Regulation of Nuclear Ca2+ Signaling by Translocation of the Ca2+ Messenger Synthesizing Enzyme ADP-ribosyl Cyclase during Neuronal Depolarization*

In neurons, voltage-gated Ca2+ channels and nuclear Ca2+ signaling play important roles, such as in the regulation of gene expression. However, the link between electrical activity and biochemical cascade activation involved in the generation of the nuclear Ca2+ signaling is poorly understood. Here we show that depolarization of Aplysia neurons induces the translocation of ADP-ribosyl cyclase, a Ca2+ messenger synthesizing enzyme, from the cytosol into the nucleus. The translocation is dependent on Ca2+ influx mainly through the voltage-dependent L-type Ca2+ channels. We report also that specific nucleoplasmic Ca2+ signals can be induced by three different calcium messengers, cyclic ADP-ribose, nicotinic acid adenine dinucleotide phosphate (NAADP), both produced by the ADP-ribosyl cyclase, and inositol 1,4,5-trisphosphate (IP3). Moreover, our pharmacological data show that NAADP acts on its own receptor, which cooperates with the IP3 and the ryanodine receptors to generate nucleoplasmic Ca2+ oscillations. We propose a new model where voltage-dependent L-type Ca2+ channel-induced nuclear translocation of the cytosolic cyclase is a crucial step in the fine tuning of nuclear Ca2+ signals in neurons.

such as gene transcription and synaptic plasticity, the mechanism involved in its generation remains largely unexplored (1, 9 -11).
Nucleoplasmic Ca 2ϩ changes can result either from passive diffusion through the nuclear pore following cytosolic Ca 2ϩ increases or from mobilization of Ca 2ϩ sequestered into the nuclear envelope by various intracellular messengers (12,13). In the latter process, cytosolic and nucleoplasmic Ca 2ϩ can be independently regulated (12,14,15). The differential control of the cytosolic and nuclear Ca 2ϩ involves, in principle, several factors including cytosolic Ca 2ϩ buffering and Ca 2ϩ sequestration by the endoplasmic reticulum and the mitochondria (12, 14 -18). The nucleus is equipped with all the necessary machinery for generating nuclear Ca 2ϩ signaling, including Ca 2ϩ pumps and release channels (12). Several groups have identified functional inositol 1,4,5-trisphosphate (IP 3 ) 2 and/or ryanodine receptors in the inner membrane of the nuclear envelope (19 -23) or in the outer membrane of the contagious nucleoplasmic reticulum (23)(24)(25).
NAADP and cADPR, two Ca 2ϩ releasing messengers are produced by multifunctional enzymes of the ADP-ribosyl cyclase family (26 -28). Several of these enzymes have been purified and cloned, including the ectoenzymes CD38 and CD157, and a soluble cyclase from the sea mollusk Aplysia (28). The surface antigen CD38 has been mostly found at the plasma membrane with the catalytic site exposed to the extracellular space (29). To solve this topological paradox, several authors have proposed that CD38 could generate the synthesis of extracellular cADPR, which then may enter into the cell interior possibly through nucleoside transporters (30). In addition, intracellular locations of CD38 have been reported such as in nucleus where no links with physiological stimuli were reported and no evidence for a nuclear role for NAADP (19,22). The synthesis of NAADP through the base-exchange reaction by this enzyme requires exclusive acidic conditions. In the case of the Aplysia cyclase, although the base-exchange reaction depends critically on the pH, it appears that NAADP synthesis is possible at more neutral pH (26). In mammals, additional enzymes exist and CD38 may not be regarded as the principal * The work was supported by grants from the Association Française contre les Myopathies and the Association pour la Recherche sur le Cancer (to J.-M. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1 and S2. 1 To whom correspondence should be addressed. cyclase (31)(32)(33). For example, a novel ADP-ribosyl cyclase for cADPR synthesis has recently been detected in the brain (31) and one for NAADP synthesis in myometrial cells (34). Recently a soluble form has been purified in sea urchin (35). Clearly, our understanding of the physiological role of the expanding ADP-ribosyl cyclase family remains poor and the cyclase from Aplysia remains the only prototypical form for the soluble enzyme characterized in neurons so far. Pioneering work on the neurobiology of learning has been performed on Aplysia californica. For example, it has been found that synaptic potentiation at sensory-motor neuron synapses require Ca 2ϩ entry through L-type Ca 2ϩ channels and/or Ca 2ϩ mobilization from intracellular stores sensitive to IP 3 and ryanodine depending of the potentiation type (7, 36 -38). Our previous studies have shown that the soluble ADP-ribosyl cyclase is not only present in the neurons of the Aplysia buccal ganglion, but that its product cADPR, an endogenous modulator of the ryanodine receptors, can enhance the evoked synaptic transmission (39). More recently, we showed that NAADP, the other product of the cyclase, can increase neurotransmitter release at well identified cholinergic synapses (40). To further delineate the Ca 2ϩ signaling pathway mediated by the Aplysia cyclase, we looked at its distribution in the Aplysia nervous system that has not been characterized.
In this study, we have revealed for the first time that in resting conditions, the soluble Aplysia ADP-ribosyl cyclase is localized in the cytosol of the soma of neurons. Importantly, we have found that depolarization of Aplysia neurons induces the translocation of ADP-ribosyl cyclase from the cytosol into the nucleus. The translocation is dependent on Ca 2ϩ influx through the voltage-dependent L-type Ca 2ϩ channels. We also show that the neuronal nucleus is a Ca 2ϩ stores, which could generate specific nucleoplasmic Ca 2ϩ signals in response to three different calcium messengers, cADPR, NAADP, both produced by the ADP-ribosyl cyclase, and IP 3 . Finally, we propose that the translocation of the Ca 2ϩ signaling enzyme to nucleus following neuronal depolarization may provide a new link between the electrical activity and the biochemical cascades leading to nuclear Ca 2ϩ signals generation.

MATERIALS AND METHODS
Preparation of Isolated Nuclei-Most of the experiments were done on nuclei isolated from abdominal, pedal, and pleural ganglia of adult A. californica and Aplysia punctata. The animals were first perfused with an isotonic MgCl 2 solution and the ganglia, after extraction, were pinned in a chamber bathed with artificial seawater (ASW) (NaCl 460 mM; KCl 10 mM; CaCl 2 11 mM; MgCl 2 25 mM; MgSO 4 28 mM; Tris-HCl buffer 10 mM; pH 7.8). The connective tissue packing the neurons was sharply removed with fine forceps. Then the neurons were gently broken in a homogenizer to extract the nuclei in an intracellular buffer with the following composition: KCl 450 mM; K 2 HPO 4 2 mM; HEPES 50 mM; MgCl 2 4 mM; CaCl 2 /EGTA (depending of the experiments); pH 7.2, adjusted with 5 N KOH. Some experiments were done on nuclei from the ganglia of A. punctata. No species differences were observed.
Confocal Imaging-Preparations of nuclei were incubated with the different fluorescent Ca 2ϩ probes. For the experiments performed on the nuclear envelope, Mag Fluo4 AM was used ( ex ϭ 494 nm and em ϭ 516 nm) (K d ϭ 22 M). The free Ca 2ϩ concentration of the intracellular medium was strongly buffered to 400 nM using 1.4 mM Ca 2ϩ and 2 mM EGTA. During loading, nuclei were incubated with 30 M Mag Fluo4 AM and 5 mM ATP for 1 h in the intracellular medium.
To measure Ca 2ϩ changes in the nucleoplasmic space, we loaded the nuclei with 10-kDa Fluo-4 dextran ( ex ϭ 494 nm and em ϭ 516 nm) (K d ϭ 3 M) at a concentration of 20 M together with 3 mM ATP for 1 h. The intracellular medium was buffered to 100 nM free Ca 2ϩ . All experiments were performed at room temperature in the same Ca 2ϩ /EGTA buffers used for loading the probe with a final ATP concentration of 3 mM. The loaded nuclei were placed on a polylysine-coated coverslip in a chamber. Ca 2ϩ changes in the lumen of the nuclear envelope were measured using an inverted Leica SP2 RS confocal microscope (objective ϫ40 and ϫ20) and the data analyzed with the 2.5 Leica confocal software.
Most of the nucleoplasmic Ca 2ϩ measurements were done using the same confocal imaging system. But some were also performed using a right Olympus microscope (objective ϫ40) coupled with a CCD camera and a xenon lamp. Images were analyzed with the Axon Imaging Workbench software. For all experiments, Ca 2ϩ concentration changes are expressed in fluorescence ratio as f/f 0 (fluorescence/fluorescence at the beginning of the experiment).
Neuron depolarization was achieved by the addition of KCl (110 mM) for 20 min (with equimolar substitution of NaCl in control experiments). Ganglia were incubated for at least 30 min in the presence of Ca 2ϩ chelator (BAPTA-AM, 50 M, Molecular Probes) or Ca 2ϩ channel inhibitors (nifedipine, 10 M from Sigma, and -conotoxins CN VIIA and S VI B, 5 M, kind gift of Dr. J. Molgo) before KCl depolarization. In the case of conotoxins, the ganglia were incubated in Ca 2ϩ -free ASW for 30 min and then reperfused with standard ASW before KCl stimulation.
To measure Ca 2ϩ changes in the intact Aplysia neurons, we used as previously described intracellular injections of Rhod-2 (K d 570 nM), selected as the best Ca 2ϩ -sensitive probe in Aplysia neurons that exhibit autofluorescence at some excitation wavelengths (40). Intracellular injections were performed using air pressure pulses delivered by controlled air valves (Pico Pump, WPI). The Rhod-2 concentration in the injecting pipette was at 1 mM and neurons were injected until they appeared dark red with an estimated dye concentration of 100 M. Neurons were depolarized by addition of 60 mM KCl (with equimolar substitu-  OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 JOURNAL OF BIOLOGICAL CHEMISTRY 27861 tion of NaCl). To visualize DNA in the Aplysia ganglia, we used the cell-permeant SYTO 13 Green from Molecular Probes at a concentration of 5 M ( ex ϭ 488 nm and em ϭ 509 nm).

Nuclear Translocation of the ADP-ribosyl Cyclase
Sequence Analysis-The sequence of the Aplysia ADP-ribosyl cyclase was scanned for consensus motifs homologies, using the Prosite Data base on the ExPASy Proteomic server.

Localization of the Aplysia Cyclase in Intact Nervous
Ganglia-The Aplysia cyclase was first discovered in the ovotestis (28,42), but its distribution in the nervous system of Aplysia has not been characterized. Typical Aplysia neurons contain a large nucleus surrounded by a thin cytoplasm (Fig.  1A). The localization of the enzyme was revealed by confocal microscopy using a specific antibody raised against the cyclase (41). In resting conditions, the cyclase was almost exclusively localized in the thin cytosol surrounding the nucleus of the soma (Fig. 1B, n ϭ 75/81). To investigate whether the cyclase distribution could change during neuronal activities, the neurons were depolarized for 20 min by the addition of KCl, which evoked a Ca 2ϩ response (Fig. 1D, n ϭ 7). Under depolarizing conditions, nuclear localization of the enzyme was dramatically enhanced (Fig. 1C, n ϭ 35/42). This depolarization-induced Ca 2ϩ elevation and translocation of the cyclase was prevented by prior incubation of the ganglia with the Ca 2ϩ chelator BAPTA-AM (50 M, 30 min, n ϭ 28/29) (Fig. 2, A, B, and E, n ϭ 4). The next step was to investigate what types of voltage-gated Ca 2ϩ channels could be involved in our observation. Previous studies from our laboratory on Aplysia neurons have shown that during neuronal depolarization the Ca 2ϩ currents are reduced nearly 40 -50% in the presence of the L-type Ca 2ϩ channel blockers nifedipine and N-type, P-type voltage-gated Ca 2ϩ channels block nearly 40% of calcium influx during depolarization by conotoxins (43). So, in Aplysia neurons, 80 -90% Ca 2ϩ entry occurred through L-type, N-type, and P-type voltage-gated Ca 2ϩ channels.
In our experiments, extracellular addition of the L-type Ca 2ϩ channel inhibitor (nifedipine 10 M, Sigma, n ϭ 33/33) likewise prevented the translocation and reduced dramatically the Ca 2ϩ responses evoked by KCl (Fig. 2, B and E, n ϭ 4). In addition, activation of the L-type Ca 2ϩ channel by Bay K-8644 at 10 M evoked nuclear translocation of the cyclase (Fig. 2C, n ϭ 24/53). In contrast, extracellular application of -conotoxins (CN VIIA and S VI B, 5 M), which block the N-and P-type Ca 2ϩ channels, did not affect the depolarization-induced translocation of the cyclase (Fig. 2D, n ϭ 15/20). These data indicate that translocation of the cyclase to the nucleus is a Ca 2ϩ -dependent process triggered by Ca 2ϩ entry through the L-type voltage-dependent channels, which, in turn, are activated by membrane depolarization.
The Nuclear Envelope Is a Ca 2ϩ Store Sensitive to Intracellular Ca 2ϩ Releasing Messengers-Cytosolic Ca 2ϩ buffering by the endoplasmic reticulum and the mitochondria (16) could insulate the nucleus (18). Although it is poorly known in neurons, the nucleus could be autonomous by mobilization of Ca 2ϩ sequestered into the nuclear envelope (11). To investigate whether the nuclear Ca 2ϩ stores are responsive to both cADPR (11) and NAADP, the two Ca 2ϩ messengers produced by the Aplysia cyclase, the large nuclei of the Aplysia neurons were isolated. The luminal Ca 2ϩ of the nuclear envelope was measured by loading Mag Fluo-4 (30 M) into the lumen, using a membrane permeant AM-form of the Ca 2ϩ -sensitive dye of low affinity (Fig. 3A). Fluorimetric measurements using confocal microscopy showed that the luminal space could be loaded with Ca 2ϩ following the addition of ATP (5 mM) suggesting that Ca 2ϩ -ATPase is involved in nuclear store refilling (Fig. 3B) (n ϭ  3). More importantly, addition of the second messengers IP 3 (5-10 M; n ϭ 8) or cADPR (5-10 M; n ϭ 8) (Fig. 3, C and D) likewise decreased the Ca 2ϩ concentration in the nuclear envelope, indicating that it is a fully functional Ca 2ϩ store.
NAADP is known to produce a biphasic response, releasing Ca 2ϩ from the sensitive stores at low concentrations while desensitizing its own receptor at high concentrations (44). Various NAADP concentrations, ranging from 10 nM to 100 M, were thus tested on the isolated nuclei (Fig. 3, E-G). With NAADP in the nanomolar range (10 -500 nM), all nuclei examined responded with decreases in Ca 2ϩ contents (n ϭ 3 for each concentration) (Fig. 3, E and F). In stark contrast, at 1 and 10 M, NAADP failed to evoke a Ca 2ϩ response in 2 of 3 nuclei tested for each concentration and at 100 M NAADP, 5 of 6 nuclei, showed no response (Fig. 3G). Finally, if the nuclear Ca 2ϩ stores were first depleted by thapsigargin (5 M), a widely used inhibitor of the sarco(endoplasmic) Ca 2ϩ -ATPase, neither IP 3 nor NAADP could release any additional Ca 2ϩ , indicating that both messengers release Ca 2ϩ from the same thapsigarginsensitive stores (Fig. 3H, n ϭ 8 and 10, respectively).
To determine whether the Ca 2ϩ release from the nuclear envelope would result in an increase of Ca 2ϩ in the nucleoplasm, a fluorescent Ca 2ϩ probe, Fluo-4 dextran (10 kDa, 20 M), was loaded into the nucleoplasm (Fig. 4A). Addition of IP 3 (10 M; n ϭ 21), cADPR (5 M; n ϭ 14), or NAADP (500 nM; n ϭ 37) at the optimal concentrations determined above indeed evoked a modest (supplemental Fig. S1A) but significant increase of the nucleoplasmic Ca 2ϩ concentration (Fig.  4, B-D). In many nuclei, intranuclear Ca 2ϩ fluctuation events called oscillations were observed on the top of a Ca 2ϩ transient response following application of the second messengers (Fig. 4. B-D). The oscillations were characterized by amplitude of about 0.1 ⌬F/F and duration of about 10 s and occurred in bursts. The duration of the Ca 2ϩ oscillation period varied among the nuclei and the frequency was estimated by dividing the number of oscillations by the duration of the burst. In 64% of the investigated nuclei, NAADP evoked 0.7 Ϯ 0.06 oscillations/min for 1340 Ϯ 140 s (n ϭ 18/28), whereas IP 3 evoked 1.1 Ϯ 0.14 oscillations/min for  OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 the mean frequency evoked by IP 3 and NAADP was statistically significant (p ϭ 0.012). The mean frequency of cADPR-evoked Ca 2ϩ oscillations was 0.9 Ϯ 0.13/min, for 1075 Ϯ 290 s, in 75% of the nuclei (n ϭ 6/8), which was not significantly different from either of the two means listed above.

Nuclear Translocation of the ADP-ribosyl Cyclase
The nuclear Ca 2ϩ oscillations suggest that the release mechanisms involved are capable of self-limiting by feedback. Indeed, none of the three Ca 2ϩ messengers were able to totally discharge the nuclear stores. Addition of a Ca 2ϩ ionophore (ionomycin, 5 M) elicited a much larger increase. Also consistent with feedback is the transient nature of the Ca 2ϩ increase (supplemental data Fig. S1B). The decline is most likely due to diffusion of Ca 2ϩ out of the nuclei through the nuclear pores, following cessation of Ca 2ϩ release. As shown in supplemental data Fig. S1A, after the release induced by IP 3 had reached the peak and started to decline, addition of Ca 2ϩ (10 mM) elicited a rapid and large increase in the nucleoplasmic Ca 2ϩ concentration, which could be buffered back by EGTA, indicating that the nuclear pores are permeant to Ca 2ϩ , in agreement with previ- ous work (13,16,21,45,46). Similar results were obtained after the NAADP-or cADPR-induced release (n ϭ 6).
In contrast to nanomolar concentrations (Fig. 5A), 100 M NAADP failed to induce an increase of the nucleoplasmic Ca 2ϩ (n ϭ 9), consistent with the biphasic behavior described above (Fig. 3). However, subsequent addition of cADPR (5 M) still resulted in a nucleoplasmic Ca 2ϩ increase (n ϭ 5) (Fig. 5A). Thus, selfdesensitization of the NAADP release mechanism did not impair the effect of cADPR on these nuclei, indicating that cADPR and NAADP target different release mechanisms in the nuclear envelope. The specificity of NAADP-evoked Ca 2ϩ release was verified by testing two inactive analogs of NAADP, nicotinamide adenine dinucleotide phosphate (NADP, 500 nM) and nicotinic acid adenine dinucleotide (NAAD, 500 nM). Both were ineffective; either applied alone or in combination (n ϭ 5) (Fig. 5B).
The Nuclear Ca 2ϩ Store Supports Ca 2ϩ Releasing Receptor Interactions-Previous studies showed that NAADP targets acidic Ca 2ϩ stores, such as lysosomes and endosomes (47,48). The isolated nuclei we used appeared to be free of these acidic compartments (supplemental data Fig. S1C).
As shown in Fig. 3H, blocking the endoplasmic Ca 2ϩ pump with thapsigargin (30 min, 5 M) can deplete the nuclear stores even in the presence of ATP. After store depletion, none of the three messengers, IP 3 , cADPR, nor NAADP, were able to evoke any changes of nucleoplasmic Ca 2ϩ in 16 of 20 nuclei tested (IP 3 , n ϭ 6; cADPR n ϭ 4; NAADP, n ϭ 10) (Fig. 5, D-H).
Our previous study has shown that in intact Aplysia neurons the NAADP receptor itself is not sensitive to heparin or ryanodine (40) but in some intact cells, the cytosolic Ca 2ϩ signals evoked by NAADP require the activation of adjacent IP 3 and/or ryanodine receptors through a Ca 2ϩ -induced Ca 2ϩ release mechanism (32, 49 -54). Heparin (50 g/ml) was used to block the IP 3 receptor, which indeed inhibited or drastically reduced the IP 3 -induced Ca 2ϩ release (Fig. 6D) in 19 nuclei tested. The heparin treatment had no or very little effect on cADPR-elicited Ca 2ϩ responses (n ϭ 8) (Fig. 6E), but did block NAADP (500 nM) from evoking Ca 2ϩ release (Fig. 6F) (n ϭ 22/26). This suggests there was strong cooperation between the IP 3 and NAADP receptors in the nuclear envelope, amplifying the Ca 2ϩ signals induced by NAADP.
Treatment of the Aplysia nuclei with 500 M ryanodine did not block the Ca 2ϩ release elicited by IP 3 (10 M) (n ϭ 7) (Fig. 6G), but strongly inhibited both the cADPR (5-10 M) evoked (Fig. 6H, n ϭ 8) and NAADP (500 nM) evoked responses (Fig. 6I). In the presence of ryanodine, NAADP failed to induce a Ca 2ϩ release in 22 of 33 nuclei, whereas in the 11 responding nuclei the release was drastically reduced in amplitude (n ϭ 11). These data indicate that full NAADPinduced Ca 2ϩ responses require functioning ryanodine receptors. Another drug that has been reported to be effective in inhibiting NAADP signaling in intact cells is SKF 96365 (55,56). The drug (10 M) had no significant effect on the IP 3 -or cADPR-evoked Ca 2ϩ release (n ϭ 4 for IP 3 ; n ϭ 5 for cADPR) (Fig. 6, J and K) but blocked the NAADP-evoked response (Fig. 6L) (n ϭ 16). Our data contrast with what has been described in preparations of nuclei isolated from pancreatic acinar cells, where NAADP was suggested to target the ryanodine receptor (13).

DISCUSSION
NAADP is a Ca 2ϩ releasing molecule that has recently been added to the list of second messengers for Ca 2ϩ mobilization (57). The Ca 2ϩ releasing property of NAADP was first discovered in sea urchin eggs (58) and has since been found in other living organisms including plants and mammals (32,54,59). It is known to be involved in regulating a wide range of physiological processes (6, 60 -62). Evidence suggests it targets a distinct, but yet uncharacterized, receptor in the thapsigargin-insensitive stores, such as lysosomes or secretory granules (47,48,63,64). Interestingly, it has been recently suggested that the lysosomal Trp-like channel mucolipin I could be the target of NAADP (65). In the same way, the differential effects of SK&F 96365 we observed in our nuclear experiments indicate that the release mechanisms activated by the three Ca 2ϩ messengers have different pharmacology, suggesting separate and distinct receptors are involved (Fig. 7). Our experiments provide evidence for the nuclear envelope as a new Ca 2ϩ store sensitive to NAADP in neurons through the activation of its own receptor. This nuclear NAADP sensitivity could in principle correlate well with the Trp-like channel mucolipin I distribution because it has a putative nuclear localization sequence (66). The density of the NAADP receptor in the nucleus, however, would appear to be low, and its full effect requires the amplification of nearby IP 3 and ryanodine receptors (Figs. 6 and 7). The Ca 2ϩ store distribution can be entirely different in the synapses. Indeed, electrophysiological experiments have shown that neurotransmitter release is enhanced by IP 3 , cADPR, or NAADP injections into the intact presynaptic Aplysia neurons. In these injection experiments in the presynaptic neurons the Ca 2ϩ responses evoked by the three messengers are all comparable and independent (40). Thus, the NAADP-induced release is shown to be unaffected by blockage of either the IP 3 or ryanodine receptor, indicating the density of the NAADP receptor would be high enough in the synapses that further amplification by the Ca 2ϩ -induced Ca 2ϩ release is not necessary.
In neurons, nucleoplasmic Ca 2ϩ elevations are particularly important for transcription factor activation. Calcium has been shown to directly bind transcription factors like DREAM (4) or activate the nuclear CaM kinase pathways to regulate gene expression (67). Despite its crucial importance, in neurons, the mechanism involved in the generation of nuclear Ca 2ϩ signals is poorly known. In our experiments, we have shown that the neuronal nucleus is able to generate calcium oscillations in response to second messengers. Nuclear oscillations have been observed in the nucleus of intact starfish oocytes in response to injection of cADPR (68). Similar oscillations have also been seen in mammalian neurons in response to activation of metabotropic glutamate receptors (69). They are likely to be the result of opening and closing clusters of the release channels, as proposed for the Ca 2ϩ sparks seen in the myocytes. In nuclei from pancreatic acinar cells, no nuclear Ca 2ϩ oscillations were reported for any of the three messengers tested (13). Our study is the first to report that the three well known second messengers evoke nuclear Ca 2ϩ oscillations with distinct frequencies within a single target nucleus. Ca 2ϩ oscillations, especially when occurs in the nucleoplasm, are more efficient in activating transcription factors than a sustained Ca 2ϩ elevation (1). Interestingly, previous studies established that optimal activation of transcription factors like NF-AT is achieved when oscillation frequency is comprised in a window from 0.66/min to 2/min (70,71), which fits with the second messenger-evoked Ca 2ϩ oscillation frequencies obtained in our study. The CaM kinase II is particularly sensitive to calcium oscillations and has been characterized as a frequency decoder able to transduce these frequencies in different amounts of activity (72). In addition, in our experiments we calculated that the amplitude of the Ca 2ϩ elevation observed in our isolated nuclei is about 200 nM and is in a range reported for substantial activation of the neuronal CaM kinases of 100 -400 nM Ca 2ϩ /calmodulin concentrations (72). Finally, the frequency of Ca 2ϩ oscillations is of first importance in the physiology of neurons. For example, it has been found that a 3-fold variation of frequency of the spontaneous Ca 2ϩ spikes in activity in embryonic spinal neuronal regulates the neurotransmitter phenotype expression (73). In this respect, although we have no evidence for it, the various Ca 2ϩ oscillation frequencies observed in our study could provide a mean to discriminate among differential transcription pathways (70,74).
The presence of three different and functional Ca 2ϩ release mechanisms in a cell provides a high degree of versatility for Ca 2ϩ signaling. These release mechanisms do not need to be distributed uniformly inside the cells. The nuclear translocation of the cyclase in the Aplysia neurons described in this study documents yet a novel way for selective and specific activation of the nuclear Ca 2ϩ stores, which are the main stores in the neurons, which have very little cytoplasm (Fig. 1). In non-neuronal cells, the phospholipase C and protein kinase C, which resulted in nuclear synthesis of IP 3 and diacylglycerol have also been reported to translocate to the nucleus in response to physiological stimuli (12,20,24,75,76). Here we provide the first evidence showing that the soluble ADP-ribosyl cyclase, the enzyme responsible for the synthesis the other two Ca 2ϩ messengers, cADPR and NAADP, can specifically be induced to translocate into the nucleus. The exact mechanism involved in such a translocation is not yet elucidated. This translocation occurs in a Ca 2ϩ -dependent manner and through . Pretreatment of the nuclei with heparin (50 g/ml) inhibited the NAADP-and IP 3 -induced Ca 2ϩ responses but had little or no effect on cADPR-evoked Ca 2ϩ changes (D-F). Pretreatment of nuclei with ryanodine (500 M) reduced drastically the NAADP-and cADPR-induced Ca 2ϩ responses, but not the IP 3 -evoked Ca 2ϩ changes (G-I). Incubation of the nuclei with SK&F 96365 (10 M) inhibited the NAADP-evoked Ca 2ϩ changes (L), but did not impair the IP 3 -and cADPR-induced nucleoplasmic Ca 2ϩ changes (J and K). OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 specific L-type voltage-dependent channels activation. We have a primary analysis of the ADP-ribosyl cyclase amino acid sequence, which revealed the presence of putative consensus motifs (supplemental data Fig. S2). Indeed myristoylation and phosphorylation sites for PKC and casein kinase II are present. The myristoylation is a post-translational modification that allows protein attachment to the cytoplasmic face of the intracellular membrane. Very recently, a new concept called "calcium/myristoyl switch" has been reported for targeting protein to different organelles (77)(78)(79)(80). This mechanism implies a protein conformational change that exposes the myristoyl group following calcium entry. This new mechanism could be highly relevant to the regulation of the cellular localization of the Aplysia cyclase and opens new perspectives for future investigation.

Nuclear Translocation of the ADP-ribosyl Cyclase
Two main mechanisms have been suggested for explaining the nuclear Ca 2ϩ -dependent transcriptional cascade involving L-type Ca 2ϩ channels. One involves Ca 2ϩ entry and diffusion to the nucleus and the other involves activation of the calcium-dependent signaling protein at the mouth of the L-type Ca 2ϩ channels (80). Our study clearly brings evidence that a synthesizing messenger enzyme could be recruited by Ca 2ϩ entering specifically through the L-type Ca 2ϩ channels and then convey the information directly to the nucleus by a Ca 2ϩ -dependent translocation mechanism. Finally our novel observation that the presence of coordinated Ca 2ϩ release mechanisms are able to produce nuclear Ca 2ϩ oscillations together with translocation of the Ca 2ϩ signaling enzyme provide the necessary versatility for the neurons to respond to a wide range of stimuli.  . Nuclear translocation of Aplysia ADP-ribosyl cyclase and specific nuclear Ca 2؉ oscillations generated by the three messengers. In resting conditions, the soluble Aplysia ADP-ribosyl cyclase is localized in the thin cytosol surrounding the nucleus of the soma (A). Depolarization of Aplysia neurons induces the translocation of ADP-ribosyl cyclase from the cytosol into the nucleus. The translocation is dependent on Ca 2ϩ influx mainly through the voltage-dependent L-type Ca 2ϩ channels (B). C, specific nucleoplasmic Ca 2ϩ signals can be induced by three different calcium messengers, cADPR, NAADP, both produced by the ADP-ribosyl cyclase, and IP 3 . We show that NAADP acts on its own receptor, which cooperates with the IP 3 and ryanodine receptors to generate nucleoplasmic Ca 2ϩ oscillations.