Cooperative Formation of the Ligand-binding Site of the Inositol 1,4,5-Trisphosphate Receptor by Two Separable Domains*

Limited trypsin digestion of mouse cerebellar membrane fractions leads to fragmentation of the type 1 inositol 1,4,5-trisphosphate receptor (IP3R1) into five major components (Yoshikawa, F., Iwasaki, H., Michikawa, T., Furuichi, T., and Mikoshiba, K. (1999) J. Biol. Chem. 274, 316–327). Here we report that trypsin-fragmented mouse IP3R1 (mIP3R1) retains significant inositol 1,4,5-trisphosphate (IP3) binding activity that is comparable to the intact receptor in affinity, capacity, and specificity. This is despite the fact that the IP3-binding core (residues 226–578), which is close to the minimum for high affinity binding, is completely split into two tryptic fragments at the Arg-343 and/or Arg-345, around the center of the core. Furthermore, we have examined whether binding activity could be complemented in vitro by mixing two distinct glutathioneS-transferase (GST) fusion proteins, which were respectively composed of residues 1–343 and 341–604, almost corresponding to two split binding components, and separately expressed in Escherichia coli. The GST-fused residues 1–343 (GN) showed no binding affinity for IP3, whereas the GST-fused residues 341–604 (GC) displayed weak but definite activity with an affinity >100-fold lower than that of the native receptor. Upon mixing of both GN and GC, a high affinity site comparable to the native site appeared. We suggest that the IP3-binding pocket consists of two non-covalently but tightly associated structural domains each of which has a discrete function: the C-terminal domain alone has low affinity for IP3, whereas the N-terminal one alone is incapable of binding but is capable of potentiating binding affinity.

Inositol 1,4,5-trisphosphate (IP 3 ) 1 is a second messenger that mediates Ca 2ϩ release from intracellular stores by binding to the IP 3 receptor (IP 3 R) which is a tetrameric IP 3 -gated Ca 2ϩ release channel (1,2). Cerebellum has an extraordinary density of specific IP 3 -binding sites (3) and purified cerebellar IP 3 R protein binds IP 3 in a stoichiometric manner, namely one subunit for one IP 3 (4 -6). Ca 2ϩ release experiments using various synthetic inositol phosphates have suggested that molecular recognition of IP 3 is markedly stereospecific (7). Type 1 IP 3 R (IP 3 R1) is the neuronal type and predominates in cerebellar Purkinje cells (8 -10). It is 2749 amino acids long (molecular mass about 313 kDa) and is structurally divided into three parts as follows: a large N-terminal cytoplasmic arm region (residues 1-2275), a putative six membrane-spanning region clustered near the C terminus (residues 2276 -2589) which is thought to constitute an ion channel by forming a tetramer, and a short C-terminal cytoplasmic tail region (residues 2590 -2749) (11). A series of deletion mutants showed that the IP 3 R1 binds IP 3 within the N-terminal 650 amino acids, independently of tetramer formation (12,13).
We previously demonstrated the structural basis for molecular recognition of IP 3 by mouse IP 3 R1 (mIP 3 R1) (14). The minimum region for high affinity binding has been localized within the 353 residues, 226 -578, so that it appears to be close to the binding "core." Within the core region, we have identified 10 important basic amino acid residues all of which are well conserved in all IP 3 R family proteins cloned to date: three (Arg-265, Lys-508, and Arg-511) are critical and the other seven are required for specific binding. Nahorski and Potter (7) predicted that ionic interactions of positive charges on a binding site with negative charges on the three phosphate groups of IP 3 would make major contributions to specific recognition and binding. Thus, we have proposed that the IP 3 -binding core forms a pocket with a positively charged inner surface lining of these basic residues which recognizes and binds a negatively charged IP 3 ligand. Interestingly, all members of the IP 3 R family share extensive homology in the core sequence, except that the IP 3 R1 has an alternative splicing SI region (15 residues, 318 -332) (9,15), adjacent to which ϳ30 residues form the longest stretch of characteristic diversity within the family. An internal deletion of residues 316 -352 leads to loss of binding (13), although neither the presence nor the absence of the SI segment (mIP 3 R1SIϩ nor mIP 3 R1SIϪ subtype, respectively) significantly affects the binding, 2 suggesting that the diversed stretch is not a prerequisite for binding but that its boundary should be strictly defined.
Recently, we have shown that limited trypsin digestion of mouse cerebellar membrane fractions causes fragmentation of the mIP 3 R1 into five major trypsin-resistant polypeptides and that these five tryptic fragments I-V have tight structuralfunctional coupling because of the following: (i) co-sedimenta-tion of all four cytoplasmic peripheral fragments I-IV (40/37, 64, 76, and 40/36 kDa) together with the membrane-spanning integral fragment V (91 kDa) by centrifugation to pellet membrane proteins and immunoprecipitation with C terminal-specific antibody, and (ii) retention of IP 3 -induced Ca 2ϩ release channel activity in such fragmented mIP 3 R1. As a result of trypsinization, the IP 3 -binding region is cleaved at the carboxyl side of Arg-343 and/or Arg-345, around which is the alternative splicing SI region, into two polypeptides: fragments I and II. These results have demonstrated that the native IP 3 -binding region consists of two major tightly folded structural components connected by one exposed loop near the SI region; the latter loop is very susceptible to trypsin attack, whereas the former two components are not. Similarly, fragments III, IV, and V would be reflected by well folded conformation of the native mIP 3 R1 channel. Thus, determination of the structurefunction relationships among these components might clarify the molecular mechanism of ligand binding and channel opening of the mIP 3 R1 channel.
In the present study, we describe the structure-function relationships of the split tryptic IP 3 -binding region. We have characterized the IP 3 binding properties of trypsin-fragmented mIP 3 R1, and we have demonstrated that the tryptic mIP 3 R1, completely fragmented into five polypeptides, still retains specific IP 3 binding activity comparable to that of the intact receptor. To analyze further the split IP 3 binding components, we have separately synthesized two glutathione S-transferase (GST) fusion proteins with these two components, and we have shown that high affinity binding site can be reconstituted in vitro by complementation with both distinct fusion proteins, each of which alone has no (N-terminal component) or low affinity (C-terminal component) binding. From these data, we propose that the functional structure of the IP 3 -binding pocket consists of two well folded structural domains that are noncovalently but tightly associated and a diverse loop-like structure between these two domains.
Preparation of Membrane Fractions from Mouse Cerebellum-ddY mice (8 -10 weeks old; Nippon SLC, Japan) were anesthetized and then decapitated, and cerebella were quickly dissected. The cerebella were homogenized by 10 strokes (850 rpm) in an ice-chilled glass Teflon Potter homogenizer containing 9 volumes of 0.32 M sucrose, 1 mM EDTA, 100 M phenylmethylsulfonyl fluoride (PMSF), 10 M pepstatin A, 10 M leupeptin, and 5 mM Tris-HCl, pH 7.4. The homogenate was centrifuged at 1,000 ϫ g for 15 min at 4°C. The supernatant was re-centrifuged under the same conditions to completely remove P1 fraction. The second supernatant was centrifuged at 105,000 ϫ g for 60 min at 2°C. The precipitate (crude microsome) was resuspended with binding buffer (1 mM EDTA, 1 mM 2-mercaptoethanol, and 50 mM Tris-HCl, pH 8.0 (at 4°C)) to give a final concentration of about 15 mg/ml protein, frozen in liquid nitrogen, and stored at Ϫ80°C until use. Protein concentrations were determined by a Bio-Rad protein assay kit using bovine serum albumin as a reference.
Trypsin Digestion-Microsomal fraction (1 mg/ml protein) was incubated with the desired concentration of trypsin at 35°C in trypsinizing buffer (120 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 20 mM Tris-HCl, pH 8.0). The reaction was quenched with 10-fold weight excess of soybean trypsin inhibitor and 0.1 mM PMSF. IP 3 Binding Assay-IP 3 binding was carried out by incubating sam-ple protein with 9.6 nM (or 4.8 nM for Scatchard analysis of microsomal fraction) [ 3 H]IP 3 in 100 l of binding buffer for 10 min at 4°C. The IP 3 /protein mixture was added to 4 l of ␥-globulin (50 mg/ml) and 100 l of PEG precipitation buffer (30% (w/v) PEG6000, 1 mM EDTA, and 50 mM Tris-HCl, pH 8.0, at 4°C) and incubated for 5 min at 4°C. The protein-PEG complexes were pelleted by centrifugation at 18,000 ϫ g for 5 min at 4°C. The pellet was dissolved in 180 l of Solvable and then neutralized by adding 18 l of acetic acid. The radioactivity of the neutralized samples was measured by mixing 5 ml of Atomlight followed by counting in a liquid scintillation counter (Beckman). The specific binding was defined as total binding minus nonspecific binding that was measured in the presence of 2-100 M cold IP 3 . In Scatchard plots, lines were fit to the data by the least squares method using either the single site relationship where B is the specific IP 3 binding in pmol/mg protein, F is the concentration of the free IP 3 , K d is dissociation constant for each separate site, and B max represents the maximal number of binding sites for each site (16).
Construction, Expression, and Purification of GST Fusion Protein-Standard methods (PCR, cloning, and DNA sequencing) for construction of the mIP 3 R1 cDNA-derived plasmids were as described previously (14). DNA clones derived from PCR products were all confirmed by sequencing. To generate fusion constructs between GST and the IP 3 -binding region of mIP 3 R1, foreign nucleotide fragments as described below were attached to the mIP 3 R1 cDNA by PCR using appropriately designed primers. A BamHI site (GGATCC) was attached at the 5Ј terminus of the initiation codon (ATG) of the mIP 3 R1 for the GST-fused residue 1-343 (GN and GN(R265Q)) and at the 5Ј terminus of Arg-341 codon for the GST-fused residues 341-604 (GC), and TAA-GAATTC (stop codon ϩ EcoRI site) was attached at the 3Ј terminus of Arg-343 codon for the GN and GN(R265Q) and at the 3Ј terminus of Arg-604 codon for the GC. A fragment including the R265Q mutation was amplified using the R265Q mutant (14) as a template. Each genetically worked fragment coding the residues 1-343 (for GN), 1-343(R265Q) (for GN(R265Q)) and 341-604 (for GC) was digested by BamHI and EcoRI, and the resultant restriction fragment was cloned into the BamHI and EcoRI site of the GST fusion vector pGEX-2T. Escherichia coli JM109 strain was transformed with these expression constructs, and recombinant fusion proteins were expressed by the low temperature method that was developed to increase solubility of expressed IP 3 -binding proteins as described previously (14). Cells were harvested, resuspended in phosphate-buffered saline, and disrupted by sonication on ice. After centrifugation at 30,000 ϫ g for 60 min at 2°C, the supernatants were collected and subjected to a GST purification step through glutathione-Sepharose 4B column chromatography according to the manufacturer's protocol. The purified GST fusion proteins were subjected to SDS-PAGE analysis and IP 3 binding assay. 3 Binding Activity of the Trypsin-fragmented mIP 3 R1-We previously showed that limited trypsin digestion of mouse cerebellar membrane fractions fragmented the mIP 3 R1 into five major polypeptides (fragments I-V), in which the IP 3 -binding region was split into two fragments I and II (18). Intriguingly, these completely fragmented mIP 3 R1 still retained strong activity for IP 3 -induced Ca 2ϩ release comparable with that of the intact receptor, suggesting that even the split IP 3 -binding site was sufficient to couple with gating the fragmented channel.

Characterization of IP
To characterize further the structural and functional properties of the split IP 3 -binding site, we analyzed the mIP 3 R1 in trypsin-digested and undigested microsomal fractions of mouse cerebellum for specificity, affinity, and capacity of IP 3 binding. Microsomal fractions were digested with 5 g/ml trypsin for 4 min at 35°C and then pelleted by centrifugation. Almost all of the tryptic fragments as well as the intact mIP 3 R1 collected into the pellet. The pellet was solubilized with 1% Triton X-100 followed by centrifugation at 20,000 ϫ g for 60 min at 2°C to obtain the supernatant. The Triton extracts were then subjected to Western blot analysis using six site-specific antibodies, which specifically recognize the major tryptic fragments Ia/b (40/37 kDa), II (64 kDa), III (76 kDa), IVa/b (40/36 kDa), and V (91 kDa) (18). As shown in Fig. 1A, trypsinization re-sulted in complete fragmentation of the mIP 3 R1 into five tryptic fragments that were detectable with the site-specific antibodies (Trypsin ϩ), but when pre-mixed with 50 g/ml trypsin inhibitor and 0.1 mM PMSF prior to the digestion, no apparent mobility change was observed in the intact mIP 3 R1 band (Trypsin Ϫ). The IVa/b fragments were only detected at very low levels, since the epitope for anti-(1718 -31) located in the alternative splicing SII region (40 residues 1692-1731) of these fragments is most labile to trypsinolysis (18). The size difference between the Ia and Ib fragments has been thought to be due to alternative splicing at the SI region (15 residues 318 -332), the former derived from the SIϩ subtype and the latter from the SIϪ subtype (15,18).
By using the Triton extracts containing these tryptic fragments, the following [ 3 H]IP 3 -binding experiments were carried out. Although the IP 3 -binding core was completely separated into two fragments I and II, which were recognized by the N1 and N3 antibodies, respectively (Fig. 1A), the fragmented mIP 3 R1 (Trypsin ϩ) still retained specific IP 3 binding activity equivalent to that of the intact receptor (Trypsin Ϫ) (Fig. 1B). In [ 3 H]IP 3 binding competition experiments with various inositol phosphates, the fragmented mIP 3 R1 (Trypsin ϩ) showed comparable ligand binding specificity to the intact receptor (Trypsin Ϫ) in the order of (1,4,5)IP 3 Ͼ (2,4,5)IP 3 Ͼ (1,3,4,5)IP 4 ( Fig. 2A). Scatchard analysis showed that the capacity for IP 3 was unaffected by trypsin digestion; the B max (pmol/mg protein) was 23 for the fragmented mIP 3 R1 (Trypsin ϩ) and 22 for the intact receptor (Trypsin Ϫ) (Fig. 2B). In addition, the affinity for IP 3 (K d value) was also comparable between the fragmented mIP 3 R1 (13 nM) and the intact receptor (19 nM). These results suggested that a rigid folded conformation of the tetrameric mIP 3 R1 complex in the native membrane might preserve its inherent IP 3 -binding pocket from trypsin digestion, even though the IP 3 -binding core was split into two near the center.
Release of IP 3 Binding Activity from the Insoluble to Soluble Fraction by Prolonged Trypsin Digestion-Most IP 3 binding activity and fragments I and II were collected in an insoluble membrane fraction upon mild trypsin digestion, whereas a proportion of these fragments became soluble upon more extensive digestion. We next examined the temporal profile of IP 3 binding activity in relation to increasing time of digestion (Fig.  3). Microsomal fractions were treated with 10 g/ml trypsin for 0, 5, 10, 20, and 40 min and were separated in insoluble and soluble fractions by centrifugation at 105,000 ϫ g for 60 min. The pelleted insoluble fractions were solubilized with 1% Triton X-100 to obtain Triton extract. An equivalent volume of the soluble fraction (sup) and the Triton extract of insoluble fraction (ppt) were subjected to [ 3 H]IP 3 binding (Fig. 3A) and Western blotting assay (Fig. 3B). As the digestion time with trypsin was extended to 10 min, both the insoluble and soluble fraction showed a sharp difference in temporal changes of IP 3 binding activity; the former exhibited gradual attenuation, whereas the latter showed a gradual increase. When the digestion time was Mouse cerebellar microsomal fractions (1.5 mg of protein) in 1.5 ml of the trypsinizing buffer were digested with 5 g/ml trypsin for 4 min at 35°C in the presence (Trypsin Ϫ) and the absence (Trypsin ϩ) of 50 g/ml trypsin inhibitor and 0.1 mM PMSF. The digested samples were centrifuged at 105,000 ϫg for 60 min at 2°C to collect the insoluble membrane fractions. The pellet was resuspended in 0.5 ml of the binding buffer, solubilized by the addition of 10% (w/v) Triton X-100 to give a final detergent concentration of 1%, and rotated for 30 min at 4°C. The Triton-treated mixtures were centrifuged at 20,000 ϫ g for 60 min at 2°C to obtain the supernatants (Triton extracts). We confirmed by Western blotting that almost all of the tryptic fragments and the intact receptor were collected in these extracts (data not shown). The Triton extracts were subjected to Western blotting and IP 3 binding assay. A, the Triton X-100 extracts (2.5 g of protein) were separated by 8% SDS-PAGE and probed with N1, N3, 10A6, anti-(1718 -31), 1ML1 and 18A10 antibodies which specifically recognize the Ia/b, II, III, IVa/b and V major tryptic fragments, respectively. Intact IP 3 R and major tryptic fragments are indicated by an arrowhead and arrows, respectively. B, specific [ 3 H]IP 3 (9.6 nM) binding to the Triton extracts (45 g of protein). Nonspecific binding was measured in the presence of 2 M cold IP 3  extended to beyond 20 min, no more marked changes were observed. These temporal profiles appear to parallel the immunoblotting patterns (Fig. 3B); as the digestion time was extended to 10 min, the levels of the Ia/b and II fragments, containing the epitopes for the N1 and N3, respectively, were decreased in the insoluble fraction but increased in the soluble fraction, and thereafter (at 20 and 40 min) no apparent change was seen in both fractions. However, more extensive digestion caused loss of binding activity and of these immunoreactive fragments (data not shown). These data indicated that the two tryptic Ia/b and II fragments were tightly associated with the tryptic mIP 3 R1-membrane complex and that some portions released by the prolonged digestions appeared to retain conformation for specific binding by interaction between fragments.
Reconstitution of IP 3 Binding Activity by Complementation between Two Recombinant Binding Components-To examine the functional interaction between the two tryptic fragments I and II containing the split IP 3 -binding core, we tried to express recombinant proteins corresponding to these fragments in an E. coli system. However, it was difficult to obtain sufficient amounts of soluble proteins, especially for the fragment II (data not shown). To solve this, we constructed GST fusion proteins as shown in Fig. 4. The GN was composed of GST fused to the N-terminal residues 1-343, corresponding to almost the entire fragment I. We previously showed that Arg-265 within the IP 3 -binding core is one of three basic amino acid residues for which Gln substitutions (R265Q) caused a complete loss of binding activity (14). The GN(R265Q) had this R265Q mutation in the GN construct. The GC consisted of a GST fusion protein with residues 341-604, corresponding to almost all of the N-terminal 29-kDa subfragment derived from fragment II, which took place upon extensive trypsinolysis (18). These GST fusion proteins were expressed in E. coli and were purified through a glutathione column (Fig. 5A). [ 3 H]IP 3 binding assay showed that neither GN nor GN(R265Q) alone had any significant binding activity, whereas the GC alone showed some traces of activity (Fig. 5B).
To examine whether the two split components worked cooperatively to bind IP 3 , we performed a complementation test by which two of these separately synthesized proteins were simply mixed in vitro (right half of Fig. 5B). The combinations of GN ϩ GST, GC ϩ GST, GN(R265Q) ϩ GST, and GN(R265Q) ϩ GC had no significant change in [ 3 H]IP 3 binding. In marked contrast, the combination of GN ϩ GC displayed a dramatic increase in the specific binding activity, suggesting that the GN and the GC complemented each other to retrieve IP 3 binding activity, even though both have a large extra GST moiety at the N terminus.
To characterize further this structural and functional reconstitution of an IP 3 -binding site by mixing two separately synthesized GN and GC, we carried out Scatchard analysis as shown in Fig. 6. The GN alone exhibited no binding activity as the control GST. On the other hand, the GC alone could form a very low affinity site with K d value of 4.7 M. It was of particular interest that the mixture of GN ϩ GC displayed a plot fitted by two binding sites; the low affinity site appeared to be consistent with that of the GC alone, and the high affinity site had a K d value of 11 nM, about 400-fold lower than that of the low affinity site and comparable to that of the intact mIP 3 R1. Judging from the B max value of 40 pmol/mg protein, it was estimated that about 2% of the total GC (about 1800 pmol/mg of protein) was involved in reconstitution of this high affinity site by complementation with the GN. DISCUSSION Mild trypsin digestion of a cerebellar membrane fraction generates five major tryptic fragments, including the N-termi- nal two fragments Ia/b (40/37 kDa) and II (64 kDa), that share the entire IP 3 -binding site and are tightly associated with the insoluble membrane fraction (18). In the present study, we have shown that the trypsinized IP 3 -binding site of mIP 3 R1 retains significant affinity, specificity, and binding capacity for IP 3 , comparable to those of the intact one. These results indicate that the trypsinized IP 3 -binding site may retain functional tertiary structure. Prolonged trypsinization caused concomitant loss of IP 3 binding activity and release of fragments I and II from the insoluble mIP 3 R1-membrane complex to the soluble fractions. These results demonstrate that the folded conformation of the IP 3 -binding site is stably retained in either the insoluble or soluble form by a possible inter-fragment interaction. Joseph et al. (17) previously reported that prolonged trypsin digestions of rat cerebellar microsomes caused release of a 68-kDa fragment of IP 3 R with concomitant appearance of higher IP 3 binding activity in the soluble fraction, whereas with mild digestion the 68-kDa fragment was retained in the membrane-bound insoluble form with lower activity. In the present study, however, we have shown that the affinity of the insoluble fraction from the mildly trypsinized samples is slightly higher than that of un-trypsinized samples. We consider that the 68-kDa fragment is likely to correspond to the 64-kDa fragment II identified in the present study and that the binding activity observed by Joseph et al. (17) could be attributed to a native-like interaction between fragments I and II which are simultaneously released, although Joseph et al. (17) did not note fragment I in both the soluble and insoluble fractions. This discrepancy between our results and the previous study (17) may be due to differences in preparation of membrane fractions, trypsin digestions, and/or binding assays (filtration versus PEG precipitation).
Neither the SI plus (mIP 3 R1SIϩ) nor minus (mIP 3 R1SIϪ) subtype showed any significant alteration in IP 3 binding activity, 2 whereas deletions of any other region within the IP 3binding core tested so far completely abolished the activity (12)(13)(14), suggesting that the SI region is the only redundant part of the binding core. Arg-343 and Arg-345, which lie between fragments I and II and are close to the alternative splicing SI region (residues 318 -332), are very susceptible to Each protein (7 g) was analyzed individually: GC, GN, and GST. The mixture of GN (7 g) and GC (7 g) was analyzed: GNϩGC. In the plot of GNϩGC, the quantity of protein is tentatively expressed as that of the GC regardless of the GN, since the GN alone has no significant activity. Nonspecific binding was measured in the presence of 200 M IP 3 . An inset in the GN ϩ GC plot shows an enlargement of a plot for a high affinity site. limited trypsinolysis. Within the IP 3 -binding core of all members of the IP 3 R family, the most divergent sequences are found around the SI region. Thus, we suggest that in the vicinity of the SI region, Arg-343 and Arg-345 may form a flexible looplike structure (residues ϳ318 to ϳ345) that interconnects two non-covalently but tightly associated well folded structural components; "binding domains" I (residues 1 or N terminus to ϳ317) and II (residues ϳ346 to ϳ923) corresponding to the trypsin-resistant fragments I and II, respectively (Fig. 7).
Residues 224 -579 of the mIP 3 R1 expressed in E. coli which almost correspond to the IP 3 -binding core (residues 226 -578) show specific and high affinity IP 3 binding activity, indicating that the binding core itself forms a functional structure (14). We recently showed that more extensive trypsin digestion further split fragment II into two subfragments, an N-terminal 29-kDa and a C-terminal 38-kDa fragment, and that the carboxyl side of Arg-603 and/or Lys-604 would be the sites for this extensive trypsinization, which lie in the vicinity of the Cterminal boundary of the IP 3 -binding core (18). Therefore, we suggest that the IP 3 -binding core consists of two "core domains" I (at least residues ϳ226 to ϳ317) and II (residues ϳ346 to ϳ604) in the binding domains I and II, respectively (Fig. 7).
The present study showed that the GST-fused N-terminal binding domain I (GN) had no significant binding activity, whereas the GST-fused C-terminal core domain II (GC) exhibited low affinity. Strikingly, simple mixing of these regions led to increased binding activity. The K d value of this reconstituted high affinity IP 3 -binding site was almost comparable to that of the intact mIP 3 R1. These results suggested that the two separate GST fusion proteins mutually recognized and cooperatively formed a virtually native high affinity IP 3 -binding pocket. However, we could only reconstitute about 2% of the total input of the GC by just mixing the GN and the GC. This low rate of functional complementation may be due to physical interference by the N-terminal GST moiety attached to both fusion proteins, and/or due to low efficiency in direct interactions between GN and GC or in proper folding of expressed GN and GC, under the present conditions.
The present study provides evidence for the modular construction of the IP 3 -binding region with discrete functional domains. We suggest that the low affinity core domain II is a prototype of the IP 3 -binding structure (Fig. 7). The binding domain I, although it has no binding capacity by itself, could contribute to form a high affinity site cooperatively with the core domain II. Thus, the domain I seems to act as a "modulator" to potentiate the affinity of the IP 3 binding prototype, core domain II. The functional property of this N-terminal binding domain I, however, is a little complicated as described previously (14); a short deletion of the N-terminal 31 amino acids from the N-terminal 734 amino acid region (T734) resulted in a significant reduction in the binding activity, although the resultant mutant included the entire IP 3 -binding core sequence. Such contradictory mutational effects were also found in serial N-terminal deletions up to residue 215. However, the authentic binding activity was markedly recovered, when deleted up to the first N-terminal 220, 223, or 225 amino acids, thereby indicating that the N-terminal boundary of IP 3 -binding core is at most residue 226. Notably, the mutant lacking the first N-terminal 223 amino acids showed more than 10-fold higher affinity for IP 3 than that of the parental T734. Furthermore, a substitution of GST (26 kDa) for the first 223 amino acids of the N-terminal 604 amino acids (T604) significantly increased the affinity, as compared with the parental T604. 2 These lines of peculiar evidence led us to hypothesize that the N-terminal first ϳ225 amino acids are not directly responsible for the binding but are somewhat related to its suppression. Then, as shown in Fig. 7 we hypothesize that there are at least three functional modules, two modulators in the domain I (N-terminal suppressor and C-terminal enhancer (core domain I)) and one proto-IP 3 -binding site in the core domain II. Co-operative regulation among these modules for ligand binding may influence the channel gating.
Mignery et al. (12) reported that the N-terminal cytoplasmic region of rat IP 3 R1 expressed in COS cells displayed an altered mobility (apparent decrease in mass of Ͼ50 kDa) on gel chromatography in the presence of IP 3 and suggested that any conformational change induced upon binding to IP 3 might be involved in coupling the ligand binding to the channel gating. We suppose that relative movement of the core domains I and II non-covalently associated with each other would occur upon ligand binding as described below. We previously demonstrated the significance of the 10 basic amino acid residues (Arg or Lys) in specific IP 3 binding, all of which are well conserved within the IP 3 R family (14). Of these, Arg-265, Lys-508, and Arg-511 are critical. We thus suggested that these 10 basic residues, especially Arg-265, Lys-508, and Arg-511, contribute to form a positively charged pocket for binding to the negative charges on the three phosphate groups of IP 3 . Four of them (Arg-241, Lys-249, Arg-265, and Arg-269) are positioned in the core domain I and the other six (Arg-504, Arg-506, Lys-508, Arg-511, Arg-568, and Lys-569) in the core domain II (Fig. 7). Thus, the IP 3 -binding pocket constituting two core domains may be relatively expanded due to repulsion among the positive charges on the inner surface in the nonligand-bound state (open) and become narrow due to neutralization of the positive charges by interactions with the negative charges on IP 3 in the ligandbound state (closed). Finally, studies on the higher order struc- FIG. 7. Schematic model of the domain structure of the IP 3binding site. Trypsin-resistant fragments I and II of the mIP 3 R1 are designated as the binding domains I (residues 1 or N terminus to ϳ317) and II (residues ϳ346 to ϳ923). The N terminus of domain I in the native mIP 3 R1 could not been determined by amino acid sequencing (18). Around the region including the SI segment (residues 318 -332), Arg-343 and Arg-345 probably form a flexible loop between two domains, exposed to the outside, thereby being hypersensitive to limited trypsinolysis. The domain II consists of at least two subdomains of the N-terminal 29 kDa (probably residues ϳ346 to ϳ604) and the C-terminal 38 kDa (probably residues ϳ605 to ϳ923), which are cleaved off with a more extensive trypsinolysis (18). The IP 3 -binding core has been experimentally determined to reside within residues 226 -578, and therefore there are at least two putative core domains, I (residues at least 226 to ϳ317) and II (residues ϳ346 to ϳ604). The inner surface of the core domains may be lined with 10 basic amino acid residues for which Gln substitutions caused significant reduction of the binding activity (14). Of them, Arg-265, Lys-508, and Arg-511 are critical residues for the specific IP 3 binding, because their single amino acid substitutions to either Gln or Ala caused a loss of the activity (14). The functional roles for the ligand binding of the binding domain I, core domains I and II, and the region between residues 1 and ϳ225 are indicated in italics as described under "Discussion." tures of the IP 3 -binding site and the possible relative movement upon IP 3 binding will shed light on the molecular basis of the gating of the mIP 3 R1 channel as well as the ligand binding.