Molecular mechanism of the Orai channel activation

The Orai channel is characterized by voltage independence, low conductance and high Ca2+ selectivity and plays an important role in Ca2+ influx through the plasma membrane. How the channel is activated and promotes Ca2+ permeation are not well understood. Here, we report the crystal structure and cryo-electron microscopy reconstruction of a Drosophila melanogaster Orai mutant (P288L) channel that is constitutively active according to electrophysiology. The open state of the Orai channel showed a hexameric assembly in which six TM1 helices in the center form the ion-conducting pore, and six TM4 helices in the periphery form extended long helices. Orai channel activation requires conformational transduction from TM4 to TM1 and eventually causes the basic section of TM1 to twist outward. The wider pore on the cytosolic side aggregates anions to increase the potential gradient across the membrane and thus facilitate Ca2+ permeation. The open-state structure of the Orai channel offers insights into channel assembly, channel activation and Ca2+ permeation.


4
Orai dimer, forming an approximately hexameric assembly 17 . However, the hexameric stoichiometry of the Orai channel was brought into question by studies of singlemolecule photobleaching 18 and artificially linked hexameric concatemers of human Orai1 19 . Interestingly, recent studies of the concatenated Orai1 channel indicated that the Orai1 channel functions as a hexamer [20][21][22] . Therefore, hexameric stoichiometry is generally accepted as one of the major conformations of the Orai channel 5,23,24 .
The current crystal structure of Drosophila melanogaster hexameric Orai represents an inactive conformation 17 . The innermost transmembrane 1 (TM1) from each subunit forms a closed ion pore, and three other transmembrane helices are arranged around TM1. The selectivity filter is presumably formed by a ring of six glutamate residues on the extracellular side of the pore. The mechanism of channel activation and Ca 2+ permeation remains unclear. Here, we determined the structure of the constitutively active Drosophila melanogaster Orai (dOrai) mutant P288L using both X-ray crystallography and cryo-electron microscopy. The open state of the Orai1 structure depicts the mechanism of channel activation and Ca 2+ permeation.

The dOrai-P288L channel is constitutively active
To obtain the three-dimensional structure of constitutively active Orai, a construct of dOrai (hereafter referred to as dOrai-P288L) consisting of amino acids 132-341 with three mutations (C224S, C283T and P288L) was selected and purified ( Supplementary   Fig. 1). The Drosophila melanogaster Orai P288L mutant corresponds to the P245L 5 gain-of-function mutation of human Orai1, which causes overlapping syndromes of tubular myopathy and congenital miosis 12,25 . To verify that the dOrai-P288L channel showed constitutive activity, electrophysiology was carried out. First, we transiently transfected HEK-293T cells with a plasmid encoding the GFP-tagged dOrai-P288L channel and showed that the channel was localized to the cell surface ( Supplementary   Fig. 2a). Next, we used whole-cell patch clamp measurements to record the Ca 2+ current in HEK-293T cells expressing the dOrai-P288L channel, showing the inward rectifying Ca 2+ current with a reversal potential close to + 50 mV (Fig. 1a). Furthermore, in the absence of divalent cations, we observed clear monovalent cation currents (Fig. 1b).
Finally, we used single channel current recording in vitro to verify the activity of the purified dOrai-P288L channel. The monodisperse and high-purity protein ( Supplementary Fig. 2b) was reconstituted into a planar lipid bilayer, as reported by our previous research 26 . No current signal was observed at -100 mV without the addition of the channel protein. After the addition of the channel protein, obvious inward Ca 2+ currents of approximately 2 pA at -100 mV were recorded (Fig. 1d). The single-channel Ca 2+ currents gradually diminished as the stimulation potential depolarized from -100 mV to 0 mV. Additionally, the single-channel Ca 2+ currents of the dOrai-P288L channel could be significantly inhibited by Gd 3+ (Fig. 1e). Taken together, these results show that the dOrai-P288L channel is fully active and recapitulates the properties of the STIM-activated Orai channel. 6

Overall structure of the open channel
The crystal structure of the dOrai-P288L channel was determined at the resolution of 4.5 Å by molecular replacement using the closed Orai structure (RCSB code: 4HKR) as a model. The overall architecture of the open channel shows a hexameric assembly (Figs. 2a, 2b). Each protomer consists of four transmembrane helices (TM1-TM4, Fig.   2c). Six protomers adopt a six-fold noncrystallographically symmetric arrangement, resulting in the six TM1 helices forming the ion-conducting pore in the center. The TM2 and TM3 helices from each protomer surround the six TM1 helices, forming a fence to fix the TM1 helix positions. Remarkably, each TM4 helix forms an extra-long helix, pointing into the distance on the cytosolic side. There are two hexamers in one asymmetric unit of the dOrai-P288L crystal, bound together through coiled coil interactions between the TM4 helices ( Supplementary Fig. 3).
To further confirm the open-state conformation of the dOrai-P288L channel, we took advantage of cryo-electron microscopy (cryo-EM) methodology. Negative-stain electron microscopy was used to screen different conditions including amphipols (A8-35), detergents (DDM), and reconstitution into nanodiscs. The negatively stained dOrai-P288L channels in the nanodiscs appeared monodisperse and were further subjected to cryo analysis. The samples showed orientation bias with more than 95% oriented to the same top or bottom view. We overcame this problem by using the detergent glyco-diosgenin (GDN). Because the dOrai-P288L channel was wrapped in a thick layer of detergents, the final structure was determined at the overall resolution of 5.7 Å by single-particle cryo-EM (Supplementary Figs. 4,5). Each helix is well 7 resolved in the final density (Fig. 2d). The overall cryo-EM structure of the dOrai-P288L channel is similar to its crystal structure counterpart. The crystallographic dimer across the central pore fits well into the cross-section of the cryo-EM density (Fig. 2e).

Activation of the Orai channel
Structural comparison of the open and closed Orai structures revealed that the TM1-TM3 regions have similar architecture, while the TM4 helices are completely different (Fig. 3a). From the top view, all six TM4 helices are fully extended, and clockwise rotation occurred during opening, causing the N-terminal regions of the innermost six TM1 helices to twist outward in a counterclockwise direction. These conformational changes are also observed in the cryo-EM density of the dOrai-P288L channel (Fig. 3b). In contrast, the closed conformation of the Orai channel does not fit into our cryo-EM density ( Supplementary Fig. 6). Surprisingly, the twisted region starts from the positively charged residue K159, which corresponds to the residue K87 in human Orai1, and proceeds to the N-terminus of the TM1 helix ( Fig. 3c and Supplementary Fig. 7). We were not able to distinguish whether the remaining region of the TM1 helix undergoes a rotation operation 23 during opening in either our crystallographic structure or the cryo-EM density.
Within each protomer, we observed a conformational transduction pathway from the peripheral TM4 helix through the middle TM3 helix to the basic section of the innermost TM1 helix (Fig. 3c). To confirm that such conformational transduction occurs during opening, we made two mutants (Orai1-F257A and Orai1-L261A) of wild-8 type human Orai1. Two residues, F257 and L261, in human Orai1 correspond to F300 and L304 in dOrai, respectively, which form hydrophobic interactions between the TM3 helix and TM4 helix in the closed conformation of dOrai (Fig. 3d). Upon opening, the drastic swing of the TM4b portion presumably causes the movement of the TM3 helix ( Fig. 3c). As we expected, whole-cell patch clamp measurement showed that the Ca 2+ currents of Orai1-F257A and Orai1-L261A after activation by STIM1 were significantly lower than those of wild-type ( Fig. 3e and Supplementary Fig. 8a). We also used a Ca 2+ influx assay to cross-validate this result. Two mutants (Orai1-F257A and Orai1-L261A) completely lost the extracellular Ca 2+ influx after ER was depleted ( Supplementary Fig. 8b). To ensure that the reduced channel activity was indeed caused by the reduction in channel activation, rather than an effect on attenuated STIM1-Orai1 binding, we performed co-immunoprecipitation (co-IP) and intracellular fluorescence resonance energy transfer (FRET) experiments to verify the interaction between the Orai1 mutations and STIM1. When coexpressed with STIM1-myc, wild-type and mutant Orai1-GFP were similarly able to pull down STIM1 after ionomycin treatment ( Fig. 3f). Additionally, when YFP-tagged STIM1 and CFP-tagged wild-type and mutant Orai1 were cotransfected into HEK-293T cells, the measured apparent FRET efficiency (Eapp) values after the cells were treated with thapsigargin were similar ( Supplementary Fig. 8c). These results indicate that after interference with the interaction between the TM3 helix and the TM4 helix, Orai1 cannot be activated by STIM1 but is able to associate normally with STIM1. Furthermore, in the closed conformation of dOrai1, the interaction between the basic section of the TM1 helix and 9 the TM3 helix is mediated by three residues, L153, S154 and K157, which correspond to L81, S82 and K85 in human Orai1, respectively (Supplementary Fig. 9). Earlier reports showed that the Orai1-L81A mutant completely blocked channel function without altering the STIM1-Orai1 association, and further, the triple L81A-S82A-K85E mutant or the double L81A-S82A mutant prevented extracellular Ca 2+ influx in the constitutively active Orai1-ANSGA mutant channel 27 . The single replacement K85E in Orai1 resulted in a complete absence of STIM-dependent current in cells in response to Ca 2+ store depletion 28 . Taken together, these evidences show that the conformational transduction pathway (T4b helix --> T3 helix --> T1 helix basic section) identified in our open-state structure is critical for Orai channel activation.

Mechanism of Ca 2+ permeation
In the closed state of the dOrai structure, the basic section of the TM1 helix was identified and proposed to provide electrostatic repulsion in blocking cation transport 17,29 . During Ca 2+ permeation, this positively charged region must be neutralized or In our open-state Orai structure, the electron density peaks corresponding to anions can be found around the basic section of the TM1 helix in both the crystallographic structure ( Fig. 4a) and the cryo-EM model (Fig. 3b). A similar anion binding site was also 10 reported in the crystal structure of the closed state of dOrai 17 . These results suggest that the binding of these anions to the basic section of the TM1 helix may be necessary for Ca 2+ permeation during channel opening. With this possibility in mind, mutating these basic amino acids will reduce the binding of anions, leading to an attenuated Orai channel and thus reconciling the above functional data. In contrast, introducing more anions on the cytosolic side of the membrane may potentiate Ca 2+ permeation through activated Orai channels.
Indeed, two mutants (Orai1-R83A-K87A and Orai1-R77A-K78A) showed severely attenuated Ca 2+ currents in the STIM1-activated Orai1 channel by whole-cell patch clamp measurement (Figs. 4b, 4c and Supplementary Fig. 10a) and significantly decreased store-depleted extracellular Ca 2+ influx (Fig. 4d). The reduced channel activity was not caused by attenuated STIM1-Orai1 binding, because the Orai1 mutants pulled down STIM1 at a similar level to wild-type Orai after ionomycin treatment in a coimmunoprecipitation experiment ( Supplementary Fig. 10b). These results reinforce the argument that the basic amino acids in this region are important for Ca 2+ permeation in the Orai channel. More importantly, we added a certain amount of aspartate in the pipette solution to record the Ca 2+ currents of the STIM1-activated Orai1 channel. As expected, aspartate concentration-dependent Ca 2+ currents were observed ( Fig. 4e and Supplementary Fig. 10c). These results suggest that the basic section of the TM1 helix may aggregate negative charges to facilitate Ca 2+ permeation during the opening of the Orai channel.

Discussion
In this study, we determined the structure of the constitutively active dOrai- Therefore, we propose a model of Ca 2+ permeation in the Orai channel (Fig. 5). In the closed state of the channel, the latched TM4 helix shrinks the pore on the cytosolic side.
Positive charge repulsion and anion plugs block Ca 2+ permeation. Upon opening, the TM4 helix swing twists the basic section outward to accommodate more anions. These anions not only neutralize the positive charges to reduce charge repulsion but also increase the potential gradient across the membrane, thus facilitating Ca 2+ permeation.
This model is consistent with many published functional studies. Zhou et al identified a "nexus" site (amino acids 261-265) within the Orai1 channel that is proposed to connect the peripheral C-terminal STIM1-binding site to the Orai1 pore helices 27 . The structural comparison of the Orai channel structures between our open state and the published closed state clearly observed the conformational transduction pathway (T4b helix --> T3 helix --> T1 helix basic section), providing further evidence that the "nexus" site is most likely the trigger for channel activation. Furthermore, cholesterol was reported to interact with Orai1 channel and inhibit its activity through residues Orai1 L74 and Orai1 Y80 31 . These two residues are located within the interface between the TM1 helix and the TM3 helix. The cholesterol binding presumably interrupts the conformational transduction pathway, which explains the result that cholesterol did not affect the binding of STIM1 to Orai1 channel but attenuate the Orai1 activation 31 .
How Orai channel conducts Ca 2+ is a puzzling question. The Orai pore consists of an extracellular mouth, the selectivity filter, an unusually long hydrophobic cavity and an intracellular basic region. The Orai closed-state structure revealed that the narrowest region of the pore has a diameter of 6.0 Å, wide enough for a dehydrated Ca 2+ passing through. It seems unnecessary to rotate the pore helix for channel activation as proposed by Yamashita et al 23        Representative I-V curves of the whole-cell Ca 2+ currents of STIM1-activated human wild-type Orai1 at the indicated concentration of sodium aspartate.  supplemented with 10% fetal bovine serum (FBS, PAN) at 37 °C with 5% CO2.
Bacmid was transfected into Spodoptera frugiperda (Sf9) cells using X-tremeGENE 26 HP DNA Transfection Reagent (Roche). All plasmids were transfected into HEK293T cells using Fugene6 (Promega). Purified proteins were added to the external surface of the chip (the trans side).

Whole-cell
Acquired data were analyzed by using Origin 9.0. All experiments were performed at room temperature.      Cryo-EM structure determination and resolution assessment of the dOrai-P288L channel. a, A drift-corrected cryo-EM micrograph of the dOrai-P288L channel. b, Ctffind showed Thon rings in the Fourier spectrum of the image in panel a. c, Selected 2D class averages of the dOrai-P288L channel. d, The "gold-standard" FSC coefficient curve of the final reconstruction showed an overall resolution of 5.7 Å. e, Local resolution estimation by ResMap.  Interactions between the TM1 helix and the TM3 helix in the closed Orai structure. a, The TM1 helix and the TM3 helix are colored green and orange, respectively. The ion-conducting pore side is labeled. b, Zoom view of the specific interactions between two helices. Side chains of three residues (K157, S154 and L153) from the TM1 helix and two residues (E245 and H241) from the TM3 helix are shown. The hydrogen bonds are shown as magenta dashed line. Atoms oxygen and nitrogen are colored red and blue, respectively. Amino acids in parentheses denote human Orai1 counterparts. The atom coordinates were taken from the structure with the RCSB code 4HKR. Side chains of residues K157 and L153 were absent in original PDB file. They were manually built from the program Coot based on frequently used rotamers.