Interaction of Luminal Calcium and Cytosolic ATP in the Control of Type 1 Inositol (1,4,5)-Trisphosphate Receptor Channels*

Ca2+ within intracellular stores (luminal Ca2+) is believed to play a role in regulating Ca2+ release into the cytosol via the inositol (1,4,5)-trisphosphate (Ins(1,4,5)P3)-gated Ca2+ channel (or Ins(1,4,5)P3 receptor). To investigate this, we incorporated purified Type 1 Ins(1,4,5)P3 receptor from rat cerebellum into planar lipid bilayers and monitored effects at altered luminal [Ca2+] using K+ as the current carrier. At a high luminal [Ca2+] and in the presence of optimal [Ins(1,4,5)P3] and cytosolic [Ca2+], a short burst of Ins(1,4,5)P3 receptor channel activity was followed by complete inactivation. Lowering the luminal [Ca2+] caused the channel to reactivate indefinitely. At luminal [Ca2+], reflecting a partially empty store, channel activity did not inactivate. The addition of cytosolic ATP to a channel inactivated by high luminal [Ca2+] caused reactivation. We provide evidence that luminal Ca2+ is exerting its effects via a direct interaction with the luminal face of the receptor. Activation of the receptor by ATP may act as a device by which cytosolic Ca2+ overload is prevented when the energy state of the cell is compromised.

The second messenger inositol (1,4,5)-trisphosphate (Ins(1,4,5)P 3 ) 1 binds to Ins(1,4,5)P 3 receptors in the endoplasmic reticulum, causing release of stored Ca 2ϩ into the cell. The subsequent changes in localized levels of cytosolic-free Ca 2ϩ have been shown to exert both stimulatory and inhibitory effects on the Ins(1,4,5)P 3 receptor (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). This regulation of Ins(1,4,5)P 3 receptors by cytosolic Ca 2ϩ is highly concentrationdependent. Small elevations in cytosolic [Ca 2ϩ ] (Ͻ300 nM) potentiate Ca 2ϩ release and channel opening, whereas higher concentrations are inhibitory (4 -6).Whereas the role of cytosolic Ca 2ϩ is well documented, the potential function of intraluminal Ca 2ϩ within the stores remains controversial. A role for intraluminal Ca 2ϩ has been proposed (16) in which Ca 2ϩ content of the stores modulates the sensitivity of the Ins(1,4,5)P 3 receptor to Ins(1,4,5)P 3 , and there is some evidence supporting this (9,(17)(18)(19)(20)(21)(22)(23)(24)(25), although there is also evidence against it (8, 26 -31). A luminal high affinity Ca 2ϩ binding site has been identified in mouse type I receptor (32) mapped to the nonconserved acidic sub-region of the luminal loop between amino acids 2463 and 2528, although whether it participates in the proposed regulation is unclear. A further possible role for luminal Ca 2ϩ interaction with the Ins(1,4,5)P 3 receptor is in the control of capacitative Ca 2ϩ entry. The idea of conformational coupling between Ins(1,4,5)P 3 receptor and a Ca 2ϩ entry channel was proposed by Irvine (16) and elaborated by Berridge (33). Recently, direct evidence for interaction between Ins(1,4,5)P 3 receptors and Ca 2ϩ entry channels in controlling capacitative Ca 2ϩ entry in a variety of cells has been presented (34 -36). A key question now is whether or not Ins(1,4,5)P 3 receptors are capable of direct transduction of the signal concerning the Ca 2ϩ filling status of the endoplasmic reticulum to plasma membrane Ca 2ϩ entry channels or whether accessory proteins are involved. Planar lipid bilayer experiments allow manipulation of the free [Ca 2ϩ ] on either the cytosolic (cis) or luminal (trans) face of the Ins(1,4,5)P 3 receptor, thereby providing a means for studying intraluminal Ca 2ϩ effects under conditions where the free Ca 2ϩ concentrations on either side are clamped at a set value throughout the experiment. Bezprozvanny and Ehrlich (37) investigated the effects of trans (or intraluminal) Ca 2ϩ using cerebellar microsomes fused to planar lipid bilayers, with Sr 2ϩ , Ba 2ϩ , or Mg 2ϩ as current carrier (the use of K ϩ was precluded by the presence of K ϩ channels in the microsomal membranes). They found that, in contrast to the effects predicted by Irvine (16), intraluminal Ca 2ϩ slightly inhibited channel activity at concentrations greater than 1 mM. However, interpretation of the data is complicated by the possibility of interactions of the divalent cations used as current carriers with potential Ca 2ϩ binding sites on the luminal face of the receptor. To resolve this complication, we have used purified cerebellar Type 1 receptor inlaid into bilayers prepared by the method of Schindler (38). This has enabled us to use K ϩ as current carrier, and since the luminal face of the Ins(1,4,5)P 3 receptor is likely to be exposed physiologically to around 100 mM K ϩ , K ϩ is unlikely to interfere with any physiologically relevant properties of the receptor, unlike Sr 2ϩ and other divalent cations (39,40). Published estimates of luminal free Ca 2ϩ are quite varied depending on the method of measurement (41). However, most recent estimates seem to put the value for a full Ca 2ϩ store at around 0.5 mM (42). Accordingly, we have looked at the effects of luminal Ca 2ϩ at 1 mM (the upper end of the estimate for a full store), 0.6 mM (around the consensus estimate for a full store), 100 M (the lower end of the estimate for a full store and the consensus value for a partially empty store), and 300 nM (a totally depleted store).
In essence, we find that at 1 and 0.6 mM luminal Ca 2ϩ , there is a very rapid run-down of channel activity. This channel activity can be restored by decreasing the Ca 2ϩ on the trans side of the bilayer. Importantly, channel run-down is not ob-served when ATP is present on the cytosolic face of the receptor, and ATP addition can reactivate channels that have been inactivated. We present evidence that the inhibition by luminal Ca 2ϩ cannot be attributed to Ca 2ϩ moving through the channel and exerting its effects on the cytosolic face of the receptor. The detailed channel parameters that we describe strongly support the idea that channel behavior is dependent on luminal Ca 2ϩ concentrations and that, therefore, the receptor is a potential transducer for the filling state of intracellular Ca 2ϩ stores.

MATERIALS AND METHODS
Purification of the Ins(1,4,5)P 3 Receptor-Ins(1,4,5)P 3 receptor was solubilized in 1% CHAPS and purified from rat cerebellum using heparin affinity chromatography as described previously (15). The purified receptor showed only one band (M r ϳ230,000) on a Western blot probed with anti-Type I InsP 3 receptor antibody (rabbit polyclonal, a kind gift from Dr. Colin Taylor, Dept. of Pharmacology, University of Cambridge, UK), indicating that the channel activity measured cannot be attributed to partially degraded Ins(1,4,5)P 3 receptor (Fig. 1). Preparations of the purified receptor were also probed for the presence of calmodulin and FKBP12 (polyclonal goat antibodies, Santa Cruz Biotechnology Inc.), since these have been found to associate with the receptor, and the presence of either one might modify channel activity (44). Recombinant FKBP12 (Sigma RBI) and purified calmodulin (Sigma) were run as standards to determine the migration of FKBP12 and calmodulin, respectively. Although both proteins were readily detectable in cerebellar microsomes and brain homogenates, neither were detectable in the purified Ins(1,4,5)P 3 receptor fractions (Fig. 1). To check the sensitivity of detection of FKBP12, we also included on the blot recombinant FKBP12 standards, (Fig. 1v), which showed that we could readily detect 50 ng of FKBP12 had it been present in the purified preparation. With 20 g of receptor present, a 1:1 association of InsP 3 receptor and FKBP12 would be expected to contain 240 ng of FKBP12, whereas clearly our preparation contained substantially less than 50 ng. FKBP12 has also been found to be absent from purified Ins(1,4,5)P 3 receptor by others. 2 In all cases Western blots were developed using the enhanced chemiluminescence technique.
Formation of Bilayers and Measurement of Single Channel Activity-0.5 ml of lipid solution (Soybean lipid, Sigma; dissolved in hexane at 50 mg/ml) was evaporated onto the surface of a round-bottomed flask using a stream of nitrogen. The electrolyte solution (500 mM KCl, 1 mM CaCl 2 , 40 mM HEPES, pH 7.35), containing glass beads, was added to the flask and thoroughly shaken until a lipid vesicle suspension was formed. Teflon chambers bisected by a thin Teflon film with a hole 75-100 m in diameter were used in these experiments. The hole in the Teflon film was treated on both sides with a coating solution (10 l of hexadecane dissolved in n-pentane, 1:100) before bilayer formation. A small amount of the lipid vesicle solution was placed in each compartment (denoted cis and trans) of the chamber and left for 30 min to allow a stable monolayer to form at the air-buffer interface. The liquid levels on either side were then raised by the addition of the same electrolyte solution to form a bilayer across the hole (38). Electrical connection of the membrane to the measuring equipment was achieved using reversible electrodes (Ag/AgCl). HEDTA (10 mM) or BAPTA (1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid, 1.7 mM) was then added to the cis side to chelate the Ca 2ϩ present to a free Ca 2ϩ concentration of 0.2-0.3 M (43), the optimal concentration for channel activity (6). Purified Ins(1,4,5)P 3 (1 g⅐ml Ϫ1 ) receptor was subsequently added to the cis compartment either as a solubilized preparation or incorporated into proteoliposomes. The method for formation of proteoliposomes was to add 15 g of Ins(1,4,5)P 3 receptor to 1 ml of liposomes (made via the method described but using 500 mM KCl, 40 mM HEPES, pH 7.35, without Ca 2ϩ ), mix, and then incubate on ice for 10 min.
After the addition of 5 M Ins(1,4,5)P 3 to the cis compartment, channel activity was observed at applied potential differences ranging from ϩ100 to Ϫ100 mV using 500 mM K ϩ as the current carrier. Recordings were made via an interface (CED 1401, Cambridge Electronic Design) to a PC running PAT V7.0 software for single channel analysis supplied by J. Dempster, University of Strathclyde (45). The sampling interval was 0. 5 ms, with a 1-kHz low pass filter.
Initially, the concentration of Ca 2ϩ in the trans compartment was 1 mM. To investigate effects of lower concentrations of trans Ca 2ϩ on Ins(1,4,5)P 3 receptor channel activity, HEDTA (final concentration, 10 mM) was added to the trans compartment (the same procedure as for lowering cis [Ca 2ϩ ]), giving a concentration between 200 and 300 nM. Any activity was recorded, as before. For experiments using 100 M Ca 2ϩ in the trans compartment, it was feasible to make bilayers in the presence of 100 M Ca 2ϩ on the trans side by the Schindler (38) technique. The bilayer was thus formed with 500 mM KCl, 40 mM HEPES, pH 7.35, and 100 M CaCl 2 on the trans side and the same solution but containing 1 mM CaCl 2 on the cis side. cis Ca 2ϩ was then chelated to 300 nM as before. Attempts to form reproducible Schindler-type (38) bilayers at lower Ca 2ϩ concentrations were unsuccessful, although bilayers formed at higher Ca 2ϩ concentrations were stable after chelation of Ca 2ϩ to low values. For data analysis, channel currents recorded by the PAT software were analyzed initially by the statistical part of the software. Channel lifetime analysis was further refined using ENZFITTER.

Channel Activity at 1 mM Intraluminal (trans) Ca 2ϩ -Ini-
tially, experiments were carried out with 1 mM Ca 2ϩ on the trans side and 300 nM on the cis side. The cis side is defined as the cytosolic face of the receptor by being the side to which Ins(1,4,5)P 3 is added. A typical recording is shown in Fig. 2. No channel activity was observed until after addition of Ins(1,4,5)P 3 (5 M) to the cis compartment. After Ins(1,4,5)P 3 addition, a short burst of channel activity was seen. The time that elapsed between Ins(1,4,5)P 3 addition and commencement of channel activity varied between experiments, probably because it was not possible to mix the contents of the compartment rapidly and maintain the bilayer. However, activity normally began within 60 s and ceased completely after a further few seconds. Typical channel activity is shown in Fig. 2A, where the applied p.d. was Ϫ100 mV. The traces have been expanded, and details of the individual channels can be seen in indicating inhibition by 1 mM luminal Ca 2ϩ . However, the channel was still present and intact in the bilayer because it could be re-activated by potential changes (see below and In some experiments, reversing the applied p.d. from Ϫ100 to ϩ100 mV resulted in brief re-activation of the channel before it inactivated again. A typical example of this behavior is shown in Fig. 3. Effets of 300 nM Intraluminal (trans) Ca 2ϩ -After adjusting the trans [Ca 2ϩ ] to 300 nM, channel activity recommenced (Fig.  4A). After the initial channel activation and inactivation at Ϫ100 mV, bursts continued for the duration of the experiment or until the addition of heparin (15 g/ml) to the cis compartment. A selection of typical traces from this record was expanded and can be seen in Fig. 4B. The characteristics of the channels observed at 300 nM trans Ca 2ϩ are very different from those seen at 1 and 0.6 mM trans Ca 2ϩ . An all-points amplitude histogram of the complete record (not shown) gave a value of 29.6 Ϯ 8. Extremely long open times were also observed (in the region of tens of milliseconds, e.g. Fig. 4B, i and ii), but they were few in number, making conventional lifetime analysis difficult. They were therefore excluded from the analysis. Long open times have been observed for the Ins(1,4,5)P 3 receptor in previous studies (15). In one particular experiment at 300 nM trans [Ca 2ϩ ], the channel was seen to be open for up to 8 s continuously in a higher conductance state (approximately 55 pS) before flickering between two lower conductance levels (approximately 37 and 24 pS) and finally closing (data not shown). The channel remained closed for approximately 1 min and then suddenly opened again to the highest conductance state (again 55 pS) before exhibiting the same behavior (transition to two lower levels, then closure). Thus, in contrast to the data at 1 mM trans Ca 2ϩ , the channel at 300 nM trans Ca 2ϩ shows much less spontaneous inactivation, longer open times, and a higher conductance as well as multiple conductance states.  Fig. 2A, except that after the channel had inactivated, the applied p.d. was changed to ϩ100 mV. On the first reversal there is no resumption of channel activity. During the second gap in the trace, polarity was restored to Ϫ100 mV, then the switch to ϩ100 mV caused a brief burst of channel activity. Channel characteristics at ϩ100 mV are very similar to those at Ϫ100 mV (see expanded sections i and ii). A subsequent restoration to Ϫ100 mV caused brief channel activity in the opposite direction, and finally there was a short burst of activity on the last switch to ϩ100 mV. The large "up-and-down" deflections are switch artifacts. ] free likely to be found in partly depleted stores. Ins(1,4,5)P 3 receptor was incorporated into the bilayer as described previously via proteoliposomes, and no channel activity was observed in the absence of Ins(1,4,5)P 3 at Ϫ100 mV applied p.d. Almost immediately after the addition of 5 M Ins(1,4,5)P 3 , channel activity started (see Fig. 5A). Four sub-conductance levels were quite clearly seen as is evident both from the selection of channel traces seen in Fig. 5B, i, ii, and iii, and from the all-points amplitude histogram in Fig. 5C. The selection of traces in Fig. 5B, i, ii, and iii, were chosen to illustrate the four levels, which are approximately 12, 24, 36, and 48 pS. The 48-pS state is the highest observed conductance level and appears to be comparable with the high conductance state (55 pS) seen at 300 nM (trans) Ca 2ϩ . The all-points amplitude histogram (see Fig. 5C . Conventional lifetime analysis was carried out on the data shown, excluding extremely long life times (see below). Two populations were detected yielding values of 1.5 Ϯ 0.1 and 8.5 Ϯ 3.2 ms. As for the experiments at 300 nM trans Ca 2ϩ , extremely long openings were observed as well, too numerous to be discounted from the overall analysis. However, inclusion in the lifetime histogram would have obscured the data to such an extent that statistical analysis would not have been possible. Hence, these long lifetimes were calculated as a percentage of the total open time to give an indication of their relevance (see Table I). As can be seen from Fig. 5B, iii, the channel opens quite early in the recording to a 12-pS sub-state and remains in this state for approximately 64% of the total recording time (Table I) Table  I). Trace ii shows three transitions. The full transition from the base line (seen at the beginning of trace A, ii) corresponds to a value of 48 pS. The channel then closes to a lower state seen halfway through the trace, with a value of 24 pS. There is a momentary transition to the 12-pS level before opening to the 24-pS level for the duration of this trace. Trace iii illustrates a 36-pS level (seen as full openings in this trace) with closures to the 12-pS level. C, all-points amplitude histogram. The base line is set at 0 pA, and four additional peaks can be seen: 1.5 Ϯ 0. sitize Ins(1,4,5)P 3 receptors to Ins(1,4,5)P 3 ; hence, high trans [Ca 2ϩ ] should potentiate channel activity at lower Ins(1,4,5)P 3 concentrations. However, in the current investigation, we found no systematic differences between the threshold Ins(1,4,5)P 3 concentrations needed to see channel activity (ϳ500 nM) at 1 mM or 300 nM trans Ca 2ϩ .
Effects of Cytosolic ATP Overcoming the Inhibition by 1 mM trans Ca 2ϩ -In an earlier study on regulation by intraluminal Ca 2ϩ at the single channel level (37), an elevation in the trans Ca 2ϩ level to 1 mM reduced the single channel open probability to 67% of its control level, i.e. there was slight inhibition of Ins(1,4,5)P 3 -gated channel activity. Although, as described above, these experiments were complicated by the use of divalent cations as current carrier, there is a marked difference between this very limited inhibition and the essentially complete (after the first few seconds) inhibition that we observe. One potential factor that may contribute to this discrepancy is that in the earlier work (37) 500 M ATP was present in the cis compartment of the bilayer chamber. We therefore repeated the initial experiments at 1 mM trans Ca 2ϩ in the presence of 100 M ATP in the cis compartment (a value already shown to maximize Ins(1,4,5)P 3 receptor activity in our hands (47)). No channel activity was seen in the presence of 100 M ATP until Ins(1,4,5)P 3 was added.
However, on addition of 5 M Ins(1,4,5)P 3 channel activity was observed (Fig. 6A) at an applied p.d. of Ϯ100 mV. As can be seen from the entire recording (Fig. 6A) and representative traces of the recording (Fig. 6B), there is continuing channel activity at 1 mM trans Ca 2ϩ in the presence of ATP. From the expanded traces of the recording, different sub-states can be seen, and the channel is active at both polarities (Ϯ100 mV) for a very extended period. By comparing Fig. 6A with Fig. 2A, it is very clear that ATP on the cytosolic face of the receptor is somehow overcoming the inhibition induced by 1 mM trans Ca 2ϩ . All-points amplitude histogram analysis (not shown) from traces totaling more than 1000 s of recording time showed opening to one major conductance level (40 Ϯ 2.8 pS (S.D.)), substantially larger than in the absence of ATP, although other sub-conductance levels are evident from individual traces (see Fig. 6B and Fig. 7). The all-points histogram did not define these levels clearly, and for the majority of the recording time, the channel was open to one state. A few very long events were also seen, although the majority are short openings. A histogram for mean open times (not shown) showed two populations: 1.30 Ϯ 0.1 and 6.0 Ϯ 0.8 ms. Under the conditions of Fig. 2, it was found that the addition of ATP to the cis face led to a very rapid reactivation of the channel (Fig. 7), with channel characteristics similar to those seen when ATP was present initially. Thus, cytosolic ATP can reverse as well as prevent the inhibition due to luminal Ca 2ϩ .
Effects of Cytosolic ATP at 300 nM trans Ca 2ϩ -In a contin-uation of the previous experiments, the [Ca 2ϩ ] in the trans compartment was chelated to 300 nM using 10 mM HEDTA to see if lowering luminal [Ca 2ϩ ] had any major effects on channel activity in the presence of 100 M cytosolic (cis) ATP. As can be seen in Fig. 8A (whole recording) and from the selection of traces in Fig. 8B, channel activity continues relatively unchanged at both polarities (Ϯ 100 mV). All-points amplitude histogram analysis yielded one major peak (33.2 Ϯ 4.7 pS (S.D.)) in addition to the closed one, a slightly smaller conductance than that seen at 1 mM trans Ca 2ϩ and 100 M cis ATP. Again, opening to sub-states is not clearly defined from the all-points histogram, as the channel opens to one major conductance level. From the histogram for mean open times two populations were determined, 1.4 Ϯ 0.1 ms and 8.7 Ϯ 2.9 ms, similar to those seen previously.
By comparing experiments at 300 nM trans Ca 2ϩ in the absence and presence of cis ATP, it can be seen that in the presence of ATP the channel shows no signs of any inactivation and continues to gate steadily for very long periods of time (Fig.  8A), unlike the bursting pattern of activity seen in Fig. 4A. That is to say, irrespective of trans [Ca 2ϩ ], ATP still has a stimulatory effect. DISCUSSION A major finding reported here is that at 1 mM luminal Ca 2ϩ the Type 1 Ins(1,4,5)P 3 receptor appears to show only very transient channel activity, unless ATP is present on the cytosolic side. This is consistent with the observations in perfused, permeabilized cell preparations, where ATP increases the amount of Ca 2ϩ release at relatively high Ins(1,4,5)P 3 concentrations without changing the threshold Ins(1,4,5)P 3 concentration required for Ca 2ϩ release (48). However, a key question is whether luminal Ca 2ϩ is exerting its effect by binding to the luminal face of the receptor or in some other way. Since we  have used purified Ins(1,4,5)P 3 receptor in these experiments, we can eliminate the influence of accessory proteins, unless they are very tightly associated with the receptor and not removed during the purification.
There remains the question of whether or not luminal Ca 2ϩ is passing through the channel and having an effect on the cytosolic face, where it is known to have both activating and inhibitory effects (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). We feel that this is very unlikely for several reasons. First, Bezprozvanny and Ehrlich found a value of P Ba /P K of 6.3 in their experiments (37). The unitary current for Ca 2ϩ is lower than for Ba 2ϩ , and we have 500 mM K ϩ present as the major current carrier. This would suggest that in our experiments only about 1% of the single channel current is carried by Ca 2ϩ at 1 mM trans Ca 2ϩ , and the build up of Ca 2ϩ at the cytosolic exit from the channel will, thus, be very small. Since we have 10 mM HEDTA on the cytosolic face, the Ca 2ϩ buffering is very substantial, and precisely similar effects are observed when BAPTA (1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid) is substituted for HEDTA on the cis side. Furthermore, during the prolonged silent phase after the initial burst of channel activity, it is highly unlikely that an inhibitory Ca 2ϩ concentration would remain at the channel mouth for a sufficiently long period to prevent channel reopening over this time scale. In a few experiments, reversing the polarity of the applied p.d. so that cations would move from cytosolic to luminal face caused a transient reactivation. However, this reactivation was very short-lived and is not compatible with the prolonged activation that would be expected if inhibitory [Ca 2ϩ ] was being dissipated from the channel mouth by Ca 2ϩ ions moving back up the channel. The cause of this transient reactivation remains obscure, but it may reflect an effect of the rapid reversal of the p.d. on the conformational state of the receptor. We have also previously found that the purified receptor in bilayers is substantially less sensitive to high cytosolic Ca 2ϩ than the receptor incorporated into bilayers from microsomal membranes (15). Finally, Mak et al. (49) find that cytosolic ATP does not alter inhibition by cytosolic Ca 2ϩ of the type 1 Ins(1,4,5)P 3 receptor from Xenopus oocytes. We cannot, however, eliminate the possibility that the site of action of luminal Ca 2ϩ is actually within the channel domain of the receptor rather than on the luminal face.
In the absence of ATP there are pronounced differences in channel behavior over the range of trans [Ca 2ϩ ] free that we have used in these experiments. This implies that the Ins(1,4,5)P 3 receptor changes its conformational state in response to luminal Ca 2ϩ and is thus capable, at least in theory, of transducing changes in Ca 2ϩ store loading from the cytosolic face of the endoplasmic reticulum to proteins outside of the endoplasmic reticulum such as Ca 2ϩ entry channels. Also, particularly at high luminal [Ca 2ϩ ], the channel shows transient activation followed by inactivation. It is tempting to equate this behavior with the pronounced biphasic kinetics of Ca 2ϩ release from intracellular stores (for review, see Ref. 50), where there is an initial fast phase of Ca 2ϩ release followed by an ongoing slower phase. It should be noted that the experiments at 300 nM trans Ca 2ϩ were continuations of experiments begun at 1 mM trans Ca 2ϩ , the trans Ca 2ϩ being chelated with an appropriate HEDTA concentration. It seems, therefore, that a channel that has inactivated in the presence of high luminal Ca 2ϩ is capable of reactivation by decreasing the luminal Ca 2ϩ . Also, at 100 M trans Ca 2ϩ , the channel shows no sign of inactivation. Physiologically, these effects appear to provide a further form of positive feedback; once a store starts emptying it will continue at an accelerating rate. Decreased inhibition due to decreasing luminal Ca 2ϩ coupled with positive feedback due to released Ca 2ϩ would then act as a "push-pull" mechanism, leading to a very steep channel activation once Ca 2ϩ release is initiated. Particularly at lower trans [Ca 2ϩ ] free , several subconductance states were routinely observed. Our preparations of the Ins(1,4,5)P 3 receptor do not contain detectable levels of FKBP12. Dissociation of this protein from ryanodine receptors has been shown to promote subconductance states in that channel (51). Since the immunosupressant FK506 dissociates FKBP12 and affects Ca 2ϩ fluxes in cerebellar microsomes (52), it is possible that the presence of FKBP12 as an accessory protein in endoplasmic reticulum might lead to changes in the conductance states of the Ins(1,4,5)P 3 receptor also, although recent evidence suggests (53) that the effect on Ca 2ϩ fluxes in this system is probably due to inhibition of the Ca 2ϩ pumps.
It is clear from our observations and those of others that ATP is an extremely important regulator of the Ins(1,4,5)P 3 receptor. Under the conditions used in Fig. 7, it behaves, along with Ins(1,4,5)P 3 and Ca 2ϩ , as a co-agonist for channel activation. Physiologically, this may be very significant. As pointed out by Bezprozvanny and Ehrlich (54), activation of the receptor by ATP may act as a device to prevent cytosolic Ca 2ϩ overload when the energy state of the cell is compromised. Our results would support this view, with the addition that the effect of ATP is particularly marked when the luminal [Ca 2ϩ ] free is very high. It should also be noted that there is a close association between mitochondria and the Ca 2ϩ release sites of the endoplasmic reticulum (55) so that a population of Ins(1,4,5)P 3 receptors may be positioned to monitor directly the rate of ATP production by mitochondria. Although the observations reported here appear to rule out major activating effects of high luminal [Ca 2ϩ ] on Ins(1,4,5)P 3 -stimulated Ca 2ϩ release, this contrasts with a number of publications where this has been observed (9,(17)(18)(19)(20)(21)(22)(23)(24)(25). However, we have studied only the Type 1 receptor, and other isotypes may show different behavior.