Human CD38 ligand: A 120-KDA protein predominantly expressed on endothelial cells

Human CD38 Ligand A 120-KDA Protein Predominantly Expressed on Endothelial Cells' Silvia Deaglio,* Umberto Dianzani,t Alberto L. Horenstein,* Juan Emilio FernAndez,* Cees van Kooten,* Manuela Bragardo,t Ada Funaro,* Giovanni Garbarino,§ Francesco Di Virgilio,' Jacques Banchereau,* and Fabio Malavasi2*ll Human CD38, a pleiotropic molecule with ADP-ribosyl cyclase activity, regulates activation and growth of several cell types. I t s in vivo function is incompletely determined, mainly due to the lack of evidence concerning the existence of a single or multiple ligands. We recently observed that CD38 rules a selectin-type adhesion between lymphoid cells and HUVECs. A panel of murine mAbs raised against HUVECincluded one (Moon-1) constantly blocking theCD38-mediated adhesion of several cell lines to HUVEC. Tissue distribution studies and an extended immunohistochemical analysis on the majority of normal human tissues revealed that the Moon-1 molecule displays a unique pattern of expression, being present at high levels on resting and activated vascular endothelium, on the majority of monocytes, platelets, NK cells, and to a lesser extent on T, B, and myeloid cells. The Moon-1 structure of an apparent molecular mass of 120 kDa proved to be a ligand for human CD38, as inferred by the direct binding observed when using a chimeric mouseCD8a-humanCD38 (mCD8cu-hCD38) molecule as a probe in Western blot experiments. Furthermore, Ab-induced modulation experiments highlighted an association between the Moon-1 molecule and human CD38 on the surface of celllines coexpressingthe two structures, which also share a common regulation system of surface expression. Finally,direct ligation of Moon-1 on T cell lines caused a relevant increase in the cytoplasmic concentration of calcium ([Ca2+l,). The Journal of Immunology, 1996, 156: 727-734.

S

ignal transduction across the cell membrane has long been a subject of great interest: the model now accepted is an elaborate network of intracellular signaling systems that rule physiologic processes such as cell-cell communication, cell proliferation, and differentiation. T and B lymphocytes bear Agspecific receptors, whose structure and signal transduction functions have been investigated in great detail. However, there are many other receptorial structures on the surface membrane whose physiology is less understood ( I , 2). Human CD38 is a single-chain type 11 glycoprotein of 45 kDa, preferentially expressed by immature and activated cells of different lineages (3,4). Many hypotheses have been proposed about the in vivo function of this molecule, but no unifying model can be drawn. Early studies indicated that human CD38 could function as a channel for activation signals in T, B, and NK cells, as shown upon ligation by agonistic Abs (5). More recently, it was observed "Laboratory of Cell Biology,Department of Genetics, Biology, andMedical Chemistry, University of Torino, Torino, Italy; 'Department of Medical Science, University of Torino, at Novara, Italy; *Laboratory for Immunologic Research, Schering-Plough, Dardilly, France; $Institute of Internal Medicine, University of Torino; ¶Institute of General Pathology, Unwersity of Ferrara, Ferrara, Italy; and ,Ilnstitute of Biology and Genetics, University of Ancona, Ancona, Italy Received ior publication July 26, 1995. Accepted for publication October

16,

1995. The costs o i publication of this article were detrayed In part by the payment of page charges. Thls article must therefore be hereby marked advertisement in a( cordanre with 18 U.S.C. Section 1734 solely to indicate this fact.

' This work was supported in part by the AlRC and by the Projects ACRO (CNR), AIDS (Istituto Superiore di Saniti, Roma, Italy), Telethon (Roma, Italy), and Biotechnology (European Community, Brussels, Belgium). A.L.H., on leave of absence from Conicet, Argentina, was supported by the Fondarione Ghirotti and by the Fondarione CRT, Torino, Italy. J.E.F., on leave of absence from the Universidad de Cranada, was supported by the Ministerio de Educacibn y Ciencia de Espaiia. M.B. is supported by an AIDS Project Fellowship.

' Address correspondence and reprint

requests to Dr. F. Malavasi, Laboratorio Biologia Cellulare, via Santena 19, 10126 Torino, Italy. Copyright 0 1996 by The American Association of Immunologists

that CD38 is an ectoenzyme (6, 7), with ADP-ribosyl (ADPR)' cyclase and hydrolase activities, which lead to the conversion of P-NAD+ into cADPR and simultaneously to the hydrolysis of this molecule to ADPR (8). This function has biologic implications, since cADPR regulates cytoplasmic Ca2+ currents, probably being the natural ligand of the ryanodine receptor (9). Other observations indicate that CD38 is also involved in the strict regulation of life and death occurring in the bone marrow (BM) environment, where the molecule inhibits the growth of immature B cells on stroma (IO). On other cell types, i.e., germinal center cells, CD38 ligation prevents the cells from undergoing apoptosis in vitro (I I). The finding of a soluble form of CD38 suggested analogies with models derived from other surface receptors, which usually compete for ligands expressed either by homotypic or heterotypic cell populations (12, 13). Finally, recent data indicate that CD38 displays adhesion properties: it was shown that CD38 mediates a selectin-type adhesion between different blood populations andHUVEC, which do not express CD38 (14-16). The latter resultsuggested that HLJVEC could harbor a surface receptor for CD38. In this study, we characterize a novel murine mAb, Moon-1, that reacts with a single-chain 120-Kda molecule expressed by HUVEC, monocyte-macrophages, platelets, NK cells, and a panel of continuous lines. Moon-1 mAb blocks the selectin-type interactions mediated by CD38. The characterization of the Moon-I molecule included the analysis of its distribution in normal tissues, the regulation of its expression, and the immunomcdulatory efects exerted by preparations of soluble CD38. The availability of a recombinantform of soluble CD38 made itpossible to evidence the direct bindingoccumng with the Moon- 1 molecule immunoprecipitated from HUVEC.

' Abbreviations

used in this paper: ADPR, ADP-ribosyl; cADPR, cyclic ADPribosyl; BM, bone marrow; ATRA, all-trans-retlnoic acid; CaMlg, goat antirnouie Ig; hCD38, human CD38; mCD8, mouse CD8.

0022-1 767/96/$02.00

IDENTIFICATIONSURFACE OF A CELL

728

Materials and Methods Antibody production and purification A panel of murine mAbs was produced using HUVEC as an immunizer. 2 X IO5 HUVEC were resuspended in 200 p1 of saline and injected into the spleen of anesthetized female BALBk mice (22-25 g) through a direct opening of the peritoneum, under surgical conditions. The same set of mice underwent a chronic boosting, using 2 X IO5 cells injected i.v. at approximately 20-day intervals for 3 mo. Four days after the last injection, the spleen was removed for fusion with the X63-Ag8.653 cell line using PEG 1,500-1,700 (Sigma Italia, Milan, Italy). Spent media of the primary cultures were screened on HUVEC by means of indirect immunofluorescence and visual observation to select the binding Abs. The clones positive for binding were assessed for cell specificity by their reaction to PBMC preparations, The specific clones selected underwent a third test to evaluate their ability to inhibit the binding between CD38+ cells and HUVEC. The primary culture simultaneously fulfilling all the requisites of the selection grid was cloned three times by limiting dilution and subsequently grown as tumor ascites in pristaneprimed mice. The isotype of the selected mAb was characterized by Outcherlony testing, and the mAb was purified using an HPLC technique after purification on protein A-Sepharose (17).

Cell preparation HUVEC were isolated from segments of normal term umbilical cord veins (161,while PBMC were prepared as described in detail ( 5 ) . Platelets and granulocytes were separated by centrifugation of blood added with EDTA, as described (18). Fibroblasts were obtained from surgical specimens from different organs. Spleen samples from healthy adult individuals were obtained throughthe Department of Immunology and Transplantation, S. Giovanni Hospital, Torino, Italy. Preparations were done as described (5). The reactivity of the selected mAb was tested with different human continuous cell lines, including Jurkat and Molt-4 (T lymphoblastoids), HL-60 and U937 (myelomonocytoids), as well as several B cell lines. The majority of cell lines used for the experiments were cloned by limiting dilution to obtain phenotypically homogeneous populations.

Evaluation of adhesion properties mediated by CD38 HUVEC were cultured to confluence in a medium containing PMA (5 ng/ml, Sigma) for 2 days in 24-well flat-bottom plates (Nunc, Roskilde, Denmark), then HUVEC-coated wells were washed three times with RPMI 1640 10% FCS (RPMI-FCS) and treated with the appropriate mAb for 30 min in RPMI-FCS. Theplates were then incubated with 5 X Io" CD38' cells previously labeled with 100 pl "Cr (Amersham, Little Chalfont, Buckinghamshire, U.K.) for 1 h. Plates were spun at 800 rpm for 30 s and incubated at 37°C for 20 min. Wells were gently washed once with 1 ml of RPMI-FCS. Bound cells were lysed with 0.5 ml of 2% Triton X-I00 and radioactivity measured as described (1 6).

+

SDS-PAGE and Western blot analyses of the target structure The molecule recognized by the selected mAb was immunoprecipitated using 10 p g of the mAb, after '251-lactoperoxidase surface labeling of HUVEC or different cell lines (3) and lysis with 0.1% Nonidet P-40 (Sigma). The immunocomplexes were precipitated by means of immobilized Abs anti-Mouse Ig (40 pl of dry pellet, Sigma), washed five times, eluted, and run in 10% SDS-PAGE. Gels were dried and exposed at -7O"C, using Cronex Dupont (Sigma) intensifying screens (18). Plasma membrane proteins obtained from HUVEC cells were separated on 10% SDS-polyacrylamide and transblotted onto Hybond (Amersham) nitrocellulose membranes in a buffer consisting of 56 mM Tris, 39 mM glycine, 0.038% SDS, and 15% methanol. The blotted membrane stained with Ponceau S (Sigma) was blocked by incubation at 37°C for 3 h in 10 mM Tris-HC1, pH 7.6, 150 mM NaCI, 0.05% Triton X-100, and 5 % BSA (TBS-T). The membrane was extensively washed with TBS-T without BSA and then incubated with Moon-I ( I O pg/lane) for 12 h at 4°C. After washing it with TBS-T, Moon-I binding activity was visualized with alkaline phosphatase-conjugated goat anti-mouse Ig (1:2000 dilution) and nitroblue tetrazolium and bromo-chloroindolyl phosphate (Bio-Rad Laboratories, Milan, Italy).

Comparative analysis between the selected mAb and reference CD reagents Moon-1 mAb was analyzed for its ability to compete with reference anti-CD mAbs with similar but not identical reactivity, namely anti-CDw32 (Serotec, Oxford, U.K.), anti-CD62 (Serotec), anti-CD71 (19), and antiLCD98 (20). The

LIGAND OF CD H3U8M A N

tests included direct epitope competition using directly 1Z51-labeled mAbs in reciprocal combinations and using isotype-matched irrelevant mAbs as controls. Directly 1251-labeled Moon-1mAb was tested in competition with the panel of anti-HUVEC Abs from the Boston Workshop on Differentiation Antigens (kindly made available by J. Harlan, Seattle, WA). To eliminate the possibility of anti-Id Abs as a possible pitfall masking a true direct effect, the reaction between anti-CD38 murine mAbs and its putative ligand was evaluated in an artificial system relying upon the use of directly labeled mAbs (as was the case for IB4, an anti-CD38 mAb). This artificial system was used to monitor the modulation of the expression of surface receptors.

Tissue distribution The distribution of the molecules recognized by different mAbs was examined on different tissues. using freeze-dried sections of surgical specimens obtained from nonpathologic samples. Ab reactivity with the sections was assessed by the alkaline immunophosphatase and immunoperoxidase techniques, as reported in detail (21).

Modulation experiments Capping and cocapping experiments were performed using Jurkat and U937 cells, as described (22). In experiments in which differentiating and activating agents were compared, HL-60 cells were cultured in the medium in the presence or absence of all-trans-retinoic acid (ATRA) (100 nM, Sigma) and DMSO (1.3% v/v, Sigma) (23, 24).

Measurement of intracellular Ca2+ concentration Cells were loaded with the fluorescent Ca'+ indicator fura-2/AM (4 pM, Molecular Probes, Eugene, OR) at a concentration of 107/ml at 37°C. as previously described (25). After 15 min of incubation in the presence of the dye, cells were centrifuged, resuspended at the same concentration in fresh medium, and kept at room temperature until used. The cells were then resuspended in a saline buffer containing 125 mM NaCl, 5 mM KCI, 1 mM NaHPO,, 5.5 mM glucose, 5 mM NaHCO,, 1 mM CaCI,, 20 mM HEPES, pH 7.4, at a concentration of 106/ml, and placed in a thermostated (37°C) and magnetically stirred fluorometer (model LS50, Perkin-Elmer, Beaconsfield, Buckinghamshire, U.K.) cuvette, where [Caz+], measurements were performed. The mAbs tested, namely Moon-I,IB4 (anti-CD38), CBT3G (anti-CD3), and GaMIg, were added to the buffer at a concentration of 10 pg/ml. EGTA was added at a 2 mM concentration.

Construction of an expression plasmid for soluble mCD8cr-hCD38 The eukaryotic expression plasmid pME18S, which contains the strong SRa promoter (26). was used for the preparation of a construct expressing a soluble chimeric molecule of hCD38 and mCD8a. Full details on the preparation of the chimera containing the extracellular region of mCD8 and hCD38 are given in Reference 27.

Production of soluble mCD8whCD38 and control soluble mCD8a-hCD40L For production of soluble mCD8-hCD38hCD40L. COS cells (7 X lo6) were transfected with 20 pg of the expression plasmid, using electroporation (220 V, 960 millifarad; Bio-Rad). Two ampoules were mixed in a 150-cmZculture flask (Corning, Coming, NY) with 25 ml of RPMI-FCS. After 24 h the medium was removed and replaced with RPMI + 1% Nutridoma (Boehringer Mannheim, Mannheim, Germany). Supernatants were harvested 4 days later, centrifuged (10 min, 2000 rpm), and filtered on a 0.2-pm filter.

Evaluation of the receptorial properties of the Moon- 1 molecule The interaction between CD38 and the putative counter-receptor was deduced not only indirectly, but also through direct evidence. The Moon-l molecule was immunoprecipitated from unlabeled HUVEC and run in 10% SDS-PAGE in the same conditions as reported above. After blotting onto a nitrocellulose membrane, the isolated structure was incubated with soluble mCD8a-hCD38, and the binding activity was visualized with biotinilated rat anti-mCD8a (Boehringer Mannheim) and streptavidin-horseradish-peroxidase (Vector Laboratories, Burlingame, CA). Controls included a mCD8a-hCD40L chimera and the supernatants of untransfected or mock-transfected COS cells.

The Journalof Immunology

729 W.BL0T SDSPAGE

mAb

:

Moon-1 (lOpg/ml)

CD38 (10pgml)

i

CD44 (10pg/ml)

i

kD

SUPT-1i 1

kD 9766-

-

-924 -69 CI

31-

"

I 20

60

40

1

2 molwt

I

80140 120 100

7 K562

! ~

I

CD38 (10pglml)

FIGURE 2. Molecular analysis of Moon-1 molecule by irnmunoprecipitation and by Western blot. Moon-1 precipitated a single-chain structure of 120 kDa in nonreducing conditions (lane 2). Moon-1 mAb recognized a linear epitope on a target molecule of identical M,, as indicated by theresult of theWestern blot (lane I ) . Externallanes include migration markers. Gels were run in 10% SDS.

I

I

Analysis of the adhesion properties of the Moon- 7 molecule

I

CD98 (10pg/ml) 0

-46

I

21-

Moon-l(lOpg/ml)

0

45-

CD98 (1Opglml)

0

- 200

20

40

60

80

100 140 120

SUPT-1

.......................

I L

To ensure that Moon-I directly interacted with CD38, adhesion experiments were performed using dilutions of the Moon-I mAb, isotype-matched mAbs against molecules expressed by HUVEC (HLA class I, CD62, CDw32, CDI I a, CD98, CD44), or mAbs to irrelevant molecules (CD57, AB8.28). Only the Moon-I mAb blocked adhesion, in a dose-dependent fashion (Fig. I ) . By contrast, no binding inhibition was found in the CD38- cell line K562. None of the mAbs used in the panel of reagents displaying relevant or irrelevant features were able to mimic the effects exerted by the Moon-I molecule. Furthermore, the observed effect is not characteristic of a single line, and is reproducible in other CD38' continuous lines, including Raji and SUPT-I (both Moon-I-). Characterization of the target structure and comparative epitope analysis

0

1

10

50

100

1000

mAb (Irglml) FIGURE 1. Binding of SUPT-1 (CD38')andK562 (CD38-) continuous cell lines to endothelium in dynamic conditions. Only the Moon-1 mAb was able to block the binding to endothelial cells, quantitatively paralleling the amount of mAb added. CD44 and CD98, reactive with HUVEC and included as controls, did not provide apparent effects. 184 mAb (anti-CD38)provided a block of the binding of -50%. Results are expressed as relative percentage of binding (Le., relative to the absolute binding displayed by the control sample). One hundred percent relative binding is the cell binding in the absence of mAb (control binding) and corresponds to an absolute cell binding (i.e., the percentage of seeded cells that were bound to HUVEC) of 24% for SUPT-1 and K562. Representative data from nine (SUPT-1) and three (K562) experiments.

The analysis of the Moon-I molecule precipitated from HUVEC preparations indicated that the target structure is a single chain of 120 kDa in nonreducing conditions. This result was confirmed by Western blot analysis, which also indicated that the Moon-I mAb is specific for a linear epitope (Fig. 2). None of the mAbs included in the endothelial cellssection of the Boston Workshop proved to significantly inhibit the binding of '*'I-labeled Moon-I mAb, indicating that the Moon-I epitope is probably unique, although molecular identity cannot be ruled out. The hypothesis that CDw32 or CD62 may bear the Moon-I epitope was excluded by means of modulation experiments. which left unaltered the expression of Moon-I molecule, as well the antiCD71 and anti-CD98mAbs included as controls. Moon-1 mAb was also shown as nonreacting with IB4 or other anti-CD38 mAbs, excluding the possibility of anti-Id mAbs. Distribution of the Moon- 1 molecule in PBMC and BM samples and continuous cell lines

Results Antibody production and selection

Indirect immunofluorescence analysis of spent media of the primary cultures with HUVEC and PBMC enabled us to define a first panel of mAbs: out of 48 hybridoma supernatants reactive with endothelial cells, one, an lgGl mAb named Moon-I, was selected for its ability to interfere with CD38-mediated adhesion.

The Ag recognized by Moon-l mAb is expressed by the majority of monocytes, platelets, NK cells and discrete populations of peripheral blood lymphocytes.Two- and three-colorimmunofluorescenceresults indicated thatMoon-l reacts with subsets of both CD4' and CD8' T cells, as well as with most B and NK cells. In the T cell compartment, Moon-I preferentially stained CD45RA' CD45RO- (naive) cells, but not CD45RA- CD45RO' (memory) cells, a distribution strikingly similar to that of CD38 (Fig. 3). Coexpression of CD38 and Moon-l by T cells was therefore analyzedby three-color immunofluorescence.

730

IDENTIFICATION OF A CELL SURFACE LIGAND OF HUMAN CD38 u3r mmM2

U3r MoN886

U3r HlxtNw4

27

B

Moon-1expression by peripheralbloodlymphocytepopulations. PBMC werestained by three-color immunofluorescence using phycoerythrin-conjugatedCD4, or CD8, orCD38mAb,FITC-conjugatedCD45RAorCD45ROmAbs, andbiotin-labeled Moon-1 mAb plus RED61 3-conjugated avidin. The figure showsa two-color analysisof total lymphocytes ( A ) ,CD4+ cells ( B ) , and CD8+ cells ( C ) .Moon-1 displays an expression pattern resembling that of CD38. Quadrants on the right show that Moon-1 and CD38 are prevalently coexpressed in the different lymphocyte subpopulations. Representative data from four experiments. FIGURE 3.

These experiments showed that the two molecules were mostly coexpressed by discrete T cell subpopulations.Further, Moon-1 stained the majority of myeloid cells in BM samples, while CD34+ hemopoietic precursorcells proved negative. BMT lymphocytes displayed a reactivity pattern almostoverlapping with that shown by peripheral blood lymphocytes. Moon-1 mAb stained T (Jurkat and Molt-4) and myelomonocytoid cell lines (HL-60 and U937), while B cell lines were in most cases negative. Tissue distribution

Next we studied the reactivity of the Moon-1 molecule with a set of tissue sections from different normal organs. Results of analysis of lymph node (Fig. 4A), as well as a selection of tissues uniquely expressing CD38 (i.e., kidney, skin, and the gastrointestinal tract)

imply that the molecule, outside the lymphocyte compartment, behaves like structures almost exclusively restricted to endothelium, as can be seen in Figures 4B and 5. Associations between the Moon- 1 molecule and CD38

Uponfinding different cell linescoexpressing CD38 and the ligand under evaluation, the possibility of interactions taking place between the two structures at the plasma-membrane level was examined. Indeed,Jurkat mutantsexpressing low to nil amounts of surface CD38 simultaneously displayed a decreased expression of Moon-1 (Fig. 6). This idea was corroborated by the results of capping and cocapping experiments using Jurkat and U937 cell lines (TableI). Both Moon- 1 and IB4 have a high affinity for their targets and a strong modulatory power. After

The Journal of Immunology

731

A

A

B

FIGURE 4.

Analysis of the distribution of Moon-1 Ag in tonsillar lymph node section. Reactivity appears to be limited tothe cells lining the vascular structures (white arrows). The same section was doublestained with anti-CD38 (A, magnification X200). Staining of the Moon-I molecule expressed by the endothelial cells of a vessel,as seen at higher magnification (B, magnification X1000).

the cells had undergone iterative cycles of exposure to excess IF34 mAb, they were treated with biotin-labeled Moon-1 mAb, and the expression in-the polar aggregates was shown by means of tetramethylrhodamhe B isothiocyanate (TRlTC)avidin. Upon longer incubation of the cells with the excess mAb and more frequent medium replacements, stripping was observed, and-in the capping experiments-it was bidirectional, as shown after reversing the order of the capping molecule (datanot shown).

FIGURE 5. Analysisofthe distribution of Moon-1 Ag in a selected panel of nonlymphoid tissues, namely skin (A), and arteriae arciformi fromthekidney ( B ) andgastrointestinaltract (0.Moon-1staining is strictly limited to endothelial cellsof the vessels,not expressing theCD38 molecule, which in turn is detectable in the intestinal lamina propria cells. Table I. Results of the capping and cocapping experiments”

Regulation of expression

The observation that the majority of continuous cell lines displayed a heterogeneous pattern of Moon-1 expression prompted us to verify the hypothesis that the structure could be regulated in its surface expression. The treatment of normal cells and normal HUVEC with IL-6, IFN-y,and PMA was not followed by any detectable variation in terms of growth, differentiation, or expression of surface markers. We transferred this model to selected continuous cell lines, starting from theexperience derived from the CD38 model (23). When theHL-60 cell line is treated with ATRA, CD38 expression increases. We tested whether this was also the case for Moon-1. Indeed, after ATRA treatment, Moon-1 expression in HL-60increased, peaking at 48 h (Fig. 7). Moon-1 expression was not influenced by DMSO, as was the case also for CD38. Function

Jurkat T cell line was rendered phenotypically homogeneous by cloning in limiting dilution. Direct ligation of the Moon-1 molecule in Jurkat CD38+/Moon-l+ cells was followed, after a lag of approximately 1 min, by a sustained increase of [Ca2+],, with a shape that differed from that triggered via CD3 and CD38 on the

Cocapping Molecule Capping Molecule

CD38

HLA class I

Moon-1

~~

CD38 Moon-1

79

5

75

8

a Results are expressed as percentage of cells in which the events have been scored and are derived from three independent experiments.

same cells. A further [Ca’+], increase was triggered by cross-linking with GaMIg (Fig. 8). Analysis of a direct interaction between Moon-7 and soluble CD38

A formal proof came from the direct interaction between a chimeric soluble mCD8a-hCD38, used as a probe to highlight immunoprecipitated Moon-1 molecule in a Western blot system. mCD8a-hCD38 bound the immunoprecipitated Moon-1 molecule run in 10% SDS-PAGE in nonreducing conditions.The chimeric molecule stains the purified Moon-1 molecule, highlighting a single chain of 120 kDa (Fig. 9). The chimeric

732

IDENTIFICATION OF A CELLSURFACE LIGAND OF HUMAN CD38

A

FIGURE 6. One-parameterflowcytometric (FACScan) analysis of Moon-1 expression in Jurkat T cell line mutants selected by cloning in limiting dilution. Jurkat cell line selected for homogeneous expression of CD38at highepitope density and considered to be CD38+ ( A ) .Jurkat mutants were selected as expressing a very low amount of surface CD38 and considered to be CD38- ( B ) . Results clearly indicate asegregation of the expression of the Moon-1 molecule with CD38. x-axis = fluorescence intensity/cells; y-axis = number of cells registered/channel. Number of cells tested = 10,000.

B

FIGURE 7. Regulation of Moon-1 expression in HL-60 cell line. Cells were treated with 100 nMATRA and evaluated at different time intervals (0, 24,48, 72 h). Expression of CD38 was included as the control. Results clearly indicate that the modulation in Moon-1 expression parallels that of CD38. x-axis = fluorescence intensity/cells; y-axis = number of cells registeredkhannel. Number of cells tested = 10,000.

mCD8a-hCD40L, aswell as thesupernatants of untransfectedor mock transfectedCOS cells, did not show anydetectable binding. No binding was observed when the immunoprecipitatedMoon-1 sample was run in reducing conditions (not shown).

Discussion

different a

Several pieces of evidence indicate that human CD38 exerts part of its pleiotropic functions by regulating the interactions between hetero-

typic cells. Indeed, the results of Dianzani et al. clearly show that CD38 acts as a selectin, modulating the weak interactions between CD38+ cells and W E C (16). Recently, Kumagaiet al. have shown that similar interactions appear to take place in the BM, where ligation of CD38 induces suppression of B lymphopoiesis (10). On B lyminanalyzed phocytes aenvironment, in e.g., lymph node, CD38 acts like a channel, driving positive signals that permit the rescue of a significant number of cells from apoptosis. Taken together,

The Journal of Immunology

733

GaMl I

FIGURE 8. Directligation of Moon-1 molecule in Jurkat CD38' Moon-1 + cells was followed, aftera lag of -1 rnin, by a sustained increase in [Ca2'Ii levels. A further [Ca2'Ii rise was triggered by cross-linkingwith a goat anti-mouse Ab (CaMlg). Arrows indicate the addition of Moon-1,CaMlg,anti-CD38,anti-CD3, and ECTA. Cells were incubated in a fluorometer cuvette in a Ca'+-containing saline buffer at a concentration of 10"/rnl andstimulated with 10 Fdml for each rnAb. EGTA was added at a 2 rnM concentration.

A kD

B

C

"'

[x1631

". 97-

6645-

3l-

FIGURE 9. Westernblotanalysis

of theMoon-1 molecule irnrnunoprecipitatedfromHUVEC: thecomplexwas run in 10% SDS-PACE, transferred onto nitrocellulose, and successively incubated with mCD8ahCD38, mCD8a-hCD40L, and supernatantsfrom mock-transfected COS cells. The reactivity was verified by means of rat anti-mCD8, labeled with biotin, and the complex shown by streptavidin-horseradish peroxidase plus specific substrate. even if different in nature, these results confirm that human CD38 rules cell-cell interactions in a so far undefined way (27). A panel of murine mAbs raised against HUVEC included a reagent able to block the binding of CD38' cells to endothelium. The molecule identified by this mAb, namely Moon-I molecule, is expressed by HUVEC, the majority of monocytes, platelets, and NK cells, as well as subpopulations of normal T and B lympho-

I.

cytes. The expression of the Moon-1 molecule is an event maintained during tumoral transformation, as shown by its presence on leukemic cell lines such asJurkat, HL-60, U937. The structure appears to be a single chain of I20 kDa, as shown by the results of SDS-PAGE and Western blot analyses. Importantly, among the mAbs tested, Moon-I mAb was the only one able to interfere and reduce significantly the CD38-mediated adhesion. The vast array of challenging and isotype-matched mAbs did not show any appreciable or reproducible effect. The analysis of the distribution through the majority of normal human tissues is compatible with the notion that the Moon-I molecule is a counter-receptor for human CD38. Furthermore, the Moon-I molecule also displays a unique tissue distribution pattern, to which nothing similar has been found among the known CDs.CDw32 and CD62, even though reminiscent of Moon-I, did not block each other. Second, a formal evidence ofan interaction between the two molecules was confirmed by the results of the Western blot analysis, where the immunoprecipitated Moon-I molecule was probed with mCD8ahCD38, mCD8a-hCD40L. and supernatants from untransfected or mock-transfected COS cells: only the soluble recombinant CD38 stained the 120 kDa. The same soluble recombinant CD38 that was used for the Western blot testing failed to react in vivo or to visibly stain the cells expressing the Moon-l molecule. This negative result, however, could appear for different reasons, mainly theexpression of the molecule in a recombinant system. Evidence supporting this view was found in the experience gathered with CD40 and CD4OL (28). At the moment, this unanswered issue is approached by trying to define a source of native soluble CD38 for use in the staining and functional analysis of the ligand.

734

IDENTIFICATION OF A CELL SURFACE LIGAND OF HUMAN CD38

An intriguing observationis that manycell lines testedso far for the expression of Moon-1 produced variable surface molecule densities, ranging from almost negativeto highly positive membranes.This behavior is another feature that is shared with CD38. In some experiments, this proved to be a shortcoming and hampered the general performance of the tests.This inconveniencewas bypassedby cloning the cell linesto homogeneity. What we observed is that the expression of Moon- 1 seems to parallel thatof CD38 in manyaspects, supporting the view of a common regulation. Alternatively, the same observations of a variable expression could be taken as an indication that regulatory elements are active either during cell cycle or differentiation. The latter hypothesis is under scrutiny at the moment. In addition to distribution, an interesting feature of this molecule concerns the regulation of its expression. This wasseenmainly by exploiting functional testsperformed on HL-60 treatedwith DMSO and ATRA, which actedas differentiating agents.The results indicate that thereis an increasedexpression of the Moon-I molecule upon treatment with A m , whereasDMSOisineffective. These findings indicate thatCD38 and its ligand are not only often expressed by the same cells, but appear to be under the same regulatory influences, at least in certain cell lines. For example, the effectsof ATRA on Moon-] expression are almost identical to those reported for CD38 in myeloid cells, where either in vitro or in vivo only ATRA induces CD38 mRNA transcription and protein synthesis. Moreover, when Moon-I and CD38 are coexpressed, theyappeartobe associated. However, this is a special situation, since in the overwhelming majority of tissues analyzed, Moon-l is expressed alone and constantly marks the cells lining the vessels. Lateral association appears to be a unique feature of the lymphatic and myeloid cells. All these results are similar to those observed for other models of receptors cross-talking among different cells (2); this scenario is further supported by recent evidence that Moon-l induces a significant increase in the cytoplasmic calcium levels. This feature suggests an ability of the Moon-1 moleculeto transduce extracellular signals, this being another characteristic that is shared with CD38 (29). An alternative hypothesis is that Moon-1 belongs to the family of CD38-related molecules, linked together by molecular and functional homologies. This can positively be excluded just by considering the M , and the tissue distribution features, which are different from those reported for human and mouse BST-1 (30, 31), as well as BP-3 (32). However, the full molecular and structural characterization of this molecule will definitively exclude the possibility of having identified another member of the CD38 family.

Acknowledgments The authors thank Dr. D. Campana (St. Jude Children’s Hospital, Memphis, TN) for an accurate and critical review of the manuscript, and Becton Dickinson Italia for making FACScan equipment available. This work is dedicated to the memory of Mina R. and Virginia L.

References I . Clark, E. A., and J . A. Ledbetter. 1994. How T and B cells talk to each other. Nature 367:425. 2. June, C. H., J. A. Bluestone, L. M. Nadler, and C. B. Thompson. 1994. The 8 7 and CD28 receptor families. lmmunol. Today 15:321. 3. Terhorst, C., A.Van Agthoven, K. LeClair, P. Snow, E. L. Reinherz, and S. F. Schlossman. 1981. Biochemical studies of the human thymocyte cell-surface antigens T6, T9 and T10. Ceii 23.771. 4. Malavasi, F., F. Caligaris-Cappio, P. Dellabona, P. Richiardi, and A. 0. Carbonara. 1984. Characterization of a murine monoclonal antibody specific for human early lymphohemopoietic cells. Hum.lmmunol. 9 9 . 5. Funaro, A,, G. C. Spagnoli. C. M.Ausiello, M. Alessio, S. Roggero, D. Della, M. Zaccolo, and F. Mdavasi. 1990. Involvment of the multilineage CD38 molecde in a unique pathway of cell activation and proliferation. J. lmmunol. 14S:l3YO. 6. Howard, M., 1. C. Grimaldi, 1. F. Bazan, F. E. Lund, L. Santos-Argumedo,R. M. E. Parkhouse, T. F. Walseth, and H. C. Lee. 1993. Formation and hydrolysis of cyclic ADP-ribose catalyzed by lymphocyte antigen CD38. Science 262:1056.

7. Zocchi, E., L. Franco, L. Guida, U. Benatti, A. Bargellesi. F. Malavasi, H. C. Lee, and A. DeFlora. 1993. A single protein immunologically identified asCD38 displays NAD‘ -glycohydrolase, ADP-ribosyl cyclase and cyclic ADP-ribose hydrolase activities in human red cells. Biockem. BiqJkw Res. Commun. IY6:/4SY. 8. Lee.H.C., T. Walseth, G. T. Bratt. R. N.Hayeh, and D.L. Clapper. 19x9. Structural determination of a cyclic metabolite of NAD+ with mtracellular Ca’+mobilizing activity. J. Biol. Ckem. 264:160S. 9. Galione, A., H. C. Lee, and W. B. Busa. 1991. Ca*+-induced Ca” release in sea urchin egg homogenates: modulation by cyclic ADP-ribose. Science 253.1143. 10. Kumagai, M., E. Coustan-Smith, D. J. Murry, 0. Silvennoinen, K. G. Murti, W. E. Evans, F. Malavasi, and D. Campana. 1995. Llgation of CD3X auppreshes human B lymphopoiesls. J . Exp. Mrd. 18/:1101. I I . Zupo, S., E. Rugarli, M. Dono, G. Tamborellt, F. Malavasi, and M. Ferrarin~. 1994. CD38 s~gnallingby agonistic monoclonal antibodies prevents apoptosi\ of human germinal center B cella. Eur. J. lmmunol. 24:121X. 12. Malavasi, F., A. Funaro, S. Roggero, A. L. Horenstein, L. Calosso, and K. Mehta. 1994. Human CD38: one molecule in search of a function. lmmunol. Toduy 1595. 13. Fundro, A.. S. Roggero, A. Horenstein, L. Cnlo~so,U. Dianzani, L. B. DeMonte, E. Zocchi, L. Franco, L. Guida, C. M. Ausiello, J. Drach, K. Metha, A. Bargellesi, and F. Malavasi. 1994. Human CD38: a transmembrane glycoprotein with pleiotropic ectoenzyme functions. In Leukocyte Typing V: White Cell Difermtiation Antigens. S. F. Schlossman, L. Boumsell, W. Gilks, 1. Harlan, T. Kishimoto, C. Morimoto, J. Ritz, S. Shaw, R. Stlverstein, T. A. Springer, T. Tedder, and R. Todd, eds. Oxford University Press, Oxford. United Kingdom, p. 380. 14. Van der Menve, A. P., and A. N. Barclay. 1994. Transient intercellular adhesion: the importance of weak protein-protein interactions. Trends Bid. Sci. 1Y:354. 15. Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyte migration: the multistep paradigm. Cell 76.301. 16. Dianzani U.. A. Funaro, D. DiFranco, G. Garbarino, M. Bragardo, V. Redoglia, D. Buonfiglio, L. DeMonte, A. Pileri, and F. Malavasi. 1994. Interaction between endothelium and CD4+CD45RA+ lymphocytes: role of the human CD38 molecule. J. Immunol. 154:Y52. 17. DeMonte. L. B., P. Nisticb, R. Tecce, P. Dellabono, M. Momo, A. Anichini, M. Mariani, P. G. Natali, and F. Malavasi. 1990. Gene transfer by retrovirusderived shuttle vectorqin the generation of murine bispecific monoclonal antibodies. Proc. Nafl. Acad. Sci. USA 87:2Y4/. 18. Alessio, M., S. Roggero, A. Funaro, L. B. DeMonte. L. Peruzzi. M. Geuna. and F. Malavasi. 1990. CD38 molecule: structural and biochemical analysis on human T lymphocytes, thymocytes and plasma cells. J. lrnmunol. 145:878. 19. Peruzzi. L., G. Melioli, L. B. DeMonte, M. Alessio, M. Zaccolo, S . Roggero. G. Bene, A. Funaro, and F. Malavast. 1989. Microplate selection technique (MPST): A new method for selecting mouse transfectants expressmg human gene products. J. Immunol. Methods 123:113. 20. Bellone, G., G. Alloatt~,R. Levi. M. Geuna, C. Tetta, L. Peruzzl, M. Letate, and F. Malavasi. 1989. Identification of a new epitope of the 4MI44D7 molecular complex present on sarcolemma and isolated cardiac fibers. Eur. J. ltnmrmol. 19.1. 21. Fernandez, J. E., A. Concha, A. Aranega, F. Ruiz-Cabello, T. Cabrara, and F. Garrido. 1991. HLA Class I and 11 expression in rabdomyosarcomaa. Immunohiol. 182:440. 22. Funaro, A,, L. B. DeMonte, U. Dianzanl, M. Forni, and F. Malavasi. IYY3. Human CD38 is associated to distinct molecules which mediate transmembrane signaling in ditferent lineages. Eur. J. lmmunol. 23:2407. 23. Drach J., S. Zhao, F. Malavasi, and K. Mehta. 1993. Rapid induction of CD38 antigen on myeloid leukemia cells by all trrms-retinoic acid. Biockem. Biophw Res. Commlm. 195:545. 24 Kontani, K., H. Nishina, Y. Ohoka, K. Takahashi, and T. Katada. 1993. NAD glycohydrolase specifically induced by retinoic acid in human leukemic HL-60 cells. J. B i d . Ckem. 268:16RYS. 2s Di Virgllto, F., T. H. Steinherg, J. A. Swanson, and S. C. Silverstein. 19x8. Fura-2 secretion and sequestration in macrophages. J. Immunol. 140:YlS. 26 Takebe, Y., S. Motoharu, J. Fujisawa, P. Hay, K. Yokota, K. Arai, M. Yoshlda, and N. Arai. 1987. Sra promoter: an efficient andversatile mammalian cDNA expression system composed of the simian virus 40 early promoter and the R-US segment of human T-cell leukemia virus type one long terminal repeat. Md. cell. Biol. 8.466. 27 van Kooten, C., C . Gaillard, J . P. Galizz~,P. Hermann, F. Fosslez. J. Banchereau, and D. Blanchard. 1994. B cells regulate exprecsion of CD40L on activated T cells by lowering the mRNA level and through the release of soluble CD40. Eur. J. Immunol. 24:787. 28 Ausiello, C. M., F. Urhani, A. LaSala. A. Funaro, and F. Malavasi. 1995. CD38 ligation induces discrete cytokine mRNA expression in human cultured lymphocytes. Eur. J. Immunol. 25:1477. 29 Kirkham, P. A,, L. Santos-Argumedo. M. M. Hamett. and R. M. E. Parkhouse. 1994. Murine B-cell activation via CD38 and protein tyrosine phosphorylation. ImmunokoRy 83:513. 30 Itoh, M., K. Ishihara, H. Tomizawa,H. Tanaka, Y. Kohune, J. I s h h w a . T. Kaisho, and T. Hirano. 1994. Molecular cloning of murine BST-1 having homology with CD38 and Apl.ysia ADP-ribosyl cyclase Biockem. Biwkys. Res. Commun. 203:1309. 31 Kaisho, T., J. Ishikawa, K. Oritani, J. Inarawa, H. Tom~zawa,0 . Muraoka. T.Ochi, and T. Hirano. 1994. BST-I, a surface molecule of bonemarrow stromalcelllinesthat facilitates pre-B-cellgrowth.Proc. Nut!. Acnd. Sci. USA 91:5325. 32 Dong, C., J. Wang, P. Neame, and M. D. Cooper. 1994. The murineBP-3 gene encodes a relatlve of CD38/NAD glycohydrolase family. lnf. lmmutlol. 6:1353.