Ca2+-dependent inactivation of a store-operated Ca2+ current in human submandibular gland cells. Role of a staurosporine-sensitive protein kinase and the intracellular Ca2+ pump.

Stimulation of human submandibular gland cells with carbachol, inositol trisphosphate (IP3), thapsigargin, or tert-butylhydroxyquinone induced an inward current that was sensitive to external Ca2+ concentration ([Ca2+]e) and was also carried by external Na+ or Ba2+ (in a Ca2+-free medium) with amplitudes in the order Ca2+ > Ba2+ > Na+. All cation currents were blocked by La3+ and Gd3+ but not by Zn2+. The IP3-stimulated current with 10 microM 3-deoxy-3-fluoro-D-myo-inositol 1,4,5-triphosphate and 10 mM 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid in the pipette solution, showed 50% inactivation in <5 min and >5 min with 10 and 1 mM [Ca2+]e, respectively. The Na+ current was not inactivated, whereas the Ba2+ current inactivated at a slower rate. The protein kinase inhibitor, staurosporine, delayed the inactivation and increased the amplitude of the current, whereas the protein Ser/Thr phosphatase inhibitor, calyculin A, reduced the current. Thapsigargin- and tert-butylhydroxyquinone-stimulated Ca2+ currents inactivated faster. Importantly, these agents accelerated the inactivation of the IP3-stimulated current. The data demonstrate that internal Ca2+ store depletion-activated Ca2+ current (ISOC) in this salivary cell line is regulated by a Ca2+-dependent feedback mechanism involving a staurosporine-sensitive protein kinase and the intracellular Ca2+ pump. We suggest that the Ca2+ pump modulates ISOC by regulating [Ca2+]i in the region of Ca2+ influx.

In non-excitable cells, such as salivary gland cells, activation of Ca 2ϩ -mobilizing receptors, e.g. muscarinic-cholinergic, on the cell surface produces a biphasic increase in cytosolic [Ca 2ϩ ] ([Ca 2ϩ ] i ), 1 with an initial phase due to Ca 2ϩ release from internal Ca 2ϩ stores via IP 3 -dependent Ca 2ϩ channels in the intracellular membrane and a second phase due to Ca 2ϩ influx across the plasma membrane. Although the molecule(s) that mediates Ca 2ϩ influx across the plasma membrane has not yet been identified, there is conclusive evidence to demonstrate that it is regulated by the [Ca 2ϩ ] in the intracellular Ca 2ϩ store(s). This Ca 2ϩ influx has been termed capacitative, or store-operated Ca 2ϩ entry (SOCE) (1)(2)(3)(4). Recently, methods have been developed to measure directly the Ca 2ϩ influx current by using the whole cell patch clamp technique in the presence of high external [Ca 2ϩ ] ([Ca 2ϩ ] e ) and high internal [Ca 2ϩ ] buffer (5)(6)(7)(8)(9). Activation of inward Ca 2ϩ currents by store depletion has been reported in a number of different cell types (4,5), and it has been suggested that this current is mediated by a channel. However, the current appears to vary in different cell types. The most well studied of these currents is I CRAC (Ca 2ϩ release-activated Ca 2ϩ current) which has some unique characteristics, including very low single channel conductance, high Ca 2ϩ selectivity, inward rectification, reversal at very positive potentials, and Ca 2ϩ -dependent inactivation (5)(6)(7)(8)(9).
The precise mechanism(s) that activates and inactivates SOCE is not yet known. Two main mechanisms have been proposed for the activation. (i) The status of the internal Ca 2ϩ store is relayed to the plasma membrane via a physical interaction between the Ca 2ϩ store membrane and the plasma membrane. (ii) A diffusible messenger either released from the store during depletion or activated in response to depletion transmits the signal from ER to the plasma membrane. Two different inactivation mechanisms have been described that are either dependent on the refilling internal Ca 2ϩ store or independent of store refilling. We have previously shown that refilling of internal Ca 2ϩ stores in salivary gland cells inactivates agonist-stimulated Ca 2ϩ and Mn 2ϩ influx (10). Ca 2ϩ -dependent inactivation of Ca 2ϩ influx which is independent of internal Ca 2ϩ store has also been described by us and others in parotid acinar cells (11)(12)(13) and also for I CRAC in T-lymphocytes (8,9). We and others (12,14) have reported that in rat parotid acinar cells Ca 2ϩ influx is inactivated by the protein Ser/Thr phosphatase inhibitors, okadaic acid and calyculin A, and pretreatment of cells with staurosporine, a relatively nonspecific protein kinase inhibitor, prevents the inactivation. We have recently reported (14) that the staurosporine-sensitive protein kinase is involved in the Ca 2ϩ -induced feedback inhibition of Ca 2ϩ influx in rat parotid acinar cells. Importantly, this kinase does not appear to be involved in the refill-dependent inactivation of SOCE, i.e. refill-dependent inactivation was not prevented by staurosporine (14). Thus, we had hypothesized that Ca 2ϩ , via a protein kinase, modulates SOCE in store-depleted rat parotid acinar cells.
HSG cells, a human submandibular gland cell line, have been used to study muscarinic receptor-activated Ca 2ϩ signaling and Ca 2ϩ influx (15)(16)(17). We have demonstrated that emptying of the Ca 2ϩ store by CCh, IP 3 , or the Ca 2ϩ pump inhibitor TG (or BHQ) activates an SOCE-type Ca 2ϩ influx mechanism * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Bldg. 10 in these cells (17), which was previously monitored by measuring fura2 fluorescence or the Ca 2ϩ -dependent K ϩ current (K Ca ) (16,17). To understand further the mechanism of Ca 2ϩ influx in HSG cells, we have now directly measured the storeoperated Ca 2ϩ current. We report here the activation of an inward Ca 2ϩ current by internal Ca 2ϩ store depletion (I SOC ) in HSG cells. This is the first report of a Ca 2ϩ influx current in a salivary gland cell. I SOC is Ca 2ϩ -selective and is regulated by a Ca 2ϩ -dependent feedback inhibition, which is determined by the [Ca 2ϩ ] i in the region of the channel and a protein kinase activity. By using the K Ca current as a readout for Ca 2ϩ influx, we had previously reported that in HSG cells, [Ca 2ϩ ] i at the site of Ca 2ϩ influx is maintained at a very low level (i.e. below that required for activation of K Ca ) by the intracellular Ca 2ϩ pump (SERCA) activity. The present data demonstrate that SERCA modulates the Ca 2ϩ -dependent feedback inhibition of the store-operated Ca 2ϩ influx channel.

EXPERIMENTAL PROCEDURES
Cell Culture-HSG cells were cultured as described earlier (15)(16)(17). The cells were detached from the tissue culture dish with 0.25% trypsin, 1.0 mM EDTA (Biofluids). A single cell suspension was seeded on coverslips, kept in a 35-mm culture dish (Corning, NY), and used after 24 h. All reagents were of the highest available grade.
Patch Clamp Experiments-Patch clamp in a whole cell configuration was performed on single HSG cells attached to coverslips (17,18). Patch electrodes were made from 1.0-mm borosilicate glass tubing with filament (BF-100 -50-10, Sutter Instrument Co., Novato, CA). The resistance of the pipette was typically between 3 and 6 M⍀ when filled. The chamber was connected with an Ag-AgCl pellet through a 150 mM NaCl-containing agar bridge. Cell membrane and pipette capacitative transients were subtracted from the records by the amplifier circuitry before sampling. Voltages have not been compensated for liquid junction potentials. Membrane currents were measured with an Axopatch 200A amplifier in conjunction with pClamp 6.1 software and a Digidata 1200 A/D converter (Axon Instruments, Foster City, CA). The currents were filtered at 2 kHz (low-pass bessel filter) and sampled at an interval of 10 ms. In the step protocol, the cell was held at 0 mV and stepped from Ϫ100 mV to Ϫ40 mV in 20 mV steps. The cells were held at each step for 200 ms and at the holding potential for 312 ms. In the ramp protocol, the holding potential was ϩ35 mV, and during the ramp the membrane potential was changed from Ϫ120 mV to ϩ60 mV within 500 ms. The leakage subtraction function was used with either ramp or step protocol to minimize the leak currents. The currents were digitized and recorded directly onto the hard drive of a Dell Pentium computer. The peak currents measured in the ramp, step, and gap-free protocol were used for analysis. The I-V relationship was calculated using the peak amplitude of the current during the ramp protocol and exported to the Origin 4.1 (Microcal Software, Inc., Northampton, MA) for further analysis.
Coverslips with cells were cut to an approximately 0.5 ϫ 0.5-mm piece that was placed in a chamber (Warner Instrument Corp., Hamden, CT) that was perfused at a rate of approximately 5 ml/min while the bath solution was continuously removed by a vacuum line. Complete solution changes were achieved within 10 s. The standard extracellular solution contained (mM) 135 sodium glutamate, 1 MgCl 2 , 10 CaCl 2 , 10 glucose, and 10 HEPES, pH 7.4 (NaOH). The pipette was filled with (mM) 135 NMDG glutamate, 10 CsCl, 10 BAPTA, 1 MgCl 2 , 1 ATP, 10 HEPES, pH 7.2 (CsOH). 10 M inositol 1,4,5,trisphosphate (IP 3 ), or its non-metabolizable analog F-IP3, was included in the pipette solution (all experiments shown here were performed with F-IP 3 ). 10 mM CaCl 2 was replaced with 1 mM CaCl 2 , 10 mM BaCl 2 or Ca 2ϩ -free medium as indicated in some experiments. Tg, BHQ, or CCh was continuously administered in the cell medium by perfusion. All experiments were performed at room temperature.
[Ca 2ϩ ] i Measurements-[Ca 2ϩ ] i in HSG cells was determined by measuring fura2 fluorescence as described previously (12,14,16). All other experimental conditions are described in the text.
Data Analysis-All the values given below are the mean Ϯ S.E. of the indicated number of experiments. The Student's t test or analysis of variance test, as indicated, was used for statistical analysis of the data.

RESULTS
Ca 2ϩ Store Depletion-activated Inward Ca 2ϩ Current in HSG Cells-To measure directly SOCE in HSG cells, we have used the whole cell patch clamp protocol described by Hoth and Penner (6), with some modifications (see "Experimental Procedures"). The fast Ca 2ϩ chelator BAPTA (10 mM) was included in the patch pipette and high [Ca 2ϩ ] (10 mM) in the external medium. The inclusion of BAPTA into the cell has been shown to prevent intracellular Ca 2ϩ -dependent feedback inhibition of the inward Ca 2ϩ current, and thus facilitate detection of this current (6,7). This was also true for HSG cells. The inward current was not detected when BAPTA was omitted from the pipette solution or low concentrations of EGTA, 0.5-1.0 mM, were used (data not shown).
Stimulation of HSG cells by CCh, or direct activation of internal Ca 2ϩ release by IP 3 , caused activation of an inward current ( Fig. 1, which shows a continuous recording of the current at 0 mV using the gap-free configuration of pClamp 6.0. The current recording was initiated as soon as the whole cell configuration was established. Consistent with previous I CRAC measurements (6,7), the IP 3 -induced current had a shorter onset (15.3 Ϯ 5.4 s, n ϭ 7) of activation as compared with that of the CCh-stimulated current (30.6 Ϯ 7.3 s, n ϭ 4, including the perfusion time, Ͻ10 s). We have shown previously that in rat parotid acinar cells peak [Ca 2ϩ ] i due to internal Ca 2ϩ release is reached within 6 s after CCh stimulation of cells, whereas Ca 2ϩ influx is detected after a lag of about 10 s (19). Similar lag times have also been reported in pancreatic acinar cell (20). IP 3 -stimulated inward currents were detected in 70.2% (52/74) of the cells tested while CCh-stimulated currents were detected in 53.8% (7/13) of the cells. The maximum steady state current activated by IP 3 at 0 mV in the continuous recording protocol had an average amplitude of 14.5 Ϯ 5.2 pA (n ϭ 7). A similar pattern of current was seen with 1,4,5-IP 3 and its non-metabolizable analog, F-IP 3 . All the studies described here, including the results in Fig. 1, were carried out with F-IP 3 , to rule out possible effects due to hydrolysis or metabolism of IP 3 . For convenience we refer to this analog as IP 3 . Note that control cells (i.e. loaded with solution that did not contain IP 3 ) displayed currents in 20% of the cells (4/20) with an average amplitude of 5 Ϯ 2 pA and an onset time of 312 Ϯ 50 s.
The ramp protocol was also used to record the IP 3 -induced inward current and to determine the I-V curve (Fig. 2). The IP 3 -induced inward Ca 2ϩ current increased at negative membrane potentials, up to Ϫ120 mV, and showed inward rectification with a reversal potential more positive than ϩ10 mV. Tg, or BHQ, also activated a similar inward current although FIG. 1. Activation of an inward current by IP 3 and CCh in HSG cells. Currents were continuously recorded in HSG cells at a holding potential of 0 mV (gap-free configuration, pClamp 6.0). Recording was initiated immediately after whole cell configuration was obtained. IP 3 (F-IP 3 , 10 M) was included in the patch pipette, and CCh (100 M) was included in the perfusion medium, when the recording was initiated. These are the representative responses of HSG cells to IP 3 and CCh (n ϭ 9). with a different temporal profile (further discussed below), suggesting that depletion of internal Ca 2ϩ stores activates a cation channel in HSG cells. In some cells stimulated by IP 3 , CCh, or Tg, but not in unstimulated cells, a small inward current was detected at higher (ϩ40 mV to ϩ60 mV) membrane potentials (Fig. 2, A and B). The nature of this inward current is not presently clear. It should be noted that the internal and external solutions used blocked the Ca 2ϩ -activated K ϩ current, which has been previously described in these cells (17). Furthermore, Ca 2ϩ -activated Cl Ϫ currents are not detected in HSG cells, but only a volume-regulated outwardly rectifying Cl Ϫ current has been described (21).
To determine the permeability of the store depletion-activated channel to other cations, IP 3 -induced inward current was measured in cells perfused with Ca 2ϩ -free medium, i.e. with Na ϩ as the only cation in the medium. This condition decreased the amplitude of the inward current by 41% (n ϭ 4, Fig. 2A). Under these conditions, the estimated permeability of Na ϩ , relative to that of Ca 2ϩ , was 0.59 and 0.66 at Ϫ120 and 0 mV, respectively. However, when [Ca 2ϩ ] e was replaced by Ba 2ϩ , the inward current was only slightly less than that obtained with 10 mM Ca 2ϩ . Furthermore, when 1 mM Ca 2ϩ was used (see Fig.  4A), the peak amplitude of the current (53 Ϯ 5.1 pA) was significantly less (p Ͻ 0.05, n ϭ 3 to 5) than that measured with 10 mM Ca 2ϩ (127 Ϯ 10.8 pA), 10 mM Ba 2ϩ (109 Ϯ 7.9 pA), or Na ϩ (75 Ϯ 5.6 pA) in the medium. Thus, the store depletionactivated cation channel appeared to conduct Ba 2ϩ and Na ϩ when Ca 2ϩ ion was not present in the external medium and demonstrated cation permeability in the order Ca 2ϩ Ͼ Ba 2ϩ Ͼ Na ϩ . Importantly, these data suggest that with a normal physiological external solution, (i.e. containing 1 mM [Ca 2ϩ ] e and about 140 mM Na ϩ ), the channel is relatively more permeant to Ca 2ϩ than Na ϩ . This is further illustrated by the data in Fig.  2B showing the IV curves of the Ca 2ϩ , Ba 2ϩ , and Na ϩ currents. The IV curve of the IP 3 -induced Ca 2ϩ current was significantly different from that of the Na ϩ current (p Ͻ 0.05, analysis of variance analysis). Additionally, in the presence of La 3ϩ , the Ca 2ϩ current was almost completely inhibited and did not show any voltage sensitivity.
Previously we have shown that Ca 2ϩ channel antagonists La 3ϩ and Gd 3ϩ , but not Zn 2ϩ , blocked SOCE in HSG cells (17). Consistent with these results, IP 3 failed to evoke inward currents in the presence of 1 mM La 3ϩ (Fig. 2B). Fig. 3 shows continuous recordings of the inward current (at 0 mV and 10 mM Ca 2ϩ ). The IP 3 -activated current was greatly reduced by perfusion of La 3ϩ into the medium, and the reduction was reversed when La 3ϩ was removed (Fig. 3A). A similar inhibition was also seen with Gd 3ϩ (Fig. 3B) but not with Zn 2ϩ (data not shown). Furthermore, La 3ϩ (1 mM) also reversibly blocked IP 3 -induced activation of Na ϩ and Ba 2ϩ (n ϭ 5). In aggregate the data in Fig. 2 suggest that Ba 2ϩ , Na ϩ , and Ca 2ϩ pass through the same store-operated Ca 2ϩ channel in HSG cells. Furthermore, the divalent cation sensitivity of the inward Ca 2ϩ current is similar to that of SOCE which we have previously reported (17). These data demonstrate the activation of a storeoperated Ca 2ϩ -permeable cation channel in HSG cells. Al-FIG. 2. Ca 2؉ selectivity of the inward current. The inward cation currents activated by IP 3 (A) and their I-V curves (B) are shown. The currents were recorded using a 500-ms ramp protocol from Ϫ120 to ϩ60 mV and holding potential of ϩ35 mV. Na ϩ (150 mM) and Ba 2ϩ (10 mM) currents were recorded in a Ca 2ϩ -free medium, whereas the currents recorded in the presence of La 3ϩ (1 mM) was obtained in Ca 2ϩ (10 mM)-containing medium. though this current exhibits several characteristics similar to those of I CRAC , it is not inhibited by Zn 2ϩ (5). Thus, we have referred to this current as I SOC .
Ca 2ϩ -dependent Inactivation of I SOC -A ramp protocol was used to record I SOC with 1 mM Ca 2ϩ in the external medium and then repeated to record another trace in the same cell after [Ca 2ϩ ] e was increased to 10 mM (Fig. 4A). The average peak amplitude of the current with 1 mM Ca 2ϩ was 40.3 Ϯ 10.5% (n ϭ 3) of that with 10 mM Ca 2ϩ in the same cell. Similar effects were seen in a continuous recording of I SOC at 0 mV (Fig. 4B). The amplitude of the IP 3 -stimulated inward current decreased significantly with reduction of [Ca 2ϩ ] e concentration from 10 to 1 mM and increased again when [Ca 2ϩ ] e was changed back to 10 mM. However, with subsequent shifts in [Ca 2ϩ ] e from 10 to 1 mM and vice versa, the amplitude of the current with either 10 or 1 mM Ca 2ϩ gradually decreased, and after the third exposure to 10 mM Ca 2ϩ , the current at 1 mM Ca 2ϩ was not detectable above basal levels. The current seen with 10 mM Ca 2ϩ was decreased by about 50%. These data demonstrate that the amplitude of I SOC is dependent on [Ca 2ϩ ] e as was previously observed for Ca 2ϩ influx (17,22,23). These data also show that I SOC inactivates, either as a function of time after activation or due to exposure of cells to high [Ca 2ϩ ] e. However, the inhibition did not appear to be dependent on the presence of high [Ca 2ϩ ] e per se, since it was observed even after the 10 mM Ca 2ϩcontaining medium was replaced with 1 mM Ca 2ϩ medium.
To examine further the apparent inhibition of I SOC by Ca 2ϩ in HSG cells, we used a step protocol to measure IP 3 -generated I SOC at various times after the whole cell configuration was established (Ͻ30 s and 1-5 min). During the first sequence ( Fig. 5A), at each holding potential from Ϫ100 to Ϫ40 mV, the inward current evoked by IP 3 , via internal Ca 2ϩ release, was biphasic, with a rapid initial increase followed by a relatively slower decay. However, the amplitude of the initial current at each holding potential gradually decreased in the subsequent sequences. Two to three minutes after stimulation (Fig. 5, C and D), the peak current at any membrane potential was reduced by about 50%, and 5 min after stimulation the current could not be detected even at Ϫ100 mV (Fig. 5E).
A similar time-dependent inhibition of the inward Ca 2ϩ current was observed using repeated ramp protocols (Ͻ30 s and 1, 3, 5, and 10 min, see Fig. 6). The IP 3 -induced inward Ca 2ϩ current (with 10 mM [Ca 2ϩ ] e ) was significantly reduced at 3 and 5 min (traces II and III, respectively) after stimulation as compared with the initial response (trace I). Five minutes after stimulation, the inward current was reduced to 47 Ϯ 3.5% (n ϭ 3) of the initial current, while 10 min after stimulation, the current was barely detectable. However, the inactivation of the current was significantly delayed when the cell was in a low, more physiological, [Ca 2ϩ ] e (1 mM, Fig. 6B). The peak amplitudes of the inward current at 5 (trace II) and 10 (trace III) min after stimulation were 64 Ϯ 6.7 and 54.5 Ϯ 7.3% (n ϭ 4), respectively, of the initial current. In aggregate these data indicate that the Ca 2ϩ -dependent feedback inhibition of I SOC depends on [Ca 2ϩ ] e , and although it is detected at physiological Ca 2ϩ levels, it occurs more rapidly at higher [Ca 2ϩ ] e . Further- FIG. 4. Dependence of Ca 2؉ currents on [Ca 2؉ ] e . A, the currents stimulated with IP 3 were recorded using the ramp protocol as described for Fig. 2 in the same cell. Traces I and II were recorded in 1 mM Ca 2ϩ or 10 mM Ca 2ϩ -containing medium, as indicated. This trace is representative of the response seen in three cells. B, the IP 3 -activated current was continuously recorded as in Fig. 1, and the external [Ca 2ϩ ] concentration was shifted as shown by the corresponding bars. This is a representative trace from six cells. more, the data in Fig. 4B strongly indicate that this inhibition is not due to [Ca 2ϩ ] e per se.
We hypothesized that the inactivation of I SOC was determined by the rate (and thus the amount) of Ca 2ϩ entering the cell, i.e. when cells are pulsed at more negative membrane potentials in the presence of high [Ca 2ϩ ] e , Ca 2ϩ influx is greater and thus the rate of inactivation is also higher than seen with lower [Ca 2ϩ ] e and membrane potentials closer to 0 mV. To demonstrate this, we used the continuous recording protocol and stimulated the cells with IP 3 at a holding potential of Ϫ80 mV, instead of 0 mV. Whereas the amplitude of the initial current at Ϫ80 mV was higher than that at 0 mV, the current was transient and inactivated rapidly (in Ͻ20 s) to a level close to the resting current (Fig. 7, compare with trace in Fig. 1). This result demonstrates that the driving force for Ca 2ϩ , thus the amount of Ca 2ϩ entering the cell, modulates the inactivation of I SOC .
Role of a Staurosporine-sensitive Protein Kinase in the Ca 2ϩdependent Inactivation of SOCE-Based on our previous studies, we have suggested that the Ca 2ϩ -dependent feedback inhibition of SOCE in rat parotid gland cells is mediated via a staurosporine-sensitive protein kinase (12,14). We have also observed that other protein kinase inhibitors are ineffective in preventing the Ca 2ϩ -induced inhibition of Ca 2ϩ influx in these cells. 2 To determine whether SOCE in HSG cells is also affected by modulators of protein phosphorylation, we studied the effects of okadaic acid and calyculin A on thapsigargin-stimulated [Ca 2ϩ ] i changes in fura2-loaded HSG cells. Similar to our results with rat parotid acinar cells, both these inhibitors induced a significant inhibition of thapsigargin-stimulated Ca 2ϩ influx but not internal Ca 2ϩ release (Table I). We also examined the inhibition of Ca 2ϩ influx with 5 mM [Ca 2ϩ ] e since the patch clamp protocol uses high [Ca 2ϩ ] e . Calyculin A was more effective in blocking Ca 2ϩ influx at this higher [Ca 2ϩ ] e . The inhibition by either phosphatase inhibitor was prevented when cells were pretreated with staurosporine (data not shown). The effect of staurosporine and calyculin A was further tested on I SOC in HSG cells. Consistent with the data shown in Table I, calyculin A induced a partial, but reversible, inhibition of the IP 3 -induced Ca 2ϩ current (Fig. 8A), whereas staurosporine (500 nM) enhanced this current in an irreversible manner (Fig.  8B). Thus, the response of SOCE in HSG cells and rat parotid gland cells to modulators of protein phosphorylation is similar. Furthermore, these data also demonstrate that the effects of these agents on SOCE are not indirectly induced via effects on the membrane potential of the cell as has been suggested (5). Fig. 9A shows the effect of staurosporine on the time-dependent inactivation of I SOC in HSG cells. Cells were preincubated with the protein kinase inhibitor staurosporine (500 nM) for 10 min prior to the current measurement. Trace I was the initial recording (Ͻ30 s after whole cell configuration was estab-2 T. Sakai and I. S. Ambudkar, unpublished data.

FIG. 5. The inactivation of IP 3 -induced Ca 2؉ currents in HSG cells.
A step protocol was used to record the time course of the IP 3 -stimulated current at different holding potentials. The cell was held at 0 mV and stepped from Ϫ100 mV to Ϫ40 in 20 mV step with 200-ms intervals. The current (A) was recorded immediately after the whole cell model was established, i.e. dialysis of cells with IP 3 . The traces of B, C, D, and E were recorded at 1, 2, 3, and 5 min after. These data are representative of currents obtained in three cells. 2؉ ] e . The ramp protocol was used to record IP 3 -induced cation currents. A, the Ca 2ϩ current stimulated with IP 3 with 10 mM Ca 2ϩ in the medium was recorded immediately after dialysis of IP 3 into the cell, i.e. after whole cell model was obtained (trace I), and 3 and 5 min later, traces II and III. These are representative recordings from four cells. B, IP 3 -induced currents, with 1 mM external Ca 2ϩ , recorded initially (trace I), and 5 (trace II), and 10 min (trace III) after whole cell model was achieved. lished), while traces II and III were recorded 5 and 10 min after the initial recording. Staurosporine delayed the inactivation of I SOC significantly. The peak currents at the 5-and 10-min time points were 88.3 Ϯ 4.5 and 61.0 Ϯ 8.0% (n ϭ 4) of the initial response, respectively (Fig. 9B). These data are consistent with our studies discussed above and indicate that a staurosporinesensitive protein kinase is involved in the Ca 2ϩ -dependent inactivation of I SOC and suggest that an increase in protein phosphorylation induces a decrease in I SOC and vice versa.

FIG. 6. Dependence of the rate of inactivation of the IP 3 -stimulated cation current on [Ca
To determine whether the inactivation of I SOC is Ca 2ϩ -specific, the rate of inactivation was examined under several different conditions. The experimental protocol was similar to that described for Fig. 9A, and the amplitudes of the peak currents determined in the different experiments have been plotted in Fig. 9B. The highest rate of inactivation of I SOC was seen with 10 mM [Ca 2ϩ ] e in the medium; with 40% decrease in the current by 5 min and Ͼ90% by 10 min. The Na ϩ currents, measured in the absence of external Ca 2ϩ , did not show significant inactivation, and Ͼ80% of the initial current was still present 15 min after the cell was dialyzed with IP 3 . Ba 2ϩ currents showed a slower rate of inactivation initially (at 5 min) than that seen with either 1 or 10 mM Ca 2ϩ in the medium. After 15 min, the Ba 2ϩ current was inhibited by about 50%, similar to that seen with 1 mM [Ca 2ϩ ] e . When cells were first treated with staurosporine, despite the presence of 10 mM external Ca 2ϩ , there was Ͻ20% decrease in the current at 5 min and about 40% decrease at 15 min. Furthermore, increasing the [BAPTA] in the internal solution to 20 mM significantly reduced the rate of inactivation of I SOC . Finally, decreasing the frequency of the step protocol from 1 per min to 1 per 5 min decreased the rate of inactivation of I SOC . Ͼ50% of the current was seen 15 min after stimulation. In aggregate, these data demonstrate that the inactivation of I SOC is Ca 2ϩ -specific and is determined by the amount of Ca 2ϩ entering the cell as well as by the buffering of [Ca 2ϩ ] i .

TG-and BHQ-induced I SOC and the Effect on the IP 3 -induced
Ca 2ϩ Current-The data described above suggest that the increase in [Ca 2ϩ ] i in the region of Ca 2ϩ influx likely determines the inactivation rate of I SOC , as has been reported for I CRAC in T-lymphocytes (8,9). However, since 10 or 20 mM BAPTA was used in the pipette solution to buffer [Ca 2ϩ ] i in these experiments, this might appear to be an unlikely possibility. One possible explanation is that there is a limited amount of BAPTA in this region of the cell, which saturates as more Ca 2ϩ enters the cell (see "Addendum"). Furthermore, BAPTA from this region might not diffuse out readily. Our recent studies with HSG cells have suggested that the SERCA in the ER and the site of SOCE in the plasma membrane are in close proximity since the Ca 2ϩ which enters the cell via this pathway is rapidly transported into the ER without any increase in [Ca 2ϩ ] i in the region of Ca 2ϩ influx, as indicated by the lack of activation of K Ca (17). Similar data were also reported by Mogami et al. (24,25). These studies are consistent with earlier reports showing that [Ca 2ϩ ] i (measured using fura2) remains close to resting levels when Ca 2ϩ influx occurs during refilling of inter-   nal Ca 2ϩ stores (3,10,25,26). Additionally, internal Ca 2ϩ stores are refilled even in the presence of BAPTA, although at a somewhat slower rate (19,26). Thus, we hypothesized that in our present experimental system, [Ca 2ϩ ] i at the site of SOCE was determined by the following: (i) the amount of Ca 2ϩ entering the cell within a given time, (ii) the buffering capacity of BAPTA, and (iii) the SERCA activity. We have presented evidence for the first two points above.
To demonstrate a role for SERCA activity in the regulation of I SOC , we used the SERCA inhibitors, Tg and BHQ, which, by blocking the uptake of Ca 2ϩ into intracellular stores, allow Ca 2ϩ to "leak" out of the ER, thus activating SOCE (27). Fig.  10A shows the activation of I SOC by Tg. Additionally, when cells were preincubated with Tg (10 M) for 5-10 min, IP 3 failed to induce any further inward currents in HSG cells (n ϭ 5, data not shown). These data are important in that they demonstrate that IP 3 and thapsigargin activate the same pathway. However, the Tg-induced inward current (measured at 0 mV and 10 mM Ca 2ϩ ) was relatively transient as compared with that activated by IP 3 or CCh and was inactivated within 2 min after stimulation (similar results were obtained with BHQ, these data are not shown). It should be noted that a few cells treated with Tg or BHQ showed longer activation of I SOC (up to 3 min), whereas about half of the cells did not display any inward current (15/27). Tg-induced Ca 2ϩ current recorded using the ramp protocol also inactivated rapidly (Fig. 9B, compare traces  I and II). These data suggest that inhibition of SERCA activity increases the rate of inactivation of the I SOC .
A possible explanation of these results is that when the SERCA activity is inhibited, there is an increase in the ambient [Ca 2ϩ ] i in the region of the SOCE since Ca 2ϩ uptake into intracellular Ca 2ϩ stores is prevented. This would increase the inactivation rate of the SOCE. To test this, we examined the effect of Tg on IP 3 -induced Ca 2ϩ current. Fig. 11A shows that addition of Tg to cells displaying IP 3 -activated I SOC (continuous recording at 0 mV and 10 mM external Ca 2ϩ ) resulted in a rapid inactivation of the current. The Tg-induced inhibition of I SOC was attenuated when the cells were loaded with 20 mM BAPTA instead of 10 mM BAPTA (n ϭ 4, data not shown). The effect of thapsigargin on I SOC was further examined by using the ramp protocol (Fig. 11B). I SOC was first activated by dialysis of IP 3 into the cell (trace I); the cell was then perfused with medium containing Tg for 1-2 min, and another trace (trace II) was recorded. I SOC appeared to be completely inactivated at this FIG. 9. Effect of staurosporine and the Ca 2؉ specificity of the inactivation of IP 3 -induced I SOC . A, the cell was incubated with staurosporine (500 nM) for approximately 10 min before patch clamp recording. The IP 3 -induced current was recorded at beginning (trace I), 5 (trace II), and 15 (trace III) min after whole cell model was obtained. These traces are representative of results from three cells. B, the same ramp experiment protocol was employed to record IP 3 -induced Ca 2ϩ currents in cells perfused with l mM Ca 2ϩ , 10 mM Ca 2ϩ , with 10 mM Ca 2ϩ ϩ 20 mM BAPTA (ࡗ) in the internal solution, 10 mM Ba 2ϩ or Na ϩ (in Ca 2ϩ -free medium), or with 10 mM Ca 2ϩ in cells preincubated with staurosporine as mentioned above. The percentage changes of the currents recorded at 1, 5, 10, and 15 min compared to the initial current are shown. The data shown are the mean and S.E. from three to four cells in each group. time. Note that the IP 3 -induced Ca 2ϩ current typically decreased by about 50% at 5 min after stimulation (see Fig. 6A).

DISCUSSION
The data presented above demonstrate the activation of an inward Ca 2ϩ current in HSG cells by conditions resulting in the depletion of the internal Ca 2ϩ store(s). This is the first report of this current in salivary gland cells, a widely used cell type for studying mechanisms of Ca 2ϩ signaling and Ca 2ϩ influx (2, 3, 10 -17). Inward currents activated in response to store depletion have been measured in a number of cell types (4,5). The best characterized of these currents is the I CRAC , which has been detected in mast cells, RBL cells, and T-lymphocytes. These previous studies suggest that I CRAC is mediated by a Ca 2ϩ channel, which is gated by the depletion of the internal Ca 2ϩ store(s) (5). We have referred to the internal Ca 2ϩ store depletion-activated inward Ca 2ϩ current in HSG cells as I SOC since several characteristics were noted that distinguished it from I CRAC .
Unlike I CRAC , I SOC in HSG cells was not sensitive to Zn 2ϩ , but was inhibited by Gd 3ϩ , consistent with our previous studies with HSG cells. Furthermore, okadaic acid and calyculin A inhibited I SOC in HSG cells. Conversely, okadaic acid, but not calyculin A, prevented refill-dependent inactivation of I CRAC in T-lymphocytes. We have also shown that staurosporine increased I SOC in HSG cells and prevented the Ca 2ϩ -induced feedback inhibition. We have described similar effects of staurosporine on Ca 2ϩ influx in HSG cells (this report) and rat parotid acinar cells (12,14). Importantly, these data demonstrate that the characteristics of I SOC and SOCE in HSG cells are similar (17) which indicates that they represent the same mechanism. Additionally, several characteristics of I SOC , e.g. sensitivity to phosphatase inhibitors and staurosporine, are similar to those of SOCE in rat parotid acinar cells. However, like I CRAC , SOCE in rat parotid acinar cells is inhibited by low concentrations of Zn 2ϩ (5,28). Furthermore, a Ca 2ϩ -conducting nonspecific cation channel activated by store depletion in rat pancreatic acinar cells is not inhibited by La 3ϩ or Gd 3ϩ (29). In aggregate, these data demonstrate marked differences between the SOCE and store-operated cation currents in different cell types in their sensitivity to divalent and trivalent cations and effects of modulators of protein phosphorylation. These characteristics could provide a useful a tool to identify different "isoforms" of this putative Ca 2ϩ channel.
I SOC , like I CRAC , displayed the following: (i) selectivity for Ca 2ϩ (over Na ϩ ) and permeability to Na ϩ and Ba 2ϩ in the absence of Ca 2ϩ , (ii) inward rectification at positive potentials with reversal at positive potentials more than ϩ10 mV, and (iii) dependence on external [Ca 2ϩ ]. Our data show that external [Ca 2ϩ ] determines the initial amplitude of I SOC and the rate of inactivation of the current. In the case of I CRAC , both Ca 2ϩ -dependent potentiation as well as inactivation have been reported (5). We have not examined the possible potentiation effects of Ca 2ϩ on I SOC in this study. However, we have examined the inactivation of I SOC that was detected in the continued presence of a non-metabolizable analog of IP 3 , using three different experimental protocols. In the step protocol, a biphasic current was stimulated, with an initial peak increase followed by a steep inactivation. The initial amplitude of the current and the subsequent inactivation rate was greater at more negative membrane potentials and higher [Ca 2ϩ ] e . In the ramp protocol, the current appeared to inactivate rapidly as the membrane potential was shifted from negative to positive potentials. It can be suggested that this decrease in the current amplitude reflects combined effects of the decrease in driving force for Ca 2ϩ influx and possible fast inactivation mechanism(s) as has been reported for I CRAC in lymphocytes (8,9). Notably, pretreatment of HSG cells with staurosporine did not affect the pattern of I SOC during the first ramp (compare trace 1 in Figs.  2A and 9A). Thus, we do not believe that the Ca 2ϩ -dependent inactivation mechanism we have described above contributes to the decrease in the amplitude of the current during the first ramp. However, it is likely that this inactivation mechanism accounts for the somewhat exaggerated curvature of the I-V curve (for the Ca 2ϩ current, see Fig. 2B) at the more negative membrane potentials. Similar observations were previously made by Hoth and Penner (6). Importantly, the present data clearly demonstrate a slower, time-dependent, inactivation of I SOC in HSG cells (seen in both the ramp as well as step protocols) that is induced by Ca 2ϩ and to a lesser extent by Ba 2ϩ , but not by Na ϩ . In the continuous recording protocol, a rapid run down of the current was seen with high [Ca 2ϩ ] e at more negative membrane potentials. Lewis and co-workers (8,9) have described Ca 2ϩ -dependent slow and fast inactivation mechanisms of I CRAC in T-lymphocytes. The fast inactivation was attributed to the [Ca 2ϩ ] i near the mouth of the channel and the slow inactivation to the Ca 2ϩ -dependent activation of a phosphatase enzyme, which was sensitive to okadaic acid but not calyculin A. However, our data presently do not establish that the fast and slow inactivation mechanisms of I SOC in HSG cells are mutually distinct. Importantly, in HSG cells both phosphatase inhibitors inactivate I SOC , whereas the protein kinase inhibitor, staurosporine, prevents the inactivation. Thus I CRAC and I SOC may be fundamentally different, with respect to the regulation of their activity and to nature of the molecular components.
We have shown that I SOC inactivates within 5 or 10 min in cells subjected to repeated step or ramp protocols, respectively. The possibility that high external [Ca 2ϩ ] per se inhibits Ca 2ϩ influx can be ruled out since IP 3 -induced I SOC measured with 10 mM Ca 2ϩ in the medium was inactivated when the holding potential was Ϫ80 mV but not 0 mV. Thus, the rate of inactivation of I SOC depends on the driving force for Ca 2ϩ influx, i.e. on the amount of Ca 2ϩ entering the cell. In aggregate these data are consistent with the suggestion that an increase in the [Ca 2ϩ ] i in the region of the SOCE induces a feedback inhibition of Ca 2ϩ influx (14,19). Furthermore, the data show that the inactivation is attenuated by pretreatment of the cells with staurosporine, suggesting the involvement of a protein kinase in the inactivation. There are two possible explanations for these results. (i) Ca 2ϩ directly activates the protein kinase or inhibits the protein phosphatase, which in either case would result in an increase in the protein phosphorylation level. (ii) Alternatively, the effect of Ca 2ϩ and phosphorylation are independent, and the Ca 2ϩ -dependent inactivation cannot be exerted when the protein is dephosphorylated. The present data do not distinguish between these two possibilities.
Our present and previous data (17) and the studies reported by Mogami et al. (24,25) suggest that the driving force for Ca 2ϩ influx and the SERCA activity regulate the ambient [Ca 2ϩ ] i in the region of the putative store-operated Ca 2ϩ channel (SOC). Such a suggestion might seem unlikely in the present study, since the cells were dialyzed with a solution containing 10 mM BAPTA, which tightly buffers changes in [Ca 2ϩ ] i . However, as discussed above, there may be a limited amount of BAPTA in this subplasma membrane region of the cell, which would tend to saturate as more Ca 2ϩ enters the cell (see "Addendum"). It is reasonable to predict that there is an equilibrium between free BAPTA, [Ca 2ϩ ] i , and Ca 2ϩ -bound BAPTA in the region of the cell near the SOC (see Fig. 12 and discussion below). Our data demonstrate that despite the presence of BAPTA, this equilibrium is affected by the activity of the SERCA pump and the rate of Ca 2ϩ entering the cell. In the ramp and step protocols, repeated application of a high driving force, although for a short time, could drive sufficient amounts of Ca 2ϩ into the cell to alter the BAPTA-Ca 2ϩ equilibrium in the region of the SOC. This suggestion is consistent with our observations that (i) increasing intracellular [BAPTA] to 20 mM (Fig. 9B) and (ii) increasing the time interval between the step sequences (data not shown) decreased the rate of inactivation of I SOC .
BAPTA has been shown to effectively clamp [Ca 2ϩ ] i and block global increases in [Ca 2ϩ ] i (21,26). The subplasma membrane [Ca 2ϩ ] i has also been reported to be clamped by BAPTA since the activation of Ca 2ϩ -dependent Cl Ϫ or K ϩ currents is prevented in BAPTA-loaded cells, including HSG cells (2,17). However, it is possible that significant changes in [Ca 2ϩ ] i can occur in localized subplasma membrane regions of BAPTAloaded cells, i.e. in the region of Ca 2ϩ influx. As mentioned above, such changes depend on the amount of Ca 2ϩ introduced into the region and the BAPTA concentration. That BAPTA in fact reaches this region of the cell is inferred from previous studies showing that it takes longer for the internal Ca 2ϩ stores to be refilled in BAPTA-loaded cells and that the rate of refilling can be increased at higher [Ca 2ϩ ] e (19,22,26). This is consistent with our previous data showing that SERCA has a very high affinity for Ca 2ϩ and has a high pump rate at resting [Ca 2ϩ ] i (30). On the other hand the plasma membrane Ca 2ϩ pump (PMCA) is relatively less active at resting [Ca 2ϩ ] i (20,30,31). Thus, it is not clear what contribution, if any, its activity has in the regulation of [Ca 2ϩ ] i in localized subplasma membrane regions in BAPTA-loaded cells. However, most likely, under physiological conditions, when BAPTA is not present, the PMCA activity would also have a role in decreasing the [Ca 2ϩ ] i in this region.
We have recently shown (17) that Ca 2ϩ entering the cell is rapidly accumulated into the internal Ca 2ϩ store by the SERCA activity thus preventing significant diffusion of Ca 2ϩ in the subplasma membrane region. According to the model we have proposed, in cells where the IP 3 -sensitive channel has been activated, Ca 2ϩ which enters via SOC is taken up into the internal Ca 2ϩ stores and then released at localized regions of the cell by the IP 3 -sensitive Ca 2ϩ channel. Mogami et al. (24) have proposed a similar model and have further suggested that when internal Ca 2ϩ stores are depleted, the SERCA activity is increased. This is consistent with our present data since such an increase in SERCA activity would facilitate the rapid uptake of Ca 2ϩ into the store, either in the presence or absence of BAPTA in the cytosol. Thus, the increase in the rate of inactivation of I SOC that was induced by Tg and BHQ directly demonstrates a functional association between SERCA and the SOC (Fig. 12). We suggest that SERCA modulates I SOC by regulating the [Ca 2ϩ ] i in the region of the SOC. Low [Ca 2ϩ ] i would be achieved when stores are depleted and SERCA activity is high; this would increase I SOC . Increases in [Ca 2ϩ ] i would be induced when internal Ca 2ϩ stores are refilled and SERCA activity is deceased; this would initiate the feedback inactivation of I SOC . The feedback inactivation would also be initiated under any other condition in which [Ca 2ϩ ] i in the region of SOC is increased, thus providing the cell with an effective protective mechanism against uncontrolled Ca 2ϩ influx.
In summary, our data demonstrate directly that the SERCA activity is involved in regulating the rate of inactivation of I SOC . These data could also offer an explanation for the observation in a number of cells, including in HSG cells (shown above), that thapsigargin is not very effective in stimulating inward Ca 2ϩ currents. According to our present results this could be due to an accelerated inactivation of the Ca 2ϩ channel. However, such a functional interaction between SERCA and SOC might not be present in all types of cells. Alternatively, the current might be more sensitive to other mechanisms of regulation, i.e. protein phosphorylation. In this case, the detection of the current would depend on the presence of such regulatory mechanisms and the rate at which they are activated.