Research ArticleCell Biology

The Membrane-Bound Enzyme CD38 Exists in Two Opposing Orientations

See allHide authors and affiliations

Science Signaling  11 Sep 2012:
Vol. 5, Issue 241, pp. ra67
DOI: 10.1126/scisignal.2002700


The transmembrane enzyme CD38, a multifunctional protein ubiquitously present in cells, is the main enzyme that synthesizes and hydrolyzes cyclic adenosine 5′-diphosphate-ribose (cADPR), an intracellular Ca2+-mobilizing messenger. CD38 is thought to be a type II transmembrane protein with its carboxyl-terminal catalytic domain located on the outside of the cell; thus, the mechanism by which CD38 metabolizes intracellular cADPR has been controversial. We developed specific antibodies against the amino-terminal segment of CD38 and showed that two opposing orientations of CD38, type II and type III (which has its catalytic domain inside the cell), were both present on the surface of HL-60 cells during retinoic acid–induced differentiation. When activated by interferon-γ, human primary monocytes and the monocytic U937 cell line exhibited a similar co-distribution pattern. Site-directed mutagenesis experiments showed that the membrane orientation of CD38 could be converted from a mixture of type II and type III orientations to all type III by mutating the cationic amino acid residues in the amino-terminal segment of CD38. Expression of type III CD38 construct in transfected cells led to increased intracellular concentrations of cADPR, indicating the importance of the type III orientation of CD38 to its Ca2+ signaling function. The identification of these two forms of CD38 suggests that flipping the catalytic domain from the outside to the inside of the cell may be a mechanism regulating its signaling activity.


CD38 is a 45-kD transmembrane protein that was initially identified by antibody typing of lymphocytes (1, 2). The abundance of CD38 on the surface of lymphocytes changes depending on their differentiation stage, and ligation of CD38 with antibodies leads to lymphocyte activation (3). In addition, CD38 protects lymphocytes from infection with HIV by competitive binding to and sequestration of the primary receptor for viral entry, CD4 (4). That CD38 is also a signaling enzyme was first indicated by its sequence similarity with Aplysia adenosine 5′-diphosphate (ADP)–ribosyl cyclase (5, 6). It has since been well established that CD38 is a multifunctional enzyme that catalyzes not only the synthesis of cyclic ADP-ribose (cADPR) from nicotinamide adenine dinucleotide (NAD+) but also the hydrolysis of cADPR to form ADP-ribose (710). In addition, in acidic conditions, CD38 catalyzes the generation of nicotinic acid adenine dinucleotide phosphate (NAADP) from nicotinamide adenine dinucleotide phosphate (NADP+) (11). Both cADPR and NAADP are well recognized as Ca2+-mobilizing messengers in cells, targeting separate intracellular Ca2+ stores (1214). The central role of CD38, as well as of the messengers it produces, in regulating a wide range of physiological functions is indicated by the multiple defects revealed in CD38-knockout mice, including impairment in the chemotaxis of neutrophils (15), defective oxytocin release, and aberrant social behavior (16).

The molecular orientation of CD38 in the plasma membrane has been unresolved ever since its signaling functions were first established (17). In some cells, lymphocytes particularly, CD38 is found on the cell surface as a type II membrane protein with its catalytic C-terminal domain facing outward, which is followed by a single segment spanning the plasma membrane and a short N-terminal segment of 21 amino acid residues inside the cell (1, 3). It is thus not entirely clear how, given that the catalytic domain is on the outside of the cell, CD38 can produce and metabolize cADPR, an intracellular messenger. It has been suggested that cells can release the substrate NAD+ and that ecto-CD38, or nucleoside transporter in the plasma membrane, can function as an active transporter of the product cADPR back into the cells to mobilize the intracellular Ca2+ stores (18). Similar transport mechanisms for NAD+ and cADPR can potentially operate also in intracellular endosomal organelles (19).

Here, we explored a simpler alternative and investigated whether CD38 exhibited an alternative type III orientation, in which the C-terminal catalytic domain would face the cytoplasm, with the N-terminal tail facing the outside of the cell. We generated specific monoclonal and polyclonal antibodies against the N-terminal tail sequence of CD38. Our results showed that specific cell activation by agonists induced the generation of both type II and type III CD38. We found that a form of CD38 that was constructed as a type III protein was catalytically active in increasing cellular cADPR concentrations. Together, these results show that the type III orientation of CD38 is crucial to the signaling function of this enzyme.


Initially identified as an antigen on lymphocytes, CD38 is thought to be expressed on the cell surface as a type II transmembrane protein with the C-terminal segment of 256 residues exposed extracellularly (1). A soluble form of recombinant CD38 consisting of only the C-terminal segment has been produced in yeast and is fully active, having both the cADPR-synthesizing and the cADPR-hydrolyzing activities observed for native CD38 (20, 21). The location and structure of the catalytic site of CD38 have now been fully characterized by crystallography and mutagenesis experiments (20, 2224). That the site of cADPR production is extracellular whereas the site of its action is intracellular raises a topological paradox that has not been fully resolved (25).

If CD38 exhibited a type III orientation, with its catalytic C-domain inside the cell, then this would be a simple solution to the paradox. We investigated this possibility by developing specific antibodies against the N-terminal segment of CD38. BLAST analysis showed that the sequence of the first 21 amino acid residues of human CD38 is unique. We made use of this and generated both monoclonal and polyclonal antibodies (N-term) against the N-terminal sequence of human CD38. Monoclonal antibodies against the N-terminal segment were generated by hybridoma technology, and we selected the four best hybridoma clones. We tested the specificities of the monoclonal antibodies by Western blotting analysis of extracts from human embryonic kidney (HEK) 293T cells transfected with plasmid encoding full-length human CD38. The monoclonal N-term antibody recognized only a single band of ~43 kD (fig. S1A). Preincubation of the N-term antibody with the N-terminal peptide of CD38 completely blocked recognition. The same protein band was also recognized by OKT10, which is a widely used monoclonal antibody (C-term) specific for the C-terminal domain of CD38. Similar specificity was seen with endogenous CD38 in human HL-60 cells after they were induced to undergo differentiation with retinoic acid (RA) (fig. S1A). To further demonstrate that the N-term antibody indeed recognized CD38, we transfected HeLa cells with a plasmid encoding a fusion protein consisting of CD38 linked to enhanced green fluorescent protein (EGFP). Immunofluorescence staining with the N-term antibody showed colocalization with EGFP fluorescence, and this staining was completely blocked by preincubation of the N-term antibody with the N-terminal peptide of CD38 (fig. S1B). We also performed similar Western blotting and immunostaining analyses with a rabbit polyclonal antibody specific for the CD38 N-terminal peptide and confirmed its specificity (fig. S2). All of the results described in this study have been verified with both antibodies.

We previously demonstrated a temporal increase in the abundance of CD38 in HL-60 cells induced to undergo granulocytic differentiation by RA (26). To determine whether type III CD38 was produced during this process, we incubated live and intact RA-treated HL-60 cells with the monoclonal N-term antibody. We found a distinct population of cells that had a strong immunofluorescence signal. We sorted these cells by flow cytometry, and microscopic analysis showed that the immunofluorescence was localized on the cell surface (Fig. 1A). Preincubation of the monoclonal N-term antibody with the CD38 N-terminal peptide totally eliminated the immunostaining, and we could not detect any fluorescent cells by flow cytometry, which indicated the specificity of the antibody (Fig. 1, A and B). As a control, we immunostained RA-treated cells with IB4, a commonly used monoclonal C-term antibody against CD38, and subjected them to similar analysis. The fluorescence micrograph shows similar peripheral staining of CD38 compared to that of the N-term antibody (Fig. 1A).

Fig. 1

Presence of type III CD38 on the surface of RA-treated HL-60 cells. HL-60 cells were induced to differentiate to granulocytes with 1 μM RA. After 4 days of induction, cells were incubated with monoclonal N-term antibody against CD38 to reveal the presence of type III CD38 on the cell surface. Cell integrity during staining was monitored by costaining with the cell-impermeant DNA stain PI. Only cells that showed immunofluorescence for CD38 but did not have nuclear staining were considered truly positive and were collected by flow cytometry. (A) Micrographs of RA-treated HL-60 cells showing surface immunofluorescence (green), indicating the presence of type III CD38. Preincubation of the monoclonal antibody with the N-terminal peptide (N-term + Pep) blocked the immunostaining, indicating antibody specificity. Cells also exhibited the type II isoform of CD38, as shown by positive staining with IB4, a monoclonal C-term antibody against CD38. Cells treated with 0.1% Triton X-100 (N-term + Triton) to deliberately compromise membrane integrity showed both immunofluorescence for CD38 and nuclear staining (red). Scale bar, 10 μm. (B) Flow cytometry plots of two control treatments. Preincubation of the N-term antibody with the N-terminal peptide (+ Pep) blocked the immunostaining (top). Cell permeabilization with Triton X-100 (+ Triton) rendered all cells stained positively by the nuclear stain PI and also by N-term antibody (bottom). All live cells were gated inside the blue square, and cells positive for staining with the C-term or N-term antibodies were gated in the green polygon. (C) Flow cytometric analysis of the surface expression of type III CD38 (top) and type II CD38 (bottom). Intact HL-60 cells were harvested during the time course of treatment with 1 μM RA. At the indicated times, live cells were stained as described in Materials and Methods and analyzed by flow cytometry. (D) Time courses of the production of type II and type III CD38 during the treatment of HL-60 cells with RA. Data are representative of three independent experiments.

To ensure that the observed immunofluorescence was entirely a result of the binding of the N-term antibody to the outer surface of the plasma membrane, and thus that it was detecting only type III CD38, we routinely monitored cell integrity during antibody labeling by co-incubating the cells with a cell-impermeable stain for nucleic acid, propidium iodide (PI). Live cells capable of excluding a small molecule such as PI should be impermeable to large antibodies as well. Indeed, live and intact cells showed staining of the membrane with the N-term antibody and no nuclear staining with PI, whereas cells treated with a detergent to disrupt the plasma membrane showed both staining patterns (Fig. 1A). On the basis of these results, we considered only cells with immunostaining but no nucleic acid staining as positive. This procedure was also used in flow cytometric analysis, in which the vertical axis corresponds to staining by the N-term antibody, whereas the horizontal axis corresponds to nuclear staining (Fig. 1, B and C). Permeabilizing the cells with Triton X-100 made essentially all of the cells positive not only with PI but also with the N-term antibody (Fig. 1B) because the N-term antibody was then able to access the cytoplasmic domain of type II CD38.

We then analyzed the time course of the production of type III CD38 after induction of cellular differentiation (Fig. 1, C and D). We observed cells showing positive staining for type III CD38 1 day after induction, a proportion that increased to about 20% of the total cells after 4 days. The increased abundance of type III CD38 was concomitant with that of type II CD38, which was found in essentially all cells 4 days after induction. We obtained similar results from experiments with the polyclonal N-term antibody (fig. S3). To examine whether type II and type III CD38 were both produced upon cellular stimulation, we co-incubated RA-treated cells with both the N-term antibody and IB4, a C-term antibody against CD38. Immunofluorescence analysis indicated that both types of CD38 were indeed present on the surface of the same cells (Fig. 2A).

Fig. 2

Two distinct populations of CD38 detected by N-terminal antibodies. (A) Intact HL-60 cells, after 4 days of treatment with 1 μM RA, were harvested and incubated with a polyclonal N-term antibody (N-term, green) in the presence of PI for monitoring cell integrity. Cells were then fixed with 2% paraformaldehyde (PFA) and counterstained with IB4, which is specific for the C terminus of CD38 (C-term, red). (B) Capping of type III CD38 with the polyclonal N-term antibody was performed followed by counterstaining with the C-term antibody IB4, as described in Materials and Methods. White arrows indicate the capped type III CD38. (C) Preincubation of the polyclonal N-term antibody with the N-terminal peptide of CD38 blocked the immunostaining, indicating antibody specificity. (D to E) Experiments were performed as described in (B) to (C) except that a monoclonal N-term antibody against CD38 was used instead. White arrows indicate capped type III CD38. (F) Capped CD31 colocalized with CD38 (yellow spot indicated the white arrow in the merged micrograph). Scale bars, 10 μm (A and D to F) and 5 μm (B and C). Data are representative of three independent experiments.

We next showed that both types of CD38 found on the cell surface were separate entities in experiments with an antibody to induce capping of surface proteins and segregate them. Cells were first incubated with the polyclonal N-term antibody at 4°C, and then the cells were subsequently warmed to 37°C to enable energy-dependent capping. Type III CD38 was capped in a few localized areas on the cell surface, whereas type II CD38 was scattered throughout the cell periphery (Fig. 2B). Merging of both images showed no colocalization of the CD38 isoforms. As before, preincubation with the N-terminal peptide blocked staining only of the type III isoform (Fig. 2C). Similar capping results were seen in experiments with the monoclonal N-term antibody (Fig. 2D), although the extent of capping of type III CD38 was not as substantial as that with the polyclonal N-term antibody (Fig. 2B). The N-terminal peptide again blocked detection of only the type III isoform of CD38 (Fig. 2E). Positive controls were performed with an antibody against CD31, which is known to co-cap with CD38 (27). Indeed, we found substantial overlap between CD31 and CD38 in these experiments (Fig. 2F).

To determine how general the production of type III CD38 was, we used a different cell type, the human monocytic cell line U937, and a different stimulus, interferon-γ (IFN-γ). Activation of U937 cells with IFN-γ induced the production of both type II and type III CD38 (fig. S4). The percentage of type III–positive cells was ~12% after 3 days of treatment. In addition to examining mammalian cell lines (Fig. 1 and fig. S4), we wanted to investigate the possibility that type III CD38 was also found in primary cells. Immunofluorescence staining with the monoclonal N-term antibody showed that type III CD38 was found in human peripheral blood mononuclear cells (PBMCs) (5.30 ± 2.05% of cells were positive). The specificity of staining was confirmed by blocking with the N-term peptide, which reduced the population of positive cells to 0.81 ± 0.87% (Fig. 3A). It is likely that those PBMCs that were positive for the N-term antibody were monocytes because U937 cells, a monocytic cell line, also express type III CD38 (fig. S4). Moreover, we also investigated the relative distribution of type III and type II CD38 isoforms in PBMCs by costaining with the N-term and C-term antibodies. To monitor cell integrity, we used another impermeant DNA stain, 4′,6-diamidino-2-phenylindole (DAPI), which exhibits blue fluorescence. Flow cytometric analysis indicated the coexpression of type III and type II CD38 in ~1.66 ± 0.16% of PBMCs (Fig. 3B). In comparison, ~25% of cells had type II CD38 alone, which may be an underestimate because of the use of a weaker red fluorophore (Alexa Fluor 555–conjugated donkey anti-rabbit antibody). Indeed, when we analyzed the immunostaining by the C-term antibody with the stronger green fluorophore (Alexa Fluor 488–conjugated donkey anti-rabbit antibody) under the same conditions, we found that at least 53% of the cells were positive for type II CD38 (fig. S5). About 1.5% of the cells exhibited only type III CD38. This percentage may be an overestimate because cells with low abundance of type II CD38 in this population may have been missed because of the weaker red fluorophore used. Microscopic observations of the sorted cells that exhibited both type II and type III CD38 indicated that the expression pattern varied from cell to cell, but generally showed punctate patterns on the cell surface. Both types of CD38 did not necessarily colocalize on the surface, as shown by the analysis of merged images (Fig. 3C).

Fig. 3

Coexpression of type II and type III CD38 on the surface of human PBMCs. Human PBMCs were purified by Ficoll-Hypaque sedimentation and incubated with the monoclonal N-term antibody (N-term, green) in the presence of PI to monitor cell integrity or were incubated with both the monoclonal N-term antibody and the polyclonal C-term antibody (C-term, red) in the presence of DAPI to monitor cell integrity. (A) Flow cytometric analysis of the surface expression of type III CD38 (green area, left) and showing that preincubation of the N-term antibody with the N-terminal CD38 peptide (+ Pep) blocked immunostaining (right). The bar chart shows the mean percentage of cells that had type III CD38. n = 3 experiments; *P < 0.05 by t test. (B) Flow cytometric analysis of the costaining of type III (green) and type II (red) CD38 (left) in PBMCs and of the blocking effect of preincubating the antibodies with both the N-terminal peptide and recombinant CD38 protein (+ Block) (right). The green area indicates cells that have only type III CD38, whereas the red area indicates cells that have only type II CD38. The orange area indicates cells that have both type II and type III CD38. The bar chart shows the mean percentage of cells that had both type II and type III CD38. n = 3 experiments; *P < 0.05 by t test. (C) Micrographs of positive cells from costaining of type III (green) and type II (red) CD38, showing surface immunofluorescence indicating the presence of type III and type II CD38. Scale bar, 10 μm. Data are representative of three independent experiments.

It is generally believed that the polarity of membrane proteins is determined predominantly by charged residues flanking the hydrophobic core of the transmembrane segment (28). The side with the most net positive charges is generally cytosolic, following the “positive-inside rule.” For CD38, the number of positive charges is about the same on both sides of the transmembrane segment, suggesting the possibility that both polarities can be found. To determine whether the positive-inside rule applied to CD38, we made a construct by attaching a fluorescent tag, reduction-oxidation–sensitive GFP (roGFP), to the C terminus of CD38, and we mutated all four of the positively charged amino acid residues on the N-terminal side of CD38 to serine or aspartate residues (mutCD38-roGFP) (Fig. 4A). As a control, we attached roGFP to the C terminus of wild-type CD38 (CD38-roGFP) (Fig. 4A). We transfected HEK 293T cells with plasmids encoding either construct. We observed that mutCD38-roGFP was found mainly inside the cells (Fig. 4B), and its pattern of expression was similar to that of endoplasmic reticulum–targeted red fluorescent protein (ER-RFP) (Fig. 4, C and D), an ER marker coexpressed in the same cells. Indeed, both proteins showed substantial colocalization (Fig. 4D). The distribution pattern of mutCD38-GFP was different from that of CD38-roGFP (Fig. 4E), which was essentially all at the plasma membrane and distinct from the ER (Fig. 4, F and G).

Fig. 4

Microscopic analysis of HEK 293T cells transfected with plasmids encoding CD38 constructs. (A) Diagram of the constructs showing the sequence of CD38 and mutCD38 on both sides of the transmembrane domain (TM), with the positive charges highlighted in red and the negative charges highlighted in green. (B to G) HEK 293T cells were transiently cotransfected with plasmid encoding the ER marker ER-RFP (red) together with plasmids encoding either mutCD38-roGFP (green) or CD38-roGFP (green), as indicated. Overlay images in (D) and (G) indicate that mutCD38-roGFP showed 84 ± 9% colocalization with ER-RFP as determined by image analysis of four different regions, whereas wild-type CD38-roGFP did not. Scale bars, 10 μm. Data are representative of three independent experiments.

We next used the fluorescence protease protection assay (29) to determine the membrane topology of the mutCD38 (Fig. 5A). After permeabilization of cells with digitonin, trypsin was added, which effectively cleaved and reduced the fluorescence of mutCD38-roGFP. In contrast, the fluorescence of the ER-luminal marker ER-RFP in the same cells was protected from trypsin by the ER membrane, and its red fluorescence remained stable. Another test for the topology of the mutCD38 was to determine whether it was glycosylated. Native CD38 is glycosylated at four sites in the C-terminal domain (1). Glycosylation generally occurs inside the ER. If the mutCD38 had a type III orientation, its C-terminal domain would be facing the cytosol and it would not be glycosylated. We found that the band corresponding to wild-type CD38-roGFP was larger than that of mutCD38-roGFP (Fig. 5B), and treatment with a deglycosylating enzyme [peptide N-glycosidase (PNGase)] reduced the band size to that corresponding to mutCD38. The mutCD38-roGFP protein, which was similar in mass to deglycosylated wild-type CD38-roGFP, was not sensitive to PNGase, indicating it was not glycosylated.

Fig. 5

Membrane orientation of mutCD38-roGFP in transfected cells. (A) HEK 293T cells were cotransfected with plasmids encoding mutCD38-roGFP (green) and ER-RFP (red), and cells were then analyzed by fluorescence microscopy. Digitonin (50 μM) was then added to permeabilize the cells, which were then washed and treated with 20 μM trypsin, which cleaved the roGFP-tag from mutCD38-roGFP, resulting in a decrease in green fluorescence, whereas the control ER-RFP (red), which was found inside the ER, was protected. The inset shows the averaged fluorescence changes in 12 measurements. The data for the right bars were from the 8-min time point. (B) Western blotting analysis of wild-type CD38-roGFP and mutCD38-roGFP. Deglycosylation with PNGase resulted in a reduction of the size of band corresponding to wild-type CD38-roGFP. In contrast, mutCD38-roGFP was smaller in size than was wild-type CD38 and was not sensitive to PNGase. (C) Analysis of the redox state sensed by the roGFP-tag. HEK 293T cells were transfected with the plasmids encoding wild-type CD38-roGFP, mutCD38-roGFP, or cytosolic roGFP (cyto roGFP). The ratio of the 390- and 480-nm fluorescence measurements of roGFP was determined. The ratio measured for CD38-roGFP was higher than that of either mutCD38-roGFP or cytosolic roGFP, indicating that it sensed the oxidized state of the extracellular medium. Further oxidation with diamide did not increase the ratio, but the ratio was decreased by reduction with DTT. In contrast, the lower ratio values of both mutCD38-roGFP and cytosolic roGFP were readily increased by diamide, indicating that both sensed the reductive state of the cytosol. Data are representative of three independent experiments.

A further test of the membrane topology of CD38 was to make use of the fluorescence property of the tag, roGFP, which we attached to the C terminus of mutCD38. The roGFP protein is a ratiometric fluorescence probe for redox potential (30). We found that the value of the measured fluorescence ratio of mutCD38-roGFP (~0.6) was similar to that of roGFP found in the cytosol (~0.4), which indicated that both proteins sensed the reductive state of the cytosol (Fig. 5C). Because roGFP was attached to the C terminus of mutCD38-roGFP, the results thus confirmed that mutCD38 was in the type III orientation. For wild-type CD38-roGFP expressed on the cell surface (Fig. 4E), the roGFP ratio was ~1.2, consistent with the oxidized environment of the extracellular medium. Further support for the membrane topology came from the addition of the oxidizing agent diamide to the cells, which did not result in an increase in the roGFP ratio of CD38-roGFP, but effectively increased the roGFP ratio when it was added to cells expressing either mutCD38-roGFP or cytosolic roGFP. These effects were reversed by subsequent reduction with dithiothreitol (DTT). Together, these results indicate that the mutations indeed altered the membrane orientation of CD38, switching it from a type II to a type III protein.

In the type III orientation, the catalytic domain of CD38 faces the cytosol, an environment that is different from the extracellular medium to which the catalytic domain of type II CD38 is exposed. Thus, it was important to determine whether type III CD38 was catalytically functional. We could not use either HL-60 or U937 cells for the required experiments, because both cell types have both forms of CD38 and it is difficult to distinguish them. We therefore focused on type III mutCD38-roGFP expressed in HEK 293T cells.

CD38 has six disulfide bonds in its C-terminal catalytic domain, and all of them are critical for its catalytic activity, except for the last one, which occurs between Cys287 and Cys296 (22, 31). We previously showed in experiments with the monoclonal antibody HB-7 that a soluble form of CD38 in the cytosol has all of the critical disulfide bonds (31), indicating that disulfide formation in CD38 is independent of the ER. Here, we found that this was also true for type III CD38 (Fig. 6A). We also previously showed that HB-7 binds to CD38 only when the disulfide bonds are intact (31). Here, we found that HB-7 bound to mutCD38-roGFP (Fig. 6A), indicating that the mutant isoform had intact disulfide bonds. In the reduced state containing DTT, which would be expected to break all of the disulfide bonds in either CD38 isoform, we found that the binding of HB-7 to either wild-type CD38-roGFP or mutCD38-roGFP was lost.

Fig. 6

mutCD38-roGFP has intact disulfide bonds and is biologically active. (A) The monoclonal antibody HB-7 was used to detect intact disulfides in CD38. Western blotting analysis showed that HB-7 recognized mutCD38-roGFP, but not when the disulfide bonds in mutCD38-roGFP were reduced by 100 mM DTT. Similar results were seen for wild-type CD38-roGFP, which is known to have intact disulfides. (B) HEK 293T cells were transfected with plasmid encoding mutCD38-roGFP, and the amount of intracellular cADPR was measured at the indicated times. Increases in intracellular cADPR concentration coincided with the increased abundance of mutCD38-roGFP (inset). (C) When normalized to the amounts of the exogenous proteins (inset), mutCD38-roGFP was more effective than wild-type CD38-roGFP in increasing intracellular cADPR concentrations. Data are representative of three independent experiments.

The mutCD38 not only had intact disulfides, but it was also biologically active and increased the concentration of cellular cADPR in a time-dependent manner concomitant with production of the protein (Fig. 6B). Cells expressing mutCD38-roGFP contained increased amounts of cADPR compared with those expressing wild-type CD38-roGFP 48 hours after transfection (Fig. 6C, right panel). Thus, CD38 in the type III orientation was fully functional as a signaling enzyme and produced the Ca2+ signaling messenger cADPR in cells. To test whether the roGFP tag fused to mutCD38 contributed to the orientation change, we constructed a mutCD38 without the tag and expressed it in HEK 293T cells. The non-fusion mutCD38 protein localized intracellularly in ER-like membrane structures (fig. S6A). More than half of the mutCD38 proteins were not glycosylated, as indicated by their unchanged molecular mass on Western blots after treatment with PNGase (fig. S6B). The mutCD38 was more effective than wild-type CD38 in increasing the intracellular concentration of cADPR (fig. S6C). Indeed, after normalizing for protein abundance, the mutCD38 produced 60 to 70 times more cADPR than did wild-type CD38 (fig. S6C).


It is well established that the enzyme CD38 produces cADPR and NAADP, both of which are Ca2+-mobilizing messengers that target different Ca2+ stores inside cells (32, 33). Equally well established is that CD38 is responsible for regulating a wide range of physiological functions (15, 16, 34). Thus, the molecular orientation of CD38 has far-reaching implications for how it functions as a signaling enzyme. Here, we showed that cells can express both type II and type III CD38. We suggest that the type III orientation, with its catalytic domain facing the cytosol, should be more suitable than the type II isoform for performing intracellular signaling functions. In addition to its substrate NAD+, the sites of action of the products of CD38, cADPR and NAADP, are all cytosolic. That the catalytic domain is in the cytoplasm also makes it amenable to a wide range of regulatory mechanisms, such as phosphorylation or interaction with various proteins. It could be argued that type III CD38 may only play a minor role, because only 10 to 20% of the cells in a population have this form (Fig. 1 and fig. S4), and it is even less abundant in primary PBMCs (Fig. 3). The low percentage could be because of the sensitivity of our immunofluorescence assays, limiting our ability such that we can detect only those cells with the greatest amounts of type III CD38. Because it is a signaling enzyme, it is likely not necessary for CD38 to be present in large amounts.

Mechanisms have been put forward that suggest how type II CD38 might influence cytosolic signaling despite having its catalytic domain on the outside of the cell (35, 36). These involve transporting the substrate NAD+ and the product cADPR across plasma or organellar membranes (18, 19). These mechanisms are potentially slow and may be more suitable for long-term regulation, such as cell proliferation, as has been described (37, 38). On the other hand, type III CD38 would be more suitable for a rapid response to an acute stimulation, such as hormone receptor activation. That cells can have both forms of CD38 is consistent with the wide range of functions that are regulated by this enzyme. CD38 is not the only membrane protein that can exist in two opposing orientations. Another example is the prion protein, which also can be synthesized in two transmembrane orientations as well as in a glycosylphosphatidylinositol-linked form. Trans-acting protein factors have been described that can direct prion proteins toward different topologic fates (3942). Another case is the melanocortin-2 receptor accessory protein (MRAP), which exists in both orientations in comparable amounts (43).

An important factor in determining the membrane orientation of CD38 was the positively charged amino acid residues located near both sides of the transmembrane segment, as was postulated by the positive-inside rule (44). We found that changing the cationic residues of the N-terminal side of CD38 to serine and aspartate residues converted the type II orientation to that of type III. This finding could have important physiological implications because the N-terminal segment also has three serine residues that could be phosphorylated. The increase in negative charges from phosphorylation could be equivalent to mutating the cationic residues to aspartate as shown here, and physiological phosphorylation of the N-terminal segment of CD38 could thus influence the amount of type III CD38 found in the cells. Indeed, there is evidence that the N-terminal region of CD38 might be phosphorylated in a microglial cell line upon stimulation by abscisic acid (45).

Here, we showed that mutations of charged residues in the N-terminal tail of CD38 directed its expression in the ER in a type III orientation. That CD38 is found in organelles has been well documented, and evidence suggests that CD38 is in the type III orientation in these situations (46, 47), consistent with our results. Furthermore, we showed that the type III CD38 in the ER was more efficient than was the wild-type, type II CD38 in the plasma membrane in synthesizing cADPR. More intriguing perhaps is the hypothesized involvement of the formation of lipid droplets to facilitate the escape of polypeptides from the ER, the site of the synthesis of membrane proteins (48). This mechanism has been invoked to account for the presence of the fully glycosylated class I major histocompatibility complex, a transmembrane protein, in the cytoplasm (49). Fusion of lipid droplets carrying the extracted membrane proteins with the inner surface of the plasma membrane could potentially enable the insertion of these proteins in alternate orientations. Whether this lipid mechanism is responsible for the type III orientation of CD38 remains to be determined.

Materials and Methods


Recombinant CD38 was prepared with a yeast expression system as described previously (21). The plasmid proGFP-N1(ro1) was obtained from S. J. Remington (University of Oregon) (30). The plasmid pDsRed-ER was a gift from J. Marchant (University of Minnesota). The expression plasmids pcDNA3.1(+) and pEGFP(N1) were obtained from Clontech Laboratories Inc. Lipofectamine 2000 and trypsin were purchased from Invitrogen. Alcohol dehydrogenase from yeast, diaphorase from Clostridium kluyveri, NAD+, nicotinamide, nucleotide pyrophosphatase from Crotalus atrox venom, resazurin, tri-n-octylamine, digitonin, diamide, DTT, PNGase, poly-l-lysine, and PI were obtained from Sigma. Chloroform was purchased from Merck. Western blotting detection regents were from Millipore and GE Healthcare. Monoclonal and polyclonal antibodies against the first 21 amino acid residues of the N terminus of CD38 were custom-made by Absea and Genemed Synthesis Inc. The monoclonal antibody was produced from the most potent clone of a hybridoma, affinity-purified, and used for live-cell immunostaining and Western blotting analysis. A mouse monoclonal antibody against CD38 (T16) and a mouse monoclonal antibody against CD31 were from Santa Cruz Biotechnology and were used for Western blotting analysis and capping experiments, respectively. The antibodies IB4 and OKT10 were gifts from F. Malvasi (University of Torino Medical School). Mouse monoclonal anti-GFP antibody was obtained from Sigma. Alexa Fluor–conjugated donkey anti-mouse or anti-rabbit immunoglobulin G (IgG), horseradish peroxidase–conjugated goat anti-mouse or anti-rabbit IgG, and DAPI were obtained from Invitrogen. HEK 293T cells were a gift from J. B. Yue (The University of Hong Kong) and were grown in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (FBS), penicillin (100 U/ml), and streptomycin (100 μg/ml). The cultures were maintained at 37°C in a humidified atmosphere of 5% CO2.

Culture of HL-60 and U937 cells and induction of cellular differentiation

HL-60 and U937 cells were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% FBS, penicillin (100 U/ml), and streptomycin (100 μg/ml). The cultures were maintained at 37°C in a humidified atmosphere of 5% CO2. Before induction of differentiation by RA, log-phase HL-60 cells were diluted to a concentration of 2 × 105 cells/ml. A final concentration of 1 μM RA, diluted from a stock solution of 10 mM in dimethyl sulfoxide (DMSO), was added to the HL-60 cells. Cells treated with the same concentration of DMSO served as controls. At different times after RA treatment, the cells were harvested by centrifugation at 200g for 5 min. Similarly, log-phase U937 cells were diluted to 2 × 105 cells/ml before induction with IFN-γ at a final concentration of 500 U/ml, which was diluted from a 40 × 104 U/ml stock solution in 50 mM Hepes (pH 7). Cells treated with the same volume of vehicle served as controls. At various times, cells were harvested by centrifugation at 200g for 5 min. HL-60 and U937 cells were gifts from J. Wong (Hong Kong University of Science and Technology) and L. C. Chan (The University of Hong Kong), respectively.

Immunostaining of live cells and flow cytometry

HL-60 cells (1 × 106) after treatment with RA and U937 cells (1 × 106) after treatment with IFN-γ were harvested by centrifugation at 200g for 5 min at 4°C and washed once with phosphate-buffered saline (PBS) before being incubated in ice-cold PBS containing PI (10 ng/ml), monoclonal N-term antibody (5 μg), and 5% normal donkey serum for 30 min on ice to prevent internalization of the antibody. PI, a cell-impermeant DNA stain, was added to monitor cell integrity during the incubation. Antibodies and PI were then removed by centrifugation at 100g for 5 min at 4°C, and the cells were washed twice with ice-cold PBS. Cells were then fixed in 2% PFA for 15 min, washed twice with PBS, and incubated with Alexa Fluor 488–conjugated donkey anti-mouse antibody (at 1:1000 dilution) at room temperature for 30 min. The fluorescently labeled cells were analyzed with an LSR II flow cytometer (BD Biosciences) and sorted with a FACSAria (BD Biosciences) equipped with FACSDiva software (BD Biosciences). Only those cells with positive immunolabeling signals but no PI signals were considered as true positives. The immunofluorescence staining of these cells was entirely a result of binding of the antibody to the external surface, because these cells were able to exclude PI and were thus intact during the incubation. As another control, cells were also deliberately permeabilized with 0.1% Triton X-100 during the incubation with the antibodies and PI. These cells all showed nuclear staining by PI. Specificity of immunostaining was shown by preincubation of the antibodies in the presence of a 10-fold excess concentration of the N-terminal peptide overnight at 4°C, which resulted in complete blockage of the staining.

Isolation of primary human PBMCs and live-cell co-immunostaining

PBMCs were isolated by Ficoll-Hypaque (GE Healthcare) gradient centrifugation of whole blood samples obtained from the Hong Kong Red Cross Blood Transfusion Service. Briefly, human whole blood samples were diluted with an equal volume of PBS. The diluted blood samples were layered on 13 ml of Ficoll-Hypaque and centrifuged at 400g for 40 min at room temperature. The upper layer of plasma was collected to prepare autologous serum, and the thin layer of PBMCs was isolated and washed with PBS. Red cell lysis buffer [ammonium chloride (8.26 g/liter), potassium bicarbonate (1 g/liter), and EDTA (0.037 g/liter)] was added and incubated for 5 min at room temperature. The PBMCs were washed with PBS three times and centrifuged at 100g for 5 min at room temperature. Cell pellets were resuspended in RPMI 1640 medium supplemented with 10% autologous serum, penicillin (100 U/ml), and streptomycin (100 μg/ml). The PBMCs (5 × 107 cells per flask) were incubated at 37°C in 5% CO2 and were collected 19 hours later with a 40-μm cell strainer to get rid of cell clusters. Cells that attached to the bottom of the flask were incubated in ice-cold PBS/EDTA (0.02%) for 30 min at 4°C, and then the flasks were tapped firmly to dislodge the cells. The cells were then washed twice with PBS and centrifuged at 100g for 5 min at room temperature. The PBMCs (1 × 107 to 1.5 × 107) were used for live-cell staining. The cells were costained with monoclonal N-term antibody (15 μg) and polyclonal antibody (6 μg) against recombinant CD38 at 4°C in the presence of DAPI (70 ng/ml), a cell-impermeant DNA stain, to monitor cell integrity. The cells were then fixed with 2% PFA and incubated with Alexa Fluor 488–conjugated donkey anti-mouse IgG (for the monoclonal N-term antibody) and Alexa Fluor 555–conjugated donkey anti-rabbit IgG (for the polyclonal antibody). Specificity of staining was verified by blocking the antibodies with either the N-term peptide (at 10-fold excess, 150 μg) or recombinant CD38 protein (at 10-fold excess, 60 μg), as appropriate. Controls with secondary antibody alone and isotype controls such as rabbit IgG and mouse IgG were used to set the baseline for flow cytometry. Flow cytometric analysis and sorting were performed on a BD FACSAria or a BD FACSAria SORP with FACSDiva software in the Faculty Core Facility, The University of Hong Kong. All data were analyzed with FlowJo software (TreeStar).

Immunofluorescence microscopy and Western blotting analysis

Aliquots of RA-treated HL-60 cells or IFN-γ–treated U937 cells after live-cell immunostaining were centrifuged onto collagen-coated coverslips with a Shandon Cytospin centrifuge (Thermo Scientific). Sorted human PBMCs were allowed to attach to poly-l-lysine–coated coverslips by gravity. The coverslips were then mounted onto a glass slide with ProLong Gold (Invitrogen). Fluorescence images of HL-60 cells were recorded with a laser confocal microscope (Olympus IX 71, Olympus) equipped with a 60× PlanApo oil objective and Fluoview software (Olympus). Fluorescence images of human PBMCs cells were recorded with a Zeiss LSM 710 confocal laser scanning microscope equipped with a 63×/1.4 oil objective and Zen 2009 software (Carl Zeiss). HEK 293T cells were grown on poly-d-lysine–coated dishes (FluoroDish, World Precision Instruments Inc.). Forty-eight hours after transfection with Lipofectamine 2000, images were acquired with a Zeiss LSM 510 META confocal laser scanning microscope. Western blotting analysis was performed according to standard procedures (50). (We are grateful to S. Bruzzone for the suggestion of using nonreducing conditions when performing Western blotting analysis with the IB4 and OKT10 antibodies.) Prestained molecular mass standards were resolved on the same SDS–polyacrylamide gel electrophoresis (SDS-PAGE) gels in all Western blots to indicate molecular mass standards. For detection of disulfide bonds in CD38, reducing and nonreducing SDS-PAGE were performed as previously described (31). Antibodies were used at different dilutions: Ab77, 1:2000; HB-7, 1:1000; IB4, 0.3 μg/ml; OKT10, 0.3 μg/ml.

Capping experiments

After 4 days of treatment with RA, as described earlier, HL-60 cells (1 × 106) were harvested at 200g for 5 min and incubated with monoclonal (5 μg) or polyclonal N-term (0.5 μg) antibodies or with antibody against CD31 (3 μg) for 30 min on ice. Cells were then washed twice with PBS and incubated with Alexa Fluor 488–conjugated donkey anti-rabbit IgG on ice. Cells were then warmed to 37°C for 45 min to induce capping, and the process was stopped by the addition of ice-cold PBS containing 0.5% bovine serum albumin and 0.1% sodium azide. Cells were fixed with 2% PFA, counterstained with the C-term antibody IB4 (0.5 μg), and then further incubated with Alexa Fluor 555–conjugated donkey anti-mouse IgG. Cells were analyzed under an Olympus laser confocal microscope.

Construction of the proGFPN1-CD38 and proGFPN1-mutCD38 plasmids

The complementary DNA (cDNA) encoding full-length wild-type human CD38 was subcloned into the plasmid proGFP-N1 at the Kpn I and Hind III restriction sites to generate proGFPN1-CD38, which encodes CD38 fused with roGFP at the C terminus. To construct proGFPN1-mutCD38, we generated the mutations K13D, R17D, R20S, and R21D in proGFPN1-CD38 by mutagenesis reactions with the QuikChange Multi Site-Directed Mutagenesis Kit (Agilent Technologies) according to the instructions of the manufacturer. The N-terminal 21–amino acid residue sequences were MANCEFSPVSGDKPCCRLSRR and MANCEFSPVSGDDPCCDLSSD for wild-type CD38 and mutCD38, respectively.

Ratiometric measurement of roGFP and its fusion protein

Diamide and DTT were added at final concentrations of 500 μM and 10 mM, respectively, to HEK 293T cells transiently transfected with proGFP-N1, proGFPN1-CD38, or proGFPN1-mutCD38. The cells were monitored with an Olympus IX81 microscope. Dual excitation ratio imaging was acquired with D390/22X and D480/20 excitation filters, a 505LP dichroic mirror, and a D535/25 emission filter (Chroma Technology Corp.) alternated by a fast filter changer. Images were collected with an iXon3 EMCCD camera (Andor Technology). Data were collected and processed with the CellR software (Olympus). Fluorescence excitation ratios were obtained by dividing integrated intensities obtained from manually selected portions of the imaged regions of intact whole cells collected with 390 ± 10 nm and 480 ± 20 nm excitation filters after appropriate background correction. Background correction was performed by subtracting the intensity of a nearby cell-free region from the signal of the imaged cell.

Fluorescence protease protection assay

This assay was performed as described previously (29) with minor modifications. Briefly, HEK 293T cells were cotransfected with proGFP-N1 or proGFPN1-mutCD38 and pDsRed-ER, seeded on plates coated with poly-l-lysine (10 μM), and monitored with an Olympus IX81 microscope. The fluorescence signals were collected with both 480-nm excitation/535-nm emission and 535-nm excitation/610-nm emission filters. Digitonin at a final concentration of 50 μM (the concentration was experimentally determined on each batch of cells) was added to the bath. After 1 min of incubation and two washes with KHM buffer (110 mM potassium acetate, 20 mM Hepes, 2 mM MgCl2) to remove the digitonin, 20 μM trypsin was applied. The green and red signals were analyzed with the CellR system.

Measurement of the intracellular production of cADPR

HEK 293T cells (3 × 105) were seeded in a 3.5-cm dish and transfected with proGFPN1-mutCD38 or proGFPN1-CD38 as described earlier. After 48 hours or as indicated in the figure legends, the cells were harvested. Cell pellets were lysed with 0.6 M perchloric acid. After centrifugation, the endogenous cADPR in the supernatant was measured by a cycling assay (51), and the pellets were redissolved in 1× Laemmli buffer (300 μl), of which 30 μl was resolved by SDS-PAGE and analyzed by Western blotting with the anti-CD38 antibody Ab77. Protein bands were visualized with the Amersham ECL Chemiluminescence Kit on a ChemiDoc XRS+ imager (Bio-Rad). Image Lab software (Bio-Rad) was used to calculate relative quantification of the specific bands recognized by Ab77. Results given as picomoles of cADPR per unit of recombinant protein were calculated by dividing the total amount of cADPR in one well by the relative amount of the recombinant protein.

Statistical analysis

Data shown in each figure are means ± SD from at least three experiments. Statistical significance (P < 0.05) was determined by t test with SigmaStat 3.1 (Systat Software).

Supplementary Materials

Fig. S1. Validation of the monoclonal N-term antibody.

Fig. S2. Validation of the polyclonal N-term antibody.

Fig. S3. The polyclonal N-term antibody demonstrated the presence of type III CD38 in RA-treated HL-60 cells.

Fig. S4. Detection of both forms of CD38 on the surface of IFN-γ–treated monocytes.

Fig. S5. Immunostaining of human PBMCs with the C-term antibody against CD38.

Fig. S6. The mutCD38 protein is biologically active.

References and Notes

Acknowledgments: We thank J. M. Garcia and C. Shek from the Institute of Pasteur (Hong Kong) for advice and operation of the LSRII flow cytometer. The IB4 and OKT10 antibodies were gifts from F. Malavasi, University of Torino Medical School. Funding: This work was supported by grants from the Hong Kong General Research Fund (nos. 768408, 769309, 770610, and 771011) and the National Natural Science Foundation of China/Research Grants Council (NSFC/RGC) Joint Research Scheme (no. N_HKU 722/08) to H.C.L. Author contributions: Y.J.Z. and C.M.C.L. performed the experiments and analyzed the results, and H.C.L. conceived the study, designed the experiments, and wrote most of the paper. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article