Nicotinic Acid Adenine Dinucleotide Phosphate (NAADP)-mediated Calcium Signaling*

Nicotinic acid adenine dinucleotide phosphate (NAADP)2 is a metabolite of NADP that was first identified as the most potent Ca2 stores mobilizing molecule in sea urchin egg homogenates more than a decade ago (1, 2). It has since been shown to be effective in a wide variety of cells, from plant to animal, including human (reviewed in Refs. 3–5). Its mechanism of action is distinct from those of cyclic ADP-ribose (cADPR) and inositol trisphosphate (IP3), and the stores it targets are separate as well (reviewed in Ref. 6). Recent evidence establishes that NAADP fulfills the criteria of being a second messenger for mobilizing Ca2 stores (reviewed in Ref. 7). This article reviews the structure, functions, and the enzymatic synthesis of this new addition to the list of Ca2 messengers.


Structure and Ca 2؉ Release Activity of NAADP
NAADP is a simple derivative of NADP, with the only modification being the conversion of the amide of the nicotinamide group to a carboxyl group (Fig.  1A). The two compounds differ by only one mass unit and have identical proton NMR and UV spectra (2). The slight structural change, nonetheless, makes NAADP the most potent Ca 2ϩ messenger ever described (2,8). A series of analogs of NAADP has been synthesized to probe the structural determinants of its Ca 2ϩ releasing activity. In addition to the carboxyl group, the amino group at the adenine ring and the 2Ј-phosphate are both critical to its biological activity (9) (circled in Fig. 1A). Attaching a caged group to the 2Ј-phosphate produces an inactive analog (10) that can regenerate NAADP and elicit large Ca 2ϩ changes upon photolysis in many cells (e.g. sea urchin eggs, Fig. 1B), providing convincing evidence that the NAADP-sensitive mechanism is present and functional in live cells (8,11). Removal of the 2Ј-phosphate by phosphatases or hydrolysis by nucleotide phosphodiesterase also inactivates its biological activity (10). The general presence of these enzymes in cells assures rapid removal of NAADP after its signaling function is completed.
That NAADP is effective in mobilizing Ca 2ϩ stores was first demonstrated in sea urchin egg homogenates, a simple and stable system for readily accessing the Ca 2ϩ stores of the cells (1,2). Results indicated that the extent of the Ca 2ϩ released by NAADP was similar to those induced by cADPR or IP 3 , the two other mobilizing mechanisms present also in the homogenates, but the NAADP mechanism could be distinguished pharmacologically by its insensitivity to specific antagonists that selectively blocked the other two (1, 2). Similar Ca 2ϩ release activity was seen in microsomes isolated from a variety of mammalian (12)(13)(14)(15) and plant cells (16). A novel property of the NAADP mechanism is its self-desensitization. Treatment with high concentrations of NAADP, or prolonged incubation with subthreshold concentrations, rendered the homogenates irresponsive to NAADP (11,17). That this property may provide a physiological mechanism for spatiotemporal Ca 2ϩ memory in cells has been demonstrated (18). In lieu of a specific antagonist, this desensitization property has now been widely used as a diagnostic tool for ascertaining if an observed Ca 2ϩ change is mediated by NAADP (8,11,19,20).
Desensitization is generally associated with receptor-mediated processes. The structure-function studies of NAADP, showing strong dependence of the Ca 2ϩ release activity on minor changes in the structure of the molecule, also are consistent with its action being mediated by a receptor ((9) and Fig. 1A). That this is the case was first shown directly by specific binding of NAADP to sea urchin egg microsomes. The measured affinity constant of the binding was consistent with the half-maximal concentration of its Ca 2ϩ release activity (11). The binding also exhibited similar desensitization behavior, which was found to require the presence of physiological concentrations of K ϩ (11,21). Ongoing effort shows that the NAADP receptor in the eggs appears to be a protein of 470 kDa, but the exact molecular identity of this novel receptor remains to be elucidated (22).
The NAADP receptor is widely distributed and is present in mammalian tissues, such as brain (23) and heart (13) as well. Particularly intriguing is its regional distribution seen in the brain as revealed by autoradiography, which shows concentration of the receptor in the thalamus, hippocampus, midbrain, and the medullar region (23).

Ca 2؉ Stores Targeted by NAADP
Density gradient fractionation of sea urchin egg homogenates first showed that the NAADP-sensitive stores had a broader distribution than the endoplasmic reticulum (ER) stores sensitive to cADPR and IP 3 and were particularly concentrated in the denser fractions, suggesting separate stores were targeted by NAADP (2,24). Stratifying cellular organelles by centrifugation in live sea urchin eggs confirmed that the NAADP-sensitive stores (Fig. 1C) were separable and could be moved to a pole distal to the nucleus, where the ER and the cADPR-and IP 3 -sensitive stores (Fig. 1, D and E) were located (25).
Marker enzyme analyses identified the NAADP-sensitive fractions in the sea urchin egg homogenates as lysosome-like organelles that possessed a thapsigargin-insensitive Ca 2ϩ transport system distinct from that of the ER (26,27). Instead of utilizing a Ca 2ϩ -ATPase, this system employed a combination of a bafilomycin-sensitive V-H ϩ -ATPase and a Ca 2ϩ /H ϩ -exchanger for the sequestration of Ca 2ϩ via a vesicular proton gradient (27). Thus treatments with bafilomycin as well as with ionophores that discharged the vesicular proton gradient depleted the stores and selectively inhibited the NAADP-dependent Ca 2ϩ release. Likewise, lysosomotrophic agents that disrupted lysosomes, such as glycylphenylalanine 2-naphthylamide (GPN), could induce Ca 2ϩ increase in sea urchin eggs and selectively eliminated the NAADP-sensitive stores without affecting the ER Ca 2ϩ stores that were sensitive to cADPR and IP 3 (27). Similar selective effects of lysosomotrophic agents on the NAADPsensitive Ca 2ϩ mobilization were seen also in mammalian cells, such as pancreatic acinar and ␤-cells (MIN6) (28) as well as in smooth muscles (29).
The NAADP mechanism, however, is not exclusively associated with lysosomes. In MIN6 cells, store-targeting expression of the Ca 2ϩ -sensor protein, aequorin, identified insulin containing secretory granules as NAADP-sensitive Ca 2ϩ stores as well (30). These granules, however, were also acidic organelles like lysosomes. Fractionation studies in cauliflower homogenates showed colocalization of the NAADP-sensitive stores with ER markers and not with the bafilomycin-sensitive ATPase, a vacuolar marker (16). The nuclear envelope had also been found to possess the NAADP mechanism, which is a functioning Ca 2ϩ store possessing both Ca 2ϩ release and thapsigargin-sensitive sequestration systems (31).
Additionally, NAADP has been observed to be capable of stimulating Ca 2ϩ influx in both invertebrate eggs as well as in mammalian T-cells (32)(33)(34). Whether the influx is an indirect consequence of mobilization of the Ca 2ϩ stores that are closely associated with the plasma membrane, or whether it is due to direct activation of membrane Ca 2ϩ channels by NAADP, remain to be settled. It thus appears that the NAADP mechanism is quite widely distributed among cellular organelles.

Messenger Functions of NAADP
That NAADP is a naturally occurring molecule in cells was first shown in sea urchin sperm (35). The endogenous concentration measured was unexpectedly high in this terminally differentiated cell that lacks most of the normal organelles. This has led to the proposal that the sperm may serve as vehicles for delivery of NAADP to the egg at fertilization. Consistently, contact of the sperm with the jelly components surrounding the eggs greatly increased the endogenous NAADP levels (33). Measurements of the NAADP level in eggs immediately after fertilization showed large increases that appeared to correlate with the flash of Ca 2ϩ increase in the egg cortex seen immediately after sperm-egg fusion. Photolysis release of only NAADP, but not Ca 2ϩ , cADPR, or IP 3 , from their respective caged analogs preloaded into the eggs induced cortical flashes similar to that observed at fertilization. Prior desensitization of the NAADP stores blocked the fertilization-induced cortical flash, indicating it was specifically mediated by the mobilization of the NAADP stores, perhaps those localized in the egg cortex (33).
Photolysis relase of NAADP in the eggs not only produced a large Ca 2ϩ increase (cf. Fig. 1B) but could also induce prolonged Ca 2ϩ oscillation, which neither IP 3 nor cADPR could do (10,11,36). Unexpectedly, blockage of the IP 3 and the cADPR mechanisms inhibited the NAADP-induced oscillation. Also, measurements showed that the Ca 2ϩ content of the ER stores increased after mobilization of the NAADP stores (36). These results indicate a two-pool mechanism is involved, whereby Ca 2ϩ released from the NAADP stores is sequestered by the ER stores, overloading the latter and triggering spontaneous release via the cADPR and IP 3 mechanisms (3,10,11,36). It is thus clear that Ca 2ϩ stores can be separate and have distinct mechanisms for mobilization, and yet they can functionally interact to produce Ca 2ϩ oscillation.
Ascidian oocytes also possess all three Ca 2ϩ mobilizing mechanisms. Infusion of NAADP produced rapid modulation of membrane currents similar to that observed at fertilization (37). On the other hand, cADPR could effectively induce vesicular fusion, which NAADP could not. Neither cADPR nor NAADP could elevate cytoplasmic Ca 2ϩ , which was mainly activated by IP 3 (37). These results are consistent with the NAADP and the cADPR stores being localized in the cell cortex, allowing them to effectively modulate membrane events, while the IP 3 -sensitive ER stores are mainly distributed in the cytoplasm. Selective localization of Ca 2ϩ stores can thus allow different messenger pathways to subserve distinct cellular functions.
The first mammalian cells shown to be responsive to NAADP are the pancreatic acinar cells (8). Cholecystokinin (CCK) is an important secretogogue in these cells, which also elicits local Ca 2ϩ spiking and global Ca 2ϩ waves (38). Physiological concentrations of CCK stimulated immediate cellular production of NAADP that peaked in seconds (39) and triggered localized Ca 2ϩ spiking in the secretory pole of the cells (28,39). Infusion of NAADP into the cells activated similar spiking. Prior desensitization of the NAADP receptor and lysosomtrophic agents, bafilomycin and GPN, all blocked the CCK signaling (8,28). These results establish that NAADP satisfies all the criteria of being a second messenger for mediating the signaling function of CCK.
Additionally, CCK also stimulated the cellular production of cADPR, although with a time course lagging behind the NAADP elevation (39). Infusion of cADPR into the cells also triggered Ca 2ϩ spiking but could not be blocked by prior desensitization of the NAADP receptor (8). These results are consistent with NAADP being the trigger, whose signal is then amplified by mobilizing the cADPR stores, perhaps through Ca 2ϩ -induced Ca 2ϩ release (8,38). Indeed, 8-amino-cADPR, a specific antagonist of cADPR (40), blocked Ca 2ϩ spiking induced by CCK, indicating the amplification by the cADPR stores was critical to the observed spiking (8,19,38). That NAADP was the messenger specifically for the CCK signaling was shown by the fact that acetylcholine, another secretogogue in the acinar cells, even at supramaximal concentrations, could only stimulate the production of cellular cADPR but not NAADP. The former reached peak values in about a minute, similar to the CCK-induced cADPR production (39). These results provide a clear example of the specificity of NAADP and illustrate the selective coupling of an agonist (CCK) to a second messenger pathway. This is also the case in arterial smooth muscles, where the NAADP stores were identified as lysosomes that were closely associated with the sarcoplasmic reticulum (29). Edothelin-1, a vasoconstrictor hormone, but not prostaglandin-F 2␣ , induced NAADP production. Also, bafilomycin selectively inhibited the endothelin-induced Ca 2ϩ elevation but not that induced by prostaglandin. On the other hand, thapsigargin blocked the global Ca 2ϩ wave induced by endothelin but not the Ca 2ϩ spiking that appeared to originate from the NAADP-sensitive lysosomal stores (29). These results again indicate NAADP is functioning as a trigger, which signal is subsequently amplified by mobilization of the ER stores.
A variety of other cellular functions have also been shown to involve NAADP. Pancreatic ␤-cells (MIN6) are responsive to glucose, which also induces complex Ca 2ϩ changes. The cellular NAADP level was found to elevate after glucose (41). Bafilomycin, but not thapsigargin, inhibited the glucose-induced Ca 2ϩ changes (28). Binding studies indicated the presence of the NAADP receptor in these cells (41). Photolysis release of low concentrations of NAADP elicited similar Ca 2ϩ changes in the cells, which could be inhibited by GPN (28). Desensitizing the NAADP receptor with high concentrations of NAADP inhibited the glucose-induced response, indicating the involvement of NAADP in mediating the response (41). Similarly, in human panceatic islets, insulin induced a complex Ca 2ϩ response, which could be mimicked by microinjection of low concentrations of NAADP, while high concentrations desensitized the receptor and blocked the insulin-induced Ca 2ϩ changes, again suggesting the involvement of NAADP in the process (42).
NAADP has also been found to mediate neuronal functions. In primary cultures of neurons, delivery of NAADP by liposomal fusion elicited Ca 2ϩ changes that could be blocked by bafilomycin but not by thapsigargin. Potentiation of neurite outgrowth was observed in an NAADP concentration-dependent manner after the treatment (43). Similarly, in frog neuromuscular junction, liposomal delivery of NAADP increased transmitter release in a concentration-dependent manner (44). In the buccal ganglion of Aplysia, presynaptic injections of NAADP induced Ca 2ϩ changes and transiently increased the inhibitory postsynaptic current, indicating increase in neurotransmitter release (45). These results show that not only are the NAADP-responsive cells widespread, so are the functions it mediates.

Enzymatic Synthesis of NAADP
Two enzymes have so far been shown to be capable of synthesizing NAADP (46). The first one is ADP-ribosyl cyclase (cyclase), a soluble protein of about 30 kDa purified from Aplysia ovotestis (47). It has since been crystallized and its structure solved (48) (Fig. 2A). The second enzyme is its mammalian homolog, CD38, a protein first thought to be a lymphocyte antigen but has since been found to be ubiquitously present intracellularly as well as on the surface of a variety of cells (Ref. 49 and reviewed in Ref. 50). In contrast to the cyclase, CD38 has a single transmembrane segment near its N terminus (49) (Fig. 2C). The crystal structure of its extramembrane domain has recently been solved (51). Both proteins and another homolog, CD157, were first identified as enzymes that can cyclize NAD in a head to tail fashion to produce cADPR with the release of nicotinamide (47,52,53). It is now known that both CD38 and the cyclase are multifunctional enzymes capable of catalyzing a base-exchange reaction, exchanging the nicotinamide group of NADP with nicotinic acid and producing NAADP. The exchange reaction is the dominant reaction at acidic pH (46) (Fig. 2C). Additionally, CD38 is the only known enzyme that can hydrolyze cADPR to ADP-ribose (52) (Fig. 2C). Whether CD157 can catalyze either the exchange or the hydrolysis reactions has not been determined. Despite only 25-30% sequence identity among the three proteins, they are  (11). In contrast, whole cell uniform photolysis of sea urchin eggs whose organelles had been stratified by centrifugation and loaded with either caged NAADP (C), cADPR (D), or IP 3 (E) elicited highly regionalized Ca 2ϩ changes in a pole either distal (NAADP) or proximal (cADPR and IP 3 ) to the nucleus (N, arrowheads), indicating that separable Ca 2ϩ stores were activated (25).
structurally very similar, especially at the enzymatic active sites that are all located in pockets at the central cleft of the proteins (51,53,54) (Fig. 2, A and  C). The largest divergences are seen in the two termini and four other loops (Fig. 2B).
The extramembrane portion of CD38 is an L-shaped molecule containing 255 residues and has a central cleft dividing the two domains. The N-terminal domain is composed mostly of ␣-helices, while the C-domain contains fourstrand parallel ␤-sheets (51). The secondary structures are thus essentially identical to the cyclase (48) (Fig. 2A). Also the same are the five disulfide bonds, two in the C-domain and three in the N-domain. CD38, however, has one additional disulfide linkage near the central cleft (Cys 119 -Cys 201 ). A surface potential map shows that the N terminus of the extramembrane domain is mainly positive, facilitating its interaction with the negatively charged phospholipids in the membrane (Fig. 2C).
Site-directed mutagenesis has identified the critical residues at the active site pocket (Fig. 2C). Glu 226 is most likely the catalytic residue, since any modification results in loss of activity (55). Asp 155 is critical for the synthesis of NAADP from NADP and nicotinic acid via the base-exchange reaction and changing it to glutamate greatly enhances the base-exchange reaction (55). Similarly, Glu 146 appears to control the cyclization reaction, since changing it to alanine, for example, greatly increases the production of cADPR from NAD (56). The two tryptophans, Trp 125 and Trp 189 , are likely to be responsible for interacting with the adenine and the nicotinamide groups of NAD, molding it into a folded conformation such that the two ends can be coupled to produce cADPR. The Lys 129 at the edge of the pocket is important in controlling the cADPR hydrolysis reaction (57). The surface potential at the active site is thus mainly negative because of the concentration of the critical acidic residues (Fig.  2C). The finding that CD38 can catalyze the synthesis of two structurally and functionally distinct Ca 2ϩ messengers is truly remarkable and suggests its central role in Ca 2ϩ signaling.

Perspectives
From the discovery of NAADP as a chemical derivative of NADP to its ascendancy to the rank of second messengers, the past decade has seen much progress in the understanding of the Ca 2ϩ signaling functions and the enzymatic synthesis of NAADP. Many important issues, however, remain to be resolved. Principal among them is how the external stimulus is coupled to cellular synthesis of NAADP. Intracellular CD38 presumably is responsible, but neither its orientation in internal membranes nor the regulation of its activity is known. Another important issue is the identity of the NAADP receptor. Whether it is a hitherto uncharacterized Ca 2ϩ -releasing channel or whether it is related to the ryanodine receptor as has been suggested (6,58,59) remain to be determined. Nevertheless, the progress on NAADP has strengthened considerably the current view that cells possess not only multiple Ca 2ϩ stores but also multiple messengers and signaling pathways for their mobilization. Stimuli can selectively activate certain pathways, and yet the separate Ca 2ϩ stores can functionally interact to produce integrated signals. This new found complexity and versatility of Ca 2ϩ store mobilization can surely fulfill its expected role as a major signaling mechanism.