An open pore structure of the Orai channel, finally

Abstract

Store-operated Orai channels are a primary mechanism for mobilizing Ca²⁺ signals in both non-excitable cells and excitable cells. The structure of the open channel, vital for understanding the mechanism of channel opening, is incompletely understood. We highlight a new study that unveils the structure of a constitutively active Orai mutant and takes us closer towards understanding the molecular basis of Orai channel activation.

Store-operated Orai calcium channels are found throughout the human body and mediate a wide range of essential functions from gene expression to synaptic plasticity. Classically activated by depletion of endoplasmic reticulum (ER) Ca²⁺ stores, Orai channels, including the prototypic isoform Orai1, are gated through direct interactions with the ER-resident Ca²⁺ sensing STIM proteins (STIM1 and STIM2). From clinical studies, we know that patients bearing loss-of-function mutations in the genes encoding Orai or STIM proteins suffer from a devastating immunodeficiency coupled with ectodermal dysplasia and muscle weakness [1]. Other gain-of-function Orai1 mutations are linked to tubular aggregate myopathy and thrombocytopenia, underscoring the vital importance of these channels for human health [1]. Efforts in the last decade have yielded considerable advances to understand the molecular basis of channel activation, including identification of the domains that comprise the channel pore, the gate, and determination of the closed channel structure by X-ray crystallography [2]. However, a high-resolution structure of the open channel, crucial to understanding the molecular motions that underlie channel gating, has remained elusive until now.

The latest cryo-electron microscopy structure by Hou et al. [3] of the Drosophila melanogaster Orai channel in an open configuration (Fig. 1) advances this quest and takes us closer to an appreciation of the structural basis of channel opening. Although Orai1 channels are physiologically gated by the STIM proteins, numerous Orai1 mutants that are constitutively active have been discovered in the last few years. Among these, mutations at the H134 locus in transmembrane domain 2 (TM2) facing the non-pore-lining surface of transmembrane domain 1 (TM1) approximately halfway into the membrane, are very well-characterized and produce Orai1 channels with high Ca²⁺ selectivity and other pore properties similar to those found in wildtype channels gated by STIM1 [4,5]. Early efforts to determine the H206A dOrai (equivalent to human Orai1 H134A) channel structure using cryo-EM were stymied by low resolutions (>7 Å) of the resulting structures [6]. However, the present study overcomes this problem with the clever use of an antibody Fab fragment bound to the TM1-TM2 extracellular loop to help stabilize the transmembrane helices, thereby improving the overall atomic-level resolution to 3.3 Å. Satisfyingly, the open channel structure of the pore is in very good agreement with previous functional studies. All of the amino acids found to contribute to the pore (hOrai1 R91, L95, G98, F99, V102, and E106) from cysteine accessibility experiments [7–9] were also observed to do so in the dOrai structures (Fig. 1). Importantly, the improved structure also reveals several notable aspects of the open pore conformation and significantly advances our understanding of the gating mechanism of the Orai channel.

Fig. 1.

Conformational changes underlying opening of the Orai pore. (A) A side-view of the X-ray crystal structure of dOrai (PDB ID: 4HKR). Pore-lining residues identified from structural and functional studies are labelled. (B) Side view of the cryo-EM structure of Orai_H206A (PDB ID: 7KR5). In this open state, the pore is significantly dilated due to rigid body movements of TM1 helices away from the central symmetry axis. Residues Q180, D182, and D184 in the TM1-TM2 loop that form the turret are labelled in brown. The cytoplasmic extensions of the TM1 helices are not resolved in this structure.

Starting from the extracellular region of the pore, the Fab-bound H206A structure showed that the TM1-TM2 loop, which was previously unresolved, forms a highly ordered structure resembling an inverted teepee (Fig. 1B), analogous to the extracellular turret domains found in K⁺ channels [10]. The wide opening of the turret is consistent with previous electrophysiological data indicating that large MTS reagents like MTS-TEAE (headgroup size ~8 Å) can bind to and reversibly block exogenous Cys residues introduced into this domain [7]. Its well-resolved density led Hou et al. to postulate that the turret is rigid. Yet, this conclusion may need to be tempered with previous findings showing that introduced Cys and native Asp residues in the TM1-TM2 loop can tightly bind small Cd²⁺ and La³⁺ ions, both of which require coordination with multiple (3 or more) side-chains [7,11]. Because tight binding with divalent and trivalent ions would require significant flexibility of the loop, it seems likely that the TM1-TM2 turret may retain a fair degree of flexibility to accommodate both large and small ions in the absence of the antibody Fab fragment.

Moving further into the pore, the structure reveals a fascinating nuance of the glutamate selectivity filter that confers high Ca²⁺ selectivity to Orai channels. The study notes that the glutamates side-chains may exhibit a “broken” symmetry in the new structure, with some carboxylates in the up, and others in the down position, analogous to the model previous described for voltage-gated activated Ca²⁺ channels [12] and as postulated for Orai3 [13]. Functional studies imply that the Orai pore can accommodate multiple closely-spaced Ca²⁺ ions with distinct binding affinities at physiological extracellular Ca²⁺ concentrations [13], but the structural basis of these sites is unresolved. Clustering of the glutamate side-chains into two groups could explain the presence of the multiple Ca²⁺ binding sites. Moreover, strong electrostatic repulsion between closely-spaced Ca²⁺ ions would reduce the apparent Ca²⁺ binding affinity of the selectivity filter by orders of magnitude and ensure Ca²⁺ permeation at physiological concentrations of extracellular Ca²⁺. Further tests of this idea using concatemers to titrate the number of the carboxylates in the hexameric channel should help reveal the distinct roles of each of the six glutamate residues.

In comparison to the closed pore of the wildtype dOrai channel [14], the open H206A pore shows significant dilation, with TM1 helices moving outwards away from the central symmetry axis by 1–2 Å (Fig. 1B). The dilation is particularly striking and relevant in the hydrophobic region of the pore containing critical residues F171 and V174 (hOrai1 F99 and V102), which is wider by ~2 Å (Figs. 1 and 2). Previous functional studies and molecular dynamics (MD) simulations have shown that V102 and F99 form the primary gate of the ion channel, imposing a strong energetic penalty for water and ion occupancy in the closed channel pore [9]. Ion permeation is triggered by displacement of F99 away from the ion conduction pathway, resulting in lowering the free energy barrier for water occupancy in the pore to permit ion conduction. Thus, the widening of the V102-F99 hydrophobic stretch in the new open structure reaffirms the key role for these residues in controlling pore opening.

Fig. 2.

Conformational changes at the hydrophobic gate of the Orai channel. Cross-sectional views of the dOrai pore as viewed from the extracellular side of the closed ((PDB ID: 4HKR) and open Orai_H206A (PDB ID: 7KR5) structures. The aromatic residue, F171, that forms the ion channel gate, is in a pore-facing orientation in the closed state (A) but is displaced away from the central pore axis in the open state (B).

One aspect in which the new H206A structure appears to diverge from experimental studies is in the precise nature of the conformational change underlying opening of the hydrophobic gate. Previous functional work supported by MD simulations indicated that a slight rotation (15–20°) of the hydrophobic TM1 domain together with modest pore dilation helps move the primary channel gate formed by F99 away from the pore axis [5,9]. However, significant helical rotation was not observed in the new H206A structure. Rather, only outward dilation mediated by rigid body movements of the TM1 helices was seen [3] (Fig. 2). In the electrophysiological experiments, rotation of the TM1 helix was inferred based on differential accessibility of thiol reagents (Cd²⁺) to cysteines introduced at G98 and F99 in open versus closed channels [9]. This was also supported by MD simulations showing that TM1 undergoes rotation/twisting and pore dilation in H206S/C open channels [5]. Although unanticipated effects arising from the introduced Cys side-chains at G98 and F99 cannot be ruled out, it seems difficult to explain the differential binding to G98 and F99 by pore dilation alone, which would be expected to decrease Cd²⁺ coordination at both positions. One possibility is that the extensive contacts between the antibody Fab domain bound to the TM1-TM2 turret and the extracellular side of TM1 may constrain spontaneous TM1 rotation, but this concern is alleviated by their calcium imaging showing that acute administration of the Fab does not block store-operated calcium entry [3]. It is also possible that modest rotation of the helix is not detected at the 3.3 Å resolution of the H206A structure. As the authors note, regardless of the precise details of the pore widening process, whether by pore dilation coupled with TM1 rotation, or dilation alone, the structure provides strong structural evidence for the involvement of the hydrophobic domain in pore opening and F99, in particular, as the primary channel gate.

Two natural questions arising from the study relate to the basis of the Orai channel’s dynamic ion selectivity, and the orientation of the cytoplasmic TM4 extension helices when the channel is open. From functional studies, we know that the ion selectivity and the unitary conductance of Orai1 are not fixed but plastic, and strongly modulated by STIM1 [15,16]. The structural basis of how STIM1 induces these changes is a fascinating question. The availability of both closed and open pore structures now offers an opportunity to mechanistically understand how the energetics of ion binding, permeation, and selectivity are regulated by channel opening. Second, the crystal structure of the closed Orai structure shows that the TM4 C-terminal extensions engage in anti-parallel coiled-coil interactions [14]. Although these regions are unresolved in the present study, a previous lower resolution structure of H206A Orai from Hou et al. indicated that the TM4 extension helices straighten to become continuous with TM4 [6]. Because the TM4 extensions constitute the primary STIM1 binding sites, a higher resolution structure showing the orientation of the amino acids in the domain and how they change in the active channel is of major interest. Resolution of this structure, ideally in complex with STIM1, represents the next major challenge. Based on innovative use of Fabs and remarkable successes of the structural approaches used by Hou et al., there is good reason to believe that these structures may also be soon unveiled to usher in a new era for the field.