| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Mount Sinai Bone Program, Departments of Medicine and Geriatrics, and Bronx Veterans Affairs Geriatrics Research Education and Clinical Center (GRECC), New York, New York 10029, USA;
* School of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA;
Physiological Laboratory, Cambridge, UK; and
University of Meikai, Saitama, Japan
2Correspondence: Division of Endocrinology, PO 1055, Annenberg 5, Mount Sinai School of Medicine, One Gustave Levy Place, New York NY 10029, USA. E-mail: mone.zaidi{at}mssm.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
cADPr conversion to cause cytosolic Ca2+ release. High levels of CD38 were found in the plasma membranes, endoplasmic reticulum, and nuclear membranes of osteoblastic MC3T3-E1 cells. More important, intracellular CD38 was colocalized with target ryanodine receptors. The cyclase also converted a NAD+ surrogate, NGD+, to its fluorescent product, cGDPr (Km 
5.13 µM). NAD+ also triggered a cytosolic Ca2+ signal. Similar results were obtained with NIH3T3 cells, which overexpressed a CD38-EGFP fusion protein. The 
-49-CD38-EGFP mutant with a deleted amino-terminal tail and transmembrane domain appeared mainly in the mitochondria with an expected loss of its membrane localization, but the NAD+-induced cytosolic Ca2+ signal was preserved. Likewise, Ca2+ release persisted in cells transfected with the Myr--49-CD38-EGFP or -49-CD38-EGFP-Fan mutants, both directed to the plasma membrane but in an opposite topology to the full-length CD38-EGFP. Finally, ryanodine inhibited Ca2+ signaling, indicating the downstream activation of ryanodine receptors by cADPr. We conclude that intracellularly expressed CD38 might link cellular NAD+ production to cytosolic Ca2+ signaling.—Sun, L., Adebanjo, O. A., Koval, A., Anandatheerthavarada, H. K., Iqbal, J., Wu, X. Y., Moonga, B. S., Wu, X. B., Biswas, G., Bevis, P. J. R., Kumegawa, M., Epstein, S., Huang, C. L.-H., Avadhani, N. G., Abe, E., Zaidi, M. A novel mechanism for coupling cellular intermediary metabolism to cytosolic Ca2+ signaling via CD38/ADP-ribosyl cyclase, a putative intracellular NAD+ sensor.
Key Words: endoplasmic reticulum • ryanodine receptor • myristoylation • osteoblasts
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. One proposed function is that of an enzyme: a cyclase that catalyzes the conversion of the cellular intermediary metabolite nicotinamide adenine dinucleotide (NAD+) to the Ca2+-releasing second messenger cyclic adenosine diphosphate ribose (cADPr) (3
4
5
6
7)
. cADPr activates a family of Ca2+ release channels known as ryanodine receptors that are expressed ubiquitously in eukaryotic cell microsomal membranes (8)
. It is thought that cADPr generated from CD38 catalysis plays a critical role in Ca2+-induced insulin release in pancreatic ß cells (9
, 10)
. Mice deficient in CD38 thus display impaired glucose tolerance (11)
. It has recently been shown that CD38 gates nucleoplasmic Ca2+ influx on the inner nuclear membrane through the activation of inner membrane ryanodine receptors (12)
.
Despite a compelling role for CD38 in Ca2+ homeostasis, there remains a crucial mechanistic paradox (13)
. That surface CD38 has its catalytic site facing outward has rendered it difficult to envisage how extracellular NAD+cADPr catalysis could be translated into a rapidly developing intracellular Ca2+ signal (14)
. A variety of hypotheses have been proposed. One is that NAD+ binding to CD38 causes its internalization and that cADPr is then generated cytosolically (15
, 16)
. Another hypothesis is that cADPr is generated extracellularly and then translocated through the enzyme itself (17)
. Despite this evolving data, several critical questions relating to the role of CD38 in cellular Ca2+ signaling continue to arise. A key question is whether CD38 can catalyze cADPr formation at other more favorable locations, such as the endoplasmic reticulum, where cADPr would be in spatial proximity to target ryanodine receptors.
We show that in addition to being located on the plasma and nuclear membranes, functional CD38 cyclase is expressed on the endoplasmic reticulum and that its distribution matches that of ryanodine receptors. Second, we demonstrate that NAD+-induced Ca2+ release requires CD38 and that it occurs through the activation of ryanodine-sensitive Ca2+ release channels. Third, we provide evidence through the expression of several mutated CD38 constructs that surface localization of the cyclase is not required for NAD+-induced Ca2+ release. We are not, however, implying that plasma membrane CD38 is mechanistically redundant. We demonstrate that a full cytosolic Ca2+ response to NAD+ can be triggered after solely intracellular CD38 expression. We thus speculate that endoplasmic reticular (ER) CD38 senses cytosolic NAD+ concentration changes and cyclizes its substrate to cADPr, which then activates the juxtaposed ryanodine receptors.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
43 kDa CD38 protein, but recognized with equal affinity the CD38-EGFP complex (O. A. Adebanjo et al., unpublished data). C1586 also inhibited the cyclase activity of CD38 (12)
. The polyclonal anti-ryanodine receptor antiserum Ab34 was raised in the rabbit against a short peptide sequence corresponding to the consensus calmodulin binding domain of the ryanodine receptor protein family (kindly provided by F. Anthony Lai, University of Cardiff, Wales) (18)
. It therefore does not differentiate between the three known isoforms—isoforms I, II, and III (18)
. We have also used two dyes, Dil and boron dipyrromethene difluoride, for delineating the plasma membrane and Golgi, respectively (Molecular Probes, San Diego, CA). Dil is a long-chain dialkycarbocyanine that has been used traditionally as a neuronal tracer and that does not alter cell physiology. Boron dipyrromethene difluoride is a ceramide analog shown to produce selective staining of the Golgi (43)
.
Expression of CD38-EGFP fusion proteins
We have previously cloned and sequenced a novel member of the CD38 family from a rabbit osteoclast library (19
, 20)
. A 298-bp cDNA coding region in the cloned insert specified a new CD38 homologue. Its predicted amino acid sequence was 60% similar to mouse and rat CD38 and 50% homologous to the human sequence. Plasmid encoding rabbit CD38 cDNA was used as a template to amplify full-length CD38 and its amino-terminal truncated variants. Upstream primers 5'-CGCAAGCTTGGCCCCGCAGAAC CCGCA-3' (5' CD38), 5'-GGGAAGCTTCACGGCGCGCTATGTCTAAGCAAGTCCTGATCGTC-3' (5' -17-CD38), or 5'-GGGAAGCTTCACGGCGCGCTATGGGCGCCACCGACCACGTCTCT-3' (5' -49-CD38) and downstream primer, 5'-AGTGGGCCCACATCAGGACGCTGCAAG-3' (3' CD38) were used to amplify sequences encoding CD38, -17-CD38-EGFP, or -49-CD38-EGFP; the latter variants lacked 17 or 49 amino acids at the amino terminus, respectively. The PCR products were digested with HindIII and ApaI at the appropriate restriction sites, purified, and ligated into the mammalian expression vector pEGFP-N1 (Clontech, Palo Alto, CA), which contains EGFP downstream of its multiple cloning site. To express the myristoylated form of CD38-EGFP and -49-CD38-EGFP fusion proteins, we used upstream primers 5'-GGGAAGCTTCACGGCGCGCTATGGGATCATCAAAATCAAAGCCAAAGGACCCATCACAACGAGGAATTCCCGACTACGAGTTCAGCCCC-3' and 5'-GGGAAGCT TCACGGCGCGCTATGGGATCATCAAAATCAAAGCCAAAGGACCCATCACAACGAGGATTCCCGGCGCCACCGACCACGTCTCT-3', respectively. The resulting PCR products were cloned in the same vector as described above and sequenced to ensure in-frame ligation. Mouse fibroblastic NIH3T3 cells were transiently transfected with the respective plasmids or empty vector using Superfect Transfection Reagent (QIAGEN, Valencia, CA) per manufacturer’s protocol. After 24 h, we looked for EGFP fluorescence and CD38 immunoreactivity simultaneously by confocal microscopy, as indicated below, to study the expression of the CD38-EGFP constructs. We optimized the transfection protocol to achieve a 20–30% efficiency.
Fluorescence localization studies using confocal microscopy
We first investigated the expression of CD38 and ryanodine receptors in CD38+ MC3T3-E1 cells. Cells were gently permeabilized by incubation for 60 s with Triton X-100 (0.001%, v/v), then incubated with normal goat serum (in 10 mM PBS, 1:10, pH 7.4, for 15 min) in multiwell dishes and washed with HBSS (Gibco-BRL). The cells were subsequently incubated either without antibody, with nonimmune mouse (or rabbit) IgG, or with anti-CD38 antibody (Sigma; C1586, raised in the mouse), with the anti-ryanodine receptor antibody (Ab34, raised in the rabbit), or with C1586 + Ab34 for colocalization studies (all in DMEM, 1:100). For a negative cell control, CD38- NIH3T3 fibroblasts were incubated with the same antibodies. After 6 h incubation, the coverslips were rinsed gently with HBSS, drained, and reincubated with goat FITC-conjugated anti-rabbit IgG or rhodamine-conjugated anti-mouse IgG or with both (for colocalization experiments) (in HBSS for 60 min). These coverslips were then washed gently and drained. A laser confocal Leica microscope was used to visualize the intracellular staining using FITC and rhodamine filters, as appropriate. The green and red digital images, respectively, were merged to detect any overlapping distribution of the two proteins (yellow).
We then examined CD38- NIH3T3 cells transfected with EGFP alone or a variety of CD38-EGFP fusion constructs (see foregoing text). Green fluorescence from EGFP was monitored by confocal microscopy using an FITC filter. To confirm localization of the CD38 by immunostaining, we permeabilized the cells by a 60 s incubation with Triton X-100 (0.001%, v/v, 60 s) and stained the same cells with the anti-CD38 antibody C1586. The cells were then counterstained with rhodamine-conjugated anti-mouse IgG. To obtain confirmatory evidence for plasma membrane (surface) expression, in a separate experimental set, we stained nonpermeabilized NIH3T3 cells transfected with full-length CD38-EFGP, -49-CD38-EGFP, or Myr--49-CD38-EGFP constructs. Intact live cells were thus incubated with C1586, followed by a rhodamine-conjugated anti-mouse IgG. The green and red images obtained were merged, as above. Live NIH3T3 cells did not take up trypan blue, suggesting that antibody access into the cytosol was unlikely. CD38 expression was confirmed in cells transfected with full-length CD38 but without EGFP by immunostaining as described for MC3T3-E1 cells.
To further confirm plasma membrane localization, we stained the cells with the plasma membrane-specific dye Dil by incubating the cells per manufacturer’s protocol. The two confocal micrographic images, namely, red (Dil) and green (EGFP), were then merged. Finally, similar experiments were performed with the Golgi-specific dye boron dipyrromethene difluoride to obtain spatial information on localization of the respective constructs to the Golgi.
Membrane isolation and Western blot analysis
Membranes were prepared from cultured cells as described (12
, 19)
. Mitochondrial fractions were prepared as described earlier (21)
. SDS-PAGE was performed using 12% separating and 4% stacking polyacrylamide gels with a minigel system (Bio-Rad Laboratories, Hercules CA). ER membrane and cytosolic fractions (as well as the nuclear membrane and mitochondrial fractions for some experiments) (30 to 50 µg protein) were heated for 5 min at 95°C in Laemmli’s sample buffer (2% SDS, 2% v/v ß-mercaptoethanol, 10% v/v glycerol, and 50 mg/l bromphenol blue in 0.1 M Tris-HCl buffer, pH 6.8). Electrophoresis was performed at 20 mAmps per gel. The resolved proteins were stained with Coomassie brilliant blue (Sigma) or transferred electrophoretically onto OPTITRAN-supported nitrocellulose membrane at 15°C for 1 h at 100 mV. The membranes were blocked with Tween 20 (0.3%, v/v) in PBS at 20°C and incubated with anti-EGFP antibody (1:3000, Sigma). After rinsing, the blot was incubated for 1 h with HRP-conjugated anti-mouse antibody and the blot was developed using Pierce SuperSignal Ultra Chemiluminescence Kit, per manufacturer’s instruction.
ADP-ribosyl cyclase (NGDcGDPr) assay
ADP-ribosyl cyclase activity was measured in cell fractions prepared from MC3T3-E1 cells as well as NIH3T3 cells transfected with various EGFP-CD38 constructs using a described method (19
, 37)
. We measured the cyclization of the NAD+ surrogate NGD+ to its fluorescent derivative cGDPr. Membrane or cytosolic fractions were incubated for 20 min at 37°C in 20 mM Tris-HCl (pH 7.4) with 100 µM NGD. The reaction was stopped with 5 µl of 100% trichloroacetic acid. Fluorescence in the supernatant was measured using a high-sensitivity spectrofluorometer (
ex=300 nm; em=410 nm). The amount of cGDPr formed was plotted as mean ± SD in nmol-1·ml-1·mg protein-1. To establish specificity, we preincubated membranes for 30 min with the anti-CD38 antibody C1586 (1:1000, Sigma, 37°C) or NAD+ (400 µM, 25°C) before NGD+ application. Mouse IgG2 was used as a control for the antibody experiments. Finally, Eddie-Hofstee plots were constructed and estimates of Vmax and Km obtained.
Microspectrofluorimetric measurements of cytosolic Ca2+
To measure cytosolic Ca2+, MC3T3-E1 or NIH3T3 cells were dispersed onto 22 mm, 0 grade glass coverslips (Libro/ICN, Aurora, OH). The coverslips were incubated for 30 min at 37°C with 10 µM fura 2/AM (Molecular Probes) in serum-free M199-H, then washed in M199-H and transferred to a Perspex bath positioned on the stage of an Ion Optix microspectrofluorimeter. The latter was constructed from an inverted microscope (Diaphot, Nikon, Telford, UK) that uses a high-resolution, low light-intensified CCD camera (SONY, Tokyo, Japan). Only EGFP-positive cells within a given microscope field (assessed using a FITC filter) were selected along with a ‘cell-less’ field for background counts. Transfection efficiencies were 25%. Cells were then exposed alternatively to excitation wavelengths of 340 or 380 nm approximately every second and the emission was sampled at 510 nm. The effect of the test substance (1 mM NAD+ with or without 5 µM ryanodine) was assessed by exposing cells to prewarmed solutions (extracellular [Ca2+]=1.25 mM). Photon counts were recorded every second on a computer to give the ratio of emitted intensities due to excitation at 340 and 380 nm, F340/F380.
The cytosolic Ca2+ measuring system was calibrated using an established protocol for intracellular calibration. Fura-2-loaded cells were bathed in Ca2+-free, EGTA-containing solution containing 130 mM NaCl, 5 mM KCl, 5 mM glucose, 0.8 mM MgCl2, 10 mM HEPES, and 0.1 mM EGTA. Ionomycin (5 µM) was first applied to obtain the minimum ratio due to lowest cytosolic Ca2+ (Rmin) and the maximum fluorescence intensity at 380 nm (Fmax); 1 mM CaCl2 was then applied with 5 µM ionomycin to obtain values of the maximum ratio due to an elevated cytosolic Ca2+ (Rmax) and minimum fluorescence intensity at 380 nm (Fmin). The dissociation constant Kd for Ca2+ and fura 2 at 20°C, an ionic strength of 0.1 M and a pH of 6.85, is 224 nM. The values were substituted into the equation: [Ca2+] = Kd x [(R-Rmin)/(Rmax-R)] x [(Fmax/Fmin)].
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). In contrast, ryanodine receptor localization was restricted to the ER and nuclear membranes. Merging the two digital images revealed a near-complete overlap in the distribution of CD38 and ryanodine receptors (yellow). NIH3T3 cells showed no evidence of CD38 expression by either immunocytochemistry or Western blotting, confirming previous results (not shown; ref 17
, 36
).
|
We next established the functionality of the expressed CD38 by measuring the conversion of the substrate NGD+, a surrogate for NAD+. The assay measures the cyclization of NGD+ to the nonhydrolyzable fluorescent derivative cGDPr (19
, 37)
. CD38 is a bifunctional enzyme: it not only catalyzes NAD+cADPr cyclization, but also hydrolyzes cADPr to ADPr (4)
. In contrast, cGDPr is nonhydrolyzable, so its measurement as a single reaction product represents a distinct advantage over measuring cADPr (38)
. Figure 2
A, B shows that cGDPr formation was dependent on the time of incubation and on added membrane protein concentration. Figure 2C
shows that the anti-CD38 antibody inhibited cGDPr formation, indicating that the detected cyclase was CD38. Preimmune mouse IgG2a failed to attenuate cGDPr formation, as expected. Finally, to confirm that NGD+ and NAD+ bound to the same catalytic site on the CD38 molecule, we carried out competition assays. Figure 2D
shows that formation of cGDPr was inhibited competitively by NAD+. Thus, in the presence of NAD+, the Km increased (as expected) from 5.1 to 19.4 µM whereas Vmax remained unchanged (Fig. 2E
). Note that ER membranes prepared from CD38- NIH3T3 cells did not catalyze NGD+cGDPr cyclization.
|
Localization of the CD38-EGFP fusion protein in NIH3T3 cells
CD38- NIH3T3 cells were transfected with a CD38-EGFP fusion construct. Green fluorescence from EGFP was monitored by confocal microscopy using an FITC filter. To confirm CD38 localization, we first permeabilized the cells by a 60 s incubation with Triton X-100 (0.001%, v/v), then stained the same cells with the anti-CD38 antibody C1586. The cells were counterstained with rhodamine-conjugated anti-mouse IgG. The green and red images obtained were then merged.
Figure 3
a shows the dense intracellular expression of the CD38-EGFP fusion protein detected as either EGFP fluorescence (green) or CD38 immunostaining (red). Figure 3b
demonstrates localization of immunoreactive CD38 to the plasma membrane not easily seen when EGFP fluorescence is monitored. Note there was also a perinuclear localization of both EGFP fluorescence and CD38 immunostaining confirming previous results (12)
. Several controls attest to the specificity of the CD38 staining. 1) Untransfected NIH3T3 cells, whether permeabilized or intact, did not immunostain. 2) Cells transfected with EGFP alone showed diffuse green fluorescence (c.f. Figs. 3
j and 9a–c), without CD38 immunostaining; this is consistent with cytosolic EGFP noted on Western blots, c.f. Figure 4
). 3) A control that omitted the anti-CD38 antibody did not stain as expected.
|
|
To further confirm a plasma membrane localization of CD38, we carried out the same experiments without permeabilization, i.e., on intact live cells. Even though diffuse intracellular EGFP fluorescence would be present, we thought that CD38 immunostaining would be localized only to the cell membrane because the cell would remain impermeable to the anti-CD38 antibody. This was indeed the case: Fig. 3e
shows that in intact cells, CD38 immunostaining was restricted to the cell surface whereas EGFP fluorescence was localized intracellularly and to the plasma membrane. The cells continued to exclude trypan blue, confirming an impermeable cell membrane. It is noteworthy that the anti-CD38 antibody recognizes the large carboxyl-terminal catalytic domain that is known to face extracellularly (12)
. Thus, the latter experiment not only confirmed the surface localization of CD38, but also suggested that its catalytic site is extracellular.
In separate experiments, we further confirmed plasma membrane localization by using a plasma membrane-specific dye, Dil. Figure 3g
shows EGFP localization and Fig. 3h
represents plasma membrane staining with Dil. An overlapping distribution is obvious by merging the red and green images, as shown in Fig. 3i
. Figure 3j-l
indicates transfection with EGFP alone, whereby it is obvious that plasma membrane localization is not seen. Figure 3m-o
shows control transfections with empty vector alone.
Next we examined for CD38-EGFP fusion protein localization to cellular subfractions: ER and nuclear membranes and the cytosol by Western blotting with an anti-EGFP antibody. Figure 4
shows that immunoreactive EGFP (Mr 18 kDa) rather than the CD38-EFGP fusion protein (Mr 63 kDa) was localized to the cytosol. In contrast, there was a strong band corresponding to immunoreactive CD38-EGFP in the ER fraction and a weaker band in nuclear membranes. The identity of the weak bands of lower Mr is unclear; possibly they represent degradation products. Using the NGD+cGDPr assay, we could confirm that cyclase activity was restricted to the ER and nuclear membranes rather than the cytosol.
Taken together, the results suggest that when expressed in CD38- NIH3T3 cells, CD38 is driven to the plasma membrane, endoplasmic reticulum, and nuclear membranes without any free enzyme in the cytosol. Second, the membrane association of the expressed CD38 mimics that of the endogenous enzyme in CD38+ MC3T3-E1 cells. Third, the results confirm our observations on the localization of CD38 to the nuclear membrane as a catalyst for intranuclear cADPr generation and Ca2+ signaling (12)
.
Ca2+ signaling through the expressed CD38-EFGP fusion protein
We and others have shown that NAD+-induced Ca2+ release is through the intermediacy of CD38 (3
, 19)
. By cyclizing NAD+, CD38 results in cADPr generation, which in turn interacts with microsomal membrane ryanodine receptors to gate Ca2+ release from intracellular stores (see ref 30
for a review). In osteoclasts, this response is blocked by ryanodine, an alkaloid that inhibits ryanodine receptor activation (19)
. In isolated nuclei, a specific cADPr receptor antagonist, 8 amino-cADPr similarly blocks NAD+-induced Ca2+ release (12)
. Having shown that CD38-EGFP possesses cyclase activity (i.e., the ability to catalyze NGD+ to cGDPr), we further investigated whether NAD+ could activate Ca2+ release in NIH3T3 cells transfected with the CD38-EGFP fusion protein.
Figure 5
A shows an important result, which is that NAD+ did not result in Ca2+ release in untransfected NIH3T3 cells. This is confirmatory of data by Franco et al. (17)
and suggests that CD38 is a prerequisite for NAD+-induced Ca2+ release. Figure 5B
shows a cytosolic Ca2+ response elicited in cells transfected with CD38 that closely mimicked the response obtained in CD38+ MC3T3-E1 cells (not shown). Successful transfection of cells with CD38 (without EGFP) was confirmed by immunocytochemistry with the anti-CD38 antibody C1586 (not shown).
|
Figure 5C-E
shows the variety of cytosolic Ca2+ responses obtained when cells transfected with the CD38-EGFP construct were exposed to NAD+. Transfection efficiency was 25% and measurements were made only in cells that showed EGFP fluorescence. The responses were either rapidly monophasic, monophasic after a latency (mean latency 2 min), or oscillatory. Application of ryanodine (5 µM) before NAD+ application attenuated the response magnitude (Fig. 5F
). As shown in Table 1
, statistical comparison of mean cytosolic Ca2+ levels (peak minus basal cytosolic Ca2+) of the treatment groups indicated no significant differences between responses elicited in MC3T3-E1 cells, NIH3T3 cells transfected with CD38, or NIH3T3 cells transfected with CD38-EGFP (P=0.74 and P=0.86, c.f. MC3T3-E1 cells, respectively). However, there was significant attenuation of the cytosolic Ca2+ response of CD38-EGFP transfectants pretreated with 5 µM ryanodine vs. those that were not (P=0.0001). This result strongly suggests that the cytosolic Ca2+ response to NAD+ is mediated through ryanodine receptor activation. Consistent with the latter functional evidence is our morphological demonstration for the colocalization of CD38 and ryanodine receptors in MC3T3-E1 cells.
|
Expression and function of CD38 mutants in NIH3T3 cells
Having successfully expressed CD38 in NIH3T3 cells, we determined whether the extracellular catalytic site was necessary for the function of CD38 in Ca2+ signaling. It has remained unclear how the extracellular conversion of NAD+ to cADPr is translated into a cytosolic Ca2+ signal (13)
. To better understand the molecular basis of this process, we made EGFP fusion proteins containing several mutated CD38s (Fig. 6
). We deleted the first 17 amino acids at the amino terminus (-17-CD38-EGFP) as well as these 17 along with 32 other transmembrane residues (-49-CD38-EGFP) (Fig. 6C, F
, respectively). We expected that although -17-CD38-EGFP would still localize to the membrane, the -49-CD38-EGFP construct would not be membrane associated. We then added a fanestrylation (Fan) or myristoylation (Myr) signal to the carboxyl and amino termini, respectively, to both full-length CD38-EGFP and -49-CD38-EGFP. We predicted that although the distribution of CD38-EGFP-Fan or Myr-CD38-EGFP would remain unchanged when applied to the deletion construct, -49-CD38-EGFP, these membrane association signals would reestablish membrane localization. Note, however, that the carboxyl-terminal catalytic site would now become cytosolic rather than extracellular. Thus, by driving CD38 to the plasma membrane in a reverse topology, we would in essence abolish any extracellular catalysis of NAD+.
|
We studied the expression of the CD38-EGFP mutant constructs in NIH3T3 cells by examining EGFP fluorescence (Fig. 7
). There was no significant difference between the expression pattern of CD38-EGFP, CD38-EGFP-Fan, Myr-CD38-EGFP, and -17-CD38-EGFP. A diffuse ER and perinuclear fluorescence together with some plasma membrane localization or a ‘membrane pattern’ was visualized with all four of these constructs. However, cells transfected with the deletion mutant -49-CD38-EGFP showed globular fluorescence without any membrane localization (Fig. 7a
). The same cells were fixed and immunostained with the anti-CD38 antibody (red) (Fig. 7b
). Merging the red and green images produced a complete overlap of the globular staining (Fig. 7c
). When live, EGFP-positive, trypan blue-negative cells without fixation or permeabilization were stained in a separate set of experiments with the same antibody, no staining was observed (Fig. 7e
), quite unlike the membrane staining seen in Fig. 3e
. Together, the results suggest that CD38 was not detected when probed using an antibody from the extracellular surface of the plasma membrane in -49-CD38-EGFP transfectants.
|
When membrane association signals Fan (not shown) or Myr were added to this construct, the fluorescence partly reverted back to a membrane pattern (Fig. 7f
), indicating that the membrane association sequences were able to direct CD38 localization to the plasma membrane as well. In fact, some globular staining was still evident. Membrane localization was better visualized on CD38 immunostaining (Fig. 7g
): an overlapping pattern of staining and EGFP fluorescence was noted in Fig. 7h
. However, when live, EGFP-positive, trypan blue-negative transfectants (Fig. 7i
) were incubated with the antibody, no immunostaining was observed (Fig. 7j
). Thus, despite our finding of a plasma membrane localization of Myr--49-CD38-EGFP in permeabilized cells, the catalytic epitope was not extracellular: thus, nonpermeabilized cells failed to bind the antibody when applied to the extracellular surface of the intact plasma membrane. This is consistent with the membrane localization of the Myr--49-CD38-EGFP construct, but in an opposite topology, i.e., with its catalytic site facing cytosolically rather than extracellularly.
Western blotting shown in Fig. 8
confirmed our EGFP data. CD38-EGFP (Mr 63 kDa) was localized primarily to the microsomal membrane fraction; it showed no cytosolic localization. Microsomal membranes from cells expressing the -17-CD38-EGFP mutant construct displayed an intense band at 63 kDa. There was no immunoreactivity in the cytosol. Microsomal membranes from cells expressing the -49-CD38-EGFP construct did not show an immunoreactive band at the expected 63 kDa position. However, when -49-CD38-EGFP was myristoylated, it became membrane associated and appeared in membrane blots.
|
The question therefore was, If not in the cytosol, where was the -49-CD38-EGFP protein localized? This was investigated in two ways. We first examined for any aberrant processing of this construct by looking for its localization relative to the Golgi. We compared the localization of EGFP (green) and the Golgi-specific dye boron dipyrromethene difluoride in cells transfected with CD38-EGFP and its mutants by high-resolution confocal microscopy. Figure 9
d–f shows there was a clear localization of the CD38-EGFP protein to the plasma membrane and the Golgi. -17-CD38-EGFP followed a similar pattern of localization consistent with its conserved transmembrane domain (Fig. 9g-i
). -49-CD38-EGFP, however, showed globules (c.f. Fig. 7a-c
) that did not localize either to the plasma membrane or the Golgi, indicating an aberrant localization of this construct (Fig. 9j-l
). Figure 9m-o
indicates that when a myristoylation signal was added, the Myr--49-CD38-EGFP construct showed a pattern of localization similar to the full-length or -17-CD38-EGFP constructs, consistent with an ER and plasma membrane localization.
|
In view of the absence of -49-CD38-EGFP in the Golgi, we next examined for its localization to the mitochondria. Western blots of mitochondrial membranes probed with the anti-CD38 antibody showed strongly positive bands (60 kDa) in membranes from the -49-CD38-EGFP transfectants, whereas only weak or no bands were observed with membranes from -17-CD38-EGFP transfectants (Fig. 8
, rightmost panel). This provided clear evidence for the mitochondrial localization of the -49-CD38-EGFP construct.
It is becoming increasingly clear that some bona fide endoplasmic reticular targeted cytochrome P450 proteins contain chimeric amino-terminal signals for targeting to mitochondria (21
22
23)
. A similar mechanism is used for mitochondrial localization of certain soluble proteins, such as GST isoenzymes and protein kinases (24)
. The NH2 termini of the CD38 proteins from rat, human, and rabbit in fact mimic the chimeric signals of cytochrome P450. Specifically, there are long 
-helical structures spanning 48 to 55 amino-terminal residues, and the sequence immediately carboxyl terminus these contains 20 to 29 amino acids rich in basic residues, resembling the cryptic mitochondrial targeting signals of the P450 apoproteins (21
, 22
, 23)
. These structures are consistent with -17-CD38-EGFP, which contains part of the amino-terminal -helical region, being poorly targeted to the mitochondria. In contrast, efficient mitochondrial localization is achieved with -49-CD38-EGFP in which the cryptic signal is fully exposed. It has indeed been demonstrated directly that the cryptic signal in cytochrome P450 is activated by the endopeptidase-mediated removal of 33 residues at the NH2 terminus. It remains to be seen, however, whether a similar mechanism is used to direct the CD38 mutant to the mitochondria. Our results are nevertheless supported by earlier studies showing such mitochondrial localization of CD38 certain physiological conditions (25
, 26)
.
Finally, we studied the functionality of each expressed fusion protein in terms of its ability to 1) convert NGD+ to cGDPr and 2) trigger cytosolic Ca2+ release in response to NAD+. All immunocytochemically detectable constructs formed cGDPr in the NGD+cGDPr assay (Fig. 8)
. All constructs released Ca2+ from intracellular stores akin to the full-length CD38-EGFP (Fig. 10
) or even native CD38 (in MC3T3-E1 cells) (Table 1)
. Because -49-CD38-EGFP that was not plasma membrane associated, as well as constructs such as Myr--49-CD38-EGFP whose catalytic site was facing intracellularly despite being membrane associated, showed evidence of NAD+-induced Ca2+ release, it is clear that extracellular surface catalysis of NAD+ was not the only mechanism for intracellular Ca2+ signaling. We have not measured the extracellular conversion of NGD+cGDPr because of the poor transfectability of these cells and high background noise.
|
Certain specific features should also be emphasized. First, a larger proportion of cells transfected with the Myr--49-CD38-EGFP construct showed oscillatory responses (12/18 vs. 3/24 in CD38-EGFP transfected cells) (Fig. 1C
). The significance of this remains unclear. Second, a proportion of Ca2+ signals associated with the Myr--49-CD38-EGFP and -49-CD38-EGFP-Fan construct was biphasic rather than monophasic. This means that the initial rapid elevation of cytosolic Ca2+ was followed by a more sustained elevation above basal (Fig. 10D, F
). Third, the mean change () in cytosolic Ca2+ in the -49-CD38-EGFP transfectants lacking the transmembrane domain was significantly higher than that of CD38-EGFP, -49-CD38-EGFP-Fan or Myr--49-CD38-EGFP (P=0.0207 vs. CD38-EGFP) (Table 1)
. Finally, the mean cytosolic Ca2+ in the -17-CD38-EGFP transfectants was significantly lower than that of CD38-EGFP (P=0.0027) (Table 1)
; the significance of this is unclear.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 27)
have investigated its function as a cell surface molecule, either an orphan receptor or an ectoenzyme. Studies have shown, for example, that CD38 is a counter-receptor for a lymphocyte adhesion molecule, CD31 (28)
. In addition, surface-expressed CD38 possesses several ectoenzymatic activities including ADP-ribosyl cyclase, cADPr hydrolase, and NAD+ glycohydrolase activities (3
4
5
6
7
, 29)
. Most studies have therefore attempted to unravel the apparent paradox of how the production of an unusually cationic molecule, cADPr, at the outer aspect of the plasma membrane can elicit an intracellular Ca2+ signal (13
, 30)
. Internalization of CD38 via a plasma membrane invagination after binding to NAD+ and the subsequent closure of the formed vesicle represent one proposed mechanism (15
, 16)
.
Franco et al. (17)
have provided strong evidence that CD38 is indeed the channel for transmembrane influx of cADPr. This makes biological sense and also explains the direct effects of extracellularly applied cADPr on Ca2+ release in cells such as splenic lymphocytes, cerebellar cells, and osteoclasts (4
, 31
, 32)
. The highly cationic cADPr could potentially associate tightly with the anionic head groups of membrane phospholipids, allowing the central cavity of CD38 to form a hydrophilic channel for its transport (17
, 30)
. This would clearly require dimeric association as well as the presence of extracellular NAD+, likely released from neighboring apoptotic cells (16
, 33
34
35)
. Our Western blots do not reveal the presence of dimeric complexes in either CD38+ MC3T3-E1 or CD38- NIH3T3 cells transfected with CD38. However, we used more drastic SDS-PAGE conditions, which could potentially dissociate dimers (33)
.
Our data suggest an additional mechanism that would depend on the presence of functional intracellular CD38. We demonstrate that the ER membranes of CD38+ MC3T3-E1 cells contain abundant immunoreactive CD38. Transfection of otherwise CD38- NIH3T3 cells with a CD38-EGFP fusion protein results in substantial ER membrane localization of the expressed protein. This localization pattern is almost identical to that of the endogenous protein, and provides strong evidence for cellular mechanisms that direct newly synthesized CD38 to the ER membrane in addition to plasma and nuclear membranes (12)
. We further show that CD38 expressed at the latter location possesses full cyclase activity and can release Ca2+ through cADPr generation.
The question therefore arises, Is ER CD38 somehow linked to cellular Ca2+ homeostasis? Colocalization of ryanodine receptors with CD38 would provide a biologically plausible and highly efficient system for the ‘juxtacrine’ control of cytosolic Ca2+ through cADPr produced locally from NAD+. We have shown here through confocal microscopy of CD38+ MC3T3-E1 cells that the distribution of the cyclase and ryanodine receptors is virtually overlapping. Moreover, ryanodine potently inhibits NAD+-induced Ca2+ release, suggesting that the effects of NAD+ are mediated through cADPr-induced ryanodine receptor activation.
That ER CD38 catalyzes intracellular NAD+ cyclization to cADPr with subsequent Ca2+ release suggests that the enzyme may function as a sensor for cellular NAD+. This raises an important topology question. Extrapolating from its intranucleoplasmic and plasma membrane topologies, it is likely that the catalytic site of ER CD38 is intravesicular. The presumed NAD+ transporter in the lumenal membrane should allow NAD+ access into the lumen for subsequent catalysis (16
, 36)
. The cADPr formed would then diffuse out into the cytosol, presumably through CD38 itself, and interact with the cytoplasmically located cADPr binding domain of the ryanodine receptor (17)
. This would prevent unrestricted consumption of the cytosolic NAD+.
It is clear from our results that NAD+ does gain intracellular access to cause Ca2+ release. Thus, even in the absence of plasma membrane CD38, as is the case with our mutant, -49-CD38-EGFP, full cytosolic Ca2+ responses were obtained. Such is also the case when fanestrylated or myristoylated -49-CD38-EGFP is expressed at the plasma membrane with the CD38 catalytic site facing cytosolically. These results indicate strongly that the action of NAD+ is also exerted intracellularly. In this respect, a bidirectional saturable NAD+ transporter has been described through systematic studies of NAD+ influx and efflux in CD38- cells. Most significant increases in cellular NAD+ levels were noted within 5 min of the extracellular application of NAD+ (16)
, although our studies indicate that such an influx is kinetically much more rapid. We observed cytosolic Ca2+ release within seconds even in transfectants that lacked CD38 surface expression. There was, however, an occasional response latency of several minutes when the full-length molecule was expressed at the plasma membrane in addition to the endoplasmic reticulum. The molecular basis underlying this delay remains unclear.
Although NAD+ can apparently traverse the cell membrane, our premise, based primarily on the abundance of CD38 in the endoplasmic reticulum and its colocalization with ryanodine receptors, is that CD38 ‘senses’ NAD+ and catalyzes NAD+cADPr cyclization intracellularly. The presence of the ER NAD+ transporter (16)
would permit NAD+ transport into the lumen where catalysis can take place. cADPr would then be released cytosolically through CD38 itself and induce Ca2+ release by action on ryanodine receptors. This will complete a ‘juxtacrine’ control mechanism. These events likely represent a ‘rapid’ pathway for translating changes in intermediary metabolism (reflected by changes in cellular NAD+ levels) into changes in cytosolic Ca2+ levels. One such example arises from our demonstration that cellular stress triggered by either mitochondrial DNA depletion or metabolic inhibitors results in changes in cytosolic Ca2+ levels, as well as changes in the expression of the ryanodine receptor gene and genes for several transcription factors (39)
. Should NAD+ play a role in translating such metabolic stress signals into changes in gene transcription via the CD38/NAD+/cADPr/Ca2+ pathway, this would certainly represent a unique and hitherto unrecognized mechanism for cell Ca2+ regulation and gene transcription in the broadest sense. The proposed regulation is likely more precise, at least theoretically, should NAD+ activate the inner nuclear membrane CD38 and result in cADPr generation and Ca2+ release within the nucleoplasm (12)
. Rapid sensing of NADP, another intermediary metabolite, and its conversion to the Ca2+-releasing second messenger NAADP+ may also be a function of intracellular CD38. Note that NAADP+-induced Ca2+ release is distinguishable pharmacologically from cADPr- or IP3-triggered release and involves the activation of a different set of Ca2+ stores (40
41
42)
.
Notwithstanding these ‘rapid’ transduction pathways triggered by intracellular CD38, a unifying hypothesis would make cell surface catalysis, followed by the ‘slower’ cytosolic translocation of cADPr, an equally important phenomenon. This will ensure not only that NAD+cADPr conversion occurs intracellularly, but also that NAD+ released extracellularly from cells either after apoptosis or due to efflux through the NAD+ transporter is converted to cADPr. It is also possible (particularly in actively metabolizing cells) that once the cellular CD38 enzymatic pool is saturated, cytosolic NAD+ concentrations rise to a level that permits efflux into extracellular space across a concentration gradient through the bidirectional NAD+ transporter (16)
. In this case, the extruded NAD+ would interact with surface CD38 on the same cell or on neighboring cells. Surface-expressed CD38 could then be a potentially important autocrine or paracrine regulator of Ca2+ signaling. Indeed, we will better understand the function of NAD+ in cellular Ca2+ homeostasis once we can identify cells that have a dinucleotide transporter (i.e., target cells for NAD+). This should also help us examine whether NAD+ has a role in cell-to-cell communication. Obtaining a molecular characterization of the NAD+ transporter thus remains the next logical step.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication August 24, 2001. Accepted for publication October 12, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, C). The data are expressed as percentages of control (mean±SD). E) Competitive inhibition of ADP-ribosyl cyclase activity by 400 µM NAD+ (
) to 19.4 µM in the presence of NAD+ (
