Cyclic ADP Ribose-Dependent Ca2+ Release by Group I Metabotropic Glutamate Receptors in Acutely Dissociated Rat Hippocampal Neurons

Group I metabotropic glutamate receptors (group I mGluRs; mGluR1 and mGluR5) exert diverse effects on neuronal and synaptic functions, many of which are regulated by intracellular Ca2+. In this study, we characterized the cellular mechanisms underlying Ca2+ mobilization induced by (RS)-3,5-dihydroxyphenylglycine (DHPG; a specific group I mGluR agonist) in the somata of acutely dissociated rat hippocampal neurons using microfluorometry. We found that DHPG activates mGluR5 to mobilize intracellular Ca2+ from ryanodine-sensitive stores via cyclic adenosine diphosphate ribose (cADPR), while the PLC/IP3 signaling pathway was not involved in Ca2+ mobilization. The application of glutamate, which depolarized the membrane potential by 28.5±4.9 mV (n = 4), led to transient Ca2+ mobilization by mGluR5 and Ca2+ influx through L-type Ca2+ channels. We found no evidence that mGluR5-mediated Ca2+ release and Ca2+ influx through L-type Ca2+ channels interact to generate supralinear Ca2+ transients. Our study provides novel insights into the mechanisms of intracellular Ca2+ mobilization by mGluR5 in the somata of hippocampal neurons.


Introduction
The group I metabotropic glutamate receptors (mGluRs), which include mGluR1 and mGluR5, play important roles in regulating intrinsic excitability and synaptic plasticity [1,2,3]. Importantly, intracellular Ca 2+ contributes to various aspects of mGluRmediated effects. Enhancement of neuronal excitability [4,5,6] and long-term depression mediated by mGluR (mGluR-LTD) [7] were shown to be blocked by intracellular dialysis of BAPTA, and the involvement of Ca 2+ -dependent proteins such as PICK1 and NCS-1 in mGluR-LTD has recently been demonstrated [8,9,10]. In addition, mGluR triggers retrograde endocannabinoid signaling, an effect that is greatly enhanced by increases in Ca 2+ [11,12]. However, the signaling pathways and the source of Ca 2+ that contributes to these diverse effects have not yet been clearly elucidated.
It is well known that group I mGluRs mobilize Ca 2+ from intracellular stores in hippocampal neurons [13,14]. As group I mGluRs are coupled to Gq proteins [15,16], Ca 2+ mobilization may involve the phospholipase C (PLC)/inositol-3-triphosphate (IP 3 ) signaling pathways [1]. Indeed, the synergistic or supralinear Ca 2+ release by group I mGluR stimulation paired with backpropagating action potential (AP) was shown to be from IP 3 receptor (IP 3 R)-sensitive intracellular stores in apical dendrites of CA1 hippocampus [17,18]. However, studies in midbrain dopaminergic neurons demonstrated that intracellular Ca 2+ mobilization by group I mGluRs required cyclic ADPR ribose (cADPR)/ryanodine receptors (RyRs) as well as IP 3 /IP 3 Rs [19]. The role of cADPR in mGluR-mediated Ca 2+ signaling is supported by the study showing that the glutamate-induced stimulation of ADP-ribosyl cyclase occurs preferentially in NG108-15 neuroblastoma/glioma hybrid cells over-expressing mGluR1, 3, 5, and 6 [20]. It is not yet clear if this finding could also be extended to hippocampal neurons, but considering the frequent involvement of PLC-independent signaling pathways in several effects of group I mGluRs [21,22,23,24], the possibility that Ca 2+ mobilization by group I mGluR may be mediated by signal pathways other than PLC/IP 3 Rs should be tested in hippocampal neurons.
Ca 2+ signaling in neurons is highly compartmentalized, with Ca 2+ having distinctive roles in each section [25,26,27]. Mechanisms involved in axonal and dendritic Ca 2+ signaling have been extensively studied due to their importance in the regulation of neurotransmitter release and synaptic plasticity [28,29,30]. Somatic Ca 2+ signals also play important roles in regulating cellular excitability, synaptic plasticity and gene expression [7,31,32], but the mechanisms involved in somatic Ca 2+ signals are not well studied. As different neuronal compartment may have distinct Ca 2+ signaling machinery, results obtained from dendrites or axons may not extend to the somatic Ca 2+ signals. Therefore, separate studies of somatic Ca 2+ signals are warranted.
In the current study, we directly investigated the signaling pathways underlying somatic Ca 2+ mobilization by group I mGluRs. Using microfluorometric Ca 2+ measurements in the somata of acutely dissociated rat hippocampal neurons loaded with Fura 2-AM, we discovered that stimulation of group I mGluRs induces the cADPR-dependent Ca 2+ mobilization from ryanodine-sensitive stores. Our results represent a novel mechanism for Ca 2+ mobilization by group I mGluRs in hippocampal neurons.

Results
mGluR5 is responsible for DHPG-induced Ca 2+ release from intracellular stores To investigate the mechanisms underlying Ca 2+ increase by group I mGluR stimulation, acutely dissociated hippocampal CA1 neurons were loaded with 2 mM Fura 2-AM for microfluorometry experiments. The application of 50 mM (RS)-3,5-dihydroxyphenylglycine (DHPG), a specific group I mGluR agonist, to these cells rapidly increased intracellular Ca 2+ concentrations in the somata. The amplitude of DHPG-induced Ca 2+ increase (Ca DHPG ) was highly variable among cells, ranging from ,20 nM to ,500 nM (mean = 97.567.8 nM, n = 168), but Ca DHPG values obtained from the same cell upon repetitive application of DHPG at 2 min intervals yielded consistent data (Ca DHPG,2 /Ca DHPG,1 = 98.665.1%, n = 6). To investigate the mechanism of DHPG-induced Ca 2+ increase, we regarded Ca DHPG,1 as the control and applied various experimental conditions prior to the second application of DHPG. The relative amplitude of Ca DHPG,2 compared with Ca DHPG,1 (Ca DHPG,2 / Ca DHPG,1 ) was obtained to study the contribution of each variable to the DHPG-induced Ca 2+ increase.
The amplitude of the second Ca 2+ transient was significantly suppressed by the selective mGluR5 antagonist, MPEP (25 mM), but not by the mGluR1 antagonist LY367385 (100 mM), indicating that mGluR5 is responsible for the DHPG-induced Ca 2+ increases in hippocampal CA1 neurons ( Figure 1A & 1B). DHPG-induced Ca 2+ transients were not affected by the removal of external Ca 2+ or the inhibition of receptor-operated Ca 2+ entry by SKF96365 (10 mM), but they were markedly suppressed when cells were pretreated with the sarcoplasmic/endoplasmic reticulum Ca 2+ -ATPase (SERCA) inhibitor thapsigargin (2 mM) for 5 min, indicating that DHPG mobilizes Ca 2+ from its intracellular stores ( Figure 1C-1E). mGluR5-induced Ca 2+ release from intracellular stores is independent of PLC-IP 3 We next tested whether the PLC/IP 3 signaling pathways link mGluR5 and Ca 2+ mobilization. Interestingly, DHPG-induced Ca 2+ release was not affected by the PLC inhibitor U73122 (1 mM; Figure 2A & 2C). Conversely, muscarinic receptor-mediated Ca 2+ transients induced by the muscarinic receptor agonist carbachol (CCh; 10 mM) were completely inhibited by U73122 ( Figure 2B & 2C), confirming that U73122 effectively inhibited the PLC pathway under our experimental conditions. Subsequently, we loaded intact cells with heparin (20 mg/ml in the electroporation pipette), a competitive antagonist of the IP 3 Rs [33], using a singlecell electroporator. Loading of heparin was confirmed by coadministration of the fluorescent compound Alexa Fluor-488 ( Figure 2D). This manipulation completely inhibited the induction of Ca 2+ transients by CCh, but not those by DHPG ( Figure 2E). For quantitative analyses, we measured the first DHPG-induced Ca 2+ transient in a Fura 2-AM-loaded neuron, patched the same neuron with a Fura 2 (10 mM)-containing pipette with or without heparin (1 mg/ml), and re-applied DHPG to elicit a second Ca 2+ transient. We confirmed that the amplitudes of the second Ca 2+ transients, which were measured at a holding potential of 260 mV, did not differ from those of the first Ca 2+ transients (103.2614.7%, n = 4) ( Figure 2F, left bar). The inclusion of heparin did not affect DHPG-induced Ca 2+ transients (97.1624.6%, n = 4) ( Figure 2F, right bar).
PLCb1 and PLCb4 are known to mediate group I mGluR signaling in the brain [34,35]; therefore, we tested whether DHPG induces Ca 2+ transients in hippocampal CA1 neurons isolated from mice lacking the PLCb1 or PLCb4 subunits. As illustrated in Figure 2G, DHPG was still able to induce Ca 2+ transients in cells from PLCb1 or PLCb4 knockout mice. These results suggest that mGluR5 induces Ca 2+ release independently of PLC/IP 3 signaling pathways.

Glutamate-induced Ca 2+ influx is mediated by L-type Ca 2+ channels activated by AMPA receptor-mediated depolarization
Our results demonstrated a predominant role for the cADPR/ RyR signaling pathway in mGluR5-induced Ca 2+ release in the somata of hippocampal neurons. In the brain, glutamate is the natural neurotransmitter that stimulates mGluRs; glutamate can activate both mGluRs and ionotropic glutamate receptors (iGluRs), which include AMPA and NMDA receptors. NMDA receptors and the Ca 2+ -permeable AMPA receptors are possible sources of Ca 2+ entry into the cell. In addition, glutamate-induced membrane depolarization should activate voltage-gated Ca 2+ channels (VGCCs) to cause Ca 2+ influx. Indeed, we confirmed that glutamate depolarizes the membrane potential by 29.064.6 mV (n = 4). As shown in Figure 4A, the applications of DHPG and glutamate (30 mM) on the same cell demonstrated that the amplitude of glutamate-induced Ca 2+ transients (Ca Glu ) was significantly greater than Ca DHPG . The average Ca DHPG was 92.8610.3 nM from 74 cells, whereas Ca Glu was 284.3623.6 nM from the same population ( Figure 4B, p,0.01). The Ca DHPG / Ca Glu ratio was calculated to be 39.063.6% (n = 74).
Subsequently, experiments were performed to determine the source of Ca 2+ entry when cells were stimulated with glutamate. Repetitive application of glutamate at 2 min intervals yielded reproducible Ca Glu values (Ca Glu,2 /Ca Glu,1 = 103.668.7%, n = 5). When the bath solution was replaced with a Ca 2+ -free solution prior to the second application of glutamate, the Ca Glu,2 was significantly decreased so that the Ca Glu,2 /Ca Glu,1 ratio was 50.469.6% (n = 6, Figure 5A). Unexpectedly, the addition of the NMDA receptor blocker AP-5 before the second application of glutamate had no effect, and the Ca Glu,2 /Ca Glu,1 ratio was 101.765.0% (n = 7, Figure 5B). In contrast, the addition of the AMPA receptor blocker CNQX prior to the second application of glutamate was as effective as the removal of Ca 2+ , producing a Ca Glu,2 /Ca Glu,1 ratio of 45.167.1% (n = 8, Figure 5C), suggesting that AMPA receptors may be involved in glutamate-induced Ca 2+ influx. However, the addition of 1-naphthyl acetyl spermine (NASPM, 10 mM), a specific blocker of the Ca 2+ -permeable AMPA receptor, had no significant effect on glutamate-induced Ca 2+ transients ( Figure 5D). These results suggest that neither NMDA nor AMPA receptors are involved in the observed Ca 2+ influx pathway, but that AMPA receptor activation may trigger Ca 2+ influx through VGCCs by depolarizing the membrane potential. In support of this, we found that glutamate-induced membrane depolarization was 5.061.2 mV (n = 3) in the presence of CNQX, which is significantly less than that observed in the control (29.064.6 mV, n = 4, p,0.05).
To identify the subtype of VGCC responsible for the calcium influx, we tested the effects of specific pharmacological inhibitors. We found that nimodipine (10 mM), an L-type Ca 2+ channel blocker, significantly reduced glutamate-induced Ca 2+ transients ( Figure 6A). In contrast, the amplitude of Ca Glu was not significantly affected by v-conotoxin GVIA (1 mM), v-agatoxin IVA (200 nM), and NiCl 2 (100 mM), indicating that N-type, P/Q type, T-type and R-type Ca 2+ channels are not involved ( Figure 6B). These data indicate that in glutamate application cADPR/RyR-dependent Ca 2+ release does not interact with the Ca 2+ influx through L-type Ca 2+ channels To understand the complexity of mGluR-mediated Ca 2+ signaling, it is necessary to examine the interaction between mGluR-induced Ca 2+ release and glutamate-induced Ca 2+ influx. It has been shown that Ca 2+ entry through VGCCs interacts synergistically with IP 3 to enhance mGluR-mediated Ca 2+ release in apical dendrites of hippocampal CA1 neurons [17,18]. Supralinear Ca 2+ release by DHPG along with either membrane potential depolarization or NMDA receptor activation was also demonstrated in primary cultured hippocampal neurons [13]. Conversely, Topolnik et al. (2009) demonstrated that dendritic Ltype Ca 2+ channels are enhanced by mGluR5-induced Ca 2+ mobilization from ryanodine-sensitive stores in the GABAergic interneurons of the hippocampus [53]. Notably, cADPR was shown to enhance L-type Ca 2+ channels induced by both orthograde and retrograde pathways in NG108-15 cells [54]. Therefore, we analyzed Ca 2+ transients by DHPG and glutamate, under the assumption that Ca Glu represents the sum of Ca 2+ influx (Ca Influx ), Ca 2+ mobilization by mGluR5 (Ca DHPG ) and the supralinear Ca 2+ transients (Ca SUP ) by synergistic effects of mGluR5 and Ca 2+ influx (Ca Glu = Ca Influx +Ca DHPG +Ca SUP ).
We estimated Ca SUP by subtracting the sum of Ca DHPG (representing RyR-dependent release; indicated by the red boxes in Figure 7A & 7B) and Ca Influx , which was estimated from the glutamate-induced Ca 2+ transients in the presence of MPEP or ryanodine (indicated by blue boxes, Figure 7A & 7B) from Ca Glu in the control condition (Ca SUP = Ca Glu 2Ca DHPG 2Ca Influx ). In this   series of experiments, Ca DHPG was 34.3610.2% of Ca Glu (n = 6, Figure 7A) and 39.7612.9 of Ca Glu (n = 4, Figure 7B), which is comparable to the values shown in Figure 4 (39.063.6%, n = 74). The estimated Ca Influx was found to be 58.9610.2% of Ca Glu (n = 6, Figure 7A) in MPEP and 52.4619.0% of Ca Glu (n = 4, Figure 7B) in ryanodine. Thus, the sum of Ca DHPG and Ca Influx was close to Ca Glu (93.2610.5% in Figure 7A and 92.169.1% in Figure 7B), and Ca SUP was found to be negligible ( Figure 7C). Because previous reports have demonstrated the role of IP 3 Rs in synergistic Ca 2+ release by mGluR and backpropagating APs [17,18], we tested the effect of U73122 (1 mM) but found no significant effect. Ca Glu,2 in the presence of U73122 was 93.668.8% of Ca Glu,1 (n = 8, Figure 7D), suggesting that the PLC/IP 3 pathway does not contribute to either Ca DHPG or Ca SUP . Taken together, these results suggest that, in the somata of hippocampal neurons, cADPR/RyR-dependent Ca 2+ mobilization by mGluR5 and Ca 2+ influx through the L-type Ca 2+ channels do not interact to generate supralinear Ca 2+ transients.

Discussion
We have demonstrated the mechanisms underlying the mGluR5-induced Ca 2+ mobilization in the somata of hippocampal neurons. Our results indicate that the cADPR signaling pathways are responsible for the mGluR5-induced Ca 2+ mobilization from ryanodine-sensitive stores. In addition, we found that glutamateinduced Ca 2+ influx via the L-type Ca 2+ channels does not interact with mGluR5-induced Ca 2+ mobilization to cause a supralinear Ca 2+ increase. These results provide novel insights into the mechanisms for group I mGluR-induced Ca 2+ mobilization in the somata of hippocampal neurons. cADPR has long been known to be an endogenous Ca 2+releasing messenger [45,46,51], and the role of cADPR in neuronal Ca 2+ signaling has previously been identified [41,43,44]. Involvement of cADPR in mGluR-mediated Ca 2+ signaling was previously demonstrated in midbrain dopamine neurons [19]. This study demonstrated that, in the presence of synaptic blockers (except for mGluR), synaptic stimulation of the dopamine neurons evoked Ca 2+ waves originating in dendrites 10-50 mm away from the soma, and that the mGluR-induced Ca 2+ waves were inhibited only when both cADPR and PLC/IP 3 signaling pathways were inhibited. It was thus concluded that mGluR-mediated Ca 2+ mobilization involves two pathways mediated by cADPR and IP 3 in a redundant manner. Our results differ, in that only the cADPR signaling pathway and ryanodine-sensitive stores contributed to Ca 2+ mobilization by mGluR5 in the somata of hippocampal neurons (Figures 2 & 3). However, these results do not mean that the Ca 2+ stores in hippocampal neuron somata are insensitive to IP 3 , as muscarinic receptor-mediated Ca 2+ mobilization was mediated by PLC/IP 3 signaling pathway ( Figure 2). Possibly, Ca 2+ stores in hippocampal neurons are fundamentally sensitive to both IP 3 and cADPR, but signaling pathways that regulate these mediators can differ depending on cell types and subcellular localization. It will be of interest to test the contribution of the cADPR-mediated Ca 2+ releases from RyRs to dendritic Ca 2+ signaling in the hippocampus.
Dendritic Ca 2+ signaling induced by group I mGluR has been extensively studied in CA1 hippocampus, and the results obtained in these studies suggest the involvement of the PLC/IP 3 signaling pathway [17,18]. Remarkably, large amplitude Ca 2+ increases induced by repetitive synaptic stimulation, which were considered to be attributable to IP 3 -induced Ca 2+ release, were precisely confined to the large apical dendrite shaft at the branch point of oblique dendrites [55]. This result indicates that even in dendrites of the same neuron, Ca 2+ signaling mechanisms are spatially segregated. In the present study, we used acutely dissociated hippocampal neurons with thick apical dendrites of ,50 mm. We measured Ca 2+ signals only from somata in response to bath application of mGluR agonist or glutamate. Thus, our results represent somatic Ca 2+ release mechanisms without interference from dendritic Ca 2+ signaling mechanisms. We have provided solid evidence that in the somata of hippocampal CA1 pyramidal neurons, cADPR-mediated Ca 2+ releases from RyRs serve as the predominant mechanism in mGluR-induced Ca 2+ release. It should also be noted that there has not been a direct examination of somatic Ca 2+ release machinery despite the fact that somatic Ca 2+ signals have distinctive roles, such as protein synthesis and gene expression [31,32]. Further studies are required to test the possibility that dendritic and somatic Ca 2+ release mechanisms may be distinct from each other in hippocampus.
One of the interesting features reported for mGluR-mediated Ca 2+ release in dendrites is that Ca 2+ entry through VGCCs interacts synergistically with IP 3 , and supralinearly increase the GluR-mediated Ca 2+ release in apical dendrites of hippocampal  . Glutamate-induced Ca 2+ influx does not interact with DHPG-induced Ca 2+ mobilization. (A) Ca 2+ influx by glutamate was determined by comparing the amplitudes of two glutamate-induced Ca 2+ transients in the presence or absence MPEP, which blocks DHPGinduced Ca 2+ mobilization (blue box), and was compared to Ca 2+ mobilization (red box). (B) Ca 2+ influx by glutamate was determined by comparing the amplitudes of two glutamate-induced Ca 2+ transients in the presence or absence of ryanodine, which blocks DHPG-induced Ca 2+ mobilization (blue box), and was compared to Ca 2+ mobilization (red box). (C) Bar graphs summarizing the relative contribution of Ca 2+ mobilization (red), Ca 2+ influx (blue) and estimated supralinear Ca 2+ mobilization (black) in both conditions. (D) Ca Glu was not inhibited by U73122. Scale bars indicate 10 sec (horizontal) and 50 nM (vertical). doi:10.1371/journal.pone.0026625.g007 CA1 neurons [17,18]. Supralinear Ca 2+ release by DHPG and either membrane potential depolarization or NMDA receptor activation was also demonstrated in primary cultured hippocampal neurons [13]. However, we showed that cADPR/RyRdependent Ca 2+ release by mGluR5 was not supralinearly increased by Ca 2+ influx in the somata of CA1 pyramidal neurons (Figure 7). This suggests that, unlike IP 3 -dependent Ca 2+ releases, cADPR-dependent Ca 2+ release through RyRs is not potentiated by Ca 2+ influx. However, we still need to consider another type of possible synergism between cADPR/RyR-dependent Ca 2+ release and L-type Ca 2+ channels, as demonstrated in previous studies. In NG108-15 cells transfected with mGluRs, direct applications of cADPR enhanced Ca 2+ influx through L-type Ca 2+ channels [54]. In addition, dendritic Ca 2+ transients evoked by back-propagating action potentials, which are mediated by VGCCs, were potentiated by mGluR5-mediated Ca 2+ release and PKC activation in hippocampal oriens-alveus interneurons [53]. These findings suggested that the RyRs-sensitive Ca 2+ releases enhance L-type Ca 2+ channels via PKC-dependent mechanisms. An interesting observation in this study is that the potentiation occurs exclusively in specific microdomains of dendrites; possible absence of similar microdomains in the somata of CA1 pyramidal neurons, which needs to be determined in future studies, would explain the lack of a synergistic interaction found in this study (Figure 7).
In this study, the experiments were performed using acutely dissociated neurons, as dissociated neurons have several advantages in studying signaling mechanisms in the somata. In this preparation, indirect effects or presynaptic components can be excluded. Furthermore, rapid application and wash-out of drugs are guaranteed. It is very difficult to obtain healthy cells by enzymatic dissociation method from rats over 2 weeks old, and therefore we used immature rats (P7-P14). However, glutamate signaling is still developing at this age, and the Ca 2+ mobilization mechanisms found in the current study may not extend into the somatic mechanism of adult neurons. To exclude this possibility, we examined the DHPG-induced Ca 2+ release mechanisms in the somata of CA1 pyramidal neurons in brain slices from 4-week-old rats (data not shown), and found that the mechanisms were consistent with those found in acutely dissociated immature neurons. Therefore, the mechanisms of mGluR5-induced somatic Ca 2+ mobilization found in the current study may be extended into at least young adult neurons.
We have demonstrated that AMPA receptors and L-type Ca 2+ channels, but not NMDA receptors, are responsible for the Ca 2+ influx by glutamate. It was shown previously that NMDAdependent Ca 2+ entry evokes Ca 2+ increases primarily in spines, which are more concentrated with oblique dendrites [56,57]. Nakamura et al. [55] also showed that synaptic stimulation evoked Ca 2+ influx by NMDA receptors exclusively at oblique dendrites, whereas backpropagating APs evoke Ca 2+ increase at all dendritic locations. Thus, in acutely dissociated neurons that usually lack oblique dendrites the role of NMDA receptors in Ca 2+ influx should be limited and membrane potential depolarization by AMPA receptors and the opening of L-type Ca 2+ channels may be responsible for Ca 2+ influx instead. Another notable finding is that the contribution of Ca 2+ influx to Ca Glu was larger than that of Ca 2+ release (Figure 7), signifying the importance of L-type Ca 2+ channels in somatic Ca 2+ signaling.
In summary, we investigated the Ca 2+ mobilization mechanisms by group I mGuRs using acutely dissociated hippocampal neurons. As discussed, the signaling pathways revealed in the current study may represent what occurs in the somata of hippocampal CA1 neurons, and this may be distinct from dendritic Ca 2+ release machinery, which has been extensively characterized by other groups. The nucleus, as well as other important intracellular organelles, resides in the somata. Therefore, Ca 2+ -dependent molecules regulating cellular excitability and synaptic plasticity may be regulated by the cADPR/RyR-dependent Ca 2+ release by group I mGluRs in hippocampal CA1 neurons.

Ethics Statement
Protocols were approved by the Animal Care Committee at Seoul National University (SNU-080107-7). Animal handling was conducted in accordance with national and international guidelines. The number of animals used was minimized, and all necessary precautions were taken to mitigate pain or suffering.

Preparation of acutely isolated hippocampal neurons
Hippocampal CA1 pyramidal neurons were isolated as described previously [21]. Briefly, 7 to 14-day-old Sprague-Dawley rats (14 to 34 g) were decapitated under pentobarbital anesthesia. The brain was quickly removed and submerged in ice-cold artificial cerebrospinal fluid (ACSF, see below) saturated with 95% O 2 and 5% CO 2 . Transverse hippocampal slices (400 mm thick) were prepared using a vibratome (VT1200, Leica). After a 30 min recovery period at 32uC, the slices were treated with protease type XIV (1 mg/5 ml, Sigma) for 30-60 min, and with protease type X (1 mg/5 ml, Sigma) for 10-15 min at 32uC. The slices were allowed to recover during a 1-hour incubation period at room temperature. The CA1 region was identified and punched out under a binocular microscope (SZ40, Olympus), placed in a recording chamber containing normal Tyrode (NT) solution (see below) and mechanically dissociated using a Pasteur pipette to release individual neurons. The dissociated neurons were allowed to adhere to the bottom of the recording chamber for 10-20 min. Cells were identified as pyramidal neurons by their typical large pyramidalshaped cell body with a thick apical dendritic stump of ,50 mm under an inverted microscope (IX70, Olympus). The isolation of hippocampal CA1 neurons from PLCb1 or PLCb4 knockout mice (generated as described in [35]) was performed as above.
(RS)-3,5-DHPG, LY367385, MPEP, SKF96365, CNQX, AP-5, ryanodine, TTX were purchased from Tocris. U73122 was purchased from Biomol. Fura 2, Fura 2-AM and 8-NH 2 -cADPR were obtained from Molecular Probes, and v-conotoxin-GVIA and v-agatoxin-IVA were from Anygen. All other drugs were purchased from Sigma. Stock solutions of drugs were made by dissolving in deionized water or DMSO according to manufacturer's specifications and were stored at 220uC. On the day of the experiment one aliquot was thawed and used. The final concentration of DMSO in solutions was maintained below 0.1%.

Calcium measurements
Acutely dissociated hippocampal CA1 neurons were loaded by incubation with 2 mM Fura 2-AM plus 0.01% Pluronic F-127 in NT solution for 10 min at room temperature. For fluorescence excitation, we used a polychromatic light source (xenon-lamp based, Polychrome-IV; TILL-Photonics), which was coupled to the epi-illumination port of an inverted microscope (IX70, Olympus) via a quartz light guide and a UV condenser. Microfluorometry was performed with a 406 water immersion objective (NA 1.15, UAPO 406 W/340, Olympus) and a photodiode (TILL-Photonics).

Calibration of Ca 2+ measurements
Calibration parameters were determined using in vivo calibration as described in [58]. The effective dissociation constant of Fura 2 (K eff ) was calculated from K eff = [Ca 2+ ](R max 2R int )/ (R int 2R min ), where [Ca 2+ ] was entered as 231 nM (assuming a dissociation constant (K d ) of BAPTA of 222 nM at pH 7.2). The estimated R min , R max and K eff (mM) measured using an inverted microscope were typically 0.27, 3.95 and 0.93, respectively. A standard two-wavelength protocol was used for fluorescence measurement of cells. Fluorescence intensity was measured at 1 Hz with double wavelength excitation at 340 nm (F 340 ) and 380 nm (F 380

Electrophysiology
Current clamp recordings of membrane potential were performed using an EPC-10 amplifier (HEKA Elektronik) at room temperature. Membrane potentials were recorded from acutely dissociated hippocampal CA1 neurons in a conventional whole cell configuration at a sampling rate of 10 kHz filtered at 1 kHz. Data were acquired using an IBM-compatible computer running Pulse software v8.67 (HEKA Elektronik). The patch pipettes were pulled from borosilicate capillaries (Hilgenberg-GmbH) using a Narishige puller (PC-10, Narishige). The patch pipettes had a resistance of 3-5 megaohms when filled with abovementioned K-based pipette solutions.

Single cell electroporation
The loading of heparin and 8-NH 2 -cADPR was performed by single cell electroporation. Micropipettes were pulled as described above and were filled at their tips with NT solutions containing heparin (20 mg/ml) or 8-NH 2 -cADPR (100 mM) plus Alexa Fluor-488 (200 mM, Molecular Probes). Micropipettes were controlled by a micromanipulator (Burleigh) to reach cells, and square electric pulses generated with an electroporator (Axoporator 800A; Molecular Devices/MDS Analytical Technologies) were applied to transfer the mixture into the cells.

Data analysis
Data were analyzed using IgorPro (version 4.1, WaveMetrics) and Origin (version 6.0, Microcal) software. Statistical data are expressed as the mean 6 S.E., where n represents the number of cells studied. The significance of differences between the peaks was evaluated using a Student's t-test with confidence levels of p,0.01 (**) and p,0.05 (*).