Stable Activation of Single Ca 2 1 Release-activated Ca 2 1 Channels in Divalent Cation-free Solutions*

,

In a variety of cells, binding of ligands such as neurotransmitters, hormones, or growth factors to receptors on the cell surface generates intracellular calcium signals. Receptor-mediated release of Ca 2ϩ from IP 3 1 -sensitive intracellular stores triggers an influx of Ca 2ϩ into the cell via store-operated channels (SOCs) in the plasma membrane, a process termed "capacitative calcium entry" (1,2). The resulting increase in intracellular Ca 2ϩ concentration regulates important processes, ranging from cell growth and differentiation to apoptosis and cell death. Molecular candidates for SOCs are the mammalian homologues of the Drosophila TRP (transient receptor potential) protein (3). Expression of the related genes, however, has resulted in varying patterns of response, including constitutive activity, augmentation of capacitative calcium entry, apparent direct activation by IP 3 , and activation by diacylglycerol (4). Therefore, the mechanism by which emptying of intracellular Ca 2ϩ stores activates SOCs remains elusive. Recent findings have favored a "direct coupling" model, whereby IP 3 receptors in the store membranes sense the Ca 2ϩ -filling status of the stores and transfer this information via direct coupling to subunits of SOCs in the plasma membrane (5)(6)(7)(8).
For the present, the best knowledge of SOCs comes from electrophysiological studies of a specific current associated with SOCs, known as the Ca 2ϩ release-activated calcium current (I crac ) (9). Despite extensive characterization of the macroscopic CRAC currents, the single-channel signature is poorly understood. This is because the channels are presumed to have a minute single-channel conductance, too low to be resolved at the single-channel level (10). During whole cell measurements in Jurkat T-lymphocytes, recently, Kerschbaum and Cahalan (11) were able to detect for the first time single-channel events which they attributed to CRAC channels. Their approach was to measure currents in the complete absence of intra-and extracellular divalent cations. They reasoned that under these conditions CRAC channels, like voltage-gated Ca 2ϩ channels should pass monovalent cations indiscriminately and with enhanced single-channel conductance (12)(13)(14). Indeed, in the absence of divalent cations large nonselective cation currents were observed, and during the initial stage of activation, singlechannel conductances of 36 to 40 pS were measured in whole cell mode (11). However, the relationship of these large nonselective cation currents to the process of store depletion, and thus their identity as CRAC channels, has not been definitively established.
Here we report similar single-channel activities measured for the first time in excised plasma membrane patches, using a mast cell line (RBL-2H3 m1) and divalent cation-free solutions. Our whole cell data are similar in some respects to those from Jurkat T-lymphocytes. In addition, because of direct access to the cytoplasmic surface of the channels, we have elucidated kinetic and regulatory properties of the channels. In particular, we have sought to determine whether or not these cation channels truly represent CRAC channels. In divalent cation-free solutions excised channels of 25 to 39 pS conductance showed stable activation, independent of IP 3 or IP 3 receptors, and were blocked by 2-APB, a relatively specific inhibitor of SOC activation (15,16). The effects of 2-APB, as well as the modulation of channel activity by Ca 2ϩ provide evidence that the Na ϩ -conducting channels are in fact single CRAC channels. Hence, this report offers new insight into the possible regulation mechanism of native SOCs and provides a novel approach to investigate these channels under low divalent conditions.

MATERIALS AND METHODS
Cell Culture-Rat basophilic leukemia cells (RBL-2H3 m1), an immortalized mucosal mast cell line expressing m1 muscarinic receptors, were obtained from Dr. M. Beaven, National Institutes of Health (17). The cells were cultured in Earle's minimal essential medium with Earle's salts, 10% fetal bovine serum, 4 mM L-glutamine, 50 units/ml penicillin, and 50 mg/ml streptomycin (37°C, 5% CO 2 ). For experi-* 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. ments, cells were plated onto glass coverslips and used 12-36 h there after.
Electrophysiology-Patch clamp experiments were performed at 20 -22°C in the tight-seal whole cell, cell-attached and inside-out configurations (18). Patch pipettes were pulled from borosilicate glass (Corning glass, 7052) and fire polished. Membrane currents, filtered at 1-2 kHz, were recorded using an Axopatch-200B amplifier (Axon Instruments, Burlingame, CA). Voltage clamp protocols were implemented and data acquisition performed with pCLAMP 7.0 Software (Axon Instruments). Solution changes were accomplished by bath perfusion. The time required for a complete change was around 2 s. All voltages were corrected for a liquid junction potential.
For whole cell experiments with Ca 2ϩ as charge carrier, unless stated otherwise, the patch pipette (2-5 megaohm) solutions had the following composition (in mM): 140 Cs ϩ -aspartate, 2 MgCl 2 , 1 MgATP, 10 Cs ϩ -BAPTA (with free calcium set to 100 nM, calculated using MaxChelator software, version 6.60), 10 HEPES (pH 7.2 with CsOH). The bath solutions contained (in mM) 140 NaCl, 4.7 KCl, 10 CsCl, 1,13 MgCl 2 , 10 glucose, 10 CaCl 2 , 10 HEPES (pH 7.2 with NaOH). Divalent free whole cell measurements, with Na ϩ as charge carrier, were done with pipettes containing (in mM) 128 Cs ϩ -aspartate, 12 Cs ϩ -BAPTA, 0.9 CaCl 2 (free Ca 2ϩ ϳ5 nM), 10 HEPES (pH 7.2 with CsOH), and bath solutions containing (in mM) 150 Na ϩ methane sulfonate, 2 EDTA, 10 HEPES (pH 7.2 with NaOH). CRAC channels were opened by passive storedepletion with high BAPTA (12 mM) in the pipette solution, or by addition of ionomycin or thapsigargin to the bath. Cells were held at a potential of 0 mV. Every 1, 2, or 5 s either voltage ramps from Ϫ100 to ϩ60 mV or voltage steps from 0 to Ϫ100 mV were delivered for 200 ms. Currents were sampled at 5 kHz during voltage ramps and 25 kHz during voltage steps. All whole cell data were corrected for leak currents. Leak current was collected in response to voltage ramps or steps immediately after establishing the whole cell configuration.
During cell-attached and inside-out recordings data were collected from 10-s records at the given membrane potential, digitized at 5 or 10 kHz and filtered digitally for analysis and presentation. The pipette (5-10 megaohm) solutions contained (in mM) 150 NaCl, 2 EDTA, 10 HEPES (pH 7.2 with NaOH). In cell-attached experiments the bath solution contained (in mM) 145 KCl, 5 NaCl, 10 MgCl 2 , 10 HEPES (pH 7.2 with KOH) to nullify the cell's resting potential. 90 s before excision of the patch, the bath was perfused with intracellular solution, which contained (in mM) 145 K ϩ -glutamate, 5 NaCl, 2 EDTA, 10 HEPES (pH 7.2 with KOH). Single-channel analysis was performed with the pCLAMP 6 software. p o values were calculated for consecutive 1-s periods.

CRAC Currents and Monovalent Currents in RBL-2H3 m1
Cells-Ca 2ϩ release-activated Ca 2ϩ currents can be activated by a variety of procedures that share the common property of emptying intracellular IP 3 -sensitive stores. Stores can be de-pleted actively by exposure to receptor agonists that elevate IP 3 levels, external application of Ca 2ϩ -ionophores like ionomycin, or the presence of IP 3 in the patch pipette solution (19). Passive methods for store depletion rely on a constitutive and ill-defined leak of Ca 2ϩ from stores. If store refilling is prevented by inhibition of SERCA pumps or by high concentrations of cytoplasmic Ca 2ϩ chelators, the stores gradually lose their Ca 2ϩ .
In the whole cell measurements shown in Fig. 1, we used the latter method and activated CRAC channels by breaking into the cell, while including high concentrations of the Ca 2ϩ -chelator, BAPTA (12 mM), in the pipette solution. With external Ca 2ϩ present (Fig. 1a), Ca 2ϩ ions permeate through the open CRAC channels and the Ca 2ϩ -influx during progressive channel activation could be monitored as a developing small inward current, of 30 to 60 pA (2-3 pA/pF) (n ϭ 5). Single-channel currents underlying the macroscopic Ca 2ϩ currents were too small to be resolved, even at Ϫ120 mV membrane potential.
With divalent free solutions and Na ϩ as the permeant ion, however, single-channel events have been observed during whole cell measurements in Jurkat T-lymphocytes (11). We first sought to determine whether similar currents could be observed with rat basophilic leukemia (RBL-2H3 m1) cells. The current development in Fig. 1b illustrates activation of a large inward current under divalent cation-free conditions (n ϭ 6). The latency for current activation after break in, and the development time to reach the peak macroscopic current, were characteristically 1.5 to 2 times longer for the current in divalent-free solutions, than for I crac in Ca 2ϩ -containing solutions. In the absence of divalent cations, dialysis of the cell with the pipette solution typically activated macroscopic currents up to 2000 pA (ϳ130 pA/pF), or ϳ40-fold greater than I crac . The current-voltage relationship of the fully developed current showed a modest inward rectification. External addition of 10 M Mg 2ϩ inhibited the monovalent currents in a voltage-dependent manner (data not shown).
During the initial activation of the large cation current (see Fig. 1b), we were able to detect single-channel openings in RBL-2H3 m1 cells (Fig. 1c). We recorded single-channel inward currents of 3.3 to 4 pA, during voltage steps to Ϫ100 mV, giving conductances of 33 to 40 pS (n ϭ 6, assuming a reversible potential of 0 mV, see below). Based on the whole cell current amplitudes and the unitary current of the single-channels, we calculated the number of channels per cell to be 260 to 500 FIG. 1. CRAC currents and monovalent currents in RBL-2H3 m1 cells. a and b, during whole cell recordings (representative plots are shown), intracellular Ca 2ϩ stores were passively depleted with 12 mM BAPTA after going whole cell at 0 s. Currents, measured during voltage-steps to Ϫ100 mV were plotted versus time, either with 10 mM Ca 2ϩ (n ϭ 5) (a) or 150 mM Na ϩ (n ϭ 6) (b) as charge carriers. The solutions used in a and b were the same, except, for the Ca 2ϩ recordings, EDTA in the bath was substituted with 10 mM Ca 2ϩ . b, only in divalent free solutions, single-channel events could be resolved during the voltage steps. c shows a sample trace, taken from a voltage step to Ϫ100 mV, applied at 287 s (star) in the measurement in b. The unitary current amplitude is ϳ4 pA. channels (n ϭ 6). This suggests an average surface density of ϳ0.22 channels per m 2 .
Stable Activation of Single Cation Channels in Excised Plasma Membrane Patches-To gain access to the cytoplasmic surface of the channels, and to limit the number of investigated channels, we attempted to record the activity of individual channels in excised plasma membrane patches. Sodium ions (150 mM) were used as the primary charge carriers. Starting from cell-attached patches with no channel activity, we excised the patches into divalent cation-free solutions, while clamping the cell's membrane potential to Ϫ73 mV. In 65 out of 95 inside-out patches this procedure induced a channel activity, with an average conductance of 31 pS (25 pS to 39 pS) (Fig. 2, a and b). We conclude that the channel activity seen after excision corresponds to the channels underlying the macroscopic current activated by divalent cation-free solutions; the single-channel conductance found in excised patches was similar to the conductance recorded in the whole cell mode (Fig.  1c), the channels appear to have similar lack of selectivity, and the channels have similar, high open probability under total divalent cation-free conditions. When patches were excised into divalent-free media, the channels apparently needed no additional factors to be activated, such as IP 3 or exogenously added IP 3 receptors (7,20). However, if fragments of endoplasmic reticulum stores were still attached to the patches, the absence of ATP, Ca 2ϩ , and Mg 2ϩ from the intracellular solutions could have induced store depletion, and thus caused or contributed to channel activation.
Continuous channel activity began an average of 70 to 80 s after excision (Fig. 2c) and was stable during recording time for at least 20 min, even after extensive washing. In 32% of the experiments the progressive opening of 6 and more channels in a patch could be observed, which may suggest clustering of CRAC channels in specific membrane domains.
Single-channel Properties of Excised Channels, Activated by Divalent Cation-free Solutions-We examined the single-channel properties of channels in inside-out patches, in divalent cation-free solutions. Fig. 3a shows recordings of 2 active channels in an excised inside-out patch, at different membrane potentials. The channels had a high open probability (0.94 Ϯ 0.03 at Ϫ113 mV, n ϭ 5). The duration of the closures from the open state decreased with hyperpolarization, from ϭ 1.19 Ϯ 0.13 (Ϯ S.E.) ms at Ϫ73 mV to ϭ 0.50 Ϯ 0.06 (Ϯ S.E.) ms at Ϫ113 mV (n ϭ 7 cells). From traces such as those shown in Fig.  3a, current amplitudes were measured manually or determined from all points amplitude histograms (Fig. 3b), over a voltage range from Ϫ113 to ϩ67 mV. The resulting currentvoltage relationships (Fig. 3c) gave an average single-channel conductance of 31 pS (25 to 39 pS). Moreover, the IV-relationship was linear with a reversal potential close to 0 mV. As the major cation in the intracellular solution was K ϩ , while that in the pipette was Na ϩ , this is indicative of a channel that passes monovalent cations indiscriminately (12). The IV-relationship is similar to that for the whole cell current, except the latter is slightly inwardly rectifying, due to a small effect of voltage on open probability (not shown). Another specific channel feature, the mean open time, was calculated to be 14.9 Ϯ 1.2 (Ϯ S.E.) ms (n ϭ 7) (Fig. 3d).
Effect of Ca 2ϩ -store Depletion on the Activation of Monovalent Currents and CRAC Currents-The results in Fig. 1 show that the activation of the large cation current by divalent cation-free solutions was considerably slower than the activation of I crac , despite the expectation that passive store emptying by BAPTA should occur at a similar rate in the two conditions. This would indicate that the activation of cation channels by divalent cation-free solutions involves a mechanism other than, or in addition to store depletion. Alternatively, it is possible that the absence of divalent cations slows the mechanism for coupling store depletion to channel activation. To address these possibilities, we carried out experiments in which Ca 2ϩ -stores were emptied and CRAC channels thus activated prior to going whole cell or excision into divalent cation-free media. Before sealing on the cells, we applied either 1 M of the SERCA inhibitor, thapsigargin plus 500 nM of the ionophore ionomycin, or 500 nM ionomycin alone. The incubation time with these drugs was minimally 5 min, by which time stores should be completely depleted and the currents should have fully developed (21). Subsequently, we patched the cells, and established either the whole cell configuration with either Ca 2ϩ or Na ϩ as charge carriers, or excised the patch (inside-out mode) at Ϫ73 mV V m into divalent-free solutions. In the whole cell experiments, the pipette solutions included high concentrations of BAPTA to passively deplete stores in the control cells.
FIG. 2. Single-channel activity in excised plasma membrane patches. a, single-channel activity was elicited in 65 out of 95 attempts by excision of patches into low buffered divalent cation-free solutions, at Ϫ73 mV V m in the inside-out mode. b, a sample trace of an excised patch at Ϫ73 mV V m is shown at a higher resolution, for comparison with the whole cell single-channel recording in Fig. 1c. c, Continuous channel activity in the insideout mode started, under the given conditions, after an average time of 70 -80 s (n ϭ 34). Following convention, unitary inward currents are displayed as downward deflections.
Prior depletion of Ca 2ϩ -stores resulted in preactivation of I crac with about 2 pA/pF (7/7) when measured with Ca 2ϩ as the charge carrier (Fig. 4, a and b, inset). However, directly after switching to Na ϩ in divalent free solutions, the current only increased 4-fold (Fig. 4b, inset), corresponding to about 10% of activation compared with the peak macroscopic divalent free currents (Fig. 4b). This 4-fold activation may correspond to the current seen in Jurkat cells by Kerschbaum and Cahalan (11) in the absence of external divalent cations, but with Mg 2ϩ present in the cytoplasm. However, in the RBL-2H3 cells no larger, transient inward current was observed. Similar kinetics for RBL-1 cells have been reported by Fiero et al. (22). Following this 4-fold jump, which was only seen in the preactivated cells, a slower activation to the full, maximally conducting state was observed, and the kinetics of this process were similar in control and preactivated cells. Likewise, the appearance of single-channel openings following excision into divalent cationfree solutions occurred with a similar delay when stores were previously emptied (Fig. 4c).
Taken together, these results demonstrate that the delay time for whole cell current activation by divalent cation-free solutions, as well as the delay time for channel activation following excision into divalent cation-free solutions does not appear to depend on depletion of Ca 2ϩ -stores. However, they do not rule out a requirement for initial store depletion, followed by a subsequent amplification step, which is rate-limiting. Nonetheless, the mechanism by which observable single-channel activation occurs appears to be at least somewhat distinct from the mechanism of activation of I crac by store depletion. Note, however, that activation of an intermediate activity state of the channels, similar to that seen with low divalent cations outside and with Mg 2ϩ inside (11,22), did appear to be hastened by prior depletion of Ca 2ϩ -stores. The significance of this observation will be considered subsequently under "Discussion." Modulation of Channel Activity in Excised Patches by Internal Ca 2ϩ , Mg 2ϩ , and pH-To further address the issue of whether the channels activated by divalent cation-free solutions represent CRAC channels, we next examined the modulation of these channels by internal divalent cations and by pH to compare their behavior which was shown previously for CRAC channels.
Krause et al. (23) reported from whole cell studies in RBL-1 cells, that I crac activation changed from spontaneous to store-operated mode, if the cytosolic [Ca 2ϩ ] is elevated to more than 50 nM. The spontaneous I crac was inactivated at a resting cytosolic [Ca 2ϩ ] of 105 nM. On the basis of their finding, after reaching full channel activity in the inside-out mode, we perfused the bath with intracellular solution containing 100 nM free Ca 2ϩ (Fig. 5a). This concentration completely inhibited channel activity in 50.5 Ϯ 9.5 s in 4 patches, while in 3 other patches channel activity was unaffected. In the 4 patches in which channel activity was inhibited, restoration of low Ca 2ϩ conditions reversed the inhibition in only 1 of the 4 experiments. The inhibitory effect of Ca 2ϩ was rather specific and not due to inhibition of monovalent currents by divalent ions per se, because similarly low concentrations of free Mg 2ϩ failed to inhibit. However, if 100 M Mg 2ϩ or more was added to the inner surface of the active channels, inhibition developed in 13.4 Ϯ 3.1 s (12 out of 12 patches) (Fig. 5b). The activity was restored by divalent free buffers (not shown). In both cases, as the inhibition developed, single channel conductance was unaffected indicating that the divalent cations suppress open probability.
In the whole cell mode in Jurkat T-lymphocytes, the inactivation of Na ϩ current by intracellular Mg 2ϩ could be reduced by increasing the intracellular pH (14). The experiment in Fig.  5c demonstrates a similar effect of increasing pH for the channels in excised patches. After inhibition by 100 M Mg 2ϩ (pH 7.2), channel activity could be rescued by increasing the pH in the perfusion solution to pH 8.2 (4 out of 7 cells). The time needed for complete reversal in the cells that responded to the pH change was 42 Ϯ 10 s. We also carried out experiments similar to those described by Kerschbaum and Cahalan (14) in which the channels were inhibited by addition of a much higher concentration of Mg 2ϩ (2.4 mM). Channel activity was blocked in 7 of 7 experiments, but in three attempts there was no recovery on shifting to pH 8.0 (not shown).
From the experiments depicted in Fig. 5, some interesting observations can be made. First, the results provide evidence that the single-channels are the channels underlying the large whole cell current in divalent cation-free solutions. Second, the inhibition of the channels with Ca 2ϩ and Mg 2ϩ concentrations in the physiological range explains the inability to detect the same channels in the cell-attached mode. In the cell-attached mode, neither addition of 1 M thapsigargin (n ϭ 14) nor addition of 500 nM ionomycin (n ϭ 13) evoked detectable singlechannel activity. Note that under these conditions, whole cell currents (I crac , pA magnitude) can be seen indicating that single channels must be opening; but these channels are not detectable because in the presence of Ca 2ϩ or Mg 2ϩ the single channel conductance and/or frequency of opening is below the level of detection. Third, the ability of low, physiological concentrations of Ca 2ϩ to inhibit the channels is consistent with their identity as CRAC channels.

Effects of the IP 3 -receptor Antagonist 2-APB (2-Aminoethoxydiphenyl Borane) on I crac Activity-Recent reports have sup-
ported a conformational coupling model for SOC activation, whereby IP 3 receptors in the endoplasmic reticulum sense the Ca 2ϩ filling status of the stores and transfer this information via direct coupling to SOCs in the plasma membrane (5-7). Ma et al. (16) concluded from their studies, using the membrane permeant IP 3 receptor inhibitor 2-APB that IP 3 receptors are essential not only for opening of SOCs, but also for maintaining their activation. On this basis, we examined the effects of 2-APB on macroscopic currents, and on channel activity in excised patches.
The results shown before in Fig. 2 demonstrate that no added IP 3 or added IP 3 receptors are necessary to maintain activity of CRAC channels in excised patches under divalent cation-free conditions. This is in contrast to previous reports demonstrating that Trp-3 channel activity (24) and store-operated channel activity in A431 cells (7) were rapidly lost in the absence of added IP 3 and IP 3 receptors. This may indicate that once activated under divalent-free conditions, CRAC channel activity is independent of Ca 2ϩ -store depletion and coupling to IP 3 receptors (although we cannot unequivocally rule out the stable retention of fragments of endoplasmic reticulum and IP 3 receptors). Thus, we were initially not surprised to find that addition of 100 M 2-APB to the medium surrounding activated channels in excised patches in most instances failed to inhibit channel activity (for example, Fig. 6d). However, with this protocol 2-APB is added to the cytoplasmic side of the plasma membrane, while in previous studies of Ma et al. (16), the drug was always added to the outside. Experiments illustrated in Fig. 6,  a and b, show that in the whole cell configuration, with either Ca 2ϩ or Na ϩ as charge carrier, 2-APB blocked the current completely when applied externally, but had significantly less effect when applied in the pipette. Finally, although 2-APB blocked channel activity when applied to the cytoplasmic side of excised patches in only 2 of 11 experiments (Fig. 6d), if  (12 out of 12). c, the full channel activity could be rescued by increasing the pH to pH 8.2. The time needed was 42 Ϯ 10 s (4 out of 7). included in the patch pipette, channel activation did not occur in any of 10 separate experiments (Fig. 6c).
These results suggest that 2-APB may be acting as a direct blocker of SOCs, rather than as an IP 3 receptor antagonist as supposed in earlier studies. However, because of the protocols used, we considered the possibility that once the channels were activated by divalent cation-free solutions, they would lose their sensitivity to the effects of 2-APB as an IP 3 receptor inhibitor. Thus, we also added 2-APB to the external medium before establishment of a surface membrane seal with the patch pipette and therefore prior to excision (Fig. 6e). However, channel activation was seen in 3 of 5 patches in this series of experiments, and this is the expected frequency of successes from control experiments (about 70%). Thus, we conclude that at least in RBL-2H3 m1 cells, the site of inhibition of CRAC channels by 2-APB is external and may be on the channels themselves. It seems unlikely that the drug's action involves IP 3 receptors under the plasma membrane.
Although not as yet thoroughly investigated in this regard, 2-APB may be relatively selective for CRAC channels among plasma membrane Ca 2ϩ channels. The current study shows that CRAC channels are directly blocked by this drug, but L-type calcium channels (15), Trp-3 channels (16), and arachidonic acid-activated channels are not. 2 Thus, the sensitivity of the channels to 2-APB seen under divalent cation-free conditions can be taken as an additional piece of evidence that they do in fact represent a high conductance state of CRAC channels. Also, based on the stability of channel activity in divalent cation-free solutions, we conclude that when activated in this way, interactions between channels and IP 3 -liganded IP 3 receptors are not required for activation, or at least for maintenance of the activated state.

DISCUSSION
Our studies on RBL-2H3 m1 cells have shown that in the complete absence of intra-and extracellular divalent cations large macroscopic inward currents develop, when measured in whole cell mode. This behavior is similar to that described for Jurkat T-lymphocytes by Kerschbaum and Cahalan (11). During the initial phase of activation, single-channels with conductances of 33 to 40 pS could be detected, resulting in an estimate of 260 to 500 channels per cell, calculated from the ratio of macroscopic to unitary currents. By excision into divalent cation-free media we found similar channels in inside-out patches, with conductances of 25 to 39 pS (averaging 31 pS). Since the conductances for single-channels in excised patches and in whole cell measurements are similar to each other, and the channels have similar open probabilities and lack of selectivity, we conclude that they are the same channels.
Whole cell currents as well as currents in excised patches were blocked by the drug 2-APB, a relatively specific inhibitor of SOC activation (16). Channel activity in excised patches was also inhibited by low Ca 2ϩ concentrations, as low as 100 nM. Similarly low concentrations of Mg 2ϩ did not inhibit, showing that the inhibiting effect of Ca 2ϩ is specific, and not due to a general inhibition of monovalent currents by divalent cations. Furthermore, the inhibition of channel activity in excised patches caused by 100 M Mg 2ϩ could be rescued by increasing the pH in the intracellular solution. A similar pH effect was shown for whole cell monovalent currents in Jurkat cells (14); however, in contrast to the findings in Jurkat cells, in RBL-2H3 cells inhibition by higher (2.4 mM) concentrations of Mg 2ϩ could not be reversed.
We have also carried out experiments to investigate the role of intracellular Ca 2ϩ store depletion on the kinetics of activation of I crac and the large currents and single-channels activated under divalent cation-free conditions. Our results indicate that two currents were preactivated by store depletion, I crac and an intermediate-sized current which is similar to one seen with no divalent cations outside but with Mg 2ϩ inside. However, the large current in whole cell mode, and the singlechannel activity in excised patches did not develop more rapidly when stores were first depleted. So, given this independence of store depletion for the current and single-channel kinetics, what is the evidence that the channels underlying the large monovalent currents are in fact CRAC-channels? We believe that they are indeed CRAC channels due to the following observations. (i) The ability of low, physiological concentrations of Ca 2ϩ to inhibit the channels in excised patches is consistent with their identity as CRAC channels. We know of no other channel which is so exquisitely sensitive to inhibition by Ca 2ϩ . (ii) We found that the channels observed under divalent cation-free conditions are sensitive to 2-APB; this drug c, if 2-APB was included in the pipette solutions, in all experiments (10 out of 10) the channels remained silent after patch excision. d, 2-APB infrequently had an inhibitory effect (2 out of 11), when added to active inside-out patches. The patches remained active in 9 out 11 trials. e, cells were also pretreated with 2-APB for 5 min before sealing. After patch excision into divalent free solutions with 2-APB present, the channel activity remained in 3 out of 5 patches. Therefore, the success rate for activation was the same as in untreated cells (60 -70%). The closed state of the channels in c-e is indicated as c.
appears to be relatively selective for CRAC channels among membrane Ca 2ϩ channels in inhibiting SOC activation; for example, the drug does not block voltage-dependent calcium channels or arachidonic acid-activated channels (15). 2 Therefore, although the investigated channels seem likely to be CRAC channels, activation by store depletion may be unnecessary in the complete absence of divalent cations. Alternatively, activation by store depletion may precede subsequent further activation by divalent cation-free solutions such that the latter is rate-limiting. 3 Although the kinetics of currents activated by divalent cation-free solutions appear independent of store depletion, there is another line of evidence that may link these currents to the process of calcium store depletion. While the large, macroscopic current was not pre-activated by store depletion, a current much smaller than the maximal macroscopic, yet larger than I crac was observed in the pre-activated cells when extracellular divalent cations were removed. This current may correspond to the intermediate sized current seen by Kerschbaum and Cahalan (11) in the absence of external divalent cations, but with intracellular Mg 2ϩ present. In the latter case, it was shown that the intermediate current resulted from single channel openings identical to those underlying the larger macroscopic current. In our experiments, an intermediate current was activated by calcium store depletion (Fig. 4B, inset), and if the single channels underlying this current are the same as those for the macroscopic current, then this is further evidence that the large openings are CRAC channels. However, we have not been successful in observing single channel openings in RBL-2H3 cells with Mg 2ϩ present.
According to the conformational coupling model for SOC activation, IP 3 receptors are essential for activation and maintenance of channel activity (5,6,16). Recent findings have also implicated a requirement for IP 3 in this activation mechanism (7). However, there is data in conflict with this idea, at least in some cell types. For example, heparin is known not to inhibit SOC activation (25) or I crac (9), and SOC activation occurs in cells apparently devoid of IP 3 receptors (26). In the current study, once excised single-channels were activated by divalent cation-free solutions, they did not run down for at least 20 min. This stable activation, elicited solely by excision into divalent cation-free media, was independent of added IP 3 or IP 3 receptors. However, the previous studies implicating coupling of SOCs to IP 3 receptors were not performed in the complete absence of divalent cations. Thus, two distinct modes of activation may exist, one requiring coupling to IP 3 receptors, and one dependent on removal of divalent cations and independent of IP 3 receptors. This conclusion is similar to one drawn by Krause et al. (23) who concluded that low cytoplasmic Ca 2ϩ per se was sufficient to activate an I crac -like current in RBL-1 cells.
Another important finding from the current work, with significant impact on the interpretation of results in previous investigations, is the effect of 2-APB on the divalent cation-free activated CRAC channels. The relatively specific blockade of CRAC channels by 2-APB has until now been attributed to its antagonistic effect on IP 3 receptors (15), thereby blocking their conformational coupling to the channels (16). Our results from whole cell and excised patch measurements, however, demon-strate that 2-APB inhibits CRAC channels also under conditions where their activation and maintenance of activity appear not to be IP 3 or IP 3 receptor dependent. Additionally, we find that the site of inhibition of CRAC channels by 2-APB is likely on the external side of the channels themselves rather than on IP 3 receptors under the plasma membrane. However, Ma et al. (16) clearly demonstrated that 2-APB could only block TRP3 channels when activated through an IP 3 -dependent mechanism, and not when activated more directly by diacylglycerol. We have confirmed this finding with our own independently generated TRP3-expressing cell line. 2 Thus, while we doubt that 2-APB blocks CRAC channels by inhibiting IP 3 receptors, we consider it still an open question as to whether the drug acts by blocking CRAC channels directly, or by interfering with their mechanism of activation.
Taken together, the findings in the current study demonstrate that the current through single CRAC channels can be amplified and observed under divalent cation-free conditions in RBL-2H3 m1 cells. The mechanism of this activation is not known. It may result simply from the dissociation of divalent cations from regulatory sites on or near the channels, or it could be due to the dissociation of another factor (or factors) whose regulation of the channels depends on divalent cations. Soluble factors do not seem to be absolutely required for inactivation by Ca 2ϩ and Mg 2ϩ since these cations blocked channel activity when added several minutes after excision. However, we were generally unsuccessful in reversing the Ca 2ϩ block by returning the patches to a low-Ca 2ϩ medium. Furthermore, we found that following excision of patches in media containing 100 nM [Ca 2ϩ ], lowering of [Ca 2ϩ ] did not lead to the appearance of channel activity (data not shown). This may indicate that Ca 2ϩ binding to inhibitory sites is irreversible in the absence of a factor or structure that is lost following patch excision.
Regardless of the underlying mechanism, the use of divalent cation-free solutions has provided conditions for the first measurement and characterization of single CRAC channel activity in excised plasma membrane patches. Understanding the behavior of CRAC channels in the absence of divalent cations may provide an important first step for the design of future experiments addressing the mechanisms of regulation of CRAC channels under more physiological conditions.