Identification of Cyclic ADP-ribose-dependent Mechanisms in Pancreatic Muscarinic Ca2+ Signaling Using CD38 Knockout Mice*

We showed that muscarinic acetylcholine (ACh)-stimulation increased the cellular content of cADPR in the pancreatic acinar cells from normal mice but not in those from CD38 knockout mice. By monitoring ACh-evoked increases in the cytosolic Ca2+ concentration ([Ca2+] i ) using fura-2 microfluorimetry, we distinguished and characterized the Ca2+ release mechanisms responsive to cADPR. The Ca2+ response from the cells of the knockout mice (KO cells) lacked two components of the muscarinic Ca2+ release present in wild mice. The first component inducible by the low concentration of ACh contributed to regenerative Ca2+spikes. This component was abolished by ryanodine treatment in the normal cells and was severely impaired in KO cells, indicating that the low ACh-induced regenerative spike responses were caused by cADPR-dependent Ca2+ release from a pool regulated by a class of ryanodine receptors. The second component inducible by the high concentration of ACh was involved in the phasic Ca2+ response, and it was not abolished by ryanodine treatment. Overall, we conclude that muscarinic Ca2+signaling in pancreatic acinar cells involves a CD38-dependent pathway responsible for two cADPR-dependent Ca2+ release mechanisms in which the one sensitive to ryanodine plays a crucial role for the generation of repetitive Ca2+ spikes.

Cyclic ADP-ribose (cADPR), 1 first found in sea urchin eggs, mobilizes Ca 2ϩ by a mechanism independent of the inositol 1,4,5-trisphosphate (IP 3 ) pathway (1) and may act on the Ca 2ϩinduced Ca 2ϩ release mechanism as an endogenous modulator (2)(3)(4)(5)(6)(7)(8). That both cADPR and its synthetic enzyme are ubiquitously present in mammalian and invertebrate tissues (9,10) suggests that cADPR is a type of cellular Ca 2ϩ -mobilizing messenger, playing a role in Ca 2ϩ signaling in a variety of cells. The best characterized mediator for the signal is IP 3 (11), and the possibility has been raised that cADPR is a different class of mediator. How cells differentiate between these signals is a question of fundamental importance. One possibility is that they act on their respective receptors in different Ca 2ϩ pools. The major purpose of the present study is to determine their contributions to overall Ca 2ϩ signaling. We started by depicting the precise concentration response curves for Ca 2ϩ mobilization with or without cADPR. This could be achieved by using CD38 knockout mice because the majority of cADPR has been ascribed to the activity of ADP-ribosyl cyclase of CD38 (12). By estimating the change in the cellular content of cADPR and monitoring the muscarinic Ca 2ϩ response in pancreatic acinar cells, either from normal or CD38 knockout mice (13), we have successfully separated the CD38-and therefore cADPRdependent component from the inositol phosphate-sensitive component. Thus, this paper provides insight into the mechanism of cADPR-dependent Ca 2ϩ mobilization mediated by CD38 activities, which is independent of IP 3 -dependent Ca 2ϩ mobilization in muscarinic acetylcholine (ACh) receptor signaling.

EXPERIMENTAL PROCEDURES
Mice-Mice lacking CD38 were generated by homologous recombination. The generation and genotyping of the mice have been described in detail previously (13). The mice used for each experiment were derived from ICR background and were from the same litter or the same family.
Cell Preparation-Fragments of pancreatic tissue from both normal ICR (CD38 ϩ/ϩ ) and CD38 knockout ICR mice (CD38 Ϫ/Ϫ ) were excised and treated with enzymes at 37°C. Single pancreatic acinar cells used for the fluorescence and/or whole-cell current measurements were prepared using collagenase (200 units/ml; Wako, Osaka, Japan) for 3 min and trypsin (0.5 mg/ml; Sigma type XI) for 2 min and then with the same collagenase for 1 min, similar to the procedure described previously (14). The pancreatic acini used for the measurements of the cell content of cADPR and IP 3 were prepared using collagenase (1000 units/ ml) alone for 10 min.
Experimental Solutions and Materials-We used a Ca 2ϩ -free solution to avoid any contribution of external Ca 2ϩ to the signaling in pancreatic acinar cells. The Ca 2ϩ -free solution contained (in mM): 140 NaCl, 4.2 KCl, 1.13 MgCl 2 , 10 glucose, 10 HEPES (pH 7.2 adjusted with NaOH). Fluorescence experiments were carried out using a solution containing 0.5 mM EGTA. Pretreatment of the cells with ryanodine (Calbiochem) was performed at 37°C for 5 min using the Ca 2ϩ -free solution (15). The reagent was removed by washing the cells several times with the experimental solution before the fluorescence measurements (16).
Fura-2 Loading and Fluorescence Measurements-The single pancreatic acinar cells were incubated with 1 M fura-2/AM (Dojin Chemical Institute, Kumamoto, Japan) for 40 min at 37°C and then washed several times with the normal solution and kept at room temperature until use. The cells, attached to a glass coverslip, were placed in a small chamber (400 l) mounted on the stage of an inverted microscope and perfused continuously (0.5 ml/min) with a stream of the experimental solution. This arrangement permitted a rapid exchange of the bathing solution (17). The fluorescence was measured by an epifluorescence inverted microscope system described previously (18,19). Single acinar cells were alternately illuminated at excitation wavelengths of 340 and 380 nm by a rotating sector mirror at 3-ms-1-s intervals. The emission was monitored at 510 nm. Cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] i ) changes were monitored as changes in the fluorescence ratio for excitation at 340 and 380 nm (F 340 /F 380 ). The [Ca 2ϩ ] i was determined from the ratio of fluorescence traces at the two excitation wavelengths (340/ 380 nm) (20).
Whole-cell Injection of cADPR, Ca 2ϩ , and IP 3 -The single pancreatic acinar cells were subjected to standard whole-cell patch-clamp recordings for the injection of reagents (IP 3 , cADPR, and Ca 2ϩ ), in a manner similar to the procedure described previously (21,22). Patch pipettes were pulled from plain hematocrit glass capillaries by a two-stage puller (PP-83, Narishige, Tokyo) coated with beeswax. The current signals were amplified by a List EPC-7 amplifier (List Electronics, Darmstadt, Germany), appropriately low pass filtered (200 -500 Hz) and displayed on an oscilloscope screen and a digital chart recorder. The cells were immersed with external solution containing (mM): 145 NaCl, 4 KCl, 1 CaCl 2 , 2 MgCl 2 , 10 HEPES at pH 7.2 (by NaOH). Patch pipettes were filled with a solution (pipette solution) containing (mM): 144 KCl, 2 MgCl 2 , 0.1 EGTA, 10 HEPES at pH 7.2 (by NaOH). For loading of the reagents, 20 M cADPR (Molecular Probes), 20 M IP 3 (Wako, Osaka), or 200 M CaCl 2 were included in the pipette solution. In the last case, EGTA was eliminated from the pipette solution. The pipette resistance when filled with the pipette solution was 3-5 megaohms. The establishment of the whole-cell recordings was monitored with repetitive voltage pulses superimposed on the holding potential, and it appeared as a sudden increase of capacitative current during the application of a train of short suction onto the patch pipette. The pulses were interrupted soon after the current responses started. The series resistance of the whole-cell recordings was 10 -15 megaohms, which was measured with a trimmer system of the amplifier after the current response evoked by the reagent recovered to the resting level. All the experiments were carried out at room temperature (24°C).
Western Blot Analysis-Proteins were extracted with 9 volumes of 100 mM Tris-HCl (pH 7.6), 1 mM EDTA, 1% (v/v) Triton X-100 supplemented with Complete TM (one table/50 ml, Roche Diagnostics GmbH, Mannheim, Germany) from mouse tissues. The proteins (50 g) were separated on a 10% SDS-polyacrylamide gel and transferred onto a polyvinylidene fluoride membrane. After blocking with 5% skim milk, the membrane was incubated at room temperature for 1 h with an anti-CD38 polyclonal antibody raised against a peptide fragment of mouse CD38 (residues 283-301, Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The concentration of antibody was 0.4 g/ml with 5% skim milk. After rinsing, the membrane was further incubated at room temperature for 1 h with a secondary antibody labeled with horseradish peroxidase and developed using an ECL detection system (Amersham Pharmacia Biotech) as described (6,22).
Measurement of cADPR by Radioimmunoassay-The cADPR content was measured in either the presence or absence of ACh as described (23). Briefly, the suspension of enzymatically dispersed pancreatic acini (about 20 mg wet weight/ml) in 1-ml test tubes was incubated for 60 s (with or without ACh), the supernatant was discarded after centrifugation at 600 ϫ g for 1 min at 4°C, and the preparation was then stocked at Ϫ80°C. The acinar preparation was homogenized in 9 volumes of perchloric acid solution (2.5%, v/v). The homogenates were stored at Ϫ80°C and later defrosted and centrifuged for 10 min at 13,000 ϫ g. The supernatant (330 -350 l) of the homogenates was then mixed with 150 l of a suspension of Norit A (27 mg/ml in H 2 O, Nacalai tesque, Kyoto, Japan). After a 30-min incubation at 37°C, the samples were again centrifuged, and the supernatant was discarded. The pellet was washed three times with 1.0 ml of H 2 O, resuspended in a pyridineethanol-H 2 O mixture (10:50:40, v/v/v), and incubated for 120 min at 37°C. After a further centrifugation, the supernatant was collected and evaporated (Speedvac, Savant Instrument Inc., Farmingdale, NY). The evaporated materials eluted from Norit A charcoal were resuspended with 50 l of H 2 O and then incubated at 25°C for 2 h with bovine alkaline phosphatase (Sigma) and venom phosphodiesterase (Worthington Biochemical) at final concentrations of 50 and 2 units/ml, respectively, in 100 mM imidazole-HCl (pH 7.5), 2 mM MgCl 2 , 100 mM NaCl, and 400 mM KCl. This enzyme treatment was effective for degrading nucleotides that weakly cross-react with the anti-cADPR antibody (24). The reaction was terminated by adding a solution of trichloroacetic acid (at a final concentration of 4%, w/v) and kept on ice for 20 min. A clear supernatant was obtained after centrifugation at 13,000 ϫ g for 10 min. An aliquot (10 -20 l) of the supernatant was immediately neutralized with a solution of 2 M Tris base and subjected to the procedure of cADPR measurement. As a control, another aliquot was heated at 95°C for 10 min before being neutralized and analyzed by the radioimmunoassay. This heat treatment converted cADPR to ADP-ribose, resulting in no cross-reaction with the anti-cADPR antibody. The immunoreactivity of all the samples was abolished by the heat treatment. The recovery of cADPR, monitored by the recovery of [ 3 H]cADPR added in each homogenate, was 74.2 Ϯ 2.92% (n ϭ 8). Correction was introduced for the recovery of cADPR.
Measurement of IP 3 -The cellular content of IP 3 was estimated with a D-myo-inositol 1,4,5-[ 3 H]trisphosphate assay system (Amersham Pharmacia Biotech) either in the presence or absence of ACh. The suspension of enzymatically dispersed pancreatic acini (about 20 mg wet weight/ml) in 1-ml test tubes was incubated for 60 s (with or without ACh), the supernatant was discarded after centrifugation at 600 ϫ g for 1 min at 4°C, and the preparation was then stocked at Ϫ80°C. The cells were homogenized in 200 ml of ice-cold 10% (v/v) perchloric acid with an ultrasonic processor (Astrason ® , Heat Systems, Farmingdale, NY) for 10 s. Then, the samples were neutralized by the addition of 1.5 M KOH, 60 mM HEPES. After sedimentation of KClO 4 by centrifugation at 2000 ϫ g for 15 min at 4°C, the supernatant was used for the IP 3 measurement. The assay was performed according to the manufacturer's instructions.
Statistics-Statistical significance was analyzed with Student's t test. Data are expressed as means Ϯ S.E.

RESULTS
The CD38 Expression, cADPR Content, and the Changes of cADPR in Response to ACh in Pancreatic Acinar Cells from Wild (CD38 ϩ/ϩ ) and Knockout Mice (CD38 Ϫ/Ϫ )-By Western blot analysis using a polyclonal antibody against mouse CD38, we detected no CD38 expression in the pancreatic acinar cells from CD38 knockout mice, but its expression was detected from those of the wild type mice (Fig. 1A). In addition, by radioimmunoassay, we scarcely detected cADPR in the cells from the knockout mice, but it was detected in the wild type mice (Fig.  1B). Thus, the majority of the cellular cADPR content was caused by the activity of the ADP-ribosyl cyclase of CD38.
Changes in Cellular Content of cADPR and IP 3 in Response to ACh- Fig. 2 shows the changes in the cellular content of cADPR ( Fig. 2A) and IP 3 (Fig. 2B) in response to ACh ranging from 40 nM to 4 M. The cADPR content was increased by the ACh stimulation in the acini from wild mice but not in those from the knockout mice. In contrast, the IP 3 content was increased similarly in both types of acini. However, we were not certain that the method employed to measure the cADPR and IP 3 contents was sensitive enough to detect the difference in content at different ACh concentrations. The difference is shown in later sections in terms of changes in the cellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) by fura-2 microfluorimetry.
Effects of Ca 2ϩ , cADPR, and IP 3 on Ca 2ϩ Responses of Pancreatic Acinar Cells from CD38 ϩ/ϩ and CD38 Ϫ/Ϫ Mice-Pancreatic acinar cells possess two types of Ca 2ϩ activated ion channels (25). One is selective for monovalent cations (26,27) and the other for Cl Ϫ , and both are poorly dependent on membrane potentials between Ϫ60 mV and 60 mV but highly dependent on [Ca 2ϩ ] i (25,28). Thus, increases and decreases of [Ca 2ϩ ] i should be reflected by the activity of these channels. To demonstrate the presence of normal Ca 2ϩ release mechanisms responsible for IP 3 (28,29), Ca 2ϩ -induced Ca 2ϩ release (30,31), and cADPR-mediated signaling, we injected these reagents directly into the single cells with the whole-cell recording technique and monitored the Ca 2ϩ -dependent current activities (Ca 2ϩ responses). Here, at a holding potential of Ϫ40 mV achieved by the whole-cell recordings, Ca 2ϩ -dependent current responses were expected to emerge as large inward currents carried by both Cl Ϫ and monovalent cations. Shortly after the establishment of the whole-cell recordings (delay of 1-3 s) when the pipette included IP 3 , cADPR, or Ca 2ϩ , we observed a large deflection of the inward current at the membrane potential of Ϫ40 mV (Fig. 3, 3/3 in each reagent injections). However, without the reagents there were no responses (3/3). That all the reagents successfully induced Ca 2ϩ responses in the single acinar cells from both the wild and knockout mice indicated that the Ca 2ϩ releasing machinery was well preserved in both types of cells.
Close inspection of the records revealed that there might have been differences between the responses of the wild type and knockout cells to cADPR and/or IP 3 . That is, there was a longer response to cADPR and a shorter response to IP 3 in the knockout cells than in those of the wild type. The reduction in cADPR hydrolase activity in the knockout pancreatic acinar cells (13) may explain their prolonged response to cADPR. The cooperative action of cADPR for enhancing IP 3 -induced Ca 2ϩ responses (32) may explain the shortened response to IP 3 because cADPR was eliminated in the knockout cells. However, further study is needed to determine the precise mechanisms of the reagent sensitivity.
ACh-evoked Ca 2ϩ Responses from Single Pancreatic Acinar Cells of Wild (CD38 ϩ/ϩ ) and Knockout Mice (CD38 Ϫ/Ϫ )-Based on the above results, we employed fura-2 microfluorimetry and compared the ACh-induced pancreatic Ca 2ϩ responses of the wild (CD38 ϩ/ϩ ) and CD38 knockout mice (CD38 Ϫ/Ϫ ) in a wide range of ACh concentrations from 10 to 20,000 nM. We used a Ca 2ϩ -free solution containing 0.5 mM EGTA to avoid any contribution of external Ca 2ϩ to the signaling in pancreatic acinar cells in the later experiments with fura-2 microfluorimetry. Table I summarizes the results in terms of the peak magnitude, the onset time after the application of ACh, and the

FIG. 2. Changes in cellular content of cADPR (A) and IP 3 (B) in response to
ACh. Pancreatic acini of CD38 ϩ/ϩ and CD38 Ϫ/Ϫ mice were incubated in increasing concentrations of ACh, and the cellular contents of IP 3 and cADPR were measured. A, cADPR in the cells was extracted, concentrated, and measured by radioimmunoassay using an antibody against cADPR. The cADPR content was calculated by accounting for the recovery of cADPR in the extraction and concentration procedures. The control values of cADPR in CD38 Ϫ/Ϫ and CD38 ϩ/ϩ cells were 0.248 Ϯ 0.248 fmol/mg protein and 0.413 Ϯ 0.240 fmol/mg protein, respectively. Asterisk (*) indicates significant difference from the value with 0 mM ACh at p Ͻ 0.05. There were no significant differences among the values with 0.04 -4 M ACh. B, IP 3 in the cells was measured. The control IP 3 values in CD38 Ϫ/Ϫ and CD38 ϩ/ϩ cells were 4.09 Ϯ 2.65 fmol/mg protein and 3.42 Ϯ 1.46 fmol/mg protein, respectively. n ϭ 4 for each point. Vertical bars indicate S.E. Asterisk (*) and symbol (#) indicate significant difference from the values with 0 mM ACh at p Ͻ 0.05 (*) and Ͻ 0.01 (**) in CD38 Ϫ/Ϫ and at p Ͻ 0.01 (##) in CD38 ϩ/ϩ , acinar cells, respectively. There were no significant differences among the values with 0.04 -4 M ACh. The threshold concentration of ACh that induced a detectable Ca 2ϩ response was 10 nM, and the response was a transient deflection with no significant difference observed in the two types of pancreatic acinar cells. A difference appeared at 40 nM ACh. Immediately after the application of 40 nM ACh, the cells from CD38 ϩ/ϩ mice showed a repetitive Ca 2ϩ response (oscillations) during the ACh application, lasting for 2-3 min. In contrast, that of CD38 Ϫ/Ϫ was sporadic with the same ACh stimulation (Fig. 4A). However, at 400 nM ACh, the difference between the responses from CD38 ϩ/ϩ and CD38 Ϫ/Ϫ mice became less than that at 40 nM ACh, and both responses showed large phasic increases in the fura-2 signals (Fig. 4B).

FIG. 3. Activation of Ca 2؉ release mechanisms by patch-clamp whole-cell injection of Ca 2؉ , cADPR, and IP 3 into single CD38
The numerical data of the [Ca 2ϩ ] i increases in Table I were plotted against the ACh concentration (Ca 2ϩ response curve) in Fig. 5, A and B. The magnitude response curve from the knockout mice (CD38 Ϫ/Ϫ ) was smoothly graded with the increasing ACh concentrations. However, that of CD38 ϩ/ϩ showed two prominent phases separated at the concentration of 400 nM ACh. This feature was also seen in the Ca 2ϩ index (18), which reflects the mixed information of both the magnitude and frequency of the response. Both response curves were replotted in Fig. 5, C and D, where the Ca 2ϩ response from the knockout mice (CD38 Ϫ/Ϫ ) was subtracted from that of the wild type (CD38 ϩ/ϩ ) at each ACh concentration. Two prominent peaks were seen in the diagrams at the values of 40 and 4000 nM, indicating that two CD38-dependent phases, in other words, two cADPR-dependent phases, separated at 400 nM, are present in the ACh-induced release of internal Ca 2ϩ . One was induced by low ACh concentrations up to 400 nM (the first cADPR-dependent component) and the other by high concentrations of ACh over 400 nM (the second cADPR-dependent component).

Effect of Ryanodine on the ACh-induced Ca 2ϩ Release in Single Pancreatic Acinar Cells from Wild Type (CD38 ϩ/ϩ ) and
Knockout (CD38 Ϫ/Ϫ ) Mice-Because cADPR has been postulated to be an endogenous modulator of the ryanodine-sensitive Ca 2ϩ release mechanism (33-35), we next examined the effect of ryanodine on the pancreatic ACh-induced Ca 2ϩ responses from normal cells. The results were compared with the Ca 2ϩ responses from the knockout cells. Fig. 6 shows the fura-2 signals in both types of cells treated either with or without ryanodine (500 M) prior to the ACh stimulation (40 or 400 nM). The 40 nM ACh stimulation, without ryanodine, induced repetitive and sporadic transient responses in the wild and knockout mice, respectively, as described in Fig.  4A. In contrast, with ryanodine, the same stimulation induced a sporadic transient response and no response in the wild and knockout mice, respectively. Comparing these records, we noticed that the response to 40 nM ACh stimulation in the wild type cells with ryanodine resembled that of knockout mice without ryanodine (Fig. 6A). The diagram in Fig. 6C summarizes the effects of ryanodine on the ACh-induced Ca 2ϩ responses in pancreatic acinar cells from the wild mice, showing  that ryanodine abolished the first cADPR-dependent component but not the second component. The complex effects of ryanodine on the knockout cells depended on the ACh concentration. That is, at concentrations of ACh from 10 to 400 nM, we could not detect any significant effect of ryanodine on the magnitude of the Ca 2ϩ response in knockout cells (Fig. 6D). In contrast, it enhanced the response at concentrations of ACh over 400 nM. It was noteworthy that the respective concentrations of ACh corresponded to the first and second cADPR-dependent components normally present in the wild type cells.
The effect of ryanodine at concentrations of ACh corresponding to the first cADPR-dependent component could be interpreted as indicating either that ryanodine treatment depletes the first cADPR-sensitive Ca 2ϩ pool or that it inhibits the Ca 2ϩ release from the pool. Either interpretation could be deduced from the result that ryanodine eliminated the first cADPR-dependent component from the wild cells and was without effect on the knockout cells. However, we prefer the first interpretation because the basal [Ca 2ϩ ] i level was higher in the ryanodine-treated cells than that in both the wild and knockout cells without the treatment. The basal [Ca 2ϩ ] i without ryanodine was 143.2 Ϯ 3.2 nM (n ϭ 269) in the wild cells and 151.3 Ϯ 3.1 nM (n ϭ 292) in the knockout cells. With ryanodine it was 234.6 Ϯ 4.6 nM (n ϭ 68) in the wild cells and 247.5 Ϯ 2.6 nM (n ϭ 87) in the knockout cells. DISCUSSION Comparing the ACh-induced Ca 2ϩ responses in pancreatic acinar cells from normal mice (CD38 ϩ/ϩ wild type: normal cells) with those of CD38 knockout mice (CD38 Ϫ/Ϫ type: KO cells), we distinguished the CD38-and therefore cADPR-dependent component from the overall muscarinic Ca 2ϩ signaling. The major findings are that: 1) ACh stimulation increased the cellular content of cADPR in the normal cells but not in KO cells; 2) the Ca 2ϩ response curve in the normal cells was separated into two phases at 400 nM ACh; 3) in contrast, the Ca 2ϩ response curve from KO cells was a smoothly graded one, and it lacked the two components inducible by ACh at concentrations below 400 nM (the first cADPR-dependent component) and over 400 nM (the second cADPR-dependent component) usually present in normal cells; 4) the first cADPR-dependent component contributed to the generation of repetitive Ca 2ϩ spikes; 5) ryanodine treatment eliminated the first cADPR-dependent component (Fig. 6).
Physiological Significance of cADPR-dependent Ca 2ϩ Release-IP 3 and cADPR are established Ca 2ϩ -mobilizing messengers that activate internal IP 3 and ryanodine receptors, respectively (36 -38). Because we scarcely detected cADPR formation in KO cells, the majority of Ca 2ϩ release in these cells could arise from Ca 2ϩ pools insensitive to cADPR but sensitive to IP 3 and/or yet unknown messengers. Thus, the IP 3 -sensitive Concentration-response curve for the change in [Ca 2؉ ] i evoked by ACh in CD38 ؉/؉ and CD38 ؊/؊ pancreatic acinar cells. A, the peak magnitude of evoked [Ca 2ϩ ] i increases was plotted against the ACh concentration. B, a normalized estimate of the mean increase of [Ca 2ϩ ] i , the Ca 2ϩ index, was plotted against the ACh concentration. Ca 2ϩ index was calculated by integrating the area covered by the trace of fluorescence intensity from the start to 200 s after the response. *** and ** represent p Ͻ 0.001 and Ͻ 0.01 versus CD38 ϩ/ϩ , respectively. E, response from CD38 ϩ/ϩ ; q, response from CD38 Ϫ/Ϫ pancreatic acinar cells. C and D, the difference between the response from CD38 ϩ/ϩ and that of CD38 Ϫ/Ϫ shown in A and B (E minus q).
pool obviously plays an important role in pancreatic Ca 2ϩ responses, and it steadily contributes to the Ca 2ϩ release in a wide range of ACh concentrations as evidenced by the smoothly graded response curve from KO cells in Fig. 5A. The superposition of cADPR on IP 3 and/or yet unknown messengers manifested two prominences on the curve, suggesting the presence of two subdivisions of cADPR-sensitive Ca 2ϩ pools (the first and second cADPR-dependent components described above). The first cADPR-dependent Ca 2ϩ pool is strictly sensitive to ryanodine, but the second one is rather resistant as evidenced by Fig. 6C.
A large part of the Ca 2ϩ release is from the first cADPR-dependent pool at concentrations of ACh up to 40 nM, as shown in Table I and Fig. 4. The contribution of this pool promotes the sharp repetitive spikes (Fig. 4A), which could be advantageous for the fine control of Ca 2ϩ -dependent cell function (39). That the repetitive spikes were severely impaired in the knockout mice suggests that they depend on cADPR formation. Thus, it could be that cADPR is one of the crucial messengers for maintaining the repetitive Ca 2ϩ spikes. However, we cannot exclude the possible contribution of IP 3 . It has been reported in other cells that IP 3 is generated in an oscillatory manner (40). We do not know the detailed pattern of cADPR and IP 3 formation in our cells; however it is likely that the ACh stimulation generates both messengers in a repetitive or oscillatory fashion. The synchronized formation of these two messengers may generate various patterns of repetitive Ca 2ϩ spikes under cooperation with increased cellular Ca 2ϩ , which is ascribed mainly to the first cADPR-dependent Ca 2ϩ pool. The critical concentration of ACh that switched the mode of the Ca 2ϩ response from the repetitive spike to the phasic deflection was 400 nM (Table I, "Frequency"). Based on the above findings, it could be that the messenger formation is no longer repetitive but rather phasic over this concentration of ACh. The significance of the second cADPR-dependent Ca 2ϩ pool could be that it serves as the resource of the phasic release of Ca 2ϩ .
Ryanodine was able to induce Ca 2ϩ release at an exceedingly high rate (Fig. 6D) with concentrations of ACh over 400 nM in KO cells. This may have resulted from a complex mechanism that involves enhanced sensitivity to ryanodine in the second cADPR-sensitive Ca 2ϩ pool. Such an enhancement may have arisen from the constant lack of cADPR in the KO cells (Fig. 1B).
CD38 in the Ca 2ϩ Signaling of Muscarinic Receptor Stimulation-The presence of CD38 in the Ca 2ϩ release mechanism is of particular importance. The CD38-mediated Ca 2ϩ signaling may be associated directly with the receptor stimulation ( Fig.  2A). This signaling pathway may be separated from that of IP 3 since both the formation of cADPR ( Fig. 2A) and its contribution to the Ca 2ϩ release (Fig. 5A) are large at 40 nM ACh, at which concentration the contribution of IP 3 is still low. It is likely that muscarinic stimulation activates both signaling pathways in parallel and that they contribute synergistically to the overall Ca 2ϩ signaling in pancreatic acinar cells. However, the precise underlying mechanism of cADPR formation after the muscarinic ACh stimulation remains unknown.