Linking Structure to Function: Recent Lessons from Inositol 1,4,5-trisphosphate Receptor Mutagenesis

Abstract

Great insight has been gained into the structure and function of the Inositol 1,4,5 trisphosphate receptor (InsP₃R) by studies employing mutagenesis of the cDNA encoding the receptor. Notably, early studies using this approach defined the key constituents required for InsP₃ binding in the N-terminus and the membrane spanning regions in the C-terminal domain responsible for channel formation, targeting and function. In this article we evaluate recent studies which have used a similar approach to investigate key residues underlying the in vivo modulation by select regulatory factors. In addition, we review studies defining the structural requirements in the channel domain which comprise the conduction pathway and are suggested to be involved in the gating of the channel.

In the 20 years following the cloning of the type-1 inositol 1,4,5-trisphosphate receptor (InsP₃R-1) [1] analysis of the predicted amino acid sequence has proceeded with studies of mutant InsP₃R proteins in an continuing effort to link structure to function. Indeed, the initial sequence analysis of the InsP₃R-1 noted several salient features of the protein; a highly positively charged region in the N-terminus, potentially constituting requirements for the putative InsP₃ binding region; two Protein Kinase A (PKA) consensus phosphorylation sites in a predicted central cytoplasmic domain and a hydrophobic region in the C-terminus, indicating a putative transmembrane domain. Strong evidence for the latter prediction was that a deletion mutant lacking the C-terminal 534 amino acid residues was, in contrast to the membrane associated native protein, largely soluble [1]. Additional truncation mutations proved invaluable in assigning function to the N- and C- terminal regions of the protein. In particular, deletion of predicted transmembrane domains in the C-terminus resulted in a soluble, monomeric protein confirming that the transmembrane domains were involved in tetrameric assembly and endoplasmic reticulum localization [2]. This study also demonstrated that expression of the N-terminal 788 residues of the protein retained InsP₃ binding while further truncation to residue 519 eliminated binding, indicating the boundary for the InsP₃ binding domain in the N-terminus of the protein [2].

These initial findings, establishing the location of the binding site and channel have subsequently been greatly extended. Notably, using an extensive collection of glutathione S-transferase (GST) N-terminal receptor constructs and exhaustive mutagenesis, Mikoshiba and colleagues defined the binding “core” of the receptor between residues 223-608. Within this domain, the key positively charged amino acids, which coordinate the negative phosphate groups of InsP₃ to mediate high-affinity specific binding were also elucidated [3, 4]. Notably, 9 of the 12 Arginine or Lysine residues identified in these studies were later confirmed to coordinate the binding of InsP₃ in a crystal structure of the InsP₃R binding core with InsP₃ [5]. Cloning of two further family members revealed that the proteins shared high sequence similarity at the N- and C-termini coupled by a substantial intervening segment with far greater sequence diversity. This region, since termed the regulatory and coupling domain, harbors putative consensus motifs for regulation by a dizzying array of potential modulatory input, including phosphorylation, by cytoplasmic factors such as ATP and by protein binding partners (For review see [6-8]). By virtue of harboring differing subtype specific motifs, the regulatory and coupling domain has been suggested to contribute to the selective regulation of Ca²⁺ release through individual InsP₃R family members.

Despite a wealth of information regarding some aspects of the molecular requirements of InsP₃R function, as noted for example InsP₃ binding, there is a relative lack of data addressing the determinants of other features of InsP₃R structure and function. In particular, the residues in the C-terminal channel domain contributing to the conduction pathway, to the gating of the channel and those responsible for coupling binding in the N-terminus to opening of the pore are not established. Also neglected are studies unequivocally establishing that putative structural motifs are indeed directly responsible for the in vivo modulation of InsP₃R activity. This latter issue, when investigated, has provided some surprising results. In this article, we review studies which have primarily employed InsP₃R mutagenesis to gain insight into some of these unresolved issues.

The InsP₃R modulatory/coupling domain-phosphorylation of InsP₃R-1 by PKA

The presence of PKA consensus phosphorylation sites noted in the original reports, were consistent with the observation that the InsP₃ binding protein expressed in cerebellar purkinje neurons was a known prominent substrate for PKA [9]. Indeed, in retrospect, the InsP₃R-1 was known to be a phosphoprotein prior to its function as Ca²⁺ release channel being established having been identified in an early proteomic screen of brain phosphoproteins [10, 11]. In general, confirmation that a putative regulatory site has relevance for modulating the activity of the protein in vivo is a multi step process. In the case of a phosphorylation event, the protein must be shown to be phosphorylated and this observation correlated with an effect on protein function. Ideally, these two events should then be conclusively linked by abrogating the effect on activity by mutation of the phosphor-acceptor amino acid. In terms of PKA phosphorylation, early experiments physically demonstrated that InsP₃R-1 was stoichiometrically phosphorylated at two putative phosphorylation sites (S1588 and S1755 in the mouse protein). These residues are present in the canonical PKA substrate motif (RRXS/T; R=basic residue) [12, 13] (Figure 1A/B). Moreover, it was suggested that the major peripherally expressed splice variant of the InsP₃R-1, the so-called short form (S2^- InsP₃R-1), in which 40 amino acids (Figure 1B; 1692-1732) are excised from the protein between the two phosphorylation sites, was more readily phosphorylated in vitro at S1588 than the predominately neuronal S2⁺ InsP₃R harboring the 40 amino acid insert which was phosphorylated at both sites [12].

Figure 1. Consensus PKA phosphorylation substrate motifs and “Walker type A” ATP binding motifs in the InsP₃R isoforms.

The general domain structure of InsP₃R family members is shown for S2⁺ InsP₃R1 in A; S2^- InsP₃R-1 in B; InsP₃R-2 in C and InsP₃R-3 in D. In the schematic for each receptor PKA phosphorylation sites are shown above and putative ATP binding domains are shown below. Residues colored red indicate amino acids which when mutated have functional consequences. Residues colored in green indicate amino acids which when mutated did not result in effects on InsP₃R activity. Residues within the putative ATP binding domains underlined in red indicate residues which abrogate binding of ATP to GST-fusion proteins containing the particular site. In InsP₃R-1 (A and B) the PKA phosphorylation sites Serine 1589 and Serine 1755 are separated by the S2 splice region shown in blue (amino acids 1692-1732) in the top panel. In B the blue underlined residues indicate the N and C terminal boundaries of the S2 splice region. The site of the in frame deletion present in the Opthisthotonos mouse (OPT; amino acids 1732-1839) which results in deletion of a phosphorylation site and the ATPA site is indicated. Below the schematic is a diagram of the fragments of InsP₃R-1 produced by limited digestion with trypsin [46, 126, 127]. ATP binding was reported in the C-terminal 40 kDa and 91 kDa fragments [46]. In B, PKA sites and the complement of ATP binding motifs are shown in the S2^- InsP₃R-1. The amino acid numbering is based on the rat S2⁺ InsP₃R-1. In C, a functionally important PKA phosphorylation site at Serine 937 is shown. The ATPB Walker motif is also functionally relevant in InsP₃R-2. Below the schematic a diagram is shown which illustrates a subcloning strategy used to determine phosphorylation sites in InsP₃R-2. The fragments of InsP₃R2 were based on tryptic sites in InsP₃R-1. PKA phosphorylation of the 76 kDa fragment was demonstrated [45]. In D, phosphorylation sites Serine 916, Serine 934 and Serine 1832 are shown, together with the ATPB Walker motif. Mutation of each of these sites appears to be without functional consequence. Below the schematic is a diagram depicting fragments produced by limited chymotryptic digestion of InsP₃R-3 [43, 46]. ATP binding was demonstrated in the C-terminal 95 kDa fragment [46]. PKA dependent phosphorylation was shown in the N terminal 105 kDa fragment; the 70 kDa and 35 kDa fragments [43].

Numerous subsequent studies addressed the functional effects of PKA phosphorylation and resulted in contradictory findings; both potentiated and inhibitory effects on InsP₃R-1 activity where reported for the S2⁺ InsP₃R-1, while mostly inhibitory effects were reported in experiments were the S2^- InsP₃R-1 was presumed to be expressed [14-18]. These studies highlight the general problem of linking a regulatory event to the functional consequence; the activity of the InsP₃R was measured indirectly as a net Ca²⁺ flux and secondly, mutagenesis of the putative site was not performed to unequivocally prove that modification was linked to the change in activity. Obviously both these issues were not trivial given the contemporary tools available, in particular, the lack of a suitable InsP₃R null expression system to functionally test the effect of mutations in the receptor. A major advance to the field in this respect has undoubtedly been the creation of the DT40 3KO cell line [19-21], a unique, genetically tractable tool which has allowed unambiguous measurement of InsP₃R function on a null background.

Even given this substantial biochemical and functional data, mutational analyses have provided unexpected insights into the effects of phosphorylation on InsP₃R-1. A common approach has been to express and compare WT and mutant InsP₃R in the DT40-3KO background. In particular these studies have unambiguously assigned the effects of phosphorylation to particular amino acids [22]. Analyses of alanine substitutions at S1589 and S1755 in the rat receptor established that, despite both sites being phosphorylated in vitro, phosphorylation of S1755 was necessary and sufficient for markedly increasing the activity of the neuronal S2⁺ splice variant. In contrast, phosphorylation of either S1589 or S1755 could augment the activity of the peripheral S2^- isoform. Further insights into the mechanisms by which PKA enhances InsP₃R-1 activity were obtained by using “phosphomimetic” glutamate substitutions at S1589 and S1755 [23]. The utility of these mutations is that any effects on activity reflect the inherent properties of the channel and are therefore independent of any cell type specific events such as targeting of PKA through anchoring proteins [24]. The glutamate substitutions exhibited increased sensitivity following agonist stimulation and direct activation by InsP₃ uncaging. Once again, the S1755E mutation was sufficient for this effect in the S2⁺ isoform (S1589E retained activity similar to WT) and both S1589E and S1755E were required for the maximal increase in sensitivity in the context of the S2^- splice variant.

Another unique attribute of the DT40-3KO cell is that direct electrophysiological measurements of InsP₃R activity following expression of defined homotetrameric channels can be made in native membranes. These measurements can be performed in the “on nucleus” configuration pioneered by Foskett and colleagues [25-29] or by whole-cell measurements of channels expressed in the plasma membrane as detailed by Taylor and co-workers [30-32]. Mechanistic information regarding the enhancing effects of phosphorylation was provided by single channel measurements of plasma membrane InsP₃R-1 in stable DT40 cell lines. This technique, allows for direct channel measurements from a small numbers of channels in the plasma membrane (typically less than 4) while in the whole-cell patch clamp configuration. Similar to Ca²⁺ imaging experiments and lipid bilayer channels measurements [16], the primary effect of the phosphomimetic mutations, like actual phosphorylation, was to cause an increase in P_o at sub-saturating [InsP₃]. Interestingly, there was no apparent effect on InsP₃ binding or the biphasic effects of Ca²⁺ on the receptor. There were however, clear effects on the kinetics of channel opening and closing such that the phosphorylated channel predominately gated in a “bursting mode” characterized by extended periods of rapid channel openings and closings [33].

The InsP₃R modulatory/coupling domain-phosphorylation of InsP₃R-1 by PKG

The identical PKA-phosphorylation sites are also potential targets of cGMP-dependent Protein Kinase (PKG) [34-36]. Despite phosphorylation of S1755 being theoretically a preferred site of PKG due to the presence of an upstream aromatic residue (F1759), S1588 has been reported to be the predominant target of PKG in S2⁺ InsP₃R-1 [36]. Again, there is little consensus in the literature as to the site responsible for the functional effects of PKG phosphorylation. Interestingly a similar mutational analysis indicated that the splicing of the InsP₃R-1 also appeared to determine the susceptibility of phosphorylation by PKG. Activating PKG resulted in augmented Ca²⁺ signals in S2⁺ InsP₃R-1 expressing cells, an effect attributed to phosphorylation of S1755. Surprisingly, PKG had no effect in cells expressing S2^- InsP₃R-1 [23]. The parsimonious explanation for these data, given that phosphate incorporation at either site in the S2^- InsP₃R-1 (following PKA activation) results in enhanced activity, is that the particular conformation of the S2^- InsP₃R renders the protein a poor substrate for PKG. These data also indicate that the reported functional effects of PKG activation in tissues such as smooth muscle and megakaryocytes [37] which express S2^- InsP₃R-1 are likely to be the consequence of phosphorylation of other InsP₃R family members co-expressed. Alternatively, these effects could be mediated indirectly via an accessory protein such as InsP₃R-associated cGMP kinase substrate (IRAG) [38].

The InsP₃R modulatory/coupling domain- PKA phosphorylation of InsP₃R-2 and InsP₃R-3

Functional effects of PKA phosphorylation have also been attributed to phosphorylation of InsP₃R-2 and InsP₃R-3 [18, 39-42]. It should be noted however that neither receptor is stoichiometrically phosphorylated following PKA activation [18, 43]. For example, in AR4-2J pancreatoma cells, parotid acinar cells and hepatocytes expressing a predominance of InsP₃R-2, an increase in receptor activity has been demonstrated following PKA stimulation [18, 44]. In contrast, in cells expressing a majority of InsP₃R-3, such as pancreatic acinar cells and Rin M5F insulinoma cells, both an inhibitory and enhancing effect have been reported [18, 41, 42]. It should be noted that without exception, all these studies have again relied on indirect measurements of InsP₃R activity, further confounded by the fact that all the cells studied express multiple InsP₃R family members to varying extents. Moreover, although PKA consensus motifs are conserved in InsP₃R-1 of all species with genomes encoding multiple InsP₃R proteins, these consensus sites are absent in InsP₃R-2 and InsP₃R-3 protein (Figure 1C and D). Here the problem of conclusive linkage between the site responsible for cause and effect has lagged behind the determination of actual sites phosphorylated in the individual receptors.

The InsP₃R modulatory/coupling domain- PKA phosphorylation of InsP₃R-2

Scores of serine and threonine residues are present in minimal PKA consensus sequences (RXXS/T or RRXS/T) within the cytosolic domains of InsP₃R-2 making a strictly mutagenesis approach somewhat unattractive. To circumvent this issue, a subcloning approach was taken to identify phosphorylation sites in InsP₃R-2 [45]. This strategy was based on the fact that limited trypsin digestion of the InsP₃R-1 or InsP₃R-3 attacks solvent exposed sites and results in 5 (InsP₃R-1) or 4 (InsP₃R-3) globular domains [3, 43, 46]. In a similar manner to maintain tertiary structure, the sub-clones were designed around 5 predicted tryptic digest sites in InsP₃R-2. Only fragments 3 and 5, corresponding to residues 920-1583 and 1884-2701 were phosphorylated in vitro. Mutation of S937A (among 5 serine residues) eliminated phosphorylation in fragment 3 while mutation of S2633A abrogated the effect in fragment 5, identifying these residues as potential PKA phosphorylation sites in InsP₃R-2. S2633 is conserved in all InsP₃R subtypes and has in fact been demonstrated to be phosphorylated by Akt kinase in vivo [47, 48]. Because mutation of established PKA sites in both InsP₃R-1 and InsP₃R-3 abolished all the effects of PKA activation, S2633 is unlikely to represent a PKA phosphorylation site in the intact protein. In contrast, S937 is unique to the InsP₃R-2 (Figure 1C) and is conserved in all InsP₃R-2 genes in the NCBI database. Of interest, S937 was also independently identified as a phosphorylated residue in a proteomic screen of liver phosphoproteins [49]. Subsequent experiments demonstrated that an antibody raised against phosphorylated S937 recognized a protein product of appropriate size in forskolin treated COS-7 cells expressing InsP₃R-2 but not S937A InsP₃R-2 and importantly, native receptor in forskolin treated parotid acinar cells. Finally, PKA activation resulted in enhanced Ca²⁺ signals and single InsP₃R-2 channel activity in DT40-3KO cells expressing InsP₃R-2 but the effect was absent in S937A InsP₃R-2 expressing cells [45]. These experiments, taking a deliberate methodological approach establish that S937 is a bone fide, functionally relevant phosphorylation site in InsP₃R-2. Further, work is obviously necessary to establish the physiological role of this regulation in cells which express predominately InsP₃R-2.

The InsP₃R modulatory/coupling domain- PKA phosphorylation of InsP₃R-3

Wojcikiewicz and colleagues have reported that of 33 serines and 18 threonine residues present in minimal PKA consensus motifs of InsP₃R-3, only S–A substitutions in three sites (S916, S934, S1832) reduced phosphorylation of InsP₃R-3 in vivo (Figure 1D) [43]. These sites appear to represent the sole sites in the intact InsP₃R-3 as a construct harboring S-A mutations in each of the three serine residues is not phosphorylated following PKA activation. Interestingly, while S934 in particular was phosphorylated prominently in DT40-3KO cells stably expressing InsP₃R-3, no effect on Ca²⁺ release stimulated by activation of endogenous DT40 cell phospholipase C-coupled receptors was observed in populations of cells [50]. These data again highlight that the physical demonstration of a phosphorylation event does not necessarily correlate with effects on Ca²⁺ release. Further experiments directly measuring channel activity are necessary to rule out that PKA-phosphorylation of InsP₃R-3 results in subtle effects on channel activity not evident in population [Ca²⁺] measurements. An alternative is that phosphorylation of InsP₃R-3 has consequences other than directly effecting Ca²⁺ flux for example, by influencing the complement of interacting proteins/factors and thus altering function as a signaling integrator and adaptor protein.

The InsP₃R modulatory/coupling domain-regulation by adenine nucleotides

A further example where mutagenesis of “established” regulatory sites has yielded unexpected results is the modulation of receptor activity by adenine nucleotides. This form of modulation may be important in a variety of physiological and pathological situations by providing direct coupling between the metabolic status of the cell and Ca²⁺ release. ATP was initially demonstrated to increase Ca²⁺ flux [51] and the InsP₃-gated activity of channels from smooth muscle [52]. Subsequently, it was established that 8-azido-α³²P-labelled ATP bound stoichiometrically to the purified cerebellar InsP₃R at a single binding site [53] and was a potent modulator of InsP₃-induced calcium release and channel activity via the reconstituted receptor [53, 54]. A number of studies documented that sub-millimolar concentrations of ATP (and other adenine nucleotides) increase the activity of the channel, while higher concentrations appear to inhibit [55, 56]. ATP modulates all InsP₃R family members, albeit with InsP₃R specific characteristics [57-59]. For example, in Ca²⁺ release assays, where Ca²⁺ is clamped to ~ basal cytosolic levels, ATP appears to modulate Ca²⁺ release through InsP₃R-1 and InsP₃R-3 at all [InsP₃] while InsP₃R-2 is only subject to modulation at sub-maximal [InsP₃] [59, 60]. In isolation, the individual receptors also differ in their affinities for ATP modulation, with InsP₃R-2 having a ~10 fold higher sensitivity than InsP₃R-3 and InsP₃R-1 exhibiting intermediate affinity [55, 57-60]. Consistent with these data, native cells expressing a majority of InsP₃R-2 are more sensitive to ATP modulation than cells expressing predominately InsP₃R-3 [61]. Mechanistically, the inhibitory effect is thought to occur as high [ATP] inhibit InsP₃ binding to the receptor [55, 56] while the increase in activity occurs as ATP allosterically modulates the Ca²⁺ co-agonist sensitivity of the InsP₃R [27, 28]. The effects of ATP do not require hydrolysis [54-56], but are thought to be mediated by ATP binding to specific sites in the regulatory domains of the receptors [6, 25, 62].

It has been proposed, and widely accepted that ATP binds to glycine rich regions containing the sequence G-X-G-X-X-G (Figure 1). This sequence is reminiscent of Walker type A motifs common in many ATPase and GTPases [6, 25, 62, 63]. The primary sequence of InsP₃R family members contains a number of such motifs (Figure 1A-D). A region termed the ATPB site is conserved in all three family members (aa 2016-2021 in InsP₃R-1), the so-called ATPA site is unique to the InsP₃R-1 (aa 1773-1780; Figure 1 A-D) and a further site termed ATPC is introduced by excision of the S2 splice site forming S2^- InsP₃R-1 (aa 1574-1765; Figure 1B). It should be noted that it has been demonstrated quite convincingly that these sites in isolation bind ATP. For example, GST fusion proteins harboring the particular sites have been shown to bind fluorescent TNP-ATP or α³²P-ATP specifically [59, 64, 65]. Furthermore, mutation of the binding sites by a G-A substitution in the putative ATP binding pocket eliminates binding in this context [59, 65]. Perhaps the most compelling evidence for the relevance of these sites are data which show that 8-azido α³²P-ATP can be cross-linked to recombinant InsP₃R which when subjected to controlled proteolysis is found in fragments which are predicted to contain the glycine rich motifs [46]. Specifically, InsP₃R-1 was shown to have two binding sites in fragments containing ATPA and ATPB (Figure 1A), while InsP₃R-3 bound one ATP molecule in a fragment predicted to harbor ATPB (Figure 1D) [46]. No mutagenesis was carried out in these studies and thus it is formally possible that sites other than GXGXXG motifs are present in these fragments. The presence of two putative binding sites in InsP₃R-1 has also been used to explain the higher functional sensitivity of the InsP₃R-1 to ATP when compared with InsP₃R-3. This explanation is consistent with the reduced functional sensitivity to ATP modulation of the Opisthotonos (opt) mutation in InsP₃R-1 which results in a deletion of 107 amino acids including the ATPA site [66]. This mutation does however result in deletion of an extended stretch of amino acids (removal of G1732-Q1839) and thus a possibility exists that disruption of overall InsP₃R structure in opt alters ATP binding outside the immediate deleted region.

Role of the ATPC site in S2^- InsP₃R-1

Despite the weight of persuasive evidence that these domains bind ATP and that the presence of these sites correlates with regulation of function, a firm causal link is missing in the absence of mutagenesis in the full length receptors. Efforts to address this issue have been made by mutating all recognized ATP binding motifs in each of the InsP₃R family members. The primary motivation for this work was to unequivocally establish that the effects of ATP binding at particular sites indeed leads to regulation of InsP₃R function. Since the creation of the ATPC site in S2^- InsP₃R-1 is the only obvious structural difference between the S2⁺ and S2^- splice variants this alteration could potentially contribute to the previously noted splice specific regulation of these receptors [22, 66, 67]. To confirm nucleotide binding to this putative site, TNP-ATP was shown to specifically interact with GST fusion proteins containing the ATPC site. Importantly, this binding was eliminated by substitution of the central glycine residue for alanine in the binding motif (G1690A). In cells expressing S2^- InsP₃R-1, Ca²⁺ release was shown to be markedly increased by ATP, however the mutation of ATPC in the full length S2^- InsP₃R-1 (ΔATPC S2- InsP₃R-1) did not alter this modulation. Instead, the mutation eliminated PKA phosphorylation and the resultant potentiated Ca²⁺ release [65]. The conclusion from these data was two-fold; that ATP binding to ATPC appears to be necessary for a conformation of the S2^- InsP₃R-1 which is susceptible to PKA phosphorylation and secondly that ATP regulation of Ca²⁺ release per se occurs at an alternative site, presumably the ATPA or ATPB also present in InsP₃R-1.

Role of the ATPB site in InsP₃R-2 and InsP₃R-3

Experiments addressed whether the single conserved ATPB site in InsP₃R-2 and InsP₃R-3 is responsible for ATP regulation of Ca²⁺ release in these receptors [59]. To ensure disruption of the ATP binding pocket each glycine residue in InsP₃R-2 ATPB was replaced with alanine (Figure 1C; G1969A, G1971A and G1974A; termed ΔATPB InsP₃R-2). This construct eliminated ATP regulation of InsP₃-induced Ca²⁺ release. In addition, both the oscillation frequency and amplitude of Ca²⁺ signals following B cell receptor (BcR) activation were reduced in cells expressing ΔATPB InsP₃R-2 when compared to WT InsP₃R-2. These data confirm that the ATPB site in the context of InsP₃R-2 is solely responsible for mediating the effects of ATP binding and that nucleotide binding has functional consequences after stimulation of an endogenous signaling pathway. In contrast, in the InsP₃R-3 a similar mutation in ATPB (Figure 1D; G1920A, G1922A, G1925A; termed ΔATPB InsP₃R-3) regulation of InsP₃-induced Ca²⁺ release and BcR-induced Ca²⁺ signaling were unchanged [59]. These data indicate that despite evidence of ATP binding in a region that contains the ATPB site that the site itself is functionally irrelevant for regulation of Ca²⁺ release via InsP₃R-3. This finding although surprising, nevertheless appears in hindsight, consistent with InsP₃R-2 and InsP₃R-3 having markedly different characteristics of ATP modulation [58, 59].

Role of the ATPA and ATPB site in S2⁺ InsP₃R-1

In a similar fashion to InsP₃R-3, mutation of the InsP₃R-1 to create ΔATPB InsP₃R-1 (G2016A, G2018A, G2021A InsP₃R-1) resulted in no obvious alteration in the sensitivity of Ca²⁺ release to ATP [60]. Given the presumed higher affinity of the ATPA site potentially masking any role of ATPB these data were not unexpected. Remarkably, however when S2⁺ InsP₃R-1 was constructed devoid of any known ATP binding sites (mutation of G1773, G1775A,G1777A, G1780A in ATPA and G2016A, G2018A, G2021A in ATPB; S2⁺ InsP₃R-1 ΔATPA/B) the modulation of Ca²⁺ release by ATP was unaltered. Indeed, at the single channel level, monitored in the “on-nucleus patch” configuration, modulation of Po by ATP, the single channel conductance and the predominant mode of gating of the mutant channel were identical to native S2⁺ InsP₃R-1 [60]. The inescapable conclusion from these data are that the known ATP binding sites in InsP₃R-1, similar to InsP₃R-3, are not responsible for mediating the functional effects of ATP binding on Ca²⁺ release. Given the absence of any further known predicted nucleotide binding sites in the primary sequence, a novel ATP binding sequence must mediated these effects. A possibility is that the interaction is dependent on the tertiary structure of the InsP₃R. Alternatively, a cryptic ATP binding site may be exposed following the large conformational changes the receptor undergoes when binding Ca²⁺ [68-70]. Conceivably, the functional effects of ATP could also be mediated through a tightly bound accessory protein.

Mutagenesis of The InsP₃R channel domain

The original prediction of six transmembrane (TM) domains in the C-terminal segment of the receptor based on hydropathy analysis has been confirmed experimentally [71-73]. Mutagenesis within the C-terminal channel domain (extending from ~amino-acid 2270-2749 of InsP₃R-1) has primarily been carried out to investigate the permeation/gating mechanisms of the channel [74-79], to study the role of the transmembrane domains in oligomerization [72, 80, 81], and to explore potential regulatory protein-protein interactions of the cytosol-exposed C-terminal tail [16, 82-84]. In general, studies on the conduction mechanism of InsP₃R channels have been hampered by the lack of a detailed knowledge of the three dimensional structure of the channel domain, as well as the absence of convenient screening methods to analyze the behavior of large numbers of mutant channels. Hence, studies using mutagenesis approaches on the channel domain are quite limited. Figure 2A indicates the location of point mutations that have been engineered in the InsP₃R-1 pore domain together with their functional consequences. For comparison, the corresponding mutations in ryanodine receptors (RyR) are also shown. The information obtained from these mutations are summarized and discussed below.

Figure 2. The pore and C-terminal tail of InsP₃ receptors.

Panel A shows the sequence of a portion of the pore region of InsP₃Rs. The consensus residues shown in bold are identical in all 3 InsP₃R isoforms and all 3 RyR isoforms although only the sequence of RyR1 is shown in the Figure. Residues which have been mutated in InsP₃R1 are shown together with their functional consequences. The analogous mutations in RyRs are shown. For additional information see text and Ref. (9). The R2596 position is indicated by an asterisk. Panel B A homology model of the InsP₃R based on a KirBac 1.1 template is shown. The image is taken from Ref. (9) and is used with permission. Only the location of selected mutations from Panel A are shown. The linkers between S5 and the pore helix, and between the selectivity filter and S6, were not included in the model. Panel C shows a cartoon of the C-terminal tail with endogenous cysteines indicated by open circles and bars to indicate the approximate location of interacting proteins. CC = coiled coil domain. GIT; G-protein coupled receptor kinase-interacting protein, PP1-a; protein phosphates 1-a, 4.1N; neuron specific isoform of erythrocyte protein band 4.1. For additional details see text.

Permeation and selectivity of InsP₃R channels

A basic feature of InsP₃R and related ryanodine receptor (RyR) channels are high divalent cation conductances (~50pS with 50 mM Ca²⁺), even higher monovalent cation conductances (200-350pS with 140mM K⁺), and only modest discrimination between divalent and monovalent ions (P_Ca²⁺_/K⁺ ~6-8) (reviewed in [25]). It has recently been suggested that the high K⁺ conductance of RyR and InsP₃R may yield an unexpected benefit in permitting these channels to provide their own countercurrent to compensate for the charge movement due to Ca²⁺ release from intracellular stores [85]. In contrast to K⁺ channels [86] and voltage-gated Ca²⁺ channels [87], there is no compelling evidence for multi-ion occupancy of the channel, suggesting that electrostatic repulsion between ions residing in the pore is probably not the mechanism that accounts for the prodigious ion flux in InsP₃R or RyR channels (but cf.[88]). This, and other biophysical properties, have led to the prediction that these channels would have short/wide pores with rings of negatively charged residues at either end to facilitate the rapid delivery and exit of ions through the pore [25, 89].

Our knowledge of InsP₃R pore structure is primarily based on comparisons to the several K⁺ channels for which crystal structures are available. The retention of channel activity in an InsP₃R mutant from which TM 1-4 have been deleted strongly suggests that the pore of InsP₃R is distal to TM 5 [75]. Within the primary sequence of this region a number of structural features common to many other channels can be recognized. These include sequentially: an outer helix (TM 5), a pore helix, a selectivity filter and an inner helix (TM 6). A homology model of the closed InsP₃R pore has been constructed by da Fonseca and Morris using the KirBac1.1 K⁺ channel as template [76] and is shown in Figure 2B. Several of the residues that are completely conserved in all three InsP₃R and RyR isoforms within the pore-helix/ selectivity filter/ TM 6 segment have been mutated to alanine [76]. Not unexpectedly many of these mutations entirely eliminate channel function – presumably, as a result of disrupting critical structures such as the pore-helix or the selectivity filter. In particular, the removal of a negative charge at the luminal mouth of the receptor (D2550A) inactivates the channel. This mutant has been widely used by investigators as a “pore-dead” version of InsP₃R [33, 90-94]. It is of interest that mutation of the analogous aspartate in RyR1 (D4899A) or RyR2 (D4831A) does not completely inactivate the channel [95-97], possibly because there is an adjacent glutamate residue that may fulfill a similar role. In addition, RyR possess a second pair of negative charges one residue away from the first pair, which in InsP₃R are both positively charged residues (Figure 2A). The distribution of negative charges at the luminal entrance is therefore very different between the two channels and the implications of this for ion permeation and selectivity remain to be investigated. Surprisingly, the mutation of positively and negatively charged residues to alanine at the cytosolic exit of the channel has little impact on channel function with the exception of R2596 which is the only charged residue that is conserved in all InsP₃R and RyR isoforms (asterisk in Figure 2A).

The selectivity filter of the InsP₃R channel has the “signature sequence” GGGVG with the valine replaced by an isoleucine in RyR channels. Mutation of the first three glycines to alanine inactivates the InsP₃R channel whereas mutation of the fourth glycine does not [76]. Substitution of the bulky valine for the smaller isoleucine in InsP₃R increased K⁺ conductance to that seen in wild type RyR without changing selectivity [30, 33, 74]. It has been argued, based on modeling and molecular dynamics simulations, that the rather wide pore diameter of RyR (~7Å; [89]) and InsP₃R (6.5Å; [98]) means that the selectivity filter actually exerts little effect on channel selectivity [99]. Instead, the density of negative charges at the luminal entrance has been proposed to play the major role in discriminating monovalent from divalent cations [89, 99, 100]. Some support for this hypothesis is the finding that the conservative substitution D2550E eliminated selectivity without altering K⁺ conductance [74].

Some of the caveats of mutagenesis studies in InsP₃R can be illustrated with specific examples. The positioning of a highly conserved positively charged residue (R2543) at the C-terminal end of the pore helix would not seem conducive to the rapid movement of positively charged ions through the channel. However, mutation of the residue to alanine has little effect in ⁴⁵Ca²⁺ flux assays, suggesting that the side-chain projects away from the permeation path as predicted by the homology model [74, 76]. Surprisingly, the R2543A mutation alters Ca²⁺ regulation of the receptor [74, 76]. That mutations within the pore of the channel can alter Ca²⁺ regulation is not without precedent in RyR studies [97] and could indicate that some point mutations can induce global conformational changes in the protein that have secondary regulatory effects. Another concern relates to the methods used to assay the function of InsP₃R channels. Two glycine mutants in the selectivity filter were shown to be functional (G2549A) or non-functional (G2546A) in ⁴⁵Ca²⁺ flux assays and when anti-IgM was used to stimulate Ca²⁺ release in DT40 cell lines stably expressing the mutants [97]. However, both mutants were equally active when assayed by measuring K⁺ conductance in patch clamp studies of the receptors expressed in plasma membranes of DT40 cells [97]. Thus the effects of mutagenesis on the conduction properties of the channel may differ if Ca²⁺ or K⁺ is monitored as the permeant ion.

Gating mechanism of the channel

The binding of InsP₃ at the N-terminal ligand binding domain (LBD) induces a conformational change in the receptor [101, 102] but how this change is transmitted to the opening of the channel in the C-terminal segment of the protein is unknown. Early studies based on cross-linking and co-immunoprecipitation assays suggested that a 40 kDa N-terminal tryptic fragment on one monomeric subunit can be directly cross-linked to a C-terminal 95 kDa tryptic fragment of an adjacent subunit [103]. This supports the idea of a direct interaction between LBD and channel domains. Based on experiments with fusion proteins and mutagenesis studies of full-length receptors, the site of interaction in the channel domain has been narrowed to the cytosol-exposed loop between TM 4 and TM 5. This conclusion is supported by experiments in which the deletion of two 10 amino-acid segments, which together constitute the majority of the TM4,5-loop, disrupted channel function and the interaction between N and C-terminal tryptic fragments [77]. Point mutations indicate that the C-N interactions may involve polar residues particularly in the proximal portion of the TM4,5 loop [77]. By analogy with other channels [104-106] it has been proposed that the TM4,5 loop is an amphipathic helix lying parallel to the membrane that constricts the cytosolic face of the TM 6 pore-lining helix and therefore maintains the channel in its closed state [77]. The conformational change resulting from binding of InsP₃ leads to the mechanical movement of the TM4,5 loop that allows the TM 6 domains to separate and permits ion flux. A mutant InsP₃R from which the TM 1-4 helices have been removed generates a constitutively open channel [75]. Although the TM 4,5 loop is retained in this mutant, it is unlikely to be oriented correctly to perform its normal function of compressing the TM 6 helices and this probably accounts for the constitutive activity of the mutant.

The N-terminal segments of the receptor that interact with the TM 4,5 loop have not been identified but a possible candidate is the suppressor domain (aa1-223). Deletion of this domain enhances InsP₃ binding to the receptor and eliminates channel function [79]. Interestingly, the TM4,5 loop deletion mutants also display enhanced InsP₃ binding, supporting the view that the loop and the LBD can influence each other [77]. Binding measurements of synthetic InsP₃ ligands to InsP₃R constructs retaining or lacking the suppressor domain have led Taylor & colleagues [107] to propose that the energy from InsP₃ binding is used to re-arrange an intramolecular interaction between the suppressor domain and the inner ligand-binding core. Furthermore, they conclude that coupling to the channel occurs exclusively through the suppressor domain. This is consistent with the finding that mutations in the suppressor domain that diminish the ability of the suppressor domain to inhibit InsP₃ binding also decrease the open probability of the channel [107]. Despite these studies there is no direct experimental evidence that the suppressor domain is the N-terminal interaction site of the TM 4,5 loop. Alternative candidates could include other portions of the LBD, such as the α-helical domain (aa 427-605), which has the ability to gate endogenous InsP₃R when targeted to ER membranes [108]. Uchida et al., reported [79] that deletion of a segment from the regulatory domain (aa 651-1130) interferes with channel function but this mutation also showed a large decrease in InsP₃ binding affinity which could be indicative of gross structural defects in the protein. Many of the mutations that have been used to study gating have involved large deletions of the full-length protein. A point mutation in Drosophila InsP₃R (corresponding to S217F in the rat InsP₃R-1 sequence) shows an enhanced affinity for InsP₃ in Ca²⁺ flux assays without significant changes in InsP₃ binding affinity [78]. This mutation has therefore been characterized as a gain-of-function “gating” mutation. Additional point mutations in the suppressor or regulatory domains that specifically interfere with gating, without affecting InsP₃ binding, remain to be identified.

In common with other voltage-gated channels a major barrier to ion-flux is expected to be located at the constriction formed at the cytosolic end of the pore-lining TM 6 helices (Figure 2B). Identifying the residue(s) at the channel gate is problematic since the exact boundaries of the pore-lining helix have not been established. Many channels have bulky, hydrophobic residues at the gate and mutation of these residues to alanine disrupts channel function [109, 110]. There is a conserved phenylalanine residue at the end of the TM 6 helix that could potentially play such a role. However, the mutation F2592A does not inactivate the channel, suggesting that the gate of InsP₃R may be located elsewhere [76]. Crystal structures of open K⁺ channels suggest that bending of the inner pore-lining helix at a conserved glycine is part of the gating mechanism [109, 111]. A conserved glycine that could potentially function as a “gating hinge” is present in TM 6 but its mutation (G2586A) also does not inactivate the channel [76]. This suggests that if InsP₃R possess a gating hinge it is probably not located at G2586. Nevertheless, the deliberate introduction of a bend in the helix with a G2586P substitution does produce a constitutively open channel [76]. It is possible that the TM 6 helix is inherently flexible and movement at a hinge may not be involved in the gating process, as suggested for the hERG K⁺ channel [112].

Properties of the C-terminal tail

The C-terminal tail consists of ~160 amino-acids projecting into the cytosol from the putative boundary of the pore-lining TM 6 helix. Interestingly, the C-terminal tail of RyR are significantly shorter (~100 amino-acids) suggesting that this portion of the protein may have distinct functions in InsP₃R and RyR. Initial studies indicated that deletion of the terminal 13 amino-acids from the InsP₃R had no effect but deletion of 139 amino-acids eliminated channel function [16, 79]. A study using progressive deletion mutants found that channel function was lost somewhere between removal of 43 and 60 amino-acids. This study also showed that the C-tail contained a coiled-coil domain in the region being deleted and that a fusion protein of the coiled-coil could form tetramers in vitro [77]. Whether this occurs in full-length InsP₃R in vivo is not known but there is evidence that the C-tail does contribute to the stability of the InsP₃R tetramer [80]. Two highly conserved cysteine residues in the tail (C2610, C2613) have been identified whose mutation to serine or alanine inactivates the channel [76, 79]. The proximity of these residues to the terminus of the TM 6 helix has led to the suggestion that these residues may function as “gate keepers” [79]. The exact role played by the cysteines or the coiled-coil domain in channel function has not been established but it is likely that they are critical to maintaining the architecture of the C-terminal tail and providing a normal exit pathway for the conducting ion. In other channels [96, 113], including RyR [114], there is evidence for salt-bridge interactions between charged residues in the TM 4,5 loop and the C-tail. Additional approaches are required to assess whether this occurs in InsP₃R and to determine how such interactions may contribute to the mechanics of the gating process.

The C-tail is also the site of interaction of a number of regulatory proteins (Figure 2C) and the phosphorylation site of Akt kinase (S2618) [47, 48] and Polo kinase (T2656) [115]. The interacting proteins cover a range of sizes and include: cytochrome c (12kDa, [116]), BclXl (26kDa, [84]), PP1α (37kDa, [16], Huntington-associated protein 1A (70kDa,[83]), 80K-H protein kinase C substrate (80kDa, [117]), G-protein coupled receptor kinase-interacting protein (95kDa; [118]) and 4.1N (98kDa; [119]). The localization of the binding sites to specific regions of the C-tail has been carried out using C-tail fusion proteins or yeast 2-hybrid assays. It is generally assumed that these relatively large proteins have free access to the entire length of the C-tail in the intact receptor. However, reaction of cysteine substitution mutants introduced into the C-tail of full length receptors with 5kDa PEG-maleimide suggest that freely accessible residues are limited to the terminal ~35 amino-acids. Using this assay, Ca²⁺-induced conformational changes in the receptor were observed to enhance accessibility of the C-tail[68]. Structural information on the C-tail, particularly its orientation with respect to the membrane and the rest of the receptor, is needed to determine its role in InsP₃R function and regulation.

Properties of the intraluminal loop

The intraluminal loop between TM 5 and the start of the pore-helix is ~62 amino-acids long in InsP₃R compared to only ~14 amino-acids in RyR. Although the sequences show considerable variability between the three InsP₃R isoforms, all three receptors contain 2 highly conserved cysteines in the loop (C2596, C2504 in InsP₃R-1). Mutation of neither cysteine affects channel function [120]. The loop has been proposed to be the binding site for ERp44, an ER luminal protein of the thioredoxin family that inhibits InsP₃R channel function [120]. The binding appears to require both cysteines to be in their reduced state, although other determinants for binding must be involved since association with ERp44 is specific to the InsP₃R-1 isoform [120]. Recent studies have shown that the cysteines can form an intramolecular disulphide bond, at least in fusion proteins encoding the loop [121]. The disulphide bond status of the cysteines in full-length receptors and their role in redox regulation of InsP₃R in intact cells remain to be evaluated.

Concluding Remarks

Clearly, analysis of mutant InsP₃R family members using a combination of Ca²⁺ flux biochemical and electrophysiological techniques has yielded valuable information regarding InsP₃R structure and function. Nevertheless, future molecular modeling based on more detailed knowledge of the 3 dimensional structure coupled with functional analysis of mutant receptors will undoubtedly be pivotal in addressing important remaining questions. Again, a fruitful approach may be to use related ion channels and shared protein binding partners as a framework for design of constructs. Obvious subjects to address include the sites of regulation of InsP₃R by Ca²⁺. Mutation of E2100 in InsP₃R and corresponding conserved residues in RyR, significantly alters Ca²⁺ sensitivity of Ca²⁺ release [122, 123]. However, this single residue is unlikely to represent the sole determinant of the complex effects of Ca²⁺ on InsP₃R. Because of the critical importance of Ca²⁺ as a co-agonist at the InsP₃R, further defining regions of the protein responsible for this action is vital for an understanding of receptor function. By analogy with other Ca²⁺ binding motifs this will probably entail defining a pocket formed by coordinating residues within the 3 dimensional structure of the protein. Similarly, further work is also needed to gain a thorough understanding of the gating mechanism of the receptor, in particular an expanded characterization of the residues necessary for coupling InsP₃ binding to channel opening.

An additional approach is to exploit InsP₃R isoform specific differences in structure and function by the generation of chimerical receptors. For example, this paradigm has been used to investigate the particular molecular motifs responsible for the distinct InsP₃ binding affinities and Ca²⁺ sensitivities exhibited by individual InsP₃R isoforms [58, 124, 125]. An obvious further use for this approach is in elucidating legitimate nucleotide binding sites in InsP₃R-1 and InsP₃R-3. To this end an approach using the template of InsP₃R-2 ΔATPB, which is not regulated by ATP, [59] and serial substitution of portions of InsP₃R-1/-3 may be valuable in identifying the sites of nucleotide regulation of these isoforms.