Regulation of Intracellular Ca²⁺ Stores by Multiple Ca²⁺-Releasing Messengers

In all cells so far investigated directly, the concentration of Ca²⁺ inside the ER is very high (several hundred micromoles per liter) compared with the cytosol (∼0.1 μmol/l). The Ca²⁺ accumulation in the ER is due to a thapsigargin-sensitive Ca²⁺-activated ATPase (sarco-endoplasmic reticulum Ca²⁺-activated ATPase [SERCA] pump), and Ca²⁺ can be released from the store by IP₃ (10). The pancreatic β-cell is no exception. The resting Ca²⁺ concentration in the ER lumen is 100–500 μmol/l (11–13), which is in good agreement with values obtained in DRG neurons (60–270 μmol/l) (9) and pancreatic acinar cells (100–300 μmol/l) (14). As expected, the Ca²⁺ accumulation into the ER in the β-cell is ATP dependent and can be blocked by thapsigargin (11,13).

The resting Ca²⁺ concentration inside the ER depends on the equilibrium between SERCA-mediated Ca²⁺ uptake and the resting leak of Ca²⁺ from the store into the cytosol. The relationships between the ER Ca²⁺ concentration, the rate of SERCA-mediated Ca²⁺ uptake, and the leak of Ca²⁺ from the store are remarkably similar in pancreatic acinar cells (14) and DRG neurons (9). This result suggests that these relationships are most likely valid for many different cell types. The rate of Ca²⁺ uptake is maximal when the ER is maximally depleted, and during the refilling process, the rate of Ca²⁺ uptake declines as the Ca²⁺ concentration in the store increases. The passive leak of Ca²⁺ from the store increases moderately when the ER Ca²⁺ concentration increases but reaches a stable plateau at a certain ER Ca²⁺ level (Fig. 1). In both DRG neurons and pancreatic acinar cells, this equilibrium, representing the point at which the active SERCA-mediated Ca²⁺ uptake exactly balances the passive leak, occurs at an ER Ca²⁺ concentration of ∼200–300 μmol/l (Fig. 1).

IP₃ is an effective releaser of Ca²⁺ from the ER store in the pancreatic β-cells (11,15). Although IP₃ binding to the receptor is necessary for channel opening, there is also a requirement for Ca²⁺ to act as a co-agonist (16,17). In a certain Ca²⁺ concentration range, a rise in the cytosolic Ca²⁺ concentration in the presence of a submaximal IP₃ concentration causes a marked increase in the open state probability of the IP₃ receptor channel (16,17). At a higher cytosolic Ca²⁺ concentration, the relationship changes and a further rise in the Ca²⁺ concentration causes a decrease in the open-state probability of the channel (17). The positive feedback loop allows CICR also to be mediated by the IP₃ receptor, and both the positive and negative feedback loops play an important role in the generation of IP₃-mediated cytosolic Ca²⁺ spiking (18). The Ca²⁺ sensor region of the IP₃ receptor (subtype 1) has recently been identified as glutamate 2100. Substitution of this residue by aspartate results in a 10-fold decrease in the Ca²⁺ sensitivity, without affecting other properties of the channel, and abolishes the ability of the receptor to support Ca²⁺ oscillations (19).

In addition to SERCA pumps and IP₃ receptors, the ER in many cell types also contains ryanodine receptors (20). One of the classic tests for functional ryanodine receptors is to use caffeine stimulation, which sensitizes the ryanodine receptor to the Ca²⁺-induced opening of the channel (CICR), therefore emptying the ER Ca²⁺ store. Fresh direct evidence for this has recently been provided in DRG neurons (9). With regard to the β-cell, this point has been controversial. Tengholm et al. (11) concluded that there was no evidence for functional ryanodine receptors, but more recently, caffeine-induced Ca²⁺ release from the ER was demonstrated in two different insulin-secreting cell lines (INS-1 and MIN6) (13,21). Expression of ryanodine receptor protein in both MIN6 cells and primary mouse β-cells has been confirmed by specific binding of fluorescent ryanodine (13). The only ligand that is known to directly affect the open state probability of the ryanodine receptor is Ca²⁺, but in the next section, we will discuss the role of cyclic ADP-ribose (cADPR) as a modulator of this channel. It is important to realize that there is no evidence for cADPR binding directly to the ryanodine receptor (7).

There is a major controversy about the issue of ER continuity or discontinuity. The textbook view is that the ER “forms a continuous sheet enclosing a single internal space” (22). However, some authors have obtained data suggesting the existence of functional sub-compartmentalization, with different parts of the ER having different Ca²⁺ concentrations (23,24). In the pancreatic β-cells, it has also been suggested that there are regional differences in the ER Ca²⁺ concentration, which implies spatially separate sub-compartments (11). There are a number of difficult technical problems in relation to this issue (25). The best approach, by which the actual movements of Ca²⁺ in the lumen of the ER can be directly monitored, has so far only been applied to normal pancreatic acinar cells (26). This direct study shows very clearly that the ER is one continuous Ca²⁺ store that can function as a Ca²⁺ tunnel, allowing relatively rapid movement of Ca²⁺ from one end of the cell to the other (26). In this context, it is also interesting that estimates of the Ca²⁺ buffer capacity in pancreatic acinar cells have revealed a much higher value (∼1,500–2,000) in the ER than in the cytoplasm (∼20) (27). This means that the mobility of Ca²⁺ in the lumen of the ER is much higher than that in the cytoplasm, which is in agreement with the Ca²⁺ tunnel concept originally advanced by Mogami et al. (28). The textbook view of the ER as a lumenally continuous network would therefore, on the basis of the best available evidence, appear to be remarkably accurate. Nevertheless, it would of course be desirable for direct studies of the kind conducted in the pancreatic acinar cells (26) also to be carried out in other cell types, including the pancreatic β-cells. At this point in time, it cannot be excluded that different cell types have different ER organizations.

cADPR (a metabolite of NAD⁺) and nicotinic acid adenine dinucleotide phosphate (NAADP) (a metabolite of NADP⁺) were first shown to release Ca²⁺ from internal stores in sea urchin eggs (29,30). Single-channel current evidence from reconstituted ryanodine receptors of subtype 2 has shown that cADPR can markedly enhance the open-state probability of the channel (31). This result should not be taken as evidence for a direct action of cADPR on the ryanodine receptor because the ryanodine receptors studied were not purified (31). The enzyme responsible for both cADPR and NAADP synthesis is the ADPribosyl cyclase, which has been found to be widespread in mammalian tissues, including pancreatic β-cells (29,30,32–34). Three identified proteins possess ADPribosyl cyclase activity, the Aplysia cyclase, the leukocyte surface antigen CD38, and BST-1, found on bone marrow stromal cells (29,30,32). The enzyme has been located in the plasma membrane, cytosol, nucleus, and mitochondria (30,35–37).

In the pancreatic β-cell, work from Takasawa et al. (38) indicates that glucose stimulates the production of cADPR, although this result is in disagreement with data from Malaisse and colleagues (39,40). Takasawa et al. have shown that cADPR induces Ca²⁺ release in permeabilized β-cells and stimulates insulin secretion (41). In ADPribosyl cyclase CD38 knockout mice, insulin secretion in response to glucose stimulation was impaired (42), whereas overexpression of the ADPribosyl cyclase CD38 enhanced insulin secretion (33). The CD38 has been implicated as the molecular target for the autoimmune response in type 2 diabetic patients (43). However, cADPR did not release Ca²⁺ from the insulin-secreting cell line RINm5F or from islets in diabetic mice (ob/ob) (38,44,45). The RINm5F cells secrete little insulin, and these cells as well as β-cells from ob/ob mice only express low levels of ryanodine receptors and of ADPribosyl cyclase CD38 (38,41). Interestingly, β-cells from ob/ob mice were found to generate excessive firing of cytosolic Ca²⁺ transients, but this is attributed to activation of the phopholipase C signaling pathway and therefore most likely involves IP₃ rather than ryanodine receptors (46).

Ca²⁺ influx evoked in response to glucose metabolism has been proposed to trigger CICR from ryanodine-sensitive stores (47,48) and could be modulated by cADPR. However, there are experiments indicating that 8-amino-cADPR, a cADPR antagonist, does not block glucose stimulation of insulin secretion (45,49). On the whole, the role of cADPR, and indeed ryanodine receptors in insulin-secreting cells, has been, and continues to be, controversial (11,13,44,47).

In a recent study on the insulin-secreting cell line MIN6, it was shown that ryanodine receptors of subtype 2 are expressed and that caffeine acts to reduce the ER Ca²⁺ concentration. It was also shown that photolysis of caged cADPR increases the cytosolic Ca²⁺ concentration and that this effect is largely abolished by emptying the ER of Ca²⁺ by thapsigargin (13). Although these recent findings are interesting, they do not address the issue of whether cADPR is involved in mediating a physiologically relevant response.

It would be expected that glucose stimulation of the pancreatic β-cell, which elicits membrane depolarization and opening of voltage-gated Ca²⁺ channels in the surface cell membrane (2), would elicit a reduction in the ER Ca²⁺ concentration because of CICR (9). However, this is not necessarily so. Using aequorin-expressing INS-1 cells, Maechler et al. (50) showed that both K⁺ depolarization of the cell membrane and glucose stimulation do not elicit a decrease, but an increase, in the ER Ca²⁺ concentration. A similar conclusion was reached by Tengholm et al. (12). This group described the ER as a high-affinity sink for Ca²⁺. Whereas in the DRG neurons membrane depolarization elicits opposite (anti-parallel) changes in the cytosolic and ER Ca²⁺ concentrations (9), in the insulin-secreting cells, such a stimulation protocol causes parallel increases in both the cytosolic and ER Ca²⁺ concentrations (50). It is therefore clear that the CICR process is not always operational in the insulin-secreting cells. This finding raises an important question: How are the Ca²⁺ release channels in the ER controlled? Data recently obtained from pancreatic acinar cells indicate the existence of subtle combinatorial roles for Ca²⁺-releasing messengers in regulating the opening of Ca²⁺ release channels in the ER. These data may turn out also to be relevant for the control of Ca²⁺ release in other systems, including the pancreatic β-cells.

Experiments on pancreatic acinar cells provided the earliest evidence for IP₃ as a Ca²⁺-releasing messenger (51). More recently, it has become clear that the so-called new Ca²⁺-releasing agents cADPR and NAADP also play an important role in physiologically relevant Ca²⁺ signaling events in the pancreatic acinar cells (7,52). Some years ago, our group demonstrated that cADPR could evoke cytosolic Ca²⁺ spiking in the acinar cells, which could be blocked by ryanodine (53). The possible messenger role for cADPR was investigated using the specific cADPR antagonist 8-amino-cADPR (54,55). 8-Amino-cADPR specifically blocked the cADPR-evoked Ca²⁺ spikes but had virtually no effect on IP₃-evoked Ca²⁺ spiking. Indeed, the experiments performed with intracellular infusion of 8-amino-cADPR were very rewarding because the responses to physiological concentrations of the hormone cholecystokinin (CCK) (2–5 pmol/l) were blocked (54). However, at this point, a surprising discovery highlights the complexity and redundancy of Ca²⁺ signaling. We found that intracellular infusion of glucose, at concentrations from 300 μmol/l to 10 mmol/l, inhibited the cADPR-evoked Ca²⁺ spiking, whereas the IP₃-evoked Ca²⁺ spiking was potentiated (56). Even more surprisingly, the intracellular infusion of glucose, which inhibited the response to cADPR, had no effect on CCK-evoked Ca²⁺ spiking (56). The simplest hypothesis is that glucose compensates for its inhibitory effect on cADPR-elicited signaling by sensitizing the IP₃ response. In agreement with this view, 8-amino-cADPR did not block the CCK-evoked Ca²⁺ spiking in the presence of glucose, whereas heparin, an IP₃ receptor antagonist, did block the CCK response (56,57). This finding also indicates that both glucose and cADPR may share the role of amplifying the Ca²⁺ signaling response initiated by IP₃ and ryanodine receptors. These data could be related to the β-cell experiments in which glucose stimulation led to an increase of the endogenous level of cADPR (38), whereas 8-amino-cADPR failed to block the Ca²⁺ signal in response to glucose stimulation (45,49). These data do not rule out a role for cADPR as a messenger but highlight the possibility that additional messengers could be involved. One interesting candidate is NAADP because the ADPribosyl cyclase, which forms cADPR, can also form NAADP, a Ca²⁺-releasing messenger in sea urchin eggs (30).

NAADP is one of the most potent Ca²⁺-releasing messengers found so far (29,34,58). The NAADP receptor has not yet been purified, but several findings indicate the existence of a specific NAADP receptor. In sea urchin eggs, NAADP releases Ca²⁺ from a thapsigargin-insensitive Ca²⁺ store, which can be physically separated from those mobilized by IP₃ and cADPR (29,58,59). An important property of NAADP signaling is the self-desensitization mechanism induced by subthreshold NAADP concentrations in sea urchin eggs or high NAADP concentrations in mammalian cells (29,34,52,55,60). This self-desensitization property of the NAADP receptor was used to provide the first evidence for NAADP acting as an internal messenger for a hormone (34).

In mouse pancreatic acinar cells, NAADP releases Ca²⁺ and the NAADP receptor is of functional importance in the Ca²⁺ signaling response to CCK stimulation (34,60,61). Surprisingly, the NAADP action is biphasic. NAADP at a low concentration (50 nmol/l) evokes local Ca²⁺ spiking in the apical granular part of the cell (61), whereas at a high concentration (50–100 μmol/l), no Ca²⁺ signaling response is generated, suggesting complete desensitization (34,60). This self-desensitization property is specific because IP₃- and cADPR-evoked Ca²⁺ spikes were not altered by desensitizing concentrations of NAADP (34). A desensitizing concentration of NAADP blocked the response to a physiological CCK concentration but had no effect on the actions of just suprathreshold ACh or bombesin concentrations (60,62). Therefore, it has been proposed that NAADP is a second messenger for the CCK-mediated Ca²⁺ spiking and that it is physiologically relevant in pancreatic acinar cells (7,34,52,60). Finally, it has been shown that the Ca²⁺ signaling responses to CCK and NAADP share the same pharmacology because both are inhibited by the cADPR antagonist 8-amino-cADPR and the IP₃ receptor antagonist heparin. A model has been proposed in which a primary release of Ca²⁺ by NAADP is amplified by IP₃ and ryanodine receptors via a CICR process (34,52,60).

Low agonist concentrations generate mostly short-lasting local cytosolic Ca²⁺ spiking in the apical granular pole (63). This is sufficient for both exocytotic enzyme secretion and activation of Ca²⁺-dependent Cl⁻ channels in the apical membrane, which drives fluid secretion (6,25,64,65). At higher agonist concentrations, and particularly in the case of CCK stimulation, the Ca²⁺ signals become global and last much longer (63,64). This would appear to be important for growth responses (6) and apoptosis (66) as well as for pathological situations because acute pancreatitis is induced by sustained global Ca²⁺ rises (67).

The local Ca²⁺ spikes in the secretory pole of the cell are due to concerted activity of IP₃ and ryanodine receptors (common oscillator units), irrespective of the triggering agonist (60). ACh activation of these common oscillator units is triggered via IP₃ receptors, whereas the CCK responses are triggered via a different, but convergent, pathway dependent on NAADP and cADPR receptors (60). We have recently investigated how these local Ca²⁺ signals can become globalized.

The IP₃ receptors are clustered in the apical granular pole, but there is some evidence for a low density of IP₃ receptors in the basal part (64,68–70). The ryanodine receptors are present in all the regions of the cell, although with a higher density in the basolateral part (71,72). Because the NAADP receptor has not yet been characterized, no information is available about its subcellular distribution. Interestingly, low just suprathreshold concentrations of IP₃, cADPR, or NAADP elicit Ca²⁺ spiking specifically in the apical pole (53,61,64). However, at higher concentrations, they may produce global Ca²⁺ waves (34,53,64). The Ca²⁺ release receptors (channels) located in the basolateral part of the cell would appear to be relatively insensitive to their respective messengers and need to be sensitized to allow Ca²⁺ release (Fig. 2). In this respect, possible interactions between different messenger pathways could be of considerable interest.

In pancreatic acinar cells, local Ca²⁺ spikes evoked by a low concentration of ACh can be transformed into a global sustained Ca²⁺ response by cADPR or NAADP but not by IP₃. On the other hand, the response to a low concentration of CCK can be strongly potentiated by IP₃, whereas cADPR and NAADP have little effect (61). Although low concentrations of IP₃, cADPR, and NAADP each evoke local Ca²⁺ signals, they may, when acting together, be involved in global Ca²⁺ signal production. We have therefore tested the effects of different messenger combinations. We found that NAADP locally potentiates the IP₃-evoked Ca²⁺ spikes in the secretory pole of the cell. Although both NAADP and cADPR evoked local Ca²⁺ spiking in the secretory pole, when infused separately, a stronger global potentiating interaction was found when they were infused together. In this case, the Ca²⁺ signals invaded the basolateral part of the cell. Finally, NAADP strongly amplified the local Ca²⁺ release evoked by a cADPR/IP₃ mixture, eliciting a vigorous global Ca²⁺ response. These recent results (61) demonstrate that different combinations of Ca²⁺-releasing messengers can shape the spatio-temporal pattern of cytosolic Ca²⁺ signals. In these experiments, the Ca²⁺-dependent Cl⁻ current across the apical cell membrane, which is an index of the acinar fluid secretion (6,65), was also recorded. The acinar fluid secretion was markedly enhanced by infusion of a triple mixture consisting of NAADP, cADPR, and IP₃ (61).

From this work, NAADP emerges as a key messenger in the globalization of Ca²⁺ signals and therefore also for the control of secretion (61). NAADP itself has a modest effect because it only releases Ca²⁺ in the apical part of the cell. However, its unique ability to interact with ryanodine and IP₃ receptor activity allows a substantial increase in the medium excitability to further activation by either cADPR or IP₃ (Fig. 2).

The difference between the sustained global Ca²⁺ release obtained, for example, by combination of just suprathreshold concentrations of IP₃, cADPR, and NAADP and the local Ca²⁺ spiking in the granular pole seen in response to stimulation with just one of the messengers is very substantial (Fig. 2). Repetitive short-lasting local Ca²⁺ spikes are not associated with any measurable decrease in the ER Ca²⁺ concentration, whereas global sustained Ca²⁺ signals are associated with a complete emptying of the ER Ca²⁺ store (26). Addition to the intracellular solution of, for example, 50 nmol/l NAADP to a mixture of 10 μmol/l IP₃ and 10 μmol/l cADPR causes an enormous increase in the rate of Ca²⁺ release from the ER (61), resulting in fast depletion of the store. It is this radical change in the responsiveness of the ER (Fig. 2) that is so important and also, we believe, potentially relevant to considerations of Ca²⁺ homeostasis in the pancreatic β-cells.

In the pancreatic β-cells, Ca²⁺ influx through the voltage-gated Ca²⁺ channels is the primary Ca²⁺ event after glucose stimulation, and whether this Ca²⁺ influx triggers a further Ca²⁺ release (CICR) from the ER store is controversial (48,50). Certainly, the clear anti-parallel changes in the cytosolic and ER Ca²⁺ concentrations after depolarization-elicited Ca²⁺ influx, which have been observed in DRG neurons (9), have not so far been described for the pancreatic β-cells. The situation is complex because during the first declining phase of the cytosolic Ca²⁺ transient, after glucose-elicited membrane depolarization and Ca²⁺ influx, the ER mainly functions as a Ca²⁺ sink. In the second phase, the cytosolic Ca²⁺ decline slows, and this may be due to the ER releasing at least part of the Ca²⁺ initially taken up (12,73). The question of how the sensitivity of the Ca²⁺-release channels is controlled is therefore of considerable interest (Fig. 3). It has been proposed that cAMP sensitizes ryanodine receptors to Ca²⁺ and thereby promotes Ca²⁺ release from the ER (47). Results from a study on INS-1 cells indicate that the cAMP-regulated guanine nucleotide exchange factor II (Epac2) mediates CICR via sensitization of ryanodine receptors. This transduction pathway is independent of protein kinase A (21).

Like the acinar cells, the β-cells possess ADPribosyl cyclase, which is responsible for the production of both cADPR and NAADP (33,34,74). Therefore, it is likely that NAADP, which has such an important role in the acinar cells, is also physiologically important in the β-cells (Fig. 3). Although there is, at this moment, no physiological evidence for such a mechanism, one study, based on biochemical experiments in sea urchin eggs, recently proposed that cAMP switches the enzyme activity toward NAADP production, whereas cGMP switches the enzyme activity toward cADPR production (75). This hypothesis is attractive because many physiological stimuli have been shown (for example, in pancreatic acinar cells [76,77]) to stimulate both the nitric oxide/cGMP and the cAMP/protein kinase A pathways. Tools are available to investigate the NAADP pathway, but they remain limited. There is so far no specific antagonist of NAADP, and the self-desensitization property of the NAADP receptor is the only way to specifically block the NAADP pathway. With regard to the growing interest in the role of intracellular Ca²⁺ stores in excitable cells (8), it is also important to point out that L-type Ca²⁺-channel blockers have been reported to inhibit NAADP-evoked Ca²⁺ release in sea urchin eggs (7,58).

In view of the unresolved issues about the control of ER Ca²⁺ uptake and release in β-cells and the fact that there is now evidence for a role of cADPR in these cells, it would be valuable to study the role of NAADP and its interactions with other signaling pathways. There is an increasing number of cell types, including electrically excitable cells, in which NAADP has been reported to release Ca²⁺ and to have physiological roles (30,52). NAADP is involved in fertilization in ascidian eggs (78), starfish oocytes (79), and sea urchin eggs (30). NAADP is also involved in Ca²⁺ release in the brain (80) and cardiac microsomes (81), as well as in the control of activation and proliferation of human T-cells (82). It has been shown recently that NAADP enhances quantal neurosecretion at the frog neuromuscular junction (83). Interestingly, fertilization of sea urchin eggs triggers global Ca²⁺ waves, which can be markedly inhibited when both IP₃ and cADPR receptors are blocked (30,84).

Despite recent progress, many elements of the NAADP signaling pathway remain to be elucidated. We still lack information about the molecular nature of the NAADP receptor, and the endogenous levels of NAADP have so far not been measured. Further development of our knowledge about NAADP signaling may lead to new therapeutic targets for the treatment of diseases linked to disturbances of Ca²⁺ homeostasis.

Work in the laboratories of the authors was supported by the Medical Research Council (MRC) (U.K.). O.H.P. is an MRC Research Professor.

We thank Nina Burdakova for technical assistance.