Voltage-dependent conductance changes in the store-operated Ca²⁺ current I_CRAC in rat basophilic leukaemia cells

Abstract

1. Tight-seal whole-cell patch-clamp experiments were carried out in order to investigate the effects of different holding potentials on the rate of development and amplitude of the Ca²⁺ release-activated Ca²⁺ current I_CRAC in rat basophilic leukaemia (RBL-1) cells. I_CRAC was monitored at −80 mV from fast voltage ramps, spanning 200 mV in 50 ms.

2. At hyperpolarised potentials, the macroscopic CRAC conductance was lower than that seen at depolarised potentials. The conductance increased almost 5-fold over the voltage range −60 to +40 mV and was seen when the stores were depleted either by the combination of IP₃ and thapsigargin in high Ca²⁺ buffer, or passively with 10 mm EGTA or BAPTA.

3. The voltage-dependent conductance of the CRAC channels could not be fully accounted for by Ca²⁺-dependent fast inactivation, nor by other slower inhibitory mechanisms. It also did not seem to involve intracellular Mg²⁺ or the polycations spermine and spermidine.

4. Voltage step relaxation experiments revealed that the voltage-dependent conductance changes developed and reversed slowly, with a time constant of several seconds at −60 mV.

5. In the presence of physiological levels of intracellular Ca²⁺ buffers, I_CRAC was barely detectable when cells were clamped at −60 mV and dialysed with IP₃ and thapsigargin, but at 0 mV the current in low Ca²⁺ buffer was as large as that seen in high Ca²⁺ buffer.

6. Our results suggest that CRAC channels exhibit slow voltage-dependent conductance changes which can triple the current amplitude over the physiological range of voltages normally encountered by these cells. The role of this conductance change and possible underlying mechanisms are discussed.

In many cell types, the process of emptying the inositol 1,4,5-trisphosphate (IP₃)-sensitive intracellular Ca²⁺ stores activates Ca²⁺ influx through the store-operated pathway (capacitative Ca²⁺ entry; Putney, 1986; Petersen et al. 1999). Whole-cell patch-clamp recordings have revealed that store-operated Ca²⁺ entry is electrogenic, being manifested as a Ca²⁺ current. The most widely distributed and best characterised store-operated Ca²⁺ current is I_CRAC (Hoth & Penner, 1992; Parekh & Penner, 1997). Ca²⁺ entry through CRAC channels is required for refilling the Ca²⁺ stores, triggering exocytosis, activating certain transcription factors and stimulating cell proliferation (Parekh & Penner, 1997). Given the broad spectrum of kinetically distinct processes that are activated by Ca²⁺ entry through CRAC channels, one would anticipate that the activity of these channels should be extensively regulated. Indeed, several independent mechanisms have been found to curtail the activity of CRAC channels in RBL-1 (rat basophilic leukaemia) cells and Jurkat T-lymphocytes, two model systems for investigating I_CRAC. These include Ca²⁺-dependent fast inactivation (operating over tens of milliseconds) arising through a negative feedback mechanism triggered by permeating Ca²⁺ ions (Zweifach & Lewis, 1995a; Fierro & Parekh, 1999a), Ca²⁺-dependent slow (tens of seconds) inactivation independent of store refilling but requiring a global increase in cytosolic Ca²⁺ (Zweifach & Lewis, 1995b; Parekh, 1998), slow inactivation by protein kinase C (Parekh & Penner, 1995), and inhibition by sphingolipids (Mathes et al. 1998).

It has generally been accepted that I_CRAC can only be measured when high concentrations of Ca²⁺ buffers are included in the recording pipette. Recently, we have succeeded in recording robust I_CRAC under conditions of weak intracellular Ca²⁺ buffering in RBL-1 cells (Fierro & Parekh, 2000). Macroscopic I_CRAC develops in low Ca²⁺ buffer provided the sarco-endoplasmic reticulum Ca²⁺-ATPase (SERCA) pumps are blocked, for example with thapsigargin. Intracellular dialysis with a combination of IP₃ and thapsigargin in low Ca²⁺ buffer activated I_CRAC to a similar extent to that seen with high Ca²⁺ buffer. I_CRAC can therefore be activated maximally under physiological conditions of weak Ca²⁺ buffering (Fierro & Parekh, 2000).

In the presence of low intracellular Ca²⁺ buffering, we have previously applied the capacitance technique to directly record exocytosis in RBL-1 cells (Artalejo et al. 1998). We found that secretion was both GTP and intracellular Ca²⁺ dependent. The combination of IP₃ and thapsigargin (together with GTPγS) in low Ca²⁺ buffer was a very effective stimulus for triggering exocytosis in RBL-1 cells. However, we failed to detect any consistent macroscopic current under these conditions. In the light of our recent finding that large I_CRAC can routinely be activated by IP₃ and thapsigargin in low Ca²⁺ buffer (Fierro & Parekh, 2000), our inability to record I_CRAC during the capacitance recordings is puzzling. The key difference between the two reports was that cells were clamped at −60 mV for the capacitance study (Artalejo et al. 1998), rather than the routinely used 0 mV (Fierro & Parekh, 2000). A holding potential of −60 mV was chosen simply to prevent the activation of the very large GTP-gated Na⁺ current, which would have contaminated the capacitance measurements. We therefore hypothesised that the amplitude of I_CRAC is affected by the prevalent membrane potential in that hyperpolarisation somehow induces a voltage-dependent inhibition of channel activity. Clarification of this is important both for understanding the regulation of I_CRAC and for placing its role in a physiological context, since most cell types that express I_CRAC have quite hyperpolarised membrane potentials where the putative block would be most severe. In the present report, we have tested this hypothesis and our results suggest that CRAC channels exhibit voltage-dependent conductance changes.

METHODS

Rat basophilic leukaemia (RBL-1) cells were bought from the American Type Culture Collection. Cell culture was as previously described (Fierro & Parekh, 2000) and will not be elaborated here.

Patch-clamp experiments were conducted in the tight-seal whole-cell configuration at room temperature (20–25°C) as previously described (Hamill et al. 1981; Fierro & Parekh, 2000; Glitsch & Parekh, 2000). Sylgard-coated, fire-polished pipettes had DC resistances of 2.9–4 MΩ when filled with standard internal solution containing (mm): caesium glutamate 145, NaCl 8, MgCl₂ 1 and Hepes 10; pH 7.2 with CsOH. The Ca²⁺ chelators EGTA (Sigma) or BAPTA (tetracaesium salt, Molecular Probes) were added to this solution at the concentrations specified in the text. IP₃ (added to the pipette solution) was obtained from Sigma. Thapsigargin was from Alomone Laboratories. All other chemicals were purchased from Sigma. A correction of +10 mV was applied for the subsequent liquid junction potential that arose from this glutamate-based internal solution. Extracellular solution contained (mm): NaCl 145, KCl 2.8, CaCl₂ 10, MgCl₂ 2, CsCl 10, glucose 10 and Hepes 10, pH 7.4 with NaOH. Voltage ramps (−100 to +100 mV in 50 ms) were applied at 0.5 Hz from different holding potentials (stated in the text) and I_CRAC was measured at −80 mV from the ramps, as previously described (Fierro & Parekh, 2000; Glitsch & Parekh, 2000). Currents were filtered using an 8-pole Bessel filter at 2.9 kHz and digitised at 100 μs. I_CRAC was normalised for cell size by dividing the amplitudes (measured from the voltage ramps at −80 mV) by the cell capacitance. Capacitative currents were compensated before each ramp using the automatic compensation of the EPC 9–2 amplifier. All leak currents were subtracted by averaging the first two to four ramp currents, and then subtracting this from all subsequent currents. Data are presented as means ±s.e.m., and statistical evaluation was carried out using Student's unpaired t test.

During the initial stages of this work, we clamped cells at different voltages and monitored the inward current that subsequently developed. However, this approach was problematic for two reasons. First, background current subtraction is not possible when the DC current is measured continuously at a fixed voltage and where I_CRAC develops almost immediately upon break-in. Second, because I_CRAC is a small current (10–60 pA, depending on the holding potential), small changes in the leak would lead to an error in estimating the true size of I_CRAC. Such small changes in the leak would not be identifiable simply by observing the current at one set voltage. To circumvent these problems, we applied brief (50 ms) voltage ramps from −100 to +100 mV every 2 s from different holding potentials. Because I_CRAC has certain hallmarks when ramps are applied (inward rectification, voltage-independent gating over the 50 ms time scale and a reversal potential (V_rev) > +50 mV (Parekh & Penner, 1997), we were able to leak subtract the records as well as detect sudden changes in leak current (V_rev shifting to less positive values and less inward rectification). Recordings contaminated with changes in leak were discarded.

RESULTS

Recording conditions

In the experiments described below, we attempted to minimise the contributions of the previously documented inhibitory mechanisms that operate on I_CRAC, in order to examine the effect of membrane potential on I_CRAC in relative isolation. To this end, we dialysed cells with the standard internal solution supplemented with a maximal concentration of IP₃ (30 μm; Parekh et al. 1997), 10 mm EGTA/BAPTA and 2 μm thapsigargin. Under these conditions, I_CRAC is activated maximally, no refilling of stores can occur, Ca²⁺-dependent slow inactivation is largely suppressed and kinase-mediated inactivation is prevented (see Discussion for relevant references).

The size of I_CRAC depends on the holding potential

Figure 1Ai-iv shows the effects of different holding potentials (−80, −40, 0 and +40 mV) on the rate and extent of activation of I_CRAC. Cells were clamped at the given potential (V_h) immediately upon break-in and then maintained at this potential. The development of I_CRAC was monitored by applying fast voltage ramps (−100 to +100 mV in 50 ms) every 2 s. The upper panels of Fig. 1Ai-iv plot the amplitude of I_CRAC (measured from the ramps at −80 mV and normalised to cell capacitance) against time after break-in at the different holding potentials. The corresponding I-V relationships are shown below. At negative holding potentials, the size of the current was considerably smaller than that seen at more depolarised potentials. Pooled data from several cells are summarised in Fig. 1Bi. Each point represents data from 5–35 cells. At potentials more negative than −60 mV, a lower limit of −0.8 pA pF⁻¹ was obtained for the size of I_CRAC; the current density increased as the holding potential became more positive, reaching an upper limit close to −4.2 pA pF⁻¹ around +40 mV. I_CRAC was most sensitive to the holding potential over the range −20 to +20 mV. The open circles in Fig. 1Bi represent recordings from cells dialysed with a similar pipette solution except that IP₃ was omitted and Ca²⁺ was buffered at 120 nm (10 mm EGTA), in order to prevent passive depletion of the stores (Fierro & Parekh, 1999a). Under these conditions, no voltage-dependent currents were observed and the small linear leak (< −0.15 pA pF⁻¹ at −80 mV) was constant.

Figure 1. Conductance of CRAC channels depends on the holding potential.

A, cells were clamped at the indicated holding potentials (V_h; −80 mV in i, −40 mV in ii, 0 mV in iii and +40 mV in iv). I_CRAC was activated by dialysing cells with IP₃ (30 μm), 10 mm EGTA and 2 μm thapsigargin and was monitored by applying fast voltage ramps (−100 to +100 mV in 50 ms at 0.5 Hz) superimposed on the holding potential. The amplitude of the current was always measured at −80 mV from the ramps. Upper panels plot the amplitude of I_CRAC against time for each V_h (corresponding to different cells) and the lower panels show the I-V relationship obtained from the ramps. Bi, plot of the pooled data for I_CRAC (•) from several cells for each V_h. Each point represents data from between 5 and 35 cells. The relationship between current amplitude and holding potential could be reasonably well fitted with a modified Boltzmann-type equation of the form: I_CRAC (−80 mV) = 1/(1 + exp(V–V_½)/S) where V_½ is the voltage at which the current is one-half its maximal amplitude and S is the slope factor (RT/zF), where R, T and F have their usual meanings and z is the gating valency of the voltage-dependent step. For the pooled data, V_½ was −3.1 mV and z was 1.9. This is a macroscopic empirical estimate of z and is only a very rough indication. All points were significantly different from that at 0 mV (P < 0.05). The size of the non-store-operated current as a function of different holding potentials, measured for up to 300 s is also shown (^). For these recordings, cells were dialysed with a solution lacking IP₃ and Ca²⁺ was buffered at 120 nm (see text). Note that this current is constant, irrespective of the membrane potential, and very small. It would therefore not affect the voltage dependence of I_CRAC (•). Bii, summary of the relationship between holding potential and the time constant of activation (τ-activation). There was a tendency for the current to activate more quickly as the holding potential became more negative. τ-activation at −80, −60, −40 and −20 mV was significantly different from that at 0 mV, whereas τ-activation at +20 and +40 mV was not.

Although I_CRAC tended to be smaller at more negative potentials, it seemed to develop more quickly. Fitting the time course of I_CRAC activation (at different holding potentials) with a mono-exponential function indicated that the time constant of activation increased with depolarisation (Fig. 1Bii), being approximately 3.5-fold slower when the holding potential was changed from −80 to +40 mV. Although this was a general pattern, there was nevertheless a lot of variability in the behaviour of individual cells in that some activated as quickly at positive potentials as others did at more hyperpolarised ones.

For clarity of narration, we shall refer to the effect of the holding potential as hyperpolarisation-induced inhibition (HII) of I_CRAC. This arbitrary terminology does not indicate an underlying molecular mechanism.

Ca²⁺-dependent inactivation of I_CRAC does not fully explain HII

The preceding results demonstrate that the amplitude of I_CRAC (measured at −80 mV from brief voltage ramps) is reduced as the holding potential between ramps becomes more negative (up to −80 mV). Because we measured I_CRAC at −80 mV (from the ramps), the electrical driving force for Ca²⁺ entry through CRAC channels is always the same ((−80 –E_Ca) mV, where E_Ca is the equilibrium potential for Ca²⁺), irrespective of the pre-ramp holding potential. Therefore the reduction in the size of I_CRAC at negative holding potentials means that the whole-cell CRAC conductance has fallen. Negative holding potentials provide a favourable electrochemical gradient for Ca²⁺ influx and it is conceivable that the enhanced Ca²⁺ entry could inactivate CRAC channels through the well-documented process of Ca²⁺-dependent fast inactivation (Zweifach & Lewis, 1995a; Fierro & Parekh, 1999a). Ca²⁺-dependent fast inactivation arises from the build-up of a microdomain of Ca²⁺ in the vicinity of each open CRAC channel and this subsequently reduces channel activity with time constants of around 10 and 100 ms. Ca²⁺-dependent fast inactivation is little affected by the slow chelator EGTA, but can be attenuated by the faster Ca²⁺ chelator BAPTA. Reducing the amount of Ca²⁺ influx through each CRAC channel (by lowering external Ca²⁺ concentration, for example) can also reduce the extent of Ca²⁺-dependent fast inactivation. The contribution of Ca²⁺-dependent fast inactivation to HII was examined and the results are shown in Fig. 2. The left-hand and middle panels in Fig. 2A depict the time course of activation and the steady-state amplitude of I_CRAC from two different holding potentials (0 and −60 mV) for cells dialysed with either IP₃ and 10 mm EGTA (0 mV, n = 35; −60 mV, n = 16) or IP₃ and 10 mm BAPTA (0 mV, n = 5; −60 mV, n = 8). Thapsigargin (2 μm) was present in both solutions. In spite of dialysis with the faster chelator BAPTA, the current from the holding potential of −60 mV was significantly reduced compared with that at 0 mV. The right-hand panel of Fig. 2A plots the amplitude of I_CRAC versus the time constant of activation for the different Ca²⁺ chelators at the different holding potentials. The amplitude of I_CRAC at −60 mV in BAPTA-dialysed cells (point A₄) was significantly larger than the corresponding value for EGTA (A₂, P < 0.05), whereas the corresponding amplitudes at 0 mV were not significantly different. This indicates that BAPTA reduces the extent of HII, albeit modestly.

Figure 2. Reduction in the extent of fast Ca²⁺-dependent inactivation does not prevent HII of I_CRAC.

In A, the rate and extent of activation of I_CRAC at 0 and −60 mV holding potentials with IP₃+ EGTA + thapsigargin in the pipette are compared with those seen with IP₃+ BAPTA + thapsigargin. Both chelators were used at 10 mm. A₁ refers to IP₃+ 10 mm EGTA + thapsigargin at a V_h of 0 mV, A₂ to IP₃+ 10 mm EGTA + thapsigargin at a V_h of −60 mV, A₃ to IP₃+ 10 mm BAPTA + thapsigargin at a V_h of 0 mV and A₄ to IP₃+ 10 mm BAPTA + thapsigargin at a V_h of −60 mV. Each trace is from a different cell. Pooled data are summarised in the right-hand panel, which plots current amplitude against τ-activation. Each point represents data from 5–9 cells. In B, I_CRAC was activated passively following dialysis with 10 mm EGTA or BAPTA at the indicated holding potentials. IP₃ was not present in the patch pipette. Point B₁ here refers to 10 mm EGTA at a V_h of 0 mV, B₂ to 10 mm EGTA at a V_h of −60 mV, B₃ to 10 mm BAPTA at a V_h of 0 mV and B₄ to 10 mm BAPTA at a V_h of −60 mV. The right-hand panel plots the amplitude of I_CRAC against half-time to peak (time at which I_CRAC had reached half its steady-state amplitude) and this has been corrected for the delay before the current started to develop (typically around 70 s). In C, I_CRAC was activated by dialysing cells with IP₃ and 10 mm EGTA + thapsigargin under conditions where extracellular Ca²⁺ was reduced from 10 to 2 mm. Point C₁ refers to a V_h of 0 mV and C₂ to a V_h of −60 mV.

We found that the time course of activation was slightly faster at −60 mV compared with 0 mV when EGTA was the chelator (P < 0.05; compare A₁ and A₂ in Fig. 2A), whereas this was not the case with BAPTA (A₃vs. A₄). Comparing EGTA with BAPTA data at −60 mV, the respective time courses of activation were slightly different (19.0 ± 2.8 s, n = 16 and 33.9 ± 5.3 s, n = 8). Although this difference was significant (P < 0.05), the variability in BAPTA-dialysed cells was quite large and four cells activated at a rate indistinguishable from that seen in EGTA.

Because I_CRAC starts to activate within 3 s of breaking into the cell and develops with a time constant of 12–20 s at negative holding potentials (Fig. 1Bii), we were concerned that an insufficient amount of free chelator would have diffused into the cell during the time that I_CRAC developed. Furthermore, the chelator would probably be saturated because of the substantial Ca²⁺ release triggered by IP₃. With our typical series resistance of 7 MΩ and a cell capacitance of 18 pF, we calculate time constants of equilibration of 270 and 373 s for EGTA and BAPTA, respectively (Pusch & Neher, 1988). Hence cytoplasmic Ca²⁺ will be buffered only weakly as I_CRAC activates. We therefore sought an approach that would enable us to effectively load the cytoplasm with exogenous chelator before I_CRAC activated. Passive depletion of stores by dialysis with high concentrations of chelator, which we have previously characterised in detail (Fierro & Parekh, 1999b), is ideal for this because I_CRAC starts around 70 s after the onset of whole-cell recording, develops slowly (100–120 s to peak), and stores are depleted without any rise in cytoplasmic Ca²⁺. We therefore compared the amplitude of I_CRAC at −60 and 0 mV holding potentials, following passive depletion of stores with 10 mm EGTA (left-hand panel of Fig. 2B; 0 mV, n = 11; −60 mV, n = 7) or BAPTA (middle panel of Fig. 2B; 0 mV, n = 7; −60 mV, n = 6). The right-hand panel of Fig. 2B plots the amplitude of I_CRAC at −60 or 0 mV holding potential against half-time to peak for both EGTA and BAPTA. As for the experiments using IP₃, the amplitude of I_CRAC at −60 mV was significantly larger with BAPTA than EGTA (P < 0.05). Nevertheless, for both treatments I_CRAC was still substantially smaller when cells were held at −60 mV compared with 0 mV.

We then examined whether a reduction in the extent of Ca²⁺ entry could reduce the difference in I_CRAC amplitude at −60 vs. 0 mV holding potential. We therefore lowered the external Ca²⁺ concentration to 2 mm, slightly above the apparent K_D of the CRAC channel for Ca²⁺ in RBL cells (Fierro & Parekh, 2000). The results are summarised in Fig. 2C. The current was evoked by dialysing cells with IP₃ and 10 mm EGTA together with thapsigargin. Application of voltage ramps revealed that I_CRAC (measured at −80 mV from the ramps) was still more than twice as large in cells held at 0 mV (n = 6) than in those held at −60 mV (n = 6). Note that C₁ and C₂ in the right-hand panel of Fig. 2C are around 50% the amplitude of A₁ and A₂ in Fig. 2A, the latter being recorded in 10 mm Ca²⁺. Despite this reduction in current amplitude, the ratios of the currents at 0 mV (C₁/A₁) and −60 mV (C₂/A₂) were similar, indicating that the voltage dependence of I_CRAC was not suppressed by reducing the extent of Ca²⁺ entry.

Taken together, these results indicate that Ca²⁺-dependent fast inactivation alone cannot account for the HII of I_CRAC. Further evidence against Ca²⁺-dependent inactivation is described below (Fig. 4 and Discussion).

Figure 4. Voltage step relaxation studies reveal that the development of, and recovery from, HII is slow.

A, a cell was dialysed with IP₃+ 10 mm EGTA + thapsigargin and held at 0 mV. Once I_CRAC had reached steady state (•), the potential was changed to −60 mV. I_CRAC then decayed slowly to reach a new steady-state (^). Stepping back to 0 mV resulted in a mono-exponential increase in the current (•) which almost reached the level it had prior to pulsing to −60 mV. B, similar experiment to A in a different cell which was initially held at −60 mV (^) and then stepped to 0 mV (•). In C, the amplitude of I_CRAC is plotted against the activation time constant corresponding to each change in membrane potential. A₁ represents the size of the current and τ-activation when cells were held at 0 mV upon break-in. A₂ refers to the change in A₁ upon stepping to −60 mV. A₃ reflects the amplitude of I_CRAC and τ-activation when cells were held at −60 mV upon break-in and A₄ is the subsequent change when these cells were then stepped to 0 mV. Note that A₁ and A₄ are virtually identical, and A₂ and A₃ are not significantly different, indicating that the changes in I_CRAC following each voltage relaxation do not seem to depend on the preceding state. Voltage relaxation experiments in the absence of store depletion (dialysis with an internal solution without IP₃ and in which Ca²⁺ was strongly buffered at 120 nm to prevent passive store emptying) failed to generate slowly developing currents on stepping from 0 to −60 mV and vice versa. Instead, there was an instantaneous change in the current amplitude (< 10 pA). It is unlikely therefore that there was a major contribution from other conductances, including Na⁺-Ca²⁺ exchange, to the voltage relaxation experiments.

HII of I_CRAC does not reflect voltage-dependent block by intracellular Mg²⁺ or polycationic bases

Intracellular Mg²⁺ induces a voltage-dependent block of inwardly rectifying K⁺ channels and this contributes to the rectification (Matsuda et al. 1987). We entertained the possibility that Mg²⁺ might plug CRAC channels from the inside at hyperpolarised potentials, reducing channel conductance. To test this, we examined the effect of different intracellular Mg²⁺ concentrations on the rate and extent of activation of I_CRAC at 0 and −60 mV. Cells were dialysed with internal solution containing IP₃, thapsigargin and 10 mm EGTA and either 4 mm Mg²⁺ (0 mV, n = 3; −60 mV, n = 4) or 10 μm Mg²⁺ (buffered with 2 mm EDTA; 0 mV, n = 4; −60 mV, n = 7). The results were compared with those from cells dialysed with internal solution containing IP₃, thapsigargin and standard Mg²⁺ (1 mm; 0 mV, n = 5; −60 mV, n = 8). The left-hand panel of Fig. 3A compares the time course and extent of I_CRAC activation at −60 and 0 mV for 0 versus 4 mm internal Mg²⁺. Reducing or elevating internal Mg²⁺ concentration did not alter the amplitude of I_CRAC at −60 or 0 mV, nor did it affect the time constant of activation at either voltage, when compared with standard 1 mm Mg²⁺ (middle and right-hand panels of Fig. 3A).

Figure 3. HII of CRAC channels is not mediated by cytoplasmic Mg²⁺ or polycations.

A, cells were dialysed with IP₃+ 10 mm EGTA + thapsigargin and the indicated concentrations of Mg²⁺. 0 Mg²⁺ refers to a pipette solution in which Mg²⁺ was omitted and 2 mm EDTA was added together with 10 mm EGTA. The left-hand panel shows examples of recordings from cells dialysed with 0 or 4 mm Mg²⁺ at a holding potential of either 0 mV (filled symbols) or −60 mV (corresponding open symbols). The middle and right-hand panels show how the amplitude of I_CRAC and activation time constant are related to internal Mg²⁺ concentration. B shows the effects of the different Mg²⁺ concentrations on the properties of I_CRAC when the latter was evoked by passive depletion of stores (10 mm EGTA). The left-hand panel depicts data from 3 cells dialysed with 0, 1 (control) or 4 mm Mg²⁺ at a holding potential of 0 mV and the middle panel shows the relationship between Mg²⁺ concentration and current size following passive store depletion. Each point represents data from 3–6 cells. The right-hand panel shows that fast Ca²⁺-dependent inactivation is not affected by changes in intracellular Mg²⁺ levels. Cells were held at 0 mV and then stepped to −80 mV for 250 s. Scale bars refer to 50 ms and −0.5 pA pF⁻¹. In C, I_CRAC was activated following dialysis with 10 mm EGTA and 2 mm spermine or spermidine and cells were clamped at either 0 or −60 mV. The left-hand panel depicts examples of the development of I_CRAC in the presence of spermine (at either 0 (A₁) or −60 mV (A₂)) and spermidine (A₃ at 0 and A₄ at −60 mV). Data from 2 different cells are shown for each voltage. The middle graph plots the amplitude of I_CRAC against half-time (time at which I_CRAC had reached half its steady-state amplitude) for the different conditions. Filled circles represent the control (no polycation) at 0 mV and open circles correspond to the control at −60 mV. The right-hand panel shows that fast inactivation of I_CRAC is largely unaffected by the polycations compared with control upon pulsing to −80 mV. Scale bars as in B.

Because the effects of Mg²⁺ on Ca²⁺-dependent fast inactivation of I_CRAC are not known, we also examined whether intracellular Mg²⁺ affected this inactivation. We repeated the above experiments, but depleted stores passively with 10 mm EGTA instead. This was because we wanted to ensure that cytosolic Mg²⁺ concentration was reasonably similar to pipette Mg²⁺ concentration as I_CRAC started to develop. Consistent with the results of Fig. 3A where IP₃ was used, the amplitude of I_CRAC was not affected by different Mg²⁺ concentrations (Fig. 3B, middle panel, points represent data from 3–6 cells). The right-hand panel of Fig. 3B shows that I_CRAC inactivated to similar extents when 250 ms hyperpolarising steps to −80 mV were applied from a holding potential of 0 mV. The extent of inactivation was 47.3 ± 3% for control, 54.0 ± 3% for 0 Mg²⁺ and 46.7 ± 5% for 4 mm Mg²⁺. Hence cytoplasmic Mg²⁺ does not seem to alter fast inactivation.

Intracellular polycationic bases like spermine and spermidine exert voltage-dependent block of several ionic channels including the inwardly rectifying K⁺ channels (Ficker et al. 1994; Lopatin et al. 1994) and voltage-dependent Ca²⁺ channels (Scott et al. 1993). The effects of these polycations on I_CRAC have not been reported. We dialysed cells with either 2 mm spermine or 1.5 mm spermidine under conditions where I_CRAC was activated by passive store depletion and compared the properties of the current at 0 and −60 mV. Typical recordings are shown in the left-hand panel of Fig. 3C. The development of I_CRAC as well as its steady-state amplitude at either voltage were not affected by the polycations, demonstrating that polyamines did not have a clear effect on HII. The middle panel of Fig. 3C (points represent data from 4–11 cells) plots the amplitude of I_CRAC against the half-time to peak, corrected for delay. Neither parameter was significantly affected by spermine or spermidine when compared with control recordings (filled and open circles, reflecting 0 mV (n = 11) and −60 mV (n = 7), respectively) obtained in the absence of the polycations. We also examined the effects of spermine and spermidine on Ca²⁺-dependent fast inactivation (250 ms hyperpolarising steps to −80 mV from 0 mV holding potential). As shown in the right-hand panel of Fig. 3C, this inactivation was not altered by inclusion of the polycations (percentage inactivation was 40.0 ± 3, 34.3 ± 3 and 44.1 ± 2% for control, spermine and spermidine).

Voltage jump relaxations reveal that HII of I_CRAC is reversible

Figure 4 describes experiments in which we activated I_CRAC at one holding potential and then changed this potential to examine the rate and extent of relaxation to the new steady state. In Fig. 4A, the cell was initially held at 0 mV and I_CRAC was evoked by dialysing the cell with IP₃, 10 mm EGTA and thapsigargin. The current activated with a time constant of 22 s before reaching a plateau. We then changed the holding potential to −60 mV. I_CRAC decayed mono-exponentially with a time constant of around 30 s to the new steady-state level. On stepping back to 0 mV, I_CRAC increased mono-exponentially (time constant, 28 s) to attain almost the same amplitude it had reached prior to stepping to −60 mV. Figure 4B shows another recording from a different cell in which the pulse protocol was inverted compared with Fig. 4A. I_CRAC increased upon stepping to 0 mV and then relaxed mono-exponentially on stepping back to −60 mV. Pooled data from several cells are summarised in Fig. 4C, which plots the amplitude of I_CRAC at each potential against the time constant that led to this amplitude. Cells initially held at 0 mV are called A₁ (n = 5); the time constant of A₁ refers to the activation time constant of the current. These cells were then stepped to −60 mV. The amplitude and time constant of I_CRAC during this voltage relaxation are indicated by point A₂. Other cells were initially clamped at −60 mV (A₃; n = 5) and then stepped to 0 mV (A₄). Strikingly, the amplitudes and time constants of I_CRAC at −60 mV were very similar irrespective of whether the cell was initially held at −60 mV or stepped to this potential after being clamped at 0 mV. Similarly, the amplitudes and time constants at 0 mV were almost identical whether the cell had been held at 0 mV from the outset of whole-cell recording or stepped to 0 mV from −60 mV. This indicates that the CRAC channels respond to voltage relaxation jumps in a manner that is not influenced by the preceding state. No detectable hysteresis was present in the system. We repeated the voltage step relaxation experiments, but this time dialysed the cells with 10 mm BAPTA instead of EGTA. On stepping to −60 mV from a holding potential of 0 mV (once I_CRAC had reached a stable level of −2.86 ± 0.21 pA pF⁻¹, n = 5), I_CRAC relaxed with a time constant of 4.94 ± 0.8 s to reach a new steady-state amplitude that was 55.4 ± 4.7% of that at 0 mV. This time constant was significantly different from that seen with EGTA (P < 0.05).

The amplitude of macroscopic I_CRAC in low Ca²⁺ buffer is reduced at hyperpolarised potentials

Recently we have discovered that I_CRAC can be measured in the presence of physiological intracellular Ca²⁺ buffer provided SERCA pumps are blocked (Fierro & Parekh, 2000). We therefore compared the size of I_CRAC at a holding potential of −60 mV with that at 0 mV following dialysis with IP₃ and thapsigargin in low Ca²⁺ buffer (0.1 mm EGTA). As stated in the Introduction, −60 mV was chosen to mimic the conditions of Artalejo et al. 1998. Whereas we were able to detect I_CRAC consistently at a holding potential of 0 mV (7 of 7 cells), results at −60 mV were more variable (Figure. 5). In 3 of 8 cells we failed to detect any I_CRAC at all and in those where we did see the current it was quite small (−0.54 ± 0.14 pA pF⁻¹; 5 cells). Holding cells at −60 mV therefore reduces the probability of recording I_CRAC and, if the current is detectable, it has a much smaller amplitude when compared with that at 0 mV.

Figure 5. I_CRAC in low intracellular Ca²⁺ buffer is very small at a holding potential of −60 mV.

A shows the time course of I_CRAC for 2 cells held at 0 mV (filled symbols) and for 2 different cells held at −60 mV (open symbols). Internal solution contained IP₃+ 0.1 mm EGTA + 2 μm thapsigargin. Note that one of the cells held at −60 mV did not generate I_CRAC at all. In B, the amplitude histogram compares the size of I_CRAC in low Ca²⁺ buffer for the 2 different holding potentials. At a V_h of 0 mV, I_CRAC was around 5-fold larger than at a V_h of −60 mV.

DISCUSSION

Our results provide new insight into the behaviour of store-operated CRAC channels. Whole-cell CRAC conductance appears to be, directly or indirectly, voltage dependent in that hyperpolarising potentials reduce channel activity. This is somewhat unexpected because one of the hallmarks of I_CRAC is its voltage-independent gating, as evinced by a weak inward rectification during fast voltage ramps (−100 to +100 mV in 50 ms). We believe that the voltage-dependent behaviour which we have described here has been overlooked in previous reports by us and others both because a holding potential of 0 mV has usually been employed and because the voltage-dependent inhibition develops slowly. A holding potential of 0 mV for I_CRAC recordings is chosen for two main reasons. First, Ca²⁺ entry between ramps is rather small at 0 mV (< −0.5 pA pF⁻¹) and therefore the exogenous Ca²⁺ buffer (usually several millimolar) can maintain cytosolic Ca²⁺ concentration at a low level. This will prevent Ca²⁺-dependent slow inactivation as well as store refilling, both of which would curtail CRAC channel activity. Second, at 0 mV, Ca²⁺-dependent fast inactivation does not occur (Fierro & Parekh, 1999a) and therefore whole-cell CRAC conductance is high. Because the ramps are of only 50 ms duration whereas HII of I_CRAC develops with a time constant of several seconds at −60 mV depending on the Ca²⁺ chelator, the latter would not be seen in the I-V relationship obtained from ramps.

Is I_CRAC the only current recorded under our conditions?

Because we have employed different holding potentials, it is possible that other currents, in addition to I_CRAC, contribute to the size of the whole-cell current we have recorded. However, we do not think our recordings were contaminated to any significant extent by other currents for the following reasons. First, irrespective of the holding potential, the I-V relationship and reversal potential were the same and clearly reflected I_CRAC (inward rectification, voltage independent and V_rev > +50 mV). If other currents had come into play then one would have expected the I-V relationship and V_rev to change. Second, we have carried out experiments under identical conditions except stores were not depleted. No voltage-dependent currents were observed (open circles in Fig. 1Bi). Hence the voltage dependence we describe reflects properties of I_CRAC rather than a contribution from another current.

Previously described inactivation mechanisms cannot adequately account for HII of I_CRAC

Ca²⁺-dependent fast inactivation of CRAC channels, reflecting the build-up of a microdomain of high Ca²⁺ in the vicinity of each open channel, is prominent at potentials > −60 mV and is therefore a prime candidate for the HII of I_CRAC (Zweifach & Lewis, 1995a; Fierro & Parekh, 1999a) However, there are several arguments against this. First, fast inactivation (in the presence of EGTA) in RBL-1 cells develops with a double-exponential time course (τ₁ of 10 ms and τ₂ of 120 ms) and recovers with time constants of 34 and 233 ms. With BAPTA, the corresponding time constants are 9 and 250 ms (Fierro & Parekh, 1999a). However, the voltage-dependent relaxation studies of Fig. 4 demonstrate that HII of I_CRAC takes place with a time constant of around 30 s with EGTA and 5 s with BAPTA. The collapse of Ca²⁺ microdomains (which would occur on stepping back to 0 mV from −60 mV) proceeds on a sub-millisecond time scale (Neher, 1998), several orders of magnitude faster than the voltage relaxation time course. Second, fast inactivation is substantially reduced at −60 mV when BAPTA is used instead of EGTA in the pipette solution (Fierro & Parekh, 1999a). However, the results of Fig. 2B indicate that HII of I_CRAC at −60 mV is still prominent in the presence of BAPTA. Third, fast inactivation is virtually absent at < −20 mV. However, the amplitude of I_CRAC increases continuously from −20 to +40 mV, even though fast inactivation would not be involved over this range.

Although I_CRAC was slightly larger at hyperpolarised potentials under conditions where fast inactivation was reduced, and the time course of the voltage step relaxations was faster with BAPTA than EGTA, it is nevertheless clear that fast inactivation alone cannot account for the inhibition at negative potentials.

In addition to fast inactivation, other mechanisms can inhibit I_CRAC over a time frame of tens of seconds. These are mediated by protein kinase C (Parekh & Penner, 1995), a Ca²⁺-dependent slow inactivation (Zweifach & Lewis, 1995b; Parekh, 1998) and Ca²⁺-dependent store refilling (Zweifach & Lewis, 1995b; D. Bakowski & A. B. Parekh, manuscript in preparation). We do not think that any of these mechanisms plays a significant role here because we have taken precautions to minimise their effectiveness. ATP was omitted from the pipette solution (a manoeuvre that suppresses kinase inactivation; Parekh & Penner, 1995), Ca²⁺ was strongly buffered with 10 mm EGTA or BAPTA (attenuating slow inactivation; Zweifach & Lewis, 1995b; Parekh, 1998) and store refilling was prevented by the inclusion of thapsigargin in the pipette solution at a concentration which routinely activates I_CRAC on its own (indicating effective SERCA pump inhibition; Fierro & Parekh, 2000). Hence HII of I_CRAC cannot be fully accommodated into the well-characterised mechanisms already known to reduce CRAC channel activity in RBL-1 cells.

Mechanism of HII of I_CRAC

The mechanism whereby sustained hyperpolarisation reduces CRAC channel conductance is not clear. Unlike inwardly rectifying K⁺ channels and NMDA receptors, it does not seem to involve voltage-dependent block by cytoplasmic Mg²⁺ or polycations such as spermine and spermidine. Because the inhibition could be repetitively evoked in whole-cell voltage relaxation experiments over a time course of several hundred seconds, it is unlikely to involve a small diffusible factor since one would expect such a molecule to be effectively washed out of the cell into the recording pipette.

One possibility is that HII of I_CRAC reflects voltage-dependent binding (at hyperpolarised potentials) and unbinding (at depolarised potentials) of a relatively large molecule such as a protein. One potential candidate is the IP₃ receptor spanning the Ca²⁺ store membrane. It has been suggested that this receptor physically links to the store-operated Ca²⁺ channel (Berridge, 1995; Kiselyov et al. 1998; Paterson et al. 1999; see Putney, 1999, for review). Because the IP₃ receptor has a large amino terminal cytoplasmic head with many polar residues, our results could be explained by a voltage-dependent interaction between the IP₃ receptor and CRAC channels. In this scenario, hyperpolarisation would favour dissociation of the protein-protein complex, whereas depolarisation would stabilise it. Such binding and unbinding would have to be rather slow to account for a time constant of around 30 s. Binding/unbinding would also have to be Ca²⁺ dependent because the kinetics are accelerated by BAPTA. Alternatively, the inhibition at negative potentials could represent an intrinsic property of the CRAC channels themselves. In this scheme, hyperpolarisation would presumably inactivate the channels rather than deactivate them because mean closed time is shorter at hyperpolarised than at depolarised potentials for single CRAC channels with Na⁺ as the charge carrier (Kerschbaum & Cahalan, 1999). However, 45 s hyperpolarising pulses to −80 mV from 0 mV did not reveal any inactivation other than the well-described biphasic inactivation time course of CRAC (data not shown). Because certain enzymes in the plasma membrane are voltage dependent (e.g. adenylate cyclase in cerebellar neurons; Reddy et al. 1995), it is possible that changes in membrane potential alter the activity of such enzymes, resulting in a biochemical cascade that culminates in the modification of CRAC channel conductance.

The amplitude of I_CRAC depends on the holding potential despite maximal store emptying

Although we dialysed cells with a combination of agents that would rapidly empty the IP₃-sensitive Ca²⁺ stores and keep them empty (IP₃+ 10 mm EGTA/BAPTA + thapsigargin), we nevertheless obtained a range of I_CRAC amplitudes depending on the holding potential. I_CRAC amplitude at −60 mV was almost three times less than at 0 mV. The extent of the current was clearly graded with the holding potential. However, this does not reflect graded depletion of stores since the latter were essentially empty. Hence sub-maximal I_CRAC can be seen under conditions where stores are depleted, and this seems to arise from the voltage-dependent conductance changes in the macroscopic CRAC conductance. HII of I_CRAC may therefore be a powerful way to grade the extent of Ca²⁺ entry through CRAC channels, even following maximal store emptying.

Since macroscopic CRAC conductance appears to be voltage dependent, investigating the relationship between the amount of Ca²⁺ released from the stores and subsequent Ca²⁺ influx in non-voltage-clamped cells becomes complex. Without knowing the value of the membrane potential or how it changes with time, one cannot rule out the possibility that HII of I_CRAC exaggerates/masks the extent of Ca²⁺ influx depending on where the membrane potential actually is. In addition, changes in membrane potential will alter the extent of Ca²⁺ influx simply by changing the electrical driving force for Ca²⁺ entry. Hence the conclusions from such experiments may be somewhat limited (Sedova et al. 2000).

Physiological implications of HII of I_CRAC

Most RBL-1 cells have a resting potential of around −90 mV (Fierro & Parekh, 1999a), a value sufficiently negative to ensure that a substantial fraction of the CRAC channels are affected by HII. At rest it would appear therefore that HII is a major inhibitory mechanism that reduces macroscopic CRAC activity. Current-clamp recordings have revealed that stimulation of FCεRI receptors with antigen does not seem to change the membrane potential (Fernandez & Lindau, 1986; Fierro & Parekh, 1999a), which may be explained by the observation that antigen receptor stimulation activates inwardly rectifying K⁺ channels which would clamp the resting potential close to E_K (Gericke et al. 1995). HII of I_CRAC would therefore remain effective and the macroscopic CRAC conductance would be kept at a low level. On the other hand, stimulation of P_2y purinoceptors on RBL cells with ATP activates large conductance non-selective cation channels which would shift the membrane potential towards 0 mV (Obukhov et al. 1995). Such a depolarisation would reduce the effects of HII and thereby increase the whole-cell CRAC conductance around 3-fold. In current-clamped RBL cells, relatively large fluctuations in the membrane potential have been observed following application of thapsigargin (Mason et al. 1999), and this seems to reflect an interaction between inwardly rectifying K⁺ channels and the store-operated Ca²⁺ current. These depolarisations were large, and could reach −17 mV, a potential where HII is occurring. Hence stimulation of distinct cell-surface receptors may evoke different patterns of intracellular Ca²⁺ signalling depending on an interplay between their ability to change the membrane potential and hence macroscopic CRAC conductance and the subsequent electrical driving force for Ca²⁺ entry.

The impact of HII in low Ca²⁺ buffer is particularly high. At −60 mV holding potential, and in agreement with a previous report (Artalejo et al. 1998), we often failed to detect any macroscopic I_CRAC at all and, if present, the current was significantly smaller than that seen at 0 mV holding potential. These experiments were carried out in 10 mm external Ca²⁺, to increase the size of the current. In 2 mm Ca²⁺, I_CRAC is around 60% the size of that in 10 mm Ca²⁺ (Fierro & Parekh, 2000). This means that under physiological conditions of low intracellular Ca²⁺ buffer, a strongly hyperpolarised membrane potential and 2 mm external Ca²⁺, I_CRAC will be virtually undetectable (< 0.27 pA pF⁻¹). Both HII and Ca²⁺-dependent slow inactivation can account for the dramatic reduction in I_CRAC at −60 mV. Although we are currently unable to accurately estimate the contribution of each mechanism, a rough indication is that the two contribute equally because in high Ca²⁺ buffer I_CRAC is reduced around 3-fold at −60 mV (compared with at 0 mV holding potential) whereas in low Ca²⁺ buffer this is approximately 6-fold.

The fact that I_CRAC can barely be resolved in low Ca²⁺ buffer at a holding potential of −60 mV in the whole-cell configuration should not be taken to indicate that the current is of little physiological relevance. For an RBL-1 cell of 15 pF capacitance, we calculate that just a −2 pA whole-cell I_CRAC will raise cytosolic Ca²⁺ concentration at the sizeable rate of 50 nm s⁻¹ (taking accessible cell volume as 50% and a cytosolic Ca²⁺ binding ratio of 75; Neher, 1995).

Since I_CRAC is around 6-fold larger at 0 mV than at −60 mV in the presence of physiological intracellular Ca²⁺ buffer, maximal activation of the current may be accomplished through rhythmic oscillations in membrane potential. A strong hyperpolarisation would initially result in a large I_CRAC (due to the increased electrical driving force for Ca²⁺ entry) but this would subsequently decline due to both Ca²⁺-dependent inactivation and HII. A modest depolarisation would increase the conductance of CRAC channels and this would tend to offset the fall in driving force, thereby maintaining Ca²⁺ entry. A subsequent hyperpolarisation would now result in more Ca²⁺ influx before HII developed further. In this way, rhythmic depolarisations and hyperpolarisations would be more effective in sustaining Ca²⁺ entry through CRAC channels than maintaining the membrane potential at a large hyperpolarised level. The membrane potential in most non-excitable cells can fluctuate between E_K (set by inward rectifiers and Ca²⁺-dependent K⁺ channels) and 0 mV (Ca²⁺-activated non-selective cation channels). Oscillations in membrane potential in this range would impact upon the extent of HII. Our results may also help explain one puzzling paradox in the calcium signalling field. It is thought that the proteins encoded by certain members of the Trp family of genes underlie store-operated Ca²⁺ channels (Zhu & Birnbaumer, 1998). However, Trp proteins are expressed in high levels in excitable cells where I_CRAC has not been reported. Although I_CRAC might not involve a Trp product, it is nevertheless tempting to speculate that the inability to record I_CRAC in neurons (and most other excitable cells, but see Fomina & Nowycky, 1999) stems from the fact that the cells have to be clamped at very negative potentials in order to prevent activation of voltage-dependent Ca²⁺ as well as other channels. At such negative potentials, HII of I_CRAC will be pronounced and this, coupled with Ca²⁺-dependent slow inactivation, may suppress the activation of I_CRAC. In this scenario, I_CRAC is indeed present in neurons but is simply not detectable at hyperpolarised potentials.

The Ca²⁺ current through single CRAC channels is too small to be resolved. Although our macroscopic measurements provide only limited insight into channel behaviour at a microscopic level, it is instructive nevertheless to consider potential gating schemes. One such scheme is depicted below (Scheme 1).

Scheme 1.

In this model, open CRAC channels (representing an unknown number (n) of open states) can enter the rapid Ca²⁺-inactivated states (I₁-Ca and I₂-Ca, representing the biphasic recovery from fast inactivation; Fierro & Parekh, 1999a) or HII. For simplicity, we have not included slow Ca²⁺-dependent or kinase-mediated inactivation processes. Because HII occurs even under conditions in which I₁-Ca and I₂-Ca are reduced, we think open channels can enter HII directly and not necessarily via one of the other inactivated states. It is striking that the time constant for development of HII (τ_HII) is around 28 s at −60 mV, a value only slightly larger than that for activation of I_CRAC (τ_act) at a similar voltage (18 s). If τ_HII >> τ_act, then HII would contribute little to the development/extent of activation of I_CRAC over the time scale of most receptor-evoked Ca²⁺ signals. If τ_HII << τ_act, HII would act simply as a DC offset. The fact that τ_HII≈τ_act indicates that HII can alter the extent of I_CRAC even as it develops. In this regard, HII will be ideally primed to dynamically modulate the pattern of Ca²⁺ entry at all stages of current development.

Experiments are now being designed to delineate the molecular mechanisms underlying HII and to investigate whether it is a general feature of CRAC channels or restricted to RBL cells.