CD38 is expressed as noncovalently associated homodimers on the surface of murine B lymphocytes Miguel E. Moreno-Garcı ´ a 1,2 , Santiago Partida-Sa ´ nchez 3 , Julie Primack 3 , Adriana Sumoza-Toledo 2 , He ´ le ` ne Muller-Steffner 4 , Francis Schuber 4 , Norman Oppenheimer 5 , Frances E. Lund 3 and Leopoldo Santos-Argumedo 2 1 Departamentos de Biologı ´ a Celular and 2 Biomedicina Molecular, CINVESTAV-IPN, Mexico; 3 Trudeau Institute, Saranac Lake, New York, USA; 4 Laboratoire de Chimie Bioorganique, UMR 7514 CNRS/ULP, Strasbourg-Illkirch, France; 5 Department of Pharmaceutical Chemistry, UCSF, San Francisco, USA CD38 is a transmembrane glycoprotein that functions as an ectoenzyme and as a receptor. Based on the structural similarity between CD38 and ADP-ribosyl cyclase from Aplysia californica, it was hypothesized that CD38 is expressed as a homodimer on the surface of cells. Indeed, CD38 dimers have been reported, however, the structural requirements for their stabilization on the plasma mem- brane are unknown. We demonstrate that the majority of CD38 is assembled as noncovalently associated homo- dimers on the surface of B cells. Analysis of CD38 mutants, expressed in Ba/F3 cells, revealed that truncation of the cytoplasmic region or mutation of a single amino acid within the a1-helix of CD38 decreased the stability of the CD38 homodimers when solubilized in detergent. Cells expressing the unstable CD38 homodimers had diminished expression of CD38 on the plasma membrane and the half-lives of these CD38 mutant proteins on the plasma membrane were significantly reduced. Together, these results show that CD38 is expressed as noncova- lently associated homodimers on the surface of murine B cells and suggest that appropriate assembly of CD38 homodimers may play an important role in stabilizing CD38 on the plasma membrane of B cells. Keywords: B lymphocytes; CD38; homodimer stability; NAD + glycohydrolase; protein structure. CD38 is a type II transmembrane ectoenzyme expressed by many cell types [1–3]. CD38 plays an important role in calcium signaling as it catalyzes the production of several calcium mobilizing metabolites including adenosine(5¢)- diphospho(5)-b- D -ribose (ADP-Rib), cyclic adeno- sine(5¢)diphospho(5)-b- D -ribose (cADP-Rib) and nicotinic acid-adenine(5¢)diphosphate (NAADP + ) [4,5]. In addition to its role as an enzyme, CD38 can also serve as a receptor on the plasma membrane of leukocytes and lymphocytes. For example, incubation of B lymphocytes with agonistic antibodies to CD38 induces calcium mobilization, protein phosphorylation, proliferation, class switching, rescue from cell death and induction of apoptosis [1,6–10]. In order to understand the dual receptor and enzyme properties of CD38, a number of structure/function studies have been performed. These studies have been guided by analyses of two CD38 homologues, the cytosolic Aplysia califor- nica ADP-ribosyl cyclase [11,12] and the mammalian GPI-anchored NAD + glycohydrolase, CD157 [13,14]. Crystallographic and X-ray diffraction analyses of these two proteins indicated that both proteins form noncova- lently associated homodimers [15,16]. Thus, it has been proposed that CD38 is also likely to be expressed as a homodimer on the plasma membrane and, in agreement with this hypothesis, initial reports showed that high molecular mass aggregates of CD38 are formed after incubation of human erythrocytes with NAD + or 2-mercapto- ethanol [17]. In addition, it was reported that CD38 formed dimers and oligomers on the membrane of CD38 trans- fected HeLa cells [18]. It is not clear, however, whether CD38 is always present in a dimeric form on the surface of cells as many groups have reported finding only the monomeric form of CD38 [19–21]. Furthermore, it remains to be determined whether CD38 dimers are formed via covalent or noncovalent interactions between monomers. For example, it has been suggested that two extracellular cysteines in CD38 (Cys119 and Cys201 in human CD38 or Cys123 and Cys205 in mouse CD38) could form interdi- sulfide bonds between CD38 monomers [22]. In agreement with this hypothesis, studies carried out with porcine heart, rat lung and rat hepatocytes showed that under nonreduc- ing conditions CD38 forms dimers, while under reducing conditions CD38 is present in a monomeric form [23–25]. On the other hand, Umar et al. have shown that CD38 oligomers, expressed by retinoic acid stimulated HL60 cells, are covalently stabilized by transglutaminase, suggesting an alternate biochemical mechanism for the stabilization of covalent CD38 oligomers [26]. As these previous results are difficult to reconcile with one another, it is still unclear Correspondence to L. Santos-Argumedo, Departamento de Biomedicina Molecular, CINVESTAV-IPN, Av. IPN #2508 Col. Zacatenco, cp 07360, Me ´ xico D.F., Me ´ xico. Fax: + 52 55 5747 7134, Tel.: + 52 55 5061 3323, E-mail: lesantos@mail.cinvestav.mx Abbreviations:BS 3 , Bis(sulfosuccinimidyl)suberate; NP-40, Nonidet-P-40; IEF, isoelectric focusing. (Received 17 October 2003, revised 22 December 2003, accepted 20 January 2004) Eur. J. Biochem. 271, 1025–1034 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04006.x whether CD38 normally forms dimers, and if so, whether stable CD38 dimer formation is dependent on covalent bonds between monomers. In this report, we show that CD38 primarily forms homodimers on the plasma mem- brane of B lymphocytes. Furthermore, we demonstrate that the stability of the CD38 homodimers is highly dependent on the detergent used to solubilize the cells and is less dependent on the formation of interdisulphide bonds between CD38 monomers, indicating that the CD38 homodimers in B cells are likely to be stabilized via noncovalent monomer– monomer interactions. Finally, using Ba/F3 cells stably transfected with a number of different CD38 mutants, we identified two domains of the CD38 protein that confer stability to the homodimeric form of CD38. Functional analysis revealed that the half-life of these unstable CD38 homodimers on the plasma membrane was significantly less than wild-type CD38 resulting in reduced plasma membrane expression. Thus, these results suggest that assembly of CD38 homodimers may influence the stable expression of CD38 on the plasma membrane of B cells. Materials and methods Mice, B cell purification and cell lines Splenic B lymphocytes were purified from 6 to 8-week-old BALB/c, C3H/HeJ, NMRI, C57BL/6 or C57BL/6-Cd38 –/ / – [27] mice with magnetic beads coupled with antibodies to B220 (Miltenyi Biotech, Auburn, CA, USA). All research mice at CINVESTAV and Trudeau Institute were eutha- nized by CO 2 narcosis in accordance with the recommenda- tions of the Panel on Euthanasia of the AVMA and in compliance with the CINVESTAV and Trudeau Institute IACUC guidelines. The IL-3 dependent murine pro-B cell line, Ba/F3 (a generous gift from D. Campana, St. Jude Children’s Hospital, Memphis, TN, USA) was cultured in complete B cell media [21] supplemented with 10% (w/v) WEHI-3 supernatant (containing IL-3). Cell lysis, immunoprecipitation, Western blot and isoelectric focusing B cells were lysed with 10 m M Tris/HCl (pH 7.3), 2 m M Na 3 VO 4 ,0.4m M EDTA, 10 m M NaF, 1 m M phenyl- methanesulfonyl fluoride, 2 lgÆmL )1 aprotinin and leupep- tin and 1% of one of the following detergents: Nonidet-P-40 (v/v) (NP-40), Triton X-100 (v/v) (Sigma), Chaps (w/v) (Polysciences Inc., Warrington, PA, USA), deoxy-BigChap (w/v) (Pierce, Rockford, IL, USA), or digitonin (w/v) (Wako pure chemicals Ltd, Japan). Cell lysates were incubated over- nightat4°Cwith5lg of anti-mouse CD38 monoclonal antibody (NIM-R5 [1]) or a nonspecific rat IgG2a (Zymed, San Francisco, CA, USA) together with a 30 lLslurryof protein G beads (Zymed). Complexes were boiled in Laemmli buffer containing 2-mercaptoethanol or dithiotre- itol (Sigma) at the concentration indicated in the text. To analyze the samples under nonreducing conditions, the samples were suspended in the Laemmli buffer in the absence of reducing agents and then heated at 50 °Cfor3min. Immunoprecipitated proteins (25 lL) were loaded into 10% polyacrylamide gels, electrophoresed and transferred to nitrocellulose (Scleicher and Schuell, Dassel, Germany). The membranes were blocked with 5% bovine serum albumin (BSA) (Research Organics, Del Mar, CA, USA) and incubated with rabbit polyclonal antibody against CD38 [28] overnight at 4 °C followed by an anti-rabbit– HRP (DAKO, Carpinteria, CA, USA) for 2 h at room temperature. Proteins were developed using chemilumines- cence (Amersham Pharmacia Biotech, Buckingamshire, England). The two-dimensional isoelectric focusing (IEF) analysis of immunoprecipitated CD38 was performed as reported by O’Farrel [29]. Preparation of CD38 mutant cDNA constructs and Ba/F3 stable transfectants Expression vectors containing thefull length coding region of murine CD38 (CD38-WT-pME18S/neo) or the CD38 cytoplasmic region mutant (CD38-lATG-pME18S/neo) have been previously described [19,30]. CD38-E150L and CD38-G68E were generated by PCR using the CD38-WT expression vector as a template and the primers below. Restriction sites are underlined and the altered nucleotides that correspond to the replacement amino acid codons are indicated in lower case italics: CD38-E150L, primer 1: 5¢-(TACTT GGATCCAGGGAAAGATGTTCACCCTG ctGGACACCCTG)-3¢; CD38-E150L, primer 2: 5¢-(CC C TCTAGACCAGATCCTTCACGTATTAAGTCT ACACG)-3¢; CD38-G68E, primer 1: 5¢-(GACATCTTC CTCGagCGCTGCCTCATC)-3¢; CD38-G68E, primer 2: 5¢-(CCC TCTAGACCAGATCCTTCACGTATTAAGTC TACACG)-3¢; CD38-G68E, primer 3: 5¢-(GATGAGGC AGCG CTCGagGAAGATGTC)-3¢; CD38-G68E, primer 4: 5¢-(GGG GAATTCATGGCTAACTATGAATTTAGC CAG)-3¢. The E150L PCR product was digested with BamHI/XbaI and was used to replace the BamHI/XbaI fragment of CD38-WT in pME18S/neo. The two PCR products for G68E were digested with XhoI, XbaIandEcoRI and cloned by three way ligation into the EcoRI and XbaIsitesof pME18S/neo. The entire CD38 coding sequence was then sequenced in both directions to ascertain that the appropri- ate mutation was introduced and that no polymerase or cloning errors had occurred. Ba/F3 cells (5 · 10 6 ) were electroporated as described previously [21] and were cultured in Ba/F3 media containing Geneticin (G418, Gibco-BRL, Grand Island, NY, USA) at 500 lgÆmL )1 . After 10 days, the surviving CD38 + cells were single cell cloned into a 96-well plate using a FACSVantage-DIVA (Becton-Dickinson, San Jose, CA, USA). At least 20 independent clones from each transfec- tion were stained to determine CD38 expression levels and at least five individual clones were picked to expand and analyze experimentally. Measurement of cyclase and glycohydrolase activity in Ba/F3 transfectant cell homogenates Transfected Ba/F3 cells were washed with NaCl/P i , pelleted, snap frozen and stored at )70 °C. The mem- brane fraction was obtained and resuspended in 1 mL potassium phosphate buffer (50 m M ,pH6.8)usinga Dounce–Potter homogenizer (Wheaton Science Products, 1026 M. E. Moreno-Garcı ´ a et al.(Eur. J. Biochem. 271) Ó FEBS 2004 Milville, NJ, USA). Protein concentration was determined with the BCA protein assay (Pierce). The catalytic activity of CD38 in Ba/F3 cell homogenates was determined by HPLC using the radiolabeled substrates [carbo- nyl- 14 C]NAD and [adenosine-U- 14 C]NAD + as described previously [30]. To normalize the enzyme activity of the cell homogenates from the various Ba/F3 transfectants, the enzyme activity (V max ) was multiplied by a correction factor that compensated for the total protein per cell and the amount of CD38 expressed per cell. This correction factor was obtained by dividing the amount of protein per cell (1.54 · 10 )7 mg) by the amount of CD38 expressed on the membrane of each Ba/F3 cell (mean fluorescence intensities) and is represented in arbitrary units of CD38 per mg of total protein. The protein concentration per cell was determined by lysing a known number of Ba/F3 cells and determining protein concentration by Bradford ana- lysis. This was repeated multiple times and the number represents the average amount of protein (in mgs) per cell. Crosslinking with BS 3 B cells were washed, resuspended in 7 mL NaCl/P i ,and 560 lLofa25m M solution of bis(sulfosuccinimidyl)suber- ate (BS 3 , Pierce, Rockford, IL, USA) in 5 m M sodium citrate buffer was added dropwise to the cell suspensions giving a final concentration of 2 m M BS 3 . Cells were incubated for 1 h at 4 °C with gentle shaking. The reaction was stopped with 140 lL1 M Tris/HCl (pH 7.5). Cell suspensions were washed with NaCl/P i and prepared for lysis. FACS analysis To measure CD38 expression on Ba/F3 cells, 5 · 10 5 cells were stained with anti-CD38 Ig (NIM-R5-FITC, dilution 1 : 500) (Southern Biotech, Birmingham, AL, USA) for 30 min at 4 °C. The cells were analyzed by cytometry using a FACSCalibur (Becton-Dickinson, San Jose, CA, USA). Surface biotinylation of proteins To analyze the stability of CD38 on the plasma membrane of the different Ba/F3 mutants, labeling of the surface proteins with the membrane impermeable reagent sulfo-NSH-LC- biotin (Pierce) was performed as described [31] and following the manufacturer instructions. Briefly, cultured Ba/F3 cells (1 · 10 7 ) or splenic B cells (2 · 10 8 )werewashedtwotimes with sterile NaCl/P i and resuspended in 3 mL of NaCl/P i containing 0.5 mgÆmL )1 of sulfo-NHS-LC-biotin. The cells were incubated for 30 min at room temperature or on ice, followed by three washings with NaCl/P i . The Ba/F3 clones were then resuspended in complete Ba/F3 media and splenic B cells were resuspended in supplemented RPMI media containing 100 UÆmL )1 of IL-4. The cells were cultured at 37 °C, and 2 · 10 6 Ba/F3 cells or 5 · 10 7 splenic B cells were harvested at 0, 2, 10, 20 and 30 h. The cells were lysed with 0.5–1 mL of lysis buffer containing 1% (v/v) NP-40, and CD38 was immunoprecipitated as described above. Immu- noprecipitated CD38 was analyzed by Western blot using streptavidin-HRP (Sigma), and then the membrane was stripped and reanalyzed with rabbit anti-CD38 Ig and finally anti-rabbit Ig–HRP. Results CD38 forms homodimers in murine splenic B cells To address whether murine CD38 is expressed as a homodimer in splenic B cells, we purified these cells from CD38 expressing and CD38 deficient (CD38 KO) mouse strains and lysed them in buffer containing 1% (v/v) NP-40. CD38 was immunoprecipitated with an anti-mouse CD38 monoclonal antibody (NIM-R5), electrophoresed under reducing or nonreducing conditions and then analyzed by Western blot using a polyclonal rabbit anti-mouse CD38 Ig (Fig. 1A). Under nonreducing conditions, no CD38 reactive proteins were detected in the immunoprecipitates from CD38 KO cells (Fig. 1A, lane 1). In contrast, two distinct Fig. 1. CD38 forms 95 kDa homodimers in B lymphocytes. (A) B lymphocytes (5 · 10 7 ) were isolated from the indicated mouse strains, including CD38 deficient mice (CD38-KO [27]). The cells were lysed with 1% (v/v) NP-40 and CD38 was immunoprecipitated with monoclonal antibody to CD38, NIM-R5. The samples were prepared either in the absence (lanes 1–5) or presence of 5% (v/v) 2-mercapto- ethanol (lanes 6–10) and CD38 was detected by Western blot as des- cribed in Materials and methods. The relative molecular mass markers are indicated on the left of each figure. The nonspecific IgH band present in all of the reduced samples (including the CD38-KO sample) is indicated with an asterisk. (B) Immunoprecipitated CD38 from BALB/c B cell lysates was boiled in the presence of 2-mercaptoethanol (lanes 1–3) or dithiothreitol (lanes 4–6) at the concentrations shown in the figure. (C) CD38 was immunoprecipitated from splenic B cell lysates, resolved by 2D isoelectric focusing (IEF) and detected by Western blot. IEF spots of monomeric and dimeric forms of CD38 are indicated by arrows. b-ME, 2-mercaptoethanol; DTT, dithiothreitol. Ó FEBS 2004 CD38 homodimers are noncovalently stabilized in B cells (Eur. J. Biochem. 271) 1027 molecular mass forms of CD38 were observed in the immunoprecipitates from CD38-expressing cells; a 42 and a 95 kDa protein (p42 and p95) (Fig. 1A, lanes 2–5). The 42 kDa protein is the expected size of glycosylated mono- meric CD38 [19] while the 95 kDa protein is the approxi- mate size of a CD38 dimer. When the samples were boiled and reduced in 2-mercaptoethanol, we observed nonspecific bands of  68 kDa (corresponding to the immunoglobulin heavy chain present in B lymphocytes) and  200 kDa (data not shown) in all immunoprecipitates, including the sample from the CD38 KO mice (Fig. 1A, lane 6). In addition to observing the nonspecific bands, we still detected the p42 and p95 forms of CD38 in the CD38-expressing cells (Fig. 1A, lanes 7–10). This indicates that p95 was partially, although not fully, resistant to reduction by 2-mercapto- ethanol. Interestingly, even addition of higher concentra- tions of 2-mercaptoethanol or another reducing agent, dithiotreitol, did not completely ablate the p95 form of CD38 (Fig. 1B). To determine the structural composition of the p95 form of CD38, we first ruled out the possibility that the p95 form was composed of a CD38 monomer associated with the immunoglobulin heavy chain from the precipitating anti- CD38 Ig (data not shown). Next, we showed that the p95 form of CD38 was easily detected when iodoacetamide was included in all of the buffers in order to block any reactive free cysteines (data not shown). This ruled out the possibility that p95 was formed during the lysis and immunopreci- pitation process. Finally, we compared p42 and p95 for their pattern of isoelectric points by IEF and 2D polyacrylamide gel electrophoresis. For p42 we observed two dominant isoelectric points of 7.7 and 7.2 and two minor points at 7.4 and 7.1 (Fig. 1C). Analysis of p95 revealed isoelectric points of 7.7, 7.2 and 7.1 (Fig. 1C). These points were located at similar positions to the corresponding points in the p42 monomeric form. We did not detect a protein spot at 7.4 in p95; however, this protein species only represented a minor form even in the CD38 p42 monomer. Taken together, the data indicate that the p95 form of CD38 appears to represent a homodimeric form of CD38 as it is recognized by both monoclonal and polyclonal antibodies against CD38, and has essentially identical IEF points as the p42 monomer form of CD38. CD38 homodimers are expressed on the surface of splenic B cells and are destabilized when solubilized with type B surfactants (steroid-based detergents) To determine whether CD38 is normally expressed in the homodimeric form on the plasma membrane of B cells, we purified splenic B cells from normal and CD38 KO mice and treated them for 1 h with a nonpermeable chemical crosslinker, BS 3 , in order to stabilize the CD38 homodimers during the lysis and immunoprecipitation steps. As expec- ted, no CD38 protein was detected in the CD38 deficient cells (Fig. 2A, lanes 2 and 5). Similarly to the previous results, the majority of CD38 protein was of monomeric size (p42) in the cells that were not treated with BS 3 (Fig. 2A, lanes 1 and 4). In contrast, in cells that had been treated with crosslinker, CD38 was found predominantly in the p95 homodimeric form (lanes 3 and 6). As the ratio of homodimers to monomers was approximately five-fold increased when the crosslinker was used (Fig. 2A, compare lanes 1 and 3 or lanes 4 and 6), these results indicate that a large proportion of the total CD38 is expressed in a Fig. 2. CD38 is found as homodimers on the surface of splenic B lymphocytes and the stability of the dimers depends on the detergent used to solubilize the cells. (A) Purified B cells were incubated with the nonpermeable crosslinker BS 3 for 1 h at 4 °C(lanes2,3,5and6)or left untreated (lanes 1 and 4). Crosslinked cells were lysed and CD38 was immunoprecipitated from cells expressing CD38 (CD38-WT, lanes 1, 3, 4 and 6) or lacking CD38 (CD38-KO, lanes 2 and 5). Immunoprecipitated proteins were treated with (lanes 4–6) or without 5% (v/v) 2-mercaptoethanol (lanes 1–3) and CD38 was detected by Western blot. (B) Splenic B cells were solubilized with 1% (v/v) NP-40 (lanes 1, 2, 5 and 6) or Chaps (lanes 3, 4, 7 and 8) and CD38 was immunoprecipitated. The samples were heated in the presence (lanes 5–8) or absence (lanes 1–4) of 5% (v/v) 2-mercaptoethanol and CD38 was detected by Western blot. (C) B cells were solubilized with 1% (v/v) NP-40 (lane 1), Triton X-100 (lanes 2 and 3), digitonin (lanes 4 and 5), Chaps (lanes 6 and 7) or deoxy-BigChap (lanes 8 and 9). Lysates were immunoprecipitated with antibody to CD38 (NIM-R5) and CD38 was detected by Western blot. The nonspecific IgH band present in all samples, including samples immunoprecipitated with an isotype control antibody (IgG2a), is indicated with an asterisk. b-ME, 2-mercaptoethanol. 1028 M. E. Moreno-Garcı ´ a et al.(Eur. J. Biochem. 271) Ó FEBS 2004 homodimeric form on the surface of live B lymphocytes and suggest that most of the CD38 dimers must fall apart when the cells are solubilized in detergent. Crosslinkers like disuccinimidyl suberate, that have the same reactivity and spacer arm length as BS 3 (11.4 A ˚ ) also stabilized the CD38 homodimers. However, crosslinkers such as 3,3¢-dithio- bis(sulfosuccinimidyl propionate), sulfo-disulfosuccinimidyl tartarate and sulfo-bis[2-(sulfosuccinimidooxycarbonyl- oxy)ethyl]sulfone, that have the same reactivity as BS 3 but have different spacer arm lengths (12, 6.4 and 13 A ˚ , respectively), were unable to stabilize the homodimers (data not shown). These results suggest that the stabilization of CD38 homodimers by crosslinkers depends strongly on the conformation and orientation between the CD38 monomers. It has been reported that NP-40 and Triton X-100 stabilize noncovalent hetero- or homo-dimerization of proteins, while detergents like Chaps and octylglucoside disrupt these interactions [32,33]. Up to now, in all our experiments, the cells were solubilized in NP-40, a deter- gent that might help to stabilize or protect the CD38 dimers from dissociating during the solubilization process. In sharp contrast, when the B cells were solubilized with Chaps we found significantly less CD38 homodimers, whether under reducing (Fig. 2B, lanes 5–8) or nonreduc- ing (Fig. 2B, lanes 1–4) conditions. This demonstrates that the detergent used to solubilize the cells influenced the amount of CD38 homodimers that could be immuno- precipitated. To analyze whether the stabilization of CD38 dimers was a property of the family of detergents utilized, we used several different detergents to solubilize the cells. As shown in Fig. 2C, CD38 dimers were precipitated when the cells were solubilized with NP-40 or Triton X-100; detergents that belong to the polyoxyethylene family (Fig. 2C, lanes 1–3). In contrast, when the cells were solubilized with Chaps, digitonin or deoxy-BigChap, members of the steroid-based detergent family, only CD38 monomers were detected (Fig. 2C, lanes 4–9). These results suggest that CD38 homodimer stability is dependent on noncovalent interactions between CD38 monomers. Structural requirements for CD38 homodimerization To investigate the structural requirements for dimer stabilization, we determined whether different CD38 mutants were capable of forming homodimers when transfected into Ba/F3 cells. Ba/F3 cells, stably transfected with full length wild-type CD38 (CD38-WT) or with different CD38 mutants, were solubilized in NP-40 lysis buffer and CD38 was immunoprecipitated, run on SDS/ PAGE under nonreducing conditions, and detected by Western blot. A summary of the results, presented in Table 1, indicates that CD38 homodimers were present in the lysates of most of the transfectants expressing CD38 mutants, including, CD38-E150L, a CD38 active site mutant (Table 2, [34]) and CD38-C123K, a mutant that is unable to form the postulated interdisulphide bond between two CD38 monomers [22]. These data indicate that CD38 homodimers can be formed even when the active site is altered and the putative interdisulphide bridge formed betweeen CD38 monomers is disrupted. Interestingly, however, CD38 dimers were absent in lysates from two of the other mutant Ba/F3 transfectants. Table 1. Expression of CD38 homodimers in different CD38 mutants expressed in Ba/F3 pro-B cells. Each of the mutant CD38 cDNAs listed, was stably expressed in Ba/F3 cells (Materials and methods) or A20 cells as described previously [30]. The cells were solubilized in 1% (v/v) NP-40. CD38 was immunoprecipitated, run on SDS/PAGE gels under nonreducing conditions and then analyzed for the presence (Y) or absence (N) of the p42 monomer and p95 homodimer by Western blot. WT, wild-type. CD38 mutant p42 p95 WT Y Y lATG Y N C123K Y Y E150L Y Y E150Q Y Y D151V Y Y E150QD151N Y Y G68E Y N Table 2. NAD + glycohydrolase activity of membrane homogenates from Ba/F3 transfectants. Each of the mutant CD38 cDNAs listed, was stably expressed in Ba/F3 cells and the enzyme activity (V max ) of the membrane homogenates was determined as described previously [30]. The V max was adjusted to reflect differences in CD38 expression levels between the various mutants and is reported as nmol of product formed per minute per arbitrary unit of CD38. Briefly, the total amount of protein per Ba/F3 cell was determined by the Bradford method. The average amount of CD38 expressed on the membrane of each Ba/F3 cell was determined by FACS and is reported as mean fluorescence intensity (Fig. 3C shows values of each of the clones). The relative enzyme activity of each of the mutants is given in parentheses. It was determined by setting the V max adjusted activity of CD38-WT to 100% and then calculating the percentage activity of each of the mutants relative to CD38-WT. MFI, mean fluorescence intensity; WT, wild-type. Mutant V max (nmolÆmin )1 Æmg )1 protein) Protein per cell (mg protein per cell · 10 )7 ) CD38 per cell arbitrary units CD38 per cell (1/MFI · 10 )4 ) V max adjusted (nmolÆmin )1 per arbitrary units of CD38 · 10 )10 ) WT 843.2 1.54 9.0 1180.0 (100%) E150L 5.54 1.54 11.4 9.73 (0.8%) G68E 56.9 1.54 35.8 310 (26%) lATG 284.1 1.54 30.8 1350 (114%) Ó FEBS 2004 CD38 homodimers are noncovalently stabilized in B cells (Eur. J. Biochem. 271) 1029 In one of the mutants (CD38-lATG), the 22 amino acid cytoplasmic region of CD38 was replaced with a 4 amino acid tail (Met-Lys-Val-Lys), and in the second mutant (CD38-G68E), the glycine at position 68 was replaced by the polar residue glutamate. The G68 residue is within the a1-helix that has been previously postulated to be a dimer interface site in the Aplysia enzyme [16]. As shown in Fig. 3A (lanes 2 and 5), CD38 homodimers were preci- pitated from Ba/F3 transfectants expressing CD38-WT or expressing a mutant form of CD38 in which a single residueintheactivesitewasmutated(CD38-E150L). However, no CD38 homodimers were detected in immunoprecipitations from transfectants expressing the CD38-G68E or CD38-lATG mutant proteins (Fig. 3A, lanes 3 and 4). This result suggests that the cytoplasmic region and first a-helix interface region of CD38 are important for dimer stability. To determine whether these two regions were necessary for CD38 dimer stabilization on the plasma membrane of the Ba/F3 cells, the transfectants expressing CD38-lATG Fig. 3. The stability and membrane expression of CD38 homodimers is dependent on at least two separate domains of CD38. (A) Ba/F3 cells transfected with control vector, CD38-WT, CD38-G68E, CD38-lATG or CD38-E150L were lysed with 1% (v/v) NP-40 and CD38 was imu- noprecipitated, run on SDS/PAGE under nonreducing conditions and detected by Western blot. (B) The Ba/F3 transfectants listed above were treated (as described in Fig. 2) with the crosslinker BS 3 (lanes 2, 4, 6, 8 and 10) or left untreated (lanes 1, 3, 5, 7 and 9). CD38 was detected by Western blot as in Fig. 2A. (C) Ba/F3 mutants were analyzed for expression of CD38 on the plasma membrane by FACS using the antibody, NIM- R5, conjugated to FITC. Dead cells were excluded by propidium iodide incorporation. Light line histograms, nontransfected Ba/F3 cells; dark line histograms, Ba/F3 transfectants. The mean fluorescence intensity of the cells from each of the transfectants is listed above the histogram. (D) Post- nuclear supernatants were prepared from Ba/F3 clones lysed with 1% (v/v) NP-40, the protein concentration was determined and equivalent amounts of protein were run on SDS/PAGE gels under reducing conditions. CD38 and actin expression were analyzed by Western blot using rabbit polyconal anti-CD38 Ig and mouse monoclonal antibody to actin, followed by HRP-labeled anti-rabbit IgG and anti-mouse IgG, respectively. (E) Comparison of plasma membrane and total CD38 expression levels in the Ba/F3 clones. To determine the relative plasma membrane expression levels of CD38 between the different Ba/F3 clones the mean fluorescence intensity for each of the clones was determined (C) and the relative levels were normalized to that of the CD38-WT transfectant which was set at 100%. To quantitate the total CD38 expression levels for the various Ba/F3 transfectant clones, densitometric analysis of CD38 and actin Western blots (D) were performed using SIGMAGEL . LNK . The CD38 levels for each clone were first normalized to actin by dividing the densitometric value of CD38 by the densitometric value of actin. Then the relative total CD38 expression levels for each clone were normalized to CD38-WT which was set at 100%. 1030 M. E. Moreno-Garcı ´ a et al.(Eur. J. Biochem. 271) Ó FEBS 2004 or CD38-G68E were crosslinked with BS 3 , solubilized in NP-40 lysis buffer, and CD38 was detected by immuno- precipitation and Western blot (Fig. 3B). As we have previously observed, homodimers of CD38 were absent in the immunoprecipitates from the noncrosslinked CD38- lATG and CD38-G68E transfectants (lanes 5 and 9). However, when the crosslinker was added, CD38 homo- dimers could be visualized (lanes 6 and 10). Indeed, similar ratios of homodimers to monomers were observed in the crosslinked CD38-G68E and CD38-lATG mutants com- pared to crosslinked CD38-WT and CD38-E150L (compare lanes 4, 6, 8 and 10). Therefore, the cytoplasmic region and a1-helix domains are not critical for CD38 dimer stabiliza- tion in B cells, however, the two domains must contribute to the overall stability of CD38 homodimers because the mutated dimers fell apart even under ÔpermissiveÕ solubili- zation and nonreducing conditions. CD38-G68E and CD38-lATG are less efficiently expressed and have a reduced half-life on the plasma membrane The previous results indicated that CD38-lATG and CD38-G68E are not obligatory for dimer stabilization but do contribute to the overall stability of the dimers, particularly upon detergent solubilization. It has been reported that inappropriate folding of proteins or inappro- priate assembly of multimeric protein complexes can influence the surface and overall expression of these proteins in cells and can also alter the half-life of the misfolded or disorganized protein complexes [35,36]. Given that the stability of CD38-lATG and CD38-G68E homodimers is reduced when the proteins are solubilized in permissive detergents, immunoblotting and FACS experiments were performed in order to analyze the total and surface levels of CD38 in all the Ba/F3 transfectants (Fig. 3C–E). These experiments revealed that the total amount of CD38 (as assessed by immunoblotting), as well as the amount of CD38 expressed on the plasma membrane (as assessed by FACS) was similar in the CD38-WT and CD38-E150L transfected Ba/F3 cells (Fig. 3C,D). In contrast, CD38 expression levels (both total and plasma membrane levels) in transfectants expressing CD38-G68E and CD38-lATG were significantly decreased relative to CD38-WT (Fig. 3C,D). Similar results were obtained upon analysis of multiple independent Ba/F3 clones expressing CD38- lATG or CD38-G68E (data not shown). Upon densito- metric analysis of the immunoblots it became clear that the amount of CD38 expressed on the plasma membrane and the total amount of CD38 expressed by the various transfectants correlated very well with one another (Fig. 3E), strongly suggesting that the reduced cell surface expression of CD38 by the CD38-lATG and CD38-G68E transfectants was not due simply to inefficient transport of the protein to the plasma membrane. In addition, confocal microscopic analysis of CD38 expression in Ba/F3 trans- fectants revealed that neither CD38 nor any of the CD38 mutant proteins were present in intracellular compartments (data not shown). Therefore, we next considered the possibility that the reduced plasma membrane expression of the CD38-lATG and CD38-G68E mutant proteins could be due to a faster turnover rate for these mutant molecules on the plasma membrane [35]. To analyze this possibility, we performed pulse-chase experiments using normal B cells and the Ba/F3 transfectants. To first determine whether the surface half-life of CD38 in Ba/F3 cells is comparable to that of splenic B cells, we biotinylated the surface of splenic B cells and Ba/F3 clones (CD38-WT and lATG) and compared the plasma membrane half-life of CD38. In these experiments the biotinylation was carried out for 30 min on ice in order to avoid any potential internalization of biotin or biotinylated proteins. The cells were then washed to remove the reactive biotin and cultured for up to 30 h. The amount of cell surface biotin-labeled CD38 at various timepoints was determined by immuno- precipitating with anti-CD38 and immunoblotting with SA-HRP to detect the cell surface biotinylated-CD38 and anti-CD38 to detect the total CD38 pool. The plasma membrane expression of CD38 on CD38-WT Ba/F3 transfectant cells and on normal B cells was quite stable over time with a half-life of approximately 28 h on both cell types (Fig. 4A,B), indicating that the turnover rate of CD38 in both cell types is similar and comparable. In contrast, the surface half-life of CD38-lATGwaslessthanhalfthat observed for CD38 expressed by normal B cells or Ba/F3 transfectants, suggesting that this mutant protein is less stably expressed on the plasma membrane (Fig. 4A,B). To confirm these results, we repeated the biotinylation experi- ment using Ba/F3 transfectants expressing other CD38 mutant proteins. The plasma membrane expression of both CD38-WT and CD38-E150L was quite stable over time with a half-life of approximately 28 h (Fig. 5A,B). In striking contrast, the half-lives of plasma membrane bound CD38-G68E and CD38-lATG were less than half that observed with CD38-WT (Fig. 5A,B). Thus, both of the CD38 mutant proteins that form unstable homodimers also have significantly reduced stability on the plasma mem- brane, suggesting that appropriate assembly or stabilization of CD38 into homodimers may be required for its extended expression on the plasma membrane. Enzyme activity is not dependent on the presence of stable homodimers As the mutations in the cytoplasmic region and the a1-helix of CD38 affected dimer stability upon solubilization, plasma membrane expression levels and surface half-life, it was also possible that these mutations would affect the enzyme activity of the proteins. To test the enzyme activity of the CD38 mutants, membrane homogenates from the various Ba/F3 transfectants were prepared and NAD + glycohydrolase activity in the membranes was measured by HPLC. As shown in Table 2, the enzyme activity of the active site mutant, CD38-E150L was greatly decreased compared to CD38-WT. Interestingly, the glycohydrolase activities of CD38-G68E and CD38-lATG were also less than CD38-WT. As the membrane expression levels of CD38-lATG and CD38-G68E were reduced compared to CD38-WT (Fig. 3C,D), we performed a calculation to adjust the enzyme activity (V max adjusted) to reflect the amount of total protein and CD38 protein expressed on a per cell basis. Upon adjusting the enzyme activity to compensate for the membrane CD38 expression levels, we found that the enzyme activity of CD38-lATG was at least Ó FEBS 2004 CD38 homodimers are noncovalently stabilized in B cells (Eur. J. Biochem. 271) 1031 as high as CD38-WT (Table 2). These data indicate that the catalytic activity of CD38 is not dependent on the formation of stable CD38 homodimers. Interestingly, however, even when expression levels of CD38-G68E were accounted for, the NAD + glycohydrolase activity of CD38-G68E was only 27% of CD38-WT. This result shows that a single point mutation in the first a-helix of CD38, a residue that is far removed from the active site of CD38, can significantly influence CD38 enzyme activity. Discussion In this work we show that homodimers of CD38 are expressed on the surface of B lymphocytes. Although CD38 dimers that are sensitive to reducing agents have been previously reported [22–25], we found that the stability of CD38 dimers expressed in B cells correlated better with the type of detergent used to solubilize the cells (Fig. 2) than the presence or absence of reducing agents (Fig. 1). Previous reports have shown that heterodimerization of proteins such as Bax with Bcl-2 or Bax with Bcl-X L are dependent on the detergent used to solubilize the cells [32,33]. Thus, NP-40 and Triton X-100, detergents that form large micelles, stabilized the hydrophobic interactions between Bax and its partners, while Chaps and octyl glucoside, detergents that form small micelles, could not accommodate the hetero- dimers. Interestingly, we found the same pattern with CD38 homodimers in that they were stabilized in the polyoxy- Fig. 5. The mutants expressing unstable CD38 homodimers present a reduced CD38 half-life on the plasma membrane. (A) Ba/F3 transfectant cells expressing CD38-WT or each of the different mutants were sur- face labeled with sulfo-NHS-LS-biotin for 30 min at room tempera- ture. The cells were washed and then cultured at 37 °Cforan additional 30 h. 2 · 10 6 cells were harvested at 0, 2, 10, 20 and 30 h after biotin labeling. Cell viability, as measured by trypan blue exclu- sion, was over 95% at each time point. Immunoprecipitation and Western-blotting was performed as described in Materials and methods and Fig. 4. (B) Densitometric analyses using the program SIGMAGEL.LNK were performed to compare the relative amounts of biotin-labeled CD38 in each Ba/F3 clone. Densitometry was per- formed as described in Fig. 4. Fig. 4. The half-life of CD38 is the same in splenic B cells and CD38- WT Ba/F3 transfectants. Purified splenic B cells, CD38-WT and CD38-lATG Ba/F3 transfectants were labeled with sulfo-NHS-LS- biotin for 30 min on ice. The cells were washed and then cultured at 37 °C for an additional 30 h. 2 · 10 6 Ba/F3 cells or 5 · 10 7 splenic B cells were harvested at 0, 10, 20 and 30 h after biotin labeling. Cell viability, as measured by trypan blue exclusion, was over 95% for the Ba/F3 transfectants at each time point and was 97, 90, 87 and 75% for splenic B cells at 0, 10, 20 and 30 h after biotinylation, respectively. (A) At each timepoint, the cells were lysed in 1% (v/v) NP-40, CD38 was immunoprecipitated with anti-CD38 Ig, and the immunoprecipi- tated protein was analyzed by SDS/PAGE and Western blotting. The amount of plasma membrane associated (biotinylated-CD38) and total CD38 was determined by immunoprecipitation, SDS/PAGE and Western blotting. Plasma membrane biotinylated-CD38 was detected with streptavidin-HRP (left). Total immunoprecipitated CD38 was detected using the polyclonal rabbit antibody to CD38 (right). (B) Densitometric analyses using the program SIGMAGEL . LNK were performed to compare the relative amounts of biotin-labeled CD38 (membrane CD38) in each Ba/F3 clone. To determine the relative amount of biotin-labeled CD38 present in each clone, the densito- metric value of biotin-CD38 was divided by the densitometric value of total CD38. The ratio of cell surface CD38 to total CD38 at time 0 was set at 100% and all other time points were compared relative to this. 1032 M. E. Moreno-Garcı ´ a et al.(Eur. J. Biochem. 271) Ó FEBS 2004 ethylene detergents (i.e. NP-40 and Triton X-100) and were destabilized with detergents such as digitonin, Chaps and deoxy-BigChap (Fig. 2). Importantly, these differences could not be attributed to differences in the ability of the various detergents to solubilize CD38 (data not shown). Furthermore, when we used the crosslinker, BS 3 ,the majority of CD38 was ÔcapturedÕ in the homodimer form indicating that CD38 must be dimerized via noncovalent interactions that were partially disrupted when the cells were solubilized in detergent. However, it is also clear that conformation and folding of the individual CD38 mono- mers is strongly influenced by the five known intradisul- phide bonds present in each monomer and their reduction is also expected to greatly influence the stability of the noncovalently associated CD38 homodimers. As CD38 dimers appear to be stabilized via noncovalent interactions between monomers, a reasonable assumption is that mutations within the potential interface domains might alter the formation or stability of CD38 homodimers. When cells expressing two different mutant forms of CD38, a cytoplasmic region mutant and an a1-helix mutant, were solubilized under nonreducing conditions in a permissive detergent such as NP-40, we were unable to detect the presence of CD38 homodimers (Fig. 3A), suggesting that these two domains play an important role in homodimer stability. Because homodimers of CD38-G68E and CD38- lATG were observed when the cells were crosslinked with BS 3 (Fig. 3B), these data suggest that these domains are not obligate for the stabilization of the dimers, but rather must contribute to the overall stability of the dimers. Crystallo- graphic analysis of the Aplysia cyclase indicates four putative oligomerization sites including the a1, a4anda10 helices and residues between 242 and 248 [16]. Thus, we propose that multiple contact points contribute in an additive or synergistic manner to CD38 homodimer stabil- ization. Although the mutants described here are not sufficient, in themselves, to control CD38 homodimer stabilization, the results clearly demonstrate a role for the a1-helix and the cytoplasmic region in stabilizing the solubilized homodimers. The mutations in the cytoplasmic tail and a1-helix of CD38 not only affected the stability of the CD38 homo- dimer upon solubilization, but also significantly diminished the expression of the CD38 homodimer on the plasma membrane. Indeed, we were never able to isolate CD38- G68E expressing transfectants expressing levels of CD38 comparable to CD38-WT, despite screening more than 25 individual clones (data not shown). As a whole, these data suggest that CD38-G68E and CD38-lATG mutants were either inefficiently assembled and transported to the mem- brane or were less stable on the surface. Our pulse-chase experiments (Figs 4 and 5) clearly showed that the plasma membrane half-life of these CD38 mutants was significantly less than CD38-WT, suggesting that extended plasma membrane expression of CD38 may depend on the presence of stable CD38 homodimers. Although further experiments will be needed to prove this hypothesis, similar results were obtained analyzing mutants of the dipeptidylpeptidase IV CD26 [36]. The close correlation between surface expression of CD38 and total expression of CD38 suggests that the mutant forms of CD38 are not preferentially retained in the intracellular compartments (Fig. 3C–E). Furthermore, intracellular expression analysis of CD38 on Ba/F3 trans- fectants and normal B cells using confocal microscopy or subcellular fractionation and immunobloting revealed that neither CD38 nor any of the CD38 mutant proteins analyzed in this study were detected in intracellular mem- branes (data not shown). CD38 homodimers have been proposed to play a number of different functional roles including formation of a transmembrane pore, allowing for transport of cADPR into the cytosol [37]. However, it is clear that stable homodimers are not obligate for enzyme activity as the unstable CD38 homodimer mutant, CD38-lATG, had perfectly normal enzyme activity when the activity was adjusted to reflect CD38 expression levels on the membrane (Table 2). In agreement with this, we also observed NADase activity from the p42 monomeric form of CD38 (data not shown), suggesting that CD38 monomers are enzymatically active. This is in agreement with the results of Bruzzone et al. [18]. Interestingly, although mutations in the active site do not decrease the formation or stability of CD38 homodimers (Table 2), the cells expressing the unstable CD38 homodimer, CD38-G68E, had significantly decreased CD38 dependent enzyme activity (Table 2). Thus, while CD38 enzyme activity is not critically dependent on the presence of stable CD38 homodimers, it is clear that mutating a single residue in the a1-helix interface can decrease both homodimer stability and enzyme activity. In conclusion, we have shown that CD38 is normally expressed as a noncovalently associated homodimer on the plasma membrane of B cells. Mutations that affect the stability of the CD38 homodimer do not necessarily alter CD38-dependent enzyme activity; however, these mutations do result in reduced plasma membrane stability and decreased expression of CD38 on the plasma membrane. Acknowledgements The authors would like to thank Troy Randall for discussions and critical reading of this manuscript. The authors also thank Dr Jose ´ Manuel Herna ´ ndez-Herna ´ ndez for technical advice and Q. F. B. He ´ ctor Romero Ramı ´ rez for technical assistance. M. E. M G., A. S T. and L. S A. are supported by CONACyT Me ´ xico grants, # 28093N, 33497N and 40218Q. J. P., S. P-S., and F. E. L. are supported by NIH grant AI-43629 and the Trudeau Institute. References 1. Santos-Argumedo, L., Teixeira, C., Preece, G., Kirkham, P.A. & Parkhouse, R.M.E. (1993) A B lymphocyte surface molecule mediating activation and protection from apoptosis via calcium channels. J. Immunol. 15, 3119–3130. 2. Mehta, K., Shahid, U. & Malavasi, F. (1996) Human CD38, a cell-surface protein with multiple function. FASEB J. 10, 1408– 1417. 3. Lund, F.E., Cockayne, D.A., Randall, T.D., Solvason, N., Schuber, F. & Howard, M.C. (1998) CD38: a new paradigm in lymphocyte activation and signal transduction. Immunol. Rev. 161, 79–93. 4. Howard,M.C.,Grimaldi,J.C.,Bazan,J.F.,Lund,F.E.,Santos- Argumedo, L., Parkhouse, R.M.E., Walseth, T.F. & Lee, H.C. (1993) Formation and hydrolysis of cyclic ADP-ribose catalyzed by lymphocyte antigen CD38. Science 262, 1056–1059. Ó FEBS 2004 CD38 homodimers are noncovalently stabilized in B cells (Eur. J. Biochem. 271) 1033 5. Aarhus, R., Graeff, R.M., Dickey, D.M., Walseth, T.F. & Lee, H.C. (1995) ADP-ribosyl cyclase and CD38 catalyze the synthesis of a calcium-mobilizing metabolite from NADP + . J. Biol. Chem. 270, 30327–30333. 6. Kirkham, P.A., Santos-Argumedo, L., Harnett, M.M. & Park- house, R.M.E. (1994) Murine B-cell activation via CD38 and protein tyrosine phosphorylation. Immunology 83, 513–516. 7. Kitanaka, A., Suzuki, T., Ito, C., Nishigaki, H., Coustain-Smith, E., Tanaka, T., Kubota, Y. & Campana, D. (1999) CD38-medi- ated signaling events in murine pro-B cells expressing human CD38 with or without its cytoplasmic domain. J. Immunol. 162, 1952–1958. 8. Kumagai, M., Coustan-Smith, E., Murray, D.J., Silvennoinen, O., Murti, K.G., Evans, W.E., Malavasi, F. & Campana, D. (1995) Ligation of CD38 suppresses human B lymphopoiesis. J. Exp. Med. 181, 1101–1110. 9. Zubiar, M., Guirado, M., Terhorst, C., Malavasi, F. & Sancho, J. (1999) The CD3-cde transducing module mediates CD38-induced protein-tyrosine kinase and mitogen-activated protein kinase activation in Jurkat T cells. J. Biol. Chem. 274, 20633–20642. 10.Yasue,T.,Baba,M.,Mori,S.,Mizoguchi,C.,Uehara,S.& Takatsu, K. (1999) IgG1 production by sIgD+ splenic B cells and peritoneal B-1 cells in response to IL-5 and CD38 ligation. Int. Immunol. 11, 915–923. 11. Hellmich, M.R. & Strumwasser, F. (1991) Purification and char- acterization of a molluscan egg-specific NADase, a second- messenger enzyme. Cell Regul. 2, 193–202. 12. Lee, H.C. & Aarhus, R. (1991) ADP-ribosyl cyclase: an enzyme that cyclizes NAD + into a calcium-mobilizing metabolite. Cell Regul. 2, 203–209. 13. Dong, C., Wang, J., Neame, P. & Cooper, M.D. (1994) The murine BP-3 gene encodes a relative of the CD38/NAD + glyco- hydrolase family. Int. Immunol. 6, 1353–1360. 14. Itoh, M., Ishihara, K., Tomizawa, H., Tanaka, H., Kobune, Y., Ishikawa, J., Kaisho, T. & Hirano, T. (1994) Molecular cloning of murine BST-1 having homology with CD38 and Aplysia ADP- ribosyl cyclase. Biochem. Biophys. Res. Commun. 203, 1309–1317. 15. Yamamoto-Katayama, S., Ariyoshi, M., Ishihara, K., Hirano, T., Jingami, H. & Morikawa, K. (2002) Crystallographic studies on human BST-1/CD157 with ADP-ribosyl cyclase and NAD gly- cohydrolase activities. J. Mol. Biol. 316, 711–723. 16. Prasad, G.S., McRee, D.E., Stura, E.A., Levitt, D.G., Lee, H.C. & Stout, C.D. (1996) Crystal structure of Aplysia ADP ribosyl cyclase, a homologue of the bifunctional ectozyme CD38. Nat. Struct. Biol. 3, 957–964. 17. Franco, L., Zocchi, E., Calder, L., Guida, L., Benatti, U. & De Flora, A. (1994) Self-aggregation of the transmembrane glycoprotein CD38 purified from human erythrocytes. Biochem. Biophys. Res. Commun. 202, 1710–1715. 18. Bruzzone, S., Guida, L., Franco, L., Zocchi, E., Corte, G. & De Flora, A. (1998) Dimeric and tetrameric forms of catalytically active transmembrane CD38 in transfected HeLa cells. FEBS Lett. 433, 275–278. 19. Harada, N., Santos-Argumedo, L., Chang, R., Grimaldi, J.C., Lund, F.E., Brannan, C.I., Copeland, N.G., Jenkins, N.A., Heath, A.W. & Parkhouse, R.M. (1993) Expression cloning of a cDNA encoding a novel murine B cell activation marker. Homology to human CD38. J. Immunol. 151, 3111–3118. 20. Jackson, D.G. & Bell, J.I. (1990) Isolation of a cDNA encoding the human CD38 (T10) molecule, a cell surface glycoprotein with an unusual discontinuous pattern of expression during lympho- cyte differentiation. J. Immunol. 144, 2811–2815. 21. Lund,F.E.,Yu,N.,Kim,K.M.,Reth,M.&Howard,M.C. (1996) Signaling through CD38 augments B cell antigen receptor (BCR) responses and is dependent on BCR expression. J. Immunol. 157, 1455–1467. 22. Han, M.K., Kim, S.J., Park, Y.R., Shin, Y.M., Park, H.J., Park, K.J., Park, K.H., Kim, H.K., Jang, S.I., An, N.H. & Kim, U.H. (2002) Antidiabetic effect of a prodrug of cysteine, L -2- oxothiazolidine-4-carboxylic acid, through CD38 dimerization and internalization. J. Biol. Chem. 277, 5315–5321. 23. Chidambaram, N., Wong, E.T. & Chang, C.F. (1998) Differential oligomerization of membrane-bound CD38/ADP-ribosyl cyclase in porcine heart microsomes. Biochem. Mol. Biol. Int. 44, 1225– 1233. 24. Khoo, K.M. & Chang, C.F. (1998) Purification and character- ization of CD38/ADP-ribosyl cyclase from rat lung. Biochem. Mol. Biol. Int. 44, 841–850. 25. Khoo, K.M. & Chang, C.F. (2000) Localization of plasma membrane CD38 is domain specific in rat hepatocyte. Arch. Bio- chem. Biophys. 373, 35–43. 26. Umar, S., Malavasi, F. & Mehta, K. (1996) Post-translational modification of CD38 protein into a high molecular weight form alters its catalytic properties. J. Biol. Chem. 271, 15922–15927. 27. Cockayne, D.A., Muchamuel, T., Grimaldi, J.C., Muller-Steffner, H., Randall, T.D., Lund, F.E., Murray, R., Schuber, F. & Howard, M.C. (1998) Mice deficient for the ecto-nicotinamide adenine dinucleotide glycohydrolase CD38 exhibit altered humoral immune responses. Blood 92, 1324–1333. 28. Lund, F.E., Solvason, N.W., Cooke, M.P., Health, A.W., Grimaldi, J.C., Parkhouse, R.M., Goodnow, C.C. & Howard, M.C. (1995) Signaling through murine CD38 is impaired in anti- gen receptor-unresponsive B cells. Eur. J. Immunol. 25, 1338–1345. 29. O’Farrel, P.H. (1975) High resolution Two-dimensional electro- phoresis of proteins. J. Biol. Chem. 250, 4007–4021. 30. Lund, F.E. & Muller-Steffner, H.M., Yu, N., Stout, C.D., Schu- ber, F. & Howard, M.C. (1999) CD38 signaling in B lymphocytes is controlled by its ectodomain but occurs independently of enzymatically generated ADP-ribose or cyclic ADP-ribose. J. Immunol. 162, 2693–2702. 31. Cavet, M.E., Akhter, S., Murtazina, R., Sanchez de Medina, F., Tse, C.M. & Donowitz, M. (2001) Half-lives of plasma membrane Na + /H + exchangers NHE1-3: plasma membrane NHE2 has a rapid rate of degradation. Am. J. Physiol. Cell Physiol. 281, C2039–C2048. 32. Hsu, Y.T. & Youle, R.J. (1997) Nonionic detergents induce dimerization among members of the Bcl-2 family. J. Biol. Chem. 272, 13829–13834. 33. Hsu, Y.T. & Youle, R.J. (1998) Bax in murine thymus is a soluble monomeric protein that displays differential detergent-induced conformations. J. Biol. Chem. 273, 10777–10783. 34. Munshi,C.,Aarhus,R.,Graeff,R.,Walseth,T.F.,Levitt,D.& Lee, H.C. (2000) Identification of the enzymatic active site of CD38 by site-directed mutagenesis. J. Biol. Chem. 275, 21566– 21571. 35. Haardt, M., Benharouga, M., Lechardeur, D., Kartner, N. & Lukacs, G.L. (1999) C-terminal truncations destabilize the cystic fibrosis transmembrane conductance regulator without impairing its biogenesis. A novel class of mutation. J. Biol. Chem. 274, 21873–21877. 36. Fan, H., Meng, W., Kilian, C., Grams, S. & Reutter, W. (1997) Domain-specific N-glycosylation of the membrane glycoprotein dipeptidase IV (CD26) influences its subcellular trafficking, biological stability, enzyme activity and protein folding. Eur. J. Biochem. 246, 243–251. 37. Franco,L.,Guida,L.,Bruzzone,S.,Zocchi,E.,Usai,C.&De Flora, A. (1998) The transmembrane glycoprotein CD38 is a catalytically active transporter responsible for generation and influx of the second messenger cyclic ADP-ribose across mem- branes. FASEB J. 12, 1507–1520. 1034 M. E. Moreno-Garcı ´ a et al.(Eur. J. Biochem. 271) Ó FEBS 2004 . proportion of the total CD38 is expressed in a Fig. 2. CD38 is found as homodimers on the surface of splenic B lymphocytes and the stability of the dimers. that CD38 is normally expressed as a noncovalently associated homodimer on the plasma membrane of B cells. Mutations that affect the stability of the CD38