Molecular Characterization of the Inositol 1,4,5-Trisphosphate Receptor Pore-forming Segment*

Specific residues in the putative pore helix, selectivity filter, and S6 transmembrane helix of the inositol 1,4,5-trisphosphate receptor were mutated in order to examine their effects on channel function. Mutation of 5 of 8 highly conserved residues in the pore helix/selectivity filter region inactivated the channel (C2533A, G2541A, G2545A, G2546A, and G2547A). Of the remaining three mutants, C2527A and R2543A were partially active and G2549A behaved like wild type receptor. Mutation of a putative glycine hinge residue in the S6 helix (G2586A) or a putative gating residue at the cytosolic end of S6 helix (F2592A) had minimal effects on function, although channel function was inactivated by G2586P and F2592D mutations. The mutagenesis data are interpreted in the context of a structural homology model of the inositol 1,4,5-trisphosphate receptor.

Specific residues in the putative pore helix, selectivity filter, and S6 transmembrane helix of the inositol 1,4,5-trisphosphate receptor were mutated in order to examine their effects on channel function. Mutation of 5 of 8 highly conserved residues in the pore helix/selectivity filter region inactivated the channel (C2533A, G2541A, G2545A, G2546A, and G2547A). Of the remaining three mutants, C2527A and R2543A were partially active and G2549A behaved like wild type receptor. Mutation of a putative glycine hinge residue in the S6 helix (G2586A) or a putative gating residue at the cytosolic end of S6 helix (F2592A) had minimal effects on function, although channel function was inactivated by G2586P and F2592D mutations. The mutagenesis data are interpreted in the context of a structural homology model of the inositol 1,4,5-trisphosphate receptor.
Inositol 1,4,5-trisphosphate receptors (IP 3 Rs) 3 are tetrameric ligand-gated ion channels located in the endoplasmic reticulum membrane that are co-activated by IP 3 and Ca 2ϩ (reviewed in Ref. 1). The Ca 2ϩ released through these channels can activate a diverse array of physiological processes, depending on the amplitude and frequency of Ca 2ϩ release (2). The IP 3 R contains a ligand-binding domain in the N-terminal region (amino acids 226 -576) and six transmembrane segments (S1-S6; amino acids 2276 -2590) in the C-terminal region of the receptor (3,4). Several key aspects of IP 3 R function remain to be addressed, including the identification of critical residues that line the channel conduction pore and the exact mechanism of the IP 3 -activated gating process.
In common with many voltage-gated ion channels, the IP 3 R and its close relative the ryanodine receptor (RyR) contain a short luminal pore helix and a selectivity filter (1). In the IP 3 R, these are located between the S5 and S6 helices. Experimental evidence suggests that the minimal channel domain of the IP 3 R lies in the portion of the receptor encoded by the sequences distal to the S5 helix (5). Previous mutagenesis studies of this region of the IP 3 R have focused on the selectivity filter sequence 2547 GVGD 2550 (numbering according to rat type-I IP 3 R). The results showed that the V2548I mutation increased conductance through the channel (6,7), that the D2550A mutation inactivated channel function (6), and that the D2550E mutation decreased the divalent cation selectivity of the channel (6). Examination of the sequences of all three IP 3 Rs and RyR isoforms from several species in the region of the putative pore helix and selectivity filter indicates the presence of 8 highly conserved residues. In the present study, we have examined the effect of mutating these conserved amino acids on IP 3 R function.
The N-terminal ligand-binding domain and the channel domain of the receptor show a direct intersubunit interaction in the tetrameric channel (8). A model for the gating mechanism of the IP 3 R channel has been proposed in which the N-terminal region of the receptor interacts with the channel domain at sites located in the cytosol-exposed S4-S5 linker (9). The S4-S5 linker is an amphipathic helix lying parallel to the membrane and is proposed to lead to the constriction of the S6 porelining helix bundle at the cytosolic aspect of the membrane (9,10). It has been proposed that conformational changes in the ligand-binding domain resulting from IP 3 binding cause a mechanical movement of the S4-S5 linker that releases the constriction of the S6 helix bundle and allows Ca 2ϩ ions to exit the channel (9). In other ion channels, it has been suggested that the gating process is associated with the bending of the S6 helix at proline or glycine residues (11,12). A conserved glycine (Gly 2586 ) is present in the S6 helix of IP 3 Rs, and we have investigated its putative role as a molecular hinge in the present study. The possible gating role of a bulky hydrophobic residue (Phe 2592 ) at the constricted cytosolic end of the S6 helix has also been investigated. Cys 2610 and Cys 2613 in the sequence context 2610 CFIC 2613 of the C-terminal tail downstream of the transmembrane region has previously been shown to be essential for channel function (9,13). The intervening Phe 2611 and Ile 2612 residues are completely conserved in all IP 3 R and RyR isoforms. The functional importance of these conserved residues has also been examined in the present study.
The results presented here define some of the essential residues for channel function in the pore helix/selectivity filter/S6 helix segment of the IP 3 R. The data are interpreted in the context of a structural homology model of the IP 3 R. 4 Although the results confirm the critical importance of residues that are presumed to be located at the entrance and exit of the selectivity filter, they also indicate that residues that are more distant from the pore may also be important for the functional integrity of the channel. Several different mutations in the pore that caused partial effects on channel function also produced unanticipated changes in Ca 2ϩ regulation of the receptor. The data suggest that the putative glycine hinge and hydrophobic gate do not function as predicted or that alternative residues fulfill these roles. The study also confirms the specific importance of Cys 2610 and Cys 2613 in channel function rather than the sequence context of the intervening conserved residues.

EXPERIMENTAL PROCEDURES
Expression Cloning-The cDNA encoding the IP 3 R type-I in pCMV3 was a kind gift of Dr. Thomas Südhof (University of Texas Southwestern Medical Center). All amino acid numbering is with reference to the rat type-I IP 3 R (15). The splice variant used in this study was SI (Ϫ), SII (ϩ), SIII (Ϫ). All point mutants were made using the QuikChange site-directed mutagenesis kit (Stratagene) utilizing a cassette encompassing the BstBI/XbaI fragment of the type-I IP 3 R in pBluescript (Invitrogen). Mutants were confirmed by sequencing, and the BstBI/XbaI-digested inserts were subcloned into the full-length IP 3 R.
Preparation of Microsomal Vesicles-COS-7 cells were harvested in isolation buffer (320 mM sucrose, 20 mM Tris-HCl, pH 7.4, 0.5 mM EGTA). The cell suspension was then lysed by 6 -8 passes through a 26 1 ⁄ 2-gauge needle. Lysates were spun at 750 ϫ g for 5 min. The KCl concentration was brought up to 600 mM, and the supernatants were spun at 115,000 ϫ g for 50 min. The pellets were resuspended in resuspension buffer A for flux assays (500 mM sucrose, 1 mM dithiothreitol, 20 mM Tris-HCl, pH 7.8, 120 mM KCl, 50 M EGTA). 45 Ca 2ϩ Flux Assays-Assays were performed as previously described (18). Briefly, microsomal vesicles prepared from COS-7 expressing SERCA2b and IP 3 R were incubated for 25 min at 30°C in a 200 nM Ca 2ϩ buffer (unless stated otherwise) supplemented with ATP and an ATP-regenerating system, 45 Ca 2ϩ , and 2 M ruthenium red. 45 Ca 2ϩ uptake was estimated in the absence of any addition or in the presence of 10 M IP 3 or 1 M A23187. After incubation, microsomes were vacuum-filtered over a 0.3-m filter (Millipore Corp.) and washed (150 mM KCl). Filters were counted in Budget-Solve complete counting mixture (Research Products International, Corp.). An unpaired Student's t test was used to assess statistical significance of differences in flux assays of the various mutant constructs. GraphPad QuickCalcs software (available on the World Wide Web) was used for all statistical analyses.
Single-channel Recordings-DT40 cells stably transfected with either the wild type or mutant type-I IP 3 R were rinsed twice with bath solution and placed onto a 35-mm plastic tissue culture dish and allowed to adhere for 10 min. Recording electrodes were created from borosilicate glass capillary pipettes (Warner Instruments) using a laser pipette puller and fire-polished to a final resistance between 10 and 20 megaohms. Cell membrane resistances were greater than 5 gigaohms, and capacitances were greater than 8 picofarads. IP 3 R single-channel currents were recorded in the whole cell patch clamp configuration. Records were digitized at a sampling frequency of 20 kHz using an Axopatch 200B and Pclamp 9 software and filtered at 5 kHz with a four-pole Bessel filter. Bath solution con- Homology Modeling-Details on the building of the models of the IP 3 R pore in its closed and open conformations are described elsewhere. 4 In brief, these models resulted from a comprehensive amino acid sequence analysis of IP 3 R and its comparison with the sequences from RyR and K ϩ channels. The resulting sequence comparisons were used for the construction of homology models using the crystal structures of KirBac1.1 (Protein Data Base accession number 1P7B (20)) and MthK (Protein Data Base accession number 1LNQ (21)) as templates for closed and open conformations, respectively. Subse-quently, the two models were subjected to energy minimization protocols.

Role of Conserved Residues in the Pore Helix and Selectivity
Filter-IP 3 Rs and RyRs share the greatest sequence homology in the C-terminal channel domains. In common with other voltage-gated channels, the IP 3 R pore domain is made up of four structurally conserved regions, notably an outer helix (S5), a pore helix, a selectivity filter, and a pore-lining inner helix (S6). Sequence alignments of all three IP 3 R isoforms and the type-I RyR in the region encoding the putative pore helix, selectivity filter, and inner helix are shown in Fig. 1. In our initial studies, we focused on several highly conserved residues in the pore helix and selectivity filter region, between residues 2533 and 2550 ( Fig. 1). In order to aid in the interpretation of the data, we have utilized a homology model of the pore region of the IP 3 R in its closed state constructed using the KirBac1.1 channel as a template. 4 The locations of the conserved residues in this model are shown in Fig. 2, except for Asp 2550 , which is not shown, since it was not investigated in the present studies. We have previously shown that the mutation D2550A functionally inactivates the channel (22).
To test if mutations of the conserved residues to alanine affect function, we employed a 45 Ca 2ϩ flux assay using microsomal vesicles prepared from COS-7 cells co-transfected with various mutant IP 3 R constructs along with the Ca 2ϩ pump, SERCA2b. The assay measures the activity of transfected rather than endogenous IP 3 Rs and has been described in detail elsewhere (8,9,18). The first lane in Fig. 3B shows the IP 3 -mediated inhibition of 45 Ca 2ϩ flux in microsomes prepared from COS-7 cells transfected only with SERCA2b (no recombinant IP 3 Rs) and confirms the negligible contribution from endogenous IP 3 Rs in this assay (9,18). The eight pore helix and selectivity filter conserved residue mutants studied here (C2527A, C2533A, G2541A, R2543A, G2545A, G2546A, G2547A, and G2549A) were transiently expressed in transfected COS-7 cells (Fig. 3A). When co-expressed with type-III IP 3 Rs, all mutants were found to form hetero-oligomers as detected by co-immunoprecipitation assays (data not shown). The 45 Ca 2ϩ flux data in Fig.  3B indicate that the alanine substitutions fell into two groups: mutants that had a complete loss of function and mutants that displayed no loss of function or had conditional effects dependent upon the type of amino acid substitution (summarized in Fig. 1).
The mutations that were entirely nonfunctional were C2533A, G2541A, G2545A, G2546A, and G2547A (Fig. 3B). In the homology model, Cys 2533 and Gly 2541 are located at the two ends of the pore helix ( Fig. 2B). A mutation analogous to C2533A in RyR1 has been shown to be unresponsive to caffeine when expressed in dyspedic 1B5 myotubes (23). Mutations analogous to the IP 3 R mutant G2541A in mouse RyR2 (G4820A), rabbit RyR2 (G4822A), or human RyR1 (G4890R) showed diminished sensitivity to caffeine (24 -27) and did not yield channels in planar lipid bilayers (26). Crystal structures indicate that the pore helix makes strong interactions with the selectivity filter in KcsA channels (28) and with the S5/S6 helices in KCNQ1 K ϩ channels (29). It seems likely therefore that mutation of the highly conserved residues Cys 2533 and Gly 2541 of the pore helix could also alter the structure of these other domains and cause the observed functional impairments.
According to the homology model, Gly 2545 , Gly 2546 , Gly 2547 , and Gly 2549 , along with Val 2548 , would make up the selectivity filter of the IP 3 R (Fig. 2). As described above, the substitution of these glycine residues by alanine results in nonfunctional channels (Fig. 3B). Mutation of other ion channels in this region has been shown to have major effects on ion conduction (30 -33). When the position analogous to Gly 2545 was mutated in both RyR1 (34) and RyR2 (24,26), it induced a large decrease in K ϩ conductance. However, an alanine mutation of the adjacent glycine (corresponding to the IP 3 R Gly 2546 ) in RyR2 retained a K ϩ conductance of 500 pS compared with 700 pS for the wild type receptor and also retained a normal sensitivity to caffeine (24,26). The RyR mutant that would correspond to position Gly 2547 of the IP 3 R showed a K ϩ conductance of 280 pS and appeared to have undergone other functional alterations, including the absence of divalent cation selectivity and loss of regulation by Ca 2ϩ (26). In recent studies, a mutation analogous to G2547A in Drosophila IP 3 R (G2630S) has been shown to be nonfunctional when analyzed in planar lipid bilayer experiments using Ba 2ϩ as the current carrier (35). On the basis of the available data, it would appear that the severity of the effects of mutating the three adjacent glycines in IP 3 Rs is greater than the effects seen in RyRs, at least for the positions corresponding to Gly 2546 and Gly 2547 . It would be premature to conclude that these differences reflect differences in the structure of the pore of the two channels. A bulkier isoleucine residue in the selec-FIGURE 1. IP 3 R sequence alignment. Shown is sequence alignment of the pore regions of type-I, -II, and -III IP 3 Rs and type-I RyR. Areas corresponding to the pore helix, selectivity filter, and S6 pore-lining helix are marked by black lines. Residues that are 100% conserved among all IP 3 R subtypes and all RyRs are in boldface type below the alignment. The highly conserved residues that were specifically mutated in this study or have been reported in the literature fell into two functional categories: complete loss of function (grey) and no loss of function or conditional effects depending on amino acid substitution (open box).
tivity filter of RyRs replaces Val 2548 in IP 3 Rs. This may contribute to a greater stability of the selectivity filter of RyRs and a higher resistance to alanine substitution of the preceding glycines. Additionally, the apparent discrepancies between the effects of these mutations in IP 3 R and equivalent mutations in RyR could also be related to the widely different assay procedures and experimental conditions used to monitor channel function (see below).
Within the selectivity filter, the only mutant that showed no loss of function was G2549A (Fig. 3B). Mutation of the Asp 2550 position to alanine or asparagine completely inactivates the IP 3 R channel (22). The fact that the adjacent Gly 2549 is tolerant to alanine substitution at this critical location suggests that this residue may be located in a somewhat flexible region that is not at the narrowest part of the pore. This is consistent with the fact that in the homology model, the residues Gly 2545 , Gly 2546 , and Gly 2547 of each subunit of the tetrameric assembly form inter-actions with the equivalent residue of adjacent subunits, but Gly 2549 does not seem to form such interactions and therefore is likely to have higher radial flexibility. 4 However, only a limited flexibility is predicted, since substitution of bulkier residues, such as G2549C or G2549W, completely inactivates channel flux (Fig. 3B). It is noteworthy that mutations to alanine of the analogous position in mouse or rabbit RyR2 also did not affect caffeine responses or single-channel conductance measured with K ϩ as current carrier (24,26). However, the substitution of a glutamate at the corresponding position in human RyR1 completely inactivates the channel (27,36).
Two of the alanine mutants showed partial function in the 45 Ca 2ϩ flux assays (Fig. 3B). Cys 2527 and Arg 2543 had 49 and 87% of the functional activity of wild type receptors, respectively. It should be noted that the partial effects were independent of experimental variation in the protein expression of the mutant constructs (data not shown). In the homology model of the IP 3 R, Cys 2527 is in a loop at the luminal end of the pore helix (Fig.  2). The possibility that Cys 2527 and Cys 2533 may form an intrasubunit disulfide bond appears unlikely, since the effect of mutating Cys 2527 and Cys 2533 on 45 Ca 2ϩ fluxes is not equivalent (Fig. 3B). By contrast, mutation of each cysteine in RyR1 completely inactivates caffeine responses (23), but differential effects of the two mutants on other functional parameters, such as single-channel conductance, have not been examined. Complete inactivation of IP 3 Rmediated Ca 2ϩ release was also reported for Cys 2527 stably expressed in DT40 cells (37).
Although the R2543A is defined in this study as a partially functioning mutant because its response is statistically different from wild type receptors, this mutant nevertheless retains a significant functional activity. The Arg 2543 residue is situated in the homology model at the C-terminal end of the pore helix (Fig. 2). The presence of four positive charges close to the permeation pathway of a divalent cation might be anticipated to strongly impede ion flux. However, in the homology model of the IP 3 R, the side chain of Arg 2543 is oriented away from the permeation path and is stabilized by interactions with the backbone of the S5 helix (Fig. 2). 4 The equivalent position to Arg 2543 has been mutated to alanine in RyR2 and has been found to FIGURE 2. Homology model of the IP 3 R pore region. Homology models of the pore region of the IP 3 R have been calculated using KirBac1.1 as a template. 4 The eight conserved residues of the pore helix/selectivity filter region that have been mutated in the present study are color-coded. A, models of the pore region for all four subunits of the IP 3 R viewed from the luminal side (left) and along the membrane (right). B, homology model viewed along the membrane with two subunits removed to reveal the selectivity filter.
retain function (24), although the mutant does show a decreased sensitivity for caffeine in Ca 2ϩ release assays and has 70% of the single-channel conductance of wild type channels with K ϩ as current carrier (26). In highly selective K ϩ channels, a pore helix dipole with its negative end pointed in the direction of ion permeation has been proposed to stabilize ions as they transit through the pore (38). However, the substitution of a positively charged residue at the C-terminal end of the pore helix of an inwardly rectifying K ϩ channel has a minimal effect on the conduction properties of the channel (39). The exact role of the positive charge within the pore domain of IP 3 Rs/RyRs is unknown, but its removal by mutation, although not markedly affecting ion flux, does alter Ca 2ϩ regulation of the receptor (see below).

A Comparison of Several Assays to Monitor the Function of Mutant IP 3 R Channels-The analyses of mutations in RyRs
have employed a number of different assay techniques, such as measurement of the Ca 2ϩ release response of recombinant receptors to caffeine in intact cells or measurements of ion channel activity after functional reconstitution of purified channels in lipid bilayers. However, in these studies, there are several examples of mutations that have produced inconsistent results in the different assays. Some channels that behave normally in caffeine-induced Ca 2ϩ release assays have severely diminished single-channel activity when reconstituted into bilayers (e.g. D4899Q in RyR2 (40)). There are also examples in which a mutant shows no caffeine response but still has a significant single-channel activity (e.g. D4917A in RyR1 (34)). In order to assess whether the IP 3 R mutants behave in the same manner in several independent assays, we conducted experiments with two selected mutants, which in the 45 Ca 2ϩ flux assays behave as functional (G2549A) or nonfunctional (G2546A) channels. In the first assay, we transfected these mutants into DT40 chicken B-lymphocytes containing targeted deletions of all three IP 3 R isoforms (TKO DT40 cells) (41). Stable cell lines were prepared and were shown to express the mutants by immunoblotting analysis (data not shown). Fig. 4A shows measurements of cytosolic [Ca 2ϩ ] changes in these cell lines after challenge with anti-IgM to cross-link and activate B-cell receptors. The data show that the G2549A mutant gave a robust Ca 2ϩ response, whereas G2546A mutant did not, which is in agreement with the data from the microsomal 45 Ca 2ϩ flux assays. The thapsigargin-releasable Ca 2ϩ stores in both mutants was comparable with wild type cells (data not shown).
Recently, Dellis et al. have shown that DT40 cells express a low density of IP 3 Rs in the plasma membrane and that the activity of these channels can be monitored by whole-cell patch clamp methods (7). We have utilized this approach to monitor the activity of G2546A and G2549A mutant IP 3 R channels stably expressed in the TKO DT40 cells (Fig. 4B). In contrast to the results from the 45 Ca 2ϩ flux and cytosolic [Ca 2ϩ ] measurements, the G2546A inactive mutant, although having a lower P o , still retained a single-channel conductance that was similar to the wild type or the G2549A mutant (Fig. 4C). The apparent discrepancy may in part be related to the use of K ϩ as a current carrier in the patch clamp experiments. It is noteworthy that the equivalent mutants in RyR2 also retained K ϩ conductance (24,26). The pores of IP 3 Rs and RyRs are expected to be much larger than the ϳ3 Å diameter found for K ϩ channels (42,43). Thus, certain mutations that restrict the flux of the relatively large Ca 2ϩ ion may have less of an effect on the movement of the smaller K ϩ ion. With the present methodology, the technical difficulties of measuring smaller currents and retaining gigaohm seals in the presence of high concentrations of Ca 2ϩ preclude a direct assay of the Ca 2ϩ conductance of IP 3 R channels in DT40 cells. In subsequent experiments, we have confined our measurements to 45 Ca 2ϩ flux assays.
Role of Gly 2586 as a Putative Hinge in the S6 Helix-Glycine residues within the inner helix of some ion channels are predicted to act as hinges that can bend during activation gating (11,44,45). All IP 3 R and RyR isoforms contain a completely conserved glycine in the inner helix, which in IP 3 R1 is at Gly 2586 ( Fig. 1 and Fig. 5, A and B). Point mutations of Gly 2586 were made to alanine and proline and transiently transfected into COS-7 cells. The expression of these constructs is shown in Fig.  6A, and the 45 Ca 2ϩ flux data from isolated microsomes are shown in Fig. 6B. The G2586A mutant showed considerable activity, although it was statistically different from wild type receptors and therefore falls into our category of a partially functional mutant. The corresponding mutant in RyR2 (G4864A) was also found to behave normally when challenged with caffeine or ryanodine (46), which is qualitatively in agreement with our findings with the IP 3 R. The data could indicate that the conserved glycine does not function as a molecular FIGURE 3. IP 3 -mediated Ca 2؉ release from microsomal vesicles containing mutations in the pore helix and selectivity filter of the IP 3 R. A, immunoblot of microsomes expressing wild type and mutant IP 3 Rs specifically detected with an antibody (CT-1) against the C terminus of the receptor. B, 45 Ca 2ϩ uptake into microsomal vesicles prepared from COS-7 expressing SERCA2b and IP 3 R was measured as described under "Experimental Methods." Recombinant IP 3 Rs and SERCA2b segregate to the same vesicles, allowing recombinant IP 3 Rs to be studied by measuring the inhibition of 45 Ca 2ϩ accumulation caused by the addition of 10 M IP 3 (18). Opening of functional IP 3 R channels results in a decreased accumulation of 45 Ca 2ϩ . For each construct, the amount of counts accumulated in the presence of IP 3 was expressed as a percentage of the control uptake in the absence of IP 3 . Each assay was carried out in triplicate on 3-6 independent microsomal preparations. Statistical significance was assessed by Student's t test. *, significantly different (p Ͻ 0.05) from SERCA2b alone. †, significantly different from wild type. hinge or that the substitution to alanine does not markedly interfere with the normal hinge motion of the S6 helix, as found in computer simulations of the glycine substitution in the KcsA channel (47). It is also possible that the motion of the S6 helix and mechanism of gating are fundamentally different in IP 3 Rs/ RyRs from those described for K ϩ channels. In this context, Ludtke et al. (48) have analyzed high resolution reconstructions of the EM structure of RyR1 and have concluded that the inner helix in the closed state of RyR1 is already bent. It should also be noted that variable results for glycine mutagenesis have been observed even for different K ϩ channels. For example, in Shaker (45,49) and BK channels (49), the mutation of the central glycine leads to loss of function or severely reduced currents, whereas the corresponding mutation in Kir3.4 (50) has no effect. In wild type KCNQ1 (Kv7.1) channels, the position of the central glycine is already occupied by an alanine residue, whereas mutation of the central glycine in the closely related KCNQ2 (Kv7.2) results in an inactive channel (51). It has been argued that the critical nature of the central glycine in some channels may be related less to its role in helix flexibility and more to the small size of its side chain, which would prevent steric clashes with adjacent regions of the pore (52). Interestingly, an analysis of our IP 3 R models in open and closed conformations (Fig. 5) 4 reveals that indeed not much bending appears to occur in the S6 helix from its closed to its open states. Instead, the whole helix moves slightly outward, almost as a rigid body, with only a small helix unwinding at the very C-terminal end. This nevertheless results in a wide opening of a gate at the cytosolic surface of the pore (Fig. 5).
The introduction of a proline residue (G2586P) led to an apparent loss of channel function in the 45 Ca 2ϩ flux assays (Fig. 6B). This substitution has not been tested in RyRs, but the analogous mutation yields enhanced currents in Shaker (49) and NaChBac (44) and decreased currents in Kir3.4 channels (50). Several lines of evidence suggest that the G2586P IP 3 R mutant may be constitutively active. First, the microsomes from G2586P transfected COS-7 cells showed a high basal 45 Ca 2ϩ leak in the flux assays (Fig. 7A). Second, DT40-TKO cells transiently transfected with G2586P cDNA displayed an elevated basal cytosolic free [Ca 2ϩ ] compared with untransfected cells (Fig. 7, B and C). This elevated Ca 2ϩ could be reduced by removal of extracellular Ca 2ϩ . An enhanced plasma membrane Ca 2ϩ entry would be anticipated if intracellular stores are depleted by a leaky G2586P channel. G2586P transfected cells showed no releasable intracellular Ca 2ϩ stores (data not shown). Definitive electrophysiological evidence of a constitutively gated channel could not be obtained, because we were unable to generate stable DT40 cell lines from this mutant. The substitution of a glycine residue by a proline is likely to cause disruption of the helical secondary structure of S6 by the introduction of a constitutive bend. This not only may result in a local structural rearrangement but it may also interfere with the conformational rearrangements associated with channel opening. Thus, distortion of the S6 helix may be responsible for the anomalous channel function observed. Further scanning mutagenesis of additional S6 residues with alanine, proline, and glycine residues will be required to better understand the motions of the IP 3 R S6 helix during the gating transitions of this channel.
Role of Phe 2592 as a Putative Gating Residue in the S6 Helix-A general mechanism of ion conduction that has emerged from structural studies of several ion channels is that the pore-lining S6 helices closely approach each other to form a bundle on the cytosolic side of the channel, which restricts ion flux. In IP 3 Rs, we have proposed that opening of this restriction is mediated by an interaction of the S4-S5 linker with the cytosolic end of the S6 helix and that the S4-S5 linker is physically associated with residues in the IP 3 -binding domain of the receptor (9). In other channels, the gating residue present at the narrowest part of the conduction pathway is usually the side chain of a hydrophobic amino acid. For example Phe 187 , Phe 146 , and Phe 168 are proposed to perform this same role in GIRK4 (50), KirBac1.1 (20), and Kir6.2 (53) channels, respectively. Identification of an analogous residue in IP 3 Rs is complicated by uncertainties relating to the exact cytosolic boundary of the S6 helix. The homology model of the RyR2 based on the KcsA channel template has a considerably longer S6 helix than derived from our homology model based on Kir-Bac1.1 or estimates of the S6 helix boundary obtained by analysis of EM structures of RyR1 (48). A highly conserved Phe 2592 is present at an appropriate position to potentially function as a gate (Fig. 1). Based upon our homology model, this site would be located at the boundary of the assumed membrane/cytosol interface.
Mutation of Phe 2592 to an alanine produced a channel with partial function, having an activity that was ϳ70% of the wild type receptor (Fig. 6B). By contrast, the analogous mutation in Kir6.2 almost entirely eliminates gating of the channel by ATP or pH (53). Similarly, replacement of a putative isoleucine hydrophobic gate in Kir1.1 by the smaller glycine residue also abolished pH gating (54). Substitution of the IP 3 R Phe 2592 with an aspartate produced an inactive channel (Fig.  6B). Hypothetically, the introduction of negative charges at the position of the gating residue, which in the closed state should be located close to the 4-fold axis of the channel, could give rise to electrostatic repulsion between side chains and a constitutive opening of the S6 helix bundle. However, there was no evidence from 45 Ca 2ϩ uptake data that the F2592D mutant was constitutively active (data not shown). In Shaker channels, the substitution of putative hydrophobic gates with aspartate impaired trafficking of channels to the plasma membrane (55), suggesting a severe structural defect, which may also underlie the inactivation of the F2592D IP 3 R mutant. Since disruption of the hydrophobic gating ring by mutagenesis inactivates gating in other ion channels, but alanine substitution at Phe 2592 retains partial function, it is possible that Phe 2592 may not be at the narrowest point of the conduction pore of IP 3 R channels. Alternative candidates include Leu 2595 or several conserved isoleucine residues at  the end of the S6 helix (Fig. 1). It is also possible that the gate may be formed from the overall arrangement of amino acids at the C terminus of the S6 helix rather than by a single amino acid side chain. Localizing a single residue to a gate could be further complicated by the possibility that mutagenesis may also alter local interactions like those with the S4-S5 linker that are recognized to be critical for channel function (9).
Mutation of the Conserved CXXC Sequence in the C-terminal Tail-The C-terminal tail of IP 3 Rs contains two invariant cysteine residues (Cys 2610 and Cys 2613 ) in the sequence context CFIC, where the intervening Phe 2611 and Ile 2612 residues are also completely conserved in all IP 3 R and RyR isoforms (Fig.  1). The effect of mutating these residues to alanine is shown in Fig. 6B. The data indicate that the channel was inactivated when either of the conserved cysteines was mutated. This confirms the findings of Uchida et al. (13). However, the mutation of the bulky hydrophobic Phe 2611 and Ile 2612 residues did not inactivate channel function, with the mutants retaining ϳ70 and 60% of the activity observed for wild type channels. The residues corresponding to Phe 2611 and Ile 2612 in RyRs have not been mutated, but the mutation of either of the flanking analogous conserved cysteines has been shown to also inactivate RyR1 channels expressed in dyspedic myotubes (23). The CXXC motif is commonly present at the catalytic site of thiol oxidoreductases of the thioredoxin family (57,58). It has been noted that this motif is frequently present at the N-terminal segment of a helix and that intramolecular disulfide bond formation acts to stabilize this helix (14). Although the redox status of these cysteines in IP 3 Rs remains to be experimentally determined, it is possible that disulfide bond formation plays a role in stabilizing the structure of the C-terminal tail that is essential for channel function (9,13).
Ca 2ϩ Dependence of IP 3 R Partial Function Mutant Channels-Biphasic regulation by Ca 2ϩ is a well established regulatory feature of IP 3 R channels (1). We sought to examine if any of the IP 3 R mutants that showed partial function in 45 Ca 2ϩ flux assays had altered regulation by Ca 2ϩ . This was examined for three selected mutants (R2543A, G2586A, and F2592A) by carrying out the microsomal 45 Ca 2ϩ flux assays at different concentrations of buffered Ca 2ϩ and 10 M IP 3 (Fig. 8). In the case of the F2592A mutant, a higher concentration of Ca 2ϩ was required to obtain peak activity, and the inhibitory effect of Ca 2ϩ was also reduced when compared with wild type channels. The G2586A and R2543A channels appeared to be activated normally by low concentrations of Ca 2ϩ , but the inhibition by high concentrations of Ca 2ϩ was enhanced when compared with wild type channels (Fig. 8B). The data indicate that point mutations within the pore domain of IP 3 Rs can affect the regulation of receptors by Ca 2ϩ . Similar findings have been made in RyRs, where point mutations within the pore helix can influence Ca 2ϩ activation and high FIGURE 7. The G2586P mutant is constitutively active. A, microsomal vesicles were made from COS-7 cells transiently transfected with cDNA encoding the SERCA2b pump and either wild type or the G2586P mutant. Microsomes (40 g of protein) were incubated with ATP and an ATP-regenerating system and allowed to accumulate 45 Ca 2ϩ in the absence of IP 3 . Background counts obtained in the presence of A23187 were subtracted from the data. *, significantly different (p Ͻ 0.05) from wild type (n ϭ 3). B, TKO DT40 cells were transiently transfected with G2586P and a plasmid encoding the red fluorescent protein pHcRed1-N1. The representative traces show a comparison of the Fura-2 340/380 ratio in two transfected cells (containing red fluorescence) and two nontransfected cells exposed to carbachol (Cch) and Ca 2ϩ -free medium. Cumulative data quantitating the basal Fura-2 ratios from the indicated numbers of cells are shown in C. affinity ryanodine binding (34). The effects of Ca 2ϩ on the IP 3 R at the single-channel level are all mediated by changing the time the channel dwells in the closed state (1). We suggest that the point mutations may induce various structural changes in the pore region that modify the gating transitions that are triggered when Ca 2ϩ interacts with the receptor.
Conclusions-There have been relatively few studies that have used mutagenesis to analyze the function of the IP 3 R pore, and these have been confined to a few residues (22,35). The present study is the first to examine a larger set of conserved residues found in all IP 3 R and RyR isoforms. In general, the functional role of those residues that completely inactivate channel function can be rationalized in terms of their critical locations in the pore helix or selectivity filter. The interpretation of results when residues have no effects or partially impair function is less clear. Additional insights will require further mutagenesis and accompanying refinements of the available structural models. The data do not definitively exclude a role for a hinged S6 helix or a hydrophobic occluding residue in IP 3 R gating, although the particular residues selected for mutagenesis in the present study may not be critical for these functions. On a methodological note, the use of electrophysiological techniques to monitor the function of mutant IP 3 R channels has obvious potential to provide valuable information, but the limitation of using monovalent K ϩ as a current carrier may not be informative in identifying critical residues that regulate the gating and permeation of divalent Ca 2ϩ ions.