed a high degree of homology with that of CD38 (16,17), which was reported to be a surface antigen of human lymphocytes (18). From the experiments in which CD38 cDNA was expressed in mammalian cells, CD38 was shown to catalyze not only the formation of cADPR from NAD ϩ but also the hydrolysis of cADPR to ADP-ribose (ADPR) (19 -21). We have demonstrated that ATP inhibited the hydrolysis, resulting in the accumulation of cADPR (19). Furthermore, we produced transgenic mice overexpressing human CD38 in pancreatic ␤ cells and demonstrated that ATP, produced in the process of glucose metabolism, increased the accumulation of cADPR to enhance the Ca 2ϩ mobilization from the intracellular stores for insulin secretion in the transgenic islets (22).
In the present study, we expressed human CD38 in Escherichia coli and purified it to homogeneity. Using the purified CD38, we found that Lys-129 of CD38 participated in cADPR binding and that ATP competed with cADPR for the binding site, resulting in the inhibition of the cADPR hydrolase activity of CD38.

EXPERIMENTAL PROCEDURES
Purification of Soluble CD38 -We isolated a CD38 cDNA from an insulinoma of a Japanese patient by reverse transcriptase-polymerase chain reaction (PCR) (19) and used it for functional analyses of CD38 (17,19,22). The cDNA sequence was exactly the same as the CD38 sequence reported by Jackson and Bell (18) except for two base substitutions (nucleotide 213, A 3 C, and nucleotide 215, C 3 A) (19). The substitutions were also found in the corresponding region of the genome isolated from a Japanese patient (23). The CD38 cDNA (19) encoding amino acids 45-300 (17,23) was amplified by PCR using two primers, one sense (5Ј-cctcccgggAGGTGGCGCCAGACGTGGAG-3Ј, nucleotide positions 202-221) and one antisense (5Ј-gctctagaGCTCAGATCTCA-GATGTGCA-3Ј, nucleotide positions 955-974). Nonspecific nucleotides are shown in lowercase letters, and SmaI and XbaI restriction sites are underlined, respectively. After digestion by SmaI and XbaI, the PCR product was inserted into the unique XmnI-XbaI site of the pMAL-p2 vector (New England Biolabs), downstream and in-frame with the maltose-binding protein (MBP) coding sequence. The accuracy of the PCRderived recombinant CD38 construct was confirmed by sequencing. After induction of the fusion protein with isopropyl-1-thio-␤-D-galactopyranoside in E. coli, the protein was purified by affinity chromatography on cross-linked amylose resin (New England Biolabs). To separate the CD38 from MBP, the fusion protein was cleaved by Factor Xa (New England Biolabs), and the CD38 was purified by gel filtration (Sephacryl S-100HR, Pharmacia Biotech, Uppsala, Sweden). To reconstitute the enzymic activity, the purified CD38 was refolded as described (24).
Site-directed Mutagenesis-The site-directed mutants were made according to the procedure described by Reikofski and Tao (25) using an LA PCR In Vitro Mutagenesis Kit (Takara Shuzo Co. Ltd., Otsu, Japan). The synthetic oligonucleotides used for site-directed mutagenesis were as follows: 5Ј-AGCAGAATAGCAGATCTGGCC-3Ј (K129A) and 5Ј-AGCAGAATAAGAGATCTGGCC-3Ј (K129R), which correspond to the region from nucleotides 445-465 of the CD38 cDNA (19,23) and where underlined nucleotides were altered.
Enzyme Assays-ADP-ribosyl cyclase and cADPR hydrolase assays were performed as described (17,19,26) with CD38. Briefly, 50 ng of CD38 was incubated for 5 min at 37°C in 0.1 ml of phosphate-buffered saline (pH 7.4) (26, 27) with 0.2 mM NAD ϩ containing 5 Ci of [ 32 P]NAD ϩ (DuPont NEN) for ADP-ribosyl cyclase or with 0.2 mM cADPR containing 5 Ci of [ 32 P]cADPR, prepared as described previously (3,17,19,26) for cADPR hydrolase. Reaction products were analyzed by HPLC (17,19,26) using a flow scintillation analyzer (Flow-One Beta-525TR, Packard, Meriden, CT). The protein concentration was measured by the method of Bradford (28) using bovine serum albumin as a standard. * This work was supported in part by grants-in-aid for scientific research from the Ministry of Education, Science, Sports and Culture of Japan and the Kanae Foundation of Research for New Medicine. 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 U.S.C. Section 1734 solely to indicate this fact.
‡ Recipient of a fellowship from the Japan Society for the Promotion of Science.
Detection of FSBA Modification-The amount of covalent modification of CD38 with FSBA was estimated by Western blot as described previously (5,17,19,26) using an ECL detection system (Amersham Corp.). A polyclonal antibody against FSBA (Boehringer Mannheim) was used at a final concentration of 1 g/ml diluted with 5% milk powder solution as a primary antibody. The band intensity of the FSBA-labeled CD38 was measured using NIH Image software.
Identification of FSBA-labeled Peptide-0.2 mg of CD38 was labeled with 1 mM FSBA for 4 h and subjected to proteolysis by lysylendopeptidase (Wako Pure Chemical Industries, Osaka, Japan) as described previously (31). The lysylendopeptidase fragments were separated by reverse-phase HPLC using a RP C2/C18 column (3.2 ϫ 30 mm, Pharmacia Biotech) at a flow rate of 200 l/min. An aliquot (10 l) of each peak was blotted on a polyvinylidene difluoride membrane and analyzed for the presence of FSBA-labeled peptide by Western blot using an anti-FSBA antibody as described above. The peak containing the FSBAlabeled peptide was subjected to automated Edman degradation as described previously (31).
cADPR Binding Assay-1.5, 3, and 6 nmol of wild type CD38, K129A-CD38, K129R-CD38, or bovine serum albumin were incubated for 10 -30 min at 4°C in 0.1 ml of phosphate-buffered saline (pH 7.4) with 10 nmol of cADPR containing 5 Ci of [ 32 P]cADPR. During the incubation, neither CD38 nor its mutants converted cADPR to ADPR. The incubation mixture was fractionated at 4°C by a gel filtration column (PD-10, Pharmacia Biotech). 0.5-ml fractions were collected, and an aliquot (10 l) of each fraction was analyzed for radioactivity. Another aliquot (50 l) of each fraction was blotted on a polyvinylidene difluoride membrane and analyzed for the presence of CD38 by Western blot using a monoclonal antibody against human CD38 (T16, Cosmo Bio Co., Ltd., Tokyo, Japan) (17, 19, 26) as described above.

RESULTS AND DISCUSSION
Human CD38 was expressed in E. coli and purified as a soluble protein without the hydrophobic membrane domain (see "Experimental Procedures"). When NAD ϩ was used as a substrate, the purified CD38 catalyzed the formation of cADPR and ADPR (Fig. 1A). On the other hand, when cADPR was used as a substrate, the CD38 exhibited the cADPR hydrolase activity to form ADPR (Fig. 1B) as described previously (17,19). We then examined the effect of ATP on the hydrolysis of cADPR by CD38. ATP inhibited the hydrolysis in a dose-dependent manner (data not shown), which was consistent with the previous result obtained by using the COS-7 cell membrane fraction, in which CD38 was expressed, as an enzyme source (19). Kinetic analysis was performed with various concentrations of cADPR in the presence or absence of ATP, and the data were plotted in a double reciprocal manner. As shown in Fig. 2, in the absence of ATP, CD38 had a V max value of 470 Ϯ 38 nmol/min/mg of protein (n ϭ 3) and a K m value of 53 Ϯ 7.0 M (n ϭ 3) for cADPR. ATP inhibited the hydrolysis of cADPR in a competitive manner, with a K i value of 4.8 Ϯ 0.53 mM (n ϭ 3). Competitive inhibition of the cADPR hydrolysis by ATP suggests that ATP and cADPR bind to the same site of CD38.
To identify the binding site for ATP and/or cADPR, we then labeled the purified CD38 with FSBA (29,30). As shown in Fig.  3A, the CD38 was labeled in a time-dependent manner, and cADPR and ATP completely inhibited the incorporation of FSBA into CD38, indicating that the binding site for ATP and/or cADPR was affected by FSBA. We next subjected the  labeled CD38 to lysylendopeptidase digestion, and the resultant peptide fragments were separated by reverse-phase HPLC (Fig. 3B). Aliquots of all peaks were withdrawn, and the presence of FSBA-labeled peptide was analyzed by Western blot. Only peak 22 was immunoreactive (Fig. 3B, inset). Edman degradation of peak 22 gave a 7-amino acid sequence ILLWSRI with yields of 0.9 -143.4 pmol/cycle, and the phenylthiohydantoin-derivative was not identified for the 8th cycle. The 7-amino acid sequence corresponded to the sequence ILLWSRI in the theoretical lysylendopeptidase fragment (amino acid residues 122-129) in the CD38. Because it is known that FSBA can act as an electrophilic agent in covalent reactions with several classes of amino acids, including serine, tyrosine, lysine, histidine, and cysteine (29), Lys-129 is therefore implicated as the FSBA-labeled residue that escaped identification by Edman degradation.
We introduced site-directed mutation into the CD38 cDNA by replacing the lysine 129 codon with that of alanine or arginine. The resulting cDNAs were then introduced into E. coli and mutant CD38s (K129A-and K129R-CD38) were produced and purified. As shown in Fig. 4A, when the mutant CD38s were incubated with FSBA, the mutant proteins were not labeled with FSBA, indicating that Lys-129 of CD38 is actually the site of FSBA labeling. We next analyzed the enzymic activities of the mutant CD38s. As shown in Table I, the mutants did not catalyze the hydrolysis of cADPR to ADPR, whereas their catalytic activity to form cADPR and ADPR from NAD ϩ still remained. In addition, the FSBA-labeled CD38 catalyzed the formation of cADPR and ADPR from NAD ϩ but not the hydrolysis of cADPR to form ADPR (Fig. 1). We next tested the mutant CD38s for their ability to bind to cADPR. As shown in Fig. 4B, cADPR was eluted in a gel filtration chromatography around fraction 16. After incubation with wild type CD38 at 4°C, a new elution peak appeared around fraction 3 (Fig. 4B,  Wild), in which most of the immunoreactivity against CD38 was recovered (inset), indicating that the elution peak represented the CD38-cADPR complex. On the other hand, no peak around fraction 3 appeared in chromatograms using mutant CD38s (Fig. 4B, K129A and K129R), whereas the CD38 immunoreactivity was recovered around fraction 3 (inset). These results indicate that the wild CD38 did bind to cADPR but the mutant CD38s did not, and that Lys-129 participates in the cADPR/ATP binding.
On the other hand, as shown in Fig. 5, the NAD ϩ binding site of CD38 appears to be different from Lys-129 because both the site-directed mutant and FSBA-modified CD38s used NAD ϩ as a substrate to form cADPR and ADPR ( Fig. 1 and Table I). Furthermore, the formation of cADPR and ADPR from NAD ϩ by CD38 was inhibited by nicotinamide in a competitive manner with a K i value of 290 Ϯ 32 M (n ϭ 3), whereas ADPR (up to 5 mM) did not inhibit the formation of cADPR and ADPR from NAD ϩ . These results indicate that the binding of NAD ϩ to CD38 is achieved by the nicotinamide moiety rather than by the adenosine moiety. This is consistent with the previous results that CD38 catalyzed the formation of cyclic GDP-ribose from NGD ϩ (nicotinamide guanine dinucleotide) but used hardly any cyclic GDP-ribose as a substrate to form GDP-ribose (32)(33)(34).
We have proposed a model for insulin secretion by glucose via cADPR-mediated Ca 2ϩ mobilization (35). In the process of CD38 catalyzes the formation of cADPR from NAD ϩ and the hydrolysis of cADPR to ADPR. Lys-129 of CD38 is the cADPR binding site and ATP competes with cADPR for the binding site, resulting in the inhibition of the hydrolysis of cADPR. The NAD ϩ binding site appears to be different from Lys-129 as discussed in the text. [Enzyme-ADPR*] is suggested to be an enzyme-stabilized ADP-ribosyl oxocarbonium ion intermediate (35,37,38). The intermediate is thought to be attacked by H 2 O to form ADPR (33,35,39). Although at present there is no direct evidence that column. An aliquot (10 l) of all fractions was withdrawn for determination of 32 P label by liquid scintillation counting. The incubation time (from 10 -30 min) did not affect the cADPR binding to the wild type CD38, and the ratio of cADPR bound per mol of the CD38 was 0.33. On the other hand, the mutants (1.5, 3, and 6 nmol) did not show any cADPR binding. Another aliquot (50 l) of all fractions was analyzed by immunoblotting using an anti-CD38 antibody. The result of the immunoblot is shown in the each inset. An arrow indicates the position of the CD38-cADPR complex.

TABLE I
Enzymic activities of CD38 and its mutants 50 ng of wild type CD38 and 250 ng of K129A-CD38 or K129R-CD38 was incubated at 37°C in 100 l of total reaction mixture with 0.2 mM NAD ϩ containing 5 Ci of [ 32 P]NAD ϩ or 0.2 mM cADPR containing 5 Ci of [ 32 P]cADPR as described under "Experimental Procedures." Reaction products were analyzed by anion exchange HPLC using a flow scintillation analyzer (17,19,26 glucose metabolism, millimolar concentrations of ATP are generated (22), inducing cADPR accumulation by inhibiting the cADPR hydrolase activity of CD38, and cADPR acts as a second messenger for intracellular Ca 2ϩ mobilization from the endoplasmic reticulum for insulin secretion (3,5,19,22,35,36). In the present study, it has been shown that Lys-129 of CD38 participates in cADPR binding and that millimolar concentrations of ATP compete with the cADPR binding site, inhibiting the cADPR hydrolase activity of CD38. Therefore, when mutations at or around Lys-129 occur in CD38, cADPR metabolism cannot be regulated by ATP, which is generated in the process of glucose metabolism. Such a mutation in CD38 could be a predisposing factor for diabetes mellitus.