Understanding IP₃R Channels: from structural underpinnings to ligand-dependent conformational landscape

Abstract

Inositol 1,4,5-trisphosphate receptors (IP₃Rs) are ubiquitously expressed large-conductance Ca²⁺-permeable channels predominantly localized to the endoplasmic reticulum (ER) membranes of virtually all eukaryotic cell types. IP₃Rs work as Ca²⁺ signaling hubs through which diverse extracellular stimuli and intracellular inputs are processed and then integrated to result in delivery of Ca²⁺ from the ER lumen to generate cytosolic Ca²⁺ signals with precise temporal and spatial properties. IP₃R-mediated Ca²⁺ signals control a vast repertoire of cellular functions ranging from gene transcription and secretion to the more enigmatic brain activities such as learning and memory. IP₃Rs open and release Ca²⁺ when they bind both IP₃ and Ca²⁺, the primary channel agonists. Despite overwhelming evidence supporting functional interplay between IP₃ and Ca²⁺ in activation and inhibition of IP₃Rs, the mechanistic understanding of how IP₃R channels convey their gating through the interplay of two primary agonists remains one of the major puzzles in the field. The last decade has seen much progress in the use of cryogenic electron microscopy to elucidate the molecular mechanisms of ligand binding, ion permeation, ion selectivity and gating of the IP₃R channels. The results of these studies, summarized in this review, provide a prospective view of what the future holds in structural and functional research of IP₃Rs.

Graphical Abstract

1. Introduction

The resolution revolution in cryogenic electron microscopy (cryo-EM) over the last decade has ushered in an explosion of structural studies of ion channels, broadening perspective and gaining insights into mechanisms of ion transport across cellular membranes. Among them are Ca²⁺ permeable ion channels, which allow calcium ions (Ca²⁺) to flow across the lipid bilayer of cell membranes. Ca²⁺ is the most ubiquitous secondary messenger in cells. Ca²⁺ is an effective second messenger, and it is imperative that cells maintain low levels of cytosolic Ca²⁺ under resting conditions (<0.1 μM) and upon stimulation generate a temporary increase of cytosolic free Ca²⁺ in specific nanodomains. Relatively small changes in cytosolic concentration of Ca²⁺ give rise to Ca²⁺ signals that regulate a variety of cellular processes including fertilization, proliferation, differentiation, muscle contraction, neuronal signaling, and apoptosis. Most Ca²⁺ signals are due to Ca²⁺ permeable ion channels, which allow Ca²⁺ to flow across the lipid bilayer of the plasma membrane (PM) or the membranes of intracellular organelles. Inositol 1,4,5-trisphosphate receptors (IP₃Rs), the most widely expressed Ca²⁺ channels, reside in the membranes of the endoplasmic reticulum (ER) which serves as the main intracellular Ca²⁺ store.

IP₃Rs release Ca²⁺ from the ER lumen to the cytoplasm to generate Ca²⁺ signals in response to the many extracellular stimuli that evoke IP₃ formation. IP₃R-mediated Ca²⁺ signals control a vast repertoire of cellular functions ranging from gene transcription and secretion to the more enigmatic brain activities such as learning and memory. Malfunction of IP₃R signaling is implicated in numerous human diseases such as Alzheimer‟s disease, Huntington‟s disease, ataxias, heart failure, stroke and cancer [1–7]. Moreover, several disease-causing mutations have been identified in all the three IP₃R subtypes [8–12]. Consequently, efforts to understand the structure and function of IP₃R relate directly to human health and disease.

IP₃Rs are ubiquitously expressed large-conductance Ca²⁺-permeable channels, whose opening is linked to essential phospholipase C signaling pathways that mediate the formation of IP₃ [13,14]. Complex cross-talk occurring between these pathways and IP₃R channels leads to precise regulation of intracellular Ca²⁺ levels producing cytosolic Ca²⁺ signals which differ in their spatial and temporal profiles. Versatility of IP₃-mediated Ca²⁺ signals in regulation of a broad range of cellular functions arises from the properties of IP₃Rs and their cellular organization. First, IP₃Rs open and release Ca²⁺ from the ER only when they bind both IP₃ and Ca²⁺, the primary channel agonists. IP₃ binding primes IP₃Rs to respond to Ca²⁺, which then promotes channel opening. Moreover, the gating of IP₃R channels is biphasically regulated by cytosolic Ca²⁺ itself. In the presence of IP₃, small elevations of Ca²⁺ enhance channel openings, whereas a further increase in intracellular Ca²⁺ causes inhibition of IP₃-evoked channel gating. Second, functional IP₃Rs are arranged in a clustered fashion on the ER membrane such that Ca²⁺ release through a single channel can trigger the opening of neighboring channels within a cluster to generate local Ca²⁺ signals, a phenomenon termed Ca²⁺-induced Ca²⁺ Release (CICR) [15–18]. These Ca²⁺ signals propagate hierarchically in the form of Ca²⁺ blips, puffs, waves and oscillations [17,19,20]. Dual regulation by two co-agonists and clustered arrangement are key features that endow IP₃Rs with a capacity to propagate Ca²⁺ signals regeneratively between IP₃Rs as well as to other organelles (notably mitochondria, lysosomes [21,22]) and to the plasma membrane. A long outstanding question was finally resolved when the use of concatenated IP₃Rs showed that IP₃ binding to all four subunits in the tetramer is required for channel opening [23].

Besides IP₃ and Ca²⁺, a precise spatiotemporal regulation of IP₃ R-mediated Ca²⁺ signals is achieved through regulation of IP₃R gating by other small molecules (e.g., ATP, cAMP, NADH), post-translational modifications (e.g., phosphorylation, glycosylation) and protein cofactors (e.g., Ca²⁺ sensor proteins, Bcl-2 proteins, IRBIT, Homer) [24–26]. Hence, IP₃Rs work as Ca²⁺ signaling hubs through which diverse cellular inputs are processed and then converted to cytosolic Ca²⁺ signals with precise temporal and spatial characteristics. Clearly, an in-depth understanding of the complex mechanisms through which interacting partners regulate IP₃Rs demands high-resolution structural characterization of the channel complex.

In mammals, three closely related IP₃R subtypes are encoded by different genes (ITPR1–3) from which splice variants also arise. The core properties of all IP₃Rs are similar and consistent with the sequence conservation between subtypes (~70%). Functional IP₃Rs are tetrameric assemblies (~1.3 MDa) of monomers of either identical or different subtypes, resulting in channel complexes with a wide range of functional properties in vivo. Early structural studies on IP₃R were focused simply on determining the correct overall structure of IP₃R, which was controversial for many years [27–31]. The first reliable structure of tetrameric full-length IP₃R1 was determined by our group to intermediate resolution (10–15 Å) in 2011 (Fig. 1) [32,33].

Fig. 1. Milestones toward the structure of IP₃Rs.

Paving the road for understanding the molecular mechanism underlying IP₃ receptor function began with several foundational studies that described increased intracellular Ca²⁺ mobilization by receptor-activated hydrolysis of PIP₂ [96]; stimulation of Ca²⁺ release from intracellular stores by the secondary messenger molecule IP₃; and Ca²⁺ regulation of IP₃-induced Ca²⁺ release [48,97]. Subsequently, the molecular identification of IP₃R was uncovered by purification of the IP₃-binding protein from the brain, its identification as a Ca²⁺ release channel through lipid vesicle reconstitution [98,99], and molecular cloning [79,100]. These studies were highly influential in shaping the next three decades of structure-function studies of IP₃Rs. The first images of purified IP₃R1 appeared by negative stain EM, and quick-freeze deep-etch EM was used to visualize the IP₃Rs in Purkinje cells [101]. Single-particle cryo-EM and negative stain EM of detergent solubilized IP₃Rs produced several low resolution (20–40 Å) structures, which unfortunately were not self-consistent, leaving the true 3D structure of the tetrameric channel unclear [27–31]. Meanwhile, X-ray crystallography produced structures for the small expressed portions of the N-terminal ligand binding domains [44–46]. Importantly, the 3D cryo-EM structure of IP₃R1 in the closed state at ~14 Å resolution was determined and rigorously validated, thus illuminating the authentic 3D structure of IP₃R1 [32,33]. Early adoption of direct electron detectors allowed for the IP₃R channel to catch the wave of the ‘resolution revolution’ with the first near-atomic resolution structure of IP₃R produced by cryo-EM [34]. An outstanding question on how many IP₃ ligands are needed to bind to the tetrameric channel in order to open the gate was settled using concatenated IP₃Rs. Cryo-EM structure of IP₃R in the presence of channel activator, adenophostin A [34–37], as well as structures of the type 3 IP₃R isoform were determined by single particle cryo-EM [38–40]. Lipid-binding sites were identified in IP₃R1 by cryo-EM followed by identification of ATP, IP₃, Zn²⁺, and Ca²⁺ binding sites. The Ca-III_S site in the ARM3 domain was determined to be the Ca²⁺ binding site for Ca²⁺ activation of IP₃R1 and IP₃R3 by mutagenesis and single-channel studies. Machine-learning analysis of cryo-EM data revealed structural dynamics of IP₃R1 controls access to IP₃ binding domains.

Over recent years, our collective knowledge of the structure and function of IP₃Rs and Ca²⁺ signals they evoke has grown considerably. There has been revolutionary progress in the cryo-EM field leading to high-resolution structures of ion channels including IP₃R Ca²⁺ channels [34–40]. In this review, we discuss recent progress in the IP₃R structure and function that have led toward enriching our understanding of the mechanism for long-range allosteric coupling between ligand-binding and channel gating. We focus on common elements of IP₃R architecture, conformational changes and ion permeation, emphasizing mechanistic principles of gating and regulation in the family of IP₃R channels.

2. Genuine multi-modal architecture of IP₃R

After the discovery and cloning of the IP₃ receptor over four decades ago, many studies have sought to elucidate the molecular mechanisms underlying IP₃ induced Ca²⁺ release via IP₃R channels [41]. Through these years several structural techniques were utilized to characterize the structure of IP₃Rs, including X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy and single-particle cryo-EM (Fig. 1). However, to study the full-length tetrameric assembly of IP₃Rs, single-particle cryo-EM became the approach of choice due to its large size (~1.3 MDa), inherent protein flexibility, and complexities dealing with detergents in crystallization. Therefore, almost two decades ago, the low-resolution structures of the entire IP₃R1 were solved almost simultaneously by multiple groups [27–31]. However, much to the consternation of the cryo-EM and interested biochemistry communities, none of these published structures at 20–30 Å resolution agreed, even about the overall architecture of the channel. This raised the question of the credibility of single-particle cryo-EM as a tool for structural determination. The long-standing controversy about the 3D architecture of IP₃R was one of the major obstacles slowing progress of the research aiming to understand the structure-function interrelationship of this channel. The first trustworthy 3D structure of IP₃R1 was determined at ~14 Å by single-particle cryo-EM as a result of extensive work towards optimization of cryospecimen preparation [32]. Importantly, this cryo-EM structure was rigorously validated by several methods, including class-average/map comparisons, tilt-pair validation, and use of multiple refinement software packages [33]. These two key studies laid to rest a critical controversy on the IP₃R quaternary structure and provided a stepping stone for future cryo-EM studies in many groups around the world.

Merely a few short years after the validation of IP₃R1 cryo-EM structure, technical improvements to cryo-EM hardware and software aided in the burgeoning capabilities of cryo-EM to achieve near-atomic resolution structures, deemed the “resolution revolution” [42]. By harnessing these new advances, a long awaited near-atomic resolution structure of the tetrameric, full-length IP₃R1 isolated from rat cerebellum was first revealed by single-particle cryo-EM [34]. Later cryo-EM studies of recombinantly expressed human IP₃R3 protein reflected the structural conservation of the overall domain folds and assembly within the IP₃R channel family [38–40]. Currently there are multiple IP₃R1 and IP₃R3 structures of the full-length channel determined by cryo-EM in the absence (apo) or presence of regulatory molecules (e.g., Ca²⁺, IP₃, Adenophostin-A, ATP, lipids). These structures have provided valuable atomic-level insights into the complexities of the IP₃R protein machinery, the mechanisms for its regulation and how disease-causing mutations may affect channel function.

The domain architecture of the IP₃R protomer (~2700 amino acids) was first described in its entirety based on the near-atomic resolution cryo-EM structure of IP₃R1 (Figs. 2, 3) [34] and consists of large cytoplasmic (CY) assembly comprising ~90% of the protein sequence consists of two N-terminal β-trefoil domains (β-TF1 and β-TF2), three armadillo solenoid folds (ARM1-ARM3), with a helical domain (HD) inserted between ARM1 and ARM2, an intervening lateral domain (ILD) comprised of two antiparallel β sheets and two short helices, leading into the transmembrane domain (TMD) comprised of six membrane spanning alpha-helices (TM1-TM6) and two membrane associated helices (MA1-MA2). A long C-terminal helical domain (CTD) is exposed to the cytosol and connected to the TMD via the linker domain (LNK), which is sandwiched between the ILD.

Fig. 2. Overall Domain Structure of IP₃Rs.

The top panel shows a linear depiction of the IP₃R1 protomer sequence showing the location of protein structural (in colored boxes) and functional domains. Magenta bars indicate unresolved regions in IP₃R structures. The lower panel lists the domain names, color coded as in the top panel, with the amino acid residues for the domain boundaries for rat IP₃R1 (P29994), human IP₃R2 (Q14571) as modeled using AlphaFold2, human IP₃R3 (Q14573) and the structurally homologous regions found in rabbit RyR1 (P11716).

Fig. 3. Structural conservation of intracellular Ca²⁺ release channels. (A-D).

Two opposing subunits of IP₃R1 (8EAR), IP₃R3 (7T3T), IP₃R2 (AlphaFold2) and RyR1 (7M6A) viewed along the membrane plane. Domains are color-coded according to Fig. 2. Domains of RyR1 with no structure consistent with IP₃R are shown in gray. Bound IP₃ molecules are colored magenta; bound ATP is colored orange. (E) Structural comparison of three isoforms of IP₃R channel in ligand-free state. One subunit is shown, IP₃R1 - 7LHE, colored by domain; IP₃R2 - predicted model, light blue; IP₃R3s - 6DQJ, coral; 6UQK, tan.

The N-terminal domains, β-TF1, β-TF2 and ARM1, constitute the ligand binding domains (LBDs) with an IP₃ binding pocket formed at the cleft between β-TF2 and ARM1 in each subunit (Fig. 4) [34–40]. Cryo-EM structures of IP₃R revealed the arrangement of the N-terminal LBDs within the tetrameric channel assembly. The β-TF1, β-TF2 and ARM1 domains within each IP₃R subunit form a triangular assembly, similar to that seen in crystal structures [44–46] and constitute an apical portion of the cytoplasmic solenoid that encircles the CTD bundle.

Fig. 4. Properties of the IP₃ binding pocket. (A-C).

The surface electrostatic potential in the IP₃ binding pocket of IP₃R1 (8EAR), IP₃R3 (7T3T), IP₃R2 (predicted model) with corresponding residues labeled. Blue surfaces - positive charges; red surfaces - negative charges. (D) Overall negative electrostatic surface potential for the cleft between RyR1 (7M6A) domains NTD-B and NSol, structurally similar to the IP₃R β-TF2 and ARM1 domains. (E-F) Schematic plot of the IP₃ molecule interacting with surrounding residues within the ligand-bound structure of IP₃R1 (8EAR) and IP₃R3 (7T3T). Colors indicate residue properties: green - hydrophobic; purple - positively charged; light blue - polar; red - negatively charged. (G) Structural alignment and overlay of the IP₃ binding pocket in IP₃R1 (8EAR, colored by domain), IP₃R3 (7T3T, tan) and RyR1 (7M6A, gray). (H) Structural alignment and overlay of the IP₃ binding pocket in ligand free structure of IP₃R1 (7LHE, colored by domain) and IP₃R2 (modeled with Alphafold2, light green).

Multiple positively charged conservative residues from β-TF2 and ARM1 are involved in coordinating IP₃ in the binding pocket in all three IP₃R subtypes (Fig. 4). Upon binding IP₃ there is ~4 Å closure of the cleft caused by a shift of the ARM1 domain toward the β-TF2 domain to allow for the coordination of the IP₃ molecule. It has been found that mutations causing cerebellar ataxia are located within or near the IP₃-binding pocket of IP₃R1, and these pathological mutations either impair IP₃ binding or disrupt channel gating without affecting IP₃ binding [43].

The β-TF1 domain is not directly involved in the formation of the IP₃ binding pocket, but it reduces the affinity of IP₃ binding [47]. The lack of the β-TF1 results in inhibition of IP₃-induced Ca²⁺ release although this mutant IP₃R1 had 10-times higher IP₃-binding affinity than the wild type. Thus, this N-terminal domain was referred to as a suppressor domain (SD). The β-TF1 forms intra-subunit interfaces with the β-TF2 and ARM1 domains, as well as interacts with the β-TF2, ARM2 and ARM3 domains from the neighboring subunit. It was reported that the β-TF1 loop comprising residue Y167 (‘hot spot’) is essential for the Ca²⁺ release activity of IP₃Rs [48] and is located at the inter-subunit β-TF1-β-TF2 interface which undergoes conformational changes upon binding of adenophostin A (a structural analog of IP₃) [35]. Collectively, these data suggest that the β-TF1 domain might be important for stabilizing inter-domain interactions and plays an important role in coupling ligand-binding signals to channel opening. In addition, a helix-loop-helix structure (so called the ‘handle’ of the hammer-shaped SD) of the β-TF1 projects inward into the hollow solenoid of the channel and interacts with the ARM3 domain from the neighboring subunit. It was demonstrated that deletion of the handle does not affect IP₃-mediated channel activation [48]. This justifies the structure-stabilizing role of the β-TF1-ARM3 interaction. Our recent progress with analysis on how ligand-binding signals might be communicated within the conformational protein landscape are discussed in the section below.

Of note, ryanodine receptors and IP₃Rs share a modest ~20% sequence identity within their first ~700 residues [50]. Upon the determination of structures for both channels, the protein folds appear nearly identical for the N-terminal region [49]. A cleft between the RyR domains NTD-B and Nsol, equivalent to the β-TF2 and ARM1 domains in IP₃Rs, exists and is similar to the IP₃ binding pocket in IP₃Rs (Fig. 4). However, in RyR1 the residues within this cleft form a negative surface charge in contrast to the positively charged surface in IP₃Rs that contributes to IP₃ binding and underlies the rationale for why RyRs do not bind IP₃ [50].

The Ca²⁺ conduction pathway is formed in the tetrameric assembly of the channel along the central four-fold axis. The TM5 and TM6 helices form a right-handed pore bundle shaping the ion conduction pathway. The TM1-TM4 connect to the pore-forming bundle through a lateral TM4-5 amphipathic helix, which allows for a domain swapped configuration of the TMDs whereby the TM1–TM4 bundle of one subunit interacts with the pore forming helices from the neighboring subunit. This domain swapped architecture is a highly conserved structural feature for many tetrameric cation channels. The luminal entrance of the ion conduction pathway harbors the P-helix and P-loop containing the selectivity filter sequence (GGVGD) and a ring of hydrophobic side chain residues F2586 and I2590 from the TM6 helices form a constriction site in the closed channel that can prevent the flow of Ca²⁺ ions across the membrane bilayer. The G2498S mutation in the selectivity filter was found to be associated with anhidrosis in humans and abolished Ca²⁺ releasing activity in the IP₃R2 channel [51].

Lipids are integral to structure and function of ion channels residing in a lipid membrane bilayer in vivo. Recent cryo-EM studies revealed lipid molecules bound to the TM helices in IP₃R1, suggesting conserved locations of protein-bound lipids among homo-tetrameric ion channels [36]. However, the exact molecular mechanism by which lipids exert their effects on the IP₃R channel remains to be elucidated.

In the tetrameric assembly, the α-helical distal portion (residues 2692–2737) of the CTD (Fig. 2) from each protomer forms a central, left-handed helical bundle, and the CY domains are constructed around this CTD bundle to form α-helical solenoid modules, well adapted for binding multiple ligands and protein-protein interactions. Noteworthy, the N-terminal β-TF2 domain makes an inter-subunit interface with the α-helical CTD from the neighboring subunit [34–37]. Thus, the entire tetrameric IP₃R assembly is uniquely arranged around two helical bundles: a ~70 Å left-handed CTD ɑ-helical bundle is connected to the right-handed ɑ-helical TM6 bundle via the ILD/LNK domains lying at the CY-membrane interface. The allosteric nexus formed by ILD and LNK domains represents the sole direct structural link between the CY and TM domains.

It is notable that both IP₃R1 and IP₃R3 exhibit identical overall architectural arrangements (Fig. 3). However, in IP₃R3 structures the central coiled-coiled CTD ɑ-helical bundle is not resolved [38–40], yet based on studies of IP₃R1 the CTD bundle is one of essential elements for maintaining structural integrity of the channel assembly and for transmitting ligand-evoked signals from LBDs to the ion-permeation pore [34–37]. This view is consistent with functional IP₃R1-TM_RyR chimeras [46] since the RyR structure does not have an extended CTD, the reduced efficacy of IP₃ suggests that within IP₃R-TM_RyR, communication between the N-terminal β-TF2 and channel are less effective than in native IP₃R1, and with the evidence that IP₃R1 function is lost when more than 43 residues are deleted from the C-terminus [52]. It is conceivable that the methods used for expression and purification of recombinant IP₃R3 yielded the channel lacking the structural coupling between the LBDs and TMDs underlying the activation mechanism in IP₃Rs due to compromised structural integrity of the tetrameric channel assembly. Another remote possibility is that different IP₃R subtypes might explore different mechanisms for allosteric regulation of the channel activity given differences in the lengths and sequences of their C-terminal tails (Fig. 2).

While structural studies of IP₃R2 have yet to be performed, a predicted molecular model generated using Alphafold2 [53] suggests that the subtype 2 receptor shares a similar domain architecture as defined for subtypes 1 and 3 (Figs. 2, 3). Moreover, overall comparison of IP₃Rs and RyRs, the two related families of Ca²⁺ release channels, parallels many structural and functional properties in both families, and there has been fruitful synergy between the research on these important ion channels (Figs. 2, 3) [41,49]. Thus, the emerging concept for the versatility of Ca²⁺ signaling via Ca²⁺ release channels is founded in the distinctive structure of IP₃Rs. In this context, IP₃R channels can be viewed as having highly dynamic scaffolding, where binding of multiple ligands and modulatory molecules alters the conformational landscape of protein and propagates ligand-evoked signals to the channel pore.

3. Ca²⁺ binding diversifies the conformational ensemble of IP₃R

A functional hallmark of IP₃R channel gating is its biphasic regulation by Ca²⁺ in the presence of activating concentrations of IP₃, with low Ca²⁺ concentrations promoting channel activation and higher Ca²⁺ concentrations inhibiting the channel [54–56]. These observations suggest the presence of at least two functional Ca²⁺ binding sites in IP₃R with different Ca²⁺ affinities – a high affinity Ca²⁺ binding site to allow for Ca²⁺ activated release and a low affinity Ca²⁺ binding site that can inhibit Ca²⁺ release [57]. While the binding of both IP₃ and Ca²⁺ are required for channel activation, the exact mechanism of their interplay and how these co-agonists transmit the gating signal, as well as how Ca²⁺ binding can lead to inhibition has not been well understood. In the primary sequence, IP₃Rs lack any canonical Ca²⁺ binding motifs such as an EF-hand or C2 domain, which has made identification of Ca²⁺ binding sites complicated [58]. Yet, purified IP₃Rs reconstituted into lipid bilayers are regulated by Ca²⁺ in a biphasic manner strongly indicating that the IP₃Rs have intrinsic Ca²⁺ binding site/s [59].

Recently, insights into Ca²⁺ regulation of channel gating have been advanced by structural studies of full-length, tetrameric IP₃Rs by single-particle cryo-EM and validated with functional studies [37,60]. Cryo-EM studies of IP₃Rs determined in the presence of Ca²⁺ have identified Ca²⁺ binding sites within the channel in conditions to promote capturing the channel in physiologically relevant states (Supplementary Figs. 1 and 2). In IP₃R1 five Ca²⁺ binding sites per monomer were identified, three in the cytosolic LBD and Ca²⁺-sensor (S) domains (Ca-I_LBD,Ca-II_LBD, Ca-III_S) and two within the ion conduction pathway (Ca-IV, Ca-V) (Fig. 5). In IP₃R3 two sites have been described, Ca-CD and Ca-JD, the latter of which is equivalent to Ca-III_S in IP₃R1. In IP₃R1, two Ca²⁺ binding sites were identified within the apical ligand binding domains, one at the intersubunit interface of β-TF1 and β-TF2 from neighboring subunits (Ca-I_LBD) and the second formed at the intra-subunit interface between the β-TF1-β-TF2 domains (Ca-II_LBD). In Ca-I_LBD the Ca²⁺ is coordinated by carboxyl groups from the negatively charged D426 and D180 residues in βTF2 and βTF1 domains, respectively. Ca-II_LBD utilizes the oxygen atoms from the E283 side chain and backbone oxygen atoms from the surrounding βTF1 domain residues to coordinate Ca²⁺. The location of the Ca-I_LBD and Ca-II_LBD sites are consistent with predicted Ca²⁺ binding sites [44,61]. The third cytosolic Ca²⁺ binding site, Ca-III_S, is located at the interface between the ARM3 and LNK domains and is comprised of the acidic E1978, E2042 (ARM3 domain) and T2654 (LNK domain) residues in IP₃R1. Ca²⁺ binding to the Ca-III_S site has also been observed in two IP₃R3 cryo-EM structures. Moreover, cryo-EM structures of RyR1 in the presence of Ca²⁺ have also shown that conserved residues in the structurally homologous region of RyR1 also serve to coordinate Ca²⁺ (Fig. 5). Mutations in the residues constituting the Ca²⁺ binding pocket in RyR1/2 are the underlying cause of central core disease and arrhythmogenic diseases [62–64]. The Ca²⁺ binding sites, Ca-IV and Ca-V, were identified within the luminal side of the ion conduction pathway and may represent high affinity ion binding sites for the Ca²⁺ ion as the ion passes from the ER lumen to the cytosol. These sites could be involved in luminal regulation of the channel, however, their functional role remains to be established.

Fig. 5. Defining Ca²⁺ binding sites in IP₃Rs.

Overall topology of Ca²⁺ binding sites is shown for a single subunit of (A) IP₃R1 (8EAQ) and (B) IP₃R3 (6DRC) colored by domains as defined in Fig. 2. (C,D) Comparison of Ca²⁺-binding sites identified in ligand binding domains of CIA-IP₃R1 (8EAR, colored by domain) and Ca-IP₃R1 (8EAQ, lavender): (C) Ca-I_LBD, red arrow indicates the change in side chain position of D426 contributing to loss of Ca²⁺ binding to Ca-I_LBD in CIA-IP₃R1; (D) Ca-II_LBD. (E) Structural alignment and overlay of CD-Ca binding site between IP₃R3 and IP₃R1 (IP₃R1 - 8EAR, colored by domain; IP₃R3 - 6DRC/pink, 7T3T/tan) Ca²⁺ has only been observed in 6DRC. (F) Structural alignment and overlay of Ca-III_S binding site in the Ca²⁺ - sensor domain for IP₃Rs and RyR1 (IP₃R1 - 8EAR, colored by domain; IP₃R3 - 6DRC, pink; 7T3T, tan; RyR1 - 7M6A, gray). Ca²⁺ ions are shown as red spheres; corresponding residues are displayed in a stick representation and labeled.

An additional Ca²⁺ binding site (Ca-CD) in IP₃R3 has been proposed at the interface between HD and ARM2 domains (CLD domain in IP₃R3 structure) in based on the cryo-EM structure of IP₃R3 in the presence of 2 mM Ca²⁺. In order to form the Ca-CD site, it is necessary for the ARM2 domain to adopt a retracted conformation, moving ~30 Å toward the HD and reducing the inter-subunit contacts between ARM2 and β-TF1 from the neighboring subunit. Supporting the Ca²⁺ coordination within this site are main-chain oxygens atoms from R743 (HD) and E1122 (ARM2) and side chain atoms from E1125.

Several structures of IP₃Rs have now been determined by cryo-EM in the presence of Ca²⁺ and have been observed to adopt a similar retracted conformation of ARM2 (Figs. 5, 8), yet only one group has reported Ca²⁺ binding site in this domain [38]. Moreover, in other IP₃R3 cryo-EM structures, as well as in IP₃R1, the domains comprising the Ca-CD site have not been sufficiently resolved to confidently detect the bound Ca²⁺ ion at this location [37,40]. It was also demonstrated that the same region in IP₃R1 is highly dynamic exhibiting significant motions [37].

Fig. 8. ARM2 domain undergoes large conformational change.

under conditions of ligand-free (left column), Ca²⁺/IP₃/ATP activated (middle column) and high Ca²⁺ inhibited (right column). Atomic models for the tetrameric assembly of type 1 and type 3 IP₃R channel are viewed along the central four-fold axis from the cytosol with color-coded by domain, one subunit is shown with atomic surface rendering. Upper panel: IP₃R1; lower panel: IP₃R3. Conformation of ARM2 (extended or retracted) and gate (open or closed) and PDB ID are listed.

The Ca²⁺ bound IP₃R1 structures were obtained in different Ca²⁺ and ligand combinations to better understand the role of Ca²⁺ during channel gating and inhibition. In the IP₃R1 structure determined in the presence of an inhibitory Ca²⁺ concentration (20 μM), the five Ca²⁺ binding sites were occupied, while at 2 μM Ca²⁺ in the presence of IP₃ and ATP, four Ca²⁺ binding sites were occupied. Ca²⁺ was not seen in the Ca-I site under lower Ca²⁺ conditions, indicating this site could have a lower apparent affinity for Ca²⁺ than the other Ca²⁺ binding sites.

It has been proposed that binding of IP₃ serves to alter the Ca²⁺ regulation of the channel, whereby IP₃ allosterically regulates Ca²⁺ inhibition without affecting Ca²⁺ activation [65–67]. In the light of the Ca²⁺ binding sites described for IP₃R1 obtained at low and high Ca²⁺ concentrations, the Ca-I_LBD site may potentially play a role in Ca²⁺ inhibition of IP₃R1. Conformational changes within the Ca-I_LBD binding site suggest that IP₃ regulates the conformation of the Ca-I_LBD site. However, based on current available structures, the impact of IP₃ binding on the Ca-I_LBD at inhibitory Ca²⁺ concentrations is an outstanding question. Functional studies of Ca-I_LBD would be necessary to address its role in Ca²⁺ dependent inhibition. It has also been suggested that the IP₃R3 Ca-CD site plays a role in Ca²⁺ inactivation, however, further structural and functional characterization is still needed to address the role of this site in IP₃Rs.

There has long been a search for the Ca²⁺ binding site/s necessary for Ca²⁺ activation of the IP₃R channels. A previous candidate for a functional Ca²⁺ binding site was E2101. However, mutations of this residue in IP₃R1 altered both the Ca²⁺ dependent activation and inhibition of channel activity [68,69]. Cryo-EM structural analysis revealed that this residue is not involved in coordinating Ca²⁺ but may have an ancillary role in favoring Ca²⁺ binding to the Ca²⁺-sensor (Ca-III_S). The Ca-III_S binding site was shown to be occupied under activating and inhibitory Ca²⁺ conditions and mutations to the analogous site in RyR1 altered Ca²⁺ activation of the channel [70]. However, the functional role for the Ca-III_S site was unknown for IP₃R channels. Through mutagenesis and electrophysiological studies, we have now demonstrated that the Ca-III_Ssite is responsible for Ca²⁺ activation in IP₃R1 and IP₃R3. Mutation of the conserved acidic residues within the Ca-III_S binding site in subtype 1 and subtype 3 IP₃R resulted in Ca²⁺-dependent activation being significantly diminished or abrogated without altering the Ca²⁺-dependent inhibition (Fig. 6) [60]. Neutralizing the negative charge on the side chain residues dramatically shifted the Ca²⁺ dependency to a higher concentration. Importantly, the Ca-III_S mutations did not compromise the tetrameric integrity and proper localization of IP₃Rs. Moreover, the Ca-III_S site is occupied in the presence and absence of IP₃ indicating that access to the activating Ca²⁺ site is not regulated by IP₃. The originally targeted putative Ca²⁺-sensor residue, E2101, affected both activation and inhibition of the channel, indicating that it plays an allosteric role in channel regulation. Based on cryo-EM structures, the E2101 residue is nearby, but outside of the Ca-III_S binding pocket and it can likely form a hydrogen bond with neighboring residues within the ARM3 domain. Thus, mutation of E2101 and the analogous residues in IP₃Rs/RyRs could disrupt the ability to maintain the structural fidelity for the Ca-III_S binding site and signal transmission to the TM6 gating helix [37].

Fig. 6. Functional determination of the activating Ca²⁺ binding site in IP₃R1 and IP₃R3.

Single-channel current traces obtained using nuclear patch-clamp experiments from DT40-3KO cells stably expressing (A) wild-type (WT), (B) E2002D, and (C) E2002A hIP₃R1 in response to increasing [Ca²⁺] (10 nM-100 μM) at fixed concentrations of IP₃ (10 μM) (when IP₃ concentration is not indicated) and an optimal [ATP] (5 mM). The maximal open probability (P_o) for the WT and E2002D channels were obtained in the range of 0.2 to 1 μM [Ca²⁺] and diminished upon a further increase in [Ca²⁺]. The P_o for E2002D remained diminished compared to WT channels at all the tested [Ca²⁺] without a notable change in Ca²⁺ dependency. The P_o of E2002A was significantly diminished when compared to either WT or E2002D and required a higher [Ca²⁺] (3 μM) for maximal P_o. The P_o remained unaltered even upon increasing the [IP₃] to 100 μM at optimal [Ca²⁺] and [ATP] in all the stable cell lines (4^th representative trace in each panel). (D) Pooled data displaying diminished open probability of mutant IP₃R1 channels, biphasic effect of Ca²⁺ on IP₃R channel opening, and a clear rightward shift in the Ca²⁺-dependency of E2002A channel compared to WT and E2002D channels. Modified from [60].

The residues involved in the Ca²⁺ coordination of Ca-III_S are conserved in both RyRs and IP₃Rs and the functional and structural observations strongly indicate that the Ca²⁺ activation site is evolutionarily well conserved and a common underlying mechanism that governs both IP₃Rs and RyRs channel activation (Fig. 5). Mechanistically, Ca²⁺ binding to the activation site may facilitate conformational changes in IP₃R structure, promoting pore opening. However, it is still not clear whether Ca²⁺ binding is cooperative and if all the stimulatory/inhibitory Ca²⁺ binding sites must be occupied for channel opening/closure. Future experiments using concatenated receptors with one or more subunits of a tetramer harboring mutations at the Ca²⁺ binding sites will address the precise stoichiometry.

4. Allosteric nexus as a structural platform for transmitting ligand-evoked signals

The allosteric nexus is a critical junction between the cytoplasmic and transmembrane domains that serves to process and transmit the ligand-dependent regulatory signals from the cytosolic domains to the channel gating apparatus. This occurs through the intra-subunit assembly of the ILD and LNK at the membrane-cytosol interface. It has recently been shown that besides a conserved Ca²⁺-binding site (Ca-III_S), which utilizes residues from the ARM3 and LNK domains connected to the nexus, the nexus carries binding sites for zinc and ATP [36–38,40].

While it remains to be established how a conformational wave generated upon binding of the activating ligands propagates from the LBDs in the apical portion of the CY scaffold to the ILD/LNK (‘nexus’) assembly, cryo-EM structures clearly revealed that upon binding of activating ligands the allosteric nexus undergoes lateral and rotational movements that result in dilating the nexus assembly. These conformational changes appear to impose the impetus that allow to reconfigure the connected TM helices that leads to opening of the channel gate [35,37].

Recent studies provided evidence confirming activating function of the Ca-III_S site located at the ARM3 domain connected to the ILD [60]. In addition, it was reported that the Ca-III_S site undergoes IP₃-induced conformational changes and this leads to activation of the channel gating [38,40]. However, this notion is not supported by studies of IP₃R1 showing that the Ca-III_S site does not change between apo-, Ca²⁺- and IP₃-bound structures of IP₃R1 [37].

Adenine nucleotides are known to bind IP₃Rs and can enhance IP₃-induced Ca²⁺ release in a concentration dependent manner. All three IP₃R subtypes bind ATP with varying affinities, ranging from low micromolar to millimolar levels [71–73]. The addition of ATP in the presence of channel activators, IP₃ and Ca²⁺, in single channel studies revealed that ATP increased the channel open probability without affecting channel conductance [55]. Binding of ATP is proposed to tune the sensitivity of the IP₃R channel to activating Ca²⁺. ATP concentrations in the cell stay relatively high and constant (> 1mM) and as such ATP may remain constitutively bound to enhance Ca²⁺ release upon receptor stimulation. It has however been proposed that ATP⁴⁻, an ATP derivative in neutral solution, is essential for regulation of IP₃R-induced Ca²⁺ release [74], and because the levels of this form of ATP are in the micromolar range, a possibility exists that IP₃Rs are dynamically regulated by changes in ATP concentrations. In addition, many studies have demonstrated that the mitochondria and ER are in close proximity in many cells types, thus release of ATP from mitochondria in close proximity to IP₃R and regulation of the IP₃R-mediate Ca²⁺ release by ATP may have significance for intracellular signaling by providing a cross-talk between the mitochondria and the ER [75–77].

In the search for the molecular determinants necessary for ATP binding and potentiation of channel activation, several canonical ATP binding motifs (GXGXXG) were identified within the IP₃R primary sequence (ATP-A:2016–2021 in IP₃R1 S2+ and IP₃R3; ATP-B site:1773–1780 in IP₃R1 S2+, and ATP-C:1688–1732 in IP₃R1 S2−). All three putative ATP-binding sites were shown to bind ATP in recombinantly expressed fragments, however, amino acid substitutions within the motifs did not alter ATP regulation of IP₃R1 or IP₃R3 subtypes indicating that ATP modulates IP₃R-mediated Ca²⁺ release by binding to a region distinct from the glycine-rich motifs [78–80].

Cryo-EM studies of IP₃R1 and IP₃R3 proved crucial to delineating the ATP binding site. Structures of IP₃Rs in the presence of saturating ATP concentrations were solved by cryo-EM to better than 3.5 Å resolution allowing for the accurate model building for protein side-chains and identification of the ATP binding site [37,40]. In the presence of ATP, strong cryo-EM densities were observed at an interfacial region between the cytoplasmic and TM domains and could accommodate an ATP molecule. One bound ATP per IP₃R subunit was identified and is consistent with Scatchard analysis [81].

The ATP binding pocket is formed at an interface of two domains, ILD and LNK, from the same subunit, and adjacent to the cytosolically exposed end of the TM6 helix. The binding pocket is electrostatically favorable for accommodating the negatively charged phosphate tail through a series of lysine residues from ILD and LNK domains while the adenine moiety is predominantly surrounded by hydrophobic residues (Fig. 7).

Fig. 7. Ligand binding domains within the allosteric nexus at the cytosolic–lipid bilayer interface.

(A) Zoomed-in view along the membrane plane of the domains comprising the allosteric nexus interface of one IP₃R1 subunit. IP₃R1 (8EAR) is colored by domain as defined in Fig. 2. (B) Structural alignment of the IP₃R1 subunit in A with IP₃R3 (tan; 7T3T) and RyR1 (gray; 7M6A). (C-E) Upper panels show the ATP binding site for (C) IP₃R1 (8EAR), (D) IP₃R3 (7T3T) and (E) RyR1 (7M6A) overlaid with surface electrostatic potential (blue - positive charge; red - negative charge) with several residues contributing to ligand-association labeled. Lower panels show the residues that comprise the ATP binding pocket in the above panels, respectively. Colors indicate: green - hydrophobic; purple - positively charged; light blue - polar; red - negatively charged residues, gray - zinc, lines - salt bridges.

There are little observed conformational changes within the ATP binding pocket between ATP-bound and apo-state structures of IP₃R, however the ATP binding site is located at a critical junction between cytoplasmic and transmembrane domains, termed the allosteric nexus. The ATP binding site is in a unique position to impart a structural/allosteric effect in synergy with the activating Ca²⁺ binding site (Ca-III_S) located in the nearby ARM3 domain to potentiate channel openings. The ATP interactions with IP₃R1 may stabilize domains to amplify the effects of Ca²⁺ binding on the gating of the channel pore, thus increasing the apparent affinity of Ca²⁺ needed to activate the channel consistent with single channel studies [82].

While the structural fold of the ATP binding pocket is conserved among IP₃Rs, subtle differences in amino acid composition result in differences on the charged surface of the binding pocket with IP₃R1 having greater positively charged surface area available for interaction with the ATP phosphates than observed in IP₃R3 (Fig. 7). This is consistent with the much higher affinity for ATP reported for IP₃R1 than IP₃R3. Of note, sequence alignment of positively charged, polar lysine residues in IP₃R1 at 2221, 2224, 2633 positions which are identified to interact with the ATP triphosphate tail [37,40] revealed the K2224 and K2633 are conserved in all the three IP₃R subtypes. Surprisingly, the K2221 in IP₃R1 aligns with a positively charged, polar arginine (R) in IP₃R2 and a negatively charged, polar glutamic acid (E) in IP₃R3 which could perhaps contribute to the previously reported higher sensitivities of IP₃R1 and IP₃R2 to ATP as compared to IP₃R3.

Moreover, the ATP binding pocket appears to be a conserved feature between IP₃Rs and RyRs indicating that a similar underlying mechanism may prime these channels to ATP potentiation [37]. RyR channel activity is also potentiated by ATP binding, and the ATP-binding site has been identified at the interface between the thumb and forefinger (TaF) domain and CTD, which are structurally equivalent to the ILD and LNK domains in IP₃Rs. However, in RyR1 ATP adopts a slightly different binding position with the phosphate tail pointed toward the membrane plane while in IP₃Rs it lies more parallel to the membrane plane. The difference in orientation of ATP in RyR may be due to an additional positively charged residue (R4215) that can coordinate with the phosphate tail that is not observed in IP₃Rs. Recent mutagenic analysis of residues within the RyR1 ATP binding pocket supports that this ATP binding site is responsible for regulating Ca²⁺ release from RyR1 and linked to myopathies in humans [83].

Adjacent to the ATP binding domain is the C2H2-like zinc-finger domain composed of amino acids from the LNK domain (C2611, C2614, H2631, H2636 in IP₃R1; C2538, C2541, H2558 and H2563 in IP₃R3; and C2562, C2565, H2582, H2587 in IP₃R2) and the resolution of the IP₃R1 and IP₃R3 cryo-EM density maps are sufficient to model one bound zinc molecule per subunit. An equivalent Zn²⁺-finger domain was also described for RyRs in its CTD, which is structurally analogous to the IP₃R LNK domain (Fig 7). Mutations to the conserved cysteine or histidine residues necessary for tetrahedral coordination of the Zn²⁺ molecule were shown to either inhibit or abolish IP₃-induced Ca²⁺ release without affecting IP₃ binding to the IP₃R1 and abolished the response of RyR to caffeine activation [47,84]. While there is some indication that RyR2 may be regulated by Zn²⁺ [85], evidence for physiological regulation of IP₃Rs by Zn²⁺ has not been established. Given the proximity of the LNK domain to the TMDs and regulatory ligand binding sites, this zinc-binding domain clearly has an important role in influencing channel gating. It is of note that in all cryo-EM studies of IP₃Rs to date that have identified bound Zn²⁺ ions, the buffer solutions have also included the high-affinity cation chelators EDTA and/or EGTA suggesting that Zn²⁺ is either a constitutively associated cation with a high binding affinity or that the concentration of contaminating Zn²⁺ is beyond the capacity of the buffering solutions.

5. Dynamic conformational landscape of IP₃R as a basis for multi-modal regulation

IP₃R channels integrate numerous regulatory inputs including Ca²⁺ levels in the cytosol and ER lumen, ATP binding, protein phosphorylation and thiol modifications to alter channel function and fine tune the Ca²⁺ signal. However, ultimately the activation of IP₃R relies upon an intricate interplay of its primary ligands, IP₃, Ca²⁺, to initiate a structural rearrangement within the channel architecture to allow for Ca²⁺ to move across the ER membrane via a chemical gradient into the cytosol. It is obvious that channel functionality greatly benefits from the modular solenoid arrangement of the CY domains which flexibility allows IP₃Rs to accrete many auxiliary modulatory proteins in vivo [24]. In cryo-EM, the protein sample, trapped within vitreous ice under particular buffer conditions, is captured in discrete static conformations, and with sufficient number of particles, computational sorting can retrieve informative structural details about these conformations. However, there is a gap between the high-resolution structural snapshots of the channel and functional studies demonstrating different levels of IP₃R channel activity even under constant experimental conditions. Clearly, to derive functionally relevant connections between high-resolution snapshots, there is a pressing need for analysis of ligand-dependent dynamic conformational landscape of the channel protein.

Recently, new deep-learning based methods of manifold analysis have allowed for the extraction of protein dynamics, effectively generating molecular movies for the trajectories of macromolecular movements based on experimental cryo-EM 2D data without any structural interpolations [86–90]. By harnessing the computational power of deep-learning neural network approach [87] and 3D variability analysis [89], molecular motions of the IP₃R1 protein were extracted directly from cryo-EM density data [37]. These data revealed that the ARM2 domain in IP₃R1, comprising the peripheral cryo-EM densities of the cytoplasmic scaffold, exhibits an intrinsic flexibility, exploring motions between an extended or retracted conformation (Fig. 8). In the extended conformation, the ARM2 domain from one subunit makes extensive contacts with β-TF1 and ARM1 domains from the neighboring subunit. While in the retracted conformation the inter-subunit contacts are significantly reduced because the ARM2 domain retracts ~30 Å away from the neighboring subunit and shares only a minimal interface with the β-TF1 domain. We found that unliganded (apo) and high Ca²⁺ IP₃R1 structures exhibited the ARM2 domain preferentially in the extended conformation, while the IP₃R1 structure determined in the presence of IP₃, ARM2 was predominantly observed in the retracted conformation (Fig. 8).

The same motion trajectories for the ARM2 were observed for all three states (apo, Ca²⁺-bound and Ca²⁺/IP₃/ATP-bound IP₃R1s), however the amplitude of motion towards the retracted conformation increased in the presence of IP₃ [37]. Based on these observations, we conclude that IP₃-binding process in the tetrameric channel is coupled to conformational motions of the ARM2 domain, and exhibits a mechanism similar to a ‘reversible ratchet’ where ARM2 (“gear”) switches back-and-forth between ‘extended’ and ‘retracted’ conformations with the extended conformation restrictive for IP₃ binding and retracted conformation suitable for capturing IP₃ due to release of structural constraints at ARM2-βTF1 and ARM2-ARM1 interfaces between the neighboring subunits (Fig. 9).

Fig. 9. Model of allosteric regulation underlying channel activation and inhibition in IP₃Rs.

IP₃Rs exhibit structural dynamics, particularly within the ARM2 domain that switches between ‘extended’ and ‘retracted’ conformations (larger gray arrowhead indicates greater amplitude of exploring the displayed conformation). Intrinsic flexibility of ARM2 domain allows for a reversible ratcheting mechanism that impacts IP₃ binding with the ‘extended’ conformation suitable for capturing IP₃ due to release of structural constraints at interfaces between ARM2 and βTF1’ and ARM1’ from the neighboring subunit. Ca²⁺ activation requires Ca²⁺ binding to the Ca-III_S in the ARM3 domain and the Ca-II_LBD and Ca-II_LBD binding site is occupied by Ca²⁺. Ca²⁺ Inhibition of IP₃-bound-IP₃R may occur by Ca²⁺ binding to the Ca-I_LBD site or an additional Ca²⁺-binding site.

The ARM2 anchors the inter-subunit LBD interfaces thereby restricting the IP₃-binding domain dynamics that is required for capturing the ligand (Fig. 8). Furthermore, in the IP₃R lacking the β-TF1 domain the IP₃-binding pocket will display a higher degree of dynamics that correlates with observed higher IP₃-binding affinity [47]. And as such, we propose that the characteristic motions of the ARM2 and the correlated flexibility of the IP₃-binding pocket might define isoform-specific affinity of IP₃Rs. This explains why the N-terminal βTF2-ARM1 expressed and isolated as individual entities from all three IP₃R subtypes bind IP₃ with similar affinity, yet the full-length receptors display different affinities to IP₃ (IP₃R2>IP₃R1>>IP₃R3) despite conservation of the IP₃-coordinating residues across the IP₃R subtypes. It is conceivable that IP₃R subtypes exhibit different conformational dynamics underlying subtype-specific IP₃-binding.

The retracted and extended forms of ARM2 have also been observed in cryo-EM structures of IP₃R3s, where the predominant class of particles produces a cryo-EM structure in the unliganded state with ARM2 in the extended form. In the presence of channel activators and with multiple finely separated 3D classes, one small class of particles resulted in the ligand-bound activated channel with ARM2 in the retracted conformation, consistent with retracted conformation seen in IP₃R1 [40]. A third structure has also been observed to contain the retracted ARM2 [38], but is not consistent with the high Ca²⁺ (20 μM) structure of IP₃R1. The extreme Ca²⁺ concentration (2 mM Ca²⁺) may account for the differences observed.

With multiple cryo-EM structures of IP₃Rs determined under different conditions (Supplementary Figs. 1 and 2), the structural underpinnings toward IP₃R activation and inhibition have begun to emerge (Fig. 9). In the tetrameric channel the role of the dynamic ARM2 is attributed to weakening inter-subunit interactions with βTF1 and ARM1 domains which allows the IP₃-binding pocket to explore multiple conformations including the conformation capable of capturing IP₃.

Based on several proposed structures of IP₃R3 in the ‘inactive state’, it is observed that at high (~2 mM) Ca²⁺, the cytosolic subunits of the channel splay away from the central 4-fold axis [38,40]. The observed ~30 Å dilation of the LBD ring leads to a disruption of the inter-subunit interfaces and loss of structural integrity of the cytosolic scaffold. These structures raise a serious question about their physiological relevance. It becomes difficult to conceive what it would take to reconfigure the channel from such extreme–conformation to the conformation susceptible to binding of activating ligands? Another possibility is that the observed ‘inactive’ conformation reflects a detrimental disintegration of the tetrameric channel assembly under chosen experimental conditions. Further rigorous analysis of the channel conformational landscape combined with high-resolution structural determination and functional studies is needed to reconcile these observations and to establish a functional path of structural changes underlying communication of ligand-evoked signals toward channel activation and inhibition.

6. Perspectives

With the “resolution revolution” in cryo-EM, the picture of IP₃R channel gating and its multi-modal regulation seems to be developing in a straightforward manner, however, many fundamental questions still remain unanswered which curtail our understanding of the molecular mechanisms governing control of IP₃R-mediated Ca²⁺ signaling: What is the structural basis for functional coupling between the ligand binding domains during IP₃/Ca²⁺-mediated channel gating? Does the binding of Ca²⁺ to the sites identified in the cryo-EM structures underpin the biphasic regulation of IP₃R by Ca²⁺? How does the unique IP₃R architecture contribute to channel modulation by an array of regulatory proteins? What is a structural basis of isoform-specific properties of three members of the IP₃R family? How are IP₃Rs assembled into clusters on ER membranes?

Recent near-atomic resolution structures of IP₃R1 and IP₃R3 determined in the apo- and several ligand-bound states revealed common elements of the channel architecture and now begin to provide important insights into the molecular mechanisms of dual activation of the channel by IP₃ and Ca²⁺. However, many structural details are still missing due to insufficient local resolution that includes side-chain placements in some parts of the protein, conformations of flexible loops, ions and water. Of note, 13% and 20% of the protein backbone is not resolved in IP₃R1 [37] and IP₃R3 structures [38,40], respectively, including phosphorylation sites and other intrinsically flexible or disordered regions. Noteworthy, while both IP₃R1 and IP₃R3 exhibit similar architectural arrangements, the helical bundle formed by the C-terminal domains in IP₃R1 is not resolved in the IP₃R3 cryo-EM maps to make definitive conclusions about the conformational changes underlying ligand-binding and gating activation, which may be conserved among IP₃R subtypes. Each of these channel subtypes has unique gating properties, and in many instances, different subtypes assemble as heterotetrameric complexes to form the native IP₃R channel in vivo. However, due to the lack of high-resolution structure of IP₃R2, subtype specific gating and modulation remain largely incomplete.

Compositional and conformational variability among particles in single-particle analysis remains a key challenge for the field. Approaches such as classification have been around for decades [91–93], and partially address the issue by at least providing a handful of stable states, but do not fully characterize the full particle variability available from the particle population in the vast majority of cases. Single-particle cryo-EM studies have reported that the IP₃R structure undergoes conformational changes upon ligand binding [35–40], suggesting structural flexibility of IP₃R that allows the channel to switch from a closed state, capable of interacting with its ligands, to an open state, capable of transferring Ca²⁺ ions across the ER membrane.

However, all these 3D cryo-EM structures represent snapshots of defined static channel conformations, and the mechanistic insights are derived based on interpolations between discrete structures, each of them likely a mixture of states from a dynamic conformational ensemble. It is clear that the large IP₃R assembly comprising multiple flexible domains undergo complex dynamics with entropy and transitory ligand binding as the only driving factors. Thus, rather than triggering a discrete change of state, ligand binding instead alters the entropic neighborhood of conformations naturally explored by the channel. This leads to a large number of continuously varying conformations. Through use of deep learning methods implemented in the cryo-EM field [86–90,94], recent studies of IP₃R1 were able to simultaneously characterize conformational variability and structure on the channel [37], providing a structural mechanism (‘reversible ratchet’) for the susceptibility of IP₃R to binding of IP₃, based on the conformational selection of the ligand-binding pocket, which may adapt different conformations in its unbound state for which IP₃ binds selectively to one of these conformations. From this study, we can see now that instead of directly going from closed to open conformation, the pore-forming elements can adopt a mixture of conformations resulting in different levels of channel activity observed under constant experimental conditions in electrophysiological studies. The dynamics of IP₃R in vivo is extremely complicated given that the channel function is regulated by an array of modulatory molecules, ranging from ions, small chemical compounds and lipids to protein cofactors that shape the amplitude, timing, and duration of IP₃R-mediated Ca²⁺ signals [24,25,95]. Although high-resolution structural knowledge of all three subtypes of IP₃R channel is essential to gain an in-depth mechanistic understanding of IP₃R function, high-resolution structures can be biased by stabilizing measures such as ligands, mutations, leaving open questions as to the native activation process. Use of deep learning-based methods to analyze cryo-EM images can yield a distinct view of conformations that may be refractory by traditional high-resolution single-particle approach. Further systematic work combining analysis of the dynamic conformational landscape of IP₃R with site-directed mutagenesis and electrophysiological characterization will be needed to test connections between protein motions and functional paths underlying transfer of ligand-mediated allosteric information within the channel.

Supplementary Material

2

Supplementary Fig. 1. Summary of IP₃R structures determined by cryo-EM. Cryo-EM structures of IP₃R1 (A) and IP₃R3 (B, C) with PDB IDs, ligand conditions and status of the channel gate listed below.

Supplementary Fig. 2. Schematic depiction of IP₃R cryo-EM structures displayed by their Ca²⁺ conditions. PDB IDs are listed for IP₃R1 and IP₃R3 based on the Ca²⁺ present in the cryo-specimen from low (left) to high (right) Ca²⁺.

Highlights.

• The intrinsically flexible 3D architecture of IP₃R provides the premise behind the multi-modal allosteric regulation of its activity

• Ligand binding alters the entropic neighborhood of conformations naturally explored by IP₃R channels

• IP₃ binding to IP₃R relies on conformational dynamics of the ARM2 domain and is based on the conformational selection of the binding pocket

• Isoform-specific affinity of IP₃Rs might be defined by the characteristic motions of the ARM2 and the correlated flexibility of LBDs

• Binding of IP₃ primes IP₃R to endow it with a capacity to respond to Ca²⁺ which then evokes channel opening

• Binding of Ca²⁺ alters the conformational landscape of IP₃R protein to exert its biphasic functional effect.

• Ligand-evoked signals are transmitted towards the channel pore via allosteric nexus connecting the CY and TM domains