Interaction of Two Classes of ADP-ribose Transfer Reactions in Immune Signaling*

CD38 is a bifunctional ectoenzyme predominantly expressed on hematopoietic cells where its expression correlates with differentiation and proliferation. The two enzyme activities displayed by CD38 are an ADP-ribosyl cyclase and a cyclic adenosine diphosphate ribose (cADPR) hydrolase that catalyzes the synthesis and hydrolysis of cADPR. T lymphocytes can be induced to express CD38 when activated with antibodies against specific antigen receptors. If the activated T cells are then exposed with NAD, cell death by apoptosis occurs. During the exposure of activated T cells to NAD, the CD38 is modified by ecto-mono-ADP-ribosyltransferases (ecto-mono-ADPRTs) specific for cysteine and arginine residues. Arginine-ADP-ribosylation results in inactivation of both cyclase and hydrolase activities of CD38, whereas cysteine-ADP-ribosylation results only in the inhibition of the hydrolase activity. The arginine-ADP-ribosylation causes a decrease in intracellular cADPR and a subsequent decrease in Ca2+influx, resulting in apoptosis of the activated T cells. Our results suggest that the interaction of two classes of ecto-ADP-ribose transfer enzymes plays an important role in immune regulation by the selective induction of apoptosis in activated T cells and that cADPR mediated signaling is essential for the survival of activated T cells.

CD38 is a bifunctional ectoenzyme predominantly expressed on hematopoietic cells where its expression correlates with differentiation and proliferation. The two enzyme activities displayed by CD38 are an ADP-ribosyl cyclase and a cyclic adenosine diphosphate ribose (cADPR) hydrolase that catalyzes the synthesis and hydrolysis of cADPR. T lymphocytes can be induced to express CD38 when activated with antibodies against specific antigen receptors. If the activated T cells are then exposed with NAD, cell death by apoptosis occurs.

During the exposure of activated T cells to NAD, the CD38 is modified by ecto-mono-ADP-ribosyltransferases (ecto-mono-ADPRTs) specific for cysteine and arginine residues. Arginine-ADP-ribosylation results in inactivation of both cyclase and hydrolase activities of CD38, whereas cysteine-ADP-ribosylation results only in the inhibition of the hydrolase activity. The arginine-ADPribosylation causes a decrease in intracellular cADPR and a subsequent decrease in Ca 2؉ influx, resulting in apoptosis of the activated T cells. Our results suggest that the interaction of two classes of ecto-ADP-ribose transfer enzymes plays an important role in immune regulation by the selective induction of apoptosis in activated T cells and that cADPR mediated signaling is essential for the survival of activated T cells.
The type II transmembrane glycoprotein CD38 is the prototypic member of the class of adenosine diphosphate ribose (ADPR) 1 transfer enzymes known as nicotinamide adenine dinucleotide (NAD) glycohydrolases (NADases) (1)(2)(3). CD38 is a bifunctional ectoenzyme that catalyzes the formation and hydrolysis of cyclic adenosine diphosphate ribose (cADPR) (4 -6). Thus, this enzyme possesses both ADP-ribosyl cyclase and cADPR hydrolase, the net reaction of which is identical to that catalyzed by NADase.
cADPR is a potent Ca 2ϩ mobilizing agent believed to be involved in Ca 2ϩ -induced Ca 2ϩ release in a variety of cells from plants to humans (7)(8)(9). cADPR has also been shown to augment the proliferative response of activated murine B lymphocytes (5), as well as to mediate the Ca 2ϩ release associated with ATP-activated potassium currents in alveolar macrophages (10), suggesting that cADPR may function as a signaling second messenger in multiple hematopoietic cell types. Although the physiological ligand for CD38 has yet to be identified, CD38 was reported to undergo internalization through non-clathrincoated endocytic vesicles upon incubating cells with thiol compounds or NAD (11). The mechanism by which endocytic vesicles can generate cADPR was suggested by the demonstration that the CD38-catalyzed conversion of NAD to cADPR can be achieved by the influx of cytosolic NAD into the endocytic vesicles and the cADPR into the cytosol for subsequent Ca 2ϩ signaling (12).
Lymphocyte surface molecules are known to play an important regulatory role by modulation of their expression, binding activity, and/or signal transduction. The ecto-enzymes expressed on the lymphocyte surface also regulate the functions of lymphocytes. ADP-ribosylation of surface proteins on T cells by an ecto-mono-ADP-ribosyl transferase (ADPRT) results in the inhibition of cytotoxic T cell functions such as cell proliferation, cytotoxicity, and cytokine secretion (13), and the suppression of cytotoxic T cell function is closely correlated with the expression of the mono-ADPRT (14). Most ecto-NADases are known to be inhibited in the presence of their substrate, NAD (15,16). Recently, we described the NAD-dependent inactivation of 65-kDa NADase from rabbit erythrocytes by auto-ADP-ribosylation (17). In the study described here, we have explored whether CD38 is also inactivated in the presence of NAD. Our results show that ADP-ribosylation of CD38 by mono-ADPRT in activated T cells results in apoptosis as well as CD38 inactivation, suggesting that ADP-ribosylation of CD38 plays an important role in immune regulation and that cADPRmediated signaling is essential for the survival of activated T cells.

EXPERIMENTAL PROCEDURES
T and B Cell Preparation-A single cell suspension of splenocytes was prepared from normal Balb/c 6 -8-week-old mice as described previously (18). The T cells were purified by negative selection using goat anti-mouse IgG-coated Immulan beads (Biotecx Laboratories). The T cells were activated by incubation for 12 h with anti-CD3 antibody (1 g/ml; Chemicon) and interleukin-2 (100 units/ml; Perkin-Elmer Cetus). B cells were prepared by eluting cells which were bound to goat anti-mouse IgG-coated Immulan beads.
ADP-ribosylation-Activated T cells were washed with HBSS and incubated with 10 M [adenylate-32 P]NAD. The cells were lysed on ice in * This work was supported by the Korea Science and Engineering Foundation, the Ministry of Education (Genetic Engineering Program), and a grant from the Kye-Nam Memorial Fund (to U.-H. K.). 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.
Transfection-A full-length human CD38 cDNA was cloned by RT-PCR using oligonucleotide primers (5Ј-GGAAGCTTATGGCCAACTGC-GAG-3Ј, 5Ј-ACGTGTAGACTCTAGACTCGAGATCTCG-3Ј) (21). The total RNA was isolated from human lymphocytes and the first strand cDNA was synthesized using oligo(dT) [12][13][14][15][16][17][18] . The cDNA was amplified by PCR using the above primers, and the amplified 928-base pair DNA was cloned into pCR TM 3.1 vector (Invitrogen). The sequence of the cloned CD38 cDNA was confirmed. The plasmid DNA was purified and linearized by digestion with PvuI. CD38 Ϫ Jurkat cells were prepared by passing cells in anti-CD38-coated Immulan bead column. CD38 Ϫ Jurkat or HeLa cells (10 7 cells/ml) were transfected with the linearized DNA (10 g) by electroporation using an Electro square porator T820 (BTX) at 342 V and a pulse length of 99 s with 5 repeat pulses. Stable transfectants were selected for resistance to G418 (200 g/ml) and positive clones were repetitively selected with anti-CD38 antibodycoated Immulan beads.
Intracellular cADPR Measurement-HeLa cells were harvested by centrifugation at 1000 ϫ g for 5 min at 4°C. After washing with ice-cold PBS, the cells were immediately frozen in liquid nitrogen and stored at Ϫ80°C. The frozen cells were added to an 8% trichloroacetic acid solution to extract cADPR. The cADPR content was measured by a specific radioimmunoassay described elsewhere (23).
Intracellular Ca 2ϩ Measurement-HeLa cells were incubated with Fura-2 AM (4 M) in RPMI 1640 medium containing 3% fetal bovine serum for 60 min at 37°C. The Fura-2-loaded cells were then washed twice with HBSS. For the fluorometric measurement of Ca 2ϩ , 1 ϫ 10 7 cells were placed in a quartz cuvette in a thermostatically controlled cell holder at 37°C and the cell suspension was stirred continuously. Fluorescence ratios were taken with an alternative wavelength time scanning method (dual excitation at 340 and 380 nm; emission at 510 nm) using a PTI fluorometer. Calibration of the fluorescent signal in terms of Ca 2ϩ was performed as described previously (24).
Purification of ADPRTs-Splenocytes (10 9 cells) were incubated with phosphatidylinositol-specific phospholipase C (1 g/ml) purified from Bacillus cereus as described (25) in a total volume of 10 ml for 30 min at 37°C to solubilize the ADPRTs. For purification of arginine-specific ADPRT, the supernatant was dialyzed in 50 mM sodium phosphate buffer (pH 7.0), 1 mM benzamide, 0.1 mM phenylmethylsulfonyl fluoride, and 1 mM EGTA and applied to an aminobenzamide-agarose column (Sigma). The elution was performed with 50 mM sodium phosphate buffer (pH 7.0), 1 mM aminobenzamide, 0.1 mM phenylmethylsulfonyl fluoride, and 1 mM EGTA. The active fractions of arginine-ADPRT were directly applied to a concanavalin A-agarose column (3 ml bed volume) equilibrated with 50 mM Tris-HCl (pH 8.0), 1% Chaps, 0.2 M NaCl and washed with the same buffer. The bound proteins was eluted with 0.3 M methyl-␣-D-mannopyranoside. Activity fractions were concentrated and dialyzed in PBS. The arginine-specific ADPRT activity was measured as described (26). One unit of the enzyme was expressed as nanomole/ min. Specific activity was 6,300 units/mg and the yield of purified enzyme was 58 g. For purification of cysteine-specific ADPRT, the phosphatidylinositol-specific phospholipase C-treated supernatant from splenocytes (10 9 cells) was dialyzed in buffer A (50 mM potassium phosphate (pH 7.0), 30% ethylene glycol, 0.1 mM phenylmethylsulfonyl fluoride, 0.1 mM EDTA, 10 mM benzamide). The sample was applied slowly to cysteine-Sepharose column (bed volume 10 ml) at a flow rate of 0.2 ml/min. Unbound proteins were washed off the column with buffer A and bound proteins were eluted with buffer A containing 0.1 M NaCl and 0.1% 2-mercaptoethanol. Cysteine-specific ADPRT activity fractions were collected and concentrated with Amicon 10. After adding 1.7 M ammonium sulfate, the concentrate was applied to a phenyl-Sepharose column (bed volume 10 ml) at a flow rate of 1 ml/min. The separation was performed on a 1.7 to 0 M ammonium sulfate gradient at 4°C using a fast protein liquid chromatography system (Waters). The active fractions of cysteine-ADPRT were directly applied to a concanavalin A-agarose column (3 ml bed volume) equilibrated with 50 mM Tris-HCl (pH 8.0), 1% Chaps, 0.2 M NaCl and washed with the same buffer. The bound proteins was eluted with 0.3 M methyl-␣-D-mannopyranoside. Activity fractions were concentrated and dialyzed in PBS. The concentrate was applied to a high performance liquid chromatography gel filtration column and eluted with PBS. Cysteine-specific ADPRT activity was measured as described (27). One unit of the enzyme was expressed as nanomole/min. Specific activity was 2,500 units/mg and the yield of purified enzyme was 2 g.

ADP-ribosylation of CD38 Results in Inactivation of Its ADPribosyl Cyclase and cADPR Hydrolase Activities-Most ecto-
NADases are known to be inactivated in the presence of NAD (15,16). We tested whether T lymphocyte CD38 is also inactivated in the presence of NAD. We induced the expression of CD38 on T cells from Balb/c mice by activation with anti-CD3 antibodies, since CD38 on murine T cells prepared from splenocytes was not induced by other stimuli, such as concanavalin A, phytohemagglutinin, or phorbol 12-myristate 13-acetate and inonomycin (Fig. 1). When the activated T cells were incubated with various concentrations of NAD for 1 h, the ADP-ribosyl cyclase and cADPR hydrolase activities of CD38 decreased in a dose-dependent manner with increasing concentrations of NAD ( Fig. 2A). CD38 seemed to be the only enzyme responsible for the activities of both ADP-ribosyl cyclase and cADPR hydrolase, since most of these enzyme activities in the activated T cells were recovered by immunoprecipitation with anti-CD38 antibodies and inactivated by the addition of NAD as in the intact T cells (Fig. 2A, inset).
This result indicates that both of the enzyme activities of CD38 are affected by the addition of NAD. To study the mechanisms responsible for the NAD-dependent inhibition of CD38 enzyme activities, we examined the transfer of ADPR from NAD to the CD38 molecule. We incubated the activated T cells with [adenylate-32 P]NAD and immunoprecipitated with anti-CD38 antibodies. We found that the precipitate contained 32 P radioactivity (Fig. 2B, control lane). These results suggest that ADPR from NAD was transferred to CD38 and that the modification of CD38 is NAD-dependent.
Next, we tested whether CD38 was inactivated by mono-ADPRT-mediated ADP-ribosylation or by automodification by CD38. As it has been reported that mono-ADPRT is anchored on the cell membrane via a glycosylphosphatidylinositol-linkage (14, 28), we examined whether the treatment of T cells with PI-PLC would block the ADP-ribosylation of CD38. Pretreatment of T cells with phosphatidylinositol-specific phospho-lipase C prevented NAD-dependent ADP-ribosylation of CD38 (Fig. 2B), but did not affect the amount of membrane-associated CD38 and the activities of CD38. Phosphatidylinositolspecific phospholipase C treatment might not damage the T cells because there was no significant ATP release from the cells and no changes of the cell viability. This result indicates that the ADP-ribosylation of CD38 was not due to automodification but was mediated by a glycosylphosphatidylinositol-anchored cell surface mono-ADPRT.
Mammalian cells contain at least two types of mono-ADPRTs that are either arginine-or cysteine-specific (29); thus, we determined the ADPR acceptor site(s) on CD38. We treated [ 32 P]ADP-ribosylated CD38 with water (control), hydroxylamine (which specifically cleaves arginine-linked ADPR), or mercuric chloride (which specifically cleaves cysteine-linked ADPR) and found that the [ 32 P]ADPR on CD38 was removed completely only by treatment with a both hydroxylamine and mercuric chloride. Treatment with either hydroxylamine or mercuric chloride alone led to only partial release (Fig. 2C). This suggests that both arginine-and cysteine-specific mono-ADPRTs are involved in the ADP-ribosylation of CD38.
Apoptosis of Activated T Cells and CD38-transfected Cells by ADP-ribosylation of CD38 -When the activated T cells were incubated with NAD, the percentage of viable T cells was

FIG. 1. Expression of CD38 in T cells by various stimuli.
Cells (1 ϫ 10 5 cells/ ml) were treated with concanavalin A (2.5 g/ml), phytohemagglutinin (1.0 g/ml), phorbol 12-myristate 13-acetate (10 ng/ ml), and ionomycin (0.5 M), or anti-CD3⑀ monoclonal antibodies (1 g/ml). After 24 h, the cells were collected and incubated anti-CD38 monoclonal antibodies (50 g/ml) for 30 min in ice. After washing with phosphate-buffered saline, the cells were stained with fluorescein isothiocyanate-labeled anti-rat IgG and subjected to a flow cytometric analysis. significantly decreased as compared with control T cells that were incubated under the same conditions except that NAD was omitted (Fig. 3A). To determine whether the reduction of viable T cells was a result of apoptosis, we examined annexin V binding, a method for the early detection of apoptosis (30), in activated T cells incubated in the presence or absence of NAD. The percentage of apoptotic cells was significantly increased in the activated T cells by the treatment with NAD (Fig. 3B).
To determine if ADP-ribosylation of CD38 plays an important role in inducing apoptosis in T cells, we examined whether ADP-ribosylation of CD38 could induce apoptosis in two cells lines. First CD38-transfected Jurkat cells, a cloned human T cell leukemia cell line that does not normally express CD38 but has both arginine-and cysteine-specific mono-ADPRTs, was examined. When the CD38-transfected and non-transfected Jurkat cells were exposed to NAD, we found that only the CD38-transfected Jurkat cells underwent apoptosis (Fig. 3C). Next, HeLa, a human epithelial cell line that does not normally express either CD38 or mono-ADPRT activity was examined. The CD38-transfected cells also could be induced to undergo apoptosis in the presence of NAD, but only when supplemented with partially purified mono-ADPRTs from murine splenocytes (Fig. 3C). These results strongly suggest that CD38 is indeed a target molecule for ADP-ribosylation by mono-ADPRTs leading to programmed cell death and that the ADP-ribosylation of CD38 can induce apoptosis in CD38-expressing cells, regardless of the cell type.
The results of NAD-dependent apoptosis of activated T cells and CD38-transfected cells still raise the argument that NAD metabolites might affect lymphocytes. To exclude the possibility, we tested the effects of the metabolites such as adenosine, AMP, ADP, ADP-ribose, and cADPR on the apoptosis of activated T cells, but no significant changes were found in these cells (data not shown).

Arginine-specific ADPRT Is Responsible for Apoptosis of Activated T Cells-Since CD38 is expressed on both T and B
lymphocytes, we determined whether B cells also undergo apoptosis through ADP-ribosylation of CD38. We incubated B cells with NAD and apoptosis was tested by annexin V binding. We found that B cells showed no detectable apoptosis (Fig. 4A), although CD38 was ADP-ribosylated. To determine whether there are differences between the ADP-ribosylation of CD38 in T cells and B cells, we examined the ADP-ribosylation sites on the CD38 of B cells and found that the ADPR on CD38 in the B cells was completely removed by treatment with mercuric chloride alone and unaffected by treatment with hydroxylamine (data not shown), indicating that the ADP-ribosylation of CD38 in B cells occurs only on the cysteine residue. We further examined the expression of arginine-specific mono-ADPRT activity in B cells and T cells by RT-PCR and Southern blot. We found that B cells did not express arginine-specific mono-AD-PRT, whereas T cells did express this enzyme (Fig. 4B).
We then asked whether there are differences between arginine-and cysteine-ADP-ribosylation on the enzyme activity of CD38. We immunoprecipitated CD38 from B cells with anti-CD38 antibodies, incubated the immune complex with 100 M C, NAD-dependent apoptosis of CD38transfected cells. Wild-type (Mock) and CD38-transfected Jurkat or HeLa cells were incubated with 100 M NAD in RPMI medium containing 10% fetal bovine serum for 16 h, following serum deprivation for 24 h. In the case of HeLa cells, 60 units/ml arginine-ADPRT purified from splenocytes were supplemented for ADP-ribosylation. The cells were centrifuged onto slides, fixed with 1% paraformaldehyde, and stained with 8 g/ml Hoescht 33342 to identify cells that lost nuclear structure.
NAD and either arginine-or cysteine-specific mono-ADPRTs purified from T cells, and examined the ADP-ribosyl cyclase and cADPR hydrolase activities. We found that arginine-ADPribosylation of CD38 resulted in the inactivation of both the cyclase and hydrolase activities of CD38, while cysteine-ADPribosylation resulted in the inactivation of hydrolase activity but did not affect the cyclase activity (Fig. 4C). Therefore, we asked whether arginine-specific ADRPT plays an essential role in the induction of apoptosis. When B cells were incubated with NAD and arginine-specific mono-ADPRT purified from splenocytes, the cells underwent apoptosis similar to T cells after incubation with NAD ( Fig. 4A and 3B). This result suggests that the ADP-ribosylation of an arginine residue(s) on CD38 plays a critical role in the induction of apoptosis, and that this may be due to reduced cyclase activity, which presumably results in a reduced intracellular cADPR concentration ([cADPR] i ). Thus, we examined the effect of CD38 ADP-ribosylation on [cADPR] i . As predicted, prior incubation of T cells with NAD significantly reduced [cADPR] i , compared with cells that were not treated with NAD (Fig. 4D). Taken together, these results suggest that the inhibition of the cyclase activity of CD38 through ADP-ribosylation by arginine-mono-ADPRT blocks intracellular cADPR production, which subsequently results in decreases in intracellular free Ca 2ϩ in activated T cells and culminating in apoptosis.
We then asked whether R269G mutant-transfected cells are resistant to NAD-induced apoptosis because this CD38 mutant would not be ADP-ribosylated upon addition of NAD. Indeed, R269G mutant-transfected HeLa cells did not undergo apoptosis by NAD treatment. With regard to Arg mutant that retains WT levels of ADP-ribosyl cyclase but cannot serve as a substrate for the Arg-ADPRT, we generated a R269K mutant and its transfectant cell line. These cells had 70% ADP-ribosyl cyclase activity of the control WT, and the cells' CD38 could not serve as a substrate for the Arg-ADPRT, therefore these cells did not undergo any apoptosis by NAD treatment (data not shown). This result also indicates that ADP-ribosylation of CD38 is an important signal that can lead to apoptosis.
The ligation of CD38 by agonistic CD38 antibodies has earlier been shown to elicit cellular responses, including elevation in cytoplasmic Ca 2ϩ due to Ca 2ϩ influx (1,31). Also, Guse et al. (32) recently demonstrated sustained increases in [cADPR] i and Ca 2ϩ entry by the stimulation of TCR⅐CD3 complex in T lymphocytes. The mechanism by which cADPR regulated Ca 2ϩ entry was suggested to be similar to that of inositol 1,4,5trisphosphate in the capacitative Ca 2ϩ model, because cADPR mediated depletion of the Ca 2ϩ pool by releasing Ca 2ϩ from its target storage (32). Therefore, in order to examine the effect of ADP-ribosylation on CD38 ligation-induced signaling in control WT-and R269K mutant-transfected cells, we measured the [cADPR] i and intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ), following CD38 ligation in cells pretreated with or without NAD. A prior incubation of WT-transfected HeLa cells with NAD completely abolished CD38 ligation-induced increase of cADPR production (Fig. 5B) and [Ca 2ϩ ] i (Fig. 5C)  increase in [Ca 2ϩ ] i by CD38 ligation was most likely due to Ca 2ϩ influx, since the increase was completely abrogated by EGTA, a Ca 2ϩ chelator (data not shown). In contrast, prior treatment of mutant-transfected cells with NAD could not counteract the CD38 ligation-induced increases in cADPR production (Fig. 5B) and Ca 2ϩ influx (Fig. 5C). These findings were also observed in the WT-and R269K mutant-transfected Jurkat T lymphocytes (data not shown). Taken together, this result indicates that ADP-ribosylation of CD38 blocks intracellular cADPR production, which subsequently results in decreases in intracellular free Ca 2ϩ and leads to apoptosis.
The results described in the present study provide evidence for the interaction of two classes of ADP-ribose transfer enzymes in T cell signaling pathways that can lead to apoptosis.
The first requirement for this pathway is likely CD38 expression (possessing cyclase activity) and the second requirement is ADP-ribosylation of CD38 (inactivation of cyclase). The present studies using mutant enzymes can be extrapolated to activated T cells by assuming that CD38 expression (cyclase activity) can accelerate cell cycling necessary for proliferation (12). In activated T cells, CD38 may be essential for the proliferation via Ca 2ϩ influx by the action of cADPR which is generated by the ADP-ribosyl cyclase activity of CD38. However, ADP-ribosylation of CD38 results not only in cessation of cell proliferation but also in apoptosis of the cells. Thus, the mechanism by which apoptosis is induced by ADP-ribosylation of CD38 may be related to the cell cycle, because actively proliferating cells are sensitive to the apoptotic insults such as stoppage of Ca 2ϩ supply.
For our ex vivo observations to have direct physiologic relevance, NAD, an intracellular metabolite, should be available extracellularly to support the ADP-ribosylation of CD38. Extracellular NAD concentrations have been reported to be submicromolar in the plasma of mammals including mice (33), whereas micromolar NAD concentrations are likely to be needed for effective ADP-ribosylation of CD38 in the extracellular compartment ( Fig. 2A). However, a recent finding by Zocchi et al. (12) that NAD is steadily released from fibroblast and epithelial cells supports the physiologic relevance of our observations as they suggest that micromolar NAD concentrations, especially in the vicinity of the outer surface, can be reached. Furthermore, conditions, such as massive cell lysis during inflammatory immune reactions may also lead to an increase in the concentration of NAD in the plasma, consequently enhancing the ADP-ribosylation of CD38.