IP₃ Receptors: Toward Understanding Their Activation

Abstract

Inositol 1,4,5-trisphosphate receptors (IP₃R) and their relatives, ryanodine receptors, are the channels that most often mediate Ca²⁺ release from intracellular stores. Their regulation by Ca²⁺ allows them also to propagate cytosolic Ca²⁺ signals regeneratively. This brief review addresses the structural basis of IP₃R activation by IP₃ and Ca²⁺. IP₃ initiates IP₃R activation by promoting Ca²⁺ binding to a stimulatory Ca²⁺-binding site, the identity of which is unresolved. We suggest that interactions of critical phosphate groups in IP₃ with opposite sides of the clam-like IP₃-binding core cause it to close and propagate a conformational change toward the pore via the adjacent N-terminal suppressor domain. The pore, assembled from the last pair of transmembrane domains and the intervening pore loop from each of the four IP₃R subunits, forms a structure in which a luminal selectivity filter and a gate at the cytosolic end of the pore control cation fluxes through the IP₃R.

IP₃ binds to receptors that release calcium from intracellular stores. It causes a conformational change in the receptor, which is transmitted to a pore formed by the last two transmembrane domains from each of its four subunits.

A BRIEF HISTORY OF IP₃ RECEPTORS

Sidney Ringer, in his famous correction to an earlier paper, showed that Ca²⁺ entry can evoke a physiological response by demonstrating that beating of the frog heart requires extracellular Ca²⁺ (Ringer 1883). Almost a century passed before it became clear that this Ca²⁺ entry, via voltage-gated Ca²⁺ channels, was not directly responsible for contraction, but instead provided the trigger for a much larger release of Ca²⁺ from stores within the sarcoplasmic reticulum (SR). The latter is mediated by type-2 ryanodine receptors (RyR) (Fabiato 1983; Cheng et al. 1993), which like many Ca²⁺ channels, are able both to transport Ca²⁺ through an open pore and respond to it. These observations highlight two general points. First, cells call upon two sources of Ca²⁺ to evoke increases in cytosolic Ca²⁺ concentration; second, interactions between these Ca²⁺ fluxes across the plasma membrane and the membranes of intracellular stores are important determinants of the physiological response. The same points apply to the Ca²⁺ signals evoked by receptors that stimulate phospholipase C (PLC) and, thereby, formation of inositol 1,4,5-trisphosphate (IP₃).

The biochemical sequence linking these receptors to formation of IP₃ emerged in the 1980s (Michell et al. 1989; Berridge 2005), but work in the decade before had established that many receptors regulate many different responses by increasing the cytosolic Ca²⁺ concentration (Rasmussen 1970; Berridge 1975). In his influential review, Bob Michell (Michell 1975), building on work showing that many of these receptors also stimulate phospholipid turnover (Hokin and Hokin 1953), had suggested a causal link between phosphoinositide hydrolysis and Ca²⁺ signals. Here, as in many studies, the emphasis was on Ca²⁺ entry, with a consensus only slowly emerging that Ca²⁺ fluxes across both the plasma membrane and the membranes of intracellular stores contribute to cytosolic Ca²⁺ signals (Rasmussen 1970; Berridge 1975; Williams 1980; Putney et al. 1981). In the years following Michell’s review, decisive evidence, much of it coming from Mike Berridge’s elegant studies of blowfly salivary gland, established that phosphoinositide hydrolysis is, as predicted by Michell, required for PLC-linked receptors to evoke Ca²⁺ signals (Berridge and Fain 1979). The same preparation was used to show that IP₃ is the first water-soluble product of the signaling pathway (Berridge 1983). IP₃, thus, emerged as a prime candidate for the cytosolic messenger linking events at the plasma membrane to release of Ca²⁺ from intracellular stores. Paradoxically, it was to be many years before the links between receptors that stimulate PLC and Ca²⁺ entry were resolved. These came with elaboration of the pathways linking empty Ca²⁺ stores to Ca²⁺ entry, the so-called store-operated Ca²⁺ entry pathway (Putney 1997; Park et al. 2009), and recognition that many trp channels are regulated by products of PLC activity (Nilius et al. 2007). IP₃ receptors (IP₃R) also contribute more directly to Ca²⁺ entry across the plasma membrane either because, at least in some cells, IP₃R are functionally expressed in the plasma membrane (Dellis et al. 2006; Dellis et al. 2008), or perhaps through their direct interactions with other plasma membrane Ca²⁺ channels (Kiselyov et al. 1999). Here, we focus solely on Ca²⁺ release from the endoplasmic reticulum (ER) by IP₃R. Some of the key steps in the evolution of our current understanding of IP₃R are listed in Table 1.

Table 1.

Landmarks en route to a structural analysis of IP₃ receptor behavior.

	RyR	IP3R
1883	Ca2+ entry required for heart contraction.1
1953		Acetylcholine stimulates turnover of phospholipids.2
1975		Phosphoinositide hydrolysis proposed to cause Ca2+ signals.3
1977	Ca2+ waves occur at fertilization.4
1977	Ca2+-induced Ca2+ release in SR.5
1979		Phosphoinositide hydrolysis required for receptor-stimulated Ca2+ signals.6
1980	Introduction of Quin 27 and facile loading methods.8
1983		IP3 is first water-soluble product of PLC.9
1983		IP3 stimulates Ca2+ release from a non-mitochondrial store.10
1985	Ryanodine, selective RyR ligand.11
1985	Single channel records of RyR.12
1986		Frequency-coded Ca2+ spikes.13
1987		Ca2+ regulates IP3R.14,15
1987	RyR1 purified.16	IP3R1 purified.17
1988		Single channel records of IP3R.18
1989	Cloning of RyR1.19	Cloning of IP3R1.20,21
1990		Elementary Ca2+-release events.22
1993	Elementary Ca2+-release events.23
2002		Atomic structure of IBC.24
2005		Atomic structure of SD.25
2009	Atomic structure of N-terminal of RyR.26,27

¹Ringer (1883).

²Hokin & Hokin (1953).

³Michell (1975).

⁴Ridgeway et al. (1977).

⁵Endo (1977).

⁶Berridge & Fain (1979).

⁷Tsien (1980).

⁸Tsien (1981).

⁹Berridge (1983).

¹⁰Streb et al. (1983).

¹¹Sutko et al. (1985).

¹²Smith et al. (1985).

¹³Woods et al. (1986).

¹⁴Iino (1987).

¹⁵Iino (1990).

¹⁶Imagawa et al. (1987).

¹⁷Supattapone et al. (1988).

¹⁸Ehrlich & Watras (1988).

¹⁹Takeshima et al. (1989).

²⁰Mignery et al. (1989).

²¹Furuichi et al. (1989).

²²Parker & Ivorra (1990).

²³Cheng et al. (1993).

²⁴Bosanac et al. (2002).

²⁵Bosanac et al. (2005).

²⁶Amador et al. (2009).

²⁷Lobo & Van Petegem (2009).

The role of the SR as the intracellular source of Ca²⁺ signals in striated muscle was long-established (Endo et al. 1970), but there was no such agreement on the identity of the organelle from which Ca²⁺ was released in other cells. Competing claims suggested roles for mitochondria or the ER. Evidence that in resting hepatocytes only the ER contains appreciable amounts of Ca²⁺ (Burgess et al. 1983) was quickly followed by the demonstration that IP₃ evoked Ca²⁺ release from a non-mitochondrial Ca²⁺ store in permeabilized pancreatic acinar cells (Streb et al. 1983). Countless groups quickly replicated these findings in many cells, and within months it was universally accepted that the ER is the major Ca²⁺ store from which IP₃ stimulates Ca²⁺ release in most animal cells (Berridge and Irvine 1984; Berridge and Irvine 1989). Subsequent work has suggested that IP₃ may also stimulate Ca²⁺ release from the Golgi apparatus (Pinton et al. 1998), from within the nucleus (Gerasimenko et al. 1995; Echevarria et al. 2003; Marchenko et al. 2005), and perhaps also from secretory vesicles (Gerasimenko et al. 1996), but ER remains the major IP₃-sensitive Ca²⁺ store. Evidence that IP₃ stimulates Ca²⁺ efflux from the ER (rather than inhibiting Ca²⁺ uptake) and the first single channel recordings (Ehrlich and Watras 1988) established that the IP₃R is an IP₃-gated, Ca²⁺-permeable channel. The first studies of ³²P-IP₃ binding (Spät et al. 1986) were followed by purification of IP₃R from cerebellum (Maeda et al. 1988; Supattapone et al. 1988) and then cloning of the first IP₃R subtype (IP₃R1) (Furuichi et al. 1989; Mignery et al. 1989). Subsequent studies identified two additional genes encoding vertebrate IP₃R (IP₃R2 and IP₃R3) and a single gene in invertebrates (Taylor et al. 1999). It remains far from clear whether plants express related IP₃R (Krinke et al. 2007). These studies established that IP₃R are unusually large proteins, comprising tetramers of closely-related subunits, each with about 2700 amino acid residues. RyR are even larger: they, too, are tetramers, but the subunits are almost twice the size of IP₃R (∼5000 residues). This progress with identifying IP₃R together with single channel recordings of IP₃R, initially in artificial lipid bilayers and later in native membranes (Foskett et al. 2007; Rahman et al. 2009), provided the foundations from which to explore the structural determinants of IP₃R behavior. The advances toward understanding the molecular mechanisms of IP₃R behavior were accompanied by similar progress with RyR (Table 1). Recurrent themes, to which we return, are the similarities between RyR and IP₃R, and the many instances where observations of one channel family have informed further analysis of the other. Very recently, a third family of intracellular Ca²⁺ channels, unrelated to RyR and IP₃R, has been implicated in Ca²⁺ signaling. These are the two-pore channels (TPC) that are activated by NAADP and release Ca²⁺ from acidic Ca²⁺ stores, including lysosomes and endosomes (Patel et al. 2010; Zhu et al. 2010). Several trp (transient receptor protein) channels, in addition to their roles in the plasma membrane, may also mediate release of Ca²⁺ from intracellular stores (Gees et al. 2010).

Parallel to work addressing the workings of IP₃R, there was growing interest in the spatiotemporal complexity of cytosolic Ca²⁺ signals. Ca²⁺ waves were first observed during fertilization. These waves were proposed to result from Ca²⁺-induced Ca²⁺ release (CICR) and were followed by smaller repetitive Ca²⁺ transients (Ridgway et al. 1977; Gilkey 1983). It was, however, the work of Peter Cobbold that focused most attention on the complexity of intracellular Ca²⁺ signals (Woods et al. 1986). Just as the activity of a nerve is conveyed by the frequency of its action potentials, Cobbold demonstrated that in hepatocytes the concentration of the extracellular stimulus determined the frequency of the cytosolic Ca²⁺ transients. As these ideas gathered momentum (Berridge 1995), evidence accumulated in support of cells using the information provided by frequency-encoded Ca²⁺ spikes as an efficient means of regulating cellular activity (Dolmetsch et al. 1997; Li et al. 1998; Berridge et al. 2000; Dupont et al. 2003). The single greatest contributor to progress in understanding the genesis of these intracellular Ca²⁺ signals was the introduction, by Roger Tsien in 1980, of simple, minimally disruptive methods for measuring the free cytosolic Ca²⁺ concentration in intact cells (Tsien 1980; Tsien 1981). These methods, in combination with improved optical microscopy, allowed Ian Parker to begin to resolve the subcellular organization of IP₃-evoked Ca²⁺ signals (Parker and Ivorra 1990; Parker et al. 1996). He showed that as the IP₃ concentration increases, it triggers a hierarchy of elementary Ca²⁺ release events, beginning with the openings of single IP₃R (Ca²⁺ blips), progressing to the coordinated openings of a cluster of several IP₃R (Ca²⁺ puffs) and finally, with sufficient IP₃, culminating in a regenerative Ca²⁺ wave invading the entire cell (Bootman et al. 1997; Demuro and Parker 2007). The demonstration, in 1987 by Masamitsu Iino, that IP₃R are stimulated by cytosolic Ca²⁺ (Iino 1987), and the later widespread recognition that all IP₃R are biphasically regulated by cytosolic Ca²⁺ (Iino 1990; Taylor and Laude 2002), provided what has become the most widely accepted explanation for the recruitment of elementary Ca²⁺-release events. Namely, that CICR, already an established feature of RyR (Endo et al. 1970), allows an active IP₃R to propagate its activity to neighboring IP₃R.

These observations and accumulating evidence that local Ca²⁺ signals can selectively regulate local events (Rizzuto et al. 1993; Berridge et al. 2000; Dyer et al. 2005; Willoughby and Cooper 2007) prompted a re-assessment of the ways in which Ca²⁺ signals convey information. It became untenable to think of responses to graded changes in the intensity of the extracellular stimulus as being simply encoded in graded changes in global cytosolic Ca²⁺ concentration. Ca²⁺ entering the cytosol via one channel can regulate different proteins to Ca²⁺ entering via another (Berridge et al. 2000; Dyer et al. 2005; Willoughby and Cooper 2007). Hence, the spatial organization of the changes in cytosolic Ca²⁺ concentration profoundly affects the physiological response, and that presents many opportunities for delivering different Ca²⁺ signals in response to different stimuli or different stimulus intensities. The duration of each Ca²⁺ increase, whether local or global, is also important in determining not only the amplitude of the response, but also its nature, because Ca²⁺-binding proteins differ in their responses to transient and sustained signals. Finally, the frequency with which Ca²⁺ signals are delivered can determine both the nature and amplitude of the cellular response. The key point is that the versatility of Ca²⁺ as an intracellular messenger capable of regulating diverse cellular events depends largely on the spatiotemporal complexity of cytosolic Ca²⁺ signals (Berridge et al. 2000). If we are to understand how Ca²⁺ functions as a ubiquitous intracellular messenger, we must explain how IP₃-evoked Ca²⁺ signals grow from the opening of a single IP₃R to much larger events. That explanation depends, ultimately, on putting IP₃R into appropriate places within the cell, and on the interactions between IP₃ and Ca²⁺ in regulating the opening of IP₃R. In recent reviews (Taylor et al. 2009a; Taylor et al. 2009b) and original reports, we have described how IP₃R are co-translationally targeted to the ER and then retained there by sequences within their transmembrane domains (TMD) (Parker et al. 2004; Pantazaka and Taylor 2010). We have also suggested that within the ER, IP₃ causes IP₃R to assemble into small clusters within which their regulation by both IP₃ and Ca²⁺ is retuned to facilitate the Ca²⁺-mediated recruitment of IP₃R activity by an active neighbor (Rahman and Taylor 2009; Rahman et al. 2009). Here, we focus entirely on the interactions between Ca²⁺ and IP₃ in regulating IP₃R activity, and the extent to which we can explain those interactions at the structural level.

REGULATION OF IP₃ RECEPTORS BY Ca²⁺ AND IP₃

Activation of IP₃R requires both IP₃ and its permeating ion, Ca²⁺ (Finch et al. 1991; Marchant and Taylor 1997; Adkins and Taylor 1999; Taylor and Laude 2002; Foskett et al. 2007). There are reports of IP₃-independent activation of IP₃R by CaBP1 (Yang et al. 2002), a member of the neuronal Ca²⁺-sensor family, and by Gβγ subunits (Zeng et al. 2003), but the physiological relevance is unclear (Haynes et al. 2004; Nadif Kasri et al. 2004). The current consensus is that binding of IP₃ to the IP₃R is essential for its activation, but whether all four IP₃-binding sites of the tetrameric IP₃R must be occupied is unresolved. Positively cooperative responses to IP₃ in some (Dufour et al. 1997; Marchant and Taylor 1997; Tu et al. 2005a), though not all, studies (Finch et al. 1991; Watras et al. 1991; Laude et al. 2005), and delays before the first response to IP₃ that decrease with increasing IP₃ concentration (Marchant and Taylor 1997), indicate that channel opening requires occupancy of more than one IP₃-binding site. However, gating by IP₃ of heteromeric IP₃R in which at least one subunit is mutated to prevent IP₃ binding suggests that occupancy of fewer than four IP₃-binding sites may be sufficient to cause some channel opening (Boehning and Joseph 2000a). IP₃R subtypes differ in their affinities for IP₃, with the general consensus being that IP₃R2 is more sensitive than IP₃R1, and both are considerably more sensitive than IP₃R3 (Tu et al. 2005b; Iwai et al. 2007). In the cellular context, however, differences in expression level (Dellis et al. 2006; Tovey et al. 2010), subcellular distribution (Petersen et al. 1999), post-transcriptional and post-translational modifications, and association of IP₃R with accessory proteins (Patterson et al. 2004) may be more important determinants of sensitivity.

Soon after the first report of IP₃-evoked Ca²⁺ release, cytosolic Ca²⁺ was shown also to regulate IP₃R (Suematsu et al. 1984; Jean and Klee 1986); thereafter, it emerged that the effects of Ca²⁺ were biphasic, with modest increases in cytosolic Ca²⁺ concentration enhancing responses to IP₃, while higher concentrations were inhibitory (Iino 1987; Iino 1990; Finch et al. 1991; Parys et al. 1992; Marshall and Taylor 1993). This provided yet another parallel with RyR, which are also biphasically regulated by Ca²⁺ (Hamilton 2005). The coregulation of IP₃R by IP₃ and Ca²⁺ in permeabilized cells was confirmed by single-channel recordings of IP₃R1 reconstituted into lipid bilayers (Bezprozvanny et al. 1991; Striggow and Ehrlich 1996; Kaftan et al. 1997; Ramos-Franco et al. 1998a; Ramos-Franco et al. 1998b; Tu et al. 2002; Tu et al. 2005b) and in native nuclear membranes (Stehno-Bittel et al. 1995; Mak et al. 1998; Boehning et al. 2001a; Marchenko et al. 2005). In each case, the single-channel open probability (P_o) of IP₃-activated channels displayed a bell-shaped dependence on cytosolic Ca²⁺ concentration. Evidence that purified IP₃R1 could be stimulated, but not inhibited, by cytosolic Ca²⁺ (Thrower et al. 1998; Michikawa et al. 1999) raised the possibility that Ca²⁺ inhibition might be mediated by an accessory protein, although it has yet to be identified. The same explanation perhaps accounts for some reports, often derived from bilayer recordings, in which Ca²⁺ was suggested not to inhibit IP₃R2 or IP₃R3 (Horne and Meyer 1995; Hagar et al. 1998; Miyakawa et al. 1999; Ramos-Franco et al. 2000). The balance of opinion, supported by numerous studies of all three IP₃R subtypes and using both single-channel and Ca²⁺-efflux studies, is that all three IP₃R subtypes are biphasically regulated by cytosolic Ca²⁺ (Marshall and Taylor 1993; Oancea and Meyer 1996; Dufour et al. 1997; Missiaen et al. 1998; Miyakawa et al. 1999; Swatton et al. 1999; Boehning and Joseph 2000b; Mak et al. 2000; Mak et al. 2001; Tu et al. 2005a). Two independent Ca²⁺-binding sites, which differ in their interactions with different bivalent cations and in their affinities for Ca²⁺, mediate the stimulatory and inhibitory effects of cytosolic Ca²⁺ (Marshall and Taylor 1994; Striggow and Ehrlich 1996; Hajnóczky and Thomas 1997). Both sites are essential elements of many models proposed to explain regenerative Ca²⁺ signals (Lechleiter et al. 1991; Berridge 1997). This core biphasic pattern of regulation by cytosolic Ca²⁺ may be modulated by other intracellular signals (and these, too, may have contributed to some of the disparate findings) and by processing of IP₃R. Ca²⁺-dependent inhibition of IP₃R3, for example, is very sensitive to cytoplasmic ATP (Tu et al. 2005b), and the neuronal S2⁺ splice variant of IP₃R1 has a broader Ca²⁺-dependence than the peripheral S2⁻ form (Tu et al. 2002). However, IP₃ is the major influence on what Ca²⁺ does to IP₃R: The two ligands are essential co-agonists of IP₃R (Finch et al. 1991). Activation of IP₃R1 by Ca²⁺ is positively cooperative, enabling P_o to reach its maximum value over a narrow range of Ca²⁺ concentrations, suggesting that IP₃R1 may be well suited to mediating CICR and regenerative Ca²⁺ signals. Activation of IP₃R3 is less cooperative, occurs over a broader range of Ca²⁺ concentrations, and requires lesser activation, making it well suited as a trigger for Ca²⁺ release as the level of IP₃ increases (Mak et al. 2001; Foskett et al. 2007).

Foskett and colleagues have argued, from their analyses of patch-clamp recordings of nuclear IP₃R, that IP₃ decreases the sensitivity of the IP₃R to inhibition by cytosolic Ca²⁺, and that this alone is the means whereby IP₃ stimulates channel opening (Mak et al. 1998; Mak et al. 2001; Ionescu et al. 2006). This simple explanation, where IP₃ serves only to relieve tonic inhibition by resting Ca²⁺ concentrations, is impossible to reconcile with their observation that pretreatment of cells with Ca²⁺-free media abolishes Ca²⁺ inhibition without preventing IP₃ from activating IP₃R (Mak et al. 2003). This simple model was later elaborated to include at least three different Ca²⁺ sensors (Mak et al. 2003), but at the core of this revised scheme is a single Ca²⁺-binding site that switches from being inhibitory in the absence of IP₃ to stimulatory in its presence (Mak et al. 2003). The essential feature of this scheme is consistent with our initial model, derived from rapid superfusion analysis, which suggests that IP₃ both relieves Ca²⁺ inhibition and promotes binding of Ca²⁺ to a stimulatory site (Marchant and Taylor 1997; Adkins and Taylor 1999). The latter is essential for the channel to open. We, however, argue that the stimulatory and inhibitory Ca²⁺-binding sites are distinct (Marshall and Taylor 1994). We suggest, therefore, that the essential role of IP₃ is to promote Ca²⁺ binding to a stimulatory Ca²⁺-binding site. IP₃, by priming this site, allows Ca²⁺ to provide instantaneous control over whether the channel opens (Fig. 1A).

Figure 1.

Regulation of IP₃R by cytosolic and luminal Ca²⁺. (A) Binding of IP₃ (black circle) to the IP₃R determines whether a stimulatory (green) or inhibitory (red) Ca²⁺-binding site is available (Adkins and Taylor 1999). IP₃ binding causes the stimulatory site to become accessible and the inhibitory site to be concealed; binding of Ca²⁺ (blue circle) to the former then triggers opening of the channel. (B) Luminal Ca²⁺ is proposed to tune the sensitivity of the IP₃R to cytosolic IP₃ and Ca²⁺ such that full stores (i) are most sensitive to IP₃. As the IP₃R opens (ii) and the stores lose Ca²⁺, they are proposed to lose sensitivity to IP₃ until eventually the IP₃R closes, despite the continued presence of the cytosolic stimuli, trapping Ca²⁺ within the ER (iii). Conversely, stores regain their sensitivity to IP₃ as the stores refill, perhaps thereby determining the interval between Ca²⁺ spikes in stimulated cells (Berridge 2007).

The structural basis for Ca²⁺-regulation of IP₃R is unresolved: it may be either direct, via Ca²⁺ binding to a site intrinsic to the IP₃R or via an accessory Ca²⁺-binding protein (Taylor et al. 2004). Stimulation of IP₃R by cytosolic Ca²⁺ is universally observed even with purified IP₃R reconstituted into lipid bilayers (Ferris et al. 1989; Hirota et al. 1995; Michikawa et al. 1999), suggesting that this essential Ca²⁺-binding site probably resides within the primary sequence of the IP₃R. At least seven cytosolic Ca²⁺-binding sites have been identified within IP₃R1 (Sienaert et al. 1996; Sienaert et al. 1997), but the physiological relevance of these sites is unresolved. Two of the sites (residues 304-381 and 378-450) are within the IP₃-binding core, for which there is a high-resolution structure (Bosanac et al. 2002). This structure shows two surface-exposed clusters of acidic residues that overlap with residues in the second N-terminal Ca²⁺-binding region. However, point mutations of several of these acidic residues had no effect on Ca²⁺-regulation of IP₃R (Joseph et al. 2005). The remaining Ca²⁺-binding sites fall within the central region of the IP₃R (Sienaert et al. 1996; Sienaert et al. 1997). The site between residues 1347–1426 is interesting because its proximity to a calmodulin (CaM)-binding region is reminiscent of RyR, which have two CaM-binding regions within ∼200 residues of high-affinity Ca²⁺-binding sites, and a third flanked by two high-affinity Ca²⁺-binding sites (Chen and MacLennan 1994). Interactions between these sites have been proposed to contribute to regulation of RyR by Ca²⁺ and CaM (Chen and MacLennan 1994). None of the Ca²⁺-binding sites within IP₃R contain EF-hands or any other known Ca²⁺-binding motif, and none have obvious sequence similarity with similar regions in RyR. However, each site has clusters of negatively charged residues that may coordinate Ca²⁺ (Sienaert et al. 1997). There is presently no evidence to link any of these sites directly to Ca²⁺ regulation of IP₃R. The only tangible link between specific residues and Ca²⁺ regulation comes from mutagenesis of a glutamate residue that is conserved in all IP₃R and RyR. Mutation of this residue in RyR massively reduced the Ca²⁺ sensitivity of the channel (Chen et al. 1998; Li and Chen 2001). Mutation of the same residue (Glu-2100) to another acidic residue (Asp) caused a ∼5- to 10-fold decrease in the Ca²⁺-sensitivity of the IP₃R to both stimulation and inhibition, abolished oscillatory Ca²⁺ transients in response to agonist stimulation, and reduced the Ca²⁺-binding affinity of a large fragment that includes the residue (Miyakawa et al. 2001; Tu et al. 2003). A rather puzzling aspect of these results is the observation that mutation of a single residue similarly attenuates both stimulation and inhibition by Ca²⁺, when other evidence suggests that the two effects are mediated by distinct sites. This, together with the lack of direct evidence that Ca²⁺ is coordinated by the conserved glutamate, leaves open the possibility that rather than itself contributing to an essential Ca²⁺-binding site, this residue may be allosterically coupled to the site.

Ca²⁺-mediated inhibition of IP₃R is widely assumed to contribute to termination of local cytosolic Ca²⁺ signals, but it remains far from clear whether such inhibition is mediated by Ca²⁺ binding directly to IP₃R or to an associated protein (Taylor and Laude 2002). The effects of Ca²⁺ on IP₃ binding differ between subtypes: It inhibits binding to IP₃R1 (Worley et al. 1987; Supattapone et al. 1988; Joseph et al. 1989; Varney et al. 1990; Richardson and Taylor 1993; Benevolensky et al. 1994; Cardy et al. 1997; Yoneshima et al. 1997), but the effects of Ca²⁺ on IP₃ binding to IP₃R from cells expressing predominantly IP₃R2 or IP₃R3 are confused (Pietri et al. 1990; Mohr et al. 1993; Marshall and Taylor 1994; Cardy et al. 1997; Yoneshima et al. 1997; Lin et al. 2000; Swatton and Taylor 2002). These conflicting results, and evidence that purified IP₃R1 is not inhibited by Ca²⁺ (Danoff et al. 1988; Richardson and Taylor 1993; Benevolensky et al. 1994; Lin et al. 2000), lend some support to the idea that Ca²⁺ inhibition may be mediated by an accessory protein. It is, however, noteworthy that deletion of the suppressor domain (SD, residues 1-223) of IP₃R1, which appears not to include a Ca²⁺-binding site, abolishes inhibition of IP₃ binding by Ca²⁺ (Sienaert et al. 2002). This suggests that effective regulation by an accessory protein might require the SD.

Calmodulin (CaM) is one candidate for the accessory protein through which Ca²⁺ inhibition is exercised (Nadif Kasri et al. 2002; Taylor and Laude 2002). CaM is a ubiquitously expressed, EF-hand containing, Ca²⁺-binding protein that serves as the Ca²⁺-sensor for many cellular events (Gnegy 1993). All IP₃R subtypes are inhibited by Ca²⁺-CaM (Hirota et al. 1999; Michikawa et al. 1999; Missiaen et al. 1999; Adkins et al. 2000; Missiaen et al. 2000), and CaM has been shown to restore Ca²⁺ inhibition to purified IP₃R (Hirota et al. 1999; Michikawa et al. 1999; Nosyreva et al. 2002). Yet, it has proven difficult to relate these functional effects of CaM to either its effects on IP₃ binding or to identified CaM-binding sites within IP₃R. CaM inhibits IP₃ binding to IP₃R1 in a Ca²⁺-independent manner (Patel et al. 1997; Cardy and Taylor 1998), through a site that probably lies within the SD (Adkins et al. 2000; Sienaert et al. 2002). Its properties are clearly inconsistent with the ability of CaM to inhibit IP₃R function only in the presence of Ca²⁺. There is a high-affinity Ca²⁺-CaM-binding site within the central region of IP₃R1 and IP₃R2, but not IP₃R3 (Yamada et al. 1995; Lin et al. 2000). However, mutations that prevented Ca²⁺-CaM binding to this site had no affect on Ca²⁺-dependent inhibition of IP₃R (Zhang and Joseph 2001; Nosyreva et al. 2002). This evidence and the absence of the site from IP₃R3 suggest that the central Ca²⁺-CaM-binding site cannot be responsible for Ca²⁺ inhibition of IP₃R. An additional high-affinity Ca²⁺-CaM-binding site is created in IP₃R1 after removal of the S2 splice region: While this may increase the Ca²⁺-CaM sensitivity of peripheral S2⁻ IP₃R1, it is not a universal candidate for mediating Ca²⁺ inhibition of IP₃R (Islam et al. 1996; Lin et al. 2000). Recently, it was suggested that bound CaM is essential for IP₃R function because a peptide antagonist of CaM inhibited IP₃-evoked Ca²⁺ release (Nadif Kasri et al. 2006). It is now clear that this peptide acts directly on IP₃R, with no requirement for CaM (Sun and Taylor 2008). While this eliminates an essential role for tethered CaM, it raises the intriguing possibility that an endogenous CaM-like structure might be essential for IP₃R activation (Sun and Taylor 2008). In summary, all IP₃R subtypes are inhibited by Ca²⁺-CaM, but the molecular basis of this inhibition has not been established. It seems, on balance, that CaM is unlikely to be the accessory protein through which Ca²⁺ universally inhibits IP₃R. That need not preclude a role for CaM in modulating IP₃R function (Taylor and Laude 2002), just as it does for RyR (Chen et al. 1997; Fruen et al. 2000; Rodney et al. 2001), but we must look elsewhere for the site through which Ca²⁺ inhibits IP₃R.

We turn now to the luminal surface of the IP₃R, where, and again drawing parallels with RyR, we consider regulation by luminal Ca²⁺. Persuasive evidence suggests that Ca²⁺ release by RyR may be terminated before Ca²⁺ stores are entirely depleted because luminal Ca²⁺ is required to maintain RyR activity (Györke and Györke 1998; Launikonis et al. 2006; Jiang et al. 2008), possibly via its interaction with calsequestrin, a luminal high-capacity Ca²⁺-binding protein (Launikonis et al. 2006; Terentyev et al. 2006). A similar scheme has been proposed to account for two features of IP₃-evoked Ca²⁺ release: the initiation of Ca²⁺ release after the quiescent interspike interval during repetitive Ca²⁺ spikes (Berridge 2007) and quantal Ca²⁺ release via IP₃R. The latter describes the situation wherein unidirectional Ca²⁺ efflux from intracellular stores terminates before the stores have fully emptied after stimulation with submaximally effective concentrations of IP₃ without loss of their ability to respond to a further increase in IP₃ concentration (Muallem et al. 1989; Meyer and Stryer 1990; Taylor and Potter 1990; Oldershaw et al. 1991; Bootman et al. 1992; Brown et al. 1992; Combettes et al. 1992; Ferris et al. 1992; Hirota et al. 1995). The proposal is that luminal Ca²⁺ sets the gain on the regulation by cytosolic IP₃ and Ca²⁺, so that as the luminal free Ca²⁺ concentration falls, it causes the sensitivity of the IP₃R to IP₃ to fall until, as Ca²⁺ leaks from the ER, the IP₃R closes despite the continued presence of cytosolic IP₃ and residual Ca²⁺ within the ER (Irvine 1990). Conversely, as stores refill between Ca²⁺ spikes in an intact cell, the model predicts that the sensitivity of the IP₃R increases until it exceeds the threshold at which prevailing cytosolic IP₃ and Ca²⁺ concentrations become sufficient to trigger opening (Fig. 1B). Despite the enduring appeal of the model, evidence that luminal Ca²⁺ directly regulates IP₃R is not yet entirely convincing.

Stores loaded with Ca²⁺ have been shown to become more sensitive to IP₃ in some studies (Missiaen et al. 1992; Nunn and Taylor 1992; Oldershaw and Taylor 1993; Parys et al. 1993; Missiaen et al. 1994; Horne and Meyer 1995; Combettes et al. 1996; Tanimura and Turner 1996), but not in others (Combettes et al. 1992; Shuttleworth 1992; Combettes et al. 1993; van de Put et al. 1994). However, even the supportive results do not eliminate the possibility that the increased sensitivity to IP₃ arises from having Ca²⁺ pass through active IP₃R and increase their sensitivity from the cytosolic surface. Similar difficulties have plagued analyses of the effects of luminal Ca²⁺ on RyR (Tripathy and Meissner 1996; Laver 2007; Laver 2009). In bilayer recordings of IP₃R1, where essential accessory proteins may be lost, luminal Ca²⁺ failed to potentiate the Ca²⁺ release evoked by IP₃ (Bezprozvanny and Ehrlich 1994). Despite the caveats, regulation of IP₃R by luminal Ca²⁺ deserves serious consideration. A high-affinity Ca²⁺-binding site within the luminal loop linking TMD 5 and 6 (Sienaert et al. 1996) contains conserved acidic residues that could mediate luminal Ca²⁺ regulation, although the sub-µM affinity of this site for Ca²⁺ would be ill-suited to detecting likely changes in luminal Ca²⁺ concentration. Luminal accessory proteins, akin to those that regulate RyR, are another possibility, with ERp44 being one candidate. ERp44 belongs to the thioredoxin protein family and regulates IP₃R in a pH- and luminal Ca²⁺-dependent manner (Higo et al. 2005). Binding of ERp44 to the TMD5-6 loop of IP₃R inhibits channel activity, and the interaction is disrupted by high concentrations of Ca²⁺ consistent with the suggestion that luminal Ca²⁺ might enhance IP₃R activity.

To summarize, IP₃ works by tuning the Ca²⁺ sensitivity of the IP₃R: It stimulates Ca²⁺ binding to a stimulatory site and inhibits Ca²⁺ binding to an inhibitory site (Fig. 1A). Binding to the stimulatory site is the trigger for opening of the pore. The identity of neither Ca²⁺-binding site is known: The stimulatory site probably resides within the IP₃R itself, but the inhibitory site may require an accessory protein, though this is unlikely to be CaM. Luminal Ca²⁺ may further tune the sensitivity of the IP₃R to regulation by its cytosolic ligands, but this remains unproven.

STRUCTURAL DETERMINANTS OF IP₃R ACTIVATION

Judged by their primary amino acid sequences, all known IP₃R subunits are assumed to have a similar architecture. Each subunit, of about 2700 residues, comprises three major regions: the N-terminal to which IP₃ binds, the C-terminal region with its six transmembrane regions (TMD) (Galvan et al. 1999), and a large intervening sequence (Fig. 2A). Functional IP₃Rs are tetrameric, assembled either from identical subunits or from mixtures of the three subtypes and their many splice variants (Taylor et al. 1999; Foskett et al. 2007). Several structures of the entire IP₃R1 have been published, each derived from single particle analysis of images from electron microscopy (Hamada and Mikoshiba 2002; Jiang et al. 2002a; da Fonseca et al. 2003; Hamada et al. 2003; Serysheva et al. 2003; Sato et al. 2004). These studies confirm the tetrameric state of IP₃R, but variability between the structures and their relatively low resolution (∼30 Å) have, so far, limited any realistic interpretation of the structural basis of IP₃R activation (Taylor et al. 2004) (Fig. 2B). Whether structures of recombinant IP₃R will contribute to resolving this impasse remains to be seen (Wolfram et al. 2010).

Figure 2.

Major structural domains of IP₃R. (A) The three key regions defined by the primary sequence of a single IP₃R subunit are highlighted: the N-terminal with its SD and IBC, the C-terminal region with its six TMD and pore, and the large central region. Atomic structures of the SD (Bosanac et al. 2005) and IBC with IP₃ bound (Bosanac et al. 2002) are also shown. (B) Two views of the IP₃R derived from single particle analysis (da Fonseca et al. 2003) (top, from the cytosol; bottom, across the ER membrane with the ER lumen at the top). (C) A possible structure of the IP₃R pore, with a luminal selectivity filter and a constriction formed by the tepee-like structure of TMD6. Only two of the four IP₃R subunits are shown.

There has been more progress with RyR, although only recently has the resolution of these structures (∼30 Å) improved on that obtained for IP₃R. These structures of native RyR, and all three subtypes of recombinant RyR reveal a shape like a square mushroom with a very large, open cytoplasmic structure tethered to a much smaller TMD region (the stalk). At ∼30 Å resolution, the structures of the three RyR subtypes are almost indistinguishable, and because they, like the three subtypes of IP₃R, share about 65% sequence identity, it seems reasonable to suppose that the 3D structures of all IP₃R are also likely to be similar to each other. These studies of RyR have identified positions of critical residues within the 3D structure, the sites to which accessory proteins bind, and conformational changes associated with opening of the pore (Orlova et al. 1996; Serysheva et al. 2005; Wang et al. 2007; Jones et al. 2008). Activation of RyR is associated with considerable changes in both the pore and cytoplasmic regions: The four corners of the latter dip down toward the SR, while the central region lifts away from it (Samso et al. 2009). It is noteworthy, in the context of schemes for activation of IP₃R (see below), that large movements of some cytoplasmic domains of RyR1 appear to occur around hinges that link them to relatively immobile domains.

The highest resolution maps (∼10Å), although still insufficient to map 3D structure to primary sequence, have come close to defining the likely secondary structure of the pore of RyR1 (Ludtke et al. 2005; Samso et al. 2005; Samso et al. 2009). This region appears to have six α-helices (Samso et al. 2009), consistent with models of RyR that suggest six TMD (Meur et al. 2007). Along the central axis, it has a luminal constriction (probably the selectivity filter, see below) and a tepee-like assembly of four inner helices (likely to be TMD6), with the apex pointing into the cytoplasmic structure. By analogy with MthK channels, this constriction may form the gate of the RyR. Kinking of the inner helix around a central Gly residue causes splitting of the tepee and thereby opening of the channel for MthK (Jiang et al. 2002b). One structure (Samso et al. 2009) is consistent with a similar mechanism operating for RyR1, but another structure (Ludtke et al. 2005) and mutagenesis of the critical Gly (G4863 in RyR1) (Wang et al. 2003) contradict it. These insights into the possible workings of the RyR pore are significant for IP₃R, because it is within the pore region (TMD5-6) that RyR and IP₃R share the greatest sequence similarity. We turn, therefore, to the pore of the IP₃R to explore its properties and structure.

All IP₃R (like all RyR) are cation channels with extremely large conductance, but only modest selectivity for Ca²⁺ over monovalent cations (permeability ratio, P_Ca/P_K ∼ 6) (Williams et al. 2001; Foskett et al. 2007). The voltage-gated and store-operated Ca²⁺ channels that mediate Ca²⁺ entry across the plasma membrane are vastly more selective (P_Ca/P_K > 1000). In the ER, where most IP₃Rs are located, this lack of selectivity is unlikely to be a problem because Ca²⁺ is probably the only cation with an appreciable electrochemical gradient across the ER membrane. In effect, the ER Ca²⁺ pump (SERCA), by creating a steep Ca²⁺ concentration gradient across the ER membrane, assumes responsibility for determining which cations flow through an open IP₃R. Indeed, the K⁺ permeability of IP₃R and RyR may facilitate rapid Ca²⁺ release by allowing K⁺ to move into the ER to electrically compensate the efflux of Ca²⁺ (Gillespie and Fill 2008). The pore of the IP₃R, like that of RyR, is formed by the final pair of TMD (TMD5-6) and the luminal loop that links them from each of the four subunits (Ramos-Franco et al. 1999; Williams et al. 2001) (Fig. 2C). The loop includes a sequence (GGVGD in IP₃R) similar to that of the selectivity filter of K⁺ channels (Balshaw et al. 1999), consistent with the idea that the overall architecture of the pore region may be broadly similar to that of K⁺ channels (MacKinnon 2004). For both IP₃R and RyR, however, the pore must be larger and less-selective than for K⁺ channels, and probably able to accommodate only one cation at a time (Williams et al. 2001). This model for the IP₃R pore, where TMD5 (the outer helix) and TMD6 (inner helix) cradle a short pore helix and selectivity filter (Fig. 2C), is consistent with mutagenesis of residues within this region affecting ion permeation (Boehning et al. 2001b; Dellis et al. 2006; Dellis et al. 2008; Schug et al. 2008), with biophysical evidence that the narrowest region of the pore lies close to the luminal entrance of the RyR (Williams et al. 2001) and with the intermediate resolution structures of the pore region of RyR1 (Samso et al. 2009). A conserved acidic residue (D2550 in IP₃R1) at the luminal end of the selectivity filter (Fig. 2C) contributes to the modest Ca²⁺ selectivity of IP₃R (Boehning et al. 2001b; Dellis et al. 2008) and RyR (Gao et al. 2000; Wang et al. 2005; Gillespie 2008), but the structural determinants of ion selectivity and permeation by IP₃R are otherwise poorly understood. The changes in pore structure that allow it to open are also minimally understood. Indeed, mutation of the conserved Gly within TMD6 of IP₃R (G2586 in IP₃R1), which might have been thought to provide the gating hinge (Samso et al. 2009), appears not to prevent IP₃ from opening IP₃R (Schug et al. 2008). In short, aside from knowing the regions of primary sequence that form the IP₃R pore (TMD5-6) and a rather vague notion that its structure perhaps resembles that of K⁺ channels, we have only the most rudimentary knowledge of the structural determinants of how the IP₃R pore opens and selects between ions.

The conformational changes in the IP₃R that lead to opening of its pore are initiated by IP₃ binding to the IP₃-binding core (IBC, residues 224–604 in IP₃R1) (Fig. 3A). Although IP₃ is the only endogenous ligand of the IBC, there are many synthetic agonists, all of which have structures equivalent to the equatorial 6-hydroxyl and the 4- and 5-phosphate groups of IP₃ (Fig. 3A) (Rossi et al. 2010). It is noteworthy that neither of the immediate products of IP₃ metabolism, IP₂ and IP₄, binds to the IBC; both metabolic pathways are therefore effective means of terminating activation of IP₃R by IP₃. An atomic structure of the IBC with IP₃ bound (Bosanac et al. 2002) shows IP₃ held within a clam-like structure in which the phosphate groups of IP₃ are coordinated by basic residues (Fig. 3A). The two sides of the clam, the α- and β-domains, form a network of interactions with the essential groups of IP₃. The 4-phosphate is hydrogen-bonded with residues in the β-domain, the 5-phosphate forms hydrogen bonds with residues predominantly in the α-domain, and the 6-hydroxyl interacts with the backbone of a residue within the α-domain. It is easy to imagine how these interactions with IP₃ might pull the α- and β-domains together, causing the clam to close in a manner similar to glutamate binding to ionotropic glutamate receptors (Mayer 2006). Structures of the IBC without IP₃ bound are urgently needed to assess this proposal, but two lines of evidence lend circumstantial support. First, the IBC adopts a more constrained structure when it binds IP₃ (Chan et al. 2007). Second, adenophostins, which are high-affinity agonists of IP₃R (Rossi et al. 2010), retain some activity after loss of the 3-phosphate (analogous to the 5-phosphate of IP₃), probably because their adenine moiety interacts strongly with a residue in the α-domain and thereby partially mimics the clam-closure that would otherwise require the 5-phosphate to bind to the α-domain (Sureshan et al. 2009). We envisage, therefore, that when IP₃ binds to the IBC, the essential vicinal phosphate groups through their contacts with the α- and β-domains effectively cross-bridge the two sides of the clam-like structure, causing it to close, and thereby initiate the processes that will culminate in opening of the pore.

Figure 3.

Initiation of IP₃R activation by IP₃. (A) The structure of IP₃, with its critical vicinal 4,5-bisphosphate and 6-hydroxyl groups, is shown alongside the structure of the IBC with IP₃ bound. The latter shows the 4- and 5-phosphates contacting the β- and α-domains, respectively (Bosanac et al. 2002), and thereby pulling the clam into a more closed state. (B) Structure of the SD (Bosanac et al. 2005) showing possible sites of interaction with the IBC and downstream domains through which it signals to the pore. See text for further details.

It is worth commenting briefly on available antagonists of IP₃R because of their obvious value as experimental tools. There are no specific antagonists of IP₃R, although with appropriate caution some antagonists can yield useful insight (Michelangeli et al. 1995). Heparin is a competitive antagonist of IP₃ (Worley et al. 1987), although it is not membrane-permeant and, among many additional effects, it uncouples G-protein-coupled receptors from their G proteins (Dasso and Taylor 1991) and activates RyR (Ehrlich et al. 1994). 2-aminoethyl diphenylboronate (2-APB) is membrane-permeant and inhibits IP₃-evoked Ca²⁺ release without affecting IP₃ binding (Maruyama et al. 1997); its mechanism of action is unresolved. However, 2-APB also inhibits Ca²⁺ uptake and many other Ca²⁺ channels. It has recently aroused interest as a modulator of STIM and, therefore, store-operated Ca²⁺ entry (Goto et al. 2010). A screen of 2-APB analogues with selectivity for store-operated Ca²⁺ entry may yet also provide IP₃R-selective antagonists (Goto et al. 2010). Xestospongins, isolated from an Australian sponge, are high-affinity membrane-permeant inhibitors of IP₃-evoked Ca²⁺ release that do not affect IP₃ binding (Gafni et al. 1997), but they, too, have side effects (Solovyova et al. 2002). High concentrations of caffeine inhibit IP₃-evoked Ca²⁺ release (Parker and Ivorra 1991) without affecting IP₃ binding (Worley et al. 1987), but caffeine also stimulates RyR, inhibits cyclic nucleotide phosphodiesterases, and interferes with many Ca²⁺ indicators. Membrane-permeant peptide antagonists of IP₃R may provide another potential source of selective antagonists (Sun and Taylor 2008).

How IP₃ binding to the IBC leads to binding of Ca²⁺ to the IP₃R, and thereby opening of the pore, remains largely unknown, but it is clear that the suppressor domain (SD, residues 1-223 of IP₃R1), which is connected to the IBC by a flexible linkage (Chan et al. 2007), plays an essential role. The clearest evidence is that IP₃ binds to IP₃R without an SD, but it fails to open the pore (Uchida et al. 2003; Szlufcik et al. 2006). The name of this region derives from the observation that, although the SD itself is unlikely to make any direct contacts with IP₃, its presence decreases the affinity of IP₃R for IP₃ (Uchida et al. 2003). We have interpreted this effect to reflect the use of binding energy from the binding of IP₃ to the IBC to cause a conformational change within the SD. This interpretation gains considerable support from our analysis of partial agonists of the IP₃R (Rossi et al. 2009). The crux of our argument is that the energy provided by agonist binding drives both the conformational changes that lead to receptor activation and tighter binding of the ligand to its receptor. There is, therefore, a playoff between these two claims on the binding energy. Partial agonists, because they less effectively activate the receptor, divert more binding energy into stabilizing the binding, while full agonists evoke more substantial conformational changes; therefore, less binding energy remains to stabilize binding. Our results show that although full and partial agonists bind with similar affinities to the IBC, the SD causes the affinity of full agonists to decrease more than for partial agonists (Rossi et al. 2009). Quantitative analyses of these results lead to the conclusion that the most energetically costly conformational change in the IP₃R evoked by IP₃ occurs within its N-terminal (residues 1-604), and that these conformational changes pass entirely via the SD to the pore region (Rossi et al. 2009). We suggest, therefore, that the SD is the essential link between IP₃ binding to the IBC and the subsequent conformational changes that lead to opening of the pore. Without a structure of the entire N-terminal of the IP₃R, we can only speculate on the physical relationship between the IBC and SD, but our results with partial agonists and mutagenesis are consistent with three exposed loops of the SD (β2–β3, β5–β6, and β7–β8, blue in Fig. 3B) being the most likely sites of interaction with the IBC (Rossi et al. 2009).

Remarkably, and despite their rather low sequence identities (∼30%), the crystal structures of the SD from IP₃R1 (Bosanac et al. 2005) and of the analogous N-terminal regions from RyR1 and RyR2 (Amador et al. 2009; Lobo and Van Petegem 2009) are extremely similar. Several mutations associated with malignant hyperthermia and central core disease (RyR1) or catecholaminergic polymorphic ventricular tachycardia (RyR2), all of which impair the normal regulation of gating, are clustered in an exposed loop (β8–β9) of the N-terminal of RyR (Amador et al. 2009). Furthermore, and consistent with the N-terminal of the RyR mediating essential interdomain interactions, a peptide derived from this region causes RyR2 to open spontaneously, apparently by uncoupling an interaction between the endogenous loop and a central region of the RyR that includes residues 2460–2495 (Oda et al. 2005; Tateishi et al. 2009). In light of the conservation of structure between IP₃R and RyR, it is tempting to speculate that the same loop in the SD of the IP₃R (β8–β9, red in Fig. 3B) may mediate transfer of conformational changes onward toward the pore. Co-immunoprecipitation studies have suggested an interaction between the N-terminal of IP₃R1 (most likely the SD) and the pore region of an adjacent subunit (Boehning and Joseph 2000a), perhaps mediated by the cytosolic loop linking TMD4 to TMD5 (Schug and Joseph 2006). An attractive possibility, therefore, is that the SD (perhaps its β8-β9 loop) interacts directly with the short cytosolic helix linking TMD4 and TMD5, and thereby gates the pore (Schug and Joseph 2006; Rossi et al. 2009). Such an interaction would require that the SD comes very close to the pore, but the exact location of the SD within the 3D structure of either the IP₃R or RyR is unknown. The N-terminal of the RyR probably lies within the clamp region at the periphery of the large square cytoplasmic structure (Wang et al. 2007), and it does change shape during RyR activation (Samso et al. 2009). Yet, in this location the N-terminal is too far from the pore to interact directly with the TMD4-5 loop, consistent perhaps with evidence that in RyR the N-terminal may interact directly with a neighboring domain that includes residues from the central part of the primary sequence (Wang et al. 2007). These observations and the evidence that the uncoupling peptide derived from the N-terminal of RyR2 appears to interact with residues remote from the pore (Oda et al. 2005; Tateishi et al. 2009), suggest that the links between the SD and pore may, at least for RyR, be indirect.

In summary, we suggest that IP₃R activation is initiated when IP₃ binds to the IBC, and perhaps thereby causes closure of its clam-like structure. That conformational change, which must also initiate the events that allow Ca²⁺ to bind to a stimulatory site, is passed to the rest of the IP₃R entirely via the SD. The location of that Ca²⁺-binding site and, therefore, the structural links between it and the SD, are unresolved. We speculate that one face of the SD interacts directly with the IBC, and the opposite face interacts with the structure through which conformational changes pass to the pore. The pore is a relatively nonselective, large-conductance cation channel formed by the tetrameric assembly of the TMD5-6 regions of each subunit. Its structure is unresolved but likely to be broadly similar to K⁺ channels with a selectivity filter and gate at opposite ends of its membrane-spanning structure.