Permeation of monovalent cations through the non-capacitative arachidonate-regulated Ca2+ channels in HEK293 cells. Comparison with endogenous store-operated channels.

In a manner similar to voltage-gated Ca(2+) channels and Ca(2+) release-activated Ca(2+) (CRAC) channels, the recently identified arachidonate-regulated Ca(2+) (ARC) channels display a large monovalent conductance upon removal of external divalent cations. Using whole-cell patch-clamp recording, we have characterized the properties of these monovalent currents in HEK293 cells stably transfected with the m3 muscarinic receptor and compared them with the corresponding currents through the endogenous store-operated Ca(2+) (SOC) channels in the same cells. Although the monovalent currents seen through these two channels displayed certain similarities, several marked differences were also apparent, including the magnitude of the monovalent current/Ca(2+) current ratio, the rate and nature of the spontaneous decline in the currents, and the effects of external monovalent cation substitutions and removal of internal Mg(2+). Moreover, monovalent ARC currents could be activated after the complete spontaneous inactivation of the corresponding SOC current in the same cell. We conclude that the non-capacitative ARC channels share, with voltage-gated Ca(2+) channels and store-operated Ca(2+) channels (e.g. SOC and CRAC the general property of monovalent ion permeation in the nominal absence of extracellular divalent ions. However, the clear differences between the properties of these currents through ARC and SOC channels in the same cell confirm that these represent distinct conductances.

Receptor-stimulated increases in Ca 2ϩ entry in non-excitable cells are known to play a pivotal role in the generation and maintenance of the intracellular Ca 2ϩ signals responsible for the control of such diverse functions as secretion, motility, growth, proliferation, and gene expression. However, despite extensive study, the mechanisms underlying such receptorstimulated Ca 2ϩ entry are currently far from clear. To date, most studies have focused on the so-called capacitative or storeoperated model in which activation of Ca 2ϩ entry occurs as a result of the emptying of intracellular Ca 2ϩ stores (1)(2)(3). Such entry is described as occurring via store-operated Ca 2ϩ (SOC) 1 channels (4,5). The ionic conductances associated with this store-operated entry were first characterized using patchclamp techniques in mast cells and Jurkat lymphocytes (6 -8), where the resulting whole-cell currents were identified as I CRAC (for calcium release-activated calcium currents). These CRAC channels therefore represent the archetypal capacitative or store-operated channel. Although store-operated Ca 2ϩ entry appears to be an almost ubiquitous feature of cells, the biophysical characterization of SOC channels from other cell types is rather limited, and the mechanism of activation of the channels is still unknown. Moreover, it seems unlikely that such channels are the exclusive route for the receptor-stimulated entry of Ca 2ϩ in non-excitable cells (9). Recently, we have identified a novel receptor-activated Ca 2ϩ entry pathway that appears to be specifically responsible for the Ca 2ϩ entry associated with agonist stimulation at physiologically relevant concentrations (10 -12). This pathway is independent of store depletion (i.e. non-capacitative) and is instead regulated by the receptor-mediated generation of arachidonic acid (AA). Similar AA-dependent non-capacitative Ca 2ϩ entry pathways have now been identified in a variety of different cell types, including Balb/c 3T3 fibroblasts (13), A7r5 smooth muscle cells (14), and mouse parotid cells 2 . Subsequent characterization of the Ca 2ϩselective conductance associated with this pathway (named I ARC , for arachidonate-regulated Ca 2ϩ current) in HEK293 cells stably transfected with the m3 muscarinic receptor (m3-HEK cells) revealed that it showed certain features that clearly distinguished it from the store-operated current (I SOC ) in the same cells (12).
Both ARC and CRAC channels, as well as the endogenous SOC channels of m3-HEK cells (HEK-SOC channels), share the properties of being highly Ca 2ϩ -selective with relatively small magnitude currents (ϳ0.5-1 pA/pF at Ϫ80 mV), voltage-independent gating, inward rectification, and very positive reversal potentials (4,5,7,12). The maintenance of the high selectivity for Ca 2ϩ shown by these channels (as well as by voltage-gated Ca 2ϩ channels) is problematic given that, under normal conditions, Na ϩ ions vastly outnumber Ca 2ϩ ions in the extracellular medium. Despite this high selectivity for Ca 2ϩ over Na ϩ under normal conditions, a characteristic feature of both CRAC channels and voltage-gated Ca 2ϩ channels is that lowering external divalent cation concentrations to the micromolar range results in the appearance of large monovalent currents through these channels (7,15,16). Evidence from voltage-gated Ca 2ϩ channels suggests that the high selectivity for Ca 2ϩ over monovalent ions seen under normal conditions does not result from a process of rejection as from a molecular "sieve," but rather from the binding of Ca 2ϩ ions to sites within the channel. In other words, the Ca 2ϩ selectivity of these channels is Ca 2ϩ -dependent. A similar conclusion was reached in an analysis of the monovalent permeability of CRAC channels (15,16). Interestingly, we observed a similar increase in monovalent ion permeability in the nominal absence of extracellular divalent cations in our earlier studies on the ARC channels (12). This raises the possibility that, despite the diverse mechanisms of gating shown by these different Ca 2ϩ -selective channels (depolarization versus store depletion versus arachidonic acid), they all appear to share similar mechanisms for Ca 2ϩ selectivity. In this study, we examine this feature in more detail with the overall aim of attempting to further develop a unique "fingerprint" for the ARC channels to provide the means of distinguishing them from the store-operated channels in the same cells.

EXPERIMENTAL PROCEDURES
Cell Culture-Cells from the human embryonic kidney cell line HEK293 that had been stably transfected with the human m3 muscarinic receptor were a generous gift from Dr. Craig Logsdon (University of Michigan). The cells were cultured in Dulbecco's modified Eagle's medium with 10% calf serum and antibiotics in a 5% CO 2 incubator at 37°C as previously reported (11). Cells were plated on glass coverslips that formed the bottom of a patch-clamp chamber (Warner Instrument Corp., Hamden, CT) at least 24 h before experimentation.
Whole-cell Patch-clamp Recordings-Patch-clamp recordings using the standard whole-cell mode (17) were performed at room temperature (20 -22°C) using an Axopatch-1C patch-clamp amplifier (Axon Instruments, Inc., Foster City, CA). Patch pipettes were pulled from borosilicate glass (GC150-F, Warner Instrument Corp.) and fire-polished to a resistance of 3-6 megaohms when filled with internal solution. Wholecell currents were recorded using 250-ms voltage steps from a holding potential of 0 to Ϫ80 mV delivered every 2 s. Alternatively, currentvoltage relationships were recorded using 150-ms voltage ramps from Ϫ100 to ϩ30 mV. Ramps were terminated at ϩ30 mV to avoid any contribution from a depolarization-activated Cl Ϫ current. Data were sampled at 20 kHz during the voltage steps and at 5.5 kHz during the voltage ramps and digitally filtered off-line at 1 kHz. Initial traces obtained upon going whole cell (i.e. before activation of I ARC or I SOC ) were averaged and used for leak subtraction of subsequent current recordings. Changes in the external (bath) solution were by perfusion of solution through the patch-clamp chamber (rate of ϳ0.5 ml/min). For the experiments investigating fast inactivation, peak and steady-state currents were compared during hyperpolarizing pulses to Ϫ80 mV. Examination of the capacitative currents during pulses to Ϫ80 mV indicated a mean time constant for the capacitative transient of 240 Ϯ 7.6 s (n ϭ 159). To minimize contributions from these transients, peak currents (average currents over a 1-ms period) were determined at a point 3 ms after the start of the hyperpolarizing pulse. Steady-state currents were similarly determined at a point 200 ms later.
Solutions-The standard pipette (internal) solution contained 140 mM Cs ϩ acetate, 1.22 mM MgCl 2 , 1.89 mM CaCl 2 , 5 mM EGTA, and 10 mM HEPES (pH 7.2 with CsOH). The free Ca 2ϩ concentration of this solution was calculated to be 100 nM as computed with Maxchelator (18). CaCl 2 or MgCl 2 was omitted for the experiments involving Ca 2ϩ -or Mg 2ϩ -free internal solutions, respectively. Where indicated, the Cs ϩ acetate was replaced with Na ϩ acetate or NMDG ϩ acetate as appropriate. The standard extracellular solution contained 140 mM NaCl, 1.2 mM MgCl 2 , 20 mM CaCl 2 , 10 mM D-glucose, and 10 mM HEPES (pH 7.4). Both MgCl 2 and CaCl 2 were omitted from this solution for the experiments involving nominally divalent ion-free external solutions, and the osmolarity (320 mosmol/liter) was maintained with additional D-glucose. Measurements of the Ca 2ϩ concentration in this nominally divalent ion-free external solution were made spectrofluorometrically using fura-2-free acid and compared against commercially prepared standards (Molecular Probes, Inc., Eugene, OR). The values obtained ranged from 0.12 to 0.6 M. Where indicated, the NaCl was replaced with CsCl, LiCl, or NMDG-Cl as appropriate.

RESULTS
Ca 2ϩ -selective ARC Channels Become Permeable to Monovalent Cations upon Removal of External Divalent Cations-In normal external medium containing Ca 2ϩ and Mg 2ϩ in millimolar concentrations, addition of exogenous arachidonic acid (8 M) to the bath solution resulted in the activation of a small inward current at Ϫ80 mV. Consistent with our previous report (12), this current displayed a marked inward rectification, a positive (greater than ϩ30 mV) reversal potential, and an absence of any fast inactivation and was inhibited by La 3ϩ (50 M). Substitution of external sodium with NMDG ϩ had negligible effects on the current, confirming a minimal permeability to Na ϩ under these conditions. All these features are consistent with the characterization of this arachidonate-induced current as a Ca 2ϩ -selective current (I ARC ).
Upon nominal removal of external divalent cations, a much larger inward current (measured at Ϫ80 mV) was seen to develop (Fig. 1A). Under normal conditions (i.e. Na ϩ in the bath solution and Cs ϩ in the pipette solution), this current increased to reach a maximum amplitude of 28.3 Ϯ 2.7 pA/pF (n ϭ 5) or ϳ50 times the Ca 2ϩ -selective current seen in normal divalent ion-containing external medium. The observed current was entirely dependent on the presence of AA, as no significant currents were seen to develop upon nominal removal of external divalent cations without prior addition of AA. In the continued absence of extracellular divalent ions, the AA-dependent current slowly declined in magnitude (50% reduction in 229 Ϯ 23 s) (Fig. 1A). In contrast, replacing the divalent ionfree bath solution with normal divalent ion-containing medium after the current had reached its maximum value resulted in a rapid inhibition of the current to levels consistent with those seen in normal medium (i.e. ϳ0.5 pA/pF) (data not shown). We used voltage ramps from Ϫ100 to ϩ30 mV to examine the current/voltage (I/V) relationship of the AA-dependent current seen in the nominal absence of divalent cations. As shown in Fig. 1B, the I/V relationship showed a reversal potential of ϳ0 mV with significant inward and outward currents (mean current at ϩ30 mV equals 22.6 Ϯ 3.4 pA/pF, n ϭ 5). The I/V curve showed a characteristic tilde (ϳ) shape, with marked inward rectification at negative voltages and outward rectification at positive voltages. The obvious nonlinear nature of the I/V curve strongly suggested that the observed macroscopic current was not a simple nonspecific "leak" current caused by removal of extracellular divalent ions. This was further confirmed by the demonstration that substitution of extracellular Na ϩ with NMDG ϩ completely inhibited the inward AA-dependent current, but did not affect the outward current, whereas substitution of Cs ϩ in the pipette solution with NMDG ϩ resulted in the complete absence of any outward current, without affecting the magnitude of the observed inward current (see Fig. 4B). Moreover, AA-dependent inward currents were completely inhibited (mean inhibition of 95.6 Ϯ 3.5%, n ϭ 5) by extracellular La 3ϩ (50 M) and partially inhibited (50.8 Ϯ 2.2%, n ϭ 4) by extracellular Cd 2ϩ (50 M). Both of these inhibitions were reversible. This further supports the contention that the AA-dependent current observed was not a nonspecific leak and indicates that the large current seen upon nominal removal of extracellular divalent cations represents the activity of an AA-dependent monovalent cation-permeable conductance. Comparison of the magnitudes of the inward current (at Ϫ80 mV) and the corresponding outward current (at ϩ30 mV) showed a direct linear relationship (r 2 ϭ 0.89) (Fig. 1C), consistent with both currents being carried through the same channels. As both inward and outward currents were observed under these divalent ion-free conditions, it is clear that the channels responsible are appreciably permeable to both Na ϩ and Cs ϩ ions (see below); however, they show negligible conductance to NMDG ϩ .
As already noted, both voltage-gated Ca 2ϩ channels and the CRAC channel of non-excitable cells have been shown to dis-play a large permeability to monovalent ions upon removal of extracellular divalent cations (7,15,16). The features of the AA-dependent monovalent current described above are consistent with the existence of a similar phenomenon in the ARC channels we have previously identified. To examine this, we compared the magnitude of the observed AA-dependent monovalent current measured at Ϫ80 mV with that of the corresponding AA-dependent Ca 2ϩ -selective current measured in the same cell immediately prior to removal of extracellular divalent cations. The data obtained show a direct linear relationship between the magnitudes of the two AA-dependent currents (r 2 ϭ 0.88) (Fig. 1D), supporting the contention that they reflect the behavior of a common set of channels (i.e. ARC channels).
Endogenous Store-operated Channels of HEK293 Cells (HEK-SOC Channels) Also Show a Permeability to Monovalent Cations upon Removal of Extracellular Divalent Cations-As noted, the aim of this study was to further define the features of the non-capacitative arachidonate-regulated conductance that distinguish it from store-operated or capacitative conductances. The features of the macroscopic monovalent conductance that develops upon reduction of external divalent cations in the store-operated conductance (I CRAC ) of Jurkat lymphocytes have been extensively characterized (15,16). However, it appears likely that store-operated channels represent a family of conductances of which I CRAC is only one member. Therefore, it was important to be able to compare the AA-activated conductance with the endogenous SOC conductance in the same cell type. We therefore examined the effect of removal of divalent cations from the extracellular medium in cells in which the internal Ca 2ϩ stores had been passively depleted by dialyzing with a highly buffered Ca 2ϩ -free pipette solution. In the normal bath solution containing millimolar Ca 2ϩ and Mg 2ϩ , a small inward current measured at Ϫ80 mV slowly developed upon achieving whole-cell conditions. As previously reported, this current displayed all the features consistent with its characterization as a Ca 2ϩ -selective store-operated current (I SOC ), including marked inward rectification, very positive reversal potential, fast inactivation, and inhibition by La 3ϩ (50 M) (12). Essentially identical currents were observed in cells in which intracellular Ca 2ϩ stores were maximally depleted by preincubation in thapsigargin (1 M for 15 min) (12) or by inclusion of the non-metabolizable inositol 1,4,5-trisphosphate analog adenophostin A (2 M) in the pipette solution (data not shown). Subsequent exposure to nominally divalent ion-free external solution evoked an increase in the inward current measured at Ϫ80 mV to reach a maximum amplitude of 11.7 Ϯ 1.6 pA/pF (n ϭ 8) or some 20 times the current observed in normal divalent ion-containing external medium ( Fig. 2A). This storeoperated current subsequently declined fairly rapidly in the continued absence of extracellular divalent ions, showing a 50% reduction in magnitude in a mean time of 17.6 Ϯ 4.2 s (n ϭ 5).
I/V curves for the store-operated current in the nominal absence of divalent cations were obtained by application of voltage ramps from Ϫ100 to ϩ30 mV (Fig. 2B). Under standard conditions (Na ϩ in the bath solution and Cs ϩ in the pipette solution), these showed a current that reversed at ϳ0 mV with significant inward and outward currents (mean current at ϩ30 mV of 10.8 Ϯ 1.7 pA/pF, n ϭ 6). Inward rectification at negative voltages and outward rectification at positive voltages were also apparent, giving the I/V curve a tilde shape. The magnitudes of the inward (at Ϫ80 mV) and outward (at ϩ30 mV) currents in individual cells were linearly related (r 2 ϭ 0.72) (Fig. 2C). Moreover, the inward store-operated currents were reversibly inhibited by extracellular La 3ϩ (50 M), and both inward and outward currents were rapidly inhibited upon reexposure to normal divalent ion-containing medium in the bath (data not shown). Substitution of extracellular Na ϩ with NMDG ϩ completely inhibited the inward currents, but did not affect the outward currents. Similarly, substitution of cesium in the pipette solution with NMDG ϩ resulted in the complete absence of any outward current, without affecting the magnitude of the observed inward current (see Fig. 4D). Based on these data, we conclude that the observed macroscopic currents are not the result of a simple nonspecific leak and that the currents seen upon nominal removal of extracellular divalent cations after depletion of the intracellular Ca 2ϩ stores repre-FIG. 1. Development of arachidonic acid-dependent monovalent currents in the nominal absence of extracellular divalent cations. A, representative trace showing the current magnitude measured during 250-ms voltage steps to Ϫ80 mV from a holding potential of 0 mV applied every 2 s under standard conditions (Na ϩ as the principal cation in the bath solution and Cs ϩ as the principal cation in the pipette solution). In the presence of exogenous AA (8 M), large inward currents developed promptly upon nominal removal of extracellular divalent cations (divalent-free) and then subsequently slowly declined. B, current-voltage relationship of the AA-dependent current in the nominal absence of extracellular divalent cations. Currents were recorded during 150-ms voltage ramps from Ϫ100 mV to ϩ30 mV at the peak of the AA-dependent current. C, relationship between the magnitude of the inward current measured at Ϫ80 mV and the simultaneous outward current measured at ϩ30 mV. Each point represents the value from a different cell. D, relationship between the magnitude of the inward Ca 2ϩ -selective current measured at Ϫ80 mV immediately prior to removal of extracellular divalent ions and the corresponding peak inward monovalent current (also measured at Ϫ80 mV) in the same cell. sent the activity of the endogenous store-operated conductance (I SOC ). This was further confirmed by comparison of the magnitude of the macroscopic SOC current in its Ca 2ϩ -selective mode seen in the presence of divalent ions in the bath solution with that of the corresponding current seen in the same cell upon removal of external divalent ions. This showed that the magnitudes of the two currents were linearly related to each other (r 2 ϭ 0.64) (Fig. 2D).
From the above results, it is clear that both the endogenous store-operated or capacitative conductance (I SOC ) and the noncapacitative AA-regulated Ca 2ϩ -selective conductance (I ARC ) display a marked permeability to monovalent cations upon nominal removal of extracellular divalent ions, a property shared with voltage-gated Ca 2ϩ channels and the classic storeoperated Ca 2ϩ conductance (I CRAC ). The most immediately obvious difference between the macroscopic currents displayed by ARC and SOC channels is the much larger monovalent conductance shown by I ARC relative to I SOC (Fig. 3A). Thus, although the whole-cell currents observed for the two conductances when in their Ca 2ϩ -selective mode are similar in magnitude (ϳ0.5-0.6 pA/pF at Ϫ80 mV), the monovalent ARC currents seen upon removal of extracellular divalent ions are ϳ2.5 times the corresponding monovalent SOC currents (28.3 versus 11.7 pA/pF at Ϫ80 mV). Such a difference may reflect a different sensitivity of ARC and SOC channels to the very low (submicromolar) Ca 2ϩ concentrations in our nominally divalent ion-free external medium. However, such sensitivity would have to be much higher than that previously reported for CRAC channels (IC 50 for Ca 2ϩ block of Na ϩ currents ϭ 4 M) (13). Moreover, as noted above, the monovalent SOC currents displayed a fairly rapid spontaneous decline, whereas the decline in the monovalent ARC current was much slower (Fig. 3, B and C). As discussed further below, the decline in the monovalent SOC current appears to reflect some kind of "inactivation" process of unknown origin. Given this, it is possible that the inactivation process shown by the SOC channels is occurring simultaneously with the activation of the monovalent permeability upon removal of extracellular divalent ions and could therefore contribute to the relatively smaller monovalent current seen with these channels. Despite these reservations, it is clear that, like the store-operated CRAC channels (7,15,16), both the endogenous SOC channels and the ARC channels of HEK293 cells are capable of conducting monovalent ions much more readily than Ca 2ϩ ions, although there may be some relative difference in this ability between the ARC and SOC channels. Monovalent Selectivity-As noted above, both I SOC and I ARC in the nominal absence of extracellular divalent ions displayed characteristic tilde-shaped I/V curves with significant inward and outward monovalent currents under the standard conditions (Na ϩ in the bath solution and Cs ϩ in the pipette solution). Moreover, when the Cs ϩ in the pipette solution was replaced with NMDG ϩ , neither I SOC nor I ARC displayed any significant outward currents. These data indicate that both SOC and ARC channels have a significant permeability to both Na ϩ and Cs ϩ ions under these conditions. To examine the monovalent cation selectivity of the ARC and SOC channels in more detail, experiments were performed involving various monovalent cation substitutions in both the bath and pipette solutions. The data obtained are illustrated (A-D) and summarized (E and F) in Fig. 4. Estimates of the respective sequences of monovalent conductance for the two channels were then determined from the resulting macroscopic current magnitudes.
Substitution of Na ϩ in the bath solution with Cs ϩ had a statistically insignificant (p ϭ 0.07) effect on the inward current density of the ARC channels measured at Ϫ80 mV (Fig. 4, A and E). When external Na ϩ ions were replaced by Li ϩ , the inward current density at Ϫ80 mV was markedly reduced (from 28.3 Ϯ 2.7 pA/pF (n ϭ 5) to 5.0 Ϯ 2.1 pA/pF (n ϭ 6)) (Fig. 4, A and E). Outward (Cs ϩ ) currents were also affected by substitutions of the external cations (Fig. 4A). Substitution of external Na ϩ with Cs ϩ reduced the outward (Cs ϩ ) current measured at ϩ30 mV by ϳ50% to 11.9 Ϯ 3.0 pA/pF (n ϭ 6; p ϭ 0.02), whereas substitution with Li ϩ further reduced the outward current to 8.0 Ϯ 3.4 pA/pF (n ϭ 6). Clearly then, the identity of Internal Ca 2ϩ stores were depleted passively using a Ca 2ϩfree pipette solution. A, representative trace showing the inward current magnitude measured during 250-ms voltage steps to Ϫ80 mV applied every 2 s from a holding potential of 0 mV under standard conditions (Na ϩ and Cs ϩ as the principal cations in the bath and pipette solutions, respectively). Large inward currents developed upon nominal removal of external divalent cations and then spontaneously declined back to resting levels. B, current-voltage relationship of the storeoperated current in the nominal absence of extracellular divalent cations. Currents were recorded during 150-ms voltage ramps from Ϫ100 to ϩ30 mV at the peak of the monovalent store-operated current. C, relationship between the magnitude of the inward and the simultaneous outward store-operated currents (measured at Ϫ80 and ϩ30 mV, respectively). Each point represents the value from a different cell. D, relationship between the magnitude of the inward Ca 2ϩ -selective storeoperated current measured at Ϫ80 mV immediately prior to removal of extracellular divalent ions and the corresponding peak inward monovalent current (also measured at Ϫ80 mV) in the same cell. the external ion affects the ability of the internal ion to carry outward current through the ARC channels. We have not analyzed the basis for these effects in detail, but they are indicative of an interaction of the different ions within the channel, consistent with the channel possessing a multi-ion conductance pathway (19). Based on these values of relative inward current density, a selectivity sequence for the monovalent conductance of the ARC channels of Cs ϩ Ϸ Na ϩ Ͼ Ͼ Li ϩ Ͼ Ͼ NMDG ϩ is indicated. True relative permeabilities are normally analyzed from changes in reversal potentials. However, in the experiments reported here, the small magnitudes of the overall currents and the fact that the I/V curves obtained were generally rather flattened around the reversal potential made such analysis quantitatively difficult. Nevertheless, examination of the changes in reversal potentials observed indicted a similar relative permeability sequence of Cs ϩ -Na ϩ -Li ϩ -NMDG ϩ .
A small but statistically significant (p ϭ 0.05) increase in the inward current density at Ϫ80 mV upon substitution of external Na ϩ with Cs ϩ was seen with the SOC current (Fig. 4, C and  E). However, as noted above, the magnitude of the observed currents may be significantly influenced by the simultaneous rate of the inactivation process. In this case, the rate of the spontaneous inactivation of the SOC current was noticeably reduced in the presence of external Cs ϩ (current magnitude reduced to 50% in a mean time of 49.3 Ϯ 2.4 s (n ϭ 3) versus 17.6 Ϯ 4.2 s (n ϭ 5)). Thus, the apparent increased inward SOC current density in the presence of external Cs ϩ may result, at least in part, from the slower rate of inactivation seen under these conditions. Replacement of external Na ϩ with Li ϩ resulted in an approximate 50% reduction of the inward current density to 6.3 Ϯ 3.0 pA/pF (n ϭ 4) (Fig. 4C), a value very similar to that seen with ARC currents (Fig. 4E). In marked contrast to the response of ARC currents, outward (Cs ϩ ) currents through SOC channels were not significantly affected by replacement of external Na ϩ with either Cs ϩ or Li ϩ . The data obtained indicate a monovalent selectivity sequence for inward currents in the HEK-SOC channels of Cs ϩ Ϸ Na ϩ Ͼ Li ϩ Ͼ Ͼ NMDG ϩ . Once again, despite the difficulties in obtaining any precise quantitative assessment, examination of the changes in reversal potential observed indicated an identical relative permeability sequence of Cs ϩ -Na ϩ -Li ϩ -NMDG ϩ .
Substitutions were also made to the monovalent cations in the internal (pipette) solutions. Substitution of internal Cs ϩ with Na ϩ markedly reduced the outward current density measured at ϩ30 mV through the ARC channels (from 22.6 Ϯ 3.4 to 8.1 Ϯ 1.0 pA/pF, n ϭ 5) (Fig. 4, B and F). A similar, although much less dramatic decrease in outward current density at ϩ30 mV was seen with SOC channels when Na ϩ replaced internal Cs ϩ (from 10.8 Ϯ 1.7 pA/pF (n ϭ 8) to 5.2 Ϯ 0.8 pA/pF (n ϭ 6)) ( Fig. 4, D and F). Importantly, this latter response could not be caused by an effect on the rate of spontaneous inactivation, as this was much slower when Na ϩ replaced Cs ϩ in the pipette solution (time to attain 50% inactivation of 54.0 Ϯ 11.4 versus 17.6 Ϯ 4.2 s). As has already been noted, replacing internal Cs ϩ with NMDG ϩ resulted in the complete absence of any significant outward currents through either ARC or SOC channels (Fig. 4, B and D). Based on these data for outward current densities, a monovalent cation selectivity for both ARC and SOC channels of Cs ϩ Ͼ Na ϩ Ͼ Ͼ NMDG ϩ is indicated. Again, despite the caveats already mentioned, examination of the conductances. Traces represent responses from typical cells in which the decay in macroscopic current is displayed by first normalizing to their corresponding peak values measured at Ϫ80 mV. C, the times required for a 50% decline in current magnitude (mean Ϯ S.E., n ϭ 5) for SOC (white bar) and ARC (black bar) channels. changes in reversal potential observed indicated an identical relative permeability sequence of Cs ϩ -Na ϩ -NMDG ϩ .
Effects of Internal Mg 2ϩ -The similarity of the inward currents carried by Na ϩ and Cs ϩ reported here for both ARC and HEK-SOC channels (Fig. 4E) is not seen in the CRAC channels of Jurkat cells measured under the same conditions (15,16). However, it was reported that magnesium-free internal solutions increase the magnitude of the inward Na ϩ currents and, more particularly, dramatically increase the Cs ϩ permeability of the Jurkat CRAC channels such that the Na ϩ and Cs ϩ currents are essentially identical (16). It was also noted that large outward currents develop in the absence of internal Mg 2ϩ that are not observed in the presence of internal Mg 2ϩ (16). We therefore examined the effects of removal of internal Mg 2ϩ on the monovalent currents recorded for ARC and HEK-SOC channels.
Under the standard conditions of Na ϩ in the bath solution and Cs ϩ in the pipette solution, removal of internal (pipette) Mg 2ϩ had no significant effect on the magnitude of the inward (Na ϩ ) currents through ARC channels (Fig. 5A). Similarly, no significant change in the magnitude of the inward current was seen when the external cation was Cs ϩ (data not shown). How-ever, examination of the I/V relationship clearly indicated that removal of internal Mg 2ϩ reduced the degree of rectification in the inward (Na ϩ ) current (Fig. 5B). Mg 2ϩ -free internal solutions also markedly reduced the outward current (measured at ϩ30 mV) carried by Cs ϩ from 22.6 Ϯ 3.4 to 13.4 Ϯ 2.2 pA/pF (n ϭ 5 and 4, respectively; p ϭ 0.04) (Fig. 5, A and B).
When monovalent SOC currents were examined, Mg 2ϩ -free internal solutions again had no significant effect on inward currents, whether these were carried by Na ϩ (Fig. 5C) or Cs ϩ (data not shown). Unlike the ARC currents, however, removal of internal Mg 2ϩ did not significantly influence the degree of rectification of the inward current (Fig. 5D). Moreover, in marked contrast to the response seen in ARC currents, removal of internal Mg 2ϩ significantly increased the outward (Cs ϩ ) current measured at ϩ30 mV from 10.8 Ϯ 1.7 to 16.6 Ϯ 2.1 pA/pF (n ϭ 6 and 8, respectively; p ϭ 0.03) (Fig. 5, C and D). Finally, Kerschbaum and Cahalan (16) reported that removal of internal Mg 2ϩ eliminates the spontaneous inactivation of the monovalent current through the CRAC channels in Jurkat cells. However, no such effect was observed in the endogenous SOC channels of HEK293 cells measured under the standard conditions (Na ϩ in the bath solution and Cs ϩ in the pipette FIG. 4. Selectivity of ARC and SOC conductances for monovalent cations. ARC currents were activated by addition of exogenous arachidonic acid (8 M), and SOC currents were induced by passive depletion using a Ca 2ϩ -free pipette solution. A-D, shown are mean current-voltage curves illustrating the effects of substitution with different cations compared with the currents seen under standard conditions with Na ϩ as the principal cation in the bath solution and Cs ϩ as the principal cation in the pipette solution (‚). The effects of substitution of Na ϩ in the external (bath) solution with either Cs ϩ (q) or Li ϩ (E) are illustrated for the monovalent currents through ARC (A) and SOC (C) channels. Similarly, the effects of substitution of Cs ϩ in the internal (pipette) solution with either Na ϩ (q) or NMDG ϩ (E) are illustrated for monovalent currents through ARC (B) and SOC (D) channels. E, summary data illustrating the effects of different external cations on inward ARC and SOC currents measured at Ϫ80 mV (mean Ϯ S.E., n ϭ 4 -11). The principal external cation was Na ϩ (white bars), Cs ϩ (black bars), or Li ϩ (shaded bars). F, summary data illustrating the effects of different internal cations on outward ARC and SOC currents measured at ϩ30 mV (mean Ϯ S.E., n ϭ 4 -8). The principal internal cation was Cs ϩ (black bars) or Na ϩ (white bars). As can be seen in the corresponding I/V traces (B and D), there was no detectable outward current when the principal internal cation was NMDG ϩ . solution). In these cells, the times required to achieve a 50% decline in the measured current at Ϫ80 mV were 17.6 Ϯ 4.2 s in the presence of internal Mg 2ϩ and 23.6 Ϯ 4.9 s in its absence (n ϭ 5).
Activation of ARC Currents after SOC Current Inactivation-As already noted, the monovalent HEK-SOC current showed a rather rapid spontaneous decline in magnitude after activation upon removal of extracellular divalent cations (see Fig. 3B). Following such inactivation, re-exposure of the cells to normal divalent ion-containing medium (i.e. with Ca 2ϩ ) demonstrated that the Ca 2ϩ -selective current was also now absent (Fig. 6A). However, continued exposure of the cells to external divalent ions induced a gradual reactivation of the normal Ca 2ϩ -selective current over a 1-2-min period (Fig. 6A, inset). Subsequent exposure to divalent ion-free medium showed that this reappearance of the Ca 2ϩ -selective current was also associated with a reactivation of the monovalent current. The data show that the sizes of these two currents (monovalent and Ca 2ϩ -selective) were directly related to each other during this reactivation process (r 2 ϭ 0.94) (Fig. 6B), confirming that the two currents reflect the activity of the same family of channels. An essentially identical phenomenon was observed in the CRAC channels of Jurkat cells (15). Without implying any specific molecular mechanism for this spontaneous decline and in accordance with the similar decline reported for monovalent CRAC currents (15,16), we have described this phenomenon as a process of inactivation. Use of such a term is consistent with the demonstrated ability to reactivate the current by exposure to extracellular divalent ions.
The rate of decline in the corresponding monovalent ARC current after initial activation was always very much slower (typically 10 -15 times) than those seen with the HEK-SOC currents under identical conditions (see Fig. 3B). In addition, after the decline in the monovalent ARC current was complete, re-exposing the cells to normal external medium containing divalent cations for up to 5 min produced only a minimal subsequent reactivation of the monovalent ARC current (equal to 5.6 Ϯ 0.9% of the original monovalent current, n ϭ 5). These data suggest that the decline in the monovalent ARC current likely involves a rather different process from that seen with HEK-SOC currents. To distinguish between these two apparently rather different phenomena and consistent with the failure to reactivate ARC currents upon re-exposure to extracellular divalent ions, we will refer to this decline in the monovalent ARC current as a "run-down" rather than an inactivation. Again, the use of this term is not meant to signify the involvement of any specific molecular mechanism.
As noted above, the inactivated HEK-SOC current was absolutely dependent on exposure to normal divalent ion-containing solutions in the bath for its re-activation, and no "spontaneous" reactivation of the monovalent HEK-SOC current was ever seen without exposure of the cell to divalent ion-containing external medium. However, addition of exogenous AA (8 M) to a cell in which the monovalent SOC current had been completely inactivated resulted in the development of a large inward monovalent current (Fig. 6C). The appearance of this AA-dependent current was not due to the activation of any residual SOC current resulting from any failure to completely deplete the internal Ca 2ϩ stores, as an identical AA-dependent monovalent current could be activated after spontaneous inactivation of SOC currents that had been stimulated by incubation in thapsigargin or by inclusion of adenophostin A (2 M) in the pipette solution (data not shown). The magnitude of the AA-activated inward current in cells whose stores had been completely depleted by either thapsigargin or adenophostin A was not significantly different from that of the normal monovalent ARC current (26.2 Ϯ 1.1 pA/pF (n ϭ 6) versus 28.3 Ϯ 2.7 pA/pF (n ϭ 5)) and was consistently larger than the magnitude of the previously recorded monovalent SOC current in the same cell (7.5 Ϯ 0.7 pA/pF, n ϭ 6). Moreover, it displayed the characteristically slow spontaneous decline previously described for the monovalent ARC current. These data demonstrate that the ARC currents can be activated in a cell whose SOC currents have been completely inactivated and that the two currents therefore represent the activities of distinct conductances. DISCUSSION The data reported here show that ARC channels in HEK293 cells along with the endogenous SOC channels share, with CRAC channels and voltage-gated Ca 2ϩ channels, the general property of monovalent cation permeation in the presence of low external divalent ion concentrations. This transition from high Ca 2ϩ selectivity to monovalent cation permeation has been most extensively analyzed in voltage-gated Ca 2ϩ channels (20 -23). In these channels, the normally high selectivity for Ca 2ϩ is thought to involve the binding of Ca 2ϩ with high affinity to a site (or sites) in the pore of the channel. This binding enables Ca 2ϩ to act as a blocking ion to reduce nonselective monovalent currents. A similar analysis has more re-cently been applied to the store-operated CRAC channels of Jurkat lymphocytes (15,16). The proposed binding of Ca 2ϩ also plays a role in Ca 2ϩ permeation through these channels, as it explains the saturation of the Ca 2ϩ current with increasing external Ca 2ϩ concentrations. Specific models based on these principles have been developed that can simulate the selectivity, permeation, and block properties of voltage-gated Ca 2ϩ channels (20,21). However, it has recently been argued that such binding models can take several different possible forms, so the precise details are currently far from fully understood (24). Nevertheless, the general principles are reasonably clear. Selectivity in these channels is not a result of any molecular sieving process, but instead relies on the binding of ions to sites within the channel. These sites show a high affinity for Ca 2ϩ compared with Na ϩ . In the presence of external divalent cations, occupation of the sites by Ca 2ϩ precludes the permeation of Na ϩ and other monovalent cations. At the same time, Ca 2ϩ permeation is favored, despite the intrinsic high affinity of the sites, possibly by repulsive effects (strong ion-ion interactions) of ions in two closely positioned binding sites of similar high affinity (20,22) or by a "stepwise" series of sites with graded binding affinities (24). At low concentrations of external divalent cations, Ca 2ϩ is lost from the binding sites and thus permits the permeation of Na ϩ (and other monovalent cations), whose high permeation reflects their low affinity for the channel binding sites.
As such, it would seem that the overall behavior of both the ARC channels and the endogenous SOC channels of HEK293 cells is consistent with these models. Both ARC and HEK-SOC channels are highly Ca 2ϩ -selective in normal divalent cation- FIG. 6. A, representative trace showing that, after spontaneous inactivation of the monovalent SOC current, reactivation is possible, but only after exposure to external Ca 2ϩ . The inset is an expanded portion of the main graph for the period at 450 -650 s showing the reactivation of the Ca 2ϩ -selective SOC current upon re-exposure to normal divalent ion-containing medium. B, the relationship between the magnitude of the Ca 2ϩ -selective current observed immediately prior to return to a nominally divalent ion-free external medium (first arrow in A) and the subsequent reactivated monovalent current (second arrow) in individual cells. C, representative trace illustrating the sequential activation of monovalent SOC and ARC currents in the same cell. After spontaneous inactivation of the monovalent SOC current, addition of exogenous AA (8 M) induced the activation of a large monovalent ARC current.
containing external medium and display a saturating current magnitude with increasing external Ca 2ϩ concentrations. Despite this high selectivity for Ca 2ϩ , these channels reveal relatively large monovalent currents in the nominal absence of extracellular divalent cations. These monovalent currents are rapidly and markedly inhibited by extracellular Ca 2ϩ and are blocked by the same ions that block the Ca 2ϩ conductance. Moreover, in both cases, the magnitude of the measured monovalent current is directly related to the magnitude of the corresponding Ca 2ϩ current. All these features argue that the Ca 2ϩ currents measured in the presence of extracellular divalent ions and the much larger monovalent currents measured in the nominal absence of divalent ions reflect the behavior of the same channels.
The characteristic small Ca 2ϩ current and yet relatively large monovalent current of CRAC channels is therefore a feature shared by both HEK-SOC and ARC channels. Nevertheless, certain striking differences are apparent in the details of this phenomenon between these different channels. We have summarized the key features of the three channels (CRAC, HEK-SOC, and ARC) in Table I. First, although both ARC and HEK-SOC channels have broadly similar Ca 2ϩ -selective current densities compared with the CRAC channels of Jurkat cells (ϳ0.5-0.6 pA/pF for ARC and HEK-SOC channels and 0.8 pA/pF for CRAC channels) (8,12), HEK-SOC and, more particularly, ARC channels have much higher monovalent current densities than Jurkat CRAC channels. Monovalent (Na ϩ )/Ca 2ϩ current density ratios for CRAC channels range from 5 to 10 (mean ϭ 7.5 Ϯ 2.7) (14). The corresponding value for HEK-SOC channels was ϳ20 and was even larger for ARC channels (ϳ50). In CRAC channels, this ratio was profoundly influenced by removal of internal Mg 2ϩ , which markedly increased inward Na ϩ currents. Under these conditions, the Na ϩ /Ca 2ϩ current ratio for the CRAC channels becomes equal to 24.6 Ϯ 4.9 (16), a value similar to that observed for HEK-SOC channels in the presence of internal Mg 2ϩ , but still considerably less than that observed for ARC channels. Moreover, unlike Jurkat CRAC currents, removal of internal Mg 2ϩ had no significant effect on the magnitude of the inward Na ϩ currents for either HEK-SOC or ARC channels. The underlying basis for the observed differences in the relative magnitudes of the Ca 2ϩ and Na ϩ currents between the different channels remains unclear. As explained, estimates of the monovalent (Na ϩ ) currents for both HEK-SOC and CRAC channels may be underestimated due to the spontaneous inactivation phenomenon seen in both these conductances. However, it seems unlikely that this can account for all the differences seen, especially those between HEK-SOC and CRAC channels, both of which show broadly similar inactivation kinetics. Until these properties have been examined in detail at the single channel level, we can only speculate on the basis for the observed differences. They may reflect, for exam-ple, distinct permeation properties of the individual channels or differences in the respective channel kinetics (e.g. open probability) induced by divalent ion-free external medium.
With regard to the monovalent selectivity, the data on inward current densities indicate a broadly similar conductance sequence for both ARC and HEK-SOC channels of Cs ϩ Ϸ Na ϩ Ͼ Li ϩ Ͼ Ͼ NMDG ϩ , although, compared with HEK-SOC channels, ARC channels did show a much lower conductance for Li ϩ relative to that for Na ϩ or Cs ϩ . The corresponding sequence for the outward current densities for both channels is Cs ϩ Ͼ Na ϩ Ͼ Ͼ NMDG ϩ . It should be noted that these apparent conductance sequences are based on the measured macroscopic current densities and therefore do not allow for any possible effects of ion substitution on the open probability of the channels. Such analysis would require information on the behavior of single channels, which is currently unavailable. The observed monovalent selectivity sequence for both ARC and HEK-SOC channels stands in marked contrast to that reported for the CRAC channels of Jurkat cells under the same conditions (15,16). In these cells, the measured Cs ϩ conductance through the CRAC channels is much smaller than the corresponding Na ϩ conductance, with a reported ratio of inward Na ϩ to Cs ϩ currents (measured at Ϫ80 mV) of 26 Ϯ 3. The corresponding ratios for ARC and HEK-SOC channels are 0.80 and 0.74, respectively (Fig. 4). Lepple-Wienhues and Cahalan (15) noted that they sometimes observed a nonspecific leak conductance in Jurkat cells that showed a Cs ϩ permeability similar to that of Na ϩ . However, this leak showed a linear I/V relationship and lacked the characteristic inactivation seen with either CRAC or HEK-SOC currents. Given these features, it seems unlikely that the high apparent Cs ϩ permeability we have observed in the HEK293 cells results from such a similar leak current. Clearly, CRAC channels are highly selective for Na ϩ over Cs ϩ in the nominal absence of extracellular divalent cations, a feature that is not shared by either the ARC channels or the endogenous HEK-SOC channels. Importantly, Kerschbaum and Cahalan (16) reported that the very low conductance to Cs ϩ shown by CRAC channels (relative to Na ϩ ) is another feature that is dramatically affected by internal Mg 2ϩ . Although removal of internal Mg 2ϩ did significantly increase outward Cs ϩ currents through HEK-SOC channels (Fig. 5C), the effect was quantitatively much smaller than that reported for CRAC channels (16). Moreover, in marked contrast to the data from both CRAC and HEK-SOC channels, removal of internal Mg 2ϩ significantly decreased outward Cs ϩ currents through the ARC channels (Fig. 5A). Obviously, the conclusion reached by Kerschbaum and Cahalan (16) that internal Mg 2ϩ normally blocks outward current through CRAC channels clearly does not apply to ARC channels.
Another reported effect of internal Mg 2ϩ removal on CRAC channels is the loss of the spontaneous inactivation of the  (16). Although this was not seen in the HEK-SOC currents, many of the other features of the inactivation of monovalent CRAC currents were seen in the HEK-SOC currents, including the ability to reactivate the monovalent currents only after a brief period of re-exposure to external Ca 2ϩ . This reactivation process involved both Ca 2ϩ currents (measured in normal divalent ion-containing medium) and the monovalent currents (measured in divalent ion-free medium), which reactivated in synchrony, as was previously reported for CRAC channels (15). A rather different kind of decay was observed in the monovalent currents through the ARC channels. This decay was much slower than the inactivation seen in the HEK-SOC currents, and after decaying to resting levels, re-exposure to external Ca 2ϩ was able to induce only a minimal reactivation of the monovalent ARC currents. Although we have no direct information on the basis for this phenomenon, such behavior is suggestive of a run-down of the currents rather than an inactivation. Precise characterization of these two processes and the mechanisms involved must await further studies. Nevertheless, we were able to utilize the absolute requirement of exposure to extracellular Ca 2ϩ for the reactivation of the HEK-SOC currents after their spontaneous inactivation to demonstrate that ARC currents could be activated (by addition of extracellular AA) in cells in which the monovalent SOC currents had been previously fully inactivated. The subsequently activated AA-dependent monovalent currents displayed all the properties characteristic of ARC currents. Clearly then, ARC channels can be activated in a cell whose SOC channels have been completely inactivated, consistent with our earlier findings that the two conductances are entirely separate and distinct entities (12,25). Together, the data presented indicate that there are both similarities and marked differences in the behavior of the CRAC channels of Jurkat cells, the endogenous SOC channels of HEK293 cells, and non-capacitative ARC channels. We have already described the similarities and differences between the HEK-SOC and ARC channels in their normal Ca 2ϩ -conducting state (12). In this study, we have shown that, as highly Ca 2ϩ -selective conductances, these channels share the property of conversion to a monovalent cationpermeable mode upon removal of extracellular divalent cations. Comparing first the two store-operated conductances CRAC and HEK-SOC, it is clear that, although they are very similar in several respects, they also appear to display some marked differences. Examination of these differences reveals that they particularly revolve around the effects of internal Mg 2ϩ ions on the channel properties. For example, two of the most obvious differences between the CRAC and HEK-SOC conductances are their respective Na ϩ /Ca 2ϩ current density ratios and their monovalent selectivity sequence. However, these differences largely disappear when CRAC currents are measured in the absence of internal Mg 2ϩ (16). In other words, HEK-SOC channels under normal conditions behave much like CRAC channels in the absence of internal Mg 2ϩ , suggesting that the principal difference between these two store-operated conductances lies simply in the presence or absence of some Mg 2ϩ -binding site(s) or in its affinity for Mg 2ϩ . Whether this reflects the influence of differences in the cellular environment (membrane composition, presence of modulatory factors, etc.) or actual differences in the molecular identity of the channel proteins involved must await their characterization at the molecular level. However, it seems that these two store-operated conductances are probably closely related members of what is likely to be a family of similar channels.
More importantly, our ability to measure the endogenous SOC channels and ARC channels in the same cells allows a much more direct comparison between these two conductances. Here, several marked differences are apparent. These include differences in the magnitude of the ratio of monovalent to Ca 2ϩ currents, in the rate and nature of the spontaneous decline in monovalent current, and in the effects of external monovalent cation substitutions and of internal Mg 2ϩ removal on outward currents. All these features indicate that ARC and SOC currents reflect the activity of distinct conductances. Moreover, we have demonstrated that monovalent ARC currents can be activated after the complete spontaneous inactivation of SOC currents in the same cell. As such, this confirms our previous findings based on the characteristics and additive nature of the Ca 2ϩ -selective currents for these two conductances (12) and on their distinct relative abilities to activate Ca 2ϩ -sensitive adenylyl cyclases (25). Of course, the most important and fundamental distinction between these two conductances is that, unlike SOC channels, ARC channels are not activated by store depletion (12). Importantly, the individually unique characteristics displayed by these two conductances in their monovalent cation-permeable modes described here, together with the much larger macroscopic currents observed, are features that will undoubtedly prove useful in the identification of the specific activities of the respective channels under different experimental conditions and, in addition, raise the possibility of further more detailed analysis at the single channel level, as has already been demonstrated for CRAC channels (26,27).