Characteristics of a Store-operated Calcium-permeable Channel SARCOENDOPLASMIC RETICULUM CALCIUM PUMP FUNCTION CONTROLS CHANNEL GATING*

We examined the single channel properties and regulation of store-operated calcium channels (SOCC). In human submandibular gland cells, carbachol (CCh) induced flickery channel activity while thapsigargin (Tg) induced burst-like activity, with relatively lower open probability (NP o ) and longer mean open time. Tg- and CCh-activated channels were permeable to Na (cid:1) and Ba 2 (cid:1) , but not to NMDG, in the absence of Ca 2 (cid:1) . The channels exhibited similar Ca 2 (cid:1) , Na (cid:1) , and Ba 2 (cid:1) conductances and were inhibited by 2-aminoethoxydiphenylbo-rate, xestospongin C, Gd 3 (cid:1) , and La 3 (cid:1) . CCh stimulated flickery activity changed to burst-like activity by (i) addition of Tg, (ii) using Na (cid:1) instead of Ca 2 (cid:1) , (iii) using Ca 2 (cid:1) -free bath solution, or (iv) buffering [Ca 2 (cid:1) ] i with BAPTA-AM. Buffering [Ca 2 (cid:1) ] i induced a 2-fold increase in NP o of Tg-stimulated SOCC. Reducing free [Ca 2 (cid:1) ] in the endoplasmic reticulum with the divalent cation ch-elator, N,N,N (cid:1) , N (cid:1) -tetrakis(2-pyridylmethyl)ethylenedia-mine

Calcium influx into non-excitable cells is mediated via storeoperated Ca 2ϩ entry channel(s) (SOCC), 1 which are activated by the depletion of Ca 2ϩ from intracellular calcium store(s) (1)(2)(3). Physiologically, Ca 2ϩ store depletion is typically achieved following receptor-mediated activation of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) hydrolysis, generation of inositol 1,4,5-trisphosphate (IP 3 ), and release of Ca 2ϩ via the inositol trisphosphate receptor (IP 3 R) by the binding of IP 3 to the IP 3 R. According to the capacitative Ca 2ϩ entry hypothesis (1) SOCC inactivation occurs when the internal Ca 2ϩ store(s) is refilled, primarily by the uptake of Ca 2ϩ into the store(s) by the sarcoendoplasmic Ca 2ϩ pump (SERCA). Although the general assumption is that inactivation of SOCC involves a reversal of the activation mechanism, this has not yet been established. Moreover, the precise molecular mechanism(s) that senses the status of the internal Ca 2ϩ store to either activate (upon depletion), or inactivate (upon refill), SOCC is not known.
Three main mechanisms currently proposed for the activation of SOCC include involvement of a diffusible messenger, recruitment of intracellular vesicles, and a physical interaction between the SOCC and IP 3 R (4,5). According to the latter, i.e. the conformational coupling model (2), the IP 3 R acts as the sensor of the internal Ca 2ϩ store and conveys the signal to the SOCC via a conformational change that results in activating the plasma membrane channel. However, store-operated calcium influx is efficiently activated in a variety of cells by the intracellular Ca 2ϩ pump (SERCA) inhibitor, thapsigargin, that depletes intracellular Ca 2ϩ stores without changing cellular levels of IP 3 (6). Thus, there has been much discussion about the direct involvement of PIP 2 hydrolysis, IP 3 , or IP 3 R in the regulation of SOCC (3,4,7,8).
Recently studies with the Trp family of putative Ca 2ϩ channel proteins, which have been proposed as molecular components of the SOCC, have provided information regarding the possible involvement of IP 3 R and PIP 2 hydrolysis in storeoperated Ca 2ϩ influx (9,10). For example, some Trps, such as Trp-3 and Trp-6, are activated by PIP 2 hydrolysis, likely via an involvement of IP 3, diacylglycerol, or PIP 2 itself (9,10). Other Trps, such as Trp-1, can be activated by an agonist or by thapsigargin alone, i.e. without PIP 2 hydrolysis (8,11). Based on these studies, it can be suggested that agonists, which result both in an increase in IP 3 and store depletion activate a different SOCC channel than that activated by thapsigargin, which induces store depletion without increase in IP 3 . Alternatively, it can be suggested that both reagents activate the same channel. Since there is little information regarding the single channel properties of the endogenous SOCC(s) in nonexcitable cells, these possibilities have not yet been fully resolved. Previous studies using fura-2 to measure changes in [Ca 2ϩ ] i have shown that although the rate of activation of store-operated Ca 2ϩ influx by thapsigargin is slower, the final level and the characteristics of the Ca 2ϩ influx are similar to that achieved with maximal concentrations of an agonist (3,6,12). Similar conclusions have been made based on electrophysiological studies, which have mainly involved measurements of whole cell currents (3,13) and by noise analysis of whole cell currents (14,15). In these studies the SOCC-associated current was activated by the inclusion of IP 3 in the pipette solution. Thus, these studies suggest that SOCC activation might not be due to PIP 2 hydrolysis per se but rather due to IP 3 , acting either directly on the SOCC or indirectly via an effect on the IP 3 R. Other recent * 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 all correspondence should be addressed: Bldg. 10 reports are consistent with the involvement of the IP 3 R in the activation of SOCC (16 -19) although this mechanism has been recently questioned (20).
In the present study, we have examined the single channel characteristics of SOCC activated by the muscarinic agonist, carbachol, the SERCA inhibitor, thapsigargin, and the permeant divalent cation chelator, TPEN, in the human submandibular gland cell line (HSG) by using the cell-attached patch clamp technique. We have previously shown that both carbachol and thapsigargin strongly induce store-operated Ca 2ϩ entry and the store-operated Ca 2ϩ current (I SOC ) in these cells (13,21,22). The present data demonstrate that carbachol, thapsigargin, and TPEN activate the same SOCC in HSG cells. Importantly, we show that the gating of this channel is regulated by the function of SERCA, which determines the [Ca 2ϩ ] in the internal Ca 2ϩ store and in the vicinity of SOCC.

EXPERIMENTAL PROCEDURES
Cell Culture-HSG cells were cultured in Earle's minimal essential medium supplemented with 10% fetal bovine serum, 1% penicillin/ streptomycin at 37°C in 5% CO 2 For electrophysiology and [Ca 2ϩ ] i measurements, confluent cells were detached from tissue culture dishes and plated on glass coverslips. Measurements were done after 24 h. In some experiments, cells were preincubated with 2-APB, xestospongin, or BAPTA-AM for 3 to 30 min, as indicated. CCh and Tg were perfused to the bath at a rate of 5 ml/min.
Electrophysiology-Patch clamp experiments were performed in the cell-attached configuration with pipette solution containing 100 mM Na/HEPES and 2 mM CaCl 2 , pH 7.2 (HCl) . In some experiment 100 mM NMDG/HEPES was used. Total Cl Ϫ ion concentration in the pipette was estimated to be about 6 mM. Cells were bathed in a standard bath solution containing (mM): 145 NaCl, 5 KCl, 1 MgCl 2 , 1 CaCl 2 , 10 HEPES, pH 7.4 (with NaOH). This solution was replaced by a high KCl solution containing (mM) 145 KCl, 5 NaCl, 1 MgCl 2 , 2 CaCl 2 , 10 HEPES, pH 7.4 (with KOH), after the seal was formed. For measurement of Ba 2ϩ or Na ϩ permeabilities, pipette solution contained either 2 mM BaCl 2 or 2 mM NaCl, 100 mM NMDG/HEPES, pH 7.2 (HCl), and 0.1 mM EGTA. In addition, 2 mM NMDG-Cl was added to the buffer during Na ϩ current measurements in order to maintain equal [Cl Ϫ ]. Liquid junction potentials (8.0, 4.3 and 4.2 mV for Ca 2ϩ , Na ϩ , and Ba 2ϩ currents measurement, respectively) were calculated using P-Clamp7 software. Currents were recorded using an Axopatch 200A amplifier and digitized with Digidata 1200 (Axon instrument) at a rate of 4 kHz by filtering at 1 kHz. All experiments were done at room temperature.
Data Analyses-P-Clamp 7 (Axon instrument) and Origin 6 (Micro- [Ca 2ϩ ] i Measurements-Fura-2 fluorescence in single cells was measured as described earlier (11,22) by using an SLM 8000/DMX 100 spectrofluorimeter attached to an inverted Nikon Diaphot microscope with a Fluor ϫ40 oil-immersion objective. Images were acquired using an enhanced CCD camera (CCD-72, MTI) and the Image-1 software (Universal Imaging Corp., PA). Analog plots of the fluorescence ratio (340/380) in single cells are shown.

Activation of SOCC in HSG Cells by Carbachol and
Thapsigargin-HSG cells did not display any significant spontaneous currents at membrane potentials between Ϫ80 to ϩ80 mV (115/123 cells, Fig. 1A shows representative traces in a cell clamped at Ϫ40 and ϩ40 mV). However, stimulation of the cells with either CCh (1 mM) or Tg (1 M) induced channel activity in 67/123 cells (about 54%). Most of the cells activated by CCh (31/35, 94%) displayed currents with a "flicker" mode, i.e. with relatively short openings (representative traces are shown in Fig. 1B. Note that the traces in Fig. 1, A and B, were recorded from the same cell). Closed "C" (indicating the 0 current level) and open "O" states are indicated in the current recordings at Ϫ40 and ϩ40 mV (the pipette solution contained 100 mM Na-Hepes, pH 7.4, 2 mM Ca 2ϩ ). The activity was quite stable for 3 to 5 min after which there was a gradual decay. The all-point amplitude histogram for the current at Ϫ40 mV ( Fig. 1E) revealed one major peak with unitary amplitude of about Ϫ1 pA. Occasional overlapping channel openings were seen corresponding to multiples of the unitary conductance level (about Ϫ2 pA, see expanded trace in Fig. 1B which shows simultaneous openings of multiple channels). Since the number of events in the second peak (Ͼ1 pA) was relatively low, data analysis was focused on the major peak. The amplitude-voltage (I-V) relationship (Fig. 1D) of the CCh-induced major current peak (Ϫ1 pA), showed a linear fit, reversing at about 0 mV, with a slope conductance of 20 pS (each point was derived from 6 to 14 patches). The NP o of the channel with CCh was 19.6 Ϯ 2.6% (see Table I). Consistent with our previous studies of storeoperated Ca 2ϩ entry and the store-operated current (I SOC ) in HSG cells, CCh activation of the current (both inward and outward) was not observed when either Gd 3ϩ (1 mM, 6/6 cells, see Fig. 1D) or La 3ϩ (1 mM, 6/6 cells, data not shown), was included in the pipette solution. Zn 2ϩ did not block channel activity (data not shown).
Store-depletion by inhibition of SERCA with Tg also stimulated single channel events in HSG cells (Fig. 1C), although the channel characteristics were markedly different from those seen with CCh (compare traces in Fig. 1B with those in Fig.  1C). Majority of cells (26/32, 81%) stimulated with Tg displayed "burst-like" channel activity that displayed longer openings at one current level, than that stimulated by CCh. However, the major peak revealed in the all-point amplitude histogram ( Fig.  1F) with unitary current amplitude of about Ϫ1 pA, was similar to that in CCh-stimulated cells. Furthermore, Tg also stimulated occasional events corresponding to multiples of this unitary conductance level. In general, the number of these events was less than that obtained in CCh-stimulated cells and could not be clearly resolved in the all-point amplitude histogram. The NP o of channel activation with Tg (9.8 Ϯ 0.8%) was significantly less than that with CCh (see Table I). Thus, there is apparently less channel activity with Tg than with CCh. The I-V relationship of the major (Ϫ1 pA) current (Fig. 1D) revealed a linear fit with a single channel conductance of 20 pS, which was not different from that obtained with CCh. Additionally, Tg-stimulated channel current, both inward and outward, was also blocked by inclusion of Gd 3ϩ (1 mM, n ϭ 5) or La 3ϩ (1 mM, n ϭ 6), but not Zn 2ϩ , in the pipette solution (data not shown).
Characteristics of CCh-and Tg-stimulated Channel Activity in HSG Cells-To further characterize the CCh-and Tg-stimulated channel activities, we measured the relative permeability of the respective channels to various cations. In the presence of 100 mM NMDG ϩ 0.1 mM EGTA in the pipette, no inward currents were seen with either Tg or CCh (Fig. 2, A and B, n ϭ 7). However, an outward current was seen between Ϫ40 mV and ϩ80 mV. This outward current likely represents K ϩ efflux via SOCC, as it was insensitive to the Ca 2ϩ -activated K ϩ channel blocker, charybdotoxin (added to the pipette and bath solutions) but was blocked when Gd 3ϩ or La 3ϩ were added to the pipette solution (data not shown). When the pipette solution contained 100 mM NMDG ϩ 2 mM Ca 2ϩ , both inward and outward currents were seen between Ϫ80 mV and ϩ80 mV in CCh-stimulated cells (Fig. 2B). The I-V curve displayed a linear fit and reversed at 0 mV. The amplitude of this current at the various voltages was similar to that seen when 100 mM Na ϩ ϩ 2 mM Ca 2ϩ was used in the pipette solution (see Figs. 1F and 2, B and D). Increasing the [Ca 2ϩ ] in the pipette solution to 20 mM induced a linear current-voltage relationship, however, the reversal potential shifted to about ϩ20 mV (Fig.  2B). Similar results were obtained when Tg was used to stimulate SOCC ( Fig. 2A).
The permeability of the Tg-or CCh-stimulated channel to Na ϩ was measured by including 100 mM of Na-Hepes ϩ 0.1 mM EGTA or 100 mM Na-Hepes ϩ 2 mM Ca 2ϩ in the pipette solution. The I-V plots of the currents stimulated by Tg is shown in Fig. 2C and that by CCh in Fig. 2D. The calculated conductances for Na ϩ were 44 and 42 pS with CCh and Tg, respectively. When 2 mM Ca 2ϩ was included in the pipette, the current was reduced. Importantly, similar currents were obtained with 2 mM Ca 2ϩ ϩ 100 mM Na ϩ or 2 mM Ca 2ϩ ϩ 100 mM NMDG ϩ . The relative current amplitudes measured at Ϫ40 mV under these different conditions are shown in Fig. 2, G and H. These data suggest that under physiological conditions (i.e. below Ϫ10 mV, with 1 mM Ca 2ϩ ϩ 100 mM Na ϩ in the external medium) the channel activated by either CCh or Tg primarily permitted Ca 2ϩ influx. Na ϩ did not contribute to this inward current. Na ϩ influx was seen only when Ca 2ϩ was removed from the external medium. Similar results have been reported for CRAC, a calcium-release activated Ca 2ϩ channel in T lymphocytes. It has been shown that cation currents via CRAC are increased by removing divalent cations (14,15,23), the conductance changes from 0.5 pS in a medium containing Na ϩ ϩ Ca 2ϩ ϩ Mg 2ϩ to about 44 pS in a medium containing Na ϩ , without Ca 2ϩ and Mg 2ϩ . Our data show that the channel con-  2. Cation selectivity of CCh-and Tg-activated channel. Channel activity was recorded using cell attached patches with the pipette solutions as indicated in the figure, in cells stimulated with either Tg (A, C, and E) or CCh (B, D, and F). Relative current amplitudes at Ϫ40 mV are given in G. The third bar in each group represents the amplitude of the current with NMDG, which was not detectable at Ϫ40 mV. ductance, with either CCh or Tg, is increased about 2-fold when Ca 2ϩ is removed from the pipette solution (note that in our experiments the pipette solution did not contain Mg 2ϩ but the bathing solution contained 1 mM Mg 2ϩ ). While this manuscript was in preparation, Yue et al. (24) reported characteristics of the CaT1-induced cation channel activity. This channel also displayed similar Na ϩ conductance (between 40 and 50 pS) in divalent cation-free conditions. Furthermore, the relative increase in the Na ϩ current upon removal of Ca 2ϩ and Mg 2ϩ was less than that seen with CRAC. Importantly, the Na ϩ conductances of CRAC, CaT1, and the channels stimulated by either Tg or CCh in HSG cells appear to be similar. However, the cation selectivity and Ca 2ϩ conductance of CRAC is quite distinct from that seen for the channel activated in HSG cells.
Permeability of the channels to Ba 2ϩ was also examined by including 2 mM Ba 2ϩ , instead of Ca 2ϩ , with 100 mM NMDG in the pipette solution. Similar Ba 2ϩ currents were induced by Tg (Fig. 2E) and CCh (Fig. 2F). In either case, the Ba 2ϩ currents were slightly smaller than Ca 2ϩ currents (see Fig. 2G for the relative amplitudes). The calculated Ba 2ϩ conductance was 15 pS with both CCh and Tg. In aggregate, the data presented in Fig. 2 strongly indicate that CCh and Tg activate the same calcium-permeable cation channel in HSG cells. The activation of this channel by Tg suggests that it is a store-operated cation channel, SOCC. Another interesting similarity in the channel behavior between CCh-and Tg-stimulated cells was that the P o of the channel (with 2 mM Ca 2ϩ in the pipette solution) increased at more negative membrane potentials (data not shown). This behavior is also similar to that recently reported for the CaT1 channel (12).
A major problem in identifying store-operated channels is the lack of specific pharmacological tools. As we have shown above, inorganic cations, such as La 3ϩ , Ni 2ϩ , Zn 2ϩ , and Gd 3ϩ , appear to have some selectivity in blocking various calcium influx pathways. Recently, there has been much focus on the effects of 2-aminoethoxydiphenylborate (2-APB) (25), which was first reported to block store-operated calcium channels by intervening with IP 3 -induced Ca 2ϩ release (19). However, as reported more recently, 2-APB might directly block SOCC and some Trp channels (20,26) but not other calcium channels; e.g. arachidonic acid-activated calcium influx. Irrespective of how 2-APB inhibits store-operated calcium channels, Ma et al. (27) have suggested that this inhibitory effect represents an impor-tant functional similarity between invertebrate Trp channels, mammalian Trp channels, and mammalian store-operated channels. Thus, we tested the effect of 2-APB on CCh-and Tg-stimulated [Ca 2ϩ ] i changes and channel activation. In control cells both CCh and Tg induced a transient increase in [Ca 2ϩ ] i , due to internal Ca 2ϩ release in a Ca 2ϩ -free media. After reintroduction of Ca 2ϩ into the bath [Ca 2ϩ ] i was significantly increased, demonstrating activation of Ca 2ϩ influx (Fig.  3, A and E). However, when the cells were exposed to 2-APB prior to reintroduction of Ca 2ϩ , [Ca 2ϩ ] i increase due to storeoperated influx was completely inhibited (Fig. 3, B and F). This was further confirmed by measuring SOCC activity in cellattached patches in HSG cells. Cells were preincubated with 2-APB for 3 to 5 min before either CCh or Tg were added. No currents were detected in either case at membrane potentials between ϩ40 and Ϫ80 mV (Fig. 3, C and G, shows currents measured at Ϫ40 mV, n ϭ 6 in each case). We also tested the effects of xestospongin C (16, 19) on SOCC activation. As seen with 2-APB, SOCC activation by either CCh or Tg was completely blocked by xestospongin C treatment (Fig. 2, D and H,  n ϭ 6). Xestospongin C also blocked [Ca 2ϩ ] i increases due to Ca 2ϩ influx in Tg and CCh-stimulated cells (data not shown, number of cells imaged per experiment was more than 50, experiments were repeated three times). While xestospongin has been suggested as an inhibitor of the IP 3 R, it appears to be structurally similar to a dimeric form of 2-APB (28) and thus could also have direct effects on SOCC. This possibility has to be further examined. In aggregate, the data described above demonstrate that the channels activated by Tg and CCh have similar pharmacological characteristics. These data provide further evidence that Tg and CCh activate the same SOCC in HSG cells.
Effect of SERCA Inhibition on CCh-stimulated Channel Activity- Fig. 1 illustrates a dramatic difference in the channel characteristics seen in Tg-and CCh-stimulated cells. Possible explanations for this are (i) activation of different channels by Tg and CCh, and (ii) lack of internal Ca 2ϩ accumulation in the Tg-stimulated cells. The data described above strongly suggest that CCh and Tg do not activate different cation channels in HSG cells. To further rule out this possibility and to test the role of SERCA, Tg was added to cells after the channel had been activated by CCh. Tg addition induced a dramatic change in the channel behavior (Fig. 4A shows traces recorded from the same cell used in Fig. 1, A and B. Tg was added to the cell after CCh-stimulated channel activity was recorded. 7/9 cells displayed this change in channel activity.) The rapid and transient openings of the channel induced by CCh were converted to more stable activity, with longer mean open times at the Ϫ1 pA current level (see the expanded traces). The resulting channel behavior was more like that in cells treated with Tg alone (compare with Fig. 1C). Additionally, the major current level seen in the all-point histogram, Ϫ1 pA (Fig. 4B), was similar to that seen in CCh-or Tg-stimulated cells. The I-V relationship of the Ϫ1 pA current (Fig. 4C) revealed a linear fit with a single channel conductance of 20 pS, which was not different from that obtained in cells stimulated by either Tg or CCh. The mean open times calculated for CCh ϩ Tg or Tg were significantly longer than that seen with CCh (p Ͻ 0.05, n ϭ 6 for each group (see Table I). NP o of the channel with CCh was significantly (p Ͻ 0.05, n ϭ 6) higher than the channel with CCh ϩ Tg or Tg alone. These data demonstrate that inhibition of SERCA activity in CCh-stimulated HSG cells induces a dramatic change in the activity of SOCC. We suggest that the "flicker" mode of SOCC seen with CCh is due to the following sequence of events: (i) activation by IP 3 -dependent release of Ca 2ϩ from the internal Ca 2ϩ store, (ii) Ca 2ϩ -dependent feedback inhibition of the IP 3 R and SOCC, (iii) SERCA-dependent uptake of Ca 2ϩ into the store, and (iv) refill-dependent inactivation of SOCC. Thus, inhibition of SERCA by the addition of Tg prevented the refill-dependent inactivation of SOCC and increased the open time.
To determine whether the flickery channel activity in CChstimulated cells is due to recycling of Ca 2ϩ in the store, channel activity was examined with 2 mM Na ϩ and 100 mM NMDG in the pipette solution. In this condition, Na ϩ enters the cell via SOCC but is not be taken up into the stores. Thus, the channel should not be subject to effects due to recycling of Ca 2ϩ . Single channel activities in CCh-and Tg-stimulated cells are shown in Fig. 5. Note that the current amplitude, as expected, is less than that seen in Fig. 2 where the pipette solution contained 100 mM Na ϩ . Importantly, channel activity in CCh-stimulated cells did not show a flickery behavior (Fig. 5A), but rather displayed a burst-like pattern similar to that in seen in cells stimulated by Tg (Fig. 5B). Thus, gating properties of the CCh-stimulated channel was dramatically altered when Na ϩ was used as the permeant cation. These data are consistent with our suggestion that the difference in the behavior of SOCC in Tg and CCh-stimulated cells is determined by the local release and uptake of Ca 2ϩ in the store.
Effect of Ca 2ϩ Store Refill on SOCC-To further demonstrate that the effect of Tg on CCh-stimulated SOCC activity was due to inhibition of Ca 2ϩ store refilling, we altered store refilling by decreasing the ambient [Ca 2ϩ ] i using two different protocols. In the first, the cells were exposed to a Ca 2ϩ -free bath solution, keeping the Ca 2ϩ in the pipette solution as the only source of external Ca 2ϩ . The typical flicker behavior of SOCC seen with CCh in the control condition (Fig. 6A-1) was changed to one more like that seen with CCh ϩ Tg (compare traces in Fig. 6A-2 with Fig. 1C, all recordings are at Ϫ40 mV). When Ca 2ϩcontaining medium was reintroduced to the cells, the typical CCh-induced flicker current behavior was restored (Fig. 6A-3, note that the three traces in Fig. 5A were recorded from the same cell). Average mean open time for the CCh-induced chan-

FIG. 4. Effect of Tg on CCh-stimulated channel activity in HSG cells. A,
shows single channel activity following addition of Tg to CCh-stimulated cells. Note that the traces shown here were recorded from the same cell used for recording the activity prior to stimulation (Fig.  1A) and following CCh addition (Fig. 1B). An expanded trace is shown at the bottom of Fig. 4A. An all point histogram is shown in B and I-V curve is shown in C. The pipette solution used in these measurements is indicated in C. All other experimental conditions were similar to those described in the legend to  (Fig. 6C). While the characteristic longer open time of the Tg-activated channel was maintained (2.55 Ϯ 0.4 ms, n ϭ 7), NP o was significantly increased from 9.8 Ϯ 0.8 to 24.8 Ϯ 4.2%, and more openings were seen with the higher current level in BAPTA-loaded cells (see expanded trace in Fig. 6C). In aggregate, these data demonstrate that a decrease in [Ca 2ϩ ] i increases activation of SOCC by Tg. Thus, it can be suggested that the lower channel activity (lower NP o ) seen in Tg-stimulated cells is due to an inhibitory effect exerted by the relatively high [Ca 2ϩ ] i in the subplasma membrane region resulting from inhibition of the SERCA. However, the channels that are active display longer open times since refilling of the internal Ca 2ϩ store is prevented by Tg. Importantly, these data also show that NP o of SOCC does not depend on IP 3 or PIP 2 hydrolysis. Thus, our data suggest at least two types of inhibitory mechanisms in the regulation of SOCC, one by the local [Ca 2ϩ ] i and another by refilling of internal Ca 2ϩ stores. In addition, the closing of the channel in BAPTA-loaded TG-treated cells demonstrates spontaneous channel inactivation. More detailed studies will be required to fully describe the events involved in the inactivation of SOCC.
SOCC Activation by Lowering the [Ca 2ϩ ] in the Internal Ca 2ϩ Store-To exclude possible effects due to changes in [Ca 2ϩ ] i , we examined whether SOCC could be activated by directly decreasing the [Ca 2ϩ ] in the lumen of the internal Ca 2ϩ store by using the cell permeant Ca 2ϩ chelator, TPEN. Hofer et al. (29) have previously reported that TPEN induced activation of I CRAC in RBL-1 cells. As shown in Fig. 7A addition of 0.5 mM TPEN induced channel activity (6/9 cells). Addition of Tg or CCh to cells treated with TPEN did not change the pattern of channel activity (data not shown). The major current amplitude, about Ϫ1 pA (Fig. 7B), the I-V relationship (Fig. 7C), and Ca 2ϩ conductance, 22 pS, were similar to that seen in cells stimulated by either Tg or CCh. The pattern of the single channel activity was also comparable to that seen in cells treated with CCh ϩ Tg with NP o ϭ 16.1 Ϯ 1.9% and mean open time ϭ 2.75 Ϯ 0.8 ms. These data suggest that the same SOCC is activated by TPEN, Tg, and CCh. It is important to note that the NP o of SOCC activation with TPEN was higher than that with Tg and similar to that seen with CCh.
The ability of TPEN to attenuate release of calcium from internal stores was examined using fura-2 (Fig. 7D). In the absence of external Ca 2ϩ , TPEN induced a very small transient increase in the fura-2 fluorescence ratio. Addition of Tg or CCh after this increase, did not induce further release of Ca 2ϩ from internal stores. Note that when either Tg or CCh were added to HSG cells in a nominally Ca 2ϩ -free medium, a transient increase in fluorescence was seen with substantially higher peak fluorescence ratios (between 3.0 and 5.0, see Fig. 3). Thus, it is highly unlikely that the TPEN-induced small increase in [Ca 2ϩ ] i represents any substantial depletion of the internal store. Importantly, when TPEN was added to cells in the presence of external Ca 2ϩ , there was a significant increase in fura-2 fluorescence (Fig. 7E shows average fluorescence values 5 min after TPEN addition), demonstrating an increase in Ca 2ϩ influx. CCh or Tg did not induce additional increase in fluorescence in TPEN-treated cells. DISCUSSION This study describes the characteristics and regulation of an endogenous SOCC in HSG cells. Store-operated calcium channels have been identified in almost all non-excitable cells and in some excitable cells. However, there is relatively little information regarding the single channel properties of this type of calcium channel or the mechanisms that regulate its gating. According to the capacitative Ca 2ϩ entry hypothesis, SOCC is activated when internal Ca 2ϩ stores are depleted and inactivated when these stores are refilled (1,6). In addition, SOCC activity is decreased by elevations of intracellular [Ca 2ϩ ], although the exact mechanism by which Ca 2ϩ induces this feedback inhibition is not yet known (3). The data presented above demonstrate that gating of SOCC is determined by the [Ca 2ϩ ] in the ER lumen and in the region of the cell where Ca 2ϩ influx occurs. Importantly, inhibition of SERCA function dramatically alters the single channel properties of SOCC. Since SERCA-mediated uptake of Ca 2ϩ into the ER regulates [Ca 2ϩ ] in the ER and in the subplasma membrane region near the site of SOCC, we suggest that SERCA function controls the gating of SOCC.
We report that calcium-permeable cation channel activity is stimulated in HSG cells when internal Ca 2ϩ stores are depleted either in the absence, by using Tg, or in the presence, by using CCh, of an increase in IP 3 . Based on the similarities in their pharmacological characteristics, cation conductances, and rel- ative cation permeabilities, we conclude that the same SOCC is activated by CCh and Tg. In the presence of physiological levels of external Na ϩ and Ca 2ϩ and at negative membrane potentials the channels activated by Tg and CCh primarily allow Ca 2ϩ influx. In the absence of Ca 2ϩ and Mg 2ϩ in the pipette solution, the channel is permeable to Na ϩ or Ba 2ϩ , but not to NMDG. The respective conductances of the channel for Ca 2ϩ , Na ϩ , or Ba 2ϩ in CCh-stimulated cells are similar to those in Tg-stimulated cells. The relative current amplitudes are in the order Na ϩ Ͼ Ca 2ϩ ϭ Ba 2ϩ and are similar in CCh-and Tg-stimulated cells. The Na ϩ conductances of the internal Ca 2ϩ release-activated Ca 2ϩ channel (CRAC), in T lymphocytes (15), and the more recently reported CaT1 (24), are similar to what we have reported here. Furthermore, external Ca 2ϩ also blocks influx of monovalent cations via CRAC and CaT1 (15,23,24). However, several other properties of CRAC and SOCC appear to be quite distinct from each other (3,14). Unlike SOCC, CRAC displays a very small Ca 2ϩ conductance, is relatively selective for Ca 2ϩ , and is blocked by Zn 2ϩ . We have shown here, and earlier (13), that in HSG cells SOCC is blocked by Gd 3ϩ and La 3ϩ but not by Zn 2ϩ . This divalent cation sensitivity appears to be more similar to those described for Trp-1, Trp-3 (11,30,31), and CaT1 (32). Other evidence that the channels activated by CCh and Tg are store-operated is their inhibition by 2-APB and xestospongin C, which have been suggested to be relatively specific for SOCC and also possibly for some Trp channels (20,26,27).
A significant, and novel, observation made in this study is that SOCC displays markedly distinct single channel properties depending on how it is activated. When stimulated by CCh, it displays a flickery activity, with a relatively higher P o and shorter mean open time, while in Tg-stimulated cells it displays burst-like activity with relatively longer mean open time, although with a lower P o . However, when Na ϩ is used as the permeant cation instead of Ca 2ϩ , (i) CCh-stimulated channel does not display flickery activity, and (ii) the pattern of CCh and Tg stimulated activities are similar. Thus, the flickery channel activity seen in CCh-stimulated cells represents a unique gating characteristic of this channel, which is related to Ca 2ϩ . We suggest that this flickery activity is due to rapid activation and inactivation of SOCC by the following sequence of events; Ca 2ϩ release from the internal store, Ca 2ϩ -depend-ent inactivation of IP 3 R, and re-accumulation of Ca 2ϩ into the store. Consistent with this, we have shown that inhibition of SERCA activity in CCh-stimulated cells by Tg converts the flickery SOCC activity to more stable openings, with longer mean open times. A similar effect on CCh-stimulated SOCC activity is also seen when Ca 2ϩ store refilling is decreased by BAPTA loading of the cell or Ca 2ϩ is removed from the bath solution. SOCC is also activated when [Ca 2ϩ ] in the ER lumen is lowered by treating cells with the permeant divalent cation chelator, TPEN. Importantly, it displays a similar burst-like pattern of activity as seen with Tg. In aggregate, these findings clearly demonstrate that internal Ca 2ϩ store refilling, i.e. the Ca 2ϩ content of the internal store regulates the gating of SOCC. The present findings can also explain previous reports (13,33,34) showing that intracellular [Ca 2ϩ ] i buffering is required to detect agonist stimulation of the store-operated Ca 2ϩ current, whereas Na ϩ currents can be more readily measured. It is likely that in the absence of intracellular [Ca 2ϩ ] i buffering there is recycling of the internal store Ca 2ϩ due to the function of IP 3 R and SERCA which allows some degree of internal store refilling, thus preventing full activation of the current.
Another important factor that influences the gating of SOCC appears to be the local [Ca 2ϩ ] in the vicinity of calcium influx. We have shown that when SOCC is stimulated by CCh or CCh ϩ Tg, it displays a higher NP o than when stimulated by Tg. This suggests that conditions leading to PIP 2 hydrolysis and elevation of IP 3 might be more efficient in the activation of SOCC. However, when [Ca 2ϩ ] i is buffered in cells by loading with BAPTA-AM, NP o of SOCC stimulated by Tg is significantly increased and is similar to that of SOCC stimulated by CCh. Importantly, BAPTA-AM loading of HSG cells, under the conditions described above, does not, by itself, activate SOCC. Consistent with these findings, NP o of SOCC in CCh-stimulated cells was decreased by addition of Tg, although it was not affected by buffering [Ca 2ϩ ] i . These data strongly suggest that the NP o of SOCC following internal Ca 2ϩ store depletion appears to be determined by store depletion and by [Ca 2ϩ ] i rather than the increase in IP 3 or PIP 2 hydrolysis. We have reported earlier that SERCA inhibition by Tg induces an increase in the [Ca 2ϩ ] i in the subplasma membrane region in HSG cells (13,21). We suggest that this increase in the local [Ca 2ϩ ] i exerts an inhibitory effect on SOCC. This accounts for the low P o of SOCC seen in Tg-treated cells and for the Tg-induced decrease in SOCC activity in CCh-stimulated cells. Feedback inhibitory effects of [Ca 2ϩ ] i on store-operated calcium entry in HSG (13) and other cells (3) has been previously reported. This negative effect of Ca 2ϩ on SOCC might be exerted directly on the SOCC protein itself, either via binding of Ca 2ϩ to the channel protein or mediated via a Ca 2ϩ -dependent protein such as calmodulin. Irrespective of the mode of inhibition, these data suggest that for maximal SOCC activity, [Ca 2ϩ ] i in the subplasma membrane region where Ca 2ϩ influx occurs must be maintained at a low level.
The present and our previous studies (13,21) demonstrate that SERCA has a major role in the regulation of [Ca 2ϩ ] i in the subplasma membrane region of the cell. Based on the data discussed above, we suggest that SERCA function is critical in the gating of SOCC. Furthermore, following agonist-stimulated internal Ca 2ϩ store depletion, at which time SOCC is activated, SERCA activity is increased as a result of a decrease in lumenal [Ca 2ϩ ] i (35) and other regulatory mechanisms such as phosphorylation (36). Thus, it appears that cells modulate the activity of SERCA to accommodate for the Ca 2ϩ entering the cells via SOCC. This ensures that (i) [Ca 2ϩ ] i near the SOCC is maintained at low levels and (ii) internal Ca 2ϩ store(s) are not fully depleted. The latter is likely an important protective mechanism for the cell since the cell undergoes a "stress" response when internal Ca 2ϩ store are depleted for extended time periods (37). Interestingly, it appears that in agonist-stimulated cells, incomplete depletion of the store which prevents full activation of SOCC is compensated by keeping [Ca 2ϩ ] i low, which increases the NP o of SOCC. Thus, SOCC is regulated in a complex way by the [Ca 2ϩ ] in two areas of the cell, in the lumen of the internal Ca 2ϩ store and in the subplasma membrane region at the site of Ca 2ϩ influx. Importantly, [Ca 2ϩ ] in both these areas of the cell are primarily regulated by the function of SERCA. The present data do not exclude an additional role for the plasma membrane calcium pump or mitochondria in the regulation of [Ca 2ϩ ] near the plasma membrane (38,39). In fact, local [Ca 2ϩ ] i buffering by mitochondria has been suggested to be involved in regulating the gating of I CRAC in lymphocytes (39).
In conclusion, we have described the single channel properties of SOCC in HSG cells. As predicted by the capacitative Ca 2ϩ entry hypothesis this channel is activated by store depletion and its activity is determined by refilling of internal Ca 2ϩ store(s). In addition, SOCC activity is also inhibited by increases in subplasma-membrane [Ca 2ϩ ] i . Based on our present and previous data, we conclude that SERCA, by regulating both internal Ca 2ϩ store refilling and [Ca 2ϩ ] i increase, controls the gating of SOCC. We have previously reported that Ca 2ϩ entering HSG cells is rapidly taken up into the internal store with minimal diffusion in the subplasma membrane region (21). Thus, it is reasonable to hypothesize that some region of the ER is localized very close to the plasma membrane, which facilitates the regulation of [Ca 2ϩ ] i near the SOCC by SERCAmediated uptake of Ca 2ϩ into the lumen of the ER. This juxtaposition of ER and the plasma membrane could also enable relaying the signal from the store lumen to the plasma membrane to either activate or inactivate SOCC. Further studies will be required to elucidate the nature of this critical mechanism.