The endocytosis of the human CD38 molecule has been investigated in normal lymphocytes and in a number of leukemia- and lymphoma-derived cell lines. CD38 internalization was followed using radioiodinated Abs in an acidic elution endocytosis assay to monitor the effects of cross-linking on internalization processes and to quantify the ratio of the internalized molecule. Second, conventional, confocal, and electron microscopies were used to evaluate the morphologic effects induced by ligation of the molecule with Abs mimicking the natural ligand(s). The results demonstrated that internalization is a reproducible phenomenon following CD38 ligation with both agonistic and nonagonistic specific Abs and involving only a fraction of the entire amount of the surface molecule. It is independent from signal transduction as can be inferred by the observation that 1) both agonistic and non agonistic Abs are effective and 2) the dynamic of internalization is much slower than that of cellular signaling. Morphologic studies demonstrated that endocytosis induced as a result of CD38 ligation presents a very specific pathway consisting of subcellular organelles fundamental to the processing of the complex. Our data indicate that down-regulation by endocytosis may be, in parallel with shedding, a regulatory element in activation and adhesion processes mediated by CD38. However, internalization seems not to be a key step in triggering intracellular signaling; more likely, it is a negative feedback control mechanism which interrupts signal transduction or cell-cell cross-talks mediated by membrane CD38.
Human CD38 is a type II glycoprotein expressed by many cell types playing a role in cell activation and proliferation processes in mature lymphocytes (1). Furthermore, CD38 is a bifunctional ectoenzyme the extracellular domain of which bears complex catalytic properties (2); it also is involved in cell-cell interactions (3). Recent findings indicate the existence of a soluble form of the molecule that is spontaneously released from cell membrane (4). A specific ligand for CD38 has also been identified (5). The functions of CD38 probably depend on the ratio among the different forms of the molecule (i.e., transmembrane, intracellular, and extracellular soluble). Therefore, the definition of the regulatory processes that control CD38 surface expression, internalization, and shedding, on the one hand, and the identification of the role of other surface receptors that are physically and functionally related to CD38, on the other hand, are fundamental issues in the elucidation of the biologic functions of the molecule in vivo.
Many receptors undergo rapid internalization after interaction with a ligand, forming a receptor-ligand complex (6, 7); for selected membrane proteins, internalization is constitutive and may depend on the association with adaptor proteins of the coated vesicles with aromatic amino acid residues to the cytoplasmic domain of the receptor (8, 9). Several receptors may follow both constitutive and ligand-induced internalization (10, 11). In this case, the binding by the specific ligand or by agonistic Abs enhances the internalization events and can alter the pathway of internalized molecules from recycling to lysosomal degradation. Receptor internalization is characterized by distinct feature for each type of receptor. Molecules that are characterized by entry of nutrients inside the cell, e.g., transferrin receptor (12), usually recycle to the cell surface; other receptors, e.g., epidermal growth factor and insulin receptors, internalize after ligand binding, which involves protein tyrosine kinase activation (13, 14), and are usually degraded in lysosomes. This pathway represents a control mechanism that abolishes cellular signaling mediated by a membrane receptor.
The aim of this work is to provide an answer to the basic question as to the function of CD38, an ectoenzyme the substrate of which lies inside as well as outside of the cell (even if in minimal amounts), while the final product of reaction is used within the cell. The model proposed to solve this topologic paradox includes internalization as a physiologic step in the function of the molecule, induced either by soluble substrates and molecules or by surface ligands. This work hypothesis was assessed functionally by using 125I-labeled anti-CD38 mAb to monitor the effect of cross-linking on internalization processes and to quantify the ratio of internalized molecule. A consequent step was the use of confocal and electron microscopy to evaluate the morphologic effects induced by mAb mimicking the natural ligand(s) of the molecule.
Materials and Methods
Cell and cell lines
Jurkat, Supt-1, and HPB-ALL human T cell acute lymphoblastoid leukemia lines, Raji (Burkitt’s lymphoma), U937 (promyelocytic leukemia), KM-3 (pre-B cell leukemia), RPMI-8402 (pre-T cell leukemia), and NALM-1 (uncommitted leukemia) were cultured in RPMI 1640 medium supplemented with 10% FCS and antibiotics and were maintained at a concentration never exceeding 106 cells/ml. Selected experiments were performed on PBMC obtained from blood donors and purified by Ficoll-Hypaque gradient (Pharmacia, Uppsala, Sweden). MLC was prepared by mixing equal numbers of PBMC from different blood donors and was used for selected experiments 4 to 6 days after the beginning of the coculture.
Antibodies and reagents
The experiments were performed with affinity-purified anti-CD38 mAb IB4 (15) and IB6 (16) and with anti-CD3 mAb CBT3G (17) all developed in the laboratory. Cross-linking experiments were conducted with affinity-purified goat anti-mouse Ig (GaMIg).3 To analyze the role of acidic compartments in the CD38 intracellular transport, selected internalization experiments were done in the presence of lysosomotropic agents (e.g., 50 mM ammonium chloride and 100 μM chloroquine (Sigma Italia, Milan, Italy)) and the carboxylic ionophore monensin (Sigma Italia). Cells were treated with 125I-labeled mAb, washed, and resuspended in culture medium in the presence of 10 to 50 μM monensin 15 min before internalization.
Internalization assays were performed using 5 × 105 cells per assay and by binding the surface CD38 with 125I-labeled IB4- and IB6-specific mAb or control mAb (anti-CD3) at 4°C in the culture medium. pH was adjusted to 7.4 with 20 mM HEPES. Unbound radioactivity was removed by three rounds of centrifugation in cold medium. After incubation at 37°C, cell-associated radioactivity (which represents both membrane-bound and internalized mAb-CD38 complexes) was quantified in a gamma counter.
Stripping assay procedure
Acid stripping of cell surface-bound 125I-labeled IB4 (or control mAb) was performed in cold stripping medium (50 mM glycine-HCl buffer, pH 3.0, or, in selected experiments, RPMI 1640 medium with 10% FCS and 20 mM HEPES, pH 2.0). Cells were incubated for 5 min at 4°C with 100 μl of the above buffer. The stripping reaction was then stopped by the addition of 1 ml of cold RPMI 1640 medium, pH 7.2, followed by two washes in the same medium. This procedure removes 92 to 96% of total membrane-bound radioactivity, without significantly altering cell viability or surface Ag expression. All samples were assayed in triplicate. The percentage of internalized radioactivity was determined at indicated times from the ratio between acid-resistant counts per minute and total cell-associated specific counts per minute.
Cells were incubated with labeled mAb as described above. After 45 min of incubation at 4°C, affinity-purified GaMIg was added (final concentration, 1 to 5 μg/ml) maintaining the cells at 4°C for 20 min. The cells were successively centrifuged and washed twice with medium, and internalization assays were conducted as previously described.
Cells (2 × 105) were incubated with saturating amounts of phycoerythrin (PE)-mAb for 30 min at 4°C, washed twice with RPMI 1640 medium, and incubated for different intervals at 37°C. The cells were then washed twice with PBS containing 0.2% BSA and fixed in 4% paraformaldehyde, pH 7.4. The distribution of the cell surface and cytoplasmic molecules was determined by means of immunofluorescence microscopy. The average number of CD38 molecules per cell was assessed by means of a DAKO Fluorospheres kit (DAKO, Glostrup, Denmark). This method makes it possible to transform directly the relative channel number obtained by flow cytometry analysis of a cell population into the number of molecules of equivalent soluble fluorochromes (MESF). The linear regression equation correlating the channel number with the specific MESF value was calculated with a specific software (TallyCall Program) (18).
Confocal microscopy studies
Confocal microscopy studies were performed on 2 × 105 cells incubated for selected times at 37°C with saturating amounts of IB4 (or control mAb), in RPMI 1640 medium with 10% FCS. The cells were then briefly rinsed with cold PBS-0.1% BSA at 4°C and fixed at 4°C for 5 min in PBS containing 4% paraformaldehyde. After one wash in PBS-BSA, cells were permeabilized in PBS-BSA containing 0.05% saponin (permeabilization buffer) for 15 min at 37°C and successively incubated with FITC-conjugated GaMIg for 45 min at room temperature. All samples were then washed three times in permeabilization buffer and once in PBS-BSA and resuspended in mounting medium consisting of 50% glycerol in PBS. This medium was previously shown to preserve three-dimensional cellular organization (19). A confocal laser scanning microscope from Wild Leitz was used for the analysis. Fluorescein was excited by an argon laser coupled to a scanning and detection unit that allowed us to divide cells into a series of 10 to 12 vertical 0.8-μm optical cuts, corresponding to different cellular localizations from the membrane to intracytoplasmic compartments.
The internalization and intracellular movement of membrane CD38 were investigated at ultrastructural level improving a pre-embedding method (20, 21, 22). Briefly, growing 107 cells were incubated with anti-CD38 mAb (or isotype-matched control mAb, 5 μg/ml) for 10 min at 37°C under conventional culture conditions. The samples were carefully washed in warm culture medium and treated with peroxidase-conjugated GaMIg for 30 min under the standard culture conditions. Selected experiments were performed using IB4-biotin and streptavidin-peroxidase as a second step. At the end of the treatments cells were washed twice with PBS and fixed for 20 min at 4°C in PBS containing 2% paraformaldehyde, 0.5% glutaraldehyde, and 2% sucrose. Samples were then incubated with a substrate for peroxidase detection (0.05 mg/ml diaminobenzidine in 0.05 M Tris-HCl buffer, pH 7.6, containing 0.015% H2O2) for 10 min at room temperature, postfixed with 1% OsO4 for 1 h, dehydrated, and embedded in Spurr’s medium. Thin sections, light counterstained with lead citrate, were observed with a Zeiss 109 transmission electron microscope.
The direct interaction between CD38 and its ligand (CD38L) was assessed by following the internalization of the vesicles expressing CD38 in a Jurkat cell line selected by limiting dilution as CD38++/CD38L−. CD38L was transfected in mouse fibroblasts (5) and used as a source of signals without any background from different human surface molecules. CD38L-transfected and untransfected fibroblasts and Jurkat cells were mixed (10:1) in RPMI 1640 medium and centrifuged (1 min at 4°C, not exceeding 200 × g) to allow the formation of heteroconjugates. The cells were subsequently incubated for 5, 45, and 90 min at 37°C. The pellets were then prepared for the pre-embedding electron microscopy analysis by treatment with 0.04% saponin, a mild cell-permeabilizing agent that allows entrance of the anti-CD38 mAb and the GaMIg for the identification of vesicles expressing CD38.
Endocytosis of CD38 on CD38+ cell lines
The analysis of the fate of CD38 molecule after ligation was performed by means of an Ab-based endocytosis assay (23) in which cell surface CD38 was labeled at 4°C with 125I-IB4 or IB6 anti-CD38 mAb. After mAb binding, the temperature was increased to 37°C to allow CD38 to internalize by restoring membrane fluidity. Subsequently, the total cell-associated radioactivity was measured or, alternatively, the internalized radioactivity was determined at different intervals using an acid-stripping assay that almost completely removed surface-bound 125I-mAb. The analysis of CD38 endocytosis was performed prevalently on Raji B lymphoma and Jurkat T leukemia and derived clones selected for expressing high levels of surface CD38. The results of a representative endocytosis experiment with Raji B cell line are shown in Figure 1. Internalization is a reproducible phenomenon implemented by engagement of the molecule; it occurs minutes after receptor binding, with progressive increase in time. It was observed that ∼30% of radioactivity is internalized 6 h after the beginning of the experiment. Similar results were obtained with Jurkat T cells (Fig. 2) and confirmed in Supt-1 and HPB-ALL cell lines (data not shown). Identical results were obtained using agonistic (IB4) or nonagonistic (IB6) mAb reacting with distinct but partially overlapping epitopes of the CD38 molecule (24). The internalization kinetics of CD38 was delayed when compared with endocytosis via the CD3-TCR complex (Fig. 2). The highest internalization rate in T cells was ∼0.3% of membrane molecules per min at the beginning of the experiment: further, no plateau was observed at least after 6 h. Cross-linking of mAb with a secondary Ab significantly increased the internalization rate on the Jurkat line. The initial rate was 0.5% per min; 2 h of incubation at 37°C yielded 40% of radioactivity to be internalized (Fig. 2). A key question to be answered concerns the kinetics of internalization and signaling on the same cell samples. The mobilization of cytosolic Ca2+ was analyzed after CD38 ligation on a Jurkat cell clone selected for expressing high amounts of the molecule and loaded with Fluo 3-AM indicator. The results obtained indicate that CD38 ligation by the agonistic IB4 mAb gives rise to an intracytoplasmic Ca2+ flux which peaks ∼2 to 3 min after the beginning of the experiment, decreasing thereafter (Fig. 2, bottom).
CD38 internalization is a general feature of the molecule in both tumor and normal cells. The results obtained on alloactivated peripheral lymphocytes (Fig. 3) suggest that CD38 internalization is a physiologic phenomenon taking place on normal and neoplastic cells constitutively expressing CD38.
To evaluate the fate of the 125I-IB4/CD38 complexes after internalization, experiments were conducted in which internalization was analyzed as a function of incubation intervals in the presence (or absence) of 50 μM monensin. This antibiotic exerts blocking effects on processing events that take place mainly on the trans-Golgi cisternae apparatus and associated compartments (e.g., lysosomes and endosomes). This blocking allows endocytosis of cell surface structures but blocks processing in internal acidic compartments and exocytosis (25), without interfering with activities taking place at the cell surface. First, cells were treated with monensin, and no changes in the structure of the molecule (at least concerning mAb binding to CD38) were detected (data not shown). Next the internalization pathway of CD38 was analyzed in the presence of the antibiotic. Jurkat (and Supt-1) cells were labeled with 125I-IB4 or 125I-IB6 mAb, and internalization was induced by incubating the samples at 37°C in the presence of monensin for different intervals. The acid-resistant radioactivity was evaluated at each incubation time and compared with control samples. Following treatment with monensin, the internal radioactivity increased in both cell lines. The effect is time dependent: >100% increase was observed after 4 h of incubation at 37°C, whether using IB4 or IB6 mAb (Fig. 4).
Then, the internalization pathway of CD38 after mAb binding was analyzed in the presence of ammonium chloride and chloroquine. Such lysosomotropic agents increase the pH of endosomes and lysosomes without interfering with the internalization itself and block the recycling and degradation of the internalized molecules in lysosomes. Incubation in the presence of 50 mM NH4Cl or 100 μM chloroquine increased the internalized fraction of CD38 on Jurkat cells (Fig. 5). Similar results were obtained with HPB-ALL cells (data not shown). The effects of ammonium chloride were specific for CD38, given that control experiments with CBT3G (anti-CD3) were not followed by increased fraction of internalized CD3-TCR complex.
Analysis of CD38 internalization by direct immunofluorescence
Internalization of membrane CD38 inside tumor cell lines and normal activated lymphocytes was further investigated by direct immunofluorescence. To examine whether selective cross-linking of CD38 molecule mediates internalization of the receptor-ligand complexes, cells were stained at 4°C with PE-IB4 mAb and then incubated at 37°C for 15 min, 1 h, 2 h, 4 h, and 6 h; fixed with paraformaldehyde; and examined by immunofluorescence microscopy. Figure 6 reports the homogeneous distribution of surface CD38 on Supt-1 cells observed immediately after ligation (first panel); 15 min after the incubation at 37°C in the presence of IB4 mAb, the fluorescence displayed a partial clumping which progressively increased thereafter, while a significant amount of fluorescence clustered within the cytoplasm 2 h later. Similar results were obtained with a panel of anti-CD38 mAb in different cell populations selected according to different surface molecule density as determined in terms of MESF numbers (data not shown).
Confocal microscopy analysis
To exclude the possibility that the clumped CD38 was present on the cell surface and not inside the cytoplasm, the intracellular localization of the molecule was further investigated by confocal laser scanning microscopy. This method enabled us to differentiate between fluorescent signals from plasma membrane and those from cytoplasm. T cell blasts were incubated with saturating amounts of selected mAb (30 min at 4°C), successively incubated at 37°C for different intervals, washed, fixed with 4% paraformaldehyde, and permeabilized. A FITC-anti-mouse Ig was used to localize intracellular CD38. Cells maintained at 4°C were homogeneously stained on the entire membrane area; incubation in medium at 37°C brought a surface patching of the fluorescence, which was visible after a few minutes; 30 min later the fluorescence was visible inside the cells in the perinuclear area, increasing in direct relation to incubation time (Fig. 7).
Transmission electron microscopy was selected to trace the fate of CD38 molecule after Ab binding at an ultrastructural level. The first conclusion inferred from such analysis is that the molecule is expressed at high epitope density on the surface of pre-B (KM-3), pre-T (RPMI-8402), and uncommitted leukemias, e.g., NALM-1. All of these different lines display a uniform surface distribution of the molecule without evidence of peculiar areas of clustering. This situation persists as long as the samples are maintained at 4°C under the conventional culture conditions (Fig. 8,a). The translocation of the CD38 molecule was monitored after binding at 37°C by the IB4 mAb, selected because of its ability to mimic the interactions taking place with the natural ligand. The dynamics of cellular movements were followed stepwise by using the mAb. The results indicate that the CD38 molecule is internalized after incubation with mAb and cross-linking with a GaMIg. Indeed, the ligation of the molecule induces CD38 to cluster rapidly in specialized regions of the cell membrane, which are morphologically similar to coated pits (Fig. 8, b–e). These regions progressively internalize and lead to the formation of basket-like structures (Fig. 8,b). The sequence of the events described is accomplished in ∼10 min. Successively, the coated pits pinch off to form coated vesicles containing the CD38/mAb complexes, which migrate to the acidic cytoplasmic compartment. During this late step, the complex can be considered as a peripheral endosome that appears beneath the plasma membrane (Fig. 8, d–f). At this stage, the endosome containing the receptor/mAb complexes changes its morphology, developing a multivesicular body (late endosome) with compartment of uncoupling of receptor and ligand (CURL) shape. Figure 9,a shows a multivesicular body, where several CD38+ vesicles inside the body are clearly apparent. The CURL-like endosome reported in the picture shows membranes that have acquired different staining characteristics and have a tendency to open (arrows). The membrane-linked CD38/mAb complexes are less defined, appearing as multimeric aggregates. Figure 9,b depicts the presence of two late CURL-like bodies (arrows) with different stainings. The larger one appears as dumping out the CD38/mAb complexes (head arrows) in an area surrounding the microfilaments (thin arrows). The smaller endosome contains residual vesicles, and the immune complexes are less apparent. The ultrastructural analyses reported in Figures 8 and 9 demonstrate the existence of an internalization process of the CD38/mAb complexes. Intracellular movements can be visualized in defined sequence; indeed, early endosomes are strongly CD38 positive with a homogeneous positivity linked to the inner compartment. Late endosomes first display the typical shape of CURL vesicles, whereas stained complexes are later reversed in the cytoplasmic compartment. The final stages of the process are characterized by a faint appearance of complexes, mostly linked to CURL-like bodies. At the end of internalization process, the vesicles emerge as bodies without detectable evidence of the CD38/mAb complexes (long arrow in Fig. 9 c). Control experiments done with irrelevant isotype-matched murine mAb did not induce any of the morphologic variations described above
CD38 internalization after direct binding of CD38L
The interactions obtained by cocentrifuging at low speed human Jurkat cells expressing high amounts of CD38 and murine fibroblasts expressing CD38L were followed by morphologic effects visible with different amplitudes at 5-, 45-, and 90-min intervals. The effects became clearly apparent 45 and 90 min from the receptor/ligand interaction. The low amount of CD38 visible inside the internalized vesicles is likely attributable to the specific technical conditions (signal induced by mAb binding vs receptor/ligand interaction) and the need of a permeabilization step of the membranes to allow the penetration of the tracking mAb.
Notwithstanding such limitations, the results obtained clearly indicated that there is a process of internalization from the surface to the cytosol (Fig. 10), at least in the time intervals considered for the experiments. None of such effects was detected in the same Jurkat cell line exposed to untransfected fibroblasts (data not shown).
Endocytosis, recycling, and metabolism of lymphocyte surface molecules play a key role in the definition of mechanisms responsible for cell activation and immunoregulation. Several receptor molecules are constitutively internalized, such as CD3/TCR complex and surface IgG (26, 27). On the contrary, other molecules, e.g., CD45, fail to internalize after ligation with appropriate mAb or the natural ligand; a third group of molecules undergoes internalization only after interaction with the ligand (28). To complicate things even further, the intracellular fates of surface molecules after internalization are also divergent, reflecting the variety of functions exerted by different molecules and often depending on the cell type (29).
In spite of the vast body of data on the structure and the multiple functions of the CD38 molecule, including proliferation (30, 31), lymphopoiesis (32), apoptosis (33), adhesion, and cytokine production (34, 35), its physiology and metabolism remain unknown. To this aim, we investigated the fate of CD38 after ligation, using specific mAb to mimic receptor-ligand interactions. We used different technical approaches to determine whether any endocytosis takes place after CD38 ligation and consequently to characterize its intracellular pathway. Along with shedding (4), internalization represents another mechanism of down-regulation of CD38 membrane expression. Indeed, internalization is a reproducible phenomenon following CD38 ligation with agonistic (IB4) or nonagonistic (IB6) specific mAb, and it is shared by all the leukocytes analyzed thus far; moreover, it is independent on the amount of CD38 molecules constitutively expressed by different cells. The kinetics of CD38 internalization is similar to that of CD23 on B cells (11) and appears to be slower than ligand-induced internalization of several other membrane proteins (36). Furthermore, CD38 internalization never involves the entire amount of surface molecules; on the contrary, the internalized fraction represents an almost constant percentage (30–40%) of the total amount of surface molecules. This observation suggests that, as described for other surface molecules such as CD22 (10), two pools of CD38 molecules may exist on the cell surface, one of which may undergo internalization after Ab binding. The two pools may differ in structural conformation, as it might be suggested by the recent report of a high m.w. form of CD38 (37), or in interaction with other still unknown membrane molecules. Cloning and sequencing showed that CD38 has a very short cytoplasmic tail lacking conventional internalization signals and potential phosphorylation sites (38). A reasonable hypothesis is that internalization may depend on the interaction with other cell surface molecules which might be cointernalized or display lateral interaction with CD38 (39). Such putative molecules could be of submembrane localization and harbor the signal and regulation domains that CD38 lacks. Furthermore, di- or oligomerization is a common mechanism preceding internalization. Indeed, it has been reported that 1) CD38 tends to oligomerize under some circumstances (i.e., isolated CD38 undergoes a stable self-aggregation, also induced by treatment with glutathione or mercaptoethanol) (40) and also that 2) ligation with agonistic Abs and cross-linking induces or modulates the CD38-induced activation of protein tyrosine kinases (41, 42). Moreover, these observations were recently corroborated by the crystal structure determination of Aplysia ADP-ribosyl cyclase (43). The enhanced effect observed after CD38 cross-linking with a secondary Ab also suggests that oligomerization or, in general, perturbation of the constitutive conformation of the molecule due to ligand binding markedly increases its endocytosis.
The observation that both agonistic and nonagonistic mAb are able to induce internalization suggests that the CD38 intracellular transport is independent from signal transduction. It is plausible that the mAb-mediated endocytosis simulates what happens in vivo following interaction with the physiologic ligand.
Endocytosis of the radiolabeled mAb was also observed using the F(ab′)2 fragment, indicating that the uptake of the 125I-mAb is specific and independent from Fc or Fcγ receptors.
Lysosomotropic agents (e.g., ammonium chloride and chloroquine, both weak-base amines, and monensin, a carboxylic ionophore) were used to assess the functional involvement of intracytoplasmic compartments. All of these compounds are known to increase pH in acidic intracellular compartments including endosomes, thus perturbing the vesicular traffic and inhibiting both recycling and degradation of internalized molecules (44). Treatment of cells with lysosomotropic agents increased the internalized fraction of the CD38/mAb complexes. Therefore, it is plausible that membrane CD38 on T cells undergoes internalization followed, at least partially, by recycling and re-expression on the cell membrane. To investigate in depth the CD38-specific intracellular pathway, electron microscopy examination was conducted in parallel with light and confocal microscopy. Morphologic studies were consistent with the conclusions drawn from the radiolabeling studies. Endocytosis induced as a result of CD38 ligation is not a random event but presents a very specific pathway consisting of subcellular organelles fundamental in the processing of the complex. Indeed, during the first minutes, CD38/mAb complexes cluster in specialized regions of the cell membrane that are morphologically similar to coated pits and then progressively internalize in the intracellular compartment in coated vesicles. Here, the acidic endosomes are of crucial importance to the intracellular traffic of internalized molecules as they form the intermediate organelle of receptor-mediated endocytosis. After more prolonged incubation, vesicles move inside the cell to the juxtanuclear area where they tend to open, giving origin to multivesicular bodies with CURL shape, containing different amounts of CD38/mAb complexes. At the terminal stage of the internalization, the vesicles appear empty of the CD38/mAb complex, probably due to active mechanism(s) as witnessed by the formation of openings in the membrane. The CURL-like vesicles recirculate and undertake a pathway in the opposite direction, moving to the membrane in areas where surface CD38 is still present at high density. The results of the experiments performed by exposing a molecule acting as a ligand for CD38 instead of an agonistic mAb confirm the validity of the results so far obtained, either in terms of activation (30) or in terms of induction of a variety of biologic effects (35). Further, the induction of vesicle formation without the use of mAb rules out any technical criticism to the experiments performed in the present work. More importantly, the study of the effects implemented after ligand/receptor interaction represents the basis of a model extremely similar to what happens in vivo. Indeed, the molecule is stably maintained inside the vesicles at least up to 90 min, as confirmed by the results reported in Figure 10 c. The meaning of such trafficking is still to be defined, mainly in terms of interaction with cytosolic substrates, second messengers and, in general, with the signaling mechanism(s). These observations are in partial contrast with recent data reported by Zocchi et al. (45). The discrepancy could be attributed to the different models used, or to the use of nicotinamide adenine dinucleotide instead of mAb to engage the CD38 molecule: the main criticism is that nicotinamide adenine dinucleotide is a ligand physiologically present in large amounts inside the cells, while outside the cells it is detectable only in trace amounts.
CD38 has been described to transduce lineage-independent activation signals across the membrane (30, 31, 35, 44) and to be involved in adhesion processes (3). Present data indicate that CD38 down-modulation by internalization might be, along with shedding, another regulatory element in both these functions. Although the enzymatic function and the ability of signal transduction of CD38 have been extensively investigated, the link between these two activities still awaits elucidation. At the moment, it is not know whether the internalized molecules maintain the enzymatic properties, due to technical difficulties in clearly discriminating between the enzymatic activities contributed by internal or membrane compartments. Intriguingly, the enzymatic activities of CD38 are localized within the extracellular domain (4, 46, 47), raising questions concerning the physiologic source of its substrate and whether there is a connection between catalytic activities and immunologic functions attributed to the molecule. The initial hypothesis was that internalization is a crucial step in triggering intracellular signaling and that CD38 molecule can operate inside the cell. However, our data indicate that internalization follows rather than preceding cellular signaling and that it is not directly linked with triggering of the signaling. This can be inferred by the observations that 1) no differences exist between internalization after binding with agonistic or non agonistic mAb and 2) the dynamics of internalization after binding by agonistic mAb is much slower than that of the cellular signaling (48). Indeed, CD38 ligation by agonistic IB4 mAb is followed by intracytoplasmic Ca2+ currents, the peaks of which are recorded 2 to 3 min after the addition of the mAb at 37°C. At this time, the engaged molecules begin to cluster on the cell membrane and have not yet entered the cell. It is more likely that down-modulation of the activated CD38 molecules by internalization serves as a negative feedback control mechanism, which abolishes the signal transduction or cell-cell adhesion mediated by membrane CD38.
This study analyzed CD38 internalization in lymphoid cells. However, internalization and recycling might be influenced by the host environment; moreover, the proportion between membrane and intracellular CD38 might be tissue specific. Beside the trans-membrane form, CD38 has also been detected in the microsomal fraction of COS-7 cells (49). It is likely that in nonlymphoid cells, e.g., in pancreatic β cells, the intracellular fraction of CD38 is involved in cellular metabolism through the production of cyclic ADP ribose (cADPR). However, whether such intracellular pool of CD38 is constituted by internalized molecules or it belongs to molecules synthesized de novo but not moved to the cell membrane is still an open question (49).
The present data demonstrate that surface CD38 expression and its involvement in signaling and adhesion processes can also be controlled by internalization. Further, we have concluded that CD38-mediated signaling is, at least in the initial phase, independent from CD38 internalization and subsequent intracellular synthesis of cADPR.
Thanks are given to Dr. M. Morra for the determination of Ca2+ currents.
This work was supported in part by the Italian Cancer Research Association (Milan, Italy), by Telethon (Rome, Italy), and by the Special Projects Applicazioni Cliniche della Ricerche Oncologica, AIDS and TB (Istituto Superiore di Sanità, Rome, Italy). M.R. is a recipient of an AIDS Research Fellowship from the Istituto Superiore di Sanità, Rome, Italy.
Abbreviations used in this paper: GaMIg, goat anti-mouse Ig; MESF, molecules of equivalent soluble fluorochromes; cADPR, cyclic ADP ribose; CURL, compartment of uncoupling of receptor and ligand; PE, phycoerythrin; CD38L, CD38 ligand.