FCDI (fast Ca2+-dependent inactivation) is a mechanism that limits Ca2+ entry through Ca2+ channels, including CRAC (Ca2+ release-activated Ca2+) channels. This phenomenon occurs when the Ca2+ concentration rises beyond a certain level in the vicinity of the intracellular mouth of the channel pore. In CRAC channels, several regions of the pore-forming protein Orai1, and STIM1 (stromal interaction molecule 1), the sarcoplasmic/endoplasmic reticulum Ca2+ sensor that communicates the Ca2+ load of the intracellular stores to Orai1, have been shown to regulate fast Ca2+-dependent inactivation. Although significant advances in unravelling the mechanisms of CRAC channel gating have occurred, the mechanisms regulating fast Ca2+-dependent inactivation in this channel are not well understood. We have identified that a pore mutation, E106D Orai1, changes the kinetics and voltage dependence of the ICRAC (CRAC current), and the selectivity of the Ca2+-binding site that regulates fast Ca2+-dependent inactivation, whereas the V102I and E190Q mutants when expressed at appropriate ratios with STIM1 have fast Ca2+-dependent inactivation similar to that of WT (wild-type) Orai1. Unexpectedly, the E106D mutation also changes the pH dependence of ICRAC. Unlike WT ICRAC, E106D-mediated current is not inhibited at low pH, but instead the block of Na+ permeation through the E106D Orai1 pore by Ca2+ is diminished. These results suggest that Glu106 inside the CRAC channel pore is involved in co-ordinating the Ca2+-binding site that mediates fast Ca2+-dependent inactivation.
The activity of Ca2+ channels within the plasma membrane is tightly regulated through a variety of mechanisms which ensure that Ca2+ entering the cell is delivered: (i) to the correct places; (ii) at the correct time; and (iii) in the correct amount (for a review, see ). FCDI (fast Ca2+-dependent inactivation) is a negative-feedback mechanism that limits the amount of Ca2+ passing through an ion channel (, for a review see ). This type of regulation is typical for a number of voltage-gated Ca2+ channels present in excitable cells, but is also found in voltage-independent CRAC (Ca2+ release-activated Ca2+) channels expressed in non-excitable cells [4,5]. Generally it is thought that Ca2+ passing through the channel pore binds to a site 3–4 nm from the intracellular mouth of the pore, most likely located on the channel itself, and this causes channel inactivation [5,6]. The discovery and cloning of the molecular components of CRAC channels, STIM1 (stromal interaction molecule 1), a Ca2+-binding single-transmembrane domain protein which acts as the sensor of Ca2+ in the endoplasmic reticulum lumen [7–9], and Orai1, a four-transmembrane domain protein that forms the pore of CRAC channels [10–12], has enabled extensive research into the biophysical properties of this channel, including FCDI [13,14]. It has been suggested that fast inactivation of the ICRAC (CRAC current) is caused by Ca2+ binding within the pore itself, as mutations of glutamate residues within the selectivity centre of Orai1 (E106D and E190Q) or residues close to it (V102I) were shown to significantly reduce or abolish ICRAC inactivation [13,14]. Subsequently, other regions of Orai1 and STIM1 were identified as mediators of fast inactivation [15–17]. A series of seven negatively charged amino acid residues (470–485) within STIM1 constitutes the so-called CMD (CRAC modulatory domain), plus the cytosolic calmodulin-binding domain (residues 68–91) of Orai1, are both required for FCDI [15–17]. Conserved C-terminal glutamate residues within Orai1 have also been reported to mediate FCDI . The involvement of both STIM1 and Orai1 in the mechanism of FCDI of ICRAC is consistent with the dependence of ICRAC kinetics on the relative expression levels of STIM1 and Orai1 . Specifically, it has been identified that higher levels of STIM1 relative to Orai1 expression produces ICRAC with strong FCDI, similar to that observed in the native ICRAC of haematopoietic and liver cells. In contrast, higher expression of Orai1 relative to STIM1 produces ICRAC that shows reduced inactivation or, in some cases, even potentiation at negative potentials . Currents obtained from cells with higher expression levels of Orai1/STIM1 look superficially similar to those produced by V102I, E190Q and E106D mutants [13,14]. However, the possibility that these mutants may also be affected by the relative expression ratios of Orai1/STIM1 has not been investigated.
In the present study, using controlled expression ratios of Orai1/STIM1, we identify that the Orai1 mutants V102I, E106D and E190Q do not abolish fast Ca2+-dependent inactivation. Similar to the WT (wild-type) Orai1, V102I and E190Q produce ICRAC with fast inactivation that is strongly dependent on the relative expression of STIM1 and Orai1. In contrast, the kinetics of the Ca2+ current through the pore formed by the E106D Orai1 mutant, measured in the absence of any other externally permeable cations, was unaffected and independent of the ratio of STIM1/Orai1. Furthermore, the inactivation properties of the E106D Orai1 mutant were vastly different from that of WT Orai1. Specifically, E106D Orai1 is characterized by very fast inactivation kinetics and a significant shift in the voltage dependence to more positive potentials. Unlike the WT Orai1 channel, the E106D mutant is not blocked by low pH, suggesting that protonation of Glu106 blocks Ca2+ permeation through the WT Orai1 pore. Relief of the Na+ current block by Ca2+ in the E106D mutant at low pH suggests that protonation of Asp106 weakens Ca2+ binding inside the pore and confirms that Glu106 is an important residue regulating Ca2+ permeability through the channel. The pH dependence of the E106D Orai1 inactivation is consistent with the notion that Ca2+ binding to the selectivity centre not only regulates Ca2+ permeability, but also controls channel gating.
Cell culture and transfections
H4IIE rat liver cells (A.T.C.C. CRL 1548) and HEK-293T cells [human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40)] (A.T.C.C. CRL 11268) were cultured at 37°C in 5% (v/v) CO2 in air in DMEM (Dulbecco's modified Eagle's medium) supplemented with 100 μM non-essential amino acids, 2 mM L-glutamine and 10% fetal bovine serum. To express STIM1, WT Orai1 and Orai1 mutants, cells plated on glass cover slips were transfected using Polyfect (Qiagen) transfection reagent according to the manufacturer's instructions. Where given, the two plasmids were transfected in the ratios shown. Where the ratio is not shown, the ratio used was 2:1 STIM1/Orai1. Experiments were performed 24–48 h following transfection.
STIM1 (GenBank® accession number NM_003156) and Orai1 (GenBank® accession number NM_032790) were subcloned into pCMV-Sport6 and the GFP (green fluorescent protein) co-expressing vector pAdTrack-CMV . The V102I Orai1, E106D Orai1 and E190Q Orai1 mutations were generated using pCMV-Sport6-Orai1 as a template according to the protocol specified by the QuikChange® II site-directed mutagenesis kit (Stratagene). Appropriate primers were designed and synthesized (Invitrogen) to include the following mutations into Orai1: V102I, E106D or E190Q. The codon changes were respectively: GTG to ATC, GAG to GAC and GAG to CAG. All mutations were validated by DNA sequencing (Department of Haematology, Flinders University, Adelaide, Australia). The positions of the mutations in Orai1 are shown in Supplementary Figure S1 (at http://www.BiochemJ.org/bj/441/bj4410743add.htm).
Whole-cell patch clamping was performed at room temperature (23°C) using a computer based patch-clamp amplifier (EPC-9, HEKA Elektronik) and PULSE software (HEKA Elektronik). The control bath solution contained 140 mM NaCl, 4 mM CsCl, 10 mM CaCl2, 2 mM MgCl2 and 10 mM Na-Hepes adjusted to pH 7.4 with NaOH. Depletion of intracellular Ca2+ stores was achieved using 20 μM Ins(3,4,5)P3 (Sigma) added to an internal solution containing, unless otherwise stated, 130 mM caesium glutamate, 4 mM CaCl2, 1 mM MgATP, 5 mM MgCl2, 10 mM EGTA and 10 mM Na-HEPES adjusted to pH 7.2 with NaOH. Patch pipettes were pulled from borosilicate glass and fire polished to give a pipette resistance between 3 and 5 MΩ. Series resistance did not exceed 15 MΩ and was 50–70% compensated.
Following the establishment of whole cell configuration, the development of ICRAC was monitored by applying voltage ramps from −138 to +102 mV every 2 s. All voltages shown have been corrected for the liquid junction potential between the bath and electrode solutions, −18 mV when Na+ is the dominant external cation (estimated by JPCalc ). The holding potential was −18 mV throughout. Current traces were acquired at a 33 kHz sampling rate and filtered at 11 kHz. The EPC-9 amplifier compensated for cell capacitance automatically. Leakage was subtracted from all traces shown, and was obtained by blocking currents using 50 μM 2-APB (2-aminoethoxydiphenyl borate) or 10 μM La3+ (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/441/bj4410743add.htm).
Normalized instantaneous tail currents for voltage steps to −118 mV after test pulses in the range of −158 to 102 mV were used to produce the apparent Po (open probability) curves by fitting with the Boltzmann distribution with an offset of the form:
Where Pmin is an offset, V is the membrane potential, V1/2 is the half-maximal activation potential (please note that V1/2 corresponds to the inflexion point of the Po curve) and k is the slope factor.
To determine the amplitude of the instantaneous tail currents and to minimize the error due to the cell capacitance, the tail currents were fitted with a single exponential function to the beginning of the −118 mV pulse.
Macroscopic tail current amplitudes fitted with the Boltzmann equation (eqn 1) are routinely used to estimate relative Po values of different types of voltage-gated channels. This method assumes the presence of a voltage sensor, the movements of which in the electric field are coupled to the opening and closing of the channel pore. Currently there is no evidence that WT and mutant CRAC channels have such a voltage sensor. However, as Ca2+ binding to specific sites in the vicinity of the inner part of the CRAC channel pore causes FCDI , voltage-dependent movement of Ca2+ through the channel pore to these sites renders the open probability of CRAC channels voltage dependent. Phenomenologically, FCDI of CRAC channels can be described using the same approach as for voltage-gated channels (eqn 1); however, the same mechanistic interpretation does not necessarily apply.
To approximate the time course of ICRAC inactivation at negative potentials raw current traces were fitted with an equation of the form:
Where A is the amplitude of the exponential component, τ its time constant, C is the amplitude of the steady state component and t is time.
Kinetics of V102I and E109Q Orai1 mutants depend on relative STIM1 expression
When WT Orai1 and STIM1 were expressed in a heterologous expression system with an excess of STIM1, Orai1 produced ICRAC with pronounced FCDI, similar to that of native ICRAC found in a range of cell types [4,20] (Figure 1A). We have previously shown that by altering the relative amounts of Orai1 and STIM1 plasmids during transfection it is possible to reproducibly control the relative expression levels of these proteins, and that cells expressing an excess of Orai1 produced ICRAC values that showed activation at negative potentials  (Figure 1A). Using this approach, we transfected H4IIE cells with mixtures of STIM1 and V102I or E190Q Orai1 mutants in 4:1 and 1:4 ratios, and investigated the kinetics of the resulting ectopic currents using whole-cell patch clamping (Figure 1). The extent of inactivation or activation was quantified by normalization of the amplitude of the current at the end of a 200 ms, −118 mV pulse (I200) to the amplitude of the current at the beginning of the pulse (I0) to give a ratio of I200/I0.
Dependence of V102I and E190Q Orai1 currents on the relative expression levels of STIM1
Cells transfected with a ratio of 1:4 STIM1/V102I Orai1 displayed currents with very brief inactivation over the first 10 ms of the voltage step to −118 mV, followed by potentiation (Figure 1B), and were qualitatively similar to the currents reported by Spassova et al.  using a similar voltage protocol. The average I200/I0 under this transfection protocol was 1.07±0.15 (n=4). In contrast, cells transfected with a ratio of 4:1 STIM1/V102I Orai1 and subject to the same voltage protocol produced currents that rapidly inactivated before approaching a steady state by the end of 200 ms (Figure 1B), with a significantly smaller average I200/I0 of 0.57±0.03 (n=8, P<0.05).
Similar to V102I Orai1 and WT Orai1, the gating kinetics of the E190Q Orai1 mutant also showed strong dependence on the relative amounts of STIM1/Orai1 transfected (Figure 1C). Cells transfected with a ratio of 1:4 STIM1/E190Q Orai1 produced an activating current in response to voltage steps to −118 mV with an average I200/I0 of 0.83±0.07 (n=8), which was similar to the current recorded by Yamashita et al.  in 20 mM external Ca2+. Cells transfected with a ratio of 4:1 STIM1/E190Q Orai1 produced an inactivating current with a significantly smaller average I200/I0 of 0.56±0.03 (n=7, P<0.05). As E190Q Orai1 conducted both Ca2+ and Na+ , in control experiments Na+ was replaced with TEA+ (tetraethylammonium) or NMDG+ (N-methyl-D-glucamine). However, the absence of Na+ had no influence on E190Q kinetics or its dependence on the relative amounts of STIM1 (results not shown).
To ensure that our results were not influenced by an artefact of the expression system, the experiments were repeated in both H4IIE and HEK-293T cells. Similar to findings in H4IIE cells, when V102I or E190Q mutants were expressed in HEK-293T cells, both indicated the kinetics of inactivation were strongly dependent on the relative expression ratios with STIM1 (results not shown).
Experiments with a third Orai1 mutant, E106D, suggested that its properties are significantly different from those of the WT Orai1 and required thorough investigation before the effect of expression ratios could be studied.
E106D Orai1 displays fast inactivation at negative potentials
The glutamate residue at position 106 forms part of the selectivity centre of the Orai1 pore, and its mutation to aspartate renders Orai1 permeable to Na+ [11,14,21,22]. Consistent with this, E106D Orai1 co-expressed with STIM1 in either H4IIE or HEK-293 cells produced currents with an I-V (current-voltage) relationship that was significantly different from that of WT Orai1 (Figure 2A). Specifically, E106D Orai1 is characterized by having a large outward current and a less positive reversal potential compared with WT Orai1, which was consistent with the presence of a significant Cs+ conductance. Replacing Na+ with TEA+ reduced the inward current amplitude and further shifted the reversal potential to more negative potentials, confirming that the E106D pore was permeable to extracellular Na+ (Figure 2A).
Properties of the E106D Orai1 current
In our initial attempts to investigate FCDI of E106D Orai1, we used an approach similar to that used above, recording currents in response to 200 ms steps to −118 mV, in an external bath solution containing 0 mM Na+ and 100 mM Ca2+ to eliminate Na+ conductance through the E106D pore. Using a 2:1 STIM1/E106D Orai1 transfection ratio, we were unable to detect any significant inactivation of the current in agreement with a previous study  (results not shown). However, although this data confirmed that FCDI of ICRAC was strongly affected by the E106D mutation, we considered the possibility that FCDI was simply too fast to be detected or its voltage dependence was shifted too far towards depolarizing potentials to be clearly seen under these recording conditions. To overcome these limitations, we opted to investigate E106D Orai1 currents using a modified protocol that covered a wide range of voltages between −132 and 102 mV, and increased the sampling rate 10-fold to 33 kHz to enable us to record faster kinetics. We also employed a 50 ms prepulse to +62 mV to maximally activate the E106D Orai1 current, which was not required for WT Orai1 or other mutants in the present study, as they were already fully activated at the holding potential of −18 mV.
Under standard conditions in the extracellular solution containing 10 mM Ca2+ and 140 mM Na+, the E106D Orai1 current recorded in response to steps to negative potentials after a prepulse to +62 mV decayed to a steady state within 10–15 ms (Figure 2B). The rapid decay of E106D Orai1 current under similar conditions has been previously reported and proposed to be due entirely to a voltage-dependent block of Na+ entry by Ca2+, and not FCDI . Indeed, in divalent-cation-free bath solution, the Na+ current through E106D Orai1 did not show any decay at negative potentials or any voltage dependence (results not shown) . However, removal of Ca2+ would also remove FCDI; therefore, to investigate Ca2+-dependent gating of E106D Orai1 mutant, we replaced all monovalent cations in the bath with Ca2+ or NMDG+.
Using this voltage protocol and a bath solution containing no Na+ and 100 mM Ca2+, we were able to record E106D Orai1 currents which showed very fast inactivation during the steps to potentials below −80 mV (Figure 2D). As the inactivation was extremely fast (τ~0.1–0.2 ms), it could only be detected in the cells with low and stable series resistance, which were amenable to a reliable subtraction of leakage and capacitance currents. During steps to less negative potentials (~−40 mV), the inactivation became slower and clearly distinguishable from cell capacitance currents (Figure 2D, and see also Supplementary Figure S3 at http://www.BiochemJ.org/bj/441/bj4410743add.htm).
Given that inactivation of ICRAC slows at lower Ca2+ concentrations , we used a bath solution with 10 mM Ca2+ to slow down the kinetics of the E106D Orai1 current. All monovalent cations were replaced with impermeable TEA+ or NMDG+ . Under these conditions, currents recorded from E106D Orai1 in response to negative steps after a pre-pulse to +62 mV clearly showed inactivation at negative potentials (Figure 3A). The time course of current traces obtained in response to 50 ms steps to negative potentials could be approximated by the sum of an exponential component and a steady state (see the Experimental section, eqn 2). The time constant of the exponential component showed voltage dependence with slower inactivation at less negative potentials (Figure 3B). Currents recorded in response to steps to −138 mV had a small second exponential component of the inactivation, which was ignored. The apparent Po of the E106D Orai1 reached maximum at voltages above 100 mV and minimum (Pmin) at voltages below −80 mV. Compared with WT Orai1, the V1/2 value of the apparent Po curve of E106D Orai1 was shifted by approximately 80 mV towards more positive potentials (Figures 3C and 3D) .
E106D Orai1 Ca2+ current displays fast inactivation in the absence of permeable monovalent cations
To ensure that the results of the present study were not affected by the method of store depletion, we conducted control experiments using the SERCA (sarcoplasmic/endoplasmic reticulum Ca2+ ATPase) inhibitor thapsigargin (2 μM) instead of Ins(3,4,5)P3 in the pipette solution to activate ICRAC mediated by WT Orai1 and the Orai1 mutants under study. The results showed that replacing Ins(3,4,5)P3 with thapsigargin had no effect on the FCDI of WT Orai1 and the Orai1 mutants (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/441/bj4410743add.htm). This is consistent with the recent studies showing that FCDI of WT Orai1/STIM1 mediated currents activated by thapsigargin exhibit the same dependence on the relative expression levels of STIM1 and Orai1 as the current activated by Ins(3,4,5)P3 [18,23].
Ca2+ dependence of E106D Orai1 mutant gating
The results presented above show that, in the presence of 10 mM Ca2+ in the bath solution and all other cations replaced with either TEA+ or NMDG+, E106D exhibits voltage-dependent inactivation. The significant difference in the kinetics of inactivation between currents recorded in 10 mM and 100 mM external Ca2+ (compare Figures 2D and 3A) suggests that E106D inactivation might be mediated by Ca2+ passing through the channel pore, a mechanism similar to FCDI in WT Orai1. Since FCDI is believed to be a result of Ca2+ binding to a site near the intracellular mouth of the pore , we used intracellular Ca2+ buffers that bind Ca2+ at different rates as a tool to determine if inactivation of E106D is mediated by Ca2+, an approach previously used to investigate inactivation of native ICRAC in several different cell types [4,24]. Using a pipette solution containing BAPTA [1,2-bis-(o-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid], a faster Ca2+ buffer than EGTA, resulted in a larger minimum Po at very negative potentials [0.15±0.02 (n=7) compared with 0.28±0.04 (n=4), P=0.005, at −138 mV], indicating a reduced inactivation at these potentials, but also a stronger inactivation at potentials above −50 mV and therefore a 30 mV shift of the Po curve to more positive potentials (Figure 4A). A small increase in the steady state current at −118 mV (Figures 4A and 4D) was consistent with a similar finding by Yamashita et al. . There was no significant difference between the time constants of inactivation of E106D currents obtained with internal solutions using EGTA or BAPTA (Figure 4B). In contrast, with WT Orai1, using BAPTA in the pipette solution resulted in a significant 35 mV shift of the Po curve to more negative potentials (Figure 4A) and a much greater increase in the steady state amplitude (Figure 4C). Essentially, BAPTA has opposite effects on the Po curves of WT and E106D Orai1, suggesting that the underlying mechanisms of inactivation of the channels are different.
Ca2+ dependence of E106D Orai1 mutant gating
Next, we investigated the extracellular Ca2+ dependence of E106D Orai1 current inactivation. When extracellular [Ca2+] was reduced from 10 mM to 2 mM, the apparent Po increased at all potentials, the V1/2 value of the Po curve shifted from −9±3 mV (n=7) to −57±7 mV (n=3) (Figure 4E), and the time constant of current inactivation at negative potentials significantly increased (Figure 4F). These data indicate that inactivation was reduced and was slower in lower extracellular Ca2+. As outlined above, using 100 mM extracellular Ca2+ made the E106D Orai1 current inactivation very difficult to resolve from uncompensated cell capacitance and fitting such fast currents with exponentials was prone to large errors (Figure 2D). Nevertheless, the apparent Po curve obtained in 100 mM Ca2+ showed a significant shift to positive potentials, confirming that E106D Orai1 gating is Ca2+-dependent (Figure 4E). This shift in voltage dependence of E106D Orai1 current in 100 mM Ca2+ to depolarizing potentials explains why E106D current does not show any inactivation during steps from the holding potentials close to 0 mV. At 0 mV and 100 mM Ca2+, E106D Orai1 channels are already near their minimum Po value (Figure 4E), therefore stepping from 0 to even the most negative potentials does not produce any additional inactivation.
To further explore the difference between the inactivation mechanisms of E106D and WT Orai1 channels, we investigated the effects of Ca2+ substitutions by other divalent cations. We have previously reported that, when Ca2+ is replaced with Ba2+ or Sr2+ as the charge carrier, Orai1/STIM1 channels display complex conductance and gating properties . Compared with Ca2+, both Ba2+ and Sr2+ produced much less fast inactivation of expressed WT Orai1/STIM1 current and native ICRAC [5,18,24]. Both WT and E106D Orai1 displayed similar changes in conductance and gating when Ba2+ was substituted for Ca2+ as the charge carrier. Replacing Ca2+ with Ba2+ in the absence of other permeable cations resulted in a larger E106D Orai1 current that showed little inactivation at negative potentials (Figures 5A and 5B). In contrast, in the presence of Sr2+, the E106D Orai1current strongly inactivated in a manner similar to the Ca2+ current (Figures 5A and 5B). Analysis of the tail currents revealed that in the presence of Ba2+ the voltage dependence of E106D Orai1 was quite weak and could not be described by a single Boltzmann function (Figure 5C), similar to the voltage dependence of the WT Orai1 Ba2+ current . In contrast, the voltage dependence of E106D Orai1 in the presence of Sr2+ was similar to that in the presence of Ca2+ (Figure 5C), and significantly different from the voltage dependence of WT Orai1 in the presence of Sr2+ (Figure 5D).
E106D Orai1 gating in the presence of Sr2+ and Ba2+
Dependence of the E106D Orai1 mutant on relative expression of STIM1
Having established that in the absence of all permeable extracellular monovalent cations E106D Orai1 exhibits Ca2+-dependent inactivation, we investigated if the transfection ratios of STIM1 and Orai1 altered the kinetics of E106D currents. The transfection ratio was varied between 4:1 STIM1/E106D Orai1 and 1:8 STIM1/E106D Orai1. Unlike WT Orai1, E106D-mediated current showed very strong inactivation at all transfection ratios used. The relative steady state current, I50/I0, was 0.19±0.02 (n=6) and 0.20±0.03 (n=4) at −118 mV for E106D/STIM1 ratios of 4:1 and 1:4 respectively.
It has been previously demonstrated that co-transfection of WT Orai1 with a truncated version of STIM1 called CAD (CRAC activation domain) (residues 342–448) leads to expression of constitutively active CRAC currents which did not show any FCDI . We were interested to know if co-transfection of CAD with E106D Orai1 would also abolish inactivation of E106D Orai1 currents. This co-transfection produced currents identical to those produced by E106 Orai1 and WT STIM1 (results not shown), suggesting that inactivation of E106D Orai1 does not involve interaction with STIM1.
E106D Orai1 mutant dependence on pH
Endogenous ICRAC recorded in human macrophages is inhibited by low extracellular pH . ICRAC mediated by heterologously expressed WT Orai1 and STIM1 in H4IIE cells is also inhibited by low extracellular pH, with complete block at pH 5.5 or less (Figures 6A and 6B). The pKa value of WT Orai1 pH dependence is 7.8±0.1 (n=4) (Figure 6B), which is very close to that previously reported for the native ICRAC . Na+ conductance through WT Orai1 both in the presence of extracellular Ca2+ and in the absence of extracellular divalent cations displayed the same pH dependence (results not shown).
Dependence of WT Orai1 and E106D Orai1 on extracellular pH
Having altered a negatively charged residue in the E106D Orai1 pore (glutamate has a carboxylic acid side chain with a pKa value of 4.3 and overall pI value of 3.22, whereas aspartate has a carboxylic acid side chain with pKa 3.9 and overall pI value of 2.77), we investigated whether this would alter the pH dependence of the E106D Orai1 conductance compared with WT Orai1. Lowering the pH from 7.4 to 6.3 or below changed the shape of the I-V plot of E106D Orai1, increasing the amplitude of the current at negative potentials (Figure 6C). Application of voltage steps revealed that the rapid decay of E106D Orai1 current was reduced in more acidic extracellular solutions (Figure 6D), suggesting that the Ca2+ block of Na+ current was reduced. Importantly, these data indicate that the Ca2+-binding site that limits Na+ permeation may be protonated at low pH values. As the peak current amplitude of E106D Orai1 current recorded in response to voltage steps was not affected by pH (Figure 6D), but the steady state current was, we normalized steady-state current at the end of 50 ms pulse to the peak current at the beginning of the pulse (I50/I0) to build the titration curve and to obtain the apparent pKa value of the Ca2+ binding site. The Hill equation fitted to the data points revealed an apparent pKa value of 5.9±0.1 (n=3) (Figure 6E).
The elimination of the Ca2+ block of Na+ current through the E106D Orai1 pore by low pH is consistent with a reduction in Ca2+ binding relative to Na+ at the blocking site. To investigate the influence of pH on Ca2+ current through E106D Orai1, all extracellular monovalent cations were replaced by TEA+. Lowering extracellular pH resulted in some reduction of Ca2+ current; however, no significant block could be achieved even at a pH value as low as 5.3. Notably, at this pH, the shape of the I-V plot changed and the reversal potential was significantly shifted towards more negative potentials (ΔE=−16.4±1.2 mV, n=5), suggesting reduced Ca2+ permeability relative to intracellular Cs+, and supporting our hypothesis that Ca2+ binding within the pore is reduced at low pH values (Figure 7A). Application of voltage steps revealed that low pH reduced E106D Orai1 inactivation (Figure 7B). Low pH significantly increased the relative amplitude of the steady state current, reduced the voltage dependence of gating and shifted it to more negative potentials (Figure 7C).
pH dependence of E106D Orai1 Ca2+ current
The two primary conclusions from this study are: (i) mutation of Glu106, which forms the selectivity centre of the Orai1 pore, to an aspartate residue significantly changes the properties of Ca2+-dependent gating of the CRAC channel and abolishes the block of ICRAC by protons; and (ii) mutation of two other residues located in or close to the Orai1 pore, Glu190 and Val102 to glutamine and isoleucine respectively, has no effect on the dependence of FCDI on the relative expression ratios of STIM1 and Orai1. These novel findings are in contrast with results reported in two previous studies which suggested that Orai1 point mutations at residues 102 and 190 abolished the FCDI of ICRAC. Furthermore, these findings significantly expand our knowledge about the role of Glu106 in CRAC channel gating and permeation [13,14].
The exact mechanism of FCDI of CRAC channels remains unknown. Based on the characteristics of FCDI and general assumptions of Ca2+ diffusion, the Ca2+-binding site responsible for fast inactivation of CRAC channels was proposed to be 3–4 nm from the intracellular mouth of the pore . Since the cloning of the molecular components of the CRAC channel, STIM1 and Orai1, a number of studies have demonstrated that mutations in both Orai1 and STIM1 can alter FCDI, suggesting that its mechanism may require interaction between the two. An anionic region of STIM1 near the C-terminus known as CMD (residues 474–485) or IDSTIM (residues 470–491) is reported to be required for fast inactivation. Truncated STIM1 lacking this region produces ICRAC which shows no inactivation, and mutation of anionic residues in this region to neutral residues significantly reduces FCDI [15–17]. Mutations of an N-terminal Orai1 domain (residues 68–91) that prevent calmodulin binding also greatly reduce FCDI. However, mutations of Tyr80 to alanine or serine residues which conserve calmodulin binding produce fast inactivation kinetics similar to E106D Orai1 . The involvement of calmodulin in the FCDI of Orai1/STIM1-mediated current is consistent with our previous findings that expression of a calmodulin mutant unable to bind Ca2+ or the use of calmodulin inhibitor peptides reduced the FCDI of native ICRAC in liver cells . The intracellular loop between transmembrane domains 2 and 3 of Orai1 was also reported to be critical for FCDI. Mutation of this region prevents inactivation, whereas intracellular application of peptides derived from the loop blocks CRAC current, leading Srikanth et al.  to suggest that this region may block the pore of Orai1 during inactivation. Adding to this complexity, we have identified that the kinetics of the Orai1/STIM1-mediated current depends on the relative expression levels of Orai1 and STIM1 .
The dependence of FCDI on the relative expression levels of STIM1 and Orai1 is the likely cause of the apparent discrepancy of the results presented in the present study with those previously published for V102I and E190Q mutants [13,14]. In the present study we show that, as with WT Orai1, increasing the amount of V102I Orai1- or E190Q Orai1-containing plasmid in the transfection mixture produces currents that show activation at negative potentials, whereas increasing the amount of STIM1 plasmid increases the extent of inactivation of the current. Since these specific point mutants show the same behaviour as WT Orai1, we conclude that residues Glu190 and Val102 are unlikely to participate in the FCDI mechanism when expressed at physiologically relevant levels.
In contrast, the results with the E106D Orai1 mutant shown in the present study do support the notion that Glu106 of Orai1 participates in CRAC channel gating . It has been shown previously that if all extracellular cations were replaced with 110 mM Ca2+, E106D Orai1 current does not inactivate at negative potentials . However, by extending the Ca2+ concentration range and the voltage range of recording and using NMDG+ or TEA+ to replace permeable Na+, we have identified that in extracellular solutions containing 2 or 10 mM Ca2+, E106D Orai1 exhibits very fast inactivation at negative potentials, with the apparent voltage dependence of gating shifted to more positive potentials compared with WT Orai1. The strong dependence of E106D Orai1 kinetics and the apparent Po on external Ca2+ concentration demonstrates that inactivation of the E106D mutant is indeed Ca2+-dependent, and is confirmed by the lack of inactivation when extracellular Ca2+ is replaced with 10 mM Ba2+. However, several lines of evidence indicate that the Ca2+-dependence of E106D Orai1 gating is vastly different from the FCDI in the WT Orai1. First, there is no leftward shift of the Po curve in the E106D mutant when BAPTA is replaced with EGTA in the pipette solution. The present model assumes that FCDI in CRAC channels is mediated by Ca2+ that enters the cytoplasmic space though the channel pore and binds somewhere close to its intracellular mouth . This is largely based on the observation that using the fast Ca2+ chelator BAPTA instead of the slow Ca2+ chelator EGTA significantly diminishes FCDI [5,24]. The lack of such an effect of BAPTA in the E106D Orai1 mutant suggests that the Ca2+-binding site that regulates E106D Orai1 gating is different from that of WT Orai1. Secondly, unlike WT Orai1 , E106D Orai1 produces the same currents when co-expressed with either CAD or WT STIM1, which suggests that STIM1 is not involved in the inactivation of E106D current and that the E106D mutation may affect the interaction of Orai1 with STIM1. The lack of any dependence of E106D Orai1 Ca2+-dependent inactivation on the expression levels of STIM1 also supports the notion that the E106D mutation affects the interaction between Orai1 and STIM1. This is in contrast with a strong dependence of the kinetics on the relative expression ratios of STIM1 and WT, E190Q or V102I Orai1 (, Figure 1). Thirdly, although neither Ba2+ nor Sr2+ supports the FCDI of heterologously expressed WT Orai1 , inactivation of E106D Orai1 currents is supported by Sr2+ but not Ba2+. Importantly, these data provide further evidence that demonstrates that the selectivity of the binding site that regulates FCDI of WT Orai1 is different from the selectivity of the site responsible for Ca2+-dependent inactivation of the E106D mutant.
Although Ca2+-dependent inactivation of the E106D mutant and the WT Orai1 seem to have fundamental differences, there is one important similarity. The slopes of the apparent Po curves of both WT Orai1 and the E106D mutant are approximately 33. The slope of the Po curve is indicative of the number of gating charges fully traversing the electric field during voltage-dependent channel opening and closing . The voltage dependence of ICRAC is mediated by voltage-dependent movement of Ca2+ through the channel pore, and Ca2+ plays the role of a gating charge. As the slope of the Po curves of WT and E106D Orai1 mutant corresponds to an apparent gating charge of ~0.75, the Ca2+-binding site responsible for Ca2+-dependent inactivation lies somewhere closer to the outer side of the pore at an electrical distance of approximately 0.38. It is notable that the apparent valence of the voltage-dependent Ca2+ block of Na+ permeation through CRAC channels is also ~0.75 [14,29]. This places the Ca2+-binding site that regulates FCDI and the site that regulates permeation through the pore at a similar location. Regulation of FCDI by Ca2+ binding to the selectivity filter has been demonstrated in Cav1.2 voltage-gated Ca2+ channels, in addition to the well-known role of calmodulin . To explain the dependence of FCDI on both Ca2+ binding to the selectivity filter and to calmodulin tethered to the cytoplasmic tail of Cav1.2, a model has been proposed where Ca2+ binding to an intracellular site increases the affinity of the selectivity filter for Ca2+ and Ca2+ directly blocks the pore. It is possible that a similar interaction between Ca2+ binding to the selectivity centre and an intracellular Ca2+-binding site exists in WT CRAC channels. However, it seems that in the E106D mutant the intracellular Ca2+-binding site does not contribute to Ca2+-dependent inactivation, as it is not affected by BAPTA or the relative expression ratios of STIM1/Orai1. The exact mechanism of Ca2+-dependent gating of WT Orai1 and E106D mutant, however, still remains to be elucidated. Further experiments are required to identify whether a model similar to that of FCDI of voltage-gated Ca2+ channels also applies for the CRAC channel.
An unexpected finding was that the E106D mutation in Orai1 abolished the block of CRAC channels by protons (low pH). Previous work carried out in macrophages led to speculation that the residue responsible for pH dependence of ICRAC was likely to be a cysteine, due to its high pKa value of 8.2 . The results in the present study suggest that the protonatable residue mediating the pH dependence of ICRAC is Glu106 in WT Orai1 and Asp106 in the E106D mutant. Although the pKa value of glutamate and aspartate residues in aqueous environments is approximately 4.3 and 3.9, within proteins it can be as high as 8.8 and 8.1 respectively [31,32].
The results obtained at low pH support the notion that the properties of the Ca2+-binding site within the Orai1 pore can affect FCDI. The lack of a Ca2+-dependent block of Na+ permeation and reduced Ca2+ permeability relative to Cs+ at low pH is evidence that protonation of Asp106 changes Ca2+ binding in the pore by reducing the energy well for Ca2+. At the same time, low pH significantly changes the kinetics of the current by increasing the relative size of the steady-state component of the current.
The block of WT Orai1 by protons and the absence of this block in E106D Orai1 can be explained by the possible changes in pore size and Ca2+-binding energy wells caused by protonation. The pore of the WT Orai1 is quite narrow [~3.8 Å (1 Å=0.1 nm)] . Protonation of Orai1 at Glu106 would reduce the negative charge in the pore, thus reducing the energy well for Ca2+. In addition, protonation of the Glu106 residues that form a negatively charged ring may cause a slight reduction in the pore diameter due to the reduced repulsion between the side chains facing the pore. The combination of these two factors provides a plausible explanation for the observed complete block of WT Orai1 current at low pH. Owing to the smaller side-chain of aspartic acid compared with glutamic acid, and possibly due to the more negative pKa value of aspartic acid compared with glutamic acid, the pore of the E106D mutant is significantly larger (~5.3 Å) . Although protonation of Asp106 also reduces the energy well for Ca2+, the reduction of the E106D pore diameter due to protonation alone is apparently insufficient to affect ion permeation, as the pore size remains significantly larger than that of the WT Orai1. Therefore E106D retains measurable conductance at low pH.
In conclusion, the results from the present study provide novel evidence that the ring of negative charges formed by position 106 in the multimeric Orai1 pore controls not only the selectivity of the channel, but also contributes to a complex mechanism of FCDI and accounts for the pH dependence of the CRAC channel. Whether the effects of E106D mutation on ICRAC kinetics are entirely due to changes in Ca2+-binding at this site or whether this mutation also has an allosteric effect on the intracellular Ca2+-binding site mediating FCDI remains the topic of future investigations.
CRAC activation domain
Ca2+ release-activated Ca2+ modulatory domain
Ca2+ release-activated Ca2+
fast Ca2+-dependent inactivation
- HEK-293T cell
human embryonic kidney-293 cell expressing the large T-antigen of SV40 (simian virus 40)
stromal interaction molecule 1
Nathan Scrimgeour contributed to conception and design of the study, collection and assembly of data, data analysis and interpretation and wrote the paper. David Wilson contributed to conception and design of the study, data interpretation and wrote the paper. Grigori Rychkov contributed to conception and design of the study, data analysis and interpretation and wrote the paper. All authors contributed to final approval of the manuscript prior to submission.
We would like to thank Dr Linlin Ma (School of Medical Sciences, University of Adelaide, Adelaide, Australia) for the advice on mutagenesis and for providing the STIM1 CAD construct.
This work was supported by the National Health and Medical Research Council (Australia) [grant number 519115] and Senior Research Fellowship (to G.Y.R.).