Subpopulation of store-operated Ca2+ channels regulate Ca2+-induced Ca2+ release in non-excitable cells.

Ca2+-induced Ca2+ release (CICR) is a well characterized activity in skeletal and cardiac muscles mediated by the ryanodine receptors. The present study demonstrates CICR in the non-excitable parotid acinar cells, which resembles the mechanism described in cardiac myocytes. Partial depletion of internal Ca2+ stores leads to a minimal activation of Ca2+ influx. Ca2+ influx through this pathway results in an explosive mobilization of Ca2+ from the majority of the stores by CICR. Thus, stimulation of parotid acinar cells in Ca2+ -free medium with 0.5 microm carbachol releases approximately 5% of the Ca2+ mobilizable by 1 mm carbachol. Addition of external Ca2+ induced the same Ca2+ release observed in maximally stimulated cells. Similar results were obtained by a short treatment with 2.5-10 microm cyclopiazonic acid, an inhibitor of the sarco/endoplasmic reticulum Ca2+ ATPase pump. The Ca2+ release induced by the addition of external Ca2+ was largely independent of IP(3)Rs because it was reduced by only approximately 30% by the inhibition of the inositol 1,4,5-trisphosphate receptors with caffeine or heparin. Measurements of Ca2+ -activated outward current and [Ca2+](i) suggested that most CICR triggered by Ca2+ influx occurred away from the plasma membrane. Measurement of the response to several concentrations of cyclopiazonic acid revealed that Ca2+ influx that regulates CICR is associated with a selective portion of the internal Ca2+ pool. The minimal activation of Ca2+ influx by partial store depletion was confirmed by the measurement of Mn2+ influx. Inhibition of Ca2+ influx with SKF96365 or 2-aminoethoxydiphenyl borate prevented activation of CICR observed on addition of external Ca2+. These findings provide evidence for activation of CICR by Ca2+ influx in non-excitable cells, demonstrate a previously unrecognized role for Ca2+ influx in triggering CICR, and indicate that CICR in non-excitable cells resembles CICR in cardiac myocytes with the exception that in cardiac cells Ca2+ influx is mediated by voltage-regulated Ca2+ channels whereas in non-excitable cells Ca2+ influx is mediated by store-operated channels.

Ca 2؉ -induced Ca 2؉ release (CICR) is a well characterized activity in skeletal and cardiac muscles mediated by the ryanodine receptors. The present study demonstrates CICR in the non-excitable parotid acinar cells, which resembles the mechanism described in cardiac myocytes. Partial depletion of internal Ca 2؉ stores leads to a minimal activation of Ca 2؉ influx. Ca 2؉ influx through this pathway results in an explosive mobilization of Ca 2؉ from the majority of the stores by CICR. Thus, stimulation of parotid acinar cells in Ca 2؉ -free medium with 0.5 M carbachol releases ϳ5% of the Ca 2؉ mobilizable by 1 mM carbachol. Addition of external Ca 2؉ induced the same Ca 2؉ release observed in maximally stimulated cells. Similar results were obtained by a short treatment with 2.5-10 M cyclopiazonic acid, an inhibitor of the sarco/endoplasmic reticulum Ca 2؉ ATPase pump. The Ca 2؉ release induced by the addition of external Ca 2؉ was largely independent of IP 3 Rs because it was reduced by only ϳ30% by the inhibition of the inositol 1,4,5-trisphosphate receptors with caffeine or heparin. Measurements of Ca 2؉ -activated outward current and [Ca 2؉ ] i suggested that most CICR triggered by Ca 2؉ influx occurred away from the plasma membrane. Measurement of the response to several concentrations of cyclopiazonic acid revealed that Ca 2؉ influx that regulates CICR is associated with a selective portion of the internal Ca 2؉ pool. The minimal activation of Ca 2؉ influx by partial store depletion was confirmed by the measurement of Mn 2؉ influx. Inhibition of Ca 2؉ influx with SKF96365 or 2-aminoethoxydiphenyl borate prevented activation of CICR observed on addition of external Ca 2؉ . These findings provide evidence for activation of CICR by Ca 2؉

influx in non-excitable cells, demonstrate a previously unrecognized role for Ca 2؉ influx in triggering CICR, and indicate that CICR in non-excitable cells resembles CICR in cardiac myocytes with the exception that in cardiac cells Ca 2؉ influx is mediated by voltage-regulated Ca 2؉ channels whereas in non-excitable cells Ca 2؉ influx is mediated by storeoperated channels.
Agonist-evoked Ca 2ϩ signaling entails the generation of IP 3 1 to activate the IP 3 receptors (IP 3 Rs) and release the Ca 2ϩ stored in the endoplasmic reticulum (ER). Ca 2ϩ release from the ER is followed by activation of store-operated Ca 2ϩ channels (SOCs) in the plasma membrane and Ca 2ϩ influx into the cells. Subsequently, the plasma membrane Ca 2ϩ ATPase pump and the sarco/endoplasmic reticulum Ca 2ϩ ATPase pump remove Ca 2ϩ from the cytosol to stabilize [Ca 2ϩ ] i at a steady state that is determined by the relative activities of Ca 2ϩ influx and Ca 2ϩ extrusion (1,2). At physiological agonist concentrations, the Ca 2ϩ signal is in the form of repetitive Ca 2ϩ oscillations (1). In most cells, but particularly in polarized cells, the Ca 2ϩ signal initiates at discrete sites and proceeds as a propagating Ca 2ϩ wave (3)(4)(5)(6)(7). The various features and the proteins that mediate each phase of the Ca 2ϩ signal that depends on IP 3mediated Ca 2ϩ release have been extensively studied in many cell types and are reasonably well understood.
Another aspect of the Ca 2ϩ signal that is poorly understood compared with IP 3 -mediated Ca 2ϩ release is Ca 2ϩ -induced Ca 2ϩ release (CICR) in non-excitable cells. CICR is mediated by the ryanodine receptors (RyRs), the function of which is well characterized in skeletal and cardiac muscles (8) where they play a central role in excitation-contraction coupling. The family of RyRs includes three isoforms that are thought to be expressed in a cell-specific manner: RyR1 is expressed primarily in skeletal muscle, RyR2 is expressed primarily in cardiac muscle, and RyR3 is expressed ubiquitously (8). The primary activator of RyRs in muscle cells is Ca 2ϩ . However, the source of the trigger pool of Ca 2ϩ is cell-specific. In skeletal muscle, RyR1 directly interacts with the voltage-regulated L-type Ca 2ϩ channel. The L-type channel functions as a voltage sensor that causes membrane depolarization to undergo a conformational change and activate RyR1. The initial Ca 2ϩ released through RyR1 further activates RyR1 to result in the explosive Ca 2ϩ release that mediates muscle contraction (8,9). On the other hand, in cardiac muscle the L-type channel is not directly coupled to RyR2. Rather, the L-type channel is localized adjacent to RyR2. Membrane depolarization results in the activation of the L-type Ca 2ϩ channel and Ca 2ϩ influx. The incoming Ca 2ϩ serves as the trigger to activate RyR2 in the cardiac sarcoplasmic reticulum and cause the explosive Ca 2ϩ release and muscle contraction (8,10).
Accumulating molecular, biochemical, and functional evidence in recent years indicated that non-muscle cells (11), including secretory acinar cells (5,(12)(13)(14)(15), also express RyRs. However, their function and mechanism of regulation in nonmuscle cells remain elusive. There is good evidence that RyRs in non-muscle cells are activated by the nucleotide cADPR (16,17). In fact, activation of a Ca 2ϩ signal by cADPR generally is accepted as evidence for the involvement of RyRs in generating the Ca 2ϩ signal. Response to cADPR or the presence of RyRs was demonstrated in several secretory cells, including pancreatic acinar cells (12-15, 18, 19) and submandibular gland acinar and duct cells (15). The proposed main function of RyRs in these cells was to facilitate the propagation of the luminal-tobasal Ca 2ϩ wave (18 -21). All models to explain the role of RyRs in non-muscle cells depict them to be sensitized to Ca 2ϩ by cADPR but, notably, to be activated by Ca 2ϩ released from the IP 3 pool. No other pool of Ca 2ϩ is considered to participate in the regulation or activation of RyRs in non-muscle cells. The present work re-examined this premise in parotid acinar cells that have a robust CICR activity. We report that Ca 2ϩ influx through SOCs is critical for Ca 2ϩ release in parotid acinar cells and suggest that CICR in non-muscle cells occurs in a mechanism similar to that in cardiac myocytes, except that in cardiac myocytes Ca 2ϩ influx is mediated by the L-type Ca 2ϩ channel whereas in non-muscle cells Ca 2ϩ influx is mediated by SOCs.

MATERIALS AND METHODS
Materials-Cyclopiazonic acid and 1,4,5-and 2,4,5-IP 3 were purchased from Alexis. Fura-2/AM was from Tef Labs (Austin, TX). 2-aminoethoxydiphenyl borate and SKF96365 were from Calbiochem. Heparin was from Sigma. The standard bath solution A contained (in mM): 140 NaCl, 5 KCl, 1 MgCl 2 , 10 HEPES (pH 7.4 with NaOH), 10 glucose, and 0.1% bovine serum albumin. Ca 2ϩ concentration in this solution was adjusted between 0.5-7.5 mM by the addition of CaCl 2 . Ca 2ϩ -free solution was prepared by the addition of 0.2 mM EGTA to solution A. Solution A was supplemented with 10 mM sodium pyruvate, 0.02% soybean trypsin inhibitor, and 1 mg/ml bovine serum albumin to form the solution that was used for gland digestion and named PSA.
Preparation of Parotid Acini and Single Acinar Cells-Mouse parotid acini and single cells were prepared as described (22,23). In brief, to prepare acini the parotid glands were washed with PSA and minced finely with dissection scissors. The minced tissue was transferred to a flask and suspended in PSA containing 3 mg/ml collagenase P (Roche Applied Science). The flask was gassed with 100% O 2 and capped. After 2 min of digestion at 37°C, the tissue was partially dispersed by pipetteting with a 5-ml plastic pipette tip, and digestion continued for an additional 4 -5 min. The digested tissue was washed with PSA, filtered through a 75-m nylon mesh to remove the large clamps, and the acini were loaded with 5 M Fura-2 and kept on ice until use. To prepare single cells, mouse parotid glands were minced and washed once with Ca 2ϩ -free solution A. The minced tissue was incubated for 7 min at 37°C with a solution containing 0.025% trypsin that was prepared by a 1:1 dilution of a trypsin/EDTA solution from Sigma with phosphate-buffered saline. Trypsin treatment was terminated by washing the cell with PSA containing 1.5 mg/ml soybean trypsin inhibitor, and the tissue was further digested for 20 min with 70 units/ml collagenase CLSPA (Worthington). The cells were washed with PSA and kept on ice until used.
[Ca 2ϩ ] i Imaging-[Ca 2ϩ ] i was measured by imaging Fura-2-loaded parotid acini as detailed before (6,23). The Fura-2 fluorescence ratio was monitored at excitation wavelengths of 355 and 380 nm using a PTI image acquisition and analysis system. [Ca 2ϩ ] i was determined from calibration of the signals by incubating the cells in medium containing 10 mM EGTA and 5 M ionomycin and then medium containing ionomycin and 10 mM Ca 2ϩ as detailed before (4 -6). To measure Mn 2ϩ influx, cells in Ca 2ϩ -free medium were stimulated for 2.5 min and then exposed to medium containing the agonists and 0.5 mM Mn 2ϩ . Fura-2 fluorescence quench was measured at an excitation wavelength of 360 mm. Results are given as the mean Ϯ S.E. of the indicated number of experiments.
Electrophysiology-The whole cell configuration of the patch clamp technique was used to measure the Ca 2ϩ -activated Cl Ϫ or K ϩ current as a reporter of [Ca 2ϩ ] i (5). The Ca 2ϩ -activated Cl Ϫ current was recorded at a holding potential of Ϫ60 mV, whereas the Ca 2ϩ -activated K ϩ current was recorded at a holding potential of 0 mV. The bath solution for current recording was solution A, and the pipette solution contained (in mM): 140 KCl, 1 MgCl 2 , 0.2 EGTA, 5 ATP, 10 HEPES (pH 7.3 with KOH) with or without cADPR and Heparin. Seals of 6 -10 gigaohms were produced on the cell membrane, and the whole cell configuration was obtained by gentle suction or voltage pulses of 0.5 V for 0.3-1 ms. Recording started as soon as possible after establishing the whole cell configuration because the high concentrations of cADPR activated the current by rapidly releasing Ca 2ϩ from internal stores. The patch clamp output (Axopatch-1B, Axon Instruments) was filtered at 20 Hz. Record-ing and analysis were performed with patch clamp 6 and a Digi-Data 1200 interface (Axon Instruments). Recordings were at room temperature.

CICR in Parotid Acinar
Cells-CICR has been demonstrated in many non-muscle cells, yet regulation of this mode of Ca 2ϩ release is not fully understood. To study regulation of CICR, we selected the parotid acinar cells for two reasons. First, there is ample evidence for expression (12,15,25) and function (12, 15, 24 -26) of RyRs in these cells, or at least activation of Ca 2ϩ release by cADPR. Second, as illustrated in Fig. 1, cADPR is capable of releasing the entire agonist-mobilizable Ca 2ϩ pool. Thus, infusing parotid acinar cells with cADPR concentrations higher than 3 M tended to induce broad oscillatory changes in the Ca 2ϩ -activated Cl Ϫ current (Fig. 1, C and D), a reporter of [Ca 2ϩ ] i (5,6,18,19). At a concentration of 1 M cADPR induced either sporadic Ca 2ϩ oscillations (3 cells) or a single broad peak (2 cells). Stimulation of cells infused with 10 M or higher concentrations of cADPR with 1 mM carbachol resulted with a diminished response (4 cells). These results are similar to those found in submandibular acinar cells (5,27) and are notably different from those reported in pancreatic acinar cells by several groups (5, 7, 18 -20) in which cADPR was able to mobilize only a small fraction of the agonist-mobilizable pool. Hence, parotid acinar cells express RyRs mRNA and protein (12,15,25) and show robust response to cADPR ( Fig. 1) (12, 15, 24 -26) and are therefore suitable for studying the role of Ca 2ϩ influx in activation of CICR.
Another indication of robust CICR activity in parotid acinar cells was apparent from measurement of [Ca 2ϩ ] i . Fig. 1E shows that when cells bathed in medium containing 1 mM Ca 2ϩ were maximally stimulated with 1 mM carbachol there were two events of [Ca 2ϩ ] i increase, an abrupt immediate increase and a secondary increase marked by an arrow. Fig. 1F shows that removal of external Ca 2ϩ resulted in the reduction of [Ca 2ϩ ] i to basal levels. Re-addition of external Ca 2ϩ caused a rapid [Ca 2ϩ ] i increase with an overshoot. Such protocols generally are used to demonstrate the contribution of Ca 2ϩ influx to the agonist-evoked Ca 2ϩ signal. However, the rapid and large [Ca 2ϩ ] i increase seen on the addition of external Ca 2ϩ is not typical of Ca 2ϩ influx through SOCs (see below also). The abundance of RyRs in parotid acinar cells raised the possibility that the rapid increase in [Ca 2ϩ ] i is caused by activation of CICR.
Two Events of Ca 2ϩ Release-To clearly separate Ca 2ϩ release and influx we measured the effect of extracellular Ca 2ϩ on the Ca 2ϩ signal evoked by carbachol. Fig. 2A shows that the removal of external Ca 2ϩ reduced the extent and duration of the increase in [Ca 2ϩ ] i . In the absence of external Ca 2ϩ , maximal stimulation increased [Ca 2ϩ ] i from about 49 Ϯ 6 nM to only 186 Ϯ 21 nM (n ϭ 17). Subsequently, [Ca 2ϩ ] i was rapidly reduced to the basal level such that the total amount of Ca 2ϩ mobilized by the agonist (area under the curve) was reduced by nearly 90%. This indicates that in the absence of external Ca 2ϩ , and thus Ca 2ϩ influx, agonist stimulation mobilizes only a fraction of the intracellular Ca 2ϩ pool of parotid acinar cells. Ca 2ϩ entry then was initiated by the addition of various concentrations of CaCl 2 to the perfusion solution. Addition of 0.5 and 1.5 mM CaCl 2 was followed by a slow [Ca 2ϩ ] i increase. Once [Ca 2ϩ ] i reached a certain level, an abrupt change in the rate on [Ca 2ϩ ] i increase occurred (marked by an arrow), indicating a second event of Ca 2ϩ release. The secondary Ca 2ϩ release was very dramatic when the cells were exposed to an external CaCl 2 concentration of 7.5 mM. Under these conditions [Ca 2ϩ ] i increased to 435 Ϯ 39 nM, which is approximately three times the level evoked by the agonist in the absence of external Ca 2ϩ (n Ͼ 50 acini and Ͼ250 cells). Importantly, at the continued incubation with 7.5 mM Ca 2ϩ , the cells reduced [Ca 2ϩ ] i to stabilize it at a plateau of about 165 Ϯ 10 nM (n ϭ 6) above resting levels.
The response in Fig. 2A raised the question of whether 7.5 mM external Ca 2ϩ increased [Ca 2ϩ ] i primarily by the activation of CICR to release Ca 2ϩ from internal stores or whether parotid acinar cells have a very robust Ca 2ϩ influx that undergo partial desensitization/inactivation. This question can be addressed by complete depletion of the stores prior to the addition of external Ca 2ϩ . To do so, the cells were incubated with 100 M of the sarco/endoplasmic reticulum Ca 2ϩ ATPase pump inhibitor cyclopiazonic acid (CPA) for about 20 min. Fig. 2B shows that Acini perfused with Ca 2ϩ -free medium were stimulated with 1 mM carbachol (A and C) and were then perfused alternately with media containing 0.5, 1.5, 7.5, or 0 mM Ca 2ϩ (A) or alternately with media containing 7.5 or 0 mM Ca 2ϩ (C). The acini in B were incubated in Ca 2ϩ -free medium and were treated with 100 M CPA for about 20 min and then were exposed to a medium containing 100 M CPA and 7.5 mM Ca 2ϩ . addition of 7.5 mM external Ca 2ϩ to such treated cells only slowly increased [Ca 2ϩ ] i , and [Ca 2ϩ ] i stabilized at ϳ138 Ϯ 11 nM (n ϭ 5) above the resting level. In an additional protocol, cells stimulated with 1 mM carbachol were repeatedly and alternately exposed to 7.5 and 0 mM external CaCl 2 until no explosive increase in Ca 2ϩ was observed. Fig. 2C shows that the addition of 7.5 mM external Ca 2ϩ could elicit several events of rapid, transient [Ca 2ϩ ] i increases (2-3 in various experiments), after which the addition of 7.5 mM external Ca 2ϩ resulted in a slow [Ca 2ϩ ] i increase that by the fourth addition stabilized at 142 Ϯ 16 nM (n ϭ 8) above the resting level.
It is evident that the type of signal illustrated in Fig. 2 and described below is likely caused by activation of CICR by Ca 2ϩ entering the cells through the plasma membrane. This can be by activation of the RyRs and perhaps the IP 3 Rs because both are activated by Ca 2ϩ (see below). Well characterized pharmacological agents that modify the activity of the RyRs in muscle cells are caffeine and ryanodine. However, in agreement with previous work in parotid acinar cells (26,28,29), we were unable to see any stimulation of Ca 2ϩ release by using up to 20 mM caffeine and inhibition or stimulation of CICR by treating the cells with up to 250 M ryanodine. Although previous work reported inhibition of cADPR-mediated Ca 2ϩ release by ryanodine in parotid cells, all the effects were observed with micro-somes (29,30). Therefore, to determine the nature of the Ca 2ϩ signal evoked by 7.5 mM external Ca 2ϩ , we proceeded to determine the role of Ca 2ϩ release and Ca 2ϩ influx in this signal and then examined the contribution of RyRs and IP 3 Rs to the observed signal.
Minimal Store Depletion Is Sufficient to Activate a Large Ca 2ϩ Release-The next issue we addressed is the relationship between the extent of cell stimulation, Ca 2ϩ release from internal stores, and the explosive [Ca 2ϩ ] i increase evoked by 7.5 mM external Ca 2ϩ . For this, parotid acini were incubated in a Ca 2ϩ -free medium and stimulated with carbachol concentrations ranging from 0.1 M to 1 mM. Two min later, the cells were exposed to media containing the same carbachol concentrations and 7.5 mM Ca 2ϩ . Remarkably, stimulating the cells with only 0.5 M carbachol was sufficient to evoke the same [Ca 2ϩ ] i increase by 7.5 mM external Ca 2ϩ as that observed when the cells were stimulated with 1 mM carbachol (Fig. 3A). The concentration of carbachol sufficient to trigger the explosive [Ca 2ϩ ] i increase by high external Ca 2ϩ varied somewhat between experiments. When the cells were stimulated with 0.1, 0.5, and 1.5 M carbachol, 7.5 mM Ca 2ϩ induced the explosive [Ca 2ϩ ] i increase in 1/9, 7/13 and 25/25 acini, respectively, and was the same as that induced by stimulated with 1 mM carbachol in the same experiments. The minimal agonist stimulation needed to activate Ca 2ϩ release by 7.5 mM external CaCl 2 indicates that either minimal cell stimulation or minimal depletion of the internal Ca 2ϩ pool is sufficient to activate CICR. To distinguish between these possibilities and to examine the possible role of SOCs in allowing the effect of external CaCl 2 , we tested the ability of treatment with a low concentration of CPA to activate the Ca 2ϩ release by high external CaCl 2 . The results of these experiments are shown in Fig. 3B. For these experiments we selected 1 M CPA, which only partially inhibits the sarco/endoplasmic reticulum Ca 2ϩ ATPase pumps to slowly deplete the ER Ca 2ϩ store. Incubating parotid acini in Ca 2ϩ -free medium containing 1 M CPA between 2-10 min resulted in a progressively higher increase in [Ca 2ϩ ] i on the addition of 7.5 mM external Ca 2ϩ . Hence, the explosive Ca 2ϩ release observed on the addition of external Ca 2ϩ is not dependent on cell stimulation but, rather, on the Ca 2ϩ release induced by cell stimulation. Furthermore, minimal depletion of the stores by either cell stimulation or treatment with CPA is sufficient to allow the Ca 2ϩ release evoked by addition of external Ca 2ϩ .
It was of interest to determine the extent of store depletion needed to activate the pathway that mediates the effect of high external Ca 2ϩ . This can be estimated fairly well by treating the cells with CPA. First, we measured the ability of incubation with different concentrations of CPA at a set period of time to activate the Ca 2ϩ release by 7.5 mM external Ca 2ϩ . Fig. 4A shows that treatment with CPA was as effective as stimulation with carbachol in allowing the Ca 2ϩ release observed on the addition of high external Ca 2ϩ . In addition, incubation with 100 M CPA for a short period of time caused only a small [Ca 2ϩ ] i increase in cells incubated in Ca 2ϩ -free media. Furthermore, incubating acini in Ca 2ϩ -free media with 2.5 M CPA for 2 min was sufficient to allow the explosive Ca 2ϩ release caused by 7.5 mM external Ca 2ϩ , an effect that was observed in all acini treated with 2.5 M CPA for 2 min or longer (n ϭ 45 acini, Ͼ180 cells). Fig. 4B shows how the extent of store depletion by this and higher concentrations of CPA was estimated. The acini were incubated for 2 min with 2.5, 10, or 100 M CPA and then the stores were completely discharged by exposing the cells to CPA and 1 mM carbachol. It is evident that during the 2-min incubation all concentrations of CPA mobilized only a small fraction of the pool. Furthermore, from the reduced response to carbachol plus CPA, we determined that mobilization of less than 5% of the pool was sufficient to maximally activate the pathway that mediated the effect of 7.5 mM external Ca 2ϩ .
Contribution of IP 3 Rs and RyRs to CICR-Two processes can contribute to the CICR. Both the RyRs (9, 10) and IP 3 Rs (31) are activated by Ca 2ϩ with the type 3 IP 3 Rs particularly sensitive to Ca 2ϩ concentrations between 25-250 nM (31). To distinguish between the contributions of each process, we used inhibitors of the IP 3 Rs because the pharmacological agents that modify the activity of the RyRs in muscle cells had no effect on CICR in parotid acini. The lack of activation by caffeine provided us with the first compound because caffeine was shown to inhibit the IP 3 Rs in a reversible manner (7,32). This is further shown in Fig. 5A for parotid acini stimulated with carbachol. In preliminary experiments we determined the minimal concentration of caffeine required to inhibit Ca 2ϩ release evoked by 15 M carbachol to minimize nonspecific effect of caffeine, which is a potent inhibitor of IP 3 production in secretory cells (33). Fig. 5A shows that when added before cell stimulation 2.5 mM caffeine prevented the carbachol-induced Ca 2ϩ release and the explosive Ca 2ϩ increase triggered by 7.5 mM external Ca 2ϩ . Washing away the caffeine reversed the inhibition although exposure to 7.5 mM external Ca 2ϩ resulted in a slower rate of Ca 2ϩ increase. As shown in Fig. 5B, the cells were first stimulated with carbachol to deplete the stores and activate Ca 2ϩ influx and then were exposed to 2.5 mM caffeine. Although caffeine reduced the rate, it did not prevent the Ca 2ϩ increase in response to 7.5 mM external Ca 2ϩ . The slow rate of Ca 2ϩ increase in Fig. 5, A and B may suggest the contribution of IP 3 Rs to CICR or the effects of caffeine on the stimulated state (33). This was examined by testing the effect of caffeine on the response observed in cells treated with CPA because CPA depletes the stores in an IP 3 -independent manner. Incubating the cells with 2.5 mM caffeine had no effect on the Ca 2ϩ increase evoked by 7.5 mM external Ca 2ϩ in CPA-treated cells. Fig. 5C shows that 10 mM caffeine reduced the response to 7.5 mM external Ca 2ϩ by about 26 Ϯ 5% (n ϭ 5) in acini treated with 10 M CPA.
Another inhibitor of the IP 3 Rs is heparin (34). To test the effect of heparin it was infused into the cells through a patch pipette, and the activity of the Ca 2ϩ -activated K ϩ channel was measured (Fig. 5, D and E). Exposing non-stimulated cells to 7.5 mM external Ca 2ϩ had minimal effect on the current. Treating the cells with 10 M CPA in Ca 2ϩ -free media failed to activate any current, and the addition of 7.5 mM external Ca 2ϩ resulted in an abrupt but modest increase in the outward current (Fig. 5D). Infusing the cells with 200 g/ml heparin for 7-8 min completely inhibited the response to carbachol stimulation (not shown) but reduced the response to 7.5 mM external Ca 2ϩ in CPA-treated cells by only 33 Ϯ 5% (n ϭ 4). Together, the results in Fig. 5 indicate that the majority of the FIG. 4. Relationship between store depletion by CPA and CICR. A, acini in Ca 2ϩ -free medium were treated for 2 min with 0 (solid gray trace, control), 1 (dotted gray trace), 2.5 (solid black trace), 10 (dotted black trace), or 100 M CPA (dashed black trace) before exposure to media containing 7.5 mM CaCl 2 and then to Ca 2ϩ -free media. B, to estimate the extent of store depletion, acini in Ca 2ϩ -free media were treated for 2 min with 2.5 (dotted black trace), 10 (solid black trace), or 100 M CPA (solid gray trace) before exposure to Ca 2ϩ -free media containing 100 M CPA and 1 mM carbachol. After reduction of [Ca 2ϩ ] i to basal levels, the acini were exposed to media containing 7.5 and then 0 mM Ca 2ϩ .
CICR was independent of Ca 2ϩ release mediated by the IP 3 Rs. The modest effects of caffeine and heparin can be caused by the nonspecific effect of the blockers or a small Ca 2ϩ release from the IP 3 pool that required sensitization of the IP 3 Rs by Ca 2ϩ .
Site of CICR-The modest activation of K ϩ current in response to the addition of 7.5 mM external Ca 2ϩ to cells treated with CPA (Fig. 5) was unexpected in view of the robust Ca 2ϩ increase caused by the addition of 7.5 mM external Ca 2ϩ (Figs.  3 and 4). At least two mechanisms can explain this discrepancy. The first possibility is that the current was measured in single cells whereas the [Ca 2ϩ ] i was measured in acini. It is possible that the ratio of [Ca 2ϩ ] i increase in the presence and absence of external Ca 2ϩ is different in a single cell than in acini. Second, it is possible that the robust [Ca 2ϩ ] i increase caused by CICR was slower than the initial Ca 2ϩ release evoked by carbachol and occurs largely away from the plasma membrane. To distinguish between these possibilities, we measured [Ca 2ϩ ] i and the Ca 2ϩ -activated K ϩ current in single cells stimulated with carbachol. Fig. 6A shows that stimulation of cells incubated in Ca 2ϩ -free media with 2.5 M or 1 mM carbachol resulted in a rapid and a large increase in the current. Noticeably, exposing the cells to 7.5 mM external Ca 2ϩ resulted in a small increase in the current. Similar results were obtained with 4 cell preparations (and at least 8 experiments under each condition) and additional experiments with different carbachol concentrations between 1-100 M. Furthermore, essentially the same results were obtained by measuring the Ca 2ϩ -activated Cl Ϫ current, except that the response to 7.5 mM external Ca 2ϩ was even smaller relative to the initial response in the absence of external Ca 2ϩ .
An example of the Ca 2ϩ signal observed in single parotid acinar cells is shown in Fig. 6B. Carbachol stimulation of cells in Ca 2ϩ -free media resulted in a luminal-to-basal Ca 2ϩ wave (Fig. 6B, top panel). As reported before (26,35), the agonistevoked Ca 2ϩ wave in parotid acini was faster than that in pancreatic acinar cells with the maximal increase in [Ca 2ϩ ] i attained within 0.8 s of stimulation. Subsequent exposure of the cells to media containing 7.5 mM Ca 2ϩ resulted in a slower but higher [Ca 2ϩ ] i increase (Fig. 6, traces b and c). In six experiments the [Ca 2ϩ ] i increase induced by 7.5 mM external Ca 2ϩ was 2.56-Ϯ 0.36-fold higher than the initial increase measured in the absence of external Ca 2ϩ . Interestingly, the Ca 2ϩ release evoked by 7.5 mM external Ca 2ϩ also tended to start at the apical pool. However, it did not occur as a propagating luminal-to-basal Ca 2ϩ wave. Rather, the rate of [Ca 2ϩ ] i increase was slow and occurred in other parts of the cell shortly after the initial increase at the apical pole. This suggests that Ca 2ϩ influx starts at the apical pole but takes place in all regions of the basolateral membrane and that Ca 2ϩ entering the cell at all sites can trigger CICR, including Ca 2ϩ entering at the IP 3 Rs-poor basal pole. The combined results in Fig. 6 are consistent with the idea that the Ca 2ϩ increase caused by CICR occurs mainly at sites far from the plasma membrane such that it does not activate the Cl Ϫ and K ϩ channels. Dissociation between a detectable [Ca 2ϩ ] i increase and activation of the K ϩ channel was reported before (36) and was attributed to a rapid initial agonist-stimulated Ca 2ϩ increase next to the plasma membrane.
Ca 2ϩ Influx through SOCs Is Essential for CICR Triggered by Addition of External Ca 2ϩ -The need to stimulate the cells with carbachol or to treat them with CPA in order to observe the Ca 2ϩ release evoked by the addition of external Ca 2ϩ suggests the activation of Ca 2ϩ influx by the agonist and CPA. Obviously, the extent of Ca 2ϩ influx activated by partial store depletion could not be determined from Ca 2ϩ removal and re-addition protocols because of the activation of the CICR. An alternative procedure to estimate the activity of SOCs is to measure the unidirectional influx of Mn 2ϩ . Most SOCs channels admit Mn 2ϩ , and Mn 2ϩ is not a substrate for the sarco/ endoplasmic reticulum Ca 2ϩ ATPase and plasma membrane Ca 2ϩ ATPase pumps so that Mn 2ϩ influx accurately reports the plasma membrane permeability to divalent ions (37). The results of Mn 2ϩ influx measurement are given in Fig. 7. Maximal store depletion by treating the cells with 1 mM carbachol and 25 M CPA increased Mn 2ϩ influx ϳ5-fold. Stimulation with 0.5 M carbachol or treatment with 2.5 M CPA for 2.5 min increased Mn 2ϩ influx by only 7 Ϯ 3% (n ϭ 3) -fold, highlighting the fact that selective activation of Ca 2ϩ influx is required to activate the CICR.
If Ca 2ϩ influx is obligatory for the Ca 2ϩ release evoked by the addition of high external Ca 2ϩ , then inhibition of the influx should inhibit the Ca 2ϩ release. This was tested using two SOCs blockers, SKF96365 and 2-aminoethoxydiphenyl borate. Fig. 8A shows that incubating the cells with 20 and 40 M FIG. 5. CICR does not require active IP 3 Rs. A and B, acini in Ca 2ϩ -free medium were incubated with 2.5 mM caffeine before (A) or after (B) stimulation with 15 M carbachol and then alternately in Ca 2ϩ -free media and media containing 7.5 mM Ca 2ϩ . Where indicated, caffeine was removed from the perfusion solutions. C, acini were treated with 10 M CPA for 2 min before incubation with 10 mM caffeine in media containing 0 or 7.5 mM Ca 2ϩ , as indicated. D and E, the Ca 2ϩactivated K ϩ current was measured in single cells. D, the cells were exposed to media containing 7.5 Ca 2ϩ before incubation in Ca 2ϩ -free media containing 10 M CPA and then were exposed to media containing CPA and 7.5 mM Ca 2ϩ . E, the pipette solution contained 200 g/ml heparin. Treatment with CPA began shortly after break-in and the cells were exposed to 7.5 mM external Ca 2ϩ as indicated by the bar.
SKF96365 partially or maximally, respectively, inhibited the [Ca 2ϩ ] i increase observed on the addition of external CaCl 2 and that at both concentrations the inhibition was reversible. 2APB was reported to inhibit both IP 3 -mediated Ca 2ϩ release and Ca 2ϩ influx by SOCs (38) but not CICR by RyRs (39). Using the skeletal muscle cell line C2C12, we confirmed that 2ABP does not inhibit Ca 2ϩ release evoked by caffeine or by depolarization, whereas it inhibited the Ca 2ϩ signal evoked by stimulation of the P2Y2 receptors. 2 As with SKF96365, incubating the cells with 20 and 40 M 2APB partially or maximally inhibited the [Ca 2ϩ ] i increase triggered by the addition of 7.5 mM external Ca 2ϩ , indicating that Ca 2ϩ influx and [Ca 2ϩ ] i increase are essential to activate the explosive increase in [Ca 2ϩ ] i that is caused by CICR. DISCUSSION The widespread expression of RyRs in non-muscle cells (11)(12)(13)(14)(15) indicates that CICR by activation of RyRs is central to agonist-evoked Ca 2ϩ signaling. Furthermore, selective activation of CICR by G protein-coupled receptors appears to be part of the regulatory repertoire (7) that is used to generate receptor-specific Ca 2ϩ signals (2). Therefore, it is important to understand the activation mechanism of CICR and its regulation in non-muscle cells. Whereas the RyRs mediating CICR in skeletal (RyR1) and cardiac (RyR2) cells are well established (8 -10), it is still a matter of debate which RyRs are expressed in non-muscle cells and even whether they mediate the CICR. For example, using the same molecular and biochemical approaches, it was reported that pancreatic acinar cells express RyR2 but not RyR1 and RyR3 (14) or all three RyRs isoforms (13). Similarly, parotid acinar cells were reported to express 2 K. Kiselyov and S. Muallem, unpublished observation. only an RyR1 isoform (25) or only RyR3 (15). Knowledge of the RyR isoforms and their role in Ca 2ϩ signaling is equally poor in other non-muscle cells, which makes it impossible to know with certainty the molecular identity of the protein mediating CICR in non-muscle cells.
The pharmacology of CICR in non-muscle cells is equally confusing and is the subject of conflicting reports. Although caffeine activates RyRs in muscle and chromaffin cells, most studies failed to show caffeine-mediated Ca 2ϩ release in nonmuscle cells that responded well to cADPR. Similarly, in the present work we were not able to see any effect of caffeine on [Ca 2ϩ ] i by itself. In fact caffeine is a good inhibitor of IP 3mediated Ca 2ϩ release (33) and was used in several studies to distinguish between Ca 2ϩ release mediated by IP 3 and by cADPR (7,32). In the present work we confirmed inhibition of agonist-evoked Ca 2ϩ release by caffeine and used this property to show that most of the CICR was not mediated by the IP 3 Rs (Fig. 5). Similar uncertainties exist with ryanodine. Most studies on parotid acini of inhibition by ryanodine of Ca 2ϩ release induced by cADPR used permeabilized cells or microsomes (24,29,30). One study reported inhibition of large Ca 2ϩ oscillations induced by thapsigargin by 10 -50 M ryanodine but failed to find inhibition of the response to the agonist (28). A recent report showed that 500 M ryanodine was required to inhibit the Ca 2ϩ oscillations riding on the top of a Ca 2ϩ plateau evoked by low carbachol concentration without affecting the plateau (26). We were unable to inhibit any part of the Ca 2ϩ signal observed under all protocols used in the present work. Therefore, we conclude that the CICR in parotid acinar cells is not very sensitive to either caffeine of ryanodine. However, we do argue that CICR is mediated by a channel separate from the IP 3 receptor channels because it was equally well activated by treatment with a low concentration of CPA and inhibited by only 30% by inhibition of the IP 3 Rs with caffeine or heparin. Whether it is mediated by a classical RyR remains to be established.
A consensus does exist of activation of CICR by cADPR in non-muscle cells and that cADPR is a ligand that directly or indirectly activates the RyRs (16). Based on the response to cADPR, many cell types, including parotid acinar cells, do have a CICR mechanism. Furthermore, it appears that at least in parotid acinar cells, CICR plays a prominent role in agonistevoked Ca 2ϩ mobilization. Thus, in the absence of external Ca 2ϩ , the agonist-evoked [Ca 2ϩ ] i increase is markedly diminished and is short lasting (Figs. 1-3). This was largely caused by the elimination of CICR because maximal Ca 2ϩ influx through SOCs account for only a small fraction of the agonistevoked [Ca 2ϩ ] i increase measured in the presence of external Ca 2ϩ (Fig. 2). Thus, the majority of Ca 2ϩ release in parotid acini is caused by CICR. The robustness of CICR in parotid acinar cells also is evident from the effect of cADPR. In most experiments 10 M cADPR mobilized almost the entire intracellular Ca 2ϩ pool as concluded from the small effect of carbachol on Ca 2ϩ -activated Cl Ϫ current in cells incubated with 10 M cADPR (Fig. 1C). Similar findings were reported in submandibular acinar cells by us (5) and others (27), although higher concentrations of cADPR were needed in the study of Harmer et al. (27). The potent CICR mechanism in parotid acini may be of particular physiological significance in these cells. Analysis of Ca 2ϩ waves in pancreatic and parotid acini revealed that the threshold Ca 2ϩ release that triggers a propagating wave is much lower in parotid than in pancreatic acinar cells and that at modest stimulation the wave propagated about 8 times faster in parotid than in pancreatic acinar FIG. 7. Measurement of Mn 2؉ influx. Acini in Ca 2ϩ -free medium were incubated for 2.5 min with 0 (control), 2.5 M CPA, 0.5 M carbachol, or 25 M CPA and 1 mM carbachol (Car), as indicated next to each trace, before exposure to Ca 2ϩ -free medium containing the same agonists and 0.5 mM Mn 2ϩ to measure Mn 2ϩ influx. The first derivatives of the slopes were used to obtain the rate of Mn 2ϩ influx. cells (35). The increased sensitivity to stimulus intensity was attributed to a 4-fold higher expression of IP 3 Rs (35). The prominent CICR likely is the mechanism behind the faster Ca 2ϩ wave propagation in parotid acinar cells.
The most notable finding of the present work is that Ca 2ϩ influx through SOCs plays a critical role in the activation of CICR in parotid acinar cells. This conclusion is supported by several findings: (a) removal of external Ca 2ϩ markedly reduced the amplitude and the amount of Ca 2ϩ mobilized in response to agonist stimulation, an effect that is not simply caused by removal of the contribution of Ca 2ϩ influx to the [Ca 2ϩ ] i increase (Figs. 1 and 2); (b) activation of SOCs by receptor-independent depletion of the stores with CPA was sufficient to activate CICR; (c) inhibition of Ca 2ϩ influx through SOCs by SKF96365 or 2APB inhibited the activation of CICR by the addition of high external Ca 2ϩ . Although these blockers are not specific and have additional effects beside the inhibition of SOCs, they were shown to inhibit SOCs in many cell types. Because most CICR is independent of IP 3 Rs, 2-aminoethoxydiphenyl borate does not inhibit CICR by RyRs, and the major acute effect of SKF9636 is inhibition of Ca 2ϩ influx, the results with the blockers are consistent with a critical role of Ca 2ϩ influx through SOCs for activation of CICR. Together, our results indicate that Ca 2ϩ influx through SOCs activates the CICR mechanism to mediate most of the [Ca 2ϩ ] i increase observed during agonist stimulation of parotid acinar cells.
It is of note that although Ca 2ϩ release from the IP 3 pool (measured in cells stimulated in the absence of external Ca 2ϩ ) increases [Ca 2ϩ ] i to higher levels than Ca 2ϩ influx through SOCs (for example, see Fig. 2, A and C), it was not sufficient to activate CICR. This indicates that not "all [Ca 2ϩ ] i is the same" and Ca 2ϩ influx through SOCs specifically activates CICR. This conclusion received further support with the finding that Ca 2ϩ release from the IP 3 pool strongly activated the Ca 2ϩactivated Cl Ϫ and K ϩ channels, whereas CICR caused by subsequent Ca 2ϩ influx increased [Ca 2ϩ ] i about 2.5-fold higher than Ca 2ϩ release from the IP 3 pool but only modestly activated the channels. These findings suggest that most CICR occurred away from the plasma membrane and further emphasis the tight association between Ca 2ϩ influx and CICR. Compartmentalization of Ca 2ϩ release events in parotid acini were reported before as a lag in detection in Ca 2ϩ increase relative to detection of the Ca 2ϩ -activated K ϩ current (36).
An important feature of the Ca 2ϩ influx that activates CICR is that it is mediated by a selective subpopulation of SOCs channel. Thus, activation of at most 5% of maximal SOCs activity was sufficient to mediate the Ca 2ϩ influx needed to activate CICR. This population of SOCs channels was regulated by a small portion of the ER Ca 2ϩ pool. These findings are similar to two previous reports showing that a selective 5% of the IP 3 pool governs regulation of I crac (40,41). Such a relationship between Ca 2ϩ and the SOCs and CICR channels requires that both channels exist in close proximity in cellular microdomains. Such an arrangement is well established for Ca 2ϩ signaling complexes in many cell types including secretory cells (2,6) and is the hallmark of excitation-contraction coupling (8 -10). Similar to the situation in muscle, it is likely that only a subpopulation of the CICR channels is coupled to SOCs and that Ca 2ϩ released from these stores then serves to activate CICR in adjacent stores and to propagate a Ca 2ϩ wave.
In accord with such a mechanism, cADPR-mediated Ca 2ϩ release appears to be essential for the propagation of Ca 2ϩ waves (19 -21, 42). The properties of CICR as demonstrated here indicate that CICR in cardiac myocytes is a specialized mech-anism of a more general mechanism found in non-muscle cells. That is, regulation of CICR by a selective population of SOCs channels and the obligatory dependence of CICR on Ca 2ϩ influx through SOCs is equivalent to CICR in cardiac myocytes with the exception that in cardiac myocytes CICR is activated by Ca 2ϩ influx through L-type Ca 2ϩ channels whereas in parotid acinar cells Ca 2ϩ influx occurs through SOCs. If this finding can be extended to other non-muscle cells, it will generalize the mechanism of regulation of CICR and highlight the role of SOCs in shaping the agonist-evoked Ca 2ϩ signal.