The effects of inositol 1,4,5-trisphosphate (InsP3) analogues on the transient kinetics of Ca2+ release from cerebellar microsomes. InsP3 analogues act as partial agonists.

An investigation of the effects of a number of inositol trisphosphate analogues on the transient kinetics of Ca2+ release from cerebellar microsomes was undertaken. All the analogues investigated could release the total Ca2+ content of the inositol 1,4,5-trisphosphate (Ins(1,4,5)P3) mobilizable Ca2+ store; however, their potencies were substantially reduced compared to Ins(1,4,5)P3. The concentration required to induce half-maximal Ca2+ mobilization was 0.14 μM for Ins(1,4,5)P3, 1.8 μM for 3-deoxyinositol 1,4,5-trisphosphate (3-deoxyInsP3), 1.0 μM for 2,3-dideoxyinositol 1,4,5-trisphosphate (2,3-dideoxyInsP3), 24 μM for 2,3,6-trideoxyinositol 1,4,5-trisphopshate (2,3,6-trideoxyInsP3), and 2.9 μM for inositol 2,4,5-trisphosphate (Ins(2,4,5)P3). In all cases and for all concentrations tested, the inositol trisphosphate analogues induced biphasic transient release of Ca2+, which could fit to a biexponential equation assuming two independent processes. The rate constants calculated for the release process were much larger for Ins(1,4,5)P3 than the other inositol trisphosphates (the fast phase rate constant varying from 0.3 to 1.6 s−1 and the slow phase from 0.01-0.5 s−1, at concentrations between 0.03 and 20 μM Ins(1,4,5)P3). The rate constants for all other inositol trisphosphates did not appear to exceed 0.4 s−1 for the fast phase and 0.1 s−1 for the slow phase at their highest concentrations tested. The maximum amplitudes for Ca2+ release by the two phases appeared to be similar for all inositol trisphosphates (approximately 45% for the fast phase and approximately 55% for the slow phase). On comparing the rate constants for Ca2+ release at inositol trisphosphate concentrations for the analogues which all induced the same extent of Ca2+ release, it was apparent that the rates of release were independent of the extent of Ca2+ release. As the extent of Ca2+ release can be related to degree of occupancy of the binding sites, it is evident that different analogues which occupy the binding site of the receptor to the same extent can induce Ca2+ to be released at different rates. We explain this conclusion in terms of partial agonism where inositol phosphates can induce two (or more) occupied states of the channel.

Certain hormones and neurotransmitters induce cells to produce the second messenger inositol 1,4,5-trisphosphate (Ins(1,4,5)P 3 ), 1 which opens an InsP 3 -sensitive calcium channel (the InsP 3 receptor) causing the elevation of cytosolic Ca 2ϩ concentrations (1). This then activates a diverse multitude of cellular processes which depend on the cell type (1). InsP 3induced Ca 2ϩ release is a complex process which, despite considerable study, remains poorly understood. Several studies have demonstrated that the InsP 3 -induced Ca 2ϩ release is "quantal" in nature where submaximal concentrations of Ins(1,4,5)P 3 are unable to fully discharge the InsP 3 -sensitive Ca 2ϩ pool unless maximal Ins(1,4,5)P 3 concentrations are added (2,3). The mechanism for quantal Ca 2ϩ release induced by Ins(1,4,5)P 3 remains unknown, even though this process appears to be adopted by other intracellular Ca 2ϩ channels (4). Attempts to explain this mechanism have lead to the proposal of several models based on either the existence of heterogeneous Ca 2ϩ stores which contain Ca 2ϩ channels with different sensitivities to InsP 3 , discharging their Ca 2ϩ in an all-or-nothing manner (5), or homogeneous Ca 2ϩ stores which discharge their Ca 2ϩ in a regulated fashion, possibly controlled by luminal Ca 2ϩ , Ca 2ϩ gradients, or limited desensitization (6 -8).
A more elaborate approach to investigating InsP 3 -induced Ca 2ϩ release is to study the transient kinetics of this process. Such studies using permeabilized hepatocytes, basophilic leukemia cells, cerebellar microsomes, and purified-reconstituted InsP 3 receptors have shown InsP 3 -induced Ca 2ϩ release to be a relatively fast and biphasic process (9 -12). These studies have therefore proved useful in aiding our understanding into the mechanism of channel opening.
The use of InsP 3 analogues in studies of InsP 3 -induced Ca 2ϩ release and binding to the channel have also helped shed light on our understanding of the pharmacological and functional properties of this transport protein (3,13). Many of these analogues are able to release Ca 2ϩ to the same extent and in a similar manner to Ins(1,4,5)P 3 , albeit with much lower affinities/potencies and have therefore been classified as "full agonists" (3). In addition, some inositol phosphate analogues are able to bind to the channel and completely displace bound [ 3 H]Ins(1,4,5)P 3 , yet unable to induce Ca 2ϩ release, and these have been classified as "full antagonists" (3).
In this study we have investigated the effects of a variety of inositol trisphosphate analogues (all of which are full agonists) on the transient kinetics of Ca 2ϩ release with the view to understanding how these ligands influence the mechanism of channel opening.
‡ To whom correspondence should be addressed. Bio-Rad, and phosphocreatine, creatine kinase, and ATP were purchased from Boehringer Mannheim. All other reagents were of analytical grade.
In order to avoid any calcium contamination when using the inositol trisphosphate analogues, all InsP 3 analogues were dissolved in double deionized water and treated with Chelex 100 resin prior to use.
Rat cerebellar microsomes were prepared as described previously (14). Briefly, approximately 20 cerebella were homogenized in 10 volumes of buffer containing 0.32 M sucrose, 5 mM Hepes, pH 7.4, in the presence of 0.1 mM phenylmethylsulfonyl fluoride, 10 M leupeptin, 10 M pepstatin A, and 50 M benzamidine and then centrifuged for 10 min at 500 ϫ g. The pellet was resuspended in 5 volumes of the same buffer and again centrifuged as above. The resulting supernatants were pooled and centrifuged for 20 min at 10,000 ϫ g. The supernatant gained from this step was then centrifuged for 1 h at 100,000 ϫ g, and the resulting pellet was resuspended in approximately 2 ml of the buffer and snap-frozen in liquid nitrogen and stored at Ϫ70°C.
Calcium Uptake and Release Experiments-Ca 2ϩ uptake and release were measured as described in Ref. 14. Typically, 300 g of rat cerebellar microsomes were suspended in 2 ml of 40 mM Tris/phosphate buffer, pH 7.2, at 37°C in the presence of 10 mM phosphocreatine, 10 g/ml creatine kinase, and 1.25 M fluo-3. The mixture was incubated at 37°C, and Ca 2ϩ uptake was initiated by the addition of 1.5 mM MgATP. Ca 2ϩ transport across the microsomal membrane was followed by monitoring the fluorescence change of fluo-3 in a Perkin-Elmer LS-50B fluorimeter by exciting at 505 nm and measuring the emission at 525 nm. When sufficient Ca 2ϩ was taken up, the pump was inhibited by the addition of between 75 and 150 M orthovanadate (which was found to cause more than 90% inhibition (15)). When the Ca 2ϩ level reached a new steady state, InsP 3 (or the analogues) were added, and fluorescence change was measured. The amount of Ca 2ϩ release was expressed as a percentage of that which was released by 12.5 g/ml A23187. Fluorescence intensity was related to [Ca 2ϩ ] using Equation 1 where K d is the dissociation constant for Ca 2ϩ binding to fluo-3 (900 nM at 37°C (16)), F is the fluorescence intensity of the sample, and F min and F max are the fluorescence intensities of the sample in 1.5 mM EGTA and Ϸ1.7 mM CaCl 2 , respectively.

Stopped Flow Measurements of InsP 3 -induced Ca 2ϩ
Release-Rapid kinetic measurements of IICR were carried out as described in Refs. 14 and 17. Ca 2ϩ uptake by the microsomes was followed in a conventional fluorimeter and inhibited with orthovanadate once sufficient uptake had taken place. It was then transferred to a syringe A of a stopped flow spectrofluorimeter (Applied Photophysics, Model SX17 MV). Syringe B was filled with InsP 3 analogues at a concentration 10 times that required since the mixing ratio of syringe A to syringe B is 10:1. The temperature of the syringe compartment was maintained at 37°C by a circulating water bath. The fluorescence change of fluo-3 was then monitored by exciting the sample at 505 nm and measuring the emission above 515 nm using a cut-off filter. The data were collected, and an average was determined from between 8 and 12 traces. Fluorescence intensities were then correlated to [Ca 2ϩ ] by comparing the traces to identical experiments carried out in a conventional fluorimeter. The traces were analyzed using nonlinear regression analyses programs supplied by Applied Photophysics and Biosoft. InsP 3 -induced Ca 2ϩ release in the microsomal preparation under study was shown to be biphasic and fit well to a biexponential equation (Equation 2) which assumes two independent processes.
where A 1 , A 2 , k 1 , and k 2 are the amplitudes (relative extent of Ca 2ϩ release) and rate constants of Ca 2ϩ release for the fast and slow phases, respectively, and t is the time (s). Over the Ca 2ϩ concentration range for which the IICR was monitored, the fluorescence change when related to [Ca 2ϩ ] was around the K d value for Ca 2ϩ binding to fluo-3, and, over this range, the fluorescence was shown to be linearly related to [Ca 2ϩ ] (linear regression coefficient, r Ͼ 0.99). Fig. 1 shows the dose-dependent response of several InsP 3 analogues on Ca 2ϩ release. All the analogues acted as full InsP 3 agonists with respect to the extent of Ca 2ϩ release since they all appear to give (or approach) the maximal response to that observed with optimal concentrations of Ins(1,4,5)P 3 . In addition, all the analogues exhibited quantal release behavior. In the preparation used in this study, the maximal extent of Ca 2ϩ release was 23 Ϯ 3% compared with that released by A23187, which was induced by approximately 3-10 M Ins(1,4,5)P 3 . The concentration of Ins(1,4,5)P 3 causing halfmaximal response (IC 50 ) was found to be 0.14 Ϯ 0.03 M. The deoxy derivatives of InsP 3 were found to be less potent in releasing Ca 2ϩ . 3-DeoxyInsP 3 had an IC 50 of 1.8 Ϯ 0.2 M, while 2,3-dideoxyInsP 3 had an IC 50 of 1.0 Ϯ 0.2 M. The 2,3,6-trideoxyInsP 3 was much less potent, requiring greater than 0.3 mM to reach maximal release (IC 50 24 Ϯ 2 M). In addition to the deoxyInsP 3 analogues, the isomer Ins(2,4,5)P 3 was also shown to have a lower potency than Ins(1,4,5)P 3 , requiring greater than 30 M for maximum Ca 2ϩ release (IC 50 for Ca 2ϩ release 2.9 Ϯ 0.3 M). The apparent cooperativity of the extent of Ca 2ϩ release was, however, found to be similar for all analogues tested (Hill coefficients were 1.0 Ϯ 0.2 in all cases). Fig. 2 shows the time-resolved Ca 2ϩ release traces induced by Ins(1,4,5)P 3 ( Fig. 2A), Ins(2,4,5)P 3 (Fig. 2E), and the three deoxy analogues (3-deoxy-, Fig. 2B; 2,3-dideoxy-, Fig. 2C; 2,3,6trideoxy-, Fig. 2D). As shown, each individual trace could be fitted extremely well to the biexponential equation (Equation 2), and all 2 values for these fits were less than 0.1.

RESULTS
The resulting rates of Ca 2ϩ release obtained from these fits show that both fast and slow rate constants increase with FIG. 1. Quantal Ca 2؉ release from rat cerebellar microsomes. Quantal Ca 2ϩ release induced by inositol phosphates were monitored over a range of concentrations as described under "Materials and Methods." The extent of Ca 2ϩ release was plotted as a percent of that released by the inositol phosphates compared to that released by A23187. Ins(1,4,5)P 3 , q; 3-deoxyInsP 3 , f; 2,3-dideoxyInsP 3 , å; 2,3,6-trideoxyInsP 3 , ç; Ins(2,4,5)P 3 , ࡗ. The fractional Ca 2ϩ release was calculated from the maximum release induced by 20 M Ins(1,4,5)P 3 (which corresponded to 23 Ϯ 3%). The data points are the mean Ϯ S.D. of 3 or more determinations.
increasing agonist concentration and in some cases appear to saturate (Fig. 3, A and B). The fast phase rate constants for Ca 2ϩ release with Ins(1,4,5)P 3 increases from 0.3 to 1.5 s Ϫ1 at concentrations between 30 nM and 20 M (Fig. 3A) and from 0.02 s Ϫ1 to 0.5 s Ϫ1 for the slow phase over the same concentration range (Fig. 3B). A Hill coefficient of Ϸ1.0 was calculated from the rate constants of the fast phase with Ins(1,4,5)P 3 , which is in agreement with the findings of Finch et al. (18), but at variance with the findings presented in Refs. 9 and 10.
Although the deoxyInsP 3 analogues released Ca 2ϩ in a similar fashion to Ins(1,4,5)P 3 , with the release process in each case also being biphasic, the rate constants for Ca 2ϩ release were, however, much lower than those of Ins(1,4,5)P 3 . Higher concentrations of the InsP 3 analogues were required in order to attain the rate constants for Ca 2ϩ release observed with even the lowest concentrations of Ins(1,4,5)P 3 used. When comparing the rate constants for both the fast and slow phases at the maximum concentrations of the analogues used, the values were never greater than 25-30% of those observed with 10 M Ins(1,4,5)P 3 . Fig. 4, A and B, shows that the amplitudes (or extent) of Ca 2ϩ release by these analogues also increase with increasing analogue concentrations. These results demonstrate that, in this preparation, the contribution of the fast phase to the total amplitude (or extent) of Ca 2ϩ release by Ins(1,4,5)P 3 was Ϸ45% and the slow phase was Ϸ55%. The proportions each phase contributes to the total extent of Ca 2ϩ release appears to be independent of InsP 3 analogues used since fast and slow phase amplitudes all reached similar levels of release. These results clearly indicate that the contribution each phase has on the total extent of Ca 2ϩ release is predefined, for any particular preparation used, as only a maximum specified extent of Ca 2ϩ can be released in association with any given phase.
Since there are differences in the potencies between InsP 3 analogues such that different concentrations are required in order to release similar levels of Ca 2ϩ , a comparison of the fast and slow rate constants for Ca 2ϩ release at analogue concentrations which release 30% and 80% of the total InsP 3 -releasable Ca 2ϩ pool was undertaken. Fig. 5, A and B, shows the relationship between the extent of Ca 2ϩ release and the rate constants, both at 30% Ca 2ϩ release or 80% Ca 2ϩ release. This figure clearly demonstrates that there is no direct correlation between the extent of Ca 2ϩ release and the rate constants. Different analogues at concentrations which release the same extent of Ca 2ϩ do so at different rates. DISCUSSION The kinetic properties of the InsP 3 -sensitive Ca 2ϩ channel remains something of a mystery despite extensive study. The inability of low concentrations of InsP 3 to mobilize all the InsP 3 -sensitive Ca 2ϩ stores has led to the development of several theoretical models (5-8), many of which have not held up  Ins(1,4,5)P 3 , q; 3-deoxyInsP 3 , f; 2,3-dideoxyInsP 3 , å; 2,3,6-trideoxy-InsP 3 , ç; and Ins(2,4,5)P 3 , ࡗ.
to rigorous testing (3). In the hope of gaining further insights into how this channel operates, we employed stopped-flow spectrofluorimetry to study the transient kinetics of Ca 2ϩ release from cerebellar microsomes using a number of different InsP 3 analogues.
Our data show good agreement with those of a previous study by Kozikowski et al. (19) using these deoxyInsP 3 analogues. They investigated the effects of these analogues on the extent of Ca 2ϩ release and their ability to displace [ 3 H]Ins(1,4,5)P 3 and showed that Ins(1,4,5)P 3 was the most potent at inducing Ca 2ϩ release, 2,3,6-trideoxyInsP 3 was the least potent, and the others were in between. These results demonstrate the importance of hydroxy groups and their positions on the inositol ring, in activating the channel. Detailed analysis of their data showed a direct correlation between Ca 2ϩ releasing ability (concentration required for half-maximal release) and binding ability (measured as the concentration required to half-maximally displace bound [ 3 H]Ins(1,4,5)P 3 ) (Fig.  6). This correlation, which was initially reported for several other inositol phosphate analogues (20), also holds when comparing our IC 50 values for Ca 2ϩ release with the K i values determined for binding as given in Ref. 19 (Fig. 6), even though in these studies the binding measurements were done under experimental conditions different from those for Ca 2ϩ release. Such a correlation has led to the proposal that the potency of all inositol phosphates in opening the channel is directly related to occupancy, and, thus, low efficacy inositol phosphates require higher concentrations in order to occupy the binding site before they induce channel opening. It is therefore widely believed that, for full agonists, the extent of occupancy directly relates to the extent of channel opening and therefore presumably the Ca 2ϩ release process would be similar for all analogues which are occupying the receptor to the same extent. However, it is clear from Fig. 5 that the rate constants for Ca 2ϩ release are not the same for any given extent of release, they are dependent upon the structure of the InsP 3 analogue and the concentration used. From this observation it must be concluded that the InsP 3 receptor can distinguish between different InsP 3 analogues occupying the binding site and alter the rate at which Ca 2ϩ flows through the channel accordingly.
One possibility to explain such an observation would be to assume the existence of 2 occupied states of the receptor. This assumption is not inconceivable since some inositol phosphates can act as antagonists, e.g. InsP 6 etc. (3). These inositol phosphates were shown to bind to the receptor displacing [ 3 H]Ins(1,4,5)P 3 without inducing Ca 2ϩ release. Thus, in one state the receptor is occupied by an inositol phosphate antagonist which stabilizes the receptor in an occupied but "unproductive" conformation, while in the other state, stabilized in the presence of Ins(1,4,5)P 3 , the receptor adopts an occupied and "productive" conformation which leads to channel opening. Presumably these two states exist due to different contacts being made by the inositol phosphates in the binding domain of the receptor. If these two states exist in dynamic equilibrium, it could be envisaged that some inositol phosphates could in fact induce both conformational states at any given time of occupancy, sometimes the ligand would form contacts which lead to channel opening (productive state), and sometimes it would not (unproductive state) (see Scheme 1). Assuming full occupancy, in such a scheme the extent of Ca 2ϩ release would eventually be the same as that observed for Ins(1,4,5)P 3 , but by virtue of the fact that since some of the time the receptor would be in a productive state, eventually all the mobilizable Ca 2ϩ would be released. However, the rates for Ca 2ϩ release would be slower than those observed for Ins(1,4,5)P 3 . The rate constants for Ca 2ϩ release would therefore depend upon the forward and backward rate constants for these two states and thus equilibrium constant for the two states induced by a particular inositol phosphate. In such a mechanism, the equilibrium constant for Ins(1,4,5)P 3 would be much greater than that for 2,3,6-trideoxyInsP 3 .
An additional mechanism for such a process could depend upon the fact that since multiple conductance states have been reported for the InsP 3 receptor (21), different inositol phos- FIG. 5. Relationship between extent and rate constants for Ca 2؉ release for the different analogues. Rate constants for Ca 2ϩ release of the fast phase (white bar) and slow phase (black bar) by different concentrations of InsP 3 analogues which release Ca 2ϩ to the same extent. A represents the rate constants for inositol phosphates at concentrations which release Ϸ80% of the InsP 3 -mobilizable Ca 2ϩ pool. The concentrations were 1 M for Ins(1,4,5)P 3 , 5 M for 3-deoxyInsP 3 , 3 M for 2,3-dideoxy InsP 3 , 100 M for 2,3,6-trideoxy InsP 3 , and 20 M for Ins(2,4,5)P 3 . B represents the rate constants for inositol phosphates at concentrations which release Ϸ30% of the InsP 3 -mobilizable pool. The concentrations were 0.05 M for Ins(1,4,5)P 3 , 1 M for 3-deoxyInsP 3 , 0.5 M for 2,3-dideoxyInsP 3 , 10 M for 2,3,6-trideoxyInsP 3 , and 1 M for Ins(2,4,5)P 3 . phate analogues may preferentially induce the channel to adopt the lower conductance states thus reducing the rate of Ca 2ϩ efflux through the channel. High conductance states would only be induced by Ins(1,4,5)P 3 .
As the rate constants for Ca 2ϩ release never reach the values of those observed for Ins(1,4,5)P 3 , these analogues must therefore be classified as "partial agonists." Such a model involving two or more occupied states, as outlined here, has also been proposed to explain partial agonism in other types of receptors (22).
Recently, another class of inositol phosphates which includes Ins(1,3,4,6)P 4 , L-ch-Ins(2,3,5)PS 3 , and D-6-deoxy-myo-Ins(1,4,5)-PS 3 (3,23,24) have been shown to induce Ca 2ϩ release through the InsP 3 receptor, but are unable to release Ca 2ϩ to the same extent as seen with maximal concentrations of Ins(1,4,5)P 3 . These analogues have also been classified as partial agonists. Therefore, in order to avoid confusion between these two classes, we suggest that partial agonists of the type which cannot mobilize all the Ins(1,4,5)P 3 mobilizable Ca 2ϩ pool be classified as class I, whereas partial agonists, of the type described in this paper which can mobilize all the Ins(1,4,5)P 3 releasable Ca 2ϩ pool but whose rates are lower than those observed with Ins(1,4,5)P 3 , be classified as class 2. The nature of the partial agonism by class 1 is yet to be investigated; however, if we assume that each phase is due to a distinct population of Ca 2ϩ stores which have functionally different InsP 3 receptors, it would be tempting to speculate that these agents may selectively affect only one of these two populations. FIG. 6. Correlation between binding affinities and Ca 2؉ releasing potencies of the inositol phosphates. A plot of the binding affinities (measured as the K i which is the concentration of the inositol phosphate which displaces 50% of bound [ 3 H]Ins(1,4,5)P 3 and IC 50 for Ca 2ϩ release. The K i values and the IC 50 for release from permeabilized SH-SY5Y cells (q) were taken from Ref. 19. The IC 50 values determined for rat cerebellar microsomes in this study, f. SCHEME 1