Molecular and Functional Evidence for Multiple Ca2+-binding Domains in the Type 1 Inositol 1,4,5-Trisphosphate Receptor*

Structural and functional analyses were used to investigate the regulation of the inositol 1,4,5-trisphosphate (InsP3) receptor (InsP3R) by Ca2+. To define the structural determinants for Ca2+ binding, cDNAs encoding GST fusion proteins that covered the complete linear cytosolic sequence of the InsP3R-1 were expressed in bacteria. The fusion proteins were screened for Ca2+ and ruthenium red binding through the use of 45Ca2+ and ruthenium red overlay procedures. Six new cytosolic Ca2+-binding regions were detected on the InsP3R in addition to the one described earlier (Sienaert, I., De Smedt, H., Parys, J. B., Missiaen, L., Vanlingen, S., Sipma, H., and Casteels, R. (1996)J. Biol. Chem. 271, 27005–27012). Strong45Ca2+ and ruthenium red binding domains were localized in the N-terminal region of the InsP3R as follows: two Ca2+-binding domains were located within the InsP3-binding domain, and three Ca2+ binding stretches were localized in a 500-amino acid region just downstream of the InsP3-binding domain. A sixth Ca2+-binding stretch was detected in the proximity of the calmodulin-binding domain. Evidence for the involvement of multiple Ca2+-binding sites in the regulation of the InsP3R was obtained from functional studies on permeabilized A7r5 cells, in which we characterized the effects of Ca2+ and Sr2+ on the EC50 and cooperativity of the InsP3-induced Ca2+ release. The activation by cytosolic Ca2+was due to a shift in EC50 toward lower InsP3concentrations, and this effect was mimicked by Sr2+. The inhibition by cytosolic Ca2+ was caused by a decrease in cooperativity and by a shift in EC50 toward higher InsP3 concentrations. The effect on the cooperativity occurred at lower Ca2+ concentrations than the inhibitory effect on the EC50. In addition, Sr2+ mimicked the effect of Ca2+ on the cooperativity but not the inhibitory effect on the EC50. The different [Ca2+] and [Sr2+] dependencies suggest that three different cytosolic interaction sites were involved. Luminal Ca2+ stimulated the release without affecting the Hill coefficient or the EC50, excluding the involvement of one of the cytosolic Ca2+-binding sites. We conclude that multiple Ca2+-binding sites are localized on the InsP3R-1 and that at least four different Ca2+-interaction sites may be involved in the complex feedback regulation of the release by Ca2+.

The inositol 1,4,5-trisphosphate receptor (InsP 3 R) 1 is an intracellular Ca 2ϩ release channel that is modulated by various physiological ligands such as inositol 1,4,5-trisphosphate (InsP 3 ), Ca 2ϩ , nucleotides, calmodulin (CaM), FK506BP, phosphatases, and kinases (reviewed in Refs. 1 and 2). The InsP 3 R can be divided into three functionally different domains as follows: an N-terminal ligand-binding domain, a modulatory domain, and a channel domain near the C terminus (3,4). Among the various modulators, Ca 2ϩ itself plays a pivotal role and may be considered as a co-agonist exerting both positive and negative effects on InsP 3 -induced Ca 2ϩ release. Cytosolic Ca 2ϩ has a bell-shaped effect on the InsP 3 R, with low concentrations stimulating the release and high concentrations inhibiting the release (5)(6)(7)(8). Both phases are time-dependent, but the inhibition develops more slowly than the stimulation (6, 9 -11). Although these kinetically distinct effects are believed to be crucial for the generation of Ca 2ϩ oscillations and waves (12), little is known about how Ca 2ϩ interacts with the InsP 3 R. Regulation might be exerted by direct binding of Ca 2ϩ to the InsP 3 R and/or via Ca 2ϩ -sensitive protein(s), e.g. via Ca 2ϩ /CaMdependent protein kinase II and protein phosphatase 2B (13) or via a Ca 2ϩ -sensitizing factor (14). Studies in which effects of Sr 2ϩ were compared with those of Ca 2ϩ revealed that at least two interaction sites must exist: a stimulatory site that is modestly sensitive to Sr 2ϩ and an inhibitory site that is nearly insensitive to Sr 2ϩ (15). One Ca 2ϩ -binding domain has been localized on the cytosolic side of the InsP 3 R-1 (16,17), but whether this domain represents a stimulatory or an inhibitory interaction site is still unclear. Luminal Ca 2ϩ also regulates the InsP 3 -induced Ca 2ϩ release. Loading of the Ca 2ϩ stores results in a relatively more pronounced InsP 3 -induced Ca 2ϩ release (18 -21). A high affinity Ca 2ϩ -binding site was detected on the luminal loop (17), but its functional significance is not yet clear.
The aim of the present work was two-fold. First, we wanted to investigate, at the structural level, the presence of direct Ca 2ϩ -binding sites on InsP 3 R-1. Second, we wanted to functionally demonstrate multiple Ca 2ϩ interactions on the InsP 3 R by a kinetic analysis of the effects of Ca 2ϩ and Sr 2ϩ on the InsP 3induced Ca 2ϩ release.
For the first part of this study we constructed and expressed a number of GST fusion proteins that contain cytosolic fragments of the InsP 3 R sequence. The ability of these fusion proteins to bind 45 Ca 2ϩ and ruthenium red was measured by overlay procedures and verified by staining with Stains-all. These studies identified multiple Ca 2ϩ -binding domains in the InsP 3 R sequence.
In the second part of the study we accurately determined the effects of Ca 2ϩ and Sr 2ϩ on the cooperativity and on the EC 50 of the InsP 3 -induced Ca 2ϩ release in permeabilized A7r5 cells using our recently developed rapid 45 Ca 2ϩ efflux technique (22). The effects of Ca 2ϩ on the V max of InsP 3 -induced Ca 2ϩ release were not determined in the present experiments because our previous work on fully loaded stores already indicated that the Ca 2ϩ release in response to a supramaximal [InsP 3 ] (320 M) was not affected when the cytosolic [Ca 2ϩ ] was varied between 10 nM and 10 M (11,23). The stimulation of the release by cytosolic Ca 2ϩ was due to a shift in EC 50 toward lower InsP 3 concentrations. The inhibition by Ca 2ϩ was due, first, to a decrease in cooperativity, and second, to a shift in EC 50 toward higher InsP 3 concentrations. Both types of inhibition had a different [Ca 2ϩ ] dependence. Sr 2ϩ only mimicked the effect on the cooperativity indicating that both inhibitory effects were exerted at different sites. Luminal Ca 2ϩ had no effect on the EC 50 nor on the cooperativity indicating that luminal Ca 2ϩ did not exert its effect by acting on a cytosolic Ca 2ϩ -binding site. Therefore at least four Ca 2ϩ -interaction sites must be present to explain these functional data.
Purification of GST Fusion Proteins-Expression, growth, induction, and harvest of GST fusion proteins was carried out as described previously (17). Isolation of the recombinant proteins was carried out as follows: the cell cultures, resuspended in phosphate-buffered saline (PBS) containing 1% Triton X-100, were sonicated at 20 kHz, 9 times for 10 s using a MSE Ltd. (Westminster, Great Britain) probe sonicator. The sonicated material was centrifuged (10,000 rpm for 10 min in a Sorvall SS-34 rotor). At this stage, the procedure differed for the soluble GST fusion proteins and the insoluble GST fusion proteins. The supernatant of the soluble GST fusion proteins was immediately incubated with glutathione-Sepharose 4B beads, whereas the initial supernatants of the insoluble recombinant proteins were discarded, and the pellets were  (25). After a 30-min incubation on ice, the soluble material was removed by centrifugation as stated above. Triton X-100 was added to the supernatant to a final concentration of 2%. Finally, this supernatant was also incubated with the glutathione-Sepharose 4B beads for at least 30 min at 4°C. After the incubation period, the beads were washed three times with 3 ml of wash buffer 1 (1% Triton X-100 in PBS), wash buffer 2 (PBS), and wash buffer 3 (50 mM Tris⅐HCl, pH 8.0), respectively. All wash buffers were supplemented with benzamidine (0.83 mM) and phenylmethylsulfonyl fluoride (0.2 mM). The fusion proteins were eluted with 10 mM glutathione in 50 mM Tris⅐HCl, pH 8.0. Protein concentration was determined by the method of Lowry et al. (26) after protein precipitation by 10% ice-cold trichloroacetic acid. 45 Ca 2ϩ and Ruthenium Red Overlay- 45 Ca 2ϩ and ruthenium red binding was carried out as described previously (17). In brief, GST fusion proteins were separated by electrophoresis on SDS-polyacrylamide gels. After electrophoretic transfer to a polyvinylidene difluoride membrane, the membranes were washed overnight at room temperature in nominally Ca 2ϩ -free (2 to 3 M Ca 2ϩ ) incubation medium (60 mM KCl, 10 mM imidazole HCl, pH 7.0, and 5 mM MgCl 2 ) and then overlaid with 2 Ci/ml 45 CaCl 2 or 25 mg/liter ruthenium red in the same medium for 10 min at room temperature. Membranes subjected to 45 Ca 2ϩ overlay were rinsed for 5 min in ice-cold 50% ethanol/water mixture and air-dried. 45 Ca 2ϩ quantification of the bands was done by means of the Storm PhosphorImager model 840 equipped with the Image Quant NT 4.2 software (Molecular Dynamics). Membranes subjected to ruthenium red overlay were washed with incubation medium and kept in distilled water.
The 45 Ca 2ϩ overlay technique used in this study will detect high affinity Ca 2ϩ -binding sites but may fail to detect some low affinity Ca 2ϩ -binding sites, since the wash procedures may remove 45 Ca 2ϩ from kinetically labile sites (27).
Stains-all Staining of the Ca 2ϩ -binding Fusion Proteins-Stains-all staining was carried out as described previously (17). After SDS-PAGE, the gels were fixed overnight with 50% methanol, 7% acetic acid, 43% water (v/v), extensively washed with 2-propanol (24 h) and stained in the dark with a solution containing 0.0025% (w/v) Stains-all in 25% 2-propanol, 7.5% formamide, and 67.5% 30 mM Tris base, pH 8.8 (v/v), as described by Campbell et al. (28). 45 Ca 2ϩ Efflux Technique-A7r5 cells from embryonic rat aorta were used between the 7th and the 19th passage after receipt from the American Type Culture Collection (Bethesda, MD) and subcultured weekly by trypsinization. The cells were cultured at 37°C in a 9% CO 2 incubator in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 3.8 mM L-glutamine, 0.9% (v/v) non-essential amino acids, 85 IU/ml penicillin, and 85 g/ml streptomycin. The cells were seeded in 12-well dishes (4 cm 2 ; Costar Europe, Badhoevedorp, The Netherlands) at a density of approximately 10 4 cells/cm 2 . 45 Ca 2ϩ fluxes on permeabilized cells were done on a thermostated plate at 25°C. The culture medium was aspirated and replaced by 1 ml of permeabilization medium containing 120 mM KCl, 30 mM imidazole HCl, pH 6.8, 2 mM MgCl 2 , 1 mM ATP, 1 mM EGTA, and 20 g/ml saponin. This solution was removed after 10 min, and the cells were washed once with a similar saponin-free solution. 45 Ca 2ϩ uptake into the non-mitochondrial Ca 2ϩ stores was accomplished by incubation for 60 min in 2 ml of loading medium containing 120 mM KCl, 30 mM imidazole HCl, pH 6.8, 5 mM MgCl 2 , 5 mM ATP, 0.44 mM EGTA, 10 mM NaN 3 , and 100 nM free 45 Ca 2ϩ . After this phase of 45 Ca 2ϩ accumulation, the monolayers were incubated in 1 ml of efflux medium containing 120 mM KCl, 30 mM imidazole HCl, pH 6.8, 1 mM ATP, 1 mM EGTA, and 2 M thapsigargin. Additions of InsP 3 or Ca 2ϩ are indicated in the figures. This efflux medium was replaced every 6 s as described (22). At the end of the experiment the 45 Ca 2ϩ remaining in the stores was released by incubation in 1 ml of 2% (w/v) SDS for 30 min.  (17), we already identified a cytosolic and a luminal region with Ca 2ϩ -binding properties. To detect additional Ca 2ϩ -binding sites on the InsP 3 R, we performed a complete screening of the remaining cytosolic InsP 3 R-1 sequence for Ca 2ϩ -binding sites. We constructed and expressed a total of 26 GST fusion proteins that contain small fragments (56 -202 amino acids) of the InsP 3 R-1 in E. coli. These short fragments of the InsP 3 R covered the complete cytosolic N-terminal 2275 amino acids, including the S-I and the S-II splice domain (29) and the Cterminal 150 amino acids of the mouse InsP 3 R-1 (Fig. 1). Ca 2ϩ binding to these fusion proteins was examined by a 45 Ca 2ϩ overlay procedure as described earlier (17). The cytosolic Ca 2ϩbinding domain labeled d and the Ca 2ϩ -binding domain on the luminal loop labeled l ( Fig. 1) were characterized previously (17) and are therefore not further considered in the present study. Fig. 2A shows a Coomassie Blue-stained gel of parental GST (pGST) and GST fusion proteins Cyt1-14, -16, and -18. Most of the GST fusion proteins migrated as single prominent bands corresponding to the expected molecular weights. Mainly little degradation was observed after expression and purification. The only exceptions were Cyt4, -8, -12, and -18 ( Fig. 2A), which were strongly degraded. Fig. 2B demonstrates 45 Ca 2ϩ binding of the different fusion proteins covering the cytosolic domains of the InsP 3 R. Strong 45 Ca 2ϩ binding was detected to Cyt3, -5, -6, -7, and -10 (Fig. 2B). 45 Ca 2ϩ binding to Cyt7 was much stronger than 45 Ca 2ϩ binding to the other Ca 2ϩ -binding fusion proteins, probably reflecting a higher affinity of this site for Ca 2ϩ . pGST, the bacterial fusion partner, which has no known Ca 2ϩ -binding site was used as a negative control and showed no 45 Ca 2ϩ binding (Fig. 2B).
To define more precisely the sequences involved in Ca 2ϩ binding within the positive domains Cyt3, -5, -6, -7, and -10, each domain was further divided into two parts. The subfragments of these InsP 3 R sequences were expressed as GST fusion proteins and designated Cyt3a, -3b, -5a, -5b, -6a, -6b, -7a, -7b, -10a, and -10b (Fig. 1). The estimated molecular mass of most of the GST fusion proteins corresponded approximately to the predicted value, except for Cyt7b (predicted value 34 kDa) which migrated more slowly, with an apparent mobility corresponding to a polypeptide of 37 kDa (Fig. 3A). Fig. 3B shows that 45 Ca 2ϩ bound to Cyt3a, -3b, -5b, -6a, -7b, and -10a but not to -5a, -6b, -7a, and -10b. It is of interest that the first degradation product of Cyt3a also bound 45 Ca 2ϩ , whereas the second degradation product that corresponds to pGST did not. The pattern of ruthenium red binding was similar to the pattern of 45 Ca 2ϩ binding in Fig. 3B, except for fusion protein Cyt5b that failed to stain with ruthenium red (data not shown). Cyt7b stained dark blue with Stains-all, and the blue staining of the Ca 2ϩ -binding fusion proteins Cyt3a, -3b, -5b, -6a, and -10a was less intense (data not shown). In contrast, pGST and fusion proteins Cyt5a, -6b, -7a, and -10b that did not bind Ca 2ϩ stained pink with Stains-all. These data are indicative for the presence of six new 45 Ca 2ϩ -binding regions within the cytosolic sequence of the mouse InsP 3 R-1, in addition to the one defined previously (17).
The presence of multiple Ca 2ϩ -binding sites could be the molecular basis for a complex regulation of the InsP 3 R by Ca 2ϩ . We have therefore further explored the kinetics and mecha-nism of the interaction of Ca 2ϩ with InsP 3 -induced Ca 2ϩ release. We used Ca 2ϩ as well as Sr 2ϩ to discriminate and characterize between different modes of interaction with divalent cations.

Lack of InsP 3 -induced InsP 3 R Inactivation in A7r5 Cells-
The cooperativity and the EC 50 of InsP 3 -induced Ca 2ϩ release can be determined by loading the Ca 2ϩ stores to steady state with 45 Ca 2ϩ and then incubating them with a progressively increasing [InsP 3 ] in efflux medium (22). This protocol only allows an accurate determination of the cooperativity and the EC 50  The inactivation of the hepatocyte InsP 3 Rs recorded by Dufour et al. (31) already occurred at 1 nM Ca 2ϩ , whereas that observed by Hajnóczky and Thomas (30) required stimulatory cytosolic Ca 2ϩ concentrations. The experiment was therefore repeated at 1 M free Ca 2ϩ , which is an optimal effective [Ca 2ϩ ] for stimulating the Ca 2ϩ release in A7r5 cells (11). Fig. 4B shows that the final Ca 2ϩ content of the stores was again independent of whether the [InsP 3 ] was increasing or decreasing. We conclude that InsP 3 -dependent incremental inactivation of the InsP 3 R did not occur under our assay conditions. As a consequence, our protocol of adding cumulative concentrations of InsP 3 can be used to measure the cooperativity and the EC 50 .
Effect of Cytosolic Ca 2ϩ and Sr 2ϩ -Ca 2ϩ -loaded stores were incubated in thapsigargin-containing efflux medium containing 5 nM, 1 or 4 M free Ca 2ϩ and challenged with a progressively increasing [InsP 3 ]. Fig. 5A gives the decrease in store

. Expression of pGST and GST fusion proteins Cyt1-14, -16, and -18 and measurement of 45 Ca 2؉ binding.
A, Coomassie Blue staining of the expressed pGST and GST fusion proteins Cyt1-14, -16, and -18. Affinity purified fusion proteins were solubilized in SDSsample buffer and separated by 10% SDS-PAGE. The amount of protein loaded was adjusted to give similar levels of fusion proteins, as detected by Coomassie Blue staining. B, SDS-PAGE resolved proteins of a similar gel were transferred to Immobilon P membranes, and a 45 Ca 2ϩ overlay was carried out. pGST was used as negative control. 45 Ca 2ϩ overlay was carried out as described under "Experimental Procedures." FIG. 3. Expression of GST fusion proteins containing subregions of Cyt3, -5, -6, -7, and -10 and measurement of 45 Ca 2؉ binding. A, Coomassie Blue staining of the expressed GST fusion proteins Cyt3a, -3b, -5a, -5b, -6a, -6b, -7a, -7b, -10a, and -10b. All fusion proteins were affinity purified. Fusion proteins were separated by SDS-PAGE as described in the legend to Fig. 2A. B, fusion proteins were solubilized, separated electrophoretically, and transferred to an Immobilon P membrane as described in Fig. 2B, and a 45 Ca 2ϩ overlay of a blot of a similar gel was carried out.
Ca 2ϩ content as a function of the corresponding [InsP 3 ]. Increasing the free [Ca 2ϩ ] from 5 nM to 1 M resulted in a more rapid decrease in store Ca 2ϩ content during the challenge with InsP 3 . A less rapid decline in store Ca 2ϩ content occurred at 4 M free Ca 2ϩ . These data confirm the bell-shaped activation of the InsP 3 R by cytosolic Ca 2ϩ . Fig. 5B gives the rates of Ca 2ϩ release as a function of the corresponding [InsP 3 ] for the three different Ca 2ϩ concentrations of Fig. 5A. The threshold [InsP 3 ] for inducing Ca 2ϩ release was 0.1 M at 5 nM Ca 2ϩ . The threshold became 0.02 M at 1 M Ca 2ϩ and increased again to 0.5 M at 4 M Ca 2ϩ . In contrast, the release rate curves became less steep as the free [Ca 2ϩ ] increased from 5 nM to 1 and 4 M. The Hill coefficients for these traces, and also for the traces measured at intermediate Ca 2ϩ concentrations (data not shown), were calculated as described (22) and plotted by the open symbols in Fig. 6A. The Hill coefficient was 2.7 at 5 nM free Ca 2ϩ . This value is higher than that reported before (2.0; Ref. 22) because the present efflux medium also contained 1 mM ATP, which increases the cooperativity of the release (32). The Hill coefficient decreased as the free [Ca 2ϩ ] increased. We (22) and others (33) already reported a decrease in cooperativity at very high Ca 2ϩ concen-trations (10 M). The present work extends these observations, and we now report a gradual decrease in Hill coefficient as the cytosolic [Ca 2ϩ ] increased. This decrease in cooperativity occurred with an EC 50 for Ca 2ϩ of 0.21 M and was already very pronounced at 1 M Ca 2ϩ , i.e. at the optimal effective concentration for stimulating the release (11).
We have also studied the effect of a progressively increasing [InsP 3 ] in the presence of various free Sr 2ϩ concentrations. Sr 2ϩ also decreased the Hill coefficient of the InsP 3 -induced Ca 2ϩ release (Fig. 6A, closed symbols). The EC 50 value for Sr 2ϩ was 8.3 M, indicating that Sr 2ϩ was thus 40 times less potent than Ca 2ϩ in this respect.
The sensitivity of InsP 3 -induced Ca 2ϩ release with respect to InsP 3 was analyzed. The EC 50  in Fig. 6B illustrate the calculated EC 50 for InsP 3 -induced Ca 2ϩ release in the presence of various free Sr 2ϩ concentrations. Sr 2ϩ , in contrast to Ca 2ϩ , lowered the EC 50 for InsP 3 -induced Ca 2ϩ release over the entire concentration range tested (up to 100 M).
Effect of Luminal Ca 2ϩ -The InsP 3 R is stimulated by Ca 2ϩ inside the store (9, 11, 18 -21, 23, 34 -41). Some groups have proposed that this effect could be exerted at the cytosolic side of the InsP 3 R (9, 20, 42). We therefore investigated whether luminal Ca 2ϩ would have the same effect as cytosolic Ca 2ϩ on the cooperativity and the EC 50 of the release. Fig. 7A compares the effect of a gradual increase in [InsP 3 ] starting after 5 min (filled stores, closed symbols) or after 29 min of efflux (less filled stores, open symbols). The stores contained 3.6 times less Ca 2ϩ after 29 min than after 5 min of efflux. The level of store loading had no effect on the threshold for InsP 3 -induced Ca 2ϩ release (Fig. 7B). The EC 50 values were similar (0.38 M InsP 3 for the less filled stores and 0.34 M InsP 3 for the more filled stores), and the Hill coefficients were equal (2.7 for the less filled stores as well as for the more filled stores). The only difference was the rate of Ca 2ϩ release at each [InsP 3 ], which was higher for the filled stores, in accordance with our previous data (22). The stimulatory effect exerted by luminal Ca 2ϩ (no effect on the EC 50 or on the Hill coefficient) therefore completely differed from that induced by stimulatory cytosolic Ca 2ϩ concentrations (a decrease in both parameters). DISCUSSION We made major progress in the identification and localization of high affinity Ca 2ϩ -binding sites on the mouse InsP 3 R-1. These include regions 3, 5, 6, 7, and 10 ( Fig. 1), in addition to regions 15 and 17 that we described earlier (17). These five domains were further subdivided into two parts to map the Ca 2ϩ -binding sites with more precision. Region 3 was located within the InsP 3 -binding domain (residues 226 -578 (43)). More precise mapping revealed that both subregion 3a and 3b, lying between residues 304 -381 and 378 -450, respectively, bound 45 Ca 2ϩ . The other Ca 2ϩ -binding stretches were localized in the modulatory domain (between residues 651 and 2275). Three Ca 2ϩ -binding stretches, regions 5, 6, and 7, were located in the 500-amino acid region just after the InsP 3 -binding domain. 45 Ca 2ϩ binding to these regions was further localized to subregion 5b, lying between residues 660 and 745, to subregion 6a, lying between residues 741 and 849, and to subregion 7b, lying between residues 994 and 1059. Another 45 Ca 2ϩ -binding domain (region 10) was detected in the proximity of the calmodulin-binding site (amino acids 1564 -1585 (44)) and was further localized to subregion 10a, lying between residues 1347 and 1426 (Fig. 8). The present molecular screening data together with the previously described (17)  2146 (d) and in region 17 mapped to the luminal residues 2463-2528 (l) (Fig. 1) give a total of seven cytosolic 45 Ca 2ϩbinding regions and one luminal 45 Ca 2ϩ -binding region for InsP 3 R-1. The short cytosolic C-terminal extension did not bind 45 Ca 2ϩ in our assay.
One of the interesting observations in this study is the demonstration of two 45 Ca 2ϩ -binding domains within the InsP 3binding domain. Up to now, complex effects of Ca 2ϩ on the binding of InsP 3 to its receptor have been reported. An inhibition of the binding was observed in several (14,45,46) but not in all (47) studies. In hepatocytes Ca 2ϩ converts the InsP 3 R from a low affinity to a high affinity conformational state, thereby increasing the binding (10,48). Moreover, recent studies on InsP 3 Rs overexpressed in insect Sf9 cells demonstrate differences in Ca 2ϩ sensitivity of the InsP 3 binding between InsP 3 R isoforms 1 and 3 (49,50). The demonstration in this study of multiple Ca 2ϩ -binding sites in the direct neighborhood of the InsP 3 -binding domain may therefore give a clue to the understanding of these complex effects of Ca 2ϩ on InsP 3 binding.
Colocalization of the Ca 2ϩ -binding domain in region 10a and the CaM-binding domain that is located about 190 residues downstream is reminiscent of what was observed for the ryanodine receptor (RyR), another intracellular Ca 2ϩ release channel. The RyR contains two strong CaM-binding sites separated by less than 200 residues from high affinity Ca 2ϩ -binding sites, and a third CaM-binding site is flanked by a high affinity Ca 2ϩ -binding site (51). A similar arrangement is reported in the plasma membrane Ca 2ϩ -ATPase sequence, where two or three Ca 2ϩ -binding sites flank a CaM-binding site (52). The significance of this colocalization of Ca 2ϩ -and CaM-binding sites is not clear. It is possible that the Ca 2ϩ -and CaM-binding sites interact with each other and that such interactions may modulate Ca 2ϩ and/or CaM binding to these regions.
Different types of Ca 2ϩ -binding domains in proteins have been elucidated by sequence analysis. On the basis of the primary structures responsible for the actual binding of these domains, intracellular Ca 2ϩ -binding proteins can be classified in the following three categories: EF-hand proteins, Ca 2ϩ /phospholipid-binding proteins (annexins and C2 region proteins), and Ca 2ϩ storage proteins (e.g. calsequestrin, calreticulin) (53). We have compared the typical sequence motifs of these three categories of Ca 2ϩ -binding proteins with the Ca 2ϩ -binding regions that we have detected in the InsP 3 R. None of the regions showed resemblance with known motifs. Since the InsP 3 R is structurally related to the RyR, we also compared the Ca 2ϩbinding sequences of the InsP 3 R-1 and the RyR-1, which also contains seven cytosolic Ca 2ϩ -binding regions (51,54). Again we could not detect a significant sequence homology. The different Ca 2ϩ -binding amino acid sequences of the InsP 3 R all contain clusters of aspartic and glutamic acid residues, which are likely to be involved in the Ca 2ϩ binding (Fig. 8). Other Ca 2ϩ -binding proteins, like the RyR (51,54), also have clusters of acidic amino acid residues that are implicated in their Ca 2ϩbinding behavior. Comparison of the different Ca 2ϩ -binding amino acid sequences of the InsP 3 R did not reveal any apparent motifs, repeats, or consensus sequences.
The functional significance of 45 Ca 2ϩ binding to each of the particular regions 3a, 3b, 5b, 6a, 7b, 10a, d, and l remains to be investigated. Some of the detected sites may represent potentially regulatory Ca 2ϩ -binding sites. Direct binding of Ca 2ϩ may induce a profound conformational change within a single domain and as such evoke an interconversion between two (or more) receptor affinity states. Other Ca 2ϩ -binding sites may have a structural role, e.g. structural stabilization of important domains in the neighborhood or protection against proteolysis of an exposed region (52). Ca 2ϩ may also bridge adjacent domains and thus direct the relative membrane orientation and supramolecular structure.
Our observation that there are multiple Ca 2ϩ -binding sites on InsP 3 R-1 is in agreement with our detailed kinetic analysis which points to several different types of interaction of divalent cations with InsP 3 -induced Ca 2ϩ release. For that purpose, we have further characterized the well known bell-shaped activation of the InsP 3 -induced Ca 2ϩ release by cytosolic Ca 2ϩ in permeabilized A7r5 cells by studying the effect of Ca 2ϩ and Sr 2ϩ on the EC 50 and cooperativity of the release measured at different [InsP 3 ].
We first investigated whether our protocol of adding cumulative concentrations of InsP 3 could be used to measure cooperativity. Since the Ca 2ϩ content of the stores at the end of the experiment did not depend on the way of InsP 3 administration (progressively increasing or progressively decreasing) (Fig. 4), we may conclude that InsP 3 -induced inactivation of the InsP 3 R does not occur in A7r5 cells. This contrasts with findings in liver cells (30,31) and in RBL-2H3 cells (55) which may be due to the expression of mainly InsP 3 R-1 in A7r5 cells, whereas the main form in liver and in RBL-2H3 cells is InsP 3 R-2 (56,57). Alternatively, differences in assay conditions may underlie this finding.
The bell-shaped activation of the release by cytosolic Ca 2ϩ (Fig. 5A) represented effects on both the cooperativity and the EC 50 of the release process. Ca 2ϩ exerted a biphasic effect on the EC 50 with a shift toward lower InsP 3 concentrations in the presence of low Ca 2ϩ concentrations and a shift toward higher InsP 3 concentrations at Ca 2ϩ concentrations above 0.3 M. In contrast, no such biphasic effect was observed for the cooperativity, since the Hill coefficient gradually decreased as the free [Ca 2ϩ ] increased (Fig. 6). There are therefore two different mechanisms causing inhibition, a decrease in cooperativity and a shift in EC 50 toward higher InsP 3 concentrations. Sr 2ϩ could only mimic the inhibitory effect on the cooperativity but not on the EC 50 . This different dependence of both types of inhibition on the [Ca 2ϩ ] and on [Sr 2ϩ ] suggests that two different inhib- itory sites were involved. The increase in EC 50 may be related to the Ca 2ϩ -induced inhibition of InsP 3 binding (14,45,46), especially since the inhibition by Ca 2ϩ in A7r5 cells (11,23) and in cerebellum (58,59) can be overcome by increasing the [InsP 3 ].
The stimulation by cytosolic Ca 2ϩ and the associated shift in EC 50 toward lower InsP 3 concentrations probably represent the interaction with a third Ca 2ϩ -binding site. It is not clear as yet whether [ 3 H]InsP 3 binding is increased under these conditions, perhaps as a consequence of isoform diversity in the response to Ca 2ϩ (49,50).
The less steep responses in the presence of stimulatory Ca 2ϩ concentrations (1 M) are not in agreement with the generally accepted view that positive feedback by cytosolic Ca 2ϩ causes the very steep rising phase of the InsP 3 -induced Ca 2ϩ spikes in the intact cell (60). A possible explanation lies in the fact that the inhibition by Ca 2ϩ is a time-dependent process (6, 9 -11, 61). Our experiments were performed after 6.5 min incubation in the high [Ca 2ϩ ], and positive feedback by Ca 2ϩ in the intact cell is assumed to occur before the time-dependent inhibition occurs.
The mechanism by which luminal Ca 2ϩ stimulates the release is still a matter of debate. Some groups (11,17,19,62) believe that luminal Ca 2ϩ binds to a luminal site. Other groups propose a model where luminal Ca 2ϩ leaks out and acts through cytosolic binding sites on the same channel or on closely associated channels (9,20,42). We investigated whether the effect of luminal Ca 2ϩ on the EC 50 and cooperativity of the release (Fig. 7) is similar to the effects exerted by cytosolic Ca 2ϩ described above. The activation of the release by luminal Ca 2ϩ did not affect the EC 50 nor the Hill coefficient. This finding makes it unlikely that the stimulatory effects of luminal Ca 2ϩ in A7r5 cells are exerted on the cytosolic side of the receptor. We therefore continue to propose, at least for A7r5 cells, that effects of luminal Ca 2ϩ do not involve the cytosolic Ca 2ϩ -binding sites (11). Luminal Ca 2ϩ may regulate the Ca 2ϩ release from within the lumen through the luminal Ca 2ϩ -binding site on the InsP 3 R (17) or via associated proteins like calreticulin (39).
In conclusion, we have conducted experiments designed to obtain molecular as well as functional data regarding the mechanism involved in the regulation of the InsP 3 R by Ca 2ϩ . We have provided evidence that multiple Ca 2ϩ -binding regions are located on the InsP 3 R and that at least four different modes of interaction with Ca 2ϩ are involved in the complex feedback regulation of the release by Ca 2ϩ . Further characterization of the regulatory (stimulatory as well as inhibitory) Ca 2ϩ -binding sites is necessary to elucidate the structure-function relationships in the InsP 3 R.