| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH COMMUNICATION |
a Institute of Biochemistry, University of Genova, 16132, Genova, Italy
b Institute of Cybernetics and Biophysics, National Research Council, 16149, Genova, Italy
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: proteoliposomes • intracellular calcium homeostasis • transmembrane enzymes • catalytic channels
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Few data are available to explain this paradox. On one hand, CD38 was described to undergo a ligand (NAD+, thiol compounds) -mediated internalization in B lymphoid cells, a process whereby cADPR metabolism can be shifted from the cell surface to the cytosol (21). On the other hand, cADPR generated at the outer surface of some cells (17) proved to elicit functional responses in these intact cells (7). Influx of cADPR across the plasma membrane was suggested, but not demonstrated, to account for these cellular responses. Recently, Prasad et al. (22) determined the crystal structure of ADP-ribosyl cyclase from Aplysia californica, a soluble and monofunctional homolog of CD38 lacking the transmembrane and the cytosolic amino-terminal domains and devoid of cADPR-hydrolase activity. This study revealed a homodimeric structure enclosing a cavity tentatively postulated to accommodate the catalytically reactive and covalently bound intermediate, ADPR. Moreover, based on the sequence homology with the Aplysia cyclase, a 3-dimensional model of CD38 was proposed in which the homodimeric structure, with its solvent accessible cavity, is also present in the human protein (22). This model suggests the possibility of a channel being formed by juxtaposition of two CD38 monomers (19).
To investigate whether a transport activity is associated with the catalytic activity of membrane-bound CD38, we followed a biochemical approach based on the in situ study of native CD38 both on resealed right side-out membranes from human erythrocytes (that are constitutively CD38+; see ref 23) and CD38-reconstituted proteoliposomes. In both systems, CD38 behaved as a selective transporter for catalytically generated, but not for exogenously added, cADPR (or for its cADPR-hydrolase resistant analog cGDPR), which was concentrated in the two internal vesicle compartments. This novel cADPR (cGDPR) -transporting activity of CD38 received structural support by the demonstrated occurrence of tetramers and dimers of the glycoprotein in CD38-reconstituted proteoliposomes. Functional evidence for a cADPR-transporting role of membrane CD38 was provided by the significant and transient increases of cytosolic calcium that were elicited by externally added NAD+ in CD38 sense-transfected but not in antisense-transfected or constitutively CD38-, wild-type HeLa cells.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Preparation of unsealed and right side-out resealed hemoglobin-free erythrocyte membranes (ghosts)
Blood samples were obtained from normal volunteers, using heparin as anticoagulant. Leukocytes and platelets were removed with a leukocyte removal filter (Sepacell, Asahi Medical Co. Ltd., Tokyo, Japan). Washed, packed erythrocytes were hemolyzed in 10 vol of ice-cold 10 mM Tris-HCl, pH 8.0, and centrifuged at 10,000 x g for 20 min at 4°C. Erythrocyte ghosts were further washed six times in 10 vol of the same buffer to obtain complete removal of hemoglobin and once in 10 mM Tris-HCl, pH 6.5 (unsealed ghosts).
The white erythrocyte membranes (3 mg/ml, as determined by the method of Bradford, ref 26) were resealed by incubation with 0.1 volumes of 3 M NaCl containing 200,000 cpm/ml of [14C] sucrose for 45 min at 37°C. The resealed ghosts were then washed four times at 10,000 x g for 10 min with 10 volumes of 0.3 M ice-cold NaCl and three additional times in ice-cold phosphate-buffered saline (PBS). Sidedness and percentage of resealing were determined as described (27): 95% right side-out resealed ghosts were routinely obtained. The internal water space of resealed ghosts was calculated from the entrapped radioactivity and expressed as µl/mg protein content.
Assay of CD38 enzyme activities
NADase activity was assayed on NAD+, and ADP-ribosyl cyclase activity was assayed either on NAD+ or NGD+ by incubating 0.5 ml of resealed ghosts (3 mg/ml) in PBS with 0.2 mM NAD+ or NGD+, respectively. At different times, 100 µl aliquots were withdrawn, TCA deproteinized, and analyzed by HPLC (25). NADase activity was estimated to be 11 ± 1 nmol ADPR/(min·mg protein); cyclase activity was calculated to be 1.0 ± 0.1 nmol cGDPR/(min·mg protein) on NGD+ as substrate and 0.11 ± 0.02 nmol cADPR/(min·mg protein) on NAD+.
Transport assay in right side-out resealed ghosts
Resealed right side-out hemoglobin-free ghosts (3 mg/ml) containing radioactive sucrose were incubated in 1.5 ml of isotonic PBS in the presence of either NAD+, or NGD+, cADPR, or cGDPR at concentrations ranging from 20 to 50 µM (cyclic nucleotides) and from 100 µM to 6 mM (NAD+ and NGD+). At various times, 300 µl aliquots were withdrawn and resealed ghosts were separated from extravesicular medium by centrifugation at 10,000 x g for 5 min at 4°C. The supernatants were recovered and immediately extracted with 20% TCA. The pellets containing ghost vesicles were washed once in 40 ml of ice-cold PBS for 5 min at 10,000 x g in order to eliminate residual external nucleotides, resuspended in 100 µl of PBS, and TCA precipitated. Excess TCA was removed by diethylether, and extracts from both supernatants and pellets were analyzed by HPLC as described (25). An aliquot of the extracted pellet was counted for radioactivity in order to determine the intravesicular volume (see above). Nonspecific adsorption of nucleotides was estimated from their incubation with unsealed ghosts, under the same conditions as described above, and found to be almost undetectable (see 
Fig. 1).
Accordingly, the levels of nucleotides estimated by HPLC in the pellets reflect their concentrations in the intravesicular space of the resealed ghosts (see Results) For the washout experiments, resealed ghosts (3 mg/ml) were subjected to three repeated 5 min incubations at 37°C in the presence of 1 mM NGD+ (see above). At the end of each incubation period, ghosts were washed at 4°C in a large volume (40 ml) of ice-cold PBS (10,000xg for 5 min). During each incubation, two aliquots (300 µl) of the mixture were withdrawn, at 30 s and 5 min, centrifuged at 4°C for 5 min at 10,000 x g, and both the supernatant and the pellet were subjected to TCA deproteinization and HPLC analysis.
|
Entrapment of NGD+ was achieved upon resealing the white erythrocyte membranes (3 mg/ml) containing radioactive sucrose in the presence of 2 mM NGD+. NGD+-loaded right side-out resealed ghosts (3 mg/ml) were then incubated in 1 ml of PBS for 40 min at 37°C. At different times (zero, 5, 20, and 40 min), 300 µl aliquots were withdrawn and centrifuged at 10,000 x g for 5 min. The supernatant was collected and extracted with 20% TCA; the pellet was washed once in 40 ml of ice-cold 0.3 M NaCl at 10,000 x g for 5 min at 4°C, resuspended in 50 µl PBS, and TCA extracted. Deproteinized extracts of supernatant and pellet were analyzed by HPLC.
Reconstitution of purified CD38 in liposomes
Native CD38 was purified from white human erythrocyte membranes as described (5), except that ß-octylglucopyranoside was used as detergent throughout the purification instead of Triton X-100. Purified full-length CD38 (in resuspension buffer: 10 mM Tris-HCl, pH 6.5 and 1% ß-octylglucopyranoside) displayed a specific cyclase activity on NGD+ of 2.0 ± 0.15 nmol cGDPR/(min·µg).
To extract total lipids from erythrocyte membranes, 10 ml of hemoglobin-free erythrocyte membranes (3 mg/ml) were mixed with 100 ml of methanol under constant stirring at 25°C. After 30 min, 100 ml of chloroform was added to the mixture and the solution was kept under stirring conditions at 20°C for 30 min. The mixture was centrifuged at 1500 x g for 5 min. The supernatant was then mixed with 100 ml of chloroform and 60 ml of 50 mM KCl, and the mixture was cooled overnight at -20°C. The apolar lipid-containing phase, well separated from the aqueous phase, was recovered and dried in rotovapor. The total lipid content was determined using a modification of the method of Marsh and Weinster (28). Briefly, aliquots of 5 or 10 µl of total lipid samples were lyophilized and resuspended in 0.2 ml of 96% H2SO4. The mixture was heated at 150°C for 30 min, cooled to 25°C for 10 min, and diluted in 1 ml of water. The optical density was measured at 275 nm. A 2 mg/ml solution of cholesterol-egg phosphatidylcholine (1:1) was used as standard solution and aliquots of 2, 5, and 10 µl were used for calibration.
Low- and high-content CD38 proteoliposomes were prepared after the detergent dialysis procedure in order to obtain large unilamellar vesicles (29). The dried lipid film (58 mg total lipids) was resuspended with either 2.4 or 14.5 µg, respectively, of CD38 in 800 µl resuspension buffer containing 0.15 M NaCl with 6000 cpm/ml [32P] inorganic phosphate. The resulting emulsions were vortexed and sonicated in ice for 1 min under N2 (Heat Systems-Ultrasonics, Inc. W-380). The sonicated solutions were extensively dialyzed against 5 l of buffer containing 10 mM Tris-HCl, pH 6.5, 0.15 M NaCl, and 30 x 106 cpm [32P] inorganic phosphate (dialysis buffer) for 36 h at 4°C.
Control liposomes were prepared as described above without the addition of purified CD38. Both proteoliposomes and control liposomes (0.4 ml aliquots) were finally filtered through a 30 x 1 cm Sephadex G-50 column equilibrated in dialysis buffer. Proteoliposomes and liposomes were collected in the turbid fraction.
The recovery of total lipids in the proteoliposome and liposome fractions was determined in 5–10 µl, as described above. CD38 cyclase activity in the proteoliposome suspension was determined by incubation of an aliquot of the vesicular fraction in the presence of 1 mM NGD+. At various times, aliquots of the incubation mixture were withdrawn, centrifuged at 100,000 x g for 15 min at 4°C, and the supernatants were analyzed by HPLC (25).The low- and high-CD38 content liposome preparations typically exhibited a cyclase activity of 0.26 ± 0.03 nmol cGDPR/(min·ml) and 1.35 ± 0.16 nmol cGDPR/(min·ml) and a lipid content of 2.2 mg/ml and 3 mg/ml, respectively. Recoveries of 40% of enzyme activity and 54% approximately of total lipids present in the starting suspension were routinely obtained.
The internal water space of proteoliposomes and liposomes was determined by using a modification of the method of Chen and Wilson (30), which is based on the measurement of [32P] inorganic phosphate entrapped during vesicle formation. After G-50 Sephadex filtration, aliquots of liposomes and proteoliposomes (3 ml) were centrifuged at 100,000 x g for 15 min at 2°C and washed once in ice-cold 0.15 M NaCl and 10 mM Tris-HCl, pH 6.5. The pellets were resuspended in 0.25 ml of water and the entrapped [32P] inorganic phosphate was detected by radioactivity counting. The total lipid concentration was determined as described above. The internal volume of liposomes and proteoliposomes was found to be 10.6 ± 0.06 and 12.14 ± 0.09 µl/mg lipid, respectively, with a very low degree of variability among different preparations.
Entrapment of NAD+ into low-CD38 proteoliposomes and liposomes was obtained as described above in the presence, during the reconstitution step, of 20 mM NAD+ and 1 x 106 cpm of [14C] sucrose instead of [32P] inorganic phosphate. The resulting proteoliposomes and liposomes containing intravesicular NAD+ (approximately 0.2 mM) and the radioactive tracer to evaluate the percentage of lysis were incubated in 0.15 M NaCl and 10 mM Tris-HCl, pH 6.5, at 37°C for 21 h. At various times (1, 3, and 21 h), 1 ml aliquots were withdrawn and centrifuged at 100,000 x g for 15 min at 4°C. The supernatants were counted for [14C] sucrose radioactivity and analyzed by HPLC to evaluate possible nucleotide release. The pellets were resuspended in 0.25 ml of water and extracted with 20% TCA. After microfuge centrifugation and diethylether extraction, the samples were analyzed by HPLC (25). To estimate the external NAD-ase activity, NAD+-loaded proteoliposomes and liposomes were incubated in the presence of external 0.5 mM NAD+ in 0.15 M NaCl and 10 mM Tris-HCl, pH 6.5, for 60 min at 37°C. After incubation, the vesicles were centrifuged at 100,000 x g for 15 min at 4°C and the supernatants were analyzed by HPLC (25) to evaluate extravesicular ADPR production.
Transport assay in CD38-reconstituted liposomes
Low-CD38 proteoliposomes (2.2 mg lipid/ml, cyclase activity 0.26±0.03 nmol cGDPR/min/ml) and control liposomes (18 ml) were recovered in dialysis buffer and incubated at 37°C with NGD+ and cGDPR at different concentrations. At various times, 4 ml aliquots were withdrawn and centrifuged at 100,000 x g for 15 min at 4°C. The nucleotides present in the supernatant were analyzed by HPLC (25). The pellet was washed twice in 3 ml of ice-cold 10 mM Tris-HCl, pH 6.5, containing 0.15 M NaCl and resuspended in 400 microliters of water. A 10 µl aliquot of each suspension was submitted to assay of total lipids, as described above, whereas 90 µl was extracted with 20% TCA and analyzed by HPLC; 300 µl were counted for [32P] radioactivity to estimate the internal water space as described above.
High-CD38 proteoliposomes and control liposomes were incubated as above with 1 mM NAD+. Nonspecific nucleotide absorption was evaluated by directly adding 5 µM cADPR or 140 µM ADPR to the control liposomes incubated in parallel and by measuring their concentrations in the corresponding TCA extracts of the supernatants (extravesicular) and pellets (intravesicular), respectively.
SDS-PAGE and elution of enzyme activity from the gel
CD38-reconstituted proteoliposomes (2 mg lipid/ml, cyclase activity of 0.26 ± 0.03 nmol cGDPR/min/ml) and control liposomes (2 mg lipid/ml) were centrifuged at 100,000 x g for 15 min at 4°C. The pellet was solubilized in a modified Laemmli sample buffer (31) containing 1.5% sodium dodecyl sulfate (SDS) without ß-mercaptoethanol and EDTA. Molecular weight markers were prepared as the sample and run in parallel. The samples were heated at 50°C for 3 min and the gel was kept refrigerated during the run (3 h). Each longitudinal lane was then cut into transversal slices of 1 cm. Each gel slice was put into a separate dialysis tubing containing 0.75 ml of 10 mM Tris-HCl, pH 6.5, and 0.1% Triton X-100. Slices were dialyzed against 4 l of the same buffer for 18 h at 4°C. Thereafter, the content of each tube (around 1 ml) was incubated with 0.04 mM 
-NAD+ at 37°C for 4 h in the same buffer. The -ADPR produced was detected by a LS50B Fluorometer (Perkin Elmer, Norwalk, Conn.) set at 300 nm excitation and 410 nm emission.
Fluorometric detection of [Ca2+]i in intact HeLa cells incubated with external NAD+
CD38 sense- and antisense-transfected as well as constitutively CD38-, wild-type HeLa cells were Fura2 loaded and analyzed for [Ca2+]i, as described (24). Calcium measurements were performed on a field containing from 5 to 20 cell bodies. At the beginning of each experiment, cells were washed in saline zero Ca2+ solution (135 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, and 5 mM Hepes pH 7.4). Intact cells in zero Ca++ solution were exposed to NAD+ at millimolar concentrations (1 to 5) for 5 min at 25°C, and [Ca++]i variations were recorded as described (24).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Fig. 1A, incubation of resealed ghosts with external NAD+ (1 mM) in isotonic medium resulted in the progressive increase of cADPR concentration inside these vesicles, with a tendency to level off after 5 min; under these conditions, NAD+ and ADPR were also detectable inside resealed ghosts ( Fig. 1A, inset). In control experiments in which unsealed ghosts were incubated with NAD+, a minute amount of cADPR was found to be adsorbed to the membrane ( Fig. 1A).
The cADPR-hydrolase activity of CD38 and the resulting significant production of ADPR from NAD+ suggested to use NGD+ as substrate, since cGDPR is not converted by the hydrolase (33). During incubation of resealed ghosts with external NGD+ (1 mM) in isotonic conditions, intravesicular cGDPR concentrations followed the same kinetics as observed for cADPR, but were approximately 10-fold higher ( Fig. 1B). This difference is due to susceptibility of cADPR, but not of cGDPR, to the hydrolase activity of CD38. As observed with NAD+, the resealed erythrocyte ghosts were also permeable to NGD+ (Fig.1B, inset). Control experiments with unsealed ghosts incubated in the presence of NGD+ demonstrated the very limited extent of cGDPR nonspecific binding to the membrane.
Influx of cGDPR across resealed ghosts takes place only after its CD38-catalyzed formation from NGD+. This was clearly demonstrated ( Fig. 2)
by incubating the resealed ghosts in the presence of preformed cGDPR at concentrations similar to those produced by resealed ghosts incubated with NGD+ for the times indicated in Fig. 1. The amount of cGDPR bound to these vesicles was two orders of magnitude lower than the intravesicular cGDPR concentration in resealed ghosts incubated with NGD+. In the latter case, cGDPR was internalized according to a hyperbolic saturation curve ( Fig. 2); under these conditions, the NGD+ concentration also increased progressively inside the vesicles ( Fig. 1B, inset). Moreover, this experiment demonstrated that transport of cGDPR across the ghost membranes occurs against a concentration gradient, as shown by higher concentrations of intravesicular than of external cGDPR. Comparable results were obtained with cADPR, which did not permeate the erythrocyte membrane when directly added to the resealed ghosts (not shown) while being conversely translocated when generated from NAD+ as precursor ( Fig. 1A).
|
The apparent active transport of cGDPR against a concentration gradient was further investigated by means of experiments in which repeated washouts and supplementations of NGD+ were performed (see Materials and Methods). As shown in Table 1,
after the first 5 min incubation with NGD+, the extravesicular and intravesicular concentrations of cGDPR were 8.7 and 85 µM, respectively. After the first washout and reincubation for 30 s with NGD+, external cGDPR dropped to 2 µM whereas internal cGDPR increased to 106 µM. External and internal cGDPR concentrations rose to 9 and 116 µM, respectively, upon prolonging the incubation to 5 min. Finally, after an additional washout and reincubation for 5 min with NGD+, the extravesicular concentration of cGDPR again reached 7.3 µM values, whereas the internal concentration increased further to 147 µM. These data, besides confirming the concentrative nature of the cGDPR transport system, also show that it is unidirectional. This property does not hold for the dinucleotide-transporting system. Entrapment of NGD+ inside the resealed ghosts and their subsequent incubation at 37°C for 40 min consistently resulted in a measurable outflow of the encapsulated NGD+, which decreased from 0.3 to 0.22 mM. The loss of NGD+ recorded in these conditions was not paralleled by any net formation of intravesicular cGDPR, thus confirming that the cyclase activity of CD38 is only at the outer surface of the erythrocyte membranes (23).
|
The active intravesicular concentration of cyclic nucleotide generated from the corresponding dinucleotide was also investigated by experiments designed to minimize variations of the external concentrations of NAD+, cADPR, and ADPR. The experimental setting consisted of resealed erythrocyte ghosts inside a dialysis tube (Sigma), with an Mr cutoff between 12,000 and 14,000, which were exposed to 1 mM NAD+ added both in the dialysis compartment and in the 400-fold larger external volume. In these experiments, the advantage to minimize changes in nucleotide concentrations was not balanced by avoidance of repeated washings of the incubated ghosts, which were still required to eliminate the interference in the HPLC assays of intravesicular vs. extravesicular NAD+. In any case, this different approach confirmed transport of cADPR against a concentration gradient (not shown).
To dissect transport of dinucleotides and of cyclic nucleotides, respectively, across the resealed ghosts, inactivation experiments were designed and performed. Thiol compounds including ß-mercaptoethanol and GSH are known to inactivate both ectoenzyme activities of CD38 either on the native purified glycoprotein (34) or in situ in erythrocyte membranes (32). Thus, we inactivated the cyclase and the cADPR-hydrolase activities (down to 14–18% of the control activities) by incubating the resealed ghosts with 100 mM ß-mercaptoethanol for 30 min at 37°C. The washed ghosts were then incubated for 20 min at 37°C with 1 mM NGD+ and 50 µM cGDPR. The intravesicular concentration of cGDPR in the enzyme-inactivated ghosts was approximately 20% of that measured in the control vesicles, whereas no substantial difference in the extent of internalized NGD+ was recorded between the inactivated and the control vesicles. Therefore, cGDPR transport across the erythrocyte membrane requires enzymatically active CD38, but NGD+ influx does not. These data, besides ruling out any specific role of CD38 in the NGD+ permeation process, also exclude symport mechanisms for cGDPR and NGD+. Attempts to affect the NGD+-derived cGDPR transport by means of an anti-human CD38 monoclonal antibody (IB4; ref 35) were unsuccessful (not shown).
Two transporter proteins potentially relevant to cGDPR (cADPR) influx are present in the erythrocyte membrane, i.e., the band 3 protein (the major anion transporter in red cells; ref 36) and the nucleoside transporter (37). Neither 50 µM DIDS (inhibiting the pyridoxal 5-phosphate-transporting activity of band 3 protein by 96%) nor 50 µM Dipyridamole (inhibiting adenosine influx into resealed ghosts by 93%, as determined by HPLC) was able to affect cGDPR and NGD+ influx across the resealed ghosts.
Transport of cADPR and cGDPR across CD38-reconstituted proteoliposomes
To analyze the role of transmembrane CD38 in the formation of cGDPR (cADPR) and in its subsequent influx, CD38-reconstituted liposomes were prepared and investigated with respect to cGDPR (cADPR) transport. To reconstitute CD38 in a lipid bilayer closely similar to its native environment, CD38 purified from human erythrocytes was embedded into proteoliposomes composed of total lipids extracted from erythrocyte membranes (38).
To detect a possible intravesicular localization of CD38, the unambiguous comparison of total (i.e., after detergent solubilization) vs. external enzymatic activities of CD38 was necessary. However, preliminary findings showed a limited extent of enzymatic activation of CD38 by non-ionic detergents and of inactivation by ionic detergents. Therefore, as an alternative approach, 0.2 mM NAD+ was encapsulated into CD38-reconstituted proteoliposomes and control liposomes, and its stability was investigated for up to 21 h at 37°C in 10 mM Tris-HCl, pH 6.5, containing 0.15 M NaCl ( Table 2).
After 21 h incubation, a 4% lysis was observed. Taking into account the very limited spontaneous conversion of NAD+ to ADPR in control liposomes (0.33±0.06 nmol ADPR·ml-1·h-1), the net production of ADPR in the CD38-reconstituted proteoliposomes was as low as 0.49 ± 0.02 nmol ADPR/(ml/h). This apparent intravesicular NAD-ase activity, which was approximately 0.7% of the external NAD-ase activity measured in the proteoliposomes (67.3±3.2 nmol ADPR/ml/h, Table 2), was close to the rate of spontaneous conversion. The almost complete stability of NAD+ inside the CD38-reconstituted proteoliposomes demonstrated that 1) 99% of CD38 molecules are correctly oriented in the lipid bilayer, with the carboxyl-terminal catalytic region facing the extravesicular space, 2) these proteoliposomes are unilamellar vesicles because of lower than 1% intravesicular enzyme activity, 3) occurrence of intravesicular cADPR upon incubation of the proteoliposomes with NAD+ cannot be due to internal formation from NAD+ itself, and 4) CD38 per se is unable to cause any outward directed transport of NAD+ because the concentration of NAD+ inside the proteoliposomes remains unchanged during 21 h incubation.
|
Since comparable results were obtained with NGD+-loaded CD38 proteoliposomes, these artificial membranes behave differently from resealed erythrocyte ghosts where efflux of NGD+ was observed (see above). Accordingly, a transporter for dinucleotides, different from CD38, is present in erythrocyte membranes.
As shown in Fig. 3A,
incubation of proteoliposomes with external 1 mM NGD+ resulted in the almost immediate appearance of cGDPR in the TCA extracts from vesicle pellets. Vesicle-bound NGD+ was detectable both in proteoliposomes and liposomes, and in both cases was calculated to be 0.02% (calculated as pmol/mg lipid) of external NGD+ present in the incubation mixtures. This result is not due to CD38-mediated transport, but to a nonspecific adsorption of NGD+ to the lipid bilayer. The intravesicular concentration of the cyclic nucleotide reached 40 pmol cGDPR/mg lipid in 5 min and was found to level off during incubation, showing kinetics comparable to those observed in the experiments with resealed ghosts ( Fig. 1B). Incubation of proteoliposomes and liposomes in the presence of 10 µM cGDPR resulted in an equally low amount of vesicle-bound cGDPR (0.04% internal/external concentration calculated as pmol/mg lipid), which can be explained as a nonspecific adsorption of the cyclic nucleotide to the artificial membranes ( Fig. 3A). During incubation of proteoliposomes in the presence of 1 mM NGD+, the intravesicular cGDPR concentrations were assayed at different times and plotted against the corresponding external concentrations of the cyclic nucleotide. Figure 3B shows that cGDPR transport across vesicles occurred against a concentration gradient, at least at the earliest times of incubation of proteoliposomes with NGD+. This result is similar to that obtained with resealed erythrocyte ghosts ( Fig. 2).
|
The dependence of cGDPR influx into CD38 proteoliposomes upon increasing NGD+ concentrations is shown in Fig. 4.
As in the resealed erythrocyte membranes (see inset), cGDPR transport occurred according to hyperbolic saturation kinetics, although Vmax levels were comparatively different because of a much lower content of CD38 in the proteoliposomes and half-saturation was consistently observed at 0.6–0.7 mM NGD+. These values are higher than the reported Km for cyclase activity of CD38 toward NGD+ (2 µM) (33). Accordingly, even at the lowest NGD+ concentrations used to measure cGDPR influx, the cyclase activity of the CD38-reconstituted proteoliposomes was maximal, as assayed from the extravesicular concentrations of cGDPR being formed.
|
To better characterize the cyclic nucleotide-transporting properties of membrane-reconstituted CD38, the high CD38-reconstituted liposomes (see Materials and Methods) were incubated with the physiological substrate NAD+ (1 mM) in isotonic conditions, and the specific reaction products cADPR and ADPR were assayed separately in the intravesicular and extravesicular compartments ( Table 3).
An immediate burst of cADPR (1.1–0.1 µM) was found to be selectively associated with the vesicle fraction of CD38-reconstituted proteoliposomes. At both incubation times analyzed, the external concentration of cADPR was below 1% of that of ADPR, as previously observed with purified native CD38 (5). Conversely, in the intravesicular compartment, cADPR reached concentrations around 50% of those of net ADPR calculated by subtracting the amount of ADPR adsorbed to the control liposome vesicles ( Table 3). The intravesicular levels of ADPR in proteoliposomes were close to the amount of nonspecifically adsorbed nucleotide as assayed in the liposome fraction ( Table 3), whereas its concentration in the supernatant reached 173 µM after 10 min incubation. Thus, the apparent NAD-ase activity (i.e., generation of ADPR from NAD+) was very high in the supernatant, as observed with both detergent-solubilized and recombinant CD38 (33). On the contrary, when CD38 was embedded in a bilayer forming a sealed vesicle, the quantitatively minor product cADPR proved to be actively and selectively concentrated in the internal space. These results demonstrate a remarkable difference between transmembrane and soluble CD38: transmembrane CD38 features both synthesis and transport of cADPR, whereas its soluble versions catalyze formation of cADPR and of ADPR only.
|
In situ occurrence of oligomeric forms of CD38 in proteoliposomes
Native CD38 has never been reported to show structures other than a 46 kDa monomeric form, except for a high molecular mass species of CD38 (Mr 190 kDa) due to transglutaminase-catalyzed. posttranslational cross-linking identified in retinoic acid-induced HL60 cells (39). One reason for these negative results is probably the known trend of CD38 to precipitate, thereby hampering conventional procedures for exploring oligomeric proteins, including density gradient centrifugation and gel permeation experiments (21). To investigate possible oligomeric forms of CD38 in CD38-reconstituted proteoliposomes, mild denaturing conditions (1.5% SDS and heating the samples at 50°C) were used for SDS-polyacrylamide gel electrophoresis (SDS-PAGE) separation of high and low molecular mass species of CD38. Assays of the NAD-ase activity performed by measuring conversion of -NAD+ to the fluorescent product -ADPR allowed us to identify three different molecular forms of catalytically active CD38: a major species, corresponding to 46 kDa, identifiable with the monomers; and two forms, the first between 86 and 112 kDa (13% of total activity) and the second around 197 kDa (11% of total activity), accounting for dimers and tetramers, respectively ( Fig. 5).
The presence of monomers might be artefactual because of the use of SDS and of brief heat treatment (see Materials and Methods) to achieve disruption of CD38 proteoliposomes. Accordingly, the percentages of dimers and tetramers may be underestimated. However, these data demonstrate that catalytically active CD38 exists in membranes as oligomers (homodimers and homotetramers).
|
Transient increases of cytosolic calcium elicited by extracellular NAD+ in CD38+ cells
In an attempt to demonstrate a physiological role of CD38 as active transporter of its enzymatic product cADPR, CD38-transfected HeLa cells were loaded with Fura2 and incubated with 1 to 5 mM NAD+ in zero Ca2+ solutions. The cytoplasmic free calcium concentrations ([Ca2+]i) were recorded as described (24). As shown in Fig. 6A, B,
addition of 2 mM NAD+ resulted in an immediate increase of [Ca2+]i up to 50% above the basal [Ca2+]i, i.e., 40 nM. Two types of [Ca2+]i movement profiles were observed in 10 different experiments: either few [Ca2+]i oscillations followed with the same amplitude ( Fig. 6A), or [Ca2+]i decreased in a couple of minutes to the basal value with no further effect ( Fig. 6B). The responses were similar when cells were exposed to NAD+ concentrations ranging from 1 to 5 mM. Addition of 50 µM digitonin, as membrane permeabilizer after 20 min incubation with NAD+, elicited a calcium release of up to 300% of the basal value (not shown). That the digitonin effect was due to accumulation of extracellularly produced cADPR was demonstrated by the failure to observe any calcium release in transfected HeLa cells incubated without NAD+. Moreover, since these experiments were performed in measured zero Ca2+ solutions (see Materials and Methods), the digitonin-induced Ca2+ signal was not due to influx of contaminating Ca2+ from the external medium. No effect on [Ca2+]i was detectable when CD38 antisense-transfected or wild-type HeLa cells, with a basal [Ca2+]i of 26 nM (24), were incubated in the presence of 1 to 5 mM NAD+ ( Fig. 6C).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The goal of this study was to focus on the catalytic properties of CD38 and on the still unknown mechanism whereby its enzymatic product cADPR regulates the intracellular calcium homeostasis. In fact, the ectocellular localization of CD38 enzymatic activities, coupled with the known role of cADPR as an intracellular calcium mobilizer, raises questions concerning 1) the membrane-bound nature of ADP-ribosyl cyclase whose homolog in invertebrates is soluble (22), and 2) how ectocellularly generated cADPR can reach its receptor-operated target stores, thereby displaying its calcium-releasing activity (18–20).
Previous experiments were consistent with accessibility of NAD+ to the cyclase activity of transmembrane CD38 during its export from Golgi vesicles to the plasma membrane (24). Indeed, in CD38- cells transduced with full-length CD38 cDNA, a causal relationship was demonstrated between decrease of NAD+/NADH, accumulation of cADPR, and increase of cytosolic calcium (24). Confirmatory evidence for the apparent traffic of NAD+ and cADPR across intracellular membranes was provided by experiments on the ligand-induced internalization of CD38 in lymphoid cells (21), showing a shift of cADPR metabolism to an intracellular compartment. Together, these results suggest the occurrence of an inward transport of NAD+ and an efflux of cADPR from CD38-containing intracellular vesicles, both exocytotic and endocytotic ( Fig. 7).
|
A convenient way to study the transport of solutes across membranes is to produce artificial bilayers containing the putative transporter. We used proteoliposomes reconstituted with CD38 purified from two different sources (i.e., erythrocyte ghosts and membranes from CD38- transfected HeLa cells; not shown). A more physiological approach exploits resealed erythrocyte ghosts whenever the transporter protein is expressed on the erythrocyte membrane. Thus, the glucose (43), anion (44), and GSSG (45) transporters have been studied extensively on these resealed model membranes. Expression of CD38 on erythrocytes (23) enabled us to take advantage of this unique and widely used model system.
The data obtained in the present study provide additional information on the mechanisms underlying the topological paradox of the CD38/cADPR system (19, 20). Specifically, they demonstrate the dual activity of transmembrane CD38 as a generating system of cADPR (cGDPR) and a vectorial transporter of cADPR (cGDPR) itself. In both membrane systems investigated (resealed human erythrocyte membranes and CD38-reconstituted proteoliposomes), no detectable influx of directly added cADPR or cGDPR was observed. Conversely, both cyclic nucleotides accumulated inside either vesicle after their CD38-catalyzed formation upon incubation with the corresponding dinucleotide (NAD+ or NGD+) under isotonic conditions. CD38-reconstituted proteoliposomes allowed us to establish specificity, unidirectionality, and the intrinsic role of CD38 as a unique cADPR (cGDPR) transporter. This is substantiated by the closely comparable behavior of proteoliposomes prepared with CD38 purified from either erythrocyte ghosts or from CD38-transfected HeLa cells. Moreover, CD38 proteoliposomes showed that the cADPR channeling activity of transmembrane CD38 obscures hydrolysis of the cyclic nucleotide ( Table 3), which, conversely, is a major feature of detergent-solubilized CD38. That no other permeation systems for enzymatically produced cADPR (cGDPR) exist in the human erythrocyte membranes is strongly suggested by the kinetic similarity of cGDPR influx across CD38-reconstituted proteoliposomes and resealed membranes ( Fig. 4). With both membrane vesicles, the Km; of saturable cGDPR influx, measured as a two-step process of 1) NGD+ to cGDPR conversion and 2) permeation of cGDPR across either membrane, was found to be identical at 0.6 ± 0.08 mM NGD+. Moreover, the failure of externally added cGDPR (cADPR) to cross both natural and artificial membranes restricts the topology of transport to the catalytic site of CD38 where conversion of either dinucleotide to cGDPR or to cADPR takes place.
The present results clearly indicate that both cGDPR and cADPR are actively concentrated inside resealed erythrocyte membranes and CD38-reconstituted liposomes, being particularly evident at the lowest external levels of either cyclic nucleotide. No external source of energy for transport of cGDPR and cADPR is required. Therefore, the driving force for this active transport across CD38 molecules must derive from the covalently GDP-ribosylated (ADP-ribosylated) CD38 intermediate formed during catalysis and from its subsequent dissociation. The N-glycosidic linkage between nicotinamide and ribose in the NAD+ molecule is a high-energy bond (-8.2 kcal/mol at pH 7 and 25°C; see ref 46). Its cleavage apparently releases an amount of free energy sufficient to elicit both the cyclization step resulting in cADPR formation and a putative conformational change affecting the solvent-filled cavity created by juxtaposition of two or four interacting monomers. The model of CD38 structure based on homology with the ADP-ribosyl cyclase of A. californica involves the presence of a disulfide bridge unique to CD38 and localized in a hinge region connecting the two major domains of each monomer (22). The spatial localization of this disulfide, which is important for catalytic activity (47), has been postulated to play a role in related conformational changes in CD38 (22). Therefore, hinge motion during activity can be reasonably expected to provide the dynamic basis for delivery of in situ generated cADPR (cGDPR) to reach the intracellular side of the membrane, as actually observed in both experimental membrane systems.
The transmembrane occurrence of oligomeric forms of catalytically active CD38 in proteoliposomes is fully consistent with the functional and kinetic data obtained in this study. The structural organization of native CD38 when embedded in membranes is in close agreement with the reported homodimeric structure of the Aplysia ADP-ribosyl cyclase and seems to account for the hitherto unrecognized functional property of CD38 as a channel-forming glycoprotein, in addition to its known cyclase and cADPR-hydrolase activities and its reported base exchange activity (18, 48, 49). Whether such channel activity is preferentially related to homodimers or to homotetramers has not been clarified, nor is any information so far available on interconversion of different CD38 species. However, monomeric forms seem to be competent for catalytic function only, whereas oligomeric forms might be responsible for the two-step process (i.e., catalysis-transport) leading to selective influx of cGDPR or cADPR.
Since active oligomeric forms of CD38 have been identified in CD38-transfected HeLa cells (unpublished results), this cell type has been used as a model to investigate the effect of external NAD+ on [Ca2+]i variations in living cells. An immediate increase of [Ca2+]i was observed after addition of NAD+ to the external medium in the absence of external calcium. This result suggests a release of calcium from internal stores induced by transport of a significant amount of cADPR into intact CD38+ HeLa cells. The same kind of experiment performed in CD38 antisense-transfected or wild-type CD38- HeLa cells did not show any variation in [Ca2+]i, therefore proving that CD38 is responsible for both enzymatic and transport activities across the plasma membrane. Maximal [Ca2+]i response observed in CD38+ intact cells occurred within 30 s of NAD+ supplementation, in agreement with the kinetics of cADPR/cGDPR transport observed in both membrane systems investigated in this study ( Figs. 1 and 3). Moreover, addition of digitonin after 20 min incubation with external NAD+ elicited a very high calcium release in the CD38-transfected, but not in the wild-type, HeLa cells, showing that cADPR was linearly produced and accumulated in the external medium during incubation, as also observed in resealed ghost membranes and CD38-reconstituted proteoliposomes ( Figs. 1 and 3and Table 3). The different patterns of [Ca2+]i observed after NAD+ supplementation may be due to the number of cells present in the photomultiplier field of view. Groups of few cells probably respond at the same time to the external stimulus, whereas large groups of cells might be reached by NAD+ at different times ( Fig. 6).
The CD38-reconstituted proteoliposomes behave as selective catalysts responsible for cADPR (cGDPR) generation and subsequent influx while being completely impermeant to NAD+ (NGD+) and ADPR. Previously, calf spleen NAD+ glycohydrolase had been incorporated into phosphatidylcholine liposomes and shown to be asymmetrically oriented, as in the present study with detergent-solubilized CD38 (50). These NAD+ glycohydrolase-reconstituted unilamellar proteoliposomes were investigated with respect to nicotinamide-dependent transport of NAD+, but proved not to catalyze any vectorial transfer of NAD+ by transglycosidation with nicotinamide (50). Our results exclude any transport of NAD+ across CD38-reconstituted proteoliposomes. On the contrary, the resealed erythrocyte membranes exhibited a remarkable NAD+ (NGD+) -transporting activity, and ADPR crossed these membranes ( Fig. 1A, inset), in agreement with a previous report (51). Therefore, two unrelated transport systems seem to coexist in erythrocyte membranes, the first (CD38) highly specific for cADPR or cGDPR influx and the second responsible for transport of NAD+ or NGD+, whose properties are undefined. Whether the second transporter also translocates ADPR or whether two distinct transporters are involved is not yet known. Elucidation of these properties is now in progress in our laboratory. Taking into account the reversed polarity of exocytotic and endocytotic vesicles compared to the plasma membrane, the observed reversibility of NGD+ transport, as documented by efflux of the dinucleotide from NGD-loaded erythrocyte ghosts, is consistent with the scheme shown in Fig. 7; this scheme accounts for accessibility of NAD+ to the apparently secluded active site of CD38 in the intravesicular space and also for release into the cytosol of catalytically generated cADPR (21, 24). Occurrence of this unprecedented traffic of metabolites across membranes and its role in the regulation of intracellular calcium homeostasis are currently under investigation. These results seem to explain why vertebrate cells can modulate their cytosolic calcium although lacking any soluble ADP-ribosyl cyclase, whose topologically unrestricted activity might be detrimental because of extensive and fast exhaustion of intracellular NAD+/NADH (19).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Abbreviations: cADPR, cyclic adenosine diphosphate ribose; ADPR, adenosine diphosphate ribose; NAD+, nicotinamide adenine dinucleotide; NGD+, nicotinamide guanine dinucleotide; DME, Dulbecco's modified Eagle's medium; GDPR, GDP-ribose; HPLC, high-pressure liquid chromatography; PBS, phosphate-buffered saline; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TCA, trichloroacetic acid; DIDS, 4–4-di-isothiocyano-2,2,-stilbene disulfonic acid.
Received for publication March 23, 1998.
Revision received June 10, 1998.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
); cADPR bound to unsealed ghosts (
); inset: NAD+ (
) and ADPR in resealed ghosts (
). B): cGDPR in resealed ghosts (
