Identification of key amino acid residues responsible for internal and external pH sensitivity of Orai1/STIM1 channels

Changes of intracellular and extracellular pH are involved in a variety of physiological and pathological processes, in which regulation of the Ca2+ release activated Ca2+ channel (ICRAC) by pH has been implicated. Ca2+ entry mediated by ICRAC has been shown to be regulated by acidic or alkaline pH. Whereas several amino acid residues have been shown to contribute to extracellular pH (pHo) sensitivity, the molecular mechanism for intracellular pH (pHi) sensitivity of Orai1/STIM1 is not fully understood. By investigating a series of mutations, we find that the previously identified residue E106 is responsible for pHo sensitivity when Ca2+ is the charge carrier. Unexpectedly, we identify that the residue E190 is responsible for pHo sensitivity when Na+ is the charge carrier. Furthermore, the intracellular mutant H155F markedly diminishes the response to acidic and alkaline pHi, suggesting that H155 is responsible for pHi sensitivity of Orai1/STIM1. Our results indicate that, whereas H155 is the intracellular pH sensor of Orai1/STIM1, the molecular mechanism of external pH sensitivity varies depending on the permeant cations. As changes of pH are involved in various physiological/pathological functions, Orai/STIM channels may be an important mediator for various physiological and pathological processes associated with acidosis and alkalinization.

Scientific RepoRts | 5:16747 | DOI: 10.1038/srep16747 inhibition, as well as alkalosis-induced promotion of platelet aggregation is mediated by the changes of store-operated Ca 2+ entry 12 . Moreover, store-operated Ca 2+ entry was shown to mediate intracellular alkalinization in neutrophils 13 , and extracellular low pH was reported to inhibit I CRAC in macrophages 14 . In Jurkat T-lymphocytes, cytosolic alkalinization induces Ca 2+ release and Ca 2+ entry 15 , and acidic internal and external pH inhibit I CRAC 16 . In SH-SY5Y neuroblastoma cells, however, store-operated Ca 2+ entry was not affected by changes of intracellular pH, even though it was attenuated by low extracellular pH and potentiated by high extracellular pH 17 . In smooth muscle cells, extracellular acidosis decreases store operated Ca 2+ entry, whereas extracellular alkalosis potentiates it 18 . Thus, it seems that changes of both intracellular and extracellular pH regulate I CRAC activity or store-operated Ca 2+ entry, albeit there are some discrepancies among different studies.
Since regulation of I CRAC seems to play a critical role in acidosis-and alkalosis-associated physiological and pathological processes, it is essential to understand the molecular basis underlying pH regulation of I CRAC . As activation of I CRAC requires coupling of Orai and STIM as well as gating of Orai [19][20][21][22][23][24] , alterations of either the coupling of Orai/STIM or gating properties of the pore-forming subunit Orai may cause functional changes of I CRAC . Indeed, it was demonstrated that intracellular low pH caused by oxidative stress induces uncoupling of Orai1 and STIM1, thereby inhibiting I CRAC 25 , and that intracellular high pH causes store depletion, thereby activating I CRAC . Moreover, mutation of the Ca 2+ selective filter residue E106 in the channel pore (E106D) has been shown to alter acidic pH-dependent inhibition of I CRAC 26 . Furthermore, mutation of D110 and D112 (D110/112A) leads to reduced external pH sensitivity of Orai1/STIM1 27 . Whereas it is known that regulation of pore-forming subunit Orai1 by protons contributes to external pH sensitivity of Orai1/STIM1, the molecular mechanisms by which I CRAC is regulated by internal pH is not fully understood.
Here we show that internal acidosis and alkalosis, as well as external acidosis and alkalosis markedly change Orai1/STIM1 channel functions. By investigating a series of mutants generated on residues located in the channel pore region, intracellular and extracellular loops, N-and C-termini, as well as transmembrane domains (TM3), we found that, in agreement with a previous report 26 , E106 is responsible for pH o sensitivity when Ca 2+ is the permeant cation. However, we found that E106 has no influence on pH o sensitivity when Na + is the charge carrier. Unexpectedly, we identified that the amino acid residue E190 located in TM3 of Orai1 is the major sensor of pH o when Na + is the charge carrier. Furthermore, we found that H155 located in the intracellular loop is responsible for intracellular pH sensitivity. Our results indicate that internal and external pH can regulate Orai1/STIM1 channel function by modulating the pore-forming subunit Orai1. Interestingly, our results suggest that the molecular basis for pH sensitivity when Ca 2+ is the charge carrier is different from that of when Na + is the charge carrier, an experimental condition which has been used for investigating pH regulation on Ca 2+ -selective and Ca 2+ -permeable channels. Thus, caution needs to be taken when extrapolating the mechanisms of pH sensitivity obtained using Na + as the permeant cation to the physiological conditions when Ca 2+ is the charge carrier. As E106, E190, and H115 are conserved residues in all the three isoforms of Orai, it is conceivable that they are the common external and internal pH sensors of different isoforms of Orai/ STIM channels.

Results
Effects of extracellular pH on Orai1/STIM1 currents. Orai1/STIM1 currents were recorded by including high EGTA concentration in the pipette solution to passively induce store depletion. The effects of extracellular pH on Orai1/STIM1 were evaluated by perfusing the cells with external divalent free solutions (DVF) at various pHs after Orai1/STIM1 activation reached a steady-state. As shown in Fig. 1A, Orai1/STIM1 currents were elicited by a ramp protocol ranging from − 100 to + 100 mV in the DVF extracellular solutions. Current amplitude was significantly increased when the cell was exposed to high pH o . Without store depletion, Orai1/STIM1 currents were not able to be induced by high pH o , indicating that basic pH o potentiates but does not activate Orai1/STIM1 channels (Fig. S1). In contrary to the effects of alkaline pH o , acidic pH o markedly inhibited current amplitude. A concentration dependent effect of external pH on Orai1/STIM1 is shown in Fig. 1B. Current amplitude was enhanced 3-to 4-fold at pH o 9, and was inhibited to a minimal level at pH o 4.5. The effects of pH were reversible as shown in Fig. 1B. The changes of current amplitude at various pH normalized to the current amplitude at pH 7.4 are shown in Fig. 1C. The best fit of the dose-response curve yielded a pKa of 8.26 ± 0.11 (Fig. 1C). Similar pKa (8.32 ± 0.11) was also obtained by the best fit of the normalized currents in reference to the maximal current amplitude (Fig. 1D).
The results shown in Fig. 1 were obtained using DVF solution because DVF solution produces larger current amplitude. We next tested the effects of various pHs on Orai1/STIM1 currents recorded in Tyrode's solutions containing 2, 20 and 120 mM Ca 2+ , respectively. As show in Fig. 2A-D, pH o 5.5 inhibited and pH o 8.2 enhanced Orai1/STIM1 inward current independent of extracellular Ca 2+ concentrations. In non-transfected control cells, high or low pH o did not induce any current (Fig. S2). The averaged current amplitude under different conditions is shown in Fig. 2E. Two-way ANOVA analysis indicated that the effect of pH o on Orai1/STIM1 was independent of extracellular Ca 2+ concentrations. The ratios of the current amplitude at various extracellular Ca 2+ concentrations versus the current amplitude recorded in 2 mM Ca 2+ Tyrode solution at pH o 8.2 were similar to those at pH o 7.4 (Fig. 2F), suggesting that potentiation of Orai1/STIM1 by basic pH was not significantly influenced by extracellular Ca 2+ concentrations. Similarly, the normalized ratios of Ca 2+ current at different Ca 2+ concentrations versus Na + current in DVF at pH o 8.2 are well superimposed with the ratios at pH o 7.4 (Fig. 2G), further suggesting that modulation of Orai1/STIM1 channel activity by protons is not dependent on the charge carrier. Thus, we first used DVF extracellular solution to investigate the effects of pH o on Orai1/STIM1 channels.
Effects of external pH on Orai2/STIM1 and Orai3/STIM1 currents. Before we went on to investigate the molecular mechanism of pH o regulation on Orai1/STIM1, we tested if external pH regulates channel activity of Orai2/STIM1 and Orai3/STIM1. As shown in Fig. 3, Orai2/STIM1 and Orai3/STIM1 currents were significantly potentiated by basic pH o and inhibited by acidic pH o (Fig. 3A,B). The fold changes of current amplitude by normalizing current amplitude at each pH to that of pH o 7.4 are shown in Fig. 3C. The maximal increases in Orai2/STIM1 and Orai3/Sim1 at high pH are about 3 fold, similar to the maximal increase of Orai1. The dose-response curves obtained by normalizing current amplitude at each pH to the maximal current amplitude are shown in Fig. 3D. The dose-response curves of Orai1/ STIM1, Orai2/STIM1, and Orai3/STIM1 are well superimposed. The pKa obtained from the best fit of the dose-response curves are 8.32 ± 0.14 and 8.52 ± 0.21 for Orai2/STIM1 and Orai3/STIM1 respectively, similar to the pKa of Orai1/STIM1 shown in Fig. 1 (dotted lines in Fig. 3). These results indicate that Orai1/STIM1, Orai2/STIM1 and Orai3/STIM1 have similar pH o sensitivity.
Mechanisms of external pH regulation on Orai1/STIM1. Activation of Orai1/STIM1 involves coupling of Orai1 and STIM1 as well as gating of Orai1. Since the external pH enhanced Orai1/STIM1 current amplitude after the channel was fully activated, we reasoned that protons may directly modulate the pore-forming subunit Orai1. To understand the mechanism by which external protons regulate Orai1/STIM1, we generated mutations by neutralizing a series of negatively charged residues located on the external site of the channel or along the channel pore, including E106Q, D110N, D112/114N, and E190Q. We also generated the mutations E106D and E190D. The negatively charged residues E106, D110, and E190 are conserved residues in all three isoforms of Orai, whereas the negatively charged residues of D112 and D114 are only conserved in Orai1 and Orai3 (Fig. S3). The mutant E106Q produced minimal current, consistent with the inability of E106Q to produce Ca 2+ influx 20 and the dominant-negative effects of E106Q reported previously 28,29 . Thus, we did not investigate E106Q in detail. For all the other mutants, representative recordings at pH o 5.5, 7.4 and 9.5 are shown in Fig. 4A1-A5. Changes of current amplitude at each pH o are shown in Fig. 4B1-B5, and the dose-response curves obtained by normalizing .0, the potentiation of D110N was greater than that of WT but did not reach statistical significance (Fig. 4B2), and the potentiation of D112N/D114N was smaller than that of WT ( Fig. 4B3) without reaching statistical differences. The dose-response curve of E190Q was slightly right-shifted without causing significant changes in pKa (p > 0.05; Fig. 4C4), albeit the increase of E190Q current amplitude by high pH o was greater than that of WT Orai1/STIM1 (Fig. 4B4). By contrast, the dose-response curve of E190D was significantly shifted to the left, resulting in a pKa that is almost two pH units lower than that of WT Orai1/STIM1 (p < 0.05; Fig. 4C5). The significant change in pKa in the E190D indicates that E190 is essential for the extracellular pH sensitivity of Orai1/STIM1 channels in DVF solution. Since E190Q only generated a small change in pKa, whereas E190D shifted the dose-response curve significantly, it is likely that the length of the side chain rather than the charge of E190 plays a key role in pH o modulation of Orai1/ STIM1. Taken together, although D110N and D112N/D114N produce some changes in potentiation of current amplitude at very high pH o , it seems that the conserved residue E190 plays a major role in pH o sensitivity of Orai1/STIM1 when Na + is the charge carrier.
Effects of internal pH i on Orai1/STIM1 channels. To study whether internal pH (pH i ) also influences Orai1/STIM1 channel activity, we titrated the pipette solution at various pHs. At acidic pH i 5.5, Orai1/STIM1 current amplitude was only about 10% of the current amplitude at pH i 7.4 ( Fig. 5A), whereas at basic pH i 8.4, the current amplitude was almost two-fold greater than the current amplitude at pH i 7.4 (Fig. 5A). The internal pH regulates Orai1/STIM1 channel activity in a concentration-dependent manner (Fig. 5B), with the maximal increase of current amplitude by 2-fold at pH i 9 (Fig. 5B). The best fit of the dose-response curve yielded pKa of 7.46 (Fig. 5C). Thus, similar to the effects of pH o on Orai1/ STIM1, internal acidic pH inhibits and alkaline pH potentiates Orai1/STIM1 channel activities. Molecular mechanisms of internal pH sensitivity of Orai1/STIM1. To understand the mechanism by which Orai1/STIM1 channels are regulated by pH i changes, we made a series of mutations at the N-and C-termini, and at the loop between TM2 and TM3 (Fig. 6A). The mutations were made on the titratable residues including histidine (His) and glutamic acid (Glu), as well as cysteine (Cys) residues which have been shown to be involved in internal pH sensing in other channels [30][31][32] . We first tested current amplitude of each mutant at pH i 5.5 and 9.0 in comparison with the current amplitude at pH i 7.4. Representative recordings at pH i 5.5, 7.4, and 9.0 of each mutant are shown in Fig. S4. The normalized current amplitude of each single or double mutant is shown in Fig. 6B. Similar to WT Orai1/STIM1, current amplitude of all the mutants was markedly inhibited at pH i 5.5 and significantly enhanced at pH i 9.0 except for H155F. For H155F, current amplitude at pH i 9.0 was even smaller than that at pH i 7.4, whereas the current amplitude of other mutants was enhanced 1.5-to 2-fold at pH i 9.0, which was similar to the changes in WT Orai1/STIM1. At pH i 5.5, current amplitude of H155F was significantly larger than WT (p < 0.05). These results indicate that H155F is less sensitive to both acidic and basic internal pH.
We further investigated the effects of different pH i on H155F in comparison with WT Orai1/STIM1. Original recordings of WT and H155F at pH i 5.5, 7.4, and 8.4 are shown in Fig. 7A,B, and the average current amplitude is shown in Fig. 7C. The concentration-dependent changes of H155F current amplitude normalized to I pHi 7.4 are shown in Fig. 7D. It is noticeable that H155F lost response to basic pH i , and the response to acidic pH i was also significantly diminished. The pKa obtained by best fit of the dose-response curves constructed using I/I Max was 6.56 ± 0.63 for H155F and 7.47 ± 0.18 for WT Orai1/ STIM1 (Fig. 7E). The dose-response curve of H155F was left-shifted by almost one pH unit (p < 0.01). These results suggest that H155 is responsible for intracellular pH sensitivity of Orai1/STIM1. Endogenous I CRAC channel exhibits similar pH sensitivity to Orai1/STIM1. To investigate whether endogenous I CRAC channels are also sensitive to pH changes, we used RBL cells for current  concentration dependent and reversible (Fig. 8B). The pKa of I CRAC was 8.41 ± 0.10, similar to the pKa of Orai1/STIM1 shown in Fig. 1. Moreover, the I CRAC current in RBL cells was also regulated by internal acidic and alkaline pH. Current amplitude was inhibited by about 80% at pH i 5.5, and potentiated by about 1.5-2 fold at pH i 8.5 (Fig. 8D,E). The pKa (Fig. 8F) of I CRAC was also similar to that of Orai1/ STIM1 shown in Fig. 7D.

Effects of external Ca 2+ on pH sensitivity of Orai1/STIM1 mutants. DVF solution has been
commonly used for investigating pH sensitivity of different ion channels [33][34][35][36][37][38][39] . Although the pH sensitivity of WT Orai1/STIM1 is not dependent on charge carrier (Fig. 2), we tested pH sensitivity of the Orai1 mutants when Ca 2+ is the permeant cation for Orai1/STIM1 channels, since changing the relative permeability and/or selectivity may serve as a mechanism by which protons regulate Ca 2+ -permeable channels 35,40,41 including I CRAC 26,27 . We used Ca 2+ concentration (20 mM) higher than physiological Ca 2+ (2mM) in order to get larger currents. Similarly high Ca 2+ concentration for I CRAC recording has been used previously 26,27 . As shown in Fig. 9, in the presence of 20 mM external Ca 2+ , the mutants D110N and D112/114N displayed similar pH sensitivity to that of WT Orai1/STIM1, indicating that the residues D110 and D112/114 do not contribute to pH sensitivity of Orai1/STIM1, which is consistent with the results obtained in DVF solution (Fig. 4). However, the mutant E106D was insensitive to pH changes in 20 mM Ca 2+ Tyrode's solution, a feature which presumably manifests the reduced Ca 2+ permeability and selectivity as previously reported 26 . Similarly, the mutant E190Q, which largely loses Ca 2+ selectivity 20 , conducted very minimal currents and was insensitive to low pH (Fig. 9E), albeit a large inward current was elicited at pH 9.0 with an unknown mechanism which is worthy of further investigation in the future. Interestingly, the mutant E190D has similar pH sensitivity as the WT Orai1/STIM1 in 20 mM Ca 2+ solution (Fig. 9F). This is not surprising given that E190D preserves similar Ca 2+ selectivity to that of WT Orai1/STIM1 channels 20 .
Using normal Tyrode's solution containing 2 mM Ca 2+ as the external solution, we also tested whether external Ca 2+ influences pH sensitivity of the internal pH sensor of Orai1/STIM1, H155F. We found that the pH sensitivity of H155F was not altered by external Ca 2+ concentration. For example, the ratios of current amplitude at pH i 5.5 and pH i 8.4 versus pH i 7.4 were 0.34 ± 0.07 (n = 6) and 1.1 ± 0.3 (n = 6) respectively in 2 mM Ca 2+ Tyrode's solution, similar to the ratios of 0.36 ± 0.05 and 1.0 ± 0.24 (n = 6) obtained in DVF solution as shown in Fig. 7.

Discussion
In this study, we demonstrate that Orai1/STIM1 channels are regulated by both internal and external pH. Whereas acidic internal and external pH inhibit channel activity, alkaline intra-and extracellular pH dramatically increase channel activity. We identify a new residue H155 which is responsible for intracellular acidic and alkaline pH sensitivity. For extracellular pH sensitivity, we find that even though pH regulation on Orai1/STIM1 channels is independent of the charge carrier, the mechanisms underlying pH sensitivity are different with different charge carriers. While the residue E106 is responsible for the pH sensitivity when Ca 2+ is the charge carrier, we find that E190 is responsible for pH sensitivity when Na + is the charge carrier. Furthermore, we show that endogenous I CRAC in RBL cells is also inhibited by acidic pH i and pH o , and potentiated by basic pH i and pH o . As I CRAC plays an important role in various physiological functions, and given that acidosis and alkalinization are involved in a variety of

Orai1/STIM1 channel activity is potentiated by intracellular and extracellular alkalinization, and inhibited by internal and external acidosis. Regulation of the endogenous I CRAC by acidic pH
in Jurkat T-lymphocytes and by external alkaline pH in macrophages has been previously reported 14,16 . Moreover, two recent studies have shown that heterologously expressed Orai1/STIM1 currents can be inhibited by external low pH 26,27 and internal low pH 27 , and enhanced by external high pH but not by internal high pH 27 . We find that Orai1/STIM1 channels expressed in HEK-293 cells are not only inhibited by intracellular and extracellular acidic pH, but also enhanced by both intracellular and extracellular alkaline pH. The reason for the different regulation by internal high pH obtained in this study and in Beck's study is currently unknown, and will need further investigation. However, previous studies demonstrated that in Jurkat T-lymphocytes and neutrophils, cytosolic alkalinization induces Ca 2+ release and store-operated Ca 2+ entry 13,15 , consistent with the notion that alkaline internal pH potentiates Orai1/ STIM1. Similar to the regulation of over-expressed I CRAC currents, the native I CRAC is inhibited by external low pH and enhanced by high pH, with the pKa of 8.4 (Fig. 8). We show that pH regulates Orai1/STIM1 channel activities independent of charge carriers. For example, acidic pH o inhibits I CRAC currents carried by Ca 2+ ions to the same degree as it inhibits the I CRAC currents carried by Na + ions. Likewise, basic pH o enhances I CRAC currents carried by both Ca 2+ ions and Na + ions (Fig. 2). Similar results have been reported for the regulation of native I CRAC currents by pH 16 . Kerschbaum and colleagues demonstrated that I CRAC currents in Jurkat T lymphocytes are inhibited by acidic extracellular and intracellular pH, and that both Ca 2+ currents and monovalent Na + currents of I CRAC can be equally inhibited by acidic pH in a voltage independent manner 16 . The authors showed that acidic intracellular pH blocks I CRAC with a pKa of 6.8, whereas acidic extracellular pH inhibits I CRAC with a pKa of 8.2 16 . Alkaline external pH was also previously shown to enhance I CRAC currents in macrophages with a pK a of 8.2 14 . Our pKa for the external proton effects on Orai1/STIM1 is similar to that previously reported 14,16 , but our pKa (7.46) for internal pH effects is higher than that obtained by Kerschbaum and colleagues 16 . We do not yet know the reason for the different results, but it is conceivable that this discrepancy could be due to the fact that we used heterologously expressed Orai1/STIM1 channels, whereas the native I CRAC may be contributed by different isoforms of Orai and STIM 42 . Nonetheless, regulation of endogenous I CRAC and heterologously   expressed Orai/STIM currents by pH indicates that I CRAC may be an important mediator of altered Ca 2+ signaling under acidic and alkaline conditions. Mechanisms of external pH sensitivity of Orai1/STIM1 channels. By mutating a series of titratable amino acid residues located in the channel pore, in the loop between TM1 and TM2, and within TM3 43,44 , we find that the amino acid residues which are responsible for extracellular pH sensitivity are different with different permeant cations. Whereas E106D loses pH sensitivity when Ca 2+ is the charge carrier, which is consistent with the previous report 26 , E106D shows similar pH sensitivity to that of WT Orai1/STIM1 in the absence of Ca 2+ (in DVF solution) when Na + is the charge carrier. E190D, however, exhibits markedly reduced pH sensitivity in DVF solution when Na + is the permeant cation, albeit displaying similar pH sensitivity to that of WT Orai1/STIM1 in the Tyrode's solution containing Ca 2+ .
Thus, it appears that E190 is responsible for external pH sensitivity in the absence of Ca 2+ when Na + is the charge carrier, whereas E106 is responsible for external pH sensitivity when Ca 2+ is the charge carrier. Although using monovalent Na + cation in DVF solution as the charge carrier is a non-physiological condition, Na + in DVF solution has been commonly used as the permeant cations for investigating pH sensitivity of different ion channels including Ca 2+ channels and Ca 2+ -permeable channels [33][34][35][36][37][38][39] . Thus, our results provide important insights suggesting that the molecular basis for pH sensitivity can be different when using Ca 2+ as the charge carrier or the monovalent cation Na + as the charge carrier, and that precautions need to be taken when extrapolating the mechanism of pH sensitivity obtained from experimental conditions to physiological conditions. External protons can regulate ion channel functions via changing gating properties and/or influencing channel permeation [33][34][35][36]40 . Many Ca 2+ permeable channels are regulated by protons via changing Ca 2+ selectivity and permeability. For the voltage-gated Ca 2+ channels (VGCC), the Ca 2+ selectivity and pH sensitivity are conferred by the Glu residues in the pore-forming region 35,36 . Mutation of Glu by glutamine (Gln) substitution in repeats I or III produces similar low conductance single channel currents, mimicking the protonated state 35,36 . Thus, the permeant divalent cations and protons compete for the same Glu binding sites 35 . The similar competing mechanism also underlies pH modulation on TRPM6 and TRPM7 channels 41,45,46 . Yet, for TRPM6 and TRPM7 channels, binding of protons to the E1024 (TRPM6) and E1047 (TRPM7) in the channel pore releases Ca 2+ block on monovalent currents. Therefore, low pH o potentiates TRPM6 and TRPM7 currents by enhancing the monovalent current amplitude 41 . Similar to VGCCs, the Ca 2+ selective TRPV5 channels are inhibited by acidic pH o 34 . However, the pH o sensing site E522 is independent of the Ca 2+ selectivity site D648 34 . Protons modulate TRPV1 channel activity via influencing both gating and permeation properties [37][38][39][40] . Activation of TRPV1 by protons is mediated by the extracellular E600 and E648 residues 37 , whereas the divalent permeability of TRPV1 is conferred by D646 47 . TRPM2 has also been shown to be inhibited by external protons [48][49][50][51] , yet the mechanism is more complicated [48][49][50] . Although one study suggested that external protons permeate through TRPM2 and inhibit channel activity intracellularly 49 , mutagenesis results in other studies demonstrated that external protons bind to external residues around the channel pore and block the channel activity extracellularly 48,50,51 . Whereas neutralization of H958, D964, and E994 at the outer vestibule of the channel pore enhances pH sensitivity of TRPM2 by reducing external Ca 2+ sensitivity 48 , Yang and colleagues demonstrated that external protons inhibit TRPM2 in a voltage-and state-dependent manner 50,51 , and that substitution of the residues K952 and D1002 by alanine (Ala) significantly reduces inhibition of open TRPM2 channels by external protons 50 . Furthermore, the residue Q992 in the outer pore of mouse TRPM2 (mTRPM2) is crucial for the reduced sensitivity to pH o 6.0 in comparison with human TRPM2 (hTRPM2) 51 . Nonetheless, it appears that the pH sensing residues are usually located either in the channel pore, or at the vestibule of the channel pore for many Ca 2+ -selective and Ca 2+ -permeable channels 39 . For Orai1, however, we found that neutralizing the negatively charged residues D110, D112 and D114 at the vestibule did not significantly change pH o sensitivity regardless of permeant cations (Figs 4 and 9). At very high pH o (pH o 9.0, 9.4, and 10.0), D110N and D112/114N exhibited larger current than WT but without statistical significance (Fig. 4) when Na + is the charge carrier. Interestingly, when Ca 2+ is the permeant cation, substitution of the Ca 2+ selective filter residue E106 by Asp (E106D) eliminates pH o sensitivity, presumably through changes of Ca 2+ permeability and selectivity, a similar mechanism by which VGCCs and some other Ca-permeable channels are regulated by protons, as previously reported 26 . However, when Na + is the permeant cation, E106D did not produce any change in pH sensitivity, and the pH dose-response curve of E106D was well superimposed with that of WT channels. Unexpectedly, we found that when Na + is the permeant cation, mutant E190D significantly altered the pH sensitivity of Orai1/STIM1, and shifted the pH dose-response curve by almost two pH units.
How might E190 influence pH sensitivity of Orai1/STIM1 channels when Na + is the charge carrier? The E190 residue in the TM3 is not facing the channel pore or involved in the pore formation 43,52 , even though E190, W176 and G183 in the TM3 have been shown to be involved in Ca 2+ selectivity, channel gating and fast inactivation 53,54 . Since E190 is not facing the channel pore, the mechanisms by which E190 affects both Ca 2+ selectivity and/or the size of channel pore are not understood from a structural point of view 55 . Amcheslavsky and colleagues recently demonstrated that E165 in Orai3, the residue equivalent to E190 in Orai1, is involved in 2-APB induced gating of Orai3 channels 55 . The authors proposed that E165 of TM3 is directly behind and in between two TM1 domains, and when 2-APB activates Orai3, E165 moves toward the central axis of the channel, therefore affecting channel permeation and pore formation 55 . Similarly, it is conceivable that E190 in Orai1 becomes accessible when the channel opens. Since E190Q did not significantly change the pH o sensitivity, whereas E190D exhibited markedly reduced sensitivity to both acidic and basic pH o resulting in a significant shift of the pH o dose-response curve by almost two pH units, it is likely that the side chain but not the charge of E190 plays an essential role in pH o sensitivity. Thus, a plausible model is that when Orai1/STIM1 channel opens, E190 becomes accessible to the channel pore. Changing the length of the side chain of E190 may alter the size of the ion-permeation pathway, thereby influencing ion permeation when Na + is the permeant cation, resulting in changed pH sensitivity. Nonetheless, further investigation such as mutating E190 to different residues, measuring single channel conductance by noise analysis, and evaluating changes of permeability will provide more evidence in order to fully understand the underlying mechanisms.
In agreement with our result that mutant E106D eliminates pH sensitivity at high and low pH (Fig. 9) when Ca 2+ is the permeant cation, Scrimgeour and colleagues showed that E106D changes Ca 2+ selectivity as well as fast Ca 2+ -dependent inactivation 44,45 , and eliminates acidic pH inhibition of Orai1/STIM1 26 . The authors demonstrated that acidic pH inhibits Orai1/STIM1 with a pKa of 7.8 26 . Similar to our results, the authors described that Na + conductance of the WT Orai1/STIM1 channels in the presence and absence of external Ca 2+ displayed the same pH dependence 26 , though whether E106 is responsible for the pH sensitivity in the absence of Ca 2+ was not investigated 26 . Since E106D alters Ca 2+ selectivity and Ca 2+ affinity (25 μ M for WT and 490 μ M for E106D) 53 , we believe that the changes in pH sensitivity in E106D is caused by altered Ca 2+ selectivity and permeability as previously demonstrated 26 . Indeed, when we used Na + as the charge carrier, E106D is no longer sensitive to high and low pH o . Instead, we found that in the absence of Ca 2+ , mutant E190D markedly changed pH o sensitivity in comparison with the WT Orai1/STIM1 channels. Thus, our results indicate that E190 is the pH o sensor of Orai1/STIM1 when Na + is the permeant cation.
Although we found that the mutants D110N and D112/114N did not significantly alter pH o sensitivity, a recent study by Beck and colleagues demonstrated that D110/D112A shows reduced sensitivity to external pH 8.4 and pH 6.0 27 . As we neutralized the aspartic acid (Asp) at D110 and D112 to asparagine (Asn), whereas Beck and colleagues mutated Asp residues to Ala residues, it is plausible that the size of the residue at D110 and D112 plays a role in sensing external pH. Alternatively, mutating D110 and D112 simultaneously (D110/D112A) may be crucial for altering pH o sensitivity, which may explain the discrepancy between our and their results. However, since Orai2/STIM1 and Orai3/STIM1 display similar pH o sensitivity to that of Orai1/STIM1 (Fig. 3), and the negatively charged residue D112 is not conserved in all the three isoforms of Orai (Fig. S4), it would be interesting to test whether D110/112A also influences pH o sensitivity of Orai2/STIM1.
Taken together, we found that different amino acid residues are responsible for pH o sensitivity of Orai1/STIM1 under different ionic conditions. Under physiological conditions when Ca 2+ is the permeant cation, it appears that E106 26 and likely D110/D112 27 are responsible for pH o sensitivity, whereas E190 contributes to pH sensitivity when Na + is the charge carrier. Although changes of Ca 2+ and Na + selectivity and permeability are proposed as the underlying mechanism for Orai1/STIM1 regulation by external protons, we and others 26,27 have not provided direct experimental evidence yet. Indeed, it seems that the outward currents are much larger and the reversal potential is not as positive as expected at high pH o (Figs 3 and 4), suggesting that alkaline pH changes ionic selectivity and permeability. Future studies focusing on how different pH o can change Ca 2+ selectivity and permeability of WT Orai1 and its mutants, how single channel conductance of the WT and mutated channels is influenced by various pH o , and whether mutating E106 and E190 to different amino acid residues will alter the pH o sensitivity are required to fully understand the molecular mechanisms underlying pH o sensitivity of Orai1/STIM1.

Mechanisms of intracellular pH sensitivity of Orai1/STIM1 channels. Previous studies showed
that native I CRAC can be blocked by internal acidic pH 16 , but the effect of basic pH was not evaluated. We demonstrate that Orai1/STIM1 channel activity is not only inhibited by acidic pH i , but also enhanced by alkaline pH i . To further understand mechanisms of internal pH regulation of Orai1/STIM1, we generated a series of mutations on the Cys, His and Glu residues located in the N-and C-termini, as well as the intracellular loop between TM2 and TM3. Among the twelve mutants, H155F displayed markedly reduced sensitivity to both acidic and alkaline pH i , indicating that H155 at the intracellular loop of TM2 and TM3 is responsible for sensing internal pH changes.
It is remarkable that neutralizing one residue H155 diminishes the sensitivity of Orail1/STIM1 to both acidic and alkaline pH i . Since the side chains of His and phenylalanine (Phe) are similar, it appears that the charge of H155 plays an essential role in sensing the internal pH changes. Although the exact mechanism is not clear yet, it is plausible that protonation of H155 under acidic pH i conditions affects intra-or intermolecular interactions of the channel leading to conformational changes that favor channel closing, whereas deprotonation of H155 under alkaline pH i conditions causes conformational changes that favor channel opening. In a previous study, a His residue in the N-terminus has been reported to be responsible for the alkaline pH i induced activation of TRPV1 30 . TRPA1 is also activated by alkaline pH i 32 . However, unlike TRPV1, the mechanism by which alkaline pH i activates TRPA1 is through the modulation of two N-terminal Cys residues 32 . Internal pH regulation of HCN2 channel is also mediated by a His residue located in the cytosolic S4-S5 linker 31 . Protonation of H321 causes a leftward shift of the activation curve therefore reducing HCN2 current amplitude, whereas deprotonation of H321 elicits a rightward shift of the activation curve therefore enhancing HCN2 current amplitude 31 . Several other channels are also regulated by internal pH changes. For example, TRPM2 and TRPM7 are inhibited by acidic pH i 48,56 . An intracellular residue D933 located at the TM4 and TM5 linker was found to serve as the internal pH sensor of TRPM2 48 . The residue D933 contributes to TRPM2 pH i sensitivity by influencing TRPM2 channel gating as well as Ca 2+ and ADPR sensitivity 48 . Different from the mechanism by which TRPM2 is regulated by pH i , intracellular protons inhibit TRPM7 channel activity by screening the negative charges of PIP 2 56 . How might H155 located in the loop of TM2 and TM3 sense the changes of internal pH and translate the pH sensitivity to channel gating? The TM2-TM3 loop has previously been shown to play an essential role in fast inactivation of Orai1/STIM1 57 . Mutation of the four residues in the middle of the loop abolished fast inactivation of Orai1, and addition of a 37-amino acid peptide derived from the loop blocked Orai1 currents 57 . It was proposed that the intracellular loop between TM2 and TM3 of Orai1 may function as an inactivation particle which mediates fast inactivation of Orai1/STIM1 57 . Therefore, it is conceivable that protonation and deprotonation of H155 cause conformational changes of the loop between TM2 and TM3, leading to changes of Orai1/STIM1 channel activity.
Since I CRAC activation requires coupling of Orai and STIM, as well as gating of Orai1, it was previous demonstrated that acidic pH under hypoxia conditions causes uncoupling of STIM1 and Orai1 and thereby reduces current amplitude 25 . Moreover, intracellular alkalinization has been shown to inhibit Ca 2+ -ATPase (SERCA), leading to store depletion and Orai1/STIM1 activation 58 . Using high EGTA buffering condition to passively delete the store, we demonstrate that changes in internal pH directly influence pore-forming subunit Orai1 through protonation and deprotonation of H155 in the loop of TM2 and TM3, a previously unknown mechanism by which Orai1/STIM1 channel activity is regulated by pH i .

Conclusions
We demonstrate that the Orai1/STIM1 channel is regulated by changes of both intracellular and extracellular pH. Acidic internal and external pH reduce Orai1/STIM1 channel activity, whereas alkaline intracellular and extracellular pH enhance Orai1/STIM1 channel activity. Whereas E106 is responsible for external pH sensitivity when Ca 2+ is the charge carrier as previously reported, we find that the residue E190 in TM3 is the major extracellular pH sensor when Na + is the permeant cation. Moreover, we demonstrate that H155 in the intracellular TM2-TM3 loop is the intracellular pH sensor of Orai1/STIM1 channels. Similar to the pH sensitivity of over-expressed Orai1/STIM1 channels, the endogenous I CRAC is also regulated by changes of internal and external pH. Given the important roles of intracellular and extracellular pH in a variety of cellular functions, our results suggest that Orai1/STIM1 channels could be an essential mediator for pH induced modulation of physiological/pathological functions.
Cell culture and functional expression of Orai1/STIM1 and the mutants. HEK-293 cells were grown in DMEM/F12 medium supplemented with 10% FBS, 100 U/ml penicillin and 100 mg/ml streptomycin at 37 °C in a humidity-controlled incubator with 5% CO 2 . Cells were transiently transfected with wild-type (WT) Orai1/STIM1 or its mutants by Lipofectamine 2000 (Invitrogen). We used 5 μ l Lipofectamine 2000 for trasfection of the cells in a 35 mm culture dish. The GFP-containing pEGFP vector was transfected to the HEK293 cells as mock controls. Successfully transfected cells can be identified by their green fluorescence when illuminated at 480 nm. Electrophysiological recordings were conducted between 24 to 36 h after transfection. All patch-clamp experiments were performed at room temperature (20-25 °C).
RBL-2H3 cells were provided by Dr. D. Clapham (Harvard Medical School, Boston, MA). Cells were cultured in DMEM supplemented with 10% FBS and 100 U/ml penicillin and 100 mg/ml streptomycin. For electrophysiological experiments, cells were plated onto glass coverslips and used 12 h thereafter 59 . Electrophysiology. Whole-cell currents were recorded using an Axopatch 200B amplifier. Data were digitized at 10 or 20 kHz, and digitally filtered off-line at 1 kHz. Patch electrodes were pulled from borosilicate glass and fire-polished to a resistance of ~3 MΩ when filled with internal solutions. Series resistance (R s ) was compensated up to 90% to reduce series resistance errors to < 5 mV. Cells with R s > 10 MΩ were discarded 60 .
For whole cell current recording, voltage stimuli lasting 250 ms were delivered at 1 s intervals with voltage ramps or voltage steps ranging from − 120 to + 100 mV. A holding potential of 0 mV was used for most experiments, unless otherwise stated. A fast perfusion system was used to exchange extracellular solutions, with a complete solution exchange achieved in about 1 to 3 s 45 . Original traces without leak subtraction were used in most figures unless otherwise stated. For the recording traces with leak