CD38 Signaling in B Is Controlled by Its Ectodomain but Occurs Independently of Enzymatically Generated ADP-Ribose or Cyclic ADP-Ribose This information is current as of September 27, 2021. Frances E. Lund, Hélène M. Muller-Steffner, Naixuan Yu, C. David Stout, Francis Schuber and Maureen C. Howard J Immunol 1999; 162:2693-2702; ; http://www.jimmunol.org/content/162/5/2693 Downloaded from

References This article cites 39 articles, 19 of which you can access for free at: http://www.jimmunol.org/content/162/5/2693.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 27, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1999 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. CD38 Signaling in B Lymphocytes Is Controlled by Its Ectodomain but Occurs Independently of Enzymatically Generated ADP-Ribose or Cyclic ADP-Ribose1

Frances E. Lund,2*He´le`ne M. Muller-Steffner,† Naixuan Yu,* C. David Stout,‡ Francis Schuber,† and Maureen C. Howard3*

CD38 is a type II transmembrane that is expressed by many cell types including lymphocytes. Signaling through CD38 on B lymphocytes can mediate activation, proliferation, and secretion. Additionally, coligation of CD38 and the B cell Ag receptor can greatly augment B cell Ag receptor responses. Interestingly, the extracellular domain of CD38 catalyzes the conversion of NAD؉ into nicotinamide, ADP-ribose (ADPR), and cyclic ADPR (cADPR). cADPR can induce intracellular calcium release in an inositol trisphosphate-independent manner and has been hypothesized to regulate CD38-mediated signaling. Downloaded from We demonstrate that replacement of the cytoplasmic tail and the transmembrane domains of CD38 did not impair CD38 signaling, coreceptor activity, or activity. In contrast, independent point mutations in the extracellular domain of CD38 dramatically impaired . However, no correlation could be found between CD38-mediated signaling and the capacity of CD38 to catalyze an enzyme reaction and produce cADPR, ADPR, and/or nicotinamide. Instead, we propose that CD38 signaling and coreceptor activity in vitro are regulated by conformational changes induced in the extracellular domain upon ligand/substrate binding, rather than on actual turnover or generation of products. The Journal of Immunology, 1999, 162: 2693–2702. http://www.jimmunol.org/

D38 is an evolutionarily conserved type II transmem- induce the release of intracellular Ca2ϩ in mammalian cells such as brane glycoprotein that is expressed extensively on many pancreatic acinar and ␤-cells, neuronal cells, and heart and pitu- C cell types including lymphocytes (reviewed in Ref. 1). itary cells (reviewed in Ref. 12). CD38 is composed of a short cytoplasmic tail and an unremarkable Like the Aplysia cyclase enzyme, the extracellular domain of transmembrane domain, neither of which is homologous to any CD38 possesses ADP-ribosyl cyclase activity (13, 14). However, known (2–4). In contrast, the extracellular domain of unlike the Aplysia cyclase enzyme, CD38 is also able to mediate CD38 shares structural homology with a family of that the catalysis of both NADϩ and cADPR into ADP-ribose (ADPR) includes the cytosolic ADP-ribosyl cyclase enzyme isolated from through its NADϩ and cADPR hydrolase activities (13). In fact, by guest on September 27, 2021 Aplysia californica (5). The ADP-ribosyl cyclase enzyme mediates CD38 is a more efficient NADϩ glycohydrolase enzyme than the catalysis of NADϩ into cyclic ADP-ribose (cADPR)4 (6, 7). In ADP-ribosyl cyclase enzyme, as Ͼ97% of the total product pro- sea urchin egg homogenates, cADPR has been shown to induce the duced by CD38 is ADPR (13). Nevertheless, it has been hypoth- mobilization of calcium from intracellular stores by an inositol esized that the cADPR produced via the cyclase activity of CD38 trisphosphate-independent, but caffeine- and ryanodine-sensitive, may be an important regulator of (15). pathway (8–10), suggesting that cADPR may induce calcium re- In the hemopoietic system, CD38 has been shown to regulate lease from the ryanodine receptor channel complex in the endo- activation and effector functions (1, 16). We and oth- plasmic reticulum (10, 11). Cyclic ADPR has also been shown to ers have shown that engagement of CD38 using anti-CD38 Abs can induce a proliferative response in B and T lymphocytes (17– 19), and that coligation of CD38 and the Ag receptor on B cells *DNAX Research Institute, Palo Alto, CA 94304; †Laboratoire de Chimie Bioor- (BCR) can augment BCR-mediated responses (20). Additionally, ganique, Unite´ Mixte de Recherche 7514 Centre National de la Recherche anti-CD38 stimulation has been shown to be a potent inducer of Scientifique-Universite´ Louis Pasteur, Faculte´ de Pharmacie, Illkirch, France; and cytokine production in T and B cells (20–22). Finally, CD38 ‡Department of Molecular Biology, The Scripps Research Institute, La Jolla, CA 92037 clearly plays an important role in the immune system in vivo, as Received for publication September 1, 1998. Accepted for publication November animals deficient in CD38 have marked deficiencies in their ability 23, 1998. to mount humoral immune responses (23). The costs of publication of this article were defrayed in part by the payment of page To examine the molecular requirements for anti-CD38-mediated charges. This article must therefore be hereby marked advertisement in accordance activation and costimulation in murine B cells, we recently estab- with 18 U.S.C. Section 1734 solely to indicate this fact. lished an in vitro CD38 signaling model system using the A20 1 DNAX Research Institute is fully funded by Schering Plough Corp. F.S. and H.M.-S. were supported by the Centre National de la Recherche Scientifique Pro- murine B cell lymphoma (20). A20 cells are CD38 negative but gramme Physique et Chimie du Vivant (Grant 1997–4053). C.D.S. was supported by express a functional BCR complex, which, when cross-linked, will National Science Foundation Grant MCB9513421. initiate the synthesis of IL-2 (24–26). When CD38 was stably 2 Address correspondence and reprint requests to Dr. Frances E. Lund, The Trudeau expressed in A20 cells, the resultant CD38ϩ clones inducibly pro- Institute,P.O.Box59,SaranacLake,NY12983.E-mailaddress:flund@trudeauinstitute. org duced IL-2 after CD38 was cross-linked with a polyclonal Ab (polyanti-CD38) (20). Furthermore, coligation of CD38 and the 3 Current address: Anergen, Inc., 301 Penobscot Dr., Redwood City, CA 94063. BCR induced a potent synergistic IL-2 response in these cells, 4 Abbreviations used in this paper: cADPR, cyclic adenosine diphosphate ribose; ADPR, adenosine diphosphate ribose; HA, hyaluronate; BCR, B cell Ag receptor; suggesting that CD38 might function as a BCR coreceptor. Inter- ARAP, Ag receptor-associated protein. estingly, CD38-mediated signaling and coreceptor activity in this

Copyright © 1999 by The American Association of Immunologists 0022-1767/99/$02.00 2694 CD38 SIGNALING OCCURS INDEPENDENTLY OF cADPR GENERATION system was found to be completely dependent upon expression of clones. For example, 10 independently selected WILD-CD38 clones were a functional BCR complex (20). Surprisingly, a large truncation expanded in log phase for 1 wk. Equal numbers of these 10 clones were of the cytoplasmic tail of CD38 did not noticeably depress then mixed together, and 30 identical aliquots of the mixed WILD-CD38 cells were frozen (Ϫ80°C) in 10% DMSO/90% FBS. CD38-mediated signal transduction, suggesting that the cytoplas- mic tail of CD38 might be unnecessary for anti-CD38-mediated FACS staining of mixed mutant clones signaling (20). An aliquot of all the mutant mixed clones was thawed, grown in tissue In this report we have attempted to reconcile the signaling ca- culture, and then analyzed for CD38 expression on a FACS Calibur (Bec- pacity of murine CD38 with its enzymatic properties. We have ton Dickinson, Mountain View, CA) after staining. Cells were first blocked performed site-directed mutagenesis to produce mutated forms of with 1 ␮g of purified Fc block (24G2, PharMingen, San Diego, CA) and CD38 and have examined the consequences of these mutations on then stained on ice with either protein A-purified rabbit anti-mouse CD38 or purified normal rabbit IgGs. The cells were washed, stained with rat CD38-mediated signal transduction and enzyme activity in our anti-rabbit FITC (Zymed, South San Francisco, CA), and analyzed. A20 B lymphoma model system. The results are presented and discussed in the context of the predicted structure for CD38. Preparation of cross-linked Ab conjugates Cross-linked Abs were prepared as previously described (20). Briefly, 2 mg Materials and Methods of the purified rabbit anti-CD38 Ab and 8 mg of purified normal rabbit Cell lines, culture conditions, and Abs IgGs were mixed per 1.0 ml of recombinant protein A beads (Zymed). To prepare the substimulatory anti-CD38 cross-linked beads, 200 ␮g of anti- The A20/2J cell line was obtained from American Type Culture Collection CD38 Ab plus 10 mg of normal rabbit IgGs were incubated per 1.0 ml of (Manassas, VA), and the cytokine-dependent cell line HT-2 was a gift from protein A beads. To prepare the substimulatory anti-Ig beads, 4 ␮gof Dr. David Woodland (St. Jude Children’s Research Hospital, Memphis, rabbit anti-mouse IgG (Cappel) plus 10 mg of normal rabbit IgGs were Downloaded from TN). The cells were maintained as previously described (20). Polyclonal incubated per 1.0 ml of protein A beads. To prepare the cocross-linked rabbit anti-mouse CD38 antiserum was prepared by immunizing rabbits Abs, 200 ␮g of anti-CD38, 4 ␮g of anti-IgG, and 10 mg of normal rabbit with purified recombinant soluble CD38 (20). The IgG fraction of the IgG were incubated per 1.0 ml of protein A beads. All the Ab/bead com- antiserum was subsequently concentrated by ammonium sulfate purifica- binations were incubated, with rocking, at room temperature for at least 1 h tion, absorbed on the CD38-negative A20 B cell line, and then protein A and then washed four times in medium to remove unbound Ab. Freshly purified. Approximately 13% of the protein A-purified rabbit anti-CD38 prepared batches of cross-linked Ab were used in each experiment and

was specific for CD38 as measured using soluble CD38 in an ELISA. The were tested on the control CD38-negative A20 clone (NeoR). In no case http://www.jimmunol.org/ same procedure was used to purify normal rabbit IgGs from normal rabbit did the cross-linked anti-CD38 beads induce the CD38-negative cells to serum (Colorado Serum Co., Denver, CO). Affinity-purified rabbit anti- produce IL-2. Ј mouse IgG and F(ab )2 goat anti-mouse IgG were purchased from Cappel (West Chester, PA). Stimulation of CD38 mutant transfectants and IL-2 quantification Construction of CD38 mutant cDNAs An aliquot of each of the mutant mixed clones was thawed on day 0 and The wild-type CD38 expression vector (CD38-pME18S/neo) contains the put into tissue culture at 2 ϫ 105 cell/ml. The cells were maintained in log full-length coding region of the murine CD38 cDNA cloned downstream of ϫ 5 ␣ phase growth for the next 2 days and were expanded at 5 10 cells/ml the SR promoter in the pME18S/neo expression vector, as previously in fresh medium on day 3. On day 4 the cells were counted, assessed for described (3). All the different CD38 mutant constructs were generated by viability, washed, and used in experiments on that day. Triplicate cultures by guest on September 27, 2021 PCR using the primers listed below (cloning sites are underlined, and the of 2 ϫ 105 mixed clone cells/well were plated in a 200-␮l volume in altered nucleotides that correspond to the replacement amino acids codons complete medium with the appropriate stimulus. The cells were stimulated are indicated in lower case italics): 5Ј-␮ATG, 5Ј-(CCC GAA TTC ATG Ј ␮ with soluble affinity-purified F(ab )2 goat anti-mouse IgG (25–50 g/ml; aag gtg aag ATC GGT CTC GGA GTG GGT CTC CTG G)-3Ј;5Ј- ␮ ␮ Ј Cappel), PMA (5 ng/ml), and A23187 (1/80,000 dilution of 1 M) or with TFR-I, 5 -(tg gtc atc ttc ttc ctc ata ggc ttc tcc gga AGG CCG CGC TCA a 2.5-␮l packed bead volume (25 ␮l of 10% bead/medium slurry) of the CTC CTG GTG TGG)-3Ј;5Ј-␮TFR-II, 5Ј-(CCC GAA TTC atg aag gtc Ј Ј various Ab/protein A conjugates described above. Cells were stimulated aag cta tgc ttc gca gcc ata gcc ttg gtc atc ttc ttc ctc ata)-3 ;5-E150L/ for 8 h, and supernatants were removed and frozen at Ϫ80°C. To test for D151V, 5Ј-(T ACT TGG ATC CAG GGA AAG AT G TTC ACC CT G Ј Ј Ј the presence of IL-2, supernatants were thawed and used in bioassays with ctg gtc ACC CTG C)-3 ;5- E150Q/D151N, 5 -(T ACT TGG ATC CAG the IL-2-dependent cell line, HT-2. One unit of IL-2 is defined as the GGA AAG AT G TTC ACC CT G cag aac ACC CTG C)-3Ј;5Ј-CD38wt, Ј Ј amount of IL-2 required to induce half-maximal growth (as measured by 5 -(GGG GAA TTC ATG GCT AAC TAT GAA TTT AGC CAG G)-3 ; Alomar Blue, Accumed, Cleveland, OH) of 5 ϫ 103 HT-2 cells. The Alo- 3Ј-XbaI, 5Ј-(CCC TCT AGA CCA GAT CCT TCA CGT ATT AAG TCT Ј Ј Ј mar Blue assay used here could reproducibly detect IL-2 levels as low as ACA CG)-3 ;3-cys, 5 -(CC CTG GAT CCA AGT ATA TTG ATG GGC 0.2 U/ml (20). The concentration of IL-2 in the supernatants was calculated CAG GTG TTT GGA TTT GCT CCA AAA GAG AGT CTT GTT ctt using purified recombinant murine IL-2 (20 U/ml; DNAX, Palo Alto, CA) TGG TAT GG)-3Ј; and 3Ј-hyaluronate (3Ј-HA), 5Ј-(C CTG GAT Ј as the standard on all plates, and the results were expressed as the CCAAGT ATA TTG ATG GGC CAG GTG tgt GGA gct GCT CC)-3 . mean Ϯ SD. The appropriate PCR reaction products were then isolated, purified, and cloned into the PME18S/neo expression vector. The cloned products were Preparation of cell fractions sequenced in both directions to ascertain that the appropriate mutation was introduced and that no polymerase or cloning errors were present in the rest Mixed clone aliquots were thawed, expanded in tissue culture to 2 ϫ 108 of the molecule. cells over a period of 7 days, washed twice by centrifugation at 1000 ϫ g in PBS, and then frozen as a pellet at Ϫ80°C until use. To prepare the cell Generation of CD38 mutant A20 stable transfectants fractions, the pellets were diluted in 1 ml of a 50-mM potassium phosphate buffer, pH 6.8, and homogenized with a glass-glass Dounce potter. The CD38-negative A20 B lymphoma cells were transfected with the different homogenates were then centrifuged at 100,000 ϫ g for 60 min, and the CD38 expression constructs and then selected in G418 (Geneticin, Life pellets were carefully resuspended (potter) in the same volume of buffer for Technologies, Grand Island, NY) as previously described (20). The G418- enzyme activity testing. The protein concentration was determined with the resistant clones were grown out and analyzed for the expression of CD38 bicinchoninic acid assay reagent (Pierce, Rockford, IL) using BSA as a and surface Ig by immunostaining (described below) using polyclonal rab- Ј standard. bit anti-mouse CD38 and an R-phycoerythrin-conjugated goat F(ab )2 anti- mouse Ig (Southern Biotechnology Associates, Birmingham, AL). Ten Enzyme assays CD38ϩ, BCRϩ clones for each of the different CD38 mutant or wild-type transfectants were chosen and single cell subcloned. After subcloning, all The catalytic activity of CD38 was determined in the various cell fractions the clones were further selected for inducible IL-2 production after stim- by two different radiometric methods. The first one determines, under sat- 2ϩ Ј 14 14 ulation with both PMA and Ca ionophore or with F(ab )2 anti-Ig urating conditions, the release of [ C]nicotinamide from [ C-nicotin- (Cappel). amide] NADϩ and yields the total NADϩ glycohydrolase and ADP-ribosyl Mixed clones were then generated for cells expressing the recombinant cyclase activities. The second, by use of [14C-adenosine]NADϩ, measures wild-type CD38 (WILD-CD38) and for each of the different mutant CD38 the formation of [14C]ADPR and [14C]cADPR and yields the two activities The Journal of Immunology 2695 Downloaded from

FIGURE 1. Schematic of CD38 mutants generated by PCR. Mutations were generated in murine CD38 by PCR as described in detail in Materials and Methods. The sequence for WILD-CD38 is shown with the complete cytoplasmic tail (amino acids 1–23), the complete transmembrane domain (amino acids 24–45; shaded area), and a portion of the extracellular domain from amino acids 123–151. †, The two putative HA binding sites in murine CD38 (marking the canonical consensus residues within binding motif); §, Cys123; छ, Glu150 and Asp151. The various mutants are diagrammed schematically with To produce ␮ATG-CD38, the cytoplasmic tail of CD38 was replaced with the signaling inert mouse IgM BCR cytoplasmic .(ء) replacement amino acids tail, and the transmembrane domain and extracellular domain of CD38 were left intact. The ␮TFR-CD38 mutant contains the cytoplasmic tail from the IgM http://www.jimmunol.org/ BCR, the transmembrane domain from the mouse (31), and the wild-type CD38 extracellular domain. The HA-CD38 mutant eliminates all the putative HA binding sites in murine CD38 (28), one of which is located in the cytoplasmic tail and the other of which is located in the extracellular domain. To eliminate the cytoplasmic tail HA site, the cytoplasmic tail was truncated, the transmembrane domain was left intact, and two of the three canonical lysine residues in the putative extracellular HA site (amino acids 131 and 133) were mutated to serine and threonine, respectively. The extracellular enzymatic domain mutants were generated with a CD38 wild-type cytoplasmic tail and transmembrane domain and point mutations within the extracellular domain at positions 123, 150, and/or 151. These mutants are designated by their respective amino acid replacements.

separately. Alternatively, when increased sensitivity was needed, [3H- radioactivity associated with the different peaks was integrated, and the by guest on September 27, 2021 adenine]NADϩ was used to generate [3H]ADPR and [3H]cADPR. [Car- cADPR/ADPR ratio was calculated (27). bonyl-14C]nicotinamide adenine dinucleotide (35 mCi/mmol) was pur- chased from Amersham (Aylesbury, U.K.). [Adenosine-U-14C]NADϩ Calculation of adjusted enzyme activity (604 mCi/mmol) and [adenine-2,8-3H]NADϩ (4 Ci/mmol) were purchased To normalize the enzyme activity of the cell homogenates, the enzyme from New England Nuclear (Boston, MA). activities were multiplied by a correction factor that compensated for the Nicotinamide-releasing activity total protein per cell and the amount of CD38 expressed per cell. The amount of protein per cell was determined by a Bradford assay (Bio-Rad, Cell fractions were added to an assay mixture containing [carbonyl-14C]ni- Hercules CA), and the mean concentration was calculated from three in- cotinamide adenine dinucleotide (2.5 ϫ 105 dpm) and NADϩ (500 ␮M) in dependent experiments. The amount of CD38 protein expressed by each a 50-mM potassium phosphate buffer, pH 6.8 (250-␮l final volume), and cell was quantitated by FACS. The mean fluorescence intensity was de- incubated at 37°C. At given times, 100-␮l aliquots were withdrawn, and termined for individual mutant cells and was then normalized to WILD- after acidification with 4 ␮l of 50% TCA, the reaction products were an- CD38, which was set at a value of 1.0 arbitrary CD38 unit/cell. The FACS alyzed as described below. Each run (two time points; reaction progress, staining experiments were performed three times, and the average arbitrary Ͻ30%) was repeated at least three times with different time points and CD38 units per cell for each of the mutants were calculated. amounts of cell fractions.

ϩ Results NAD glycohydrolase and ADP-ribosyl cyclase activity Generation and characterization of transfectants expressing Cell fractions were added to an assay mixture containing either [adenosine- mutated forms of CD38 U-14C]NADϩ (5.0 ϫ 105 dpm) and NADϩ (final concentration, 25 ␮M) or [adenine-2,8-3H]NADϩ (5 ϫ 106 dpm) and NADϩ (final concentration, 5 Using A20 B lymphoma cells that had been stably transfected with ␮M) in a 50-mM potassium phosphate buffer, pH 6.8 (500 ␮l final vol- murine CD38, we previously demonstrated that signaling through ume), and incubated at 37°C. At given time points, 100-␮l aliquots were CD38 was unimpaired by truncation of the cytoplasmic tail to four withdrawn, and the reaction product was analyzed as described below. amino acids, suggesting that the cytoplasmic tail of CD38 may not Analysis of the reaction products by HPLC be necessary for signal transduction (20). Three other domains in the CD38 molecule have been hypothesized to play a role in sig- Product analysis was performed on TCA-precipitated aliquots of the reac- naling; these include the transmembrane domain, the putative HA tion medium (centrifuged at 10,000 ϫ g for 10 min at 4°C on Microcon-10 microconcentraters (Amicon, Danvers, MA)) using a Waters HPLC system binding sites (HA sites) (28), and the extracellular enzymatic ac- (Waters Associates, Milford, MA). Chromatography was conducted on a tive site(s) (1, 13, 16). To test which of these domains is required ϫ ␮ 300 3.9-mm Bondapak C18 column (Waters) operated at ambient tem- for anti-CD38-mediated signal transduction in B cells, we gener- perature at a flow rate of 1 ml/min. The compounds were eluted isocrati- ated A20 clones expressing CD38 molecules that had been spe- cally with a 10-mM ammonium phosphate buffer, pH 5.5, containing 1.2% (v/v) acetonitrile and were detected by their UV absorbance at 260 nm and cifically altered by site-directed mutagenesis in each of the regions by radiodetection (Flo-one, Packard-Radiometric Instruments, Meriden, described above. As diagrammed in Fig. 1, we generated two chi- CT). The reaction products were identified by coelution with standards, the meric forms of CD38 in which the cytoplasmic tail (␮ATG-CD38) 2696 CD38 SIGNALING OCCURS INDEPENDENTLY OF cADPR GENERATION or the cytoplasmic tail and transmembrane domain (␮TFR-CD38) of CD38 were completely replaced. The CD38 cytoplasmic tail in both chimeric molecules was replaced with the three amino acid sequence of the IgM BCR cytoplasmic tail (KVK), as this se- quence has been demonstrated to be inert in B cell signaling assays (29, 30). Furthermore, in ␮TFR-CD38, the transmembrane domain of CD38 was replaced with the transmembrane domain from an- other type II protein, the transferrin receptor (31) (Fig. 1). In addition to these chimeric CD38 molecules, we introduced several independent point mutations in the extracellular domain of CD38. Amino acid codons were altered by site-directed mutagen- esis at positions 123, 150, and 151, as these residues were previ- ously shown to be important for CD38-mediated enzymatic activ- ity (32, 33). These extracellular mutant molecules are referred to by their amino acid replacements (i.e., C123K-CD38, E150L/ D151V-CD38, and E150Q/D151N-CD38). Finally, we truncated the cytoplasmic tail of CD38 to three amino acids and introduced two point mutations into the only extracellular HA binding site of murine CD38 (28) to generate a CD38 mutant molecule that does Downloaded from not contain any HA binding sites (HA-CD38). FIGURE 2. CD38 expression levels on mixed mutant clones. Control cDNAs encoding each of the different forms of CD38 as well as clone (NeoR, CD38 negative) and mixed clones expressing wild-type a cDNA encoding the wild-type CD38 molecule (WILD-CD38) CD38 (WILD-CD38) or various mutant forms of CD38 were stained with were generated by PCR, cloned into the PME18S/neo mammalian either polyclonal rabbit anti-mouse CD38 (filled histogram) or purified expression vector (3), and then stably expressed in the parental rabbit IgGs (open histogram), washed, and then stained with rat anti-rabbit

CD38-negative A20 lymphoma cells (20). Multiple independent IgG-FITC. The mean fluorescence intensity (MFI; top right corner) for http://www.jimmunol.org/ CD38-expressing clones were generated from cells transfected each of the mutant clones was calculated. with WILD-CD38 or with each of the different CD38 mutant cDNAs and were tested to ensure that they inducibly produced IL-2 after treatment with PMA and Ca2ϩ ionophore treatment as adenine dinucleotide. The reaction products were then analyzed by well as after stimulation with anti-Ig reagents. At least 10 inde- radiometric HPLC. The steady state enzyme activity for the ho- pendent clones of cells expressing a particular transfected form of mogenates generated from each of the CD38 mutant cell lines was CD38 were then equally mixed (as described in Materials and expressed as nanomoles of product produced per minute per mil- Methods) to eliminate any clone biasing that might arise in the ligrams of total protein, and the relative enzyme activities of each selection process. Thus, the effect of each of the different CD38 of the homogenates was calculated as a percentage of the activity by guest on September 27, 2021 mutations could be assessed on a known, reproducible, mixed pop- of WILD-CD38 (Table I). ulation of transfected cells. Replacement of either the cytoplasmic tail (␮ATG-CD38) or the In initial experiments using the mixed clones, the CD38 expres- cytoplasmic tail and transmembrane domain (␮TFR-CD38) of sion levels were quantitated by FACS using purified polyanti- CD38 did not impair the ability of CD38 to catalyze the release of CD38 Ab or normal rabbit IgG as a control. The resultant histo- nicotinamide from NADϩ (Table I). The elimination of the two grams are shown in Fig. 2. The CD38 expression levels on the putative HA binding sites (HA-CD38) had a small, but measur- mixed clones were evaluated with the polyclonal anti-CD38 and able, effect and reduced the catalytic activity of CD38 by approx- were found to be nearly equivalent, with no more than a twofold imately 50% (Table I). In contrast, and in agreement with the find- difference in the mean fluorescence intensity (in upper right corner ings of Grimaldi et al. (32), selected point mutations at positions of each histogram) between WILD-CD38 and any of the mutant 150 and 151 greatly reduced the nicotinamide-releasing activity of clones. CD38. In fact, the nicotinamide-releasing activity of the homog- enates expressing E150L/D151V-CD38 was at least 35 times Analysis of enzyme function in CD38 mutant transfectants lower than that of those expressing WILD-CD38. This mutant was CD38 is a multifunctional enzyme that is able to catalyze the con- not catalytically inactive however, as the nicotinamide-releasing version of NADϩ into nicotinamide (nicotinamide-releasing activ- activity of the homogenate generated from the E150L/D151V- ity), ADPR (NADϩ glycohydrolase activity), and cADPR (ADP- expressing cells was approximately 10-fold higher than that of the ribosyl cyclase activity; Fig. 3). To assess the role of CD38’s homogenates of the nontransfected CD38-negative A20 control enzymatic activities on signal transduction, we first evaluated the cells (NeoR). Interestingly, a more conservative set of replacement impact of the different structural mutations described above on amino acids at positions 150 and 151 (E150Q/D151N) did not CD38’s various enzyme functions. To do this, the mixed clones dramatically alter the catalytic activity of CD38 compared with the expressing the various mutant CD38 molecules were homogenized CD38-WILD protein (70% of normal activity). Finally, the re- in the absence of detergent to generate cell homogenates that pre- placement of the cysteine residue at amino acid position 123 with served the integrity of the cellular membranes. The protein con- a lysine residue (C123K-CD38) increased the CD38-mediated nic- centration and enzyme activity of the homogenates were then de- otinamide-releasing activity approximately 2-fold over the activity termined for each of the mutants. The nicotinamide-releasing of WILD-CD38 (Table I). activity (Fig. 3), which is basically a determination of the total The nicotinamide-releasing activity of the cell homogenates was catalytic activity of CD38 (combined NADϩ glycohydrolase, normalized for the amount of total protein and CD38 protein ex- ADP-ribosyl cyclase, transglycosidation, and cADPR hydrolase pressed per cell (correction factor, Table I, column 2). The amount activities), was assayed under saturating condition by measuring of CD38 expressed by each cell was represented by the mean flu- the release of [14C]nicotinamide from [carbonyl-14C]nicotinamide orescence intensity value, which was determined by staining the The Journal of Immunology 2697

FIGURE 3. Schematic of CD38 enzymatic activities. The extracellular domain of CD38 binds NADϩ, forms the Michaelis complex, and catalytically releases nicotinamide by cleaving the nicotinamide-ribose bond. Upon the release of nicotinamide, an oxocarbonium intermediate (ADPR*CD38) is formed that can be partitioned into three enzymatic reactions. If the intermediate is attacked by H2O, ADPR is formed through the NAD glycohydrolase reaction. If the intermediate is attacked by another nucleophile (i.e., nicotinic acid, X:), a new dinucleotide is formed through the transglycosidation reaction.

Alternatively, the intermediate can be cyclized to form cADPR through the ADP-ribosyl cyclase reaction. Finally, the cADPR can be hydrolyzed by CD38 Downloaded from to form ADPR through the cADPR hydrolase reaction. cells with polyclonal anti-CD38 Abs that contain multiple epitope (ADP-ribosyl cyclase activity) was individually measured (27), specificities (Fig. 2). The adjusted activities were expressed as and the ratio of cyclization to hydrolysis was calculated (i.e., the nanomoles of product produced per minute per arbitrary unit of cADPR/ADPR ratio; Table II). Using this assay, we found that CD38 (Table I, column 4). Even after the enzyme activity of each wild-type CD38 produced approximately 1000 times more ADPR http://www.jimmunol.org/ of the homogenates was adjusted to reflect the amount of CD38 than cADPR. This was not unexpected, as recombinant soluble present, the relative activity of each of the CD38 mutant molecules murine CD38 has been shown to produce exceedingly small quan- did not alter significantly compared with the activity of the CD38 tities of cADPR (13). The amount of cADPR produced by the wild-type protein (compare fold activity in column 2 to column 4 homogenates generated from the CD38 cytoplasmic tail mutant in Table I). Importantly, the greatly reduced catalytic activity of cell line (␮ATG-CD38) and the CD38 transmembrane domain mu- E150L/D151V-CD38 could not simply be explained by either re- tant cell line (␮TFR-CD38) was essentially equivalent to that pro- duced total cellular protein content or a drastic reduction of CD38 duced by WILD-CD38. The amount of cADPR produced by HA-

protein. CD38 and C123K-CD38 homogenates was increased over that by guest on September 27, 2021 Next, to test whether any of the mutant forms of CD38 were produced by WILD-CD38, but was still 20–50 times less than the altered in their ability to produce the calcium-mobilizing second amount of ADPR produced. Interestingly, the E150Q/D151N- messenger, cADPR, the cell homogenates were tested in assays CD38 homogenate produced much larger quantities of cADPR rel- using [14C-adenosine]NADϩ as a substrate. The formation of ative to ADPR. In contrast, no detectable cADPR could be mea- [14C]ADPR (NADϩ glycohydrolase activity) and [14C]cADPR sured in either the E150L/D151V-CD38 or the CD38 negative

Table I. Total catalytic activity measurements of CD38 mutants

Enzyme Activitya Correction Factorb Adjusted Enzyme Activity CD38 Mutant nmol/min mg protein nmol/min ͩ ͪ ͫͩ ͪ ϫ 105ͬ ͫͩ ͪ ϫ 104ͬ Mixed Clones mg protein Arb. Unit CD38 Arb. Unit CD38 WILD 6.26 Ϯ 0.54 (1.0)c 6.8 4.26 (1.0)c ␮ATG 6.94 Ϯ 0.60 (1.1) 5.4 3.75 (0.9) ␮TFR 7.74 Ϯ 0.67 (1.2) 10.5 8.13 (1.91) HA 3.18 Ϯ 0.24 (0.5) 12.1 3.18 (0.9) E150L/D151V 0.17 Ϯ 0.0272 (0.03)d 15.8 0.269 (0.06) E150Q/D151N 4.14 Ϯ 0.25 (0.7) 3.7 1.53 (0.4) C123K 12.98 Ϯ 0.79 (2.1) 4.1 5.32 (1.3) NeoR 0.018 Ϯ 0.0031 (0.003) ND ND

(n Ն 3) a The catalytic activity of each of the different CD38 mutants was measured, as described in Materials and Methods, under saturating conditions with [14C-nicotinamide]NADϩ (concentration in the assay, 500 ␮M). The enzyme activities of the mem- brane fraction are expressed as [14C]nicotinamide releasing activity (radiometric HPLC assay), which corresponds to the sum of the NADϩ glycohydrolase and ADP-ribosyl cyclase activities of the CD38 mutants. This assay could detect the transformation of NADϩ as low as 5 pmol/min/mg protein. Under these experimental conditions, the nucleotide pyrophosphatase activity of the cells, which was easily distinguishable from the nicotinamide releasing activity, was not high enough to deplete the NADϩ or to alter the kinetics of latter activity. b A correction factor was introduced to account for the amount of total protein and CD38 protein expressed per cell, allowing for calculation of an “adjusted” enzyme activity that was expressed as nmol product produced per min per Arbitrary Units CD38 protein. c The relative enzyme activity of each of the mutants was determined by setting the enzyme activity of the WILD-CD38 cells at 1.0 and then calculating the fold difference in activity between WILD-CD38 and the mutants. d When using a more sensitive assay at 20 ␮M NADϩ, the nicotinamide cleaving activity of this mutant was calculated to be even less than indicated here and was only 1.4% of the CD38ϩ wild type cells. ND, Not determined as the NeoR cells do not express any detectable CD38 protein. 2698 CD38 SIGNALING OCCURS INDEPENDENTLY OF cADPR GENERATION

Table II. NAD glycohydrolase and ADP-ribosyl cyclase activities of CD38 mutants

ADP-Ribosyl Cyclase Activityb Adjusted Cyclase Activityc ͓ ͔ a CD38 Mutant cADPR nmol/min nmol/min ͫͩ ͪ ϫ 102ͬ ͫͩ ͪ ϫ 106ͬ Mixed Clones ͓ADPR͔ mg protein Arb. Unit CD38 WILD 0.0012 0.75 (1.0)d 0.51 (1.0)d ␮ATG 0.0016 1.11 (1.5) 0.60 (1.2) ␮TFR 0.0026 2.01 (2.7) 2.11 (4.1) HA 0.0450 13.69 (18.3) 16.56 (32.5) E150L/D151V Ͻ0.0010e Ͻ0.017 (0.02) Ͻ0.023 (0.045) E150Q/D151N 0.3380 104.58 (139.4) 38.69 (75.9) C123K 0.0195 24.83 (33.1) 10.18 (20.0) NeoR Ͻ0.0010e Ͻ0.0018 (0.002) ND

a The formation of ADPR (NADϩ glycohydrolase activity) and cADPR (ADP-ribosyl cyclase activity) were determined for each of the CD38 mutants (radiometric HPLC assay) by incubation of the membrane fractions with [14C-adenosine] NADϩ (25 ␮M) as described in Materials and Methods. The relative ADP-ribosyl cyclase to NADϩ glycohydrolase activity can be accurately represented by the ratio (R) of the amount of cADPR to ADPR produced by the various CD38 mutants. b To determine the ADPR-ribosyl cyclase activity (i.e., the amount of cADPR produced per min per mg protein), the total activity (Table I) was multiplied by ((R)/(R ϩ 1)). c The adjusted cyclase activity was calculated using the correction factor described in Table I and is represented as nmol per min per Arbitrary Unit of CD38. d The relative cyclase activity for each of the mutants was determined as described in Table I. Downloaded from e In the case of NeoR and E150L/D151V, no cADPR was detected, indicating that neither extract had substantial ADP-ribosyl cyclase activity. Furthermore, more sensitive assay conditions using 10-fold higher radioactivity amounts of [3H-adenine] NADϩ (final concentration of NADϩ was 5 ␮M, see Materials and Methods) still did not reveal cADPR production by these extracts, even when Ͼ80% of the labeled NADϩ had been converted to products. Thus, the cADPR/ADPR ratio of E150L/D151V and NeoR is certainly no greater than WILD-CD38 (0.12) and may be even less than this amount. ND, Not determined as the NeoR cells do not express any detectable CD38 protein. Each of these measurements were repeated in at least two independent experiments. http://www.jimmunol.org/ (NeoR) homogenates. To ensure that we were not at the limits of enzymatic properties. This panel included CD38 molecules with wild- detection of our assay, the experiments were repeated using 10- type enzymatic activities (␮ATG-CD38, ␮TFR-CD38), a CD38 mol- ϩ fold more (5.0 ϫ 106 dpm) labeled [adenine-2,8-3H]NAD . How- ecule with greatly reduced catalytic activity (E150L/D151V), and ever, even when NeoR and E150L/D151V had catabolized Ͼ80% molecules with increased ADP-ribosyl cyclase activity (E150Q/ ϩ of the 3H-labeled NAD to ADPR, no [3H]cADPR could be de- D151N, C123K, and HA-CD38). We next tested the effects of these tected (not shown), demonstrating that the cADPR/ADPR ratio for mutations on CD38-mediated signaling. To be able to directly com- NeoR and E150L/D151N is certainly no higher than that for wild- pare the anti-CD38-mediated response of all of the CD38-transfected type CD38 and, if anything, is somewhat lower. Importantly, the cell lines generated, all the different mutant mixed clones were stim- ϩ by guest on September 27, 2021 very low residual NAD -converting activity of E150L/D151V- ulated with an optimal dose of cross-linked polyanti-CD38 for 6–8 h. CD38 that was observed in the nicotinamide release assay (Table ϩ The supernatants were collected, and IL-2 production was measured I) must be primarily due to NAD glycohydrolase activity and by bioassay. The experiment was repeated seven times, and in each cannot be attributed solely to ADP-ribosyl cyclase activity. experiment the anti-CD38-mediated IL-2 response of WILD-CD38 The ADP-ribosyl cyclase activity was then calculated at maxi- was set at 1.0, and the amount of IL-2 produced by each of the cell mum velocity (Vmax) for each of the homogenates generated from lines expressing mutated forms of CD38 was calculated relative to the the CD38-expressing cell lines and is shown in the middle column wild-type response. The IL-2 response from each of the mutants from of Table II. The cyclase activity of the CD38 cytoplasmic and all seven experiments is shown as a scatter plot in Fig. 4. As expected, transmembrane mutant homogenates (␮ATG-CD38 and ␮TFR- none of the mixed clones produced IL-2 constitutively, and all of the CD38) was approximately equivalent to that of WILD-CD38. The ϩ mixed clones produced similar amounts of IL-2 after PMA and Ca2 cyclase activities of HA-CD38, E150Q/D151N-CD38, and ionophore treatment (not shown). Extending our previously published C123K-CD38 were very elevated (20- to 140-fold increased) rel- results (20), we found that complete replacement of the cytoplasmic ative to that of WILD-CD38. In contrast, the cyclase activities of tail of CD38 (␮ATG-CD38) had no deleterious effect on CD38-me- the homogenates prepared from the CD38 negative NeoR clone ␮ and the E150L/D151V-CD38 clone were at least 40- to 400-fold diated signaling. In fact, the response of the ATG-CD38-expressing lower than that of the wild-type enzyme. Given that we were un- clone to anti-CD38 was consistently improved over that of WILD- able to measure cADPR production by these homogenates, the CD38 (Fig. 4). When both the cytoplasmic tail and the transmem- ␮ actual cyclase activity is probably significantly lower than the min- brane domain of CD38 were replaced ( TFR-CD38), there was again imal calculated activity given in Table II. Finally, the cyclase ac- no effect on anti-CD38-mediated IL-2 induction, with the average tivity in each of the homogenates was normalized for CD38 and response being equivalent to that of WILD-CD38 (Fig. 4). When the total protein content (Table II, third column). Again, even after putative HA binding sites of CD38 were eliminated, the average anti- these adjustments, the cyclase activity of the different CD38 mu- CD38-mediated response was reduced to 75% that of WILD-CD38; tants did not change substantially relative to that of the CD38 however, in about half the experiments the response was equivalent to wild-type control (see fold differences in parentheses in columns 2 that of WILD-CD38 (Fig. 4). Although the HA-CD38 IL-2 response and 3 in Table II). was somewhat lower than the wild-type response, it was easily mea- surable and was at least 10-fold increased over the IL-2 detection The cytoplasmic tail, transmembrane domain, and HA binding threshold (Ͻ0.2 U/ml) and the CD38-negative control clone (NeoR) sites can be replaced without impairing CD38-mediated signal in all experiments. Thus, from these experiments it appeared as transduction though the cytoplasmic tail, the transmembrane domain, and the HA From the previous experiments it was clear that we had produced a binding sites of CD38 could be completely replaced without dramat- panel of cell lines that expressed CD38 mutant molecules with varied ically altering CD38 signaling. The Journal of Immunology 2699

FIGURE 5. CD38-mediated costimulation: assessment of CD38 mu- tants. Triplicate cultures of 2 ϫ 105 cells from each of the different mutants were stimulated with 10 ng/well cross-linked rabbit anti-mouse IgG (open bars), 500 ng/well cross-linked poly anti-CD38 (striped bars), or cocross- Downloaded from linked anti-Ig (10 ng/well) and anti-CD38 (500 ng/well; solid bars). IL-2 production was measured by bioassay after 8-h stimulation. The results are FIGURE 4. Anti-CD38 stimulation of CD38 mutants: a comparison of shown as the mean Ϯ SD and are representative of at least five independent 5 the IL-2 responses. Triplicate cultures of 2 ϫ 10 cells from all the different experiments. mutant mixed clones, from WILD-CD38, and from CD38Ϫ NeoR cells were stimulated with 5 ␮g/well cross-linked polyanti-CD38. Supernatants were harvested after 8 h, and IL-2 production was assessed. The amount of normal B cells (20, 34). Since some of the mutations in the extra- http://www.jimmunol.org/ IL-2 produced by the WILD-CD38 mixed clone in each experiment was set cellular enzymatic domain of CD38 greatly decreased CD38- at 1.0 (long bar), and the amount of IL-2 produced by each of the mutants was then calculated relative to the response made by WILD-CD38 in that mediated signaling, we next evaluated CD38-mediated costimula- experiment. This was repeated seven times for each of the clones, and each tory activity in the cells expressing mutant CD38 molecules. of the filled squares in the scatter plot represents the results from a single Mixed clones expressing ␮TFR-C38, HA-CD38, the various ex- experiment. The average response in these seven experiments for each tracellular domain point mutations, and WILD-CD38 were stimu- mutant was calculated and is represented as a small bar in each of the lanes. lated for 8 h with suboptimal doses of cross-linked anti-Ig alone The average response for ␮ATG-CD38 was 1.4, with a range of 0.75–2.70. (Fig. 5, open bars), cross-linked anti-CD38 alone (striped bars), or The average response of ␮TFR-CD38 was 1.2, with a range of 0.6–2.1. For cocross-linked anti-CD38 and anti-Ig (filled dark bars). IL-2 pro- HA-CD38, the average IL-2 response was 0.7-fold that of WILD-CD38, duction was measured and is shown in Fig. 5. In all cases, no by guest on September 27, 2021 and the range was 0.6–1.0. The average response of E150L/D151V-CD38 detectable response was made to suboptimal cross-linking of either was 0.7-fold, and the range was 0.1–1.3. The average IL-2 response of the BCR alone or CD38 alone. Interestingly, all the mixed clones E150Q/D151N was 0.2-fold, and the range was 0.1–0.4. Finally, the av- erage IL-2 response of C123K-CD38 was 0.2-fold that of WILD-CD38, did make a synergistic IL-2 response when CD38 and the BCR with a range from 0.1–0.3. In all experiments the CD38-negative NeoR were coligated, but there were differences among the various cell cells did not produce any detectable IL-2 (Ͻ0.2 U/ml) in response to anti- lines in the strength of the IL-2 response. There was no appreciable CD38 stimulation. change in the synergistic response of the ␮ATG-CD38 (not shown), ␮TFR-CD38, or HA-CD38 mixed clones compared with that of WILD-CD38 (Fig. 4). In contrast, cells expressing the other Mutations in the extracellular domain of CD38 greatly impair CD38 extracellular domain mutant molecules made an IL-2 CD38-mediated signal transduction response that was reduced 50–75% compared with that of Although none of the previously described mutations in CD38 ap- WILD-CD38. preciably affected signal transduction, an examination of cells ex- pressing CD38 extracellular mutant molecules revealed striking Anti-CD38 signaling in B cells can occur independently of defects in anti-CD38-mediated signal transduction. For example, enzyme activity the anti-CD38-induced IL-2 response of the C123K-CD38- and These data demonstrate that the sequences associated with the cy- E150Q/D151N-CD38-expressing cell lines was reduced by 80– toplasmic tail and transmembrane domain of CD38 are not re- 90% compared with that of cells expressing wild-type CD38 mol- quired for CD38-mediated signaling and coreceptor activity. Ad- ecules, with little variance in the response from experiment to ex- ditionally, the data indicate that selected point mutations in the periment. Thus, the reduced response of cells expressing these extracellular domain of CD38 can impair signaling, suggesting that CD38 extracellular domain mutants could not simply be attributed the extracellular domain is instrumental in CD38-mediated signal to the health or condition of the cells on any given day (Fig. 4). transduction. To determine whether there was any correlation be- Taken together, these experiments indicate that the extracellular tween the ability to mediate signal transduction and the ability to domain of CD38, rather than its membrane anchor or cytoplasmic produce nicotinamide (nicotinamide-releasing activity), ADPR tail, controls anti-CD38-mediated signal transduction in B (NADϩ glycohydrolase activity), or cADPR (ADP-ribosyl cyclase lymphocytes. activity), the relative signaling capacity of each of the mutants was plotted vs their relative enzyme capacities (Fig. 6). In each case, CD38-mediated coreceptor activity is regulated by its the signaling capacity and enzymatic activity of WILD-CD38 extracellular domain (filled circles) were set at 1.0. The signaling capacity and enzy- Previously we demonstrated that cocross-linking of the BCR and matic activity of the cells expressing the altered CD38 molecules CD38 can augment activation responses in both A20 cells and (filled squares) were then compared relative to those of wild-type 2700 CD38 SIGNALING OCCURS INDEPENDENTLY OF cADPR GENERATION

less well. Taken together, these data suggest that while the extra- cellular domain of CD38 is required for signal transduction, its nicotinamide-releasing activity, NADϩ glycohydrolase activity, and ADP-ribosyl cyclase activities are unlikely to be required for anti-CD38-mediated signaling in B cells. The implications of these findings for CD38-mediated signal transduction in B lymphocytes is discussed.

Discussion CD38 can catalyze the conversion of NADϩ into nicotinamide (nicotinamide-releasing activity), ADPR (NADϩ glycohydrolase activity), and, to a much lesser extent, cADPR (ADP-ribosyl cy- clase activity) (13, 14) (Fig. 3). CD38 can also hydrolyze cADPR into ADPR (cADPR hydrolase activity) (13) and can catalyze base exchange reactions (transglycosidation) between the nicotinamide group of NADϩ or NADPϩ and other acceptor compounds (35,

36) (Fig. 3). Interestingly, at least two of the products generated by Downloaded from the enzymatic activities of CD38 have been shown to regulate intracellular calcium levels (8, 35, 36), and it has been hypothe- sized that these products may control CD38-mediated signal trans- duction (1, 13, 16). To directly test this hypothesis, we introduced a number of mutations into the murine CD38 molecule and as-

sessed their impact on the enzymatic activities of CD38 and its http://www.jimmunol.org/ ability to mediate signal transduction. A comparison of the signal- ing capacity and enzymatic capabilities of cell lines expressing the mutant molecules relative to those expressing WILD-CD38 is pro- vided in Fig. 6. From the presented data it is clear that complete replacement of the cytoplasmic tail and transmembrane domain of CD38 does not alter signaling or any of the measured enzyme activities, suggest- ing that the cytoplasmic tail and transmembrane domains of CD38 are not necessary for any of its described functions. Instead, the by guest on September 27, 2021 enzyme activity and signaling properties of CD38 appear to be controlled by its extracellular domain, as independent point muta- tions in the extracellular domain of CD38 could alter both enzyme FIGURE 6. Comparison of CD38-mediated signal transduction to activity and anti-CD38-mediated signaling. For example, all the CD38 enzyme activities for CD38 mutant transfectants. CD38-mediated signaling was measured after anti-CD38 stimulation as described in Fig. 4, measured enzyme activities of one mutant, E150L/D151V-CD38, and the average response from seven separate experiments was determined. were dramatically reduced. In particular, the nicotinamide-releas- The average response for each of the mutant transfectants is represented as ing capacity of homogenates of cells expressing this mutant mol- a proportion of the WILD-CD38 IL-2 production, which was arbitrarily set ecule was only 1–3% the activity of homogenates of cells express- at 1.0 for each of the seven experiments. The nicotinamide-releasing ac- ing wild-type CD38. Since the release of nicotinamide from tivity (A) and ADP-ribosyl cyclase activity (C) of each of the mutants were NADϩ is the necessary first step for all the known enzymatic ac- ϩ calculated as shown in Tables I and II. The NAD glycohydrolase activity tivities of CD38, it is clear that this molecule is significantly cat- of the mutants (B) was calculated by multiplying the nicotinamide-releas- alytically impaired. However, no commensurate effect on signaling ing activity (Table I) by the percentage of ADPR produced (Table II). The was observed, as the anti-CD38-induced IL-2 response in cells different enzyme activities of each of the mutants are represented as a proportion of the activity in WILD-CD38 cells, which was set at 1.0 (see expressing E150L/D151V-CD38 was reduced less than twofold relative enzyme activities in Tables I and II). The relative enzyme activities compared with that in cells expressing wild-type CD38. Thus, the of each of the mutants (x-axis on all graphs) were then compared with their enzyme activities of CD38 do not appear to be absolutely required relative signaling capacities (y-axis on all graphs). to mediate anti-CD38 signaling in vitro. In agreement with this conclusion, soluble recombinant enzy- matically active CD38 was not sufficient to induce signaling in CD38. Interestingly, there was no correlation between the signal- A20 cells, even after addition of the stimulatory anti-CD38 Ab (not ing capacity of CD38 and its ability to produce nicotinamide, shown). Additionally, the direct addition of purified cADPR, cADPR, or ADPR. Most strikingly, cells expressing the catalyti- ADPR, or both metabolites to cultures of CD38-negative and ϩ cally impaired E150L/D151V-CD38 molecules made an easily de- CD38 A20 cells did not induce cytokine production (not shown). tectable IL-2 response after anti-CD38 stimulation. In fact, the Thus, not only is enzyme activity of CD38 insufficient to induce average response of this cell line was 65% that of wild-type cells, signaling, the data also suggest that the products produced via the with a range of responses that varied between 10–110% of wild- catalytic activity of CD38 do not induce cytokine production on type cells (Fig. 4). Additionally, cells expressing CD38 mutant their own or directly regulate anti-CD38-mediated signaling. This proteins that had increased cADPR production (E150Q/D151N conclusion was further strengthened when we compared the sig- and C123K) did not signal better than cells that were not capable naling capacity of the mutant CD38 molecules with their capacity of producing any cADPR (E150L/D151V) and, in fact, signaled to produce the different enzymatically generated products (Fig. 6). The Journal of Immunology 2701

For example, we found that the E150Q/D151N mutant was Ͼ100- fold better at producing cADPR than the wild-type enzyme, yet the signaling capacity of this mutant was reduced by Ͼ80%. In con- trast, the cyclase activity of E150L/D151V-CD38 was at least 50- fold lower than that of wild-type CD38, yet the signaling capacity of this mutant was reduced Ͻ2-fold. Interestingly, the mutants that made the most cADPR (C123K- FIGURE 7. Model for CD38-mediated signal transduction. We propose CD38 and E150Q/D151N-CD38) were also the most compromised that CD38 is associated extracellularly with an unknown signaling mole- in their ability to mediate signal transduction, suggesting that cule that we have termed ARAP. In this model, Ab binding, ligand binding, cADPR production might negatively regulate CD38 signaling. or NADϩ binding all cause an activating conformational change in the However, we believe that this possibility is unlikely. First, one ectodomain of CD38 that enables ARAP to leave CD38 and enter the might expect that CD38 mutants that make less cADPR than the Ag-receptor complex. ARAP, in turn, activates the BCR signaling network. wild-type enzyme should signal better than the wild-type protein. The active conformation of CD38 is shown as a dimer in this model, as the The E150L/D151V mutant, which does not make any detectable structurally related Aplysia enzyme was crystallized as a dimer (37). cADPR, can mediate signal transduction, but does not signal more efficiently than the wild-type CD38. Additionally, only a 2-fold other known enzyme activities of CD38, such as its base exchange difference in cyclase activity was seen when HA-CD38 and reaction or cADPR hydrolase activity (Fig. 3), might produce or C123K-CD38 were compared, yet there was a striking difference destroy metabolites that regulate CD38-mediated signaling. Al- in the signaling capacity of these two mutants. In contrast, an 800- though this hypothesis might be used to explain our results, we Downloaded from fold difference in cyclase production between HA-CD38 and believe the E150L/D151V-CD38 mutant argues against this pos- E150L/D151V resulted in only a 15% difference in signaling. sibility. For example, we were unable to detect cADPR hydrolysis Thus, if there is a negative correlation between cADPR production by E150L/D151V-CD38, demonstrating that this enzyme activity and signaling, there would have to be a very sharply defined border was also impaired in the mutant (not shown). Additionally, since between the amount of cADPR produced and the signaling capac- the base exchange reaction requires the release of nicotinamide ϩ ity of the cells. from NAD , and the nicotinamide-releasing capacity of E150L/ http://www.jimmunol.org/ To eliminate the concern that the polyclonal anti-CD38 Ab used D151V-CD38 was only 1–3% that of wild-type CD38, it is ex- to initiate CD38 signaling might alter the enzymatic activities of ceedingly unlikely that this mutant is able to produce any other CD38 or may not be able to bind to the mutant molecules effi- signaling metabolites. ciently, we repeated the enzyme measurements in the presence and An alternate model is shown in Fig. 7, which hypothesizes that the absence of polyclonal anti-CD38, but found no differences in the structure or conformation of CD38’s extracellular domain reg- reaction rates or the measured cADPR/ADPR ratios (not shown). ulates CD38-mediated signal transduction via CD38’s association Additionally, as shown by FACS analysis (Fig. 2), all the mutant with other signaling molecules. We have previously demonstrated CD38 molecules were recognized by the polyclonal anti-CD38 that anti-CD38-mediated signal transduction in B cells is critically Abs used in the stimulation experiments. Finally, Biacore mea- dependent on expression of the BCR. Additionally, the data sug- by guest on September 27, 2021 surements of the affinity of the polyclonal anti-CD38 for the sol- gested that the interaction between CD38 and BCR must occur uble version of one of the most repressed signaling mutants, extracellularly via an intermediary protein that we have called Ag E150Q/D151N-CD38, was identical with the affinity measured for receptor-associated protein (ARAP) (20). Thus, if the conforma- WILD-CD38, demonstrating that the mutant could be efficiently tion of the CD38 ectodomain was altered such that ARAP and bound by the polyclonal anti-CD38 Ab (not shown). Thus, we CD38 could not efficiently communicate, signal transduction were left with the conclusion that anti-CD38-mediated signaling in would be predicted to be greatly reduced. In agreement with this B lymphocytes can occur independently of CD38 enzyme activity possibility, we know that the two mutations that most affected or enzymatically generated products. signaling (C123K and E150Q/D151N) are predicted to alter the This conclusion, at first glance, is in apparent contradiction to conformation of the CD38 enzyme active site. The C123K muta- our own earlier published observation that addition of purified tion is within the solvent-exposed hinge region of CD38 and is cADPR to activated B cells enhanced their proliferative capacity believed to change the size of the CD38 active site by changing the (13). However, upon closer examination, we believe that the two conformation of the extracellular domain (37). In contrast, the res- findings are not mutually exclusive. In the earlier experiments we idues at positions 150 and 151 are predicted to be localized spa- found that B cells that had been induced to enter cell cycle by a tially within the buried hydrophobic carboxyl-terminal pocket that number of different stimuli, including LPS, anti-CD40, anti-Ig, and has been hypothesized to be the active site of this family of en- anti-CD38 (13), were all further responsive to the addition of pu- zymes (37). Interestingly, the only difference between the catalyt- rified cADPR. Thus, cADPR, perhaps through its calcium-mobi- ically dead E150L/D151V CD38 mutant and the altered enzyme lizing capacity, acted as a growth enhancer for all activated B cells. profile E150Q/D151N-CD38 mutant was the choice of amino acid Since CD38 can produce cADPR, this might argue that the enzy- replacements at residues 150 and 151. Thus, it appears that these matic activity of CD38 can enhance the proliferation of replicating residues are involved in catalysis of the oxocarbonium intermedi- B cells. However, this does not imply that cADPR is required for ate (Fig. 3). Finally, we know that mutations within the active site CD38 to initiate its own signal transduction pathway. Thus, the of the enzyme also alter global protein conformation, as at least enzymatic activities of CD38 may play an important role in pro- one of the mutants, E150L/D151V-CD38, is no longer recognized ducing metabolites, such as cADPR, that can alter cell signaling; by the mAb anti-CD38 Ab NIMR-5 (18) (data not shown). Thus, however, these same metabolites do not appear to be directly re- it is at least possible that some of the mutations that altered CD38 quired to induce anti-CD38-mediated activation and cytokine pro- conformation also altered the capacity of CD38 to associate effi- duction in B lymphocytes. ciently with ARAP, and thus diminished anti-CD38-mediated How, then, can we explain the requirement for the extracellular signaling. domain of CD38 to mediate signal transduction if cADPR, ADPR, Extracellular, conformationally regulated, signaling models of and nicotinamide production are not responsible? First, perhaps the this type have also been invoked to explain the signal transduction 2702 CD38 SIGNALING OCCURS INDEPENDENTLY OF cADPR GENERATION properties of a number of other lymphocyte ecto-, includ- 16. Lund, F., N. Solvason, J. C. Grimaldi, R. M. E. Parkhouse, and M. Howard. 1995. ing CD73 and CD26, which can also signal independently of their Murine CD38: an immunoregulatory ectoenzyme. Immunol. Today 16:469. 17. Funaro, A., G. C. Spagnoli, C. M. Ausiello, M. Alessio, S. Roggero, D. Delia, respective enzyme activities, yet require their extracellular do- M. Zaccolo, and F. Malavasi. 1990. Involvement of the multilineage CD38 mol- mains for signal transduction (38, 39). If these signaling models ecule in a unique pathway of cell activation and proliferation. J. Immunol. 145: are correct, it suggests that ecto-enzymes have evolved a unique 2390. 18. Santos-Argumedo, L., C. Teixeira, G. Preece, P. A. Kirkham, and method of interacting with the cell signal transduction machinery. R. M. E. Parkhouse. 1993. A B lymphocyte surface molecule mediating activa- Additionally, the models suggest that conformational changes in- tion and protection from apoptosis via calcium channels. J. Immunol. 151:3119. duced via ligand binding in vivo must regulate signal transduction. 19. Lund, F. E., N. W. Solvason, M. P. Cooke, A. W. Heath, J. C. Grimaldi, R. M. E. Parkhouse, C. C. Goodnow, and M. C. Howard. 1995. Signaling through If this is true, conformational changes in the ectodomain induced murine CD38 is impaired in antigen receptor unresponsive B cells. Eur. J. Im- upon binding or catalysis of substrate could potentially mediate munol. 25:1338. signaling. Thus, it is possible that NAD could serve as a signaling 20. Lund, F. E., N. Yu, K.-M. Kim, M. Reth, and M. C. Howard. 1996. Signaling through CD38 augments B cell antigen receptor (BCR) responses and is depen- ligand in vivo. Regardless, the data presented here clearly demon- dent on BCR expression. J. Immunol. 157:1455. strate that signaling through CD38, at least in vitro, is mediated 21. Ausiello, C. M., F. Urbani, F. la Sala, A. Funaro, and F. Malavasi. 1995. CD38 ligation induces discrete cytokine mRNA expression in human cultured lympho- entirely through its ectodomain, and while cADPR and ADPR may cytes. Eur. J. Immunol. 25:1477. regulate other signal transduction events, neither cADPR nor 22. Ausiello, C. M., A. la Sala, C. Ramoni, F. Urbani, A. Funaro, and F. Malavasi. ADPR production appears necessary for anti-CD38 mediated sig- 1996. Secretion of IFN-␥, IL-6, granulocyte- colony-stimulating fac- tor and IL-10 after activation of human purified T lymphocytes upon naling or coreceptor activity in murine B lymphocytes. CD38 ligation. Cell. Immunol. 173:192. 23. Cockayne, D., T. Muchamuel, J. C. Grimaldi, H. Muller-Steffner, T. D. Randall, Acknowledgments F. E. Lund, R. Murray, F. Schuber, and M. C. Howard. 1998. Mice deficient for Downloaded from the ecto-NADϩ glycohydrolase CD38 exhibit altered humoral immune re- We thank Sriram Balasubramanian and J. Christopher Grimaldi for sharing sponses. Blood 92:1324. their data on the soluble CD38 mutant (E150Q/D151N). We also thank 24. Taira, S., M. Matsui, K. Hayakawa, T. Yokoyama, and H. Nariuchi. 1987. In- terleukin 2 secretion by B cell lines and splenic B cells stimulated with calcium Troy Randall and Brian Rogerson for critical reading of the manuscript, ionophore and phorbol ester. J. Immunol. 139:2957. and Debra Cockayne for helpful discussions. 25. Justement, L. B., J. Kreiger, and J. C. Cambier. 1989. Production of multiple lymphokines by the A20. 1 B cell lymphoma after cross-linking of membrane Ig

References by immobilized anti-Ig. J. Immunol. 143:881. http://www.jimmunol.org/ 26. Kakiuchi, T., T. Tamura, Y. Gyotoku, and H. Nariuchi. 1991. IL-2 production by 1. Mehta, K., U. Shahid, and F. Malavasi. 1996. Human CD38, a cell surface protein B cells stimulated with a specific antigen. Cell. Immunol. 138:207. with multiple functions. FASEB J. 10:1408. 27. Muller-Steffner, H. M., A. Augustin, and F. Schuber. 1996. Mechanism of cy- ϩ 2. Jackson, D. G., and J. I. Bell. 1990. Isolation of a cDNA encoding the human clization of pyridine nucleotides by bovine spleen NAD glycohydrolase. J. Biol. CD38 (T10) molecule, a cell surface glycoprotein with an unusual discontinuous Chem. 271:23967. pattern of expression during lymphocyte differentiation. J. Immunol. 144:2811. 28. Nishina, H., K. Inageda, K. Takahashi, S. Hoshino, K. Ikeda, and T. Katada. 3. Harada, N., L. Santos-Argumedo, R. Chang, J. C. Grimaldi, F. E. Lund, 1994. Cell surface antigen CD38 identified as ecto-enzyme of NAD glycohydro- C. I. Brannan, N. G. Copeland, N. A. Jenkins, A. W. Heath, R. M. E. Parkhouse, lase has hyaluronate-binding activity. Biochem. Biophys. Res. Commun. 203: et al. 1993. Expression cloning of a cDNA encoding a novel murine B cell 1318. activation marker. J. Immunol. 151:3111. 29. Shaw, A. C., R. N. Mitchell, Y. K. Weaver, J. Campos-Torres, A. K. Abbas, and 4. Alessio, M., S. Roggero, A. Funaro, L. B. De Monte, L. Peruzzi, M. Geuna, and P. Leder. 1990. Mutations of immunoglobulin transmembrane and cytoplasmic F. Malavasi. 1990. CD38 molecule: structural and biochemical analysis on hu- domains: effects on intracellular signaling and antigen presentation. Cell 63:381. by guest on September 27, 2021 man T lymphocytes, thymocytes, and plasma cells. J. Immunol. 145:878. 30. Blum, J. H., T. L. Stevens, and A. L. DeFranco. 1993. Role of the mu immuno- 5. States, D. J., T. F. Walseth, and H. C. Lee. 1992. Similarities in amino acid globulin heavy chain transmembrane and cytoplasmic domains in B cell antigen sequences of Aplysia ADP-ribosyl cyclase and human lymphocyte antigen CD38. receptor expression and signal transduction. J. Biol. Chem. 268:27236. Trends Biochem. Sci. 17:495. 31. Stearne, P. A., G. A. Pietersz, and J. W. Goding. 1985. cDNA cloning of the 6. Hellmich, M. R., and F. Strumwasser. 1991. Purification and characterization of murine transferrin receptor: sequence of transmembrane and adjacent regions. a molluscan egg-specific NADase, a second-messenger enzyme. Cell Regul. J. Immunol. 134:3473. 2:193. 32. Grimaldi, J. C., S. Balasubramanian, N. H. Kabra, A. Shanafelt, J. F. Bazan, 7. Lee, H. C., and R. Aarhus. 1991. ADP-ribosyl cyclase: an enzyme that cyclizes G. Zurawski, and M. C. Howard. 1995. CD38-mediated ribosylation of proteins. ϩ NAD into a calcium mobilizing metabolite. Cell Regul. 2:203. J. Immunol. 155:811. 8. Clapper, D. L., T. F. Walseth, P. J. Dargie, and H. C. Lee. 1987. Pyridine nu- 33. Tohgo, A., S. Takasawa, N. Noguchi, T. Koguma, K. Nata, T. Sugimoto, cleotide metabolites stimulate calcium release from sea urchin egg microsomes Y. Furuya, H. Yonekura, and H. Okamoto. 1994. Essential cysteine residues for desensitized to inositol triphosphate. J. Biol. Chem. 262:9561. cyclic ADP-ribose synthesis and hydrolysis by CD38. J. Biol. Chem. 269:28555. 9. Dargie, P. J., M. C. Agee, and H. C. Lee. 1990. Comparison of Ca2ϩ mobilizing 34. Lund, F. E., D. A. Cockayne, T. D. Randall, N. Solvason, F. Schuber, and activities of cyclic ADP-ribose and inositol triphosphate. Cell Regul. 1:279. ϩ ϩ M. C. Howard. 1998. CD38: a new paradigm in lymphocyte activation and signal 10. Galione, A., H. C. Lee, and W. B. Busa. 1991. Ca2 -induced Ca2 release in sea transduction. Immunol. Rev. 161:79. urchin egg homogenates: modulation by cyclic ADP-ribose. Science 253:1143. 11. Meszaros, L., J. Bak, and A. Chu. 1993. Cyclic ADP-ribose as an endogenous 35. Aarhus, R., R. M. Graeff, D. M. Dickey, T. F. Walseth, and H. C. Lee. 1995. regulator of the non-skeletal type ryanodine receptor Ca2ϩ channel. Nature 354: ADP-ribosyl cyclase and CD38 catalyze the synthesis of a calcium mobilizing 76. metabolite from NADP. J. Biol. Chem. 270:30327. 12. Sitsapesan, R., S. J. McGarry, and A. J. Williams. 1995. Cyclic ADP-ribose, the 36. De Flora, A., L. Guida, L. Franco, E. Zocchi, S. Bruzzone, U. Benatti, ryanodine receptor and Ca2ϩ release. Trends Pharmacol. Sci. 16:386. G. Damonte, and H. C. Lee. 1997. CD38 and ADP-ribosyl cyclase catalyze the 13. Howard, M., J. C. Grimaldi, J. F. Bazan, F. E. Lund, L. Santos-Argumedo, synthesis of a dimeric ADP-ribose that potentiates the calcium mobilizing activity R. M. E. Parkhouse, T. F. Walseth, and H. C. Lee. 1993. Formation and hydro- of cyclic ADP-ribose. J. Biol. Chem. 272:12945. lysis of cyclic ADP-ribose catalyzed by lymphocyte antigen CD38. Science 262: 37. Prasad, G. S., D. E. McRee, E. A. Stura, D. G. Levitt, H. C. Lee, and D. C. Stout. 1056. 1996. Crystal structure of Aplysia ADP ribosyl cyclase, a homologue of the 14. Takasawa, S., A. Togho, N. Noguchi, T. Koguma, K. Nata, T. Sugimoto, bifunctional ectoenzyme CD38. Nat. Struc. Biol. 3:957. H. Yonekura, and H. Okamoto. 1993. Synthesis and hydrolysis of cyclic ADP- 38. Resta, R., Y. Yamashita, and L. F. Thompson. 1998. Ecto-enzyme and signaling ribose by human leukocyte antigen CD38 and inhibition of the hydrolysis by functions of lymphocyte CD73. Immunol. Rev. 161:95. ATP. J. Biol. Chem. 268:26052. 39. von Bonin, A., J. Huhn, and B. Fleischer. 1998. Dipeptidyl-peptidase IV/CD26 15. Lee, H. C., R. M. Graeff, and T. F. Walseth. 1997. ADP-ribosyl cyclase and on T cells: analysis of an alternative T-cell activation pathway. Immunol. Rev. CD38. Adv. Exp. Med. Biol. 419:411. 161:43.