NAADP Mobilizes Calcium from the Endoplasmic Reticular Ca2+ Store in T-lymphocytes*

The target calcium store of nicotinic acid adenine dinucleotide phosphate (NAADP), the most potent endogenous calcium-mobilizing compound known to date, has been proposed to reside in the lysosomal compartment or in the endo/sarcoplasmic reticulum. This study was performed to test the hypothesis of a lysosomal versus an endoplasmic reticular calcium store sensitive to NAADP in T-lymphocytes. Pretreatment of intact Jurkat T cells with glycyl-phenylalanine 2-naphthylamide largely reduced staining of lysosomes by LysoTracker Red and abolished NAADP-induced Ca2+ signaling. However, the inhibitory effect was not specific since Ca2+ mobilization by d-myo-inositol 1,4,5-trisphosphate and cyclic ADP-ribose was abolished, too. Bafilomycin A1, an inhibitor of the lysosomal H+-ATPase, did not block or reduce NAADP-induced Ca2+ signaling, although it effectively prevented labeling of lysosomes by LysoTracker Red. Further, previous T cell receptor/CD3 stimulation in the presence of bafilomycin A1, assumed to block refilling of lysosomal Ca2+ stores, did not antagonize subsequent NAADP-induced Ca2+ signaling. In contrast to bafilomycin A1, emptying of the endoplasmic reticulum by thapsigargin almost completely prevented Ca2+ signaling induced by NAADP. In conclusion, in T-lymphocytes, no evidence for involvement of lysosomes in NAADP-mediated Ca2+ signaling was obtained. The sensitivity of NAADP-induced Ca2+ signaling toward thapsigargin, combined with our recent results identifying ryanodine receptors as the target calcium channel of NAADP (Dammermann, W., and Guse, A. H. (2005) J. Biol. Chem. 280, 21394–21399), rather suggest that the target calcium store of NAADP in T cells is the endoplasmic reticulum.

The search for the target organelle and the molecular receptor for NAADP has attracted much attention in the last years. Lee et al. (10) demonstrated different subcellular localizations of the NAADP-sensitive Ca 2ϩ pool versus the InsP 3 -and cADPR-sensitive Ca 2ϩ pools in stratified sea urchin eggs. This NAADP-sensitive Ca 2ϩ pool was identified as the reserve granule in sea urchin eggs, an organelle related to lysosomes (11). In addition, pharmacological and partial biochemical characterization suggests expression of a novel NAADP receptor/Ca 2ϩ channel in sea urchin eggs (1,(12)(13)(14)(15).
In a few higher eukaryotic cell types, evidence for the involvement of ryanodine receptors (RyR) in NAADP-mediated Ca 2ϩ signaling has been published. NAADP activated purified RyR from heart and skeletal muscle in lipid planar bilayers (16,17). In addition, NAADP-induced Ca 2ϩ release from the nuclear envelope of pancreatic acinar cells and from permeabilized pancreatic acinar cells was blocked by RyR antagonists (18,19).
In human Jurkat T-lymphocytes, we have demonstrated: (i) TCR-CD3-mediated formation of NAADP (9), (ii) Ca 2ϩ signaling induced by microinjection of NAADP in a concentrationdependent fashion (6), (iii) that NAADP-induced Ca 2ϩ signaling consists of both Ca 2ϩ release and Ca 2ϩ entry (6,20), and (iv) involvement of RyR in both local and global Ca 2ϩ signaling (20,21). Here we confirm that lysosomes can be functionally destroyed using either hypertonic swelling and physical destruction by glycyl-phenylalanine 2-naphthylamide (GPN) or inhibition of the lysosomal H ϩ -ATPase using bafilomycin A1 (11). Although GPN pretreatment resulted in loss of Ca 2ϩmobilizing activity of all three second messengers, NAADP, cADPR, and InsP 3 , there was no inhibitory effect of bafilomycin A1 on any of the Ca 2ϩ -mobilizing compounds, even when endogenous Ca 2ϩ stores were previously emptied by TCR-CD3 stimulation in the presence of bafilomycin A1. In contrast, emptying the endoplasmic reticular (ER) Ca 2ϩ store by thapsigargin almost completely prevented Ca 2ϩ signaling induced by NAADP. Taken together, our data indicate a major role for the ER, but not for lysosomes, in NAADP-mediated Ca 2ϩ signaling in T cells.
Ratiometric Ca 2ϩ Imaging-The cells were loaded with Fura-2/AM as described (22) and kept in the dark at room temperature until use. Thin glass coverslips (0.1 mm) were coated first with bovine serum albumin (5 mg/ml) and subsequently with poly-L-lysine (0.1 mg/ml). Silicon grease was used to seal small chambers consisting of a rubber O-ring on the glass coverslip. Then, 60 l of buffer A containing 140 mM NaCl, 5 mM KCl, 1 mM MgSO 4 , 1 mM CaCl 2 , 1 mM Na 2 HPO 4 , 5.5 mM glucose, and 20 mM HEPES (pH 7.4), and 40 l of cell suspension (2 ϫ 10 6 cells/ ml) suspended in the same buffer were added into the small chamber (6), and the coverslip was mounted on the stage of a fluorescence microscope (Leica DMIRE2). In some experiments, GdCl 3 (10 or 100 M) was added to buffer A to block capacitative Ca 2ϩ entry; in these experiments, buffer A without Na 2 HPO 4 was used to prevent precipitation of GdPO 4 .
Ratiometric Ca 2ϩ imaging was performed as described recently (23). We used an Improvision imaging system (Tübingen, Germany) build around the Leica microscope at 100-fold magnification. Illumination at 340 and 380 nm was carried out using a monochromator system (Polychrom IV, TILL Photonics, Gräfelfing, Germany). Images were taken with a gray-scale CCD camera (type C 4742-95-12ER; Hamamatsu, Enfield, United Kingdom) operated in 8-bit mode. The spatial resolution was 512 ϫ 640 pixels at 100fold magnification. Camera exposure times were 12 ms (at 340 nm) and 4 ms (at 380 nm). The acquisition rate was adjusted to ϳ14 ratios/ min. Raw data images were stored on a hard disk. Confocal Ca 2ϩ images were obtained by off-line deconvolution (no-neighbor algorithm) using the volume deconvolution module of the Openlab software as described recently for 3T3 fibroblasts (24). The deconvolved images were used to construct ratio images (340/380 nm) pixel by pixel. Finally, ratio values were converted into Ca 2ϩ concentrations by external calibration (23). Data processing was performed using Openlab software, version 3.0.8 or 1.7.8 (Improvision).
Microinjection-Microinjections were carried out as described (25). We used an Eppendorf system (transjector type 5246, micromanipulator type 5171, Eppendorf-Netheler-Hinz, Hamburg, Germany) with Femtotips I and II as pipettes. NAADP, InsP 3 , or cADPR were diluted to their final concentration in intracellular buffer (20 mM HEPES, 110 mM KCl, 10 mM NaCl, pH 7.2) and filtered (0.2 m) before use. Injections were made using the semiautomatic mode of the system with following instrumental settings: injection pressure 60 -90 hec-  Staining of Lysosomes-Lysosomes were labeled by incubation of cells with 75 nM LysoTracker Red DND-99 for 30 min at room temperature. Cells were added into small chambers (see above), and fluorescence was determined at excitation and emission wavelengths of 575 and 590 nm using the Improvision imaging system as described above.
Then, cells were treated either with 50 M GPN, 10 M cathepsin inhibitor 1, plus 50 M GPN or with Me 2 SO (0.1% v/v or 0.2% v/v) for 15 min at room temperature. To analyze the effect of bafilomycin A1 on LysoTracker staining, untreated cells were incubated with different concentrations of bafilomycin A1 or 0.1% v/v Me 2 SO as control for 15 min at room temperature, and afterward, lysosomes were labeled with Lyso-Tracker Red DND-99 as described above (26).

RESULTS
To analyze whether in Jurkat T-lymphocytes lysosomes represent a Ca 2ϩ store sensitive to NAADP, we treated intact cells with GPN, a substrate of cathepsin C. Via osmotic lysis, the metabolic products of GPN selectively disrupt lysosomal membranes (11,27). The resulting significant loss of lysosome staining by LysoTracker Red was confirmed in Jurkat T cells (Fig.  1A). GPN increased the intracellular Ca 2ϩ concentration by lysing the lysosomes (Fig. 1, B-D, right panels). As described for sea urchin eggs (11), GPN also abolished NAADPmediated Ca 2ϩ release in T cells (Fig. 1B, right panel). However, Ca 2ϩ release by InsP 3 or cADPR was eliminated, too ( Fig. 1, C and D,  right panels). Therefore, we used cathepsin inhibitor 1, which is known to reduce the damaging effect of GPN by inhibition of cathepsin B, L, and S and also papain released from the lysosomes. Since cathepsin inhibitor 1 does not inhibit cathepsin C, the disruption of lysosomes by GPN should be unaffected (19). This was indeed confirmed since the intensity of lysosomes labeled with LysoTracker Red after incubation with cathepsin inhibitor 1 (10 M) plus GPN (50 M) was reduced similarly as compared with the control (Fig. 1A). However, treatment with cathepsin inhibitor 1 and GPN not only blocked NAADP-mediated Ca 2ϩ release, but also, Ca 2ϩ release by InsP 3 or cADPR was abolished in Jurkat T-lymphocytes (data not shown) as shown above for GPN alone. Thus, although GPN appears to be a powerful tool to destroy lysosomes, the release of lysosomal proteases or other hydrolases seems to interfere with all Ca 2ϩ release systems present in T cells. Since cADPR and InsP 3 are supposed to target the ER, the data obtained with GPN do not allow a clear-cut decision as to whether NAADP solely acts on lysosomes.
As a second approach, we investigated the effect of bafilomycin A1 on global Ca 2ϩ signals induced by NAADP. Bafilomycin A1 is an inhibitor of vacuolar H ϩ -pumps powered by ATP (28), prevents organelle acidification, and blocks the Ca 2ϩ reuptake into lysosomes via Ca 2ϩ /H ϩ -exchanger since this transport protein requires a proton gradient (11). As LysoTracker Red DND-99 is a weak base and accumulates in acidic compart- ments, preincubation with bafilomycin A1 concentration-dependently reduced or abolished labeling of lysosomes with LysoTracker Red DND-99 (26), a result that was confirmed in Jurkat T cells ( Fig. 2A). For our experiments, we used different lots of bafilomycin A1 purchased from different suppliers. Identity, purity, and content were confirmed by HPLC (Fig. 2B). Then, cells were preincubated with increasing concentrations of bafilomycin A1 or vehicle (Me 2 SO 0.1% v/v) and microinjected with 100 nM NAADP (Fig. 2, C-G). No reduction of NAADP-mediated Ca 2ϩ signaling was observed after preincubation with bafilomycin A1 at concentrations up to 1 M (Fig. 2,  C-G). In contrast, a slight increase of the NAADP-mediated Ca 2ϩ peak at 250 nM bafilomycin A1 was detected as compared with the control (Fig. 2, E and G). Bafilomycin A1 at a concentration that almost completely abolished the lysosome H ϩ gradient did not affect InsP 3 -or cADPR-mediated Ca 2ϩ signaling (Fig. 3). Since InsP 3 and cADPR have been shown to release Ca 2ϩ from the ER, this control experiment suggests that bafilomycin A1, in contrast to GPN (Fig. 1), in principle can be used to analyze the role of lysosomes in T cell Ca 2ϩ signaling.
Lysosomes are supposed to be very tight Ca 2ϩ stores, with almost no passive leak of Ca 2ϩ ions into the cytosol (11). In Jurkat T-lymphocytes, TCR-CD3 stimulation is known to increase intracellular NAADP levels (9). If lysosomes are the organelles targeted by NAADP, stimulation via the TCR-CD3 complex in the presence of bafilomycin A1 should lead to Ca 2ϩ depletion of lysosomal stores. As bafilomycin A1 inhibits the Ca 2ϩ reuptake in lysosomes, a second increase of the NAADP level via microinjection should show a reduced or abolished NAADP-mediated Ca 2ϩ release. However, Ca 2ϩ signaling by NAADP after preincubation with bafilomycin A1 and subsequent stimulation via TCR-CD3 was not different from untreated controls, indicating no major role for bafilomycin A1-sensitive compartments in NAADP-mediated Ca 2ϩ signaling (Fig. 4).
Since no evidence for an involvement for lysosomes in NAADP-induced Ca 2ϩ signaling was obtained, the alternative hypothesis (the ER as NAADP-sensitive Ca 2ϩ stores) was taken into account. Although the specific inhibitor of sarcoplasmic/endoplasmic reticulum Ca 2ϩ ATPases (SERCA), thapsigargin (29), is an effective and specific tool to deplete ER/SRtype Ca 2ϩ stores, in intact, electrically non-excitable cells, there is often the problem of capacitative Ca 2ϩ entry being switched on immediately (30,31). Thus, to avoid any effects of capacitative Ca 2ϩ entry, all further experiments were conducted in the presence of GdCl 3 in the extracellular buffer. Under these conditions, microinjection of 100 nM NAADP resulted in transient elevations of [Ca 2ϩ ] i with most of the cells returning to basal values within a few hundred seconds (Fig. 5A). Also, the mean amplitude of the initial peak induced by NAADP was signifi- cantly decreased in the absence of Ca 2ϩ entry (compare Fig. 5A, right panel, with Fig. 2C, right panel). Thapsigargin also induced a transient elevation of [Ca 2ϩ ] i , indicating rapid depletion of the ER and no concomitant activation of capacitative Ca 2ϩ entry (Fig. 5B). A second activation of thapsigargin confirmed that the ER Ca 2ϩ stores have indeed been depleted (Fig.  5B). Then, ER Ca 2ϩ stores were depleted by thapsigargin, and NAADP was subsequently microinjected (Fig. 5C); almost none of the cells responded to NAADP (7 out of 9 cells). In a complimentary approach, the intracellular store sensitive to NAADP was partially depleted by microinjection of NAADP in the presence of GdCl 3 ; then, the filling states of both the ER and the lysosomes were analyzed by the addition of either thapsigargin or GPN (Fig. 6). When compared with controls (injections of intracellular buffer instead of NAADP), the filling state of the ER was decreased slightly (Fig. 6, A, B, and E), whereas the content of the lysosomal store was unaffected (Fig. 6, C-E), indicating that NAADP releases Ca 2ϩ from the ER rather than from the lysosomes.

DISCUSSION
Although not confirmed in a great variety of cells, NAADP has been proposed to be an important second messenger in glucose-stimulated pancreatic ␤-cells (7), in cholecystokinin-mediated Ca 2ϩ signaling in pancreatic acinar cells (8), and in Ca 2ϩ signaling in Jurkat T cells stimulated via the TCR-CD3 complex (9). Thus, at least in these selected cell types, the NAADP/Ca 2ϩ signaling system may play an important role for cellular function. Therefore, important modules of this signaling system are the target organelle and target Ca 2ϩ channel of NAADP. There has been much debate as to whether NAADP acts on a novel receptor/Ca 2ϩ channel or whether RyRs respond to NAADP directly or indirectly, e.g. via a separate binding protein (reviewed in Ref. 32). For T cells, we recently demonstrated that functional RyR are required for both local and global Ca 2ϩ signals induced by NAADP (20,21). Although RyR are assumed to be localized at the ER (or SR), suggesting that the NAADP-sensitive Ca 2ϩ pool is located there, convincing data from other cell systems, including sea urchin eggs (10,11), pancreatic acinar-and ␤-cells (33), neurosecretory PC12 cells (34), and guinea pig cardiac myocytes (35), prompted us to investigate involvement of a lysosomal Ca 2ϩ store in NAADP-mediated Ca 2ϩ signaling in T cells. The first approach, lysis of lysosomes using the cathepsin substrate GPN, has successfully been used in sea urchin eggs (11) and pancreatic acinar cells (33); in sea urchin eggs, GPN-mediated bursting of lysosomes selectively abolished Ca 2ϩ signaling by NAADP but not by InsP 3 or cADPR (11). This result could not be reproduced in T cells; the Ca 2ϩmobilizing effects of all three messengers, NAADP, InsP 3 , or cADPR, were massively reduced. Since bursting of lysosomes is assumed to release proteases and other hydrolases into the cytosol, it is well imaginable that proteolytic attack of the InsP 3 receptor or the RyR is the reason for the lack of effect of InsP 3 or cADPR. Thus, although the method may work well in other cell types, such as sea urchin eggs, it appears to be not suitable for T cells.
The second approach, bafilomycin A1, yielded an even better loss of staining of the lysosomes in intact Jurkat T cells (compare Fig. 1A with Fig. 2A), indicating that the H ϩ gradient across the lysosomal membrane was effectively destroyed. Lack of lysosomal staining was almost completed at 25 nM bafilomycin A1. However, NAADP-mediated Ca 2ϩ signaling was not affected by up to 1000 nM bafilomycin A1 in T cells. In contrast, in sea urchin eggs, bafilomycin A1 inhibited Ca 2ϩ release induced by NAADP in experiments where NAADP was liber- ated from caged NAADP subsequently two times. Interestingly, the Ca 2ϩ release induced by the first uncaging of NAADP was not influenced, whereas the second one was largely reduced (11). Churchill et al. (11) suggested, "that the lysosomal Ca 2ϩ store is replete and non-leaky in resting cells, and that bafilomycin sensitivity is only revealed once the store is mobilized, making its inhibition use-dependent." However, in higher eukaryotic cell types, e.g. in pancreatic acinar cells, bafilomycin A1 inhibited cholecystokinin-induced Ca 2ϩ spiking without prior emptying of stores (33). Similarly, glucose-induced Ca 2ϩ signaling in pancreatic ␤-cells (MIN6 cells) was abrogated by bafilomycin A1 (33). Thus, in higher eukaryotic cells, the NAADP-sensitive Ca 2ϩ stores obviously are not very "non-leaky." Consequently, the lack of inhibitory effect of bafilomycin A1 on NAADP-induced Ca 2ϩ signals in T cells (Fig. 2) strongly indicates that the NAADP-sensitive Ca 2ϩ store of T cells is not the lysosomal compartment. To be on the safe side, and assuming that T cells are more similar to sea urchin eggs than to pancreatic cells concerning a nonleaky NAADP-sensitive Ca 2ϩ store, a further crucial experiment was performed (Fig. 4); T cells were first stimulated via the TCR-CD3 complex in the presence of bafilomycin A1, and then NAADP was microinjected. TCR-CD3 stimulation elevates NAADP in a biphasic manner (9), and thus, Ca 2ϩ present in NAADP-sensitive stores was released. Due to the presence of bafilomycin A1, any reuptake of Ca 2ϩ into acidic stores was inhibited. Nevertheless, there was no inhibition of subsequent Ca 2ϩ signaling induced by microinjection of NAADP, ruling out the possibility that T cells possess a very tight (ϭ non-leaky) NAADP-sensitive lysosomal Ca 2ϩ store.
In contrast to the idea that pancreatic acinar cells exclusively possess a lysosomal Ca 2ϩ store sensitive to NAADP, Gerasimenko et al. obtained evidence for NAADP-mediated Ca 2ϩ release from the nuclear envelope (18) and partially from both the ER and acidic organelles (19). In a similar approach in which the activation of capacitative Ca 2ϩ entry was blocked by GdCl 3 , we emptied ER Ca 2ϩ stores by the addition of thapsigargin. Since subsequent microinjection of NAADP did not induce Ca 2ϩ release in most of the cells investigated, it is very likely that NAADP indeed acts on an endoplasmic reticular Ca 2ϩ store in T cells. In contrast, in the complimentary experiment (prior microinjection of NAADP followed by the addition of thapsigargin), the effect of thapsigargin was reduced as compared with prior microinjection of intracellular buffer (Fig. 6). These data indicate that indeed the filling state of the ER was decreased by approximately 35% by prior injection of NAADP. A similar experiment carried out using GPN instead of thapsigargin to analyze the filling state of the lysosomes resulted in no difference regardless of whether the cells had been microinjected by NAADP or intracellular buffer. Taken together, our results demonstrate a major role for the ER, but not for lysosomes, in NAADP-mediated Ca 2ϩ signaling in T cells.