Critical regions for activation gating of the inositol 1,4,5-trisphosphate receptor.

To understand the molecular mechanism of ligand-induced gating of the inositol 1,4,5-trisphosphate (IP(3)) receptor (IP(3)R)/Ca(2+) release channel, we analyzed the channel properties of deletion mutants retaining both the IP(3)-binding and channel-forming domains of IP(3)R1. Using intrinsically IP(3)R-deficient cells as the host cells for receptor expression, we determined that six of the mutants, those lacking residues 1-223, 651-1130, 1267-2110, 1845-2042, 1845-2216, and 2610-2748, did not exhibit any measurable Ca(2+) release activity, whereas the mutants lacking residues 1131-1379 and 2736-2749 retained the activity. Limited trypsin digestion showed that not only the IP(3)-gated Ca(2+)-permeable mutants lacking residues 1131-1379 and 2736-2749, but also two nonfunctional mutants lacking residues 1-223 and 651-1130, retained the normal folding structure of at least the C-terminal channel-forming domain. These results indicate that two regions of IP(3)R1, viz. residues 1-223 and 651-1130, are critical for IP(3)-induced gating. We also identified a highly conserved cysteine residue at position 2613, which is located within the C-terminal tail, as being essential for channel opening. Based on these results, we propose a novel five-domain structure model in which both N-terminal and internal coupling domains transduce ligand-binding signals to the C-terminal tail, which acts as a gatekeeper that triggers opening of the activation gate of IP(3)R1 following IP(3) binding.

Inositol 1,4,5-trisphosphate (IP 3 ) 1 is a second messenger that is produced by hydrolysis of phosphatidylinositol 4,5-bisphosphate in response to activation by extracellular stimuli of the G protein-or tyrosine kinase-coupled receptors on the plasma membrane in various cell types (1). IP 3 mediates the release of Ca 2ϩ from intracellular storage sites such as the endoplasmic reticulum by binding to the IP 3 receptor (IP 3 R)/Ca 2ϩ release channel. IP 3 -induced Ca 2ϩ release (IICR) regulates numerous physiological processes, including fertilization, cell proliferation, development, muscle contraction, secretion, learning, and memory. In this signal transduction pathway, the IP 3 R works as a switch that converts the information carried by extracellular stimuli into intracellular Ca 2ϩ signals. IP 3 -gated intracellular Ca 2ϩ release channels are composed of four IP 3 R subunits (2). There are at least three types of IP 3 Rs (IP 3 R1, IP 3 R2, and IP 3 R3) (3), and they exist as both homo-and heterotetramers (4). The structure of IP 3 Rs has traditionally been divided into three functional domains (3,5): the N-terminal ligand-binding domain; the modulatory/coupling domain; and the C-terminal transmembrane/channel-forming domain, which contains six putative membrane-spanning regions. The transmembrane region is required for the intermolecular interaction in the formation of a tetrameric complex (6 -9), and it is likely that the C-terminal cytoplasmic region just following the putative membrane-spanning regions has a supportive role in the association among the subunits (6,9). An ion conduction pore has been proposed to be located in the hydrophobic segment between the fifth and sixth transmembrane regions (10,11). The primary sequence of the transmembrane domain adjacent to the pore-forming segment is highly homologous to that of the ryanodine receptor (RyR), another type of intracellular Ca 2ϩ release channel, suggesting that these two channels might share a common structure for the conduction of Ca 2ϩ ions.
Each IP 3 R subunit has a single high affinity IP 3 -binding site (2). The IP 3 -binding core, a minimum essential region for specific IP 3 binding (12), resides among residues 226 -578 of mouse IP 3 R1 (2749 amino acids) (13), and it contains 11 essential basic amino acids for IP 3 binding (14). The N-terminal 225 residues, which are close to the IP 3 -binding core, have been thought to function as a suppressor for IP 3 binding because their deletion from the N-terminal 734-amino acid region results in significant enhancement of IP 3 -binding activity (12,15). IICR is a positively cooperative process (16 -18), i.e. the binding of at least two IP 3 molecules to a single tetrameric IP 3 R channel is required for channel opening. IP 3 binding elicits a large conformational change in the N-terminal cytoplasmic portion of the IP 3 R (19). Furthermore, the C-terminal cytoplasmic region following the transmembrane domain is thought to be involved in the IP 3 -induced gating of the receptor because monoclonal antibody (mAb) 18A10, whose epitope is located in the C-terminal portion of mouse IP 3 R1 (13,20,21), has an inhibitory effect on IICR, without causing any decrease in the affinity of the receptor for IP 3 (21). Controlled trypsinization induces fragmentation of mouse IP 3 R1 into five major fragments, and all four N-terminal cytoplasmic fragments, which contain the IP 3 -binding core, are associated directly or indirectly with the remaining C-terminal fragment, which contains the channel domain (22). The trypsinized IP 3 R retains significant IICR activity, indicating that intramolecular interaction within a subunit and/or intermolecular interaction between neighboring subunits could effect functional coupling between IP 3 binding and channel opening (22). However, the sites of the interfaces between the cytoplasmic fragments and the channel domain and the molecular mechanism of their coupling remain to be elucidated.
IICR has been shown to occur in a quantal manner in permeabilized cells and isolated endoplasmic reticulum membranes (23,24). The addition of submaximal concentrations of IP 3 in the presence of Ca 2ϩ pump inhibitors leads to the partial release of sequestered Ca 2ϩ , and the amount of released Ca 2ϩ varies with the concentration of IP 3 (24). Although the Ca 2ϩ release terminates abruptly, because it can be reinitiated by an additional increment in IP 3 concentration (24), the rapid termination of Ca 2ϩ release is not due to ordinary inactivation or desensitization of the receptor. Purified IP 3 Rs reconstituted into lipid vesicles reveal a quantal Ca 2ϩ flux (17,25), indicating that the quantal release of Ca 2ϩ is an intrinsic property of the IP 3 R. Similar behavior was observed for the RyR, which mediates Ca 2ϩ -induced Ca 2ϩ release from intracellular Ca 2ϩ stores (26), but has not been observed for other ligand-gated ion channels on the plasma membrane, suggesting that the quantal release is a fundamental and unique property of the intracellular Ca 2ϩ release channels.
To understand the molecular basis of the ligand-induced gating of the IP 3 R, we analyzed a series of internal deletion mutants and site-directed mutants of mouse IP 3 R1 expressed in intrinsically IP 3 R-deficient R23-11 cells (27). We found that at least two regions and a cysteine residue are essential for IP 3 -dependent gating of IP 3 R1. These findings provide us with new insight into the gating mechanism of the IP 3 R.
Preparation of Membrane Fractions from Stable Cells Expressing Mouse IP 3 R1 and Its Mutants-Membrane fractions were prepared in accordance with the protocol for mouse cerebella described by Michikawa et al. (18), with minor modifications. Cells were collected by centrifugation, washed twice with cold phosphate-buffered saline, and homogenized in ice-cold homogenization buffer (5 mM NaN 3 , 0.1 mM EGTA, 1 mM 2-mercaptoethanol, and 20 mM HEPES-NaOH, pH 7.4) containing protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride, 10 M leupeptin, 10 M pepstatin A, and 10 M E-64) by 40 strokes in a chilled glass-Teflon Potter homogenizer at 1000 rpm. The homogenate was centrifuged at 1100 ϫ g for 10 min at 2°C. The supernatant was centrifuged at 100,000 ϫ g in a Beckman TLA100.3 rotor for 30 min at 2°C. The pellet was resuspended in an appropriate volume of wash buffer (600 mM KCl, 5 mM NaN 3 , 20 mM Na 4 P 2 O 5 , 1 mM 2-mercaptoethanol, and 10 mM HEPES-HCl, pH 7.2) containing protease inhibitors. The suspension was centrifuged at 1100 ϫ g for 10 min, and the supernatant was centrifuged at 63,000 ϫ g for 30 min at 2°C. The pellet was finally suspended in an appropriate volume of Ca 2ϩ release buffer (110 mM KCl, 10 mM NaCl, 5 mM KH 2 PO 4 , 1 mM 2-mercaptoethanol, and 50 mM HEPES-KOH, pH 7.2) containing protease inhibitors to a final concentration of ϳ15 mg/ml protein. Ca 2ϩ release buffer was passed over Chelex 100 (Bio-Rad) to eliminate any extra free Ca 2ϩ before use. The membrane fractions were either used immediately or frozen in liquid nitrogen and stored at Ϫ80°C until used.
[ 3 H]IP 3 Binding Assay Using Membrane Fractions-The IP 3 binding assay was performed as described previously (6). The membrane fractions (50 -200 g/tube) were incubated with 9.6 nM [ 3 H]IP 3 (PerkinElmer Life Sciences) in 100 l of binding buffer (50 mM Tris-HCl, pH 8.0, 1 mM EGTA, and 1 mM 2-mercaptoethanol) for 10 min at 4°C. After centrifugation, the pellets were dissolved in Solvable (PerkinElmer Life Sciences), and the radioactivities were measured with a scintillation counter (Beckman LS6500). Nonspecific binding was measured in the presence of 10 M unlabeled IP 3 (Dojindo Laboratories).
Fluorescence was recorded at 510 nm with alternate excitation of 340 and 380 nm (F 340 and F 380 , respectively). Using a CAF-110 spectrofluorometer (Japan Spectroscopic Co.), signals were recorded every 0.01 s with MacLab Version 3.6 (ADInstruments) at 30°C. When the Ca 2ϩ uptake induced by the addition of 1 mM ATP reached a steady level, 2 M thapsigargin was added to eliminate active Ca 2ϩ uptake through intrinsic Ca 2ϩ pumps. The rate of leakage from the membrane fractions following the addition of thapsigargin was almost linear. When the ratio of fluorescence intensity (F 340 /F 380 ) reached 1.2, corresponding to ϳ170 nM free Ca 2ϩ , various concentrations of IP 3 were added. At the end of each experiment, 2 mM CaCl 2 and 10 mM EGTA were added successively for normalization and calibration (30).
Limited Trypsin Digestion of Mutant Receptors-Limited trypsin digestion was performed as described previously (22). Microsomal fractions (0.25-5 mg/ml) of wild-type and mutant IP 3 R1-expressing cells were incubated with 0.01-10 g/ml trypsin in trypsinization buffer (120 mM KCl, 1 mM EDTA, 1 mM 2-mercaptoethanol, and 20 mM Tris-HCl, pH 8.0) at 35°C for 10 min. The reaction was terminated by the addition of 50 g/ml soybean trypsin inhibitor (Sigma) and 0.1 mM phenylmethylsulfonyl fluoride. After the addition of an equal volume of SDS-PAGE sample buffer, reaction mixtures were incubated at 55°C for 30 min. The digested proteins were separated by 8% SDS-PAGE and then analyzed by Western blotting with anti-IP 3 R1 antibodies N1, 4C11, 10A6, 1ML1, and 18A10 and the anti-(1718 -31) antibody (see Fig. 1A) (22). 3 R-deficient R23-11 Cells-As previously reported (6), we constructed 17 internal deletion mutants of mouse IP 3 R1. Among these mutants, we selected seven (Fig. 1B) containing both the IP 3 -binding region (residues 226 -578) and the putative transmembrane domain (residues 2276 -2589) to investigate the critical regions for the coupling between ligand binding and channel opening. In addition, we constructed two mutants lacking residues 1-223 and 2736 -2749, respectively (Fig. 1B). To express these mutant receptors, we introduced the mutant cDNAs into R23-11 cells and established stable cell lines by selection with 2 mg/ml G418. Fig. 2A illustrates the results from Western blot analysis of the membrane fractions prepared from these stable cell lines using anti-IP 3 R1 polyclonal antibody 1ML1, whose epitope lies within residues 2504 -2523 of IP 3 R1 (Fig. 1A) (10). All of the mutant receptors except D2610 -2748 were detected with an appropriate molecular mass ( Fig. 2A). Because D2610 -2748 was not well recognized by antibody 1ML1, we confirmed the expression of D2610 -2748 by Western blot analysis using anti-IP 3 R1 mAb 4C11 (20). As shown in Fig. 2B (open arrowhead), an additional signal with a low molecular mass was detected, indicating that degradation (or truncation) of D2610 -2748 occurs in R23-11 cells.

Expression of Deletion Mutants of IP 3 R1 in Intrinsically IP
[ 3 H]IP 3 -binding Activities of Deletion Mutant IP 3 Rs-The IP 3 -binding activities of the internal deletion mutant receptors expressed in R23-11 cells were measured by equilibrium [ 3 H]IP 3 binding analysis as described previously (6). There was no significant IP 3 -binding activity in the membrane fraction obtained from the R23-11 cells (data not shown). Therefore, we measured the ligand-binding activity of the exogenously expressed IP 3 R using membrane fractions obtained from the stable cell lines. The IP 3 -binding properties of wild-type and deletion mutant IP 3 Rs are summarized in Table I. Wild-type   (Table I). The IP 3 -binding affinity of mutant receptors D1267-2110, D1845-2042, and D1845-2216 was 2-3-fold lower (Table  I), and mutant D651-1130 had 7.5-fold lower affinity for IP 3 (Table I). Mutant D1-223 exhibited, however, significantly higher affinity for IP 3 (Table I), consistent with a previous report showing that residues 1-223 act as a suppressor for IP 3 binding (12). It has been reported that IP 3 binding to the IP 3 R is not cooperative (31,33), and the same property holds true for the wild-type receptor and all of the mutant receptors except D1-223 expressed in R23-11 cells (Table I). Both the Western blot ( Fig. 2) and IP 3 binding (Table I) analyses showed that the amount of IP 3 R protein expressed in each cell line was different. The amounts of the mutant IP 3 Rs expressed were in the range of 1.5-9.2 pmol/mg of protein (Table I); and therefore, we used two stable cell lines (KMN13 and KMN107) 2 expressing different amounts of wild-type IP 3 R1 as controls in the following experiments. The B max values for KMN13 and KMN107 were 13 Ϯ 9 and 0.77 Ϯ 0.2 pmol/mg of protein, respectively (Table I).
IICR Activity of Wild-type and Deletion Mutant IP 3 Rs-To investigate the Ca 2ϩ release activity of the mutant IP 3 Rs, IICR from the membrane fractions prepared from each stable cell line was measured in the presence of the Ca 2ϩ pump inhibitor thapsigargin. No Ca 2ϩ release was observed from membrane fractions prepared from R23-11 cells even after the addition of 10 M IP 3 (Fig. 3A), indicating that using R23-11 cells as host cells for transfection allows evaluation of definite Ca 2ϩ release activity by exogenously expressed IP 3 Rs. Fig. 3B shows the time course of the Ca 2ϩ release mediated by recombinant wildtype IP 3 R1 after the addition of various concentrations of IP 3 . Both the rate and amplitude of Ca 2ϩ release depended on the concentration of IP 3 added, indicating that IP 3 R1 expressed in R23-11 cells exhibits the quantal Ca 2ϩ release that is known to be an intrinsic property of native IP 3 R1 (17, 25). As previously reported (11), D1692-1731, which corresponds to the SII Ϫ alternative splicing variant of IP 3 R1 observed in peripheral tissues (34,35), exhibited Ca 2ϩ release after the addition of 10 M IP 3 (Fig. 3A). Among the eight artificial mutants (D1-223, D651-1130, D1131-1379, D1267-2110, D1845-2042, D1845-2216, D2610 -2748, and D2736 -2749), only D1131-1379 and D2736 -2749 possessed measurable Ca 2ϩ release activity (Fig.  3A). Under the conditions employed, we could detect IICR from membrane fractions prepared from the low level IP 3 R1-expressing KMN107 cells (Fig. 3A), which contained 0.77 Ϯ 0.2 pmol of IP 3 -binding sites/mg of protein. The expression levels of all the internal deletion mutants in each stable cell line were higher than the expression level of wild-type IP 3 R1 in KMN107 cells (Table I). These findings suggest that none of the mutants except D1131-1379 and D2736 -2749 act as IP 3 -gated Ca 2ϩ release channels.
Limited Trypsin Digestion of Mutant IP 3 Rs-Mouse cerebellar IP 3 R1 is trypsinized into five major fragments (I-V) (Fig.  1A) (22). Limited proteolysis provides direct evidence of protein folding (36). To probe the tertiary structure of the mutant IP 3 Rs, we analyzed the trypsinized fragments of recombinant IP 3 R1. Because R23-11 cells contain an intrinsic mAb 4C11reactive protein whose molecular mass is similar to that of tryptic fragment II of IP 3 R1 (data not shown), we analyzed four tryptic fragments (I and III-V) of the recombinant receptors. As shown in Fig. 4A, wild-type IP 3 R1 expressed in R23-11 cells was digested into the same four fragments, indicating that recombinant IP 3 R1 retains native structure. We found that trypsin digestion of the Ca 2ϩ -releasing mutant D2736 -2749 generated the same four trypsinized fragments (Fig. 4B), indicating that D2736 -2749 folds in the same manner as wild-type IP 3 R1. Trypsinization of the other functional mutant, D1131-1379, also generated fragments IV and V (Fig. 4C). However, D1131-1379, which exhibited markedly decreased Ca 2ϩ release activity (Fig. 3A), was digested with much lower concentrations of trypsin (Fig. 4C). This difference in trypsin sensitivity suggests that the deletion of amino acids 1131-1379 influences the structure of the C-terminal channel domain. Tryptic fragments of the three functionless mutants, D1-223, D651-1130, and D2610 -2748, are shown in Fig. 4 (D-F, respectively). Fragments III-V were generated by trypsin digestion of D1-223 (Fig. 4D). Trypsinization of D651-1130, which lacks a cleavage site between fragments II and III (22), generated fragments IV and V (Fig. 4E). Both D1-223 and D651-1130 exhibited trypsin sensitivity similar to that of wild-type IP 3 R1, suggesting that at least the C-terminal channel domains of these mutants fold correctly. By contrast, only fragments I and III were generated by trypsin digestion of D2610 -2748 (Fig. 4F), and fragments IV (40 kDa) and V (91 kDa) were not detected (open arrows). These results indicate that deletion of residues 2610 -2748 induces a significant distortion of the folding structure of the C-terminal channel domain of IP 3 R1.
cating that this region is involved in the formation of a critical structure that is required for some common functions of the intracellular Ca 2ϩ release channels. One of the remarkable features of this region is that it contains two cysteine residues that are conserved in all of the intracellular Ca 2ϩ release channels. Both the IP 3 R and RyR channels are known to be modified by sulfhydryl reagents (37)(38)(39); therefore, we analyzed whether or not the two conserved cysteine residues, Cys 2610 and Cys 2613 (Fig. 5), are essential for the gating function in mouse IP 3 R1. We generated three mutant receptors (C2610S, C2613S, and C2610S/C2613S) in which Cys 2610 and/or Cys 2613 was replaced with serine (Fig. 1B). We also constructed a mutant receptor (C1976S) in which Cys 1976 was replaced with serine (Fig. 1B). All of the mutated cDNAs were introduced into R23-11 cells, and stable cell lines expressing mutant receptors were established. Equilibrium [ 3 H]IP 3 binding assay showed that all of the cysteine mutants bound IP 3 with affinities similar to that of wild-type IP 3 R1 (Table II). The expression levels of the mutant receptors were in the range of 1-10 pmol/mg of protein (Table II), and the amounts of all of the mutant receptors expressed in each established cell line were higher than the expression level of wild-type IP 3 R1 in KMN107 cells.
We then examined the Ca 2ϩ channel activities of the mutant receptors by measuring IICR from the membrane fractions prepared from the stable cell lines. We found that substitution of serine for Cys 2610 and/or Cys 2613 completely abolished the Ca 2ϩ release activity, whereas no such effect was apparent upon substitution of serine for Cys 1976 (Fig. 6). Limited trypsin digestion of C1976S, C2613S, and C2610S/C2613S generated fragments I and III-V (Fig. 7, A, C, and D, closed arrows), indicating that these mutants retain a normal structure. By contrast, trypsin digestion of C2610S generated fragments I, III, and IV (Fig. 7B, closed arrows), but not fragment V (open arrow), indicating that the single amino acid substitution at Cys 2610 induces a significant structural alteration of the Cterminal channel domain of IP 3 R1. DISCUSSION Most cells, including mammalian cultured cells, express two or all three types of the IP 3 R (4, 40). Thus, measurement of the actual channel activities of recombinant IP 3 Rs in most cells is difficult because of the background activities of the endogenous IP 3 Rs. In this study, we used R23-11 cells, which intrinsically lack all of the three IP 3 Rs (27), as the host cells to exclude the background effects of the endogenous IP 3 Rs. Under the conditions used, neither IP 3 -binding activity nor IP 3 -elicited Ca 2ϩ release activity was detected in membrane fractions prepared from R23-11 cells (Fig. 3A). Wild-type IP 3 R1 expressed in KMN13 cells revealed affinity for IP 3 (K d ϭ 20 Ϯ 5 nM) ( Table  I) similar to that of native (31) and recombinant (6,28,32) mouse IP 3 R1 and mediated Ca 2ϩ release from microsomal vesicles in an IP 3 -dependent manner (Fig. 3B). Wild-type IP 3 R1 expressed in KMN13 cells revealed quantal Ca 2ϩ release (Fig.  3B), which is thought to be an intrinsic property of native IP 3 Rs (17,25), suggesting that recombinant IP 3 R1 in R23-11 cells functions properly. Using this system, we determined that six of the mutants investigated in this study, those lacking residues 1-223, 651-1130, 1267-2110, 1845-2042, 1845-2216, and 2610 -2748, did not exhibit any measurable Ca 2ϩ release activity, whereas the mutants lacking residues 1131-1379 and 2736 -2749 retained the activity. [ 3 H]IP 3 binding analysis showed that all of these nonfunctional mutants except D1-223 possessed lower IP 3 -binding affinity (Table I). However, 98.5% of the receptors of even the lowest affinity mutant, D651-1130, whose K d is 150 nM (Table I), would have been occupied when 10 M IP 3 was applied. Therefore, the decrease in IP 3 -binding affinity is unlikely to be the primary cause of the loss of function of these mutants. The limited trypsin digestion of the crude membrane fractions prepared from the mutant-expressing cells showed that not only the IP 3 -gated Ca 2ϩ -permeable mutants, D1131-1379 and D2736 -2749, but two nonfunctional mutants, D1-223 and D651-1130, generated fragments IV and V (Fig. 4). All of these mutants except D1131-1379 exhibited trypsin sensitivity similar to that of wild-type IP 3 R1, indicating that these mutants retain a normal folding structure in at least the C-terminal channel-forming domain. Immunocytochemical staining suggested that all of the mutants were localized on the Ca 2ϩ stores of the cells (data not shown). Hence, we concluded that at least two regions, viz. regions 1-223 and 651-1130, are required for the IP 3 -dependent gating of IP 3 R1.
The Critical Region 1-223 Is Known as a Suppressor of IP 3 Binding-Deletion of the residues on the N-terminal side of the IP 3 -binding core has a complicated effect on the IP 3 -binding activity (12). A short deletion of the N-terminal 31 residues from the N-terminal 734-amino acid region results in a significant reduction in binding activity even though the mutant includes the entire IP 3 -binding core sequence. Such effects are also found in serial N-terminal deletions up to 215 residues. However, the binding activity recovers when deleted up to the first 220, 223, or 225 amino acids. The mutant lacking the first 223 amino acids shows Ͼ10-fold higher affinity for IP 3 than does the parental N-terminal 734-amino acid region. Based on these results, Yoshikawa et al. (41) proposed that the first ϳ225 N-terminal amino acids function as a suppressor of IP 3 binding. In this study, we found that D1-223 had 10-fold higher affinity for IP 3 than did wild-type IP 3 R1 (Table I). In addition, we found that the IP 3 binding of D1-223 was positively cooperative (Table I), indicating that the intersubunit interaction may be elicited (or modified) in the tetrameric complex composed of D1-223. Limited trypsin digestion showed that the mutant is likely to retain the normal folding structure of the C-terminal channel-forming domain (Fig. 4D). Surprisingly, D1-223 did not exhibit any measurable Ca 2ϩ release activity (Fig. 3A). These data therefore clearly indicate that residues 1-223 are required for the functional coupling between IP 3 binding and channel opening.
Homer (42)-and calmodulin (43)-binding sites are localized in region 1-223 of IP 3 R1. Homer forms an adaptor complex that couples between group 1 metabotropic glutamate receptors and IP 3 Rs, and it has recently been reported to be capable of associating with RyR1 and up-regulating its Ca 2ϩ release activity (44). The Homer-binding motif (PPXXFR) is present in residues 49 -54 (PPKKFR) of mouse IP 3 R1 (42). Calmodulin interacts with residues 49 -81 and 106 -128 in a Ca 2ϩ -dependent and Ca 2ϩ -independent manner (43). These binding proteins within the critical region 1-223 may modulate the gating of IP 3 R1.
The Critical Region 651-1130 Is Close to the IP 3 -binding Core-In region 651-1130 of IP 3 R1, the alternative splicing site (between amino acids 917 and 918) referred to as SIII is present (45), but the functional significance of the SIII segment (nine residues) has not yet been elucidated. There are three possible Ca 2ϩ -binding sites within regions 660 -745, 741-849, and 994 -1059 (46), suggesting that region 651-1130 is involved in the Ca 2ϩ -dependent regulation of IP 3 R function. Recently, Bosanac et al. (14) unveiled the three-dimensional structure of the IP 3 -binding domain (residues 224 -604) that covers the IP 3 -binding core (residues 226 -578) of IP 3 R1 in the presence of IP 3 . The IP 3 -binding domain forms an asymmetric boomerang-like structure that consists of an N-terminal ␤-trefoil domain (residues 224 -436) and a C-terminal ␣-helical domain (residues 437-604) containing three armadillo repeatlike folds. IP 3 fits into a cleft formed by these two arms of the boomerang. Our data described here suggest that region 651-1130, which immediately follows the IP 3 -binding core, is essential for IP 3 -induced gating of the channel. What kind of roles does this region have? It is known that other proteins containing armadillo repeats such as ␤-catenin (47) and importins (48) have Ͼ10 repeats. Based on the analysis of the amino acid sequence of IP 3 R1, Bosanac et al. (14) suggested that, in IP 3 R1, the armadillo repeat-like folds extend to the C-terminal region of the IP 3 -binding core. It is predicted that many ␣-helical domains are formed over the entire region 651-1130 (14), and deletion of residues 1131-1379 did not abolish the IP 3 -dependent Ca 2ϩ release activity (Fig. 3A). We therefore speculate that the armadillo repeat-like fold-containing ␣-helical domain, which is essential for IP 3 -induced gating, is formed within residues 440 -1130. This region might constitute part of the bridge that connects the IP 3 -dependent conformational change in the IP 3 R with channel opening.
Recently, Hamada et al. (49) showed that the purified IP 3 R from mouse cerebellum contains two distinctive structures: a windmill-like structure and a square-shaped structure. Ca 2ϩ reversibly promotes transition from the square-to the windmill-shaped structure, with relocation of the four peripheral IP 3 -binding domains. This observation predicts the presence of a hinge region that changes its conformation drastically following Ca 2ϩ binding to the receptor. Hamada et al. (49) examined the Ca 2ϩ -dependent structural change of the purified IP 3 R by limited protease digestion analysis. The results show that a 38-kDa fragment detected using anti-IP 3 R1 mAb 4C11 is spe-cifically generated by cleavage in a solution containing CaCl 2 , but not in an EDTA-containing solution. The epitope of antibody 4C11 was mapped within residues 679 -727 in IP 3 R1 (13); and therefore, region 651-1130 found in this study is a strong candidate for the hinge region. The presence of Ca 2ϩ -binding sites within this region (46) also supports this hypothesis.
Nonfunctional Mutants D1267-2110, D1845-2042, and D1845-2216 -Because the nonfunctional mutants D1267-2110, D1845-2042, and D1845-2216 do not have cleavage sites between tryptic fragments IV and V (Arg 1931 , Arg 1923 , or Lys 1924 ) (22), we could not evaluate the folding structure of the C-terminal channel-forming domain of these mutants. Therefore, we did not determine whether the deleted regions include critical regions for activation gating or whether the deletions simply distort the structure of the receptor. Notably, the deleted regions in these mutants include (or are close to) the Ca 2ϩ sensor region found in IP 3 R1 (50). Cytoplasmic Ca 2ϩ is a coagonist for the IP 3 R (51); and thus, the IP 3 -and Ca 2ϩ -binding signals must be combined on the IP 3 R. More detailed analysis of the regions deleted in these mutants may help us to better understand the molecular mechanism of gating, in particular, the Ca 2ϩ -dependent processes during channel opening.
Region 2610 -2748 May Be Required for the Correct Folding of IP 3 R1-The C-terminal region following the sixth transmembrane region may play some part in channel gating of the IP 3 R because mAbs that recognize this region have been reported to either inhibit (21) or enhance (52) IICR. In addition, it has been suggested that the C terminus is involved in subunit assembly of the IP 3 R channel complex (6,9). It has been shown that, although a truncation mutant of IP 3 R1 that lacks all of the transmembrane regions and the succeeding C terminus (amino acids 2218 -2749) is present as a monomer, a deletion mutant that lacks only the transmembrane regions (amino acids 2112-2605) forms dimers (6), suggesting that the C-terminal 144 amino acids (positions 2606 -2749) are involved in the intersubunit interaction of IP 3 R1. Facilitation of multimer formation of mutant IP 3 Rs with two or more transmembrane regions is observed if the mutants are fused to the C-terminal 145 residues (9); however, recombinant IP 3 R1 lacking the C-terminal 145 residues forms tetramers (9), suggesting that this C-terminal region is not essential for the formation of the tetrameric channel complex. In this study, we found that deletion of amino acids 2610 -2748 completely abolished the activity of the IP 3 R channel. D2610 -2748 was not well recognized by antibody 1ML1 (Fig.  2A). Limited trypsin digestion of mutant D2610 -2748 did not generate tryptic fragments IV and V (Fig. 4F), suggesting that deletion of residues 2610 -2748 affects the folding struc- ture around the cleavage sites between tryptic fragments IV and V.
The Essential Cysteine Residue in the C-terminal Tail-We found that the site-directed mutants C2610S, C2613S, and C2610S/C2613S did not exhibit any measurable Ca 2ϩ release activity (Fig. 6). As shown in Fig. 7, limited trypsin digestion of C2613S and C2610S/C2613S generated all four tryptic fragments (I and III-V), whereas trypsinization of C2610S generated only three tryptic fragments (I, III, and IV). These results indicate that Cys 2613 is required for the functional coupling between IP 3 binding and channel opening. The results of trypsinization of C2610S suggest that substitution of Cys 2610 disrupts the correct folding in at least the C-terminal channelforming domain of the IP 3 R. This is puzzling, however, because C2610S/C2613S generated all four tryptic fragments. The effect of a single amino acid substitution at Cys 2610 could be explained if the cysteine at position 2613 in the C2610S mutant elicits an artificial modification (such as disulfide formation, S-nitrosylation, or palmitoylation) that induces a serious distortion in the folding structure of the C-terminal channel domain. These modifications might be prevented in the presence of Cys 2610 . This explanation suggests a direct or indirect structural interaction between Cys 2610 and Cys 2613 in wild-type IP 3 R1.
These cysteine residues are also conserved in the RyR family (Fig. 5). Involvement of the cysteine residues in intracellular Ca 2ϩ channel gating has been postulated on the basis of evidence that thiol reagents such as oxidized glutathione and thimerosal enhance the Ca 2ϩ -mobilizing activity of both IP 3 R and RyR channels (1). There are nine cysteine residues conserved in both families (13). As shown in Fig. 6, substitution of the conserved cysteine residue at position 1976 did not affect the activity of IP 3 R1, whereas substitution of Cys 2610 or Cys 2613 caused loss of function of IP 3 R1. It is possible that thiol reagents directly attack the cysteine residues at positions 2610 and/or 2613 and enhance channel activity. Further studies are required to elucidate the exact target sites of thiol reagents and the mechanism of the enhancement of channel gating induced by these reagents.
Cysteine residues located in the C-terminal region following the transmembrane domain are known to be involved in the gating of some ion channels on the plasma membrane, such as cyclic nucleotide-gated channels (53)(54)(55) and voltage-dependent and inwardly rectifying K ϩ channels (56,57). In the presence of oxidants, a certain C-terminal cysteine residue in both these plasma membrane channels reacts with a cysteine residue located in the N-terminal region in the same or different subunit. This reaction depends on the states of the channels, and the formation of disulfide bonds results in channel potentiation. On the N-terminal side of the transmembrane domain, the IP 3 R possesses 18 cysteine residues that are conserved in the IP 3 R family. Thus, examining the interaction between the N and C termini via the formation of disulfide bonds in the presence of some oxidants according to the gating states would be useful in understanding the conformational changes that occur during IP 3 R gating.
A Novel Five-domain Structure Model for the IP 3 R-We found that two regions, viz. regions 1-223 and 651-1130, and Cys 2613 are crucial for the IP 3 -induced gating of IP 3 R1. How do they contribute to the activation gating of IP 3 R1? Recently, the open pore conformation of K ϩ channels (MthK) was resolved at a resolution of 3.3 Å (58,59). Structural comparison between KcsA and MthK (closed and open K ϩ channels) revealed that pore-lining inner helices form the channel gate and that bending of the inner helices causes channel opening. In the bent configuration, the inner helices form a wide (12 Å) entryway. The gating hinge is a glycine residue located in the middle of the inner helices. The glycine residue is highly conserved in voltage-or ligand-gated channels with two or six membranespanning segments per subunit, suggesting that the bending of the inner helices is a common mechanism for channel opening. The IP 3 R has been proposed to have the same structural arrangement of the pore-forming domain as the voltage-gated K ϩ channels (10); therefore, the IP 3 R pore may also be equipped with the same gating mechanism. Pore-lining inner helices that contain the gating hinge in the MthK channel correspond to the sixth membrane-spanning segment of the IP 3 R. It is of interest to note that a glycine residue is also present within the sixth membrane-spanning segment of all three types of the IP 3 R. This structural similarity suggests that the channel gate is formed by the sixth membrane-spanning helix of the IP 3 R. One of the striking differences between the IP 3 R and other ligand-gated channels with six membrane-spanning segments, such as cyclic nucleotide-gated channels, is in the location of the ligand-binding site. Both the IP 3 -binding site and the Ca 2ϩbinding sites are positioned on the N-terminal side of the transmembrane domain of the IP 3 R, whereas in other ligand-gated channels, the ligand-binding sites are located in the C-terminal cytoplasmic region that is close to the pore-lining inner helices. In these ligand-gated channels, the ligand-binding signals may be transferred directly to the pore domain and cause bending of the inner helices to open the channels. The ligand-binding signals of the IP 3 R may be transferred to the pore domain in a different manner. Based on the results presented here, we propose a novel five-domain structure model in which the Cterminal tail works as a gatekeeper for activation-induced gating of the IP 3 R (Fig. 8). In this model, conformational changes in the IP 3 -binding domain caused by IP 3 binding are transmitted through both the N-terminal and internal coupling domains to the C-terminal tail, which then triggers channel opening. Cys 2613 in the C-terminal tail may be critical for receiving the IP 3 -binding signal and/or for triggering channel opening. Further studies on the structure and function of the IP 3 R using the described experimental approach may provide us with an exact answer for the long-asked question, "How does the binding of IP 3 at the N terminus gate the C-terminal Ca 2ϩ permeation pore?"