Ligand-induced internalization of CD38 results inintracellular Ca21 mobilization: role of NAD1

transport across cell membranesELENA ZOCCHI,* CESARE USAI,† LUCREZIA GUIDA,* LUISA FRANCO,*SANTINA BRUZZONE,* MARIO PASSALACQUA,* AND ANTONIO DE FLORA*,1

*Institute of Biochemistry, University of Genova, Viale Benedetto XV/1, 16132, Genova; and†Institute of Cybernetics and Biophysics, National Research Council, Via De Marini 6, 16149, Genova,Italy

ABSTRACT CD38, a transmembrane glycoproteinwidely expressed in vertebrate cells, is a bifunctionalectoenzyme catalyzing the synthesis and hydrolysis ofcyclic ADP-ribose (cADPR). cADPR is a universalsecond messenger that releases calcium from intra-cellular stores. Since cADPR is generated by CD38 atthe outer surface of many cells, where it acts intra-cellularly, increasing attention is paid to addressingthis topological paradox. Recently, we demonstratedthat CD38 is a catalytically active, unidirectionaltransmembrane transporter of cADPR, which thenreaches its receptor-operated intracellular calciumstores. Moreover, CD38 was reported to undergo aselective and extensive internalization through nonclathrin-coated endocytotic vesicles upon incubatingCD381 cells with either NAD1 or thiol compounds:these endocytotic vesicles can convert cytosolic NADinto cADPR despite an asymmetric unfavorable ori-entation that makes the active site of CD38 intrave-sicular. Here we demonstrate that the cADPR-gener-ating activity of the endocytotic vesicles results inremarkable and sustained increases of intracellularfree calcium concentration in different cells exposedto either NAD1, or GSH, or N-acetylcysteine. Thiseffect of CD38-internalizing ligands on intracellularcalcium levels was found to involve a two-step mech-anism: 1) influx of cytosolic NAD1 into the endocy-totic vesicles, mediated by a hitherto unrecognizeddinucleotide transport system that is saturable, bidi-rectional, inhibitable by 8-N3-NAD1, and character-ized by poor dinucleotide specificity, low affinity,and high efficiency; 2) intravesicular CD38-catalyzedconversion of NAD1 to cADPR, followed by out-pumping of the cyclic nucleotide into the cytosol andsubsequent release of calcium from thapsigargin-sensitive stores. This unknown intracellular traffick-ing of NAD1 and cADPR based on two distinctiveand specific transmembrane carriers for either nu-cleotide can affect the intracellular calcium ho-meostasis in CD381 cells.—Zocchi, E., Usai, C.,Guida, L., Franco, L., Bruzzone, S., Passalacqua, M.,De Flora, A. Ligand-induced internalization of CD38

results in intracellular Ca21 mobilization: role ofNAD1 transport across cell membranes. FASEB J. 13,273–283 (1999)

Key Words: pyridine nucleotides z cyclic ADP-ribose z intracel-lular calcium homeostasis z NAD1 transporter

CD38 is a bifunctional ectoenzyme present in awide number of different mammalian cells (includ-ing muscle, neuronal, and endocrine tissues) butpredominantly expressed on hemopoietic cellswhere its expression correlates with differentiationand proliferation processes (1–6). The two enzymeactivities displayed by CD38 are an ADP-ribosyl cy-clase and a cyclic ADP-ribose (cADPR)2 hydrolasethat catalyze the synthesis from NAD1 and thehydrolysis of cADPR, respectively (7–11). Underspecific conditions, i.e., at acidic pH values, CD38can also catalyze the synthesis of nicotinic acidadenine dinucleotide (NAADP1) via a base ex-change reaction between NADP1 and nicotinic acid(12). Both cADPR and NAADP1 are powerful Ca21

mobilizers in different cell types (12–14). Specifi-cally, cADPR has recently emerged as a novel secondmessenger for many agonists from plant (15) to

1 Correspondence: Institute of Biochemistry, University ofGenova, Viale Benedetto XV/1, 16132, Genova, Italy. E-mail:toninodf@unige.it

2Abbreviations: 3T3 cells, NIH 3T3 cells; cADPR, cyclicadenosine diphosphate ribose; [Ca21]i, cytosolic calciumconcentration; mAb, monoclonal antibody; NAADP1, nico-tinic acid adenine dinucleotide phosphate; GSH, glutathione(reduced form); NAC, N-acetylcysteine; 8-N3-NAD1, 8-azido-NAD1; 8-N3-AMP, 8-azido-AMP; PBL, peripheral blood lym-phocytes; DME, Dulbecco’s modified Eagle’s medium; FCS,fetal calf serum; Ig, immunoglobulin; NMN, nicotinamidemononucleotide; NGD1, nicotinamide guanine dinucleotide;NHD1, nicotinamide hypoxanthine dinucleotide; G6PD, glu-cose 6-P dehydrogenase; HPLC, high-pressure liquid chroma-tography; TCA, trichloroacetic acid; ADPR, adenosinediphosphate ribose; HLA, histocompatibility leukocyte anti-gen; FAD, flavine adenine dinucleotide.

2730892-6638/99/0013-0273/$02.25 © FASEB

human cells, where it can modulate the intracellularcalcium levels by releasing it from ryanodine-sensi-tive intracellular stores (5, 13, 16, 17). The well-established role of Ca21 in the control of suchfundamental processes of cell physiology as secre-tion, contraction, differentiation, proliferation, andapoptosis (18) suggests a possible involvement of theCD38/cADPR system in the regulation of theseevents. However, the topological problem of theectocellular production of an intracellular Ca21 mo-bilizer defies an obvious interpretation of the func-tion of the CD38/cADPR system in mammalian cells(5, 19, 20).

We recently described two different processes bywhich this topological paradox can be overcome: 1)the ligand (GSH or NAD1) -induced, vesicle-medi-ated internalization of CD38 in B-lymphocytic, hu-man Namalwa cells, which is followed by an increaseof intracellular cyclase activity and of intracellularcADPR concentration ([cADPR]i) (21); and 2) thecADPR-transporting function of membrane-boundCD38, which behaves as a catalytically active trans-porter responsible for generation and selective in-flux of cADPR across membranes (22). Both mech-anisms could in fact cooperate to yield the sameresult: import of cADPR metabolism from the cellsurface into the cytosol of CD381 cells. Though it isstill unknown whether the increased [cADPR]i pro-duced by CD38 internalization also induces an in-creased cytosolic calcium concentration ([Ca21]i),an unequivocal correlation between the two facts hasbeen demonstrated during the opposite process ofexocytotic transport of CD38 to the plasmamem-brane (23). In this case, intracellularly producedcADPR was demonstrated to be cytosolic by theincrease of basal [Ca21]i it elicits in CD381-trans-fected cells as compared to antisense-transfectedcontrols.

These data prompted us to address the ‘subcellu-lar’ topological inconsistency of the CD38/cADPRsystem: the endocytotic vesicles mediating ligand-induced CD38 internalization derive from invagina-tions of the plasmamembrane and therefore theactive site of CD38 is intravesicular, i.e., apparentlynot accessible to cytosolic NAD1. The same asym-metric organization shields the catalytic region ofCD38 inside exocytotic vesicles (5, 19); however,these vesicles in fact produce functionally activecADPR during de novo expression of CD38 (23). Toensure production of cytosolic cADPR by the intra-vesicular catalytic domain of CD38, two subsequentevents are required: NAD1 influx from the cyto-plasm into the intravesicular space, and cADPRefflux into the cytoplasm to reach its microsomalreceptors and release calcium (24–27). Whereastransport of intravesicularly produced cADPR acrossthe vesicle membrane is the result of the channel

activity of oligomeric CD38 (22), NAD1 fluxes acrossthe plasma membrane of intact cells have not yetbeen reported.

Thus, the aim of this study was twofold: to ascer-tain whether cytosolic calcium is actually increasedafter ligand-induced CD38 internalization, as itproved to be during CD38 export to the cell mem-brane (23); and, on the other hand, to elucidate themechanisms overcoming these topological prob-lems. Results obtained demonstrate that CD38 inter-nalization in various cell types is causally linked toincreased [Ca21]i. The mechanism underlying thisprocess (also accounting for enhanced [Ca21]i dur-ing CD38 exocytosis) was shown to involve influx ofcytosolic NAD1 into the endocytotic CD38-contain-ing vesicles, mediated by a hitherto unrecognizedtransmembrane pyridine dinucleotide transporter.

These data identify the occurrence of NAD1 andcADPR trafficking across cell membranes and dem-onstrate that internalization of CD38 ensures theintracellular production of functionally significantand topologically adequate (i.e., cytosolic) cADPRconcentrations in CD381 cells. Finally, the demon-stration of NAD1 fluxes across the plasmamembraneof mammalian cells and of a functional effect ofextracellular, micromolar NAD1 on [Ca21]i, medi-ated by CD38 internalization and production ofintracellular cADPR, imply a hitherto unrecognizedextracellular, hormone-like function of NAD1 onCD381 cells.

MATERIALS AND METHODS

N-Acetylcysteine (NAC), reagent grade, was kindly providedby Zambon Group S.p.A., Vicenza, Italy. Fura2-AM and thap-sigargin were purchased from Fluka, Milano, Italy. [carbonyl-14C]Nicotinamide (53 mCi/mmol) was obtained fromMoravek (Brea, Calif.). Anti-CD38 monoclonal antibodies(mAb’s) were kindly provided by Prof. F. Malavasi, Torino,Italy (IB4 and IB6), or obtained from Sigma, Milano, Italy(OKT10). Silicon oil (OEL AR 200) was purchased fromWacker-Chemie GmbH, Munchen, Germany.

8-N3-NAD1 was chemically synthesized from 8-N3-AMP andb-NMN (both purchased from Sigma), and was purified byhigh-pressure liquid chromatography (HPLC) as described(28, 29). All other chemicals were obtained from Sigma.

Cell cultures

Namalwa, Jurkat, and wild-type HeLa and NIH 3T3 cells(3T3) were purchased from ATCC (Rockville, Md.). CD38sense- and antisense-transfected HeLa and 3T3 cells (CD381

tand CD382

t, respectively) were obtained and cultured asdescribed (23). Peripheral blood lymphocytes (PBL) wererecovered after Ficoll gradient separation from freshly drawnblood samples after informed consent from healthy donors.

All cells were cultured in Dulbecco’s modified Eagle’smedium (without phenol red) (DME) supplemented with10% fetal calf serum (FCS), penicillin (100 U/ml), strepto-mycin (100 mg/ml), and glutamine (2 mM) in a humidified5% CO2 atmosphere at 37°C.

274 Vol. 13 February 1999 ZOCCHI ET AL.The FASEB Journal

Immunoflorescence microscopy

CD381t HeLa cells were grown overnight on glass coverslips.

Incubations with the various internalizing agents at millimo-lar concentrations were performed at 37°C in a petri dish witha single addition at zero time, while micromolar NAD1 or8-N3-NAD1 were continuously infused for 30 min at 25°Cthrough a 200 ml chamber at a flow rate of 50 ml/s. Allsolutions were in DME without FCS. Cells were fixed with theTriton/paraformaldehyde method (30). Nonspecific proteinbinding to cells was blocked by a 30 min incubation with 5%FCS. Cells were then incubated with anti-CD38 or anti-HLA(histocompatibility leukocyte antigen) class I (Sigma) mAb’s(10 mg/ml). A FITC-conjugated anti-mouse immunoglobulinG (IgG) mAb (Jackson Immunoresearch Laboratories Inc.,West Grove, Pa.) was used as secondary antibody, at aconcentration of 2.5 mg/ml.

Images of the samples were collected by confocal micros-copy with a BioRad MRC 1024 Instrument (krypton/argonlaser) on a Nikon DIAPHOT 200, 360 oil plan apo objectivewith NA1.4.

Cytofluorimetric analyses

CD381t HeLa cells, either untreated or treated with internal-

izing agents, were detached with trypsin and incubated withthe primary antibody against HLA class I (10 mg/ml) incomplete medium for 30 min at 4°C. Thereafter, cells werewashed and incubated with 10 mg/ml FITC-conjugated anti-mouse IgG antibody in complete medium for 30 min at 4°C.Cells were then washed and fluorescence intensity was quan-titated by flow cytometry (21). Control samples were incu-bated with the second antibody only.

Fluorimetric determination of the [Ca21]i

The experimental setting used for determination of[Ca21]i on Fura2-loaded, adherent cells has been de-scribed in detail elsewhere (23). Briefly, cells were seededon glass coverslips 18 h before the experiment. After Fura2loading of the cells, the glass coverslip was mounted in a200 ml recording chamber, mounted on the stage of aninverted microscope (Zeiss IM35, Stuttgart, Germany),continuously perfused at 25°C with solutions fed by gravitythrough solenoid microvalves, and removed by a hydraulicvacuum pump. Calcium measurements were performed onfields containing 2–10 cell bodies. At the beginning of eachexperiment, cells were washed in saline zero calciumsolution (135 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 10 mMglucose, and 5 mM Hepes, pH 7.4). Intact cells were eitherexposed to millimolar concentrations of internalizingagents under conditions of stopped flow or to micromolarconcentrations under conditions of continuous perfusionin the zero calcium solution. Calcium variations wererecorded as described (23). Nonadherent cells (Namalwa,Jurkat, and PBL) were treated similarly except that Fura2loading and calcium measurements were performed insuspension in a 2 ml cuvette (0.53106 cells/ml).

Enzymatic assays

Ectocellular cyclase activity on NGD1 (23) and protein con-tent (31) were determined on the same samples used forcalcium measurements. Glucose 6-P dehydrogenase (G6PD)and hexokinase activities were determined on sonicated cellsand on centrifuged supernatants as described (32). Thepercentage of cell lysis was calculated as the percentage of

total intracellular enzyme activity released into the superna-tant.

Determination of pyridine dinucleotide concentrations

The pyridine dinucleotide content of alkaline cell extractsand of alkalinized, centrifuged cell culture supernatants wasdetermined by a sensitive enzymatic cycling assay procedure(33), as described (23). Briefly, this assay enables the sensitive($10 pmol per assay) and specific determination of theamount of each pyridine dinucleotide species (NAD1,NADH, NADP1, and NADPH). A different assay mixture wasused to detect NAD1/NADH or NADP1/NADPH, the appro-priate substrate and enzyme being ethanol and alcoholdehydrogenase or glucose 6-phosphate and G6PD, respec-tively. An aliquot of each sample was heated to 60°C for 30min to destroy the oxidized forms. The amount of theoxidized forms was calculated from the difference betweenthe unheated (oxidized 1 reduced dinucleotides) and theheated (reduced form only) samples (33). Assays were donein duplicate; in controls (run in parallel), the enzyme wasomitted from the assay mixture. Alternatively, aliquots of eachsample were incubated 60 min at 37°C with 1 U/ml ofnucleotide pyrophosphatase (Sigma): both controls gavecomparable results and were used alternatively. All assayswere performed on freshly obtained (not frozen) samples.Protein determination was obtained on aliquots of cell ex-tracts (31).

Influx of pyridine dinucleotides into cells

Two different experimental approaches were followed in thestudies investigating pyridine dinucleotide influx into cells.As the transport proved to be bidirectional, a significantamount of the internalized nucleotides could be lost from thecells during the extensive washings necessary to remove theexcess of extracellular coenzymes. In one type of experiment,adherent HeLa or 3T3 cells (approximately 106 cells persample, seeded 24–36 h before the experiment) were rapidlywashed once in DME to remove the FCS (which could be asource of dinucleotide pyrophosphatase; L. Guida, unpub-lished data). Cells were incubated with the various pyridinedinucleotides in DME at 37°C and subsequently washed at0°C. A control flask, without cells, was incubated and washedin parallel with each sample. The number of washings wascalibrated so as to reduce the concentration of extracellulardinucleotide in the last washing of the control flask to HPLC(34) -undetectable concentrations, i.e., ,50 nM: under theseconditions, the total amount of extracellular dinucleotideremaining after removal of the last washing accounted for 1%to 10% (depending on the nucleotide type) of the intracel-lular amount of dinucleotide in untreated cells. In a parallelexperiment, cells were incubated with each dinucleotide insuspension (approximately 106 cells per sample) and theextracellular nucleotides were alternatively removed by cen-trifugation of the cells through 3 vol of silicon oil (d51.03–1.05 g/ml) for 15 s at 13,000 3 g at 25°C. Pellets werecentrifuged again through 0.5 ml of silicon oil, and cells werelysed in 200 ml of water and sonicated for 30 s (HeatSystems-Ultrasonics, Inc. W-380) in ice. The intracellulardinucleotide content was determined on alkaline cell extractswith the cycling assay (33) or, in the experiments designed todetermine Km values, on TCA extracts by HPLC (34).

In the kinetic experiments, which were performed on bothCD382 and CD381 cells, the ectocellular NADase activity ofthe latter cell types could interfere with the results by produc-ing membrane-permeable nicotinamide and ADPR (22),from which intracellular NAD1 synthesis might occur. Thus,

275CD38 INTERNALIZATION AND NAD1 TRANSLOCATION

the incubation times were brief (from 30 s to 2 min) andcontrol incubations were performed using nicotinamide andADPR at the same concentrations as those actually producedduring incubation of the CD381 cells with NAD1. Thistreatment proved not to modify the [NAD1]i.

RESULTS

GSH- and NAD1-induced internalization of CD38in CD381 cells

When HeLa cells transfected with human full-lengthCD38 (CD381

t HeLa) were incubated in the pres-ence of 10 mM GSH, a time-dependent disappear-ance of ectocellular ADP-ribosyl cyclase activity wasobserved, with 50 63, 20 6 4, and 8 64% of initialactivity remaining after 1, 12, and 24 h incubation,respectively. The decrease of surface cyclase activitywas paralleled by disappearance of membrane im-munoreactivity tested at the confocal microscopewith several anti-CD38 monoclonal antibodies andby the appearance of cytosolic dot-like staining (Fig.1), consistent with vesicle-mediated internalization(21). Closely comparable results were obtained wheneither 10 mM NAD1 (Fig. 1) or 8-N3-NAD1 (notshown) was used as internalizing agent. Ligand-induced endocytosis of CD38 proved to be specific;no internalization of HLA-class I antigens occurredin control, CD38 antisense-transfected HeLa cellsincubated with either NAD1 or GSH, as detected byboth confocal immunofluorescence and FACS anal-ysis (not shown).

Similar results were obtained on CD381t 3T3 cells,

thus confirming and extending our previous obser-vation of GSH-and NAD1-induced, vesicle-mediatedCD38 internalization in human B-lymphocytic Na-malwa cells (21). In these cells, CD38 internalizationwas found to be followed by an increase of theintracellular cADPR concentration (21). Thisprompted us to investigate the intracellular freecalcium concentration ( [Ca21]i) in CD381 cellsundergoing CD38 internalization.

CD38 internalization and increased [Ca21]i

After 3 h incubation of all cell types tested (CD381t

HeLa and 3T3 cells and Namalwa B lymphoid cells)with either NAD1 (2 mM) or GSH (10 mM), thesurface cyclase activity was reduced to approximately35% of the initial value, while the [Ca21]i wasmarkedly increased up to approximately twice thebasal concentration (Table 1). No further calciumincrease was observed upon incubation of the cellswith either NAD1 or GSH for as long as 24 h. Nosignificant differences between the various cell lineswere apparent. Comparable results were obtainedalso on peripheral blood lymphocytes (Table 1).

Neither NAD1 nor GSH produced any increase inthe [Ca21]i in CD382

t HeLa and 3T3 cells (23), thusruling out any direct effect of either CD38-internal-izing agent on the intracellular calcium levels.

The time course of the calcium increase afteraddition of the CD38-internalizing agents to CD381

t

HeLa cells was investigated. Upon addition of NAD1

(2 mM) or GSH (10 mM) to cells the [Ca21]i

increased to reach maximal values (approximatelytwice the initial concentration) within 40–45 min(Fig. 2). NAC, a stable thiol-containing molecule,elicited the same extent of calcium increase as GSHand NAD1 (Fig. 2). After removal of any internaliz-ing compound, the [Ca21]i decreased progressivelyand almost linearly, with 190, 165, and 105% of thebasal [Ca21]i being recorded after 2, 6, and 24 h,respectively (mean of four experiments). The samekinetics and extent of [Ca21]i increase were alsoobserved on CD381

t 3T3 cells and on constitutively

Figure 1. Detection of CD38 internalization by confocalfluorescence microscopy in CD381

t HeLa cells incubatedwith GSH or NAD1. CD381

t HeLa cells were incubated onglass coverslips under the conditions listed below. CD38 wasdetected by indirect immunofluorescence on fixed and per-meabilized cells. Three different monoclonal antibodies (IB4,IB6, OKT10) gave similar fluorescence patterns. Results of atypical experiment with IB4 are shown. A) Control cells,incubated without additions. B) Cells incubated with 10 mMGSH for 3 h. C) Cells incubated with 10 mM NAD1 undercontinuous flow for 30 min.

276 Vol. 13 February 1999 ZOCCHI ET AL.The FASEB Journal

CD381 Namalwa cells upon addition of each of thethree internalizing agents. This calcium increasewas not due to influx of Ca21 from the extracellu-

lar medium because all experiments were per-formed in zero calcium saline solution (see Mate-rials and Methods). Moreover, no increase in[Ca21]i was detectable upon addition of eitherNAD1 or GSH to thapsigargin-pretreated CD381

t

cells (not shown).To demonstrate that the increase in [Ca21]i is

specifically dependent on CD38 internalization,GSH was externally supplemented to cells whereCD38 internalization had been prevented by priorcovalent cross-linking of surface membrane proteins.In the absence of glutaraldehyde, exposure ofCD38t

1 HeLa cells to 10 mM GSH (40 min) resultedin an approximately 45% decrease of the ectocellu-lar cyclase activity due to CD38 internalization (insetto Fig. 3) and produced a marked (80%) andtime-dependent increase in [Ca21]i (Fig. 3). Afterglutaraldehyde pretreatment of the cells, however,the surface cyclase activity remained constant uponaddition of GSH (inset to Fig. 3). Thus, exposure ofcells to glutaraldehyde effectively prevented CD38internalization without interfering with its enzymaticactivity. Under these conditions, no [Ca21]i increasewas observed in glutaraldehyde pretreated cells afteraddition of GSH (Fig. 3).

NAD1 efflux from intact CD382 cells

These results clearly demonstrated that the calciumincrease is dependent on ligand-induced transfer of

Figure 3. Effect on [Ca21]i of pretreatment of CD381t HeLa

cells with glutaraldehyde. Adherent, Fura2-loaded, CD381t

HeLa cells (approximately 23105 cells per sample) wereincubated with (E) or without (F) 0.25% glutaraldehyde for10 min at 37°C and washed. On duplicate incubations, the[Ca21]i was continuously monitored after addition of 10 mMGSH (arrow) at 25°C, while surface cyclase activity, measuredon NGD1 (inset), was determined at the end of 40 minincubation with GSH (no55) at 25°C. Traces shown arerepresentative of results obtained in 5 consistent experi-ments. Each point is the mean of the values (no5200)recorded in 2 min.

TABLE 1. [Ca21]i in CD381 cells incubated with NAD1 orGSH

Cell typea Addition

Cyclase activityb [Ca21]ib

(% of control)

HeLat CD381 NAD1 38 6 3 185 6 15GSH 36 6 3 195 6 16

3T3tCD381 NAD1 35 6 3 198 6 16GSH 32 6 3 188 6 15

Namalwa NAD1 32 6 3 187 6 15GSH 36 6 3 170 6 13

PBL NAD1 52 6 5 200 6 18GSH 45 6 4 230 6 20

a Cells were incubated with 2 mM NAD1 or 10 mM GSH for 3 hat 37°C; ectocellular cyclase activity and [Ca21]i were determined asdescribed in Materials and Methods. b Results are expressed aspercentage of control values obtained from the corresponding celltypes incubated without NAD1 or GSH. Neither treatment deter-mined any [Ca21]i modification in CD382

t HeLa or 3T3 cells ascompared to untreated CD382

t controls. PBL, peripheral bloodlymphocytes. For absolute values of cyclase activity and [Ca21]i in thecell types listed in the Table, see ref 23.

Figure 2. Effect of CD38-internalizing agents on [Ca21]i inCD38 sense- and antisense-transfected HeLa cells. [Ca21]i wascontinuously monitored in adherent, Fura2-loaded, CD381

t(open symbols), and CD382

t (filled symbols) HeLa cells afteraddition (arrow) of 5 mM NAD1 (E), 10 mM GSH (h), or 10mM N-acetylcysteine (ƒ) at 25°C. Traces shown are representa-tive of 4 consistent experiments. Each point is the mean of thevalues (no5200) recorded in 2 min. Similar results were ob-tained with CD381

t and CD382t 3T3 cells. As reported (23), the

basal [Ca21]i is higher in CD381t cells compared with CD382

tcontrols. The inset shows the [Ca21]i after removal of GSH byextensive washing. Comparable data were observed upon remov-ing either NAD1 or NAC as internalizing agents.

277CD38 INTERNALIZATION AND NAD1 TRANSLOCATION

CD38 into the cytoplasm. After internalization, how-ever, the catalytic domain of CD38 is secluded fromcytosolic NAD1 by the endocytotic vesicle membrane(21). When internalization is triggered by exposureof cells to NAD1, availability of NAD1 to the cyclaseactivity of CD38 is ensured by endocytosed NAD1

inside the vesicles. However, GSH and NAC inducethe same extent and kinetics of [Ca21]i increase asNAD1 (Table 1 and Fig. 2) on all cell types tested.This is consistent for the occurrence of a transportmechanism for cytosolic NAD1 in the membrane ofthe CD38 internalization vesicle. Subsequent effluxof catalytically generated cADPR from the vesicle toreach its microsomal receptors is made possible bythe recently discovered cADPR-transporting activityof CD38 itself (22).

The presence of a transport mechanism responsi-ble for NAD1 influx into the endocytotic vesicleswould also allow NAD1 to be released from the cell.Thus, we investigated NAD1 efflux from wild-typeHeLa and 3T3 cells. Absence of any NAD1-hydrolyz-ing activity on the membrane of these cells waspreliminarily checked as a necessary prerequisite forthe detection of NAD1 efflux (S. Bruzzone, unpub-lished data). As illustrated in Fig. 4, repeated wash-ings of adherent HeLa cells at 37°C in DME resultedin the progressive decrease of intracellular NAD1

and NADH content ([NAD11NADH]i), which wasmore extensive during the first three medium

changes (Fig. 4). The sum of the NAD11NADHreleased into the supernatants of the sequentialwashings closely paralleled the amount of pyridinedinucleotide lost from the cells (Fig. 4). NAD1 wasthe most abundant species both in the cell extractsand in the supernatants, accounting for 90% of thetotal NAD1 1 NADH pool.

Approximately 25% of the intracellular NAD1 1NADH pool was consistently released from the cellsinto the supernatant. This NAD1 efflux was not dueto cell lysis, the extent of which in each supernatantwas below 0.5% (Fig. 4). Comparable results werealso obtained with wild-type 3T3 cells and with theCD382 murine fibroblast cell line M210B4 (notshown).

Susceptibility to inhibition is one of the key fea-tures of protein-mediated transport mechanisms. Inpreliminary experiments we attempted to inhibitefflux of metabolically labeled [14C-nicotinamide]-NAD1 from HeLa cells with extracellular unlabeledNAD1. We tested extracellular NAD1 concentra-tions of up to 0.25 mM without observing significantinhibitory effects. Higher extracellular concentra-tions of NAD1 were found to interfere with theHPLC separation of the excess [14C]-nicotinamidepresent in the cells from the NAD1 peak (34). Theremarkable [NAD1]i in these cells (7.04 61.2 nmolNAD1 1 NADH/mg protein in unwashed cells)may, however, require a much higher than 0.25 mMextracellular NAD1 concentration to create andmaintain a gradient against NAD1 efflux. To circum-vent these technical difficulties, the NAD1 analog8-N3-NAD1 was synthesized (28, 29). Addition of theNAD1 analog to the supernatant, at progressivelyincreasing concentrations (the highest being 1.2mM), markedly inhibited efflux of metabolicallylabeled [14C] NAD1 from HeLa cells (Fig. 5). Uponremoval of the analog, NAD1 efflux steadily resumed(Fig. 5). This reversible inhibition of NAD1 efflux by8-N3-NAD1, which was observed on 3T3 cells as well,rules out conclusively cell lysis as the source ofextracellular NAD1 and indicates the existence of aninhibitable, bidirectional NAD1 transport in themembrane of these cells.

These results suggested an experimental setting tomaximize NAD1 efflux from these cell lines bymaintaining a near-zero extracellular [NAD1]. Ad-herent HeLa cells were first washed to remove theintracellular exchangeable NAD1 and cells weresubsequently cultured with or without mediumchanges for 7 h. As shown in Fig. 6, a much higheramount of NAD1 was released into the supernatantwhen the medium was changed every 2 h as com-pared to cell cultures where the medium was notreplaced and the intracellular and extracellularNAD1 concentrations could equilibrate. Approxi-mately 3% of the intracellular NAD1 was released

Figure 4. NAD1 efflux from wild-type HeLa cells.Adherent,exponentially growing HeLa cells (7 flasks; 106 cells per 25cm2 flask) were either extracted in alkaline buffer (seeMaterials and Methods) without washing (zero time) orsubjected to sequential 5 min-washings at 37°C in DME. Eachflask was washed for a different number of times, from 1 to 6,as shown on the abscissa. NAD1 1 NADH content in cellextracts (■) and supernatants (�) was estimated by theenzymatic cycling procedure (33) and the total amount ofNAD1 1 NADH released was calculated as the sum of thepyridine dinucleotide present in each individual mediumchange. The percentage of cell lysis in each washing wasbelow 0.5% as determined by assay of G6PD and hexokinaseactivities in the corresponding supernatants. Results of arepresentative experiment are shown. Similar results wereobtained with wild-type 3T3 cells.

278 Vol. 13 February 1999 ZOCCHI ET AL.The FASEB Journal

into the medium per hour when the medium waschanged. The [NAD1]i at the beginning and end ofthe 7 h incubation period was not significantlydifferent: thus, biosynthesis apparently preventedNAD1 depletion inside the cells. In fact, incubationof the cells under the same culture conditions (me-dium changes every 2 h) but without nicotinamide inthe medium resulted in a 50% reduction of NAD1

release into the supernatant.

Influx of pyridine dinucleotides into intact cells

The results shown in Fig. 6 indicate that NAD1

transport depends on the concentration gradientacross the plasmamembrane. To demonstrate un-equivocally the bidirectional nature of dinucleotidetransport and to investigate its kinetic properties, weanalyzed the influx of pyridine dinucleotides intoHeLa and 3T3 cells. A short incubation time wasnecessary to prevent intracellular conversions be-tween the various dinucleotide pools. As shown inTable 2, incubation of the cells with each of the fourpyridine dinucleotide species tested resulted in atwo- to ninefold increase in intracellular concentra-

tion without any significant modifications of thelevels of the other dinucleotides. Partial loss of theinternalized nucleotides during washings was con-firmed, for each incubation experiment, by HPLCanalysis of the last washing, which indicated dinucle-otide concentrations 10- to 20-fold higher than thatpresent in the last washing of the control flask (seeMaterials and Methods). If this loss was taken intoaccount and added to the intracellular amount ofdinucleotides actually measured in cell extracts, thevalues of influx were closely comparable to thoseobtained with the silicon oil procedure. Similarresults were also obtained with 3T3 cells.

Finally, the kinetics of NAD1 influx were investi-gated both in CD382 (HeLa and 3T3) cells and inCD381 (Namalwa and Jurkat) cell lines. Influx ofNAD1 into intact HeLa cells followed hyperbolicsaturation kinetics (not shown). The apparent Km

for NAD1 transport was approximately 15 mM, sim-ilar for all cell types tested and close to valuespreviously recorded on resealed human erythrocytemembranes (22). The initial rate of NAD1 influx wascalculated to be approximately 10 nmol/(min z mgcell protein) for HeLa and 3T3 cells. These valuesare the mean from results obtained with both exper-imental settings described under Materials andMethods. Influx of a number of NAD1 analogs wasalso investigated on both HeLa and 3T3 cells: whilemodifications in the adenine moiety of the molecule

Figure 6. NAD1 efflux from HeLa cells cultured with orwithout medium changes. Adherent, exponentially growingHeLa cells (106 per sample) were washed in DME andincubated at 37°C in the same medium with (■) or without(F) medium changes (indicated by the arrows). The NAD1

concentration in the medium was determined by the cyclingassay procedure (33) and the total amount of NAD1 released(mean 6sd of 4 experiments) was calculated as the sum ofthe NAD1 found in each supernatant. Similar results wereobtained with 3T3 cells. The percentage of cell lysis in eachsupernatant was below 0.3%, as determined by assay of G6PDand hexokinase activities.

Figure 5. Inhibition of NAD1 efflux from HeLa cells by8-N3-NAD1Intracellular NAD1 was metabolically labeled byculturing HeLa cells for 5 days at 37°C in DME, 10% FCScontaining [14C]-nicotinamide (5 mCi per sample). At thetime points indicated the supernatants (0.4 ml) were replacedwith the media listed below and the amount of radioactiveNAD1 released was determined by (Œ) HPLC. Control (■).DME containing 8-N3-NAD1 at a final concentration of 0.4,0.8, and 1.2 mM in the first three medium changes, respec-tively. The arrow indicates removal of the inhibitor. The totalamount of [14C]-NAD1 released was calculated as the sum ofthe radioactive dinucleotide present in each individual super-natant. Results are expressed as percentage of the total[14C]-NAD1 (intracellular 1 extracellular) released into themedium. Results of a representative experiment, out of 4different ones, are shown.

279CD38 INTERNALIZATION AND NAD1 TRANSLOCATION

apparently did not affect transport (8-N3-NAD1,NHD1 and NGD1 were internalized similarly toNAD1), modifications at the nicotinamide ring ei-ther decreased (a-NAD1) or completely prevented(FAD) internalization (not shown). These resultsseem to indicate some extent of specificity of thispyridine dinucleotide transport system.

Role of cADPR in the increased [Ca21]i inducedby CD38 internalization

8-N3-NAD1 is a substrate of both the enzymatic andof the cyclic nucleotide-transporting activities ofCD38 (L. Franco, unpublished data), but 8-N3-cADPR is a potent antagonist of cADPR-inducedcalcium release (28, 29). We took advantage of thefact that 8-N3-NAD1 is also internalized by the pyri-dine dinucleotide-transporting system (see above) todevise an experiment aimed to causally correlateendocytosis of CD38, intracellular production ofcADPR and [Ca21]i increase. Adherent CD381

t

HeLa cells were first depleted of their cytosolic freeNAD1 by continuous perfusion with DME in therecording chamber described in Materials and Meth-ods. Cells were incubated with either 10 mM NAD1

or 20 mM 8-N3-NAD1 for 30 min under continuousflow, and the [Ca21]i and the surface cyclase activityon NGD1 were determined on the same sample. Thedecrease of surface cyclase activity was closely com-parable in NAD1- and 8-N3-NAD1-incubated cells,down to 48% and 50% of starting values, respec-tively. In addition, confocal fluorescence microscopy

confirmed a comparable extent of CD38 internaliza-tion with both dinucleotides (not shown). The[Ca21]i, however, increased only in the cells incu-bated with NAD1 while keeping fairly stable in thecells exposed to the NAD1 analog (Fig. 7). Theseresults demonstrate a causal correlation between theintracellular cADPR production by internalizedCD38 and the [Ca21]i increase. Moreover, the ex-perimental setting of the continuous flow enabled usto demonstrate that micromolar concentrations ofNAD1 are sufficient to elicit both internalization ofCD38 and enhanced [Ca21]i, provided fresh NAD1

is continuously supplied to the CD381 cells.

DISCUSSION

The original aim of this study was to investigate theeffect of ligand (NAD1 or GSH) -induced internal-ization of CD38 on [Ca21]i. A link between the twoprocesses was suggested, but not yet demonstrated,by our earlier finding that the NAD1-or GSH-in-duced specific internalization of CD38 through nonclathrin-coated endocytotic vesicles in Namalwa B-lymphocytes results in increased [cADPR]i (21).

The preliminary demonstration (Fig. 1 and Table1) that CD38 internalization occurs in different celltypes (Namalwa, CD381

t HeLa and 3T3 cells, andPBL), with a similar mechanism (NAD1- or GSH-induced, vesicle-mediated endocytosis) and withcomparable kinetics (approximately 50% being in-ternalized within 1 h incubation), supports the con-clusion that internalization is an intrinsic property of

Figure 7. Effect of 8-N3-NAD1 on [Ca21]i in CD381t HeLa

cells. Adherent HeLa cells, depleted of endogenous, ex-changeable NAD1 by extensive washings and then Fura2-loaded, were incubated in DME at 25°C in the presence ofeither 10 mM (E) or 100 mM (h) NAD1 or in the presence of20 mM 8-N3-NAD1 (Œ), under conditions of continuous flow.The [Ca21]i was continuously monitored: each point is themean of 200 values, recorded in 2 min. Traces shown arerepresentative of 3 consistent experiments.

TABLE 2. Influx of NAD1, NADH, NADP1, and NADPH intoHeLa cells

Extracellulardinucleotide

Intracellular pyridine dinucleotidea

NAD1 NADH NADP1 NADPH

(% of control)

NAD1 220 6 20 109 6 10 98 6 9 94 6 8NADH 113 6 10 300 6 30 82 6 7 110 6 10NADP1 121 6 11 101 6 10 890 6 80 126 6 10NADPH 112 6 10 91 6 8 121 6 11 230 6 15

a Wild-type HeLa cells (106 per sample) were washed once inDME and incubated for 5 min at 37°C in DME containing theindicated pyridine dinucleotide at 10 mM final concentration. Eachincubation was performed in duplicate on adherent and on sus-pended cells. At the end of the incubation, adherent cells werewashed with DME at 0°C; cells in suspension were briefly centrifugedthrough three volumes of silicon oil. The intracellular pyridinedinucleotide concentrations were determined by cycling assay (33)on alkaline cell extracts; the amount of dinucleotide lost in the lastwashing was determined by HPLC (34) and added to the intracellulardinucleotide concentration measured in adherent cells. Resultsshown are the mean 6sd of the values obtained from both experi-mental settings in three independent experiments, and are expressedas percentage of control values obtained from cells incubated withoutany addition (NAD1, 7.04 60.7 nmol/mg; NADH, 0.9 60.07 nmol/mg; NADP1, 0.16 60.01 nmol/mg; NADPH, 1.4 60.01 nmol/mg).Similar results were obtained with 3T3 cells.

280 Vol. 13 February 1999 ZOCCHI ET AL.The FASEB Journal

CD38 itself and does not depend on cell type ormembrane protein environment.

After CD38 internalization, the [Ca21]i increasedto maximal values (i.e., approximately twice theinitial concentration) within 45–60 min incubationin all cell types tested and with either internalizingagent used—NAD1, GSH, or NAC (Table 1 and Fig.2). These kinetics of calcium increase closely parallelthe process of CD38 internalization and are differentfrom the transient (seconds) and more limited(around 50%) increase of cytosolic calcium thatimmediately follows exposure of CD381-transfectedHeLa cells to NAD1 (22). The rapid elevation of[Ca21]i triggered in the latter case is due to therecently demonstrated property of CD38 to behaveas an active channel for generation and selectiveinflux of cADPR across membranes (22). On thecontrary, the sustained increase of [Ca21]i afterexposure of cells to NAD1, GSH, or NAC is mecha-nistically related to 1) specific CD38 internalization,as demonstrated by blockade of both processes byprior cross-linking of membrane proteins with glu-taraldehyde (Fig. 3), and 2) the consequently en-hanced formation of intracellular cADPR, as shownby the experiments with 8-N3-NAD1 (Fig. 7).

A significant outcome of the present investigationwas the demonstration of a hitherto unrecognizedtransport system for NAD1 in cell membranes and ofits role in the regulation of cytosolic calcium relatedto the ligand-induced CD38 internalization. Evi-dence for this transport emerged from the attemptto explain a topological inconsistency of the CD38/cADPR system during either internalization (thisstudy) or export (23) of membrane-bound CD38. Infact, due to the reverse polarity of asymmetricallyoriented CD38 molecules in the plasmamembrane,both endocytotic and exocytotic membrane vesiclesdisplay the carboxyl-terminal catalytic region ofCD38 in the intravesicular space, therefore not di-rectly accessible to its cytosolic substrate NAD1.Accordingly, for the endocytotic or exocytotic vesi-cles to behave as cADPR-dependent calcium-releas-ing systems in cells, as actually observed in differentexperimental conditions (21,23), two requirementsshould be met: 1) availability of NAD1 to the appar-ently secluded active site of CD38, and 2) release ofcatalytically generated cADPR from both types ofvesicles into the cytosol, allowing the cyclic nucleo-tide to reach its receptor-operated calcium stores(24–27). As mentioned, membrane-embeddedCD38 was shown to meet the latter requirementsince it proved to be a catalytically active transporterof cADPR, but not of NAD1, across membranes (22),by virtue of its oligomeric (dimeric and tetrameric),channel-generating structure (22, 35). The need toidentify an NAD1-translocating system prompted usto investigate NAD1 fluxes across the plasmamem-

brane, since the endocytotic vesicles derive from theplasmamembrane itself. These studies were carriedout on intact cells, mostly using the wild-type humanHeLa and murine 3T3 cells, which do not exhibitany NAD1-cleaving enzyme activity at their outersurface. Therefore, both influx and efflux of NAD1

from these cells could be investigated without anyinterference by NAD1 degradation.

The present findings, besides conclusively demon-strating that cell membranes are permeable toNAD1, also identify some properties of this noveltransport system. This proves to be a passive trans-port, because the direction of pyridine dinucleotideflux depends on the concentration gradient onlyand no apparent energy source is required. Thetransport system is characterized by low dinucleotidespecificity and follows hyperbolic saturation kinetics.No significant effect of temperature was observed ininflux and efflux experiments (not shown), thissuggesting that dinucleotide translocation occursacross a channel rather than involving a transporteroscillating between different conformations (36).

The demonstration of NAD1 fluxes through theplasmamembrane of intact mammalian cells is, toour knowledge, without precedent, although slowNAD1 leakage across the mitochondrial membraneshas been described both in plant (37) and mamma-lian cells (38). We found that approximately 25% ofintracellular NAD1 can be lost from HeLa or 3T3cells upon repeated washings: this ‘exchangeable’NAD1 (over 90% oxidized) is probably the cytosolic,free (i.e., non enzyme-bound) dinucleotide (Fig. 4).Efflux of NAD1 can be reversibly inhibited by theNAD1 analog 8-N3-NAD1 (Fig. 5), but attempts toirreversibly inhibit NAD1 efflux from HeLa cells byphotolysis were unsuccessful. However, photoaffinitylabeling does not necessarily yield stoichiometriclabeling: values between 0.01 and 1.0 mol to mol areusually reported (39).

Influx of all pyridine dinucleotides (NAD1,NADH, NADP1, and NADPH) occurs across theplasmamembrane of intact HeLa and 3T3 cells:although some degree of intracellular reequilibra-tion between the various oxidized and reduced dinu-cleotide pools takes place at longer incubation times(not shown), selective increase of the intracellularconcentration of each extracellularly added nucleo-tide was obtained by reducing the incubation time to5 min (Table 2). NAD1 influx into both CD382

HeLa and 3T3 cells and CD381 Namalwa and Jurkatcells displays hyperbolic saturation kinetics: over theshort incubation time sufficient for NAD1 influx(below 2 min), a very limited NAD1 degradationoccurred with CD381 cells. Incubation of these cellswith concentrations of nicotinamide and ADPR aslow as those produced ectocellularly from NAD1

during the incubation time failed to determine any

281CD38 INTERNALIZATION AND NAD1 TRANSLOCATION

increase of the [NAD1]i. This control confirms theexistence of NAD1 transport also in the plasmamem-brane of CD381 lymphocytic cell lines.

It is significant that micromolar NAD1 concentra-tions were effective in inducing both CD38 internal-ization and calcium increase (Figs. 1, 7). Althoughreported extracellular NAD1 concentrations are ap-proximately 10 nM in rat cerebellar interstitial fluid(6) and 50–60 nM in human blood serum (T. F.Walseth, personal communication), these values rep-resent average levels of the dinucleotide in theextracellular compartment: micromolar NAD1 con-centrations in the extracellular fluids may bereached in specific conditions (e.g., apoptosis inlymphnodes; see ref 40) and in selected districts, alsoin view of the demonstrated ability of fibroblast andepithelial cell lines to steadily release NAD1 (Fig. 6).It is likely that also micromolar concentrations ofGSH or reducing agents can induce internalizationof CD38 in vivo; a relevant observation is that oraladministration of therapeutic doses of N-acetylcyste-ine for 3 days resulted in a 30% decrease of ectocel-lular ADP-ribosyl cyclase activity in PBL (L. Guida, E.Zocchi, and A. De Flora, unpublished data).

Extracellular GSH proved not to affect the effluxof [14C]-NAD1 from metabolically labeled HeLacells (not shown). Accordingly, the presence of highGSH concentrations in the endocytotic vesicles gen-erated by GSH-induced internalization does not re-strict the influx of cytosolic NAD1 into these vesicles.The high intracellular concentrations of NAD1 inthe cell lines investigated (HeLa and 3T3) and thehigh rate of NAD1-transporting activity are conso-nant with a remarkable efficiency of cytosolic NAD1

influx into the vesicles.The identification of a still unrecognized dinucle-

otide transport system, responsible for bidirectionalNAD1 flux across membranes, and the recentlyestablished nature of CD38 as a selective and unidi-rectional cADPR transporter (22) allow us to ratio-nalize the complex interplay of intracellular traffick-ing of NAD1 and of cADPR as well. Both transportsystems seem to operate sequentially inside cells so asto bypass apparent processes of compartmentaliza-tion and to produce functionally significant, cytoso-lic cADPR concentrations during endocytosis or exo-cytosis, effective on intracellular calcium levels. Aspreviously postulated (22), this subcellular type ofstructural and functional organization centered onopposite polarity of NAD1 and cADPR transportsmay provide the dual advantage of preventing unre-stricted NAD1 consumption, while representing aflexible means for finely tuning the NAD1-depen-dent regulation of [Ca21]. Unequivocal confirma-tion of the regulatory role of NAD1 on [Ca21]i

should come from elucidation of the patterns ofNAD1-induced, cADPR-related [Ca21]i increases in

specific cell types (e.g., oscillatory vs. sustained) andfrom identification of selected molecular targets ofthese calcium movements, e.g., expression of specificgenes (41–43). Dissection between distinctive func-tional consequences of oscillatory and sustained in-creases of [Ca21]i, as possibly elicited by differentschedules of exposure and by different concentra-tions of internalizing compounds, will be importantin this respect, since CD38 per se is suitable to elicitboth types of effects as a steady cADPR channel (22)and as a catalytic component of slowly cADPR-generating vesicles, respectively.

The recognized efflux of NAD1 from the plasma-membrane of intact cells and the properties oftransmembrane CD38 suggest that the CD38-relatedregulation of [Ca21]i is not restricted to a singleCD381 cell. Rather, it may occur through transfer ofNAD1 from any cell type to CD381 cells, accordingto paracrine mechanisms whereby NAD1 could actas a hormone, involved in cell-to-cell interactions,and cADPR as its intracellular second messenger.These results, besides accounting for a hitherthounknown cross-talk of signal metabolites betweendifferent cell compartments or between differentcells, also identify potential targets for the develop-ment of new drug molecules affecting intracellularcalcium levels in various diseases characterized byimbalanced calcium homeostasis.

We are indebted to Dr. R. Graeff and to Professor F.Palmieri for helpful advice. This study was supported in partby the Italian Ministry of University and Scientific and Tech-nological Research (MURST-PRIN), by the Associazione Itali-ana per la Ricerca sul Cancro (AIRC), by the 5% Project onBiotechnology (MURST-CNR), by the CNR Target Project onBiotechnology, and by the Fondazione Cassa di Risparmio diGenova e Imperia.

REFERENCES

1. Jackson, D. G., and Bell, J. I. (1990) Isolation of a cDNAencoding the human CD38 (T10) molecule, a cell surfaceglycoprotein with an unusual discontinuous pattern of expres-sion during lymphocyte differentiation. J. Immunol. 144, 2811–2817

2. Malavasi, F., Funaro, A., Roggero, S., Horenstein, A., Calosso, L.,and Mehta, K. (1994) Human CD38: a glycoprotein in search ofa function. Immunol. Today 15, 95–97

3. Lund, F., Solvason, N., Grimaldi, J. C., Parkhouse, R. M. E., andHoward, M. (1995) Murine CD38: an immunoregulatory ec-toenzyme. Immunol. Today 16, 469–473

4. Mehta, K., Shahid, U., and Malavasi F. (1996) Human CD38, acell-surface protein with multiple functions. FASEB J. 10, 1408–1417

5. Lee, H. C., Graeff, R., and Walseth, T. F. (1995) Cyclic ADP-ribose and its metabolic enzymes. Biochimie 77, 345–355

6. De Flora, A., Guida, L., Franco, L., Zocchi, E., Pestarino, M.,Usai, C., Marchetti, C., Fedele, E., Fontana, G., and Raiteri, M.(1996) Ectocellular in vitro and in vivo metabolism of cADP-ribose in cerebellum. Biochem. J. 320, 665–672

7. Takasawa, S., Tohgo, A., Noguchi, N., Koguma, T., Nata, K.,Sugimoto, T., Yonekura, H., and Okamoto, S. (1993) Synthesisand hydrolysis of cyclic ADP-ribose by human leukocyte antigen

282 Vol. 13 February 1999 ZOCCHI ET AL.The FASEB Journal

CD38 and inhibition of hydrolysis by ATP. J. Biol. Chem. 268,26052–26054

8. Zocchi, E., Franco, L., Guida, L., Benatti, U., Bargellesi, A.,Malavasi, F., Lee, H. C., and De Flora, A. (1993) A single proteinimmunologically identified as CD38 displays NAD1 glycohydro-lase, ADP-ribosyl cyclase and cyclic ADP-ribose hydrolase activi-ties at the outer surface of human erythrocytes.Biochem. Bio-phys. Res. Commun. 196, 1459–1465

9. Howard, M., Grimaldi, J. C., Bazan, J. F., Lund, F. E., Santos-Argumedo, L., Parkhouse, R. M. E., Walseth, T. F., and Lee,H. C. (1993) Formation and hydrolysis of cyclic ADP-ribosecatalyzed by lymphocyte antigen CD38. Science 262, 1056–1059

10. Kontani, K., Nishina, H., Ohoka, J., Takahashi, K., and Katada,T. (1993) NAD glycohydrolase specifically induced by retinoicacid in human leukemic HL-60 cells: identification of the NADglycohydrolase as leukocyte cell surface antigen CD38. J. Biol.Chem. 268, 16895–16898

11. Summerhill, R. J., Jackson, D. G., and Galione, A. (1993)Human lymphocyte antigen CD38 catalyzes the production ofcyclic ADP-ribose. FEBS Lett. 355, 231–233

12. Aarhus, R., Graeff, R. M., Dickey, D. M., Walseth, T. F., and Lee,H. C. (1995) ADP-ribosyl cyclase and CD38 catalyze the synthesisof a calcium mobilizing metabolite from NADP. J. Biol. Chem.270, 30327–30334

13. Lee, H. C. (1997) Mechanisms of calcium signaling by cyclicADP-ribose and NAADP. Physiol. Rev. 77, 1133–1164

14. Lee, H. C., Graeff, R. M., and Walseth, T. F. (1997) ADP-ribosylcyclase and CD38. Multi-functional enzymes in Ca21 signaling.In: ADP-Ribosylation in Animal Tissues (Haag, F., and Koch-Nolte, F., eds) pp. 411–419, Plenum Press, New York

15. Wu, Y., Kuzma, J., Marechal, E., Graeff., R., Lee, H. C., Foster,R., and Chua, N. H. (1997) Abscisic acid signaling throughcyclic ADP-ribose in plants. Science 278, 2126–2129

16. Lee, H. C., Walseth, T. F., Bratt, G. T., Hayes, R. N., andClapper, D. L. (1989) Structural determination of a cyclicmetabolite of NAD1 with intracellular Ca21 mobilizing activity.J. Biol. Chem. 264, 1608–1615

17. Lee, H. C., Galione, A., and Walseth, T. F. (1994) CyclicADP-ribose: metabolism and calcium mobilizing function. In:Vitamins and Hormones (Litwack, G., ed) Vol. 48, pp. 199–258,Academic Press, Orlando, Florida

18. Carafoli, E. (1987) Intracellular calcium homeostasis. Annu.Rev. Biochem. 56, 395–433

19. De Flora, A., Guida, L., Franco, L., and Zocchi, E. (1997) TheCD38/cyclic ADP-ribose system: a topological paradox. Int.J. Biochem. Cell. Biol. 29, 1149–1166

20. De Flora, A., Franco, L., Guida, L., Bruzzone, S., and Zocchi, E.(1998) Ectocellular CD38-catalyzed synthesis and intracellularCa21-mobilizing activity of cyclic ADP-ribose. Cell Biochem.Biophys. 28, 45–62

21. Zocchi, E., Franco, L., Guida, L., Piccini, D., Tacchetti, C., andDe Flora, A. (1996) NAD1-dependent internalization of thetransmembrane glycoprotein CD38 in human Namalwa B cells.FEBS Lett. 369, 327–332

22. Franco, L., Guida, L., Bruzzone, S., Zocchi, E., Usai, C., and DeFlora, A. (1998) The transmembrane glycoprotein CD38 is acatalytically active transporter responsible for generation andinflux of the second messenger cyclic ADP-ribose across mem-branes. FASEB J. 13, 1507–1520

23. Zocchi, E., Daga, A., Usai, C., Franco., L., Guida, L., Bruzzone,S., Costa, A., Marchetti, C., and De Flora, A. (1998) Expressionof CD38 increases intracellular calcium concentration andreduces doubling time in HeLa and 3T3 cells. J. Biol. Chem.273, 8017–8024

24. Lee, H. C. (1991) Specific binding of cyclic ADP-ribose tocalcium-storing microsomes from sea urchin eggs. J. Biol.Chem. 266, 2276–2281

25. Lee, H. C., Aarhus, R., and Walseth, T. F. (1993) Calciummobilization by dual receptors during fertilization of sea urchineggs. Science 261, 352–355

26. Walseth, T. F., Aarhus, R., Kerr, J. A., and Lee, H. C. (1993)Identification of cyclic ADP-ribose binding proteins by photoaf-finity labeling. J. Biol. Chem. 268, 26686–26691

27. Lee, H. C., Aarhus, R., Graeff, R., Guarnack, M. E., and Walseth,T. F. (1994) Cyclic ADP-ribose activation of the ryanodinereceptor is mediated by calmodulin. Nature (London) 370,307–309

28. Walseth, T. F., and Lee, H. C. (1993) Synthesis and character-ization of antagonists of cyclic-ADP-ribose-induced Ca21 release.Biochim. Biophys. Acta 1178, 235–242

29. Walseth, T. F., Aarhus, R., Gurnack, M. E., Wong, L., Breitinger,H.-G., Gee, K. R., and Lee, H.C. (1997) Preparation of cyclicADP-ribose antagonists and caged cyclic ADP-ribose. In: Meth-ods in Enzymology (Abelson, J. N., and Simon, M. I., eds) Vol.280, 294–305

30. Higgings, D. G., and Sharp, P. M. (1988) CLUSTAL: a packagefor performing multiple sequence alignment on a microcom-puter. Gene 73, 237–244

31. Bradford, M. (1976) A rapid and sensitive method for quanti-tation of microgram quantities of protein utilizing the principleof protein-dye binding. Anal. Biochem. 72, 248–252

32. Beutler, E. (1984) Red Cell Metabolism: A Manual of Biochem-ical Methods, 3rd Ed, Grune & Stratton, New York

33. Zerez, C. R., Lee, S. J., and Tanaka, K. R. (1987) Spectrophoto-metric determination of oxidized and reduced pyridine nucle-otides in erythrocytes using a single extraction procedure. Anal.Biochem. 164, 367–373

34. Guida, L., Franco, L., Zocchi, E., and De Flora, A. (1995)Structural role of disulfide bridges in cyclic ADP-ribose relatedbifunctional ectoenzyme CD38. FEBS Lett. 368, 481–484

35. Bruzzone, S., Guida, L., Franco, L., Zocchi, E., Corte, G., and DeFlora, A. (1998) Dimeric and tetrameric forms of catalyticallyactive transmembrane CD38 in transfected HeLa cells. FEBSLett. 433, 275–278

36. Stryer, L. (1995) Biochemistry, 4th Ed, W.H. Freeman andCompany, New York

37. Neuberger, M., and Douce, R. (1983) Slow passive diffusion ofNAD1 between intact isolated plant mitochondria and suspend-ing medium. Biochem. J. 216, 443–450

38. Rustin, P., Parfait, B., Chretien, D., Bourgeron, T., Djouadi, F.,Bastin, J., Rotig, A., and Munnich, A. (1996) Fluxes of Nicotin-amide adenine dinucleotides through mitochondrial mem-branes in human cultured cells. J. Biol. Chem. 271, 14785–14790

39. Haley, B. E. (1991) Nucleotide photoaffinity labeling of proteinkinase subunits. In: Methods in Enzymology (Hunter, T., andSefton, B. M., eds) Vol. 200, 477–487

40. Zupo, S., Rugarli, S., Dono, M., Tamborelli, G., Malavasi, F., andFerrarini, M. (1994) CD38 signalling by agonistic monoclonalantibodies prevents apoptosis of human germinal center B cells.Eur. J. Immunol. 24, 1218–1222

41. Dolmetsch, R. E., Xu, K., and Lewis, R. S. (1998) Calciumoscillation increase the efficiency and specificity of gene expres-sion. Nature (London) 392, 933–936

42. Li, W. H., Llopis, J., Whitney, M., Zlokarnik, G., and Tsien, R. Y.(1998) Cell-permeant caged InsP3/ester shows that Ca21 spikefrequency can optimize gene expression. Nature (London) 392,936–941

43. Meldolesi, J. (1998) Oscillation, activation, expression. Nature(London) 392, 863–866

Received for publication August 17, 1998.Revised for publication October 16, 1998.

283CD38 INTERNALIZATION AND NAD1 TRANSLOCATION