Mechanistic basis of bell-shaped dependence of inositol 1,4,5-trisphosphate receptor gating on cytosolic calcium

Abstract

The inositol 1,4,5-trisphosphate (IP₃) receptor (IP₃R) is an intracellular Ca²⁺ release channel, and its opening is controlled by IP₃ and Ca²⁺. A single IP₃ binding site and multiple Ca²⁺ binding sites exist on single subunits, but the precise nature of the interplay between these two ligands in regulating biphasic dependence of channel activity on cytosolic Ca²⁺ is unknown. In this study, we visualized conformational changes in IP₃R evoked by various concentrations of ligands by using the FRET between two fluorescent proteins fused to the N terminus of individual subunits. IP₃ and Ca²⁺ have opposite effects on the FRET signal change, but the combined effect of these ligands is not a simple summative response. The bell-shaped Ca²⁺ dependence of FRET efficiency was observed after the subtraction of the component corresponding to the FRET change evoked by Ca²⁺ alone from the FRET changes evoked by both ligands together. A mutant IP₃R containing a single amino acid substitution at K508, which is critical for IP₃ binding, did not exhibit this bell-shaped Ca²⁺ dependence of the subtracted FRET efficiency. Mutation at E2100, which is known as a Ca²⁺ sensor, resulted in ∼10-fold reduction in the Ca²⁺ dependence of the subtracted signal. These results suggest that the subtracted FRET signal reflects IP₃R activity. We propose a five-state model, which implements a dual-ligand competition response without complex allosteric regulation of Ca²⁺ binding affinity, as the mechanism underlying the IP₃-dependent regulation of the bell-shaped relationship between the IP₃R activity and cytosolic Ca²⁺.

The inositol 1,4,5-trisphosphate (IP₃) receptor (IP₃R) is a dual-ligand–gated Ca²⁺ release channel whose opening is controlled by IP₃ and Ca²⁺ (1) and which plays a crucial role in the generation of Ca²⁺ signals that control numerous cellular processes (2). Individual IP₃R subunits possess a single IP₃ binding site (3) and multiple Ca²⁺ binding sites (4, 5), and tetrameric complexes of these subunits form functional IP₃-gated Ca²⁺ release channels (6). Cytoplasmic Ca²⁺ regulates IP₃R in a biphasic manner: Ca²⁺ release is potentiated at low Ca²⁺ concentrations but inhibited at higher Ca²⁺ concentrations (7, 8). The stimulatory effect suggests that the channels display the process of Ca²⁺-induced Ca²⁺ release, which underlies Ca²⁺ spike generation and wave propagation. In other words, the bell-shaped dependence on cytosolic Ca²⁺ is the fundamental property of IP₃R for the generation of Ca²⁺ excitability (9).

IP₃ monotonically activates the IP₃R channels at constant Ca²⁺ concentrations (10), but IP₃ dynamically changes the Ca²⁺ sensitivity of the channel (11, 12). At subsaturating concentrations of IP₃, the optimal Ca²⁺ concentration for IP₃R modulation becomes lower, whereas at very high concentrations of IP₃, channel activity persists at supramicromolar Ca²⁺ concentrations (11, 12). This mechanism of dual-ligand regulation of the IP₃R channel has attracted considerable interest, but the molecular dynamics underpinning this mechanism of IP₃R channel gating is still controversial. Marchant and Taylor (13) have proposed that IP₃ binding evokes a rapid conformational change that exposes a high-affinity Ca²⁺ binding site, to which Ca²⁺ must bind before the channel can open. This model suggests that Ca²⁺, but not IP₃, directly activates the IP₃R channel. Foskett and colleagues (12) have proposed that Ca²⁺ is the true agonist of IP₃R, whereas IP₃ acts as a regulatory factor that simply reduces the sensitivity of the receptor to the inhibition caused by high concentrations of Ca²⁺. Foskett and colleagues (14) have also proposed a triumvirate of Ca²⁺ binding sites that are involved in channel gating: IP₃-independent Ca²⁺ binding sites responsible for channel activation, IP₃-independent Ca²⁺ binding sites responsible for channel inactivation, and IP₃-dependent Ca²⁺ binding sites that have opposite functions (activation or inhibition) depending on IP₃ binding. Evidence that substitution of a glutamate residue (E2100) in type 1 IP₃R (IP₃R1) impairs (by 10-fold) Ca²⁺ sensitivity for both Ca²⁺-dependent activation and inactivation of IP₃R1 highlights a potential role of this residue in sensing Ca²⁺ (15, 16). However, whether E2100 is involved in the IP₃-induced high-affinity Ca²⁺ binding site (13) or in one of three Ca²⁺ binding sites (14) is not known. No other Ca²⁺ binding sites involved in the regulation of IP₃R channel activity have been identified. Conversely, a further model has been proposed in which IP₃ is the functional ligand, inducing conformational changes in the N-terminal IP₃ binding domain that are mechanically transmitted to the opening of the pore through an attachment to the linker region between the fourth and fifth transmembrane regions (17).

Observation of the conformational changes evoked by IP₃ and/or Ca²⁺ should facilitate understanding of the mechanism responsible for ligand regulation of IP₃R channel activity. The N-terminal IP₃ binding domain is critical for functional coupling between IP₃ binding and channel opening (17–20), and thus the relative position of the N terminus may contain the information that regulates the conductance of the IP₃R channel. In this study, we detected both IP₃ binding- and Ca²⁺ binding-induced conformational changes in the IP₃R channel by measuring FRET between N-terminally fused fluorescent proteins on individual IP₃R subunits. We found that the effects of these two ligands on the molecular conformation of the channel are different and that the combined effect is not a simple summative response. This approach offers unique insight into the mechanism of dual-ligand regulation of channel activation and inactivation of IP₃R.

Results

Fluorescent Protein–IP₃R Fusion Proteins.

To monitor the ligand-induced conformational changes in mouse IP₃R1, enhanced cyan fluorescent protein (ECFP) or an improved yellow fluorescent protein, Venus, was fused to the N terminus of IP₃R1 (cR and vR, respectively) (Fig. 1A). Stable cell lines expressing cR or vR were established from the intrinsic IP₃R-deficient DT40 cells (21). Fig. 1B shows the Western blot analysis of the membrane fractions prepared from established cells by using anti-IP₃R1 monoclonal antibody 18A10 (22). The measurements of cytosolic Ca²⁺ concentration ([Ca²⁺]) changes evoked by B-cell receptor stimulation showed that both cR and vR retain IP₃-induced Ca²⁺ release activity (Fig. S1). Cells expressing both cR and vR also exhibited Ca²⁺ increases after B-cell receptor stimulation (Fig. S1). The apparent IP₃ binding dissociation constant of vR expressed in Sf9 cells was 21 nM (n = 2), which is consistent with that of wild-type (wt) IP₃R1 (28.6 nM) (23). Therefore, the N-terminal fusion did not interfere with the IP₃ binding. IP₃-gated Ca²⁺ release channels are composed with homotetrameric and heterotetrameric complexes (6). Molecular mass of cR was measured by size-exclusion column chromatography (Fig. 1C). IP₃R proteins were solubilized with 0.1% Nonidet P-40 from microsomal membranes prepared from HeLa cells and were applied to a size-exclusion column TSK-gel G4000SW. Exogenously expressed cR was eluted in similar fractions to those of endogenous IP₃R1 and exogenous wt IP₃R1 (Fig. 1C), which was coimmunoprecipitated with cR (Fig. S2). These results indicate that fluorescent protein did not interfere with the tetrameric formation of IP₃R1. Fig. 1D shows the subcellular distribution of cR and vR expressed in the same HeLa cells. ECFP and Venus signals were almost overlapped, suggesting that both cR and vR are similarly distributed on the endoplasmic reticulum (ER) in HeLa cells.

Fig. 1.

Fluorescent protein–IP₃R1 fusions. (A) ECFP or Venus was fused to the N terminus of IP₃R1 with linker sequences. (B) Western blot analysis of cR and vR. Membrane proteins (10 μg) prepared from intrinsic IP₃R-deficient DT40 cells (cell) and stable cell lines expressing wt, cR, or vR were applied. Recombinant IP₃Rs were detected with the antibody 18A10. A molecular size marker is shown on the left in kDa. (C) Lysates prepared from nontransfected HeLa cells (cell) and cells transfected with wt and cR cDNAs (cR/wt) were subjected to size-exclusion chromatography. Fractions from 9 to 26 were analyzed by Western blot using an anti-GFP antibody and 18A10. Black and red arrowheads indicate the position of wt and cR, respectively. The results were representative of three independent experiments. (D) HeLa cells transfected with cR and vR cDNAs were excited at 440 nm, and fluorescent signals (460–510 nm for ECFP and 515–620 nm for Venus) were monitored by confocal microscopy. (Scale bar: 10 μm.)

Cytosolic Ca²⁺ dependence of fluorescent protein-tagged IP₃R1 was measured by ER luminal Ca²⁺ imaging (Fig. S3). As shown in Fig. S3A, the addition of 1 μM IP₃ rapidly reduced the amount of Ca²⁺ in the ER lumen in permeabilized HeLa cells. The rate of decrease in luminal Ca²⁺ monotonically depends on the concentration of cytosolic Ca²⁺ within the range examined (Fig. S3B). When wt IP₃R1 was expressed in HeLa cells, the rate of decrease in luminal Ca²⁺ was substantially modified, and this modification exhibited a bell-shaped dependence on cytosolic Ca²⁺ with a peak at ∼0.4 μM Ca²⁺ (Fig. S3C). The expression of the E2100Q mutant did not increase the rate of luminal Ca²⁺ decrease within the range examined (Fig. S3D), suggesting that the Ca²⁺ dependence of wt IP₃R1 channel activity directly reflects the modification of the rate of luminal Ca²⁺ decrease. We found that HeLa cells expressing vR exhibit a similar biphasic Ca²⁺ dependence on the modification of the rate of luminal Ca²⁺ decrease (Fig. S3E). These results indicate that N-terminally fluorescent protein-tagged IP₃R1 is functional on the ER membrane in permeabilized HeLa cells and that the N-terminal fusion does not alter the Ca²⁺ dependence of IP₃R1 channel activity.

The diameter of a tetrameric IP₃R is ∼20 nm (24, 25). When both cR and vR exist within a single tetramer, FRET from ECFP to Venus may occur because the Förster distance—the distance at which FRET efficiency is 50%—of the pair of ECFP and Venus is ∼5 nm. By acceptor (Venus) photobleaching (Fig. S4A), the ECFP fluorescence intensity was increased to 118.0 ± 8.8% in HeLa cells expressing both cR and vR (n = 92). In HeLa cells transiently expressing cR or vR, the amount of exogenous fluorescent IP₃R1 proteins was comparable with that of endogenous IP₃R1 (Fig. S4B). These results indicate that a substantial amount of tetrameric IP₃R complexes contain cR and vR and that the intermolecular FRET between cR and vR occurs in HeLa cells.

Measurements of FRET Efficiency at Various Ligand Conditions.

To measure IP₃-dependent conformational changes of IP₃R1, HeLa cells expressing both cR and vR were permeabilized with 60 μM β-escin after treatments with 10 μM phospholipase C inhibitor U73122 and 1 μM thapsigargin, and then internal solutions containing various concentrations of free IP₃ were applied. As shown in Fig. 2A, IP₃ slightly increased the Venus/ECFP emission ratio (FRET signal). In contrast, the physiological concentration of free Ca²⁺ in cytosol significantly decreased the FRET signal (Fig. 2B). The maximal normalized FRET signal change (ΔR/R₀) evoked by IP₃ and Ca²⁺ was 6.4% and −26.7%, respectively, and the apparent IP₃ and Ca²⁺ sensitivity was 39 ± 25 and 122 ± 19 nM, respectively (Fig. 2 C and D). In the presence of Ca²⁺, the application of IP₃ decreased the apparent Ca²⁺ sensitivity of the FRET signals in an IP₃ dose-dependent manner (Fig. 2B). The results of dual-ligand application are summarized in Fig. 2E. Because the permeabilization treatment prevented the formation of IP₃R1 clusters on the ER (26), but did not inhibit IP₃-induced Ca²⁺ release (Fig. S3), we consider that these FRET signals reflect conformational changes accompanying the channel gating of IP₃R subunits.

Fig. 2.

IP₃- and/or Ca²⁺-induced FRET signal changes. (A) Permeabilized HeLa cells expressing cR and vR were treated with the internal solution containing 5 mM EGTA and then with various concentrations of IP₃ after the time indicated by arrows. Normalized Venus/ECFP emission ratio changes (ΔR/R₀) are shown. [IP₃] is shown on the left in μM. (B) Permeabilized HeLa cells expressing cR and vR were treated with the internal solution containing 5 mM EGTA and then with various concentrations of Ca²⁺ and/or IP₃ during the period indicated by the vertical bars. Normalized Venus/ECFP emission ratio changes (ΔR/R₀) are shown. Free [Ca²⁺] is shown on the left in μM. (C) IP₃ dependence of FRET signal changes in the presence of 5 mM EGTA. Error bars correspond to SD. All values are relative to the FRET signal with zero IP₃ and 5 mM EGTA in C–E. (D) Ca²⁺ dependence of FRET signal changes in the absence of IP₃. Error bars correspond to SD. (E) Steady-state FRET signals are plotted against [Ca²⁺] and [IP₃]. The number of measurements is shown in Table S1.

From the slight convexity of the surface of the 3D plot (Fig. 2E), we noticed that the residual signals after the subtraction of Ca²⁺-induced FRET signal changes measured without IP₃ from those measured with IP₃ exhibit a bell-shaped Ca²⁺ dependence with a peak within a physiological range of Ca²⁺ (Fig. 3A). The peak level of the residual signal increased and the Ca²⁺ concentration at peak level moved to the right as IP₃ concentration ([IP₃]) increased (Fig. 3B), consistent with the IP₃ and Ca²⁺ dependence of single-channel open probability of cerebellar IP₃R recorded in planar lipid bilayers (7, 11). The residual signals after the subtraction of the IP₃-independent, Ca²⁺-induced FRET signals were not detected in HeLa cells expressing IP₃ binding-deficient K508A mutants (Fig. 3C). These results indicate that the residual signals were evoked by IP₃ binding to IP₃R1. The substitution of E2100 induced a 10-fold lower shift in Ca²⁺ sensitivity for both Ca²⁺-dependent activation and inactivation of IP₃R1 when reconstituted into planar lipid bilayers (15, 16). The residual signals in the E2100Q mutant also showed a bell-shaped Ca²⁺ dependence with ∼10-fold lower Ca²⁺ sensitivity compared with that of the wt channel (Fig. 3D). All these data strongly suggest that the positive FRET signal (i.e., increased FRET efficiency), which was uncovered by the subtraction of the IP₃-independent Ca²⁺-induced FRET signal, directly reflects the conducting activity of IP₃R1.

Fig. 3.

Subtracted FRET signals. (A) FRET signals measured with (green) or without (blue) IP₃. The Hill equations fitted to the data are shown in colored smooth curves. Subtracted signals (green minus blue) are shown in red. The difference between two Hill equations is shown in red smooth curves. Error bars correspond to SD. (B) FRET signal changes in the presence of 0.01 (red), 0.1 (yellow), 0.3 (green), 3 (cyan), and 10 (blue) μM IP₃. (C) Results from cells expressing cR(K508A) and vR(K508A) at 1 μM IP₃. (D) Results from cells expressing cR(E2100Q) and vR(E2100Q) at 1 μM IP₃. Results from cells expressing nonmutated cR and vR at 1 μM IP₃ are shown in broken lines.

Construction of a Phenomenological IP₃R Model.

In this study, we found that the activity of the IP₃R1 channel is proportional to the subtracted FRET signal, as follows:

where sFRET is a subtracted FRET signal, FRET_IP3 is the FRET signal in the presence of IP₃, and FRET₀ is the FRET signal in the absence of IP₃. When we use the Hill equation to fit the measured FRET signals with or without IP₃, we can express sFRET by the following equation:

where a is the maximal FRET change of −26.7% (Fig. 4A), and b is the apparent Ca²⁺ affinity of the FRET signal in the absence of IP₃, which was estimated to be 1.33 × 10⁻⁷ M (Fig. 4D; below). The Hill coefficient is independent of [IP₃] and was estimated to be 4 (Fig. 4B). B₀(IP₃) is the FRET signal in the absence of Ca²⁺ and is a function of [IP₃] (Fig. 4C). B₀(IP₃) can be expressed as

Fig. 4.

An IP₃R model based on the results of FRET imaging. (A) Estimated maximal FRET change. The broken line indicates a mean value of −26.7. (B) Estimated Hill coefficient. The broken line indicates a mean value of 4. (C) Estimated basal FRET change in the absence of Ca²⁺. The data were fitted with Eq. 3 (broken line). (D) Estimated apparent Ca²⁺ affinity. The data were fitted with Eq. 4 (broken line). (E) Plot of sFRET value calculated with Eqs. 2–4.

where c is the maximal FRET change in the absence of Ca²⁺ and was estimated to be 14.80% (Fig. 4C), d is the apparent IP₃ affinity in the absence of Ca²⁺ and was estimated to be 8.79 × 10⁻⁷ M (Fig. 4C), and the Hill coefficient was estimated to be 0.24 (Fig. 4C). K(IP₃) is the apparent Ca²⁺ affinity of the FRET signal and is a function of [IP₃] (Fig. 4D). K(IP₃) can be expressed as

where e is the minimum apparent Ca²⁺ affinity and was estimated to be 4.36 × 10⁻⁷ M (Fig. 4D), and f is the apparent IP₃ affinity and was estimated to be 4.70 × 10⁻⁷ M (Fig. 4D). Fig. 4E shows the sFRET values calculated by Eqs. 2–4. The calculated sFRET values provide a prediction of the activity of IP₃R on the ER in permeabilized HeLa cells. When we adjust the parameters, Eqs. 2–4 can reproduce the phenomenon of very high [IP₃] compensating for the Ca²⁺-dependent inactivation of IP₃R (Fig. S5), which was shown previously by using cerebellar IP₃R reconstituted into planar lipid bilayers (11).

Discussion

In this study, we applied an optical technique to monitor conformational changes in IP₃R channels and found that IP₃ and Ca²⁺ have opposite effects on FRET signal changes. IP₃ binding increases FRET efficiency, indicating that differentially tagged N termini within a single tetrameric channel are brought closer together by changing the distance and/or angle between them. In contrast, Ca²⁺ binding functions to decrease FRET efficiency, indicating that Ca²⁺ binding induces a relaxation of tetrameric channel complexes. This Ca²⁺-induced decrease in FRET efficiency is consistent with the results of single particle analysis showing that purified IP₃R tetramers change their shape from a tight “square” form to a relaxed “windmill” form after Ca²⁺ binding (25, 27). In this study, the magnitude of the FRET change evoked by Ca²⁺ was larger than that evoked by IP₃ (Fig. 2 C and D), a feature consistent with the results of single particle analysis in which the effect of IP₃ addition on IP₃R conformation was undetectably small (25, 27). In the presence of both IP₃ and Ca²⁺, the conformational changes within the channel subunits are not a simple summation of those evoked by each ligand, because the dual-ligand–induced FRET signal changes were not as predicted for a summative response, as shown in Fig. 5A. The same concentration of IP₃ induced different degrees of FRET signal change depending on the concentration of Ca²⁺ (Fig. 5B). A remarkable finding in this study was that, after subtraction of the FRET signal corresponding to the Ca²⁺induced conformational change from the FRET signal corresponding to the gross conformational change evoked by IP₃ and Ca²⁺ together, a bell-shaped dependence of FRET signal changes on cytosolic Ca²⁺ was revealed (Fig. 3A). The other important finding is that the maximal FRET signal changes evoked by Ca²⁺ were not affected by IP₃ concentration (Fig. 3A). If the maximal signal was increased by the addition of IP₃, the subtracted signal showed negative values at high [Ca²⁺] (Fig. 5C). The analyses of the mutant IP₃R channels suggest that the subtracted FRET signal approximates the conducting activity of the channel (Fig. 3 C and D). These results demonstrate that the relative position of the N termini provides information concerning the gating activity of the tetrameric IP₃R channel complex.

Fig. 5.

Mechanism of dual-ligand regulation of IP₃R channel gating. (A) Simulated linear summation. The FRET signal change in the absence of IP₃ (broken line) was calculated from parameters (basal FRET signal, maximal FRET change, apparent Ca²⁺ sensitivity, and Hill coefficient) estimated from the data shown in Fig. 2C. For this signal, constant FRET signal (10%) was added (continuous line). Subtracted signal (continuous line–broken line) is shown in Lower. (B) Actual measurements. The FRET signal change in the presence of 10 μM IP₃ (continuous line) was calculated from the parameters estimated from the data shown in Fig. 2C. Subtracted signal (continuous line–broken line) is shown in Lower. (C) Variable maximal FRET changes. The maximal FRET signal was changed to −35% in the presence of 10 μM IP₃. All other parameters are the same as shown in B. Subtracted signal (continuous line–broken line) is shown in Lower. (D) A five-state model of IP₃R. R₀₀₀ is the unliganded state (FRET = 0%). R₂₀₀ is a state with two Ca²⁺ binding sites occupied (FRET = 0%); R₂₂₀ is a state with four Ca²⁺ binding sites occupied (FRET = −26.7%); R₀₀₁ is a state with a single IP₃ binding site occupied and all Ca²⁺ binding sites unoccupied [FRET = B₀(IP₃)%]; and R₂₀₁ is a state with two Ca²⁺ binding sites and an IP₃ binding site occupied [FRET = B₀(IP₃)%]. (E) Steady-state FRET signals calculated according to the model shown in D with the following parameters: K₁ = k_-1/k₁ = 5.32 × 10⁻⁷ (M); K₂ = k_-2/k₂ = 5.32 × 10⁻⁸ (M); K = k_-/k₊ = 8.79 × 10⁻⁷ (M); k₁ = 2 × 10⁷ (M⁻¹⋅s⁻¹); and k₊ = 4 × 10⁸ (M⁻¹⋅s⁻¹). Broken line: 0 IP₃; continuous line: 10 μM IP₃. Experimental data of FRET signals observed in the presence of zero IP₃ (open circles) and 10 μM IP₃ (filled circles) are shown. Data are mean ± SD. The subtracted signal (continuous line–broken line) is shown in Lower.

The findings in this study allowed us to construct a phenomenological model (Eqs. 2–4) that quantitatively expresses the dependence of channel activity on [IP₃] and [Ca²⁺]. Formulation of the subtracted FRET signal showed that conformational changes accompanying channel gating can be divided into two components, which depend on [Ca²⁺] alone and both [Ca²⁺] and [IP₃] (Eq. 2). Although the sensitivity of the IP₃R channel to Ca²⁺-mediated activation is comparable in most measurements, the sensitivities to Ca²⁺ inhibition are variable depending on the experimental approaches and/or conditions used to monitor single-channel currents. IP₃Rs reconstituted into planar lipid bilayers are inhibited at [Ca²⁺] >1 μM (7), whereas IP₃Rs in patch-clamped outer nuclear membranes are inhibited at [Ca²⁺] >10 μM (12). Purified IP₃R reconstituted into planar lipid bilayers (28) and IP₃R exposed to low [Ca²⁺] (<5 nM) for a few minutes before patch-clamp experiments (29) were not inhibited at high [Ca²⁺]. These results indicate that the sensitivity to Ca²⁺ inhibition is not an intrinsic property of the IP₃R channel and is actively or passively regulated in individual cells. Therefore, parameters in Eq. 2 should be specific for the cell types examined. We demonstrated that our model, based on the results obtained from permeabilized HeLa cells, is applicable to single-channel data measured from IP₃R reconstituted into planar lipid bilayers by tuning the Hill coefficient in Eq. 2 and the parameters c, e, and f (Fig. S5). The parameter c is the maximal FRET change in the absence of Ca²⁺ (Eq. 3) and determines the channel activity at zero Ca²⁺. The parameter e is the minimum apparent Ca²⁺ affinity and determines the lower limit of the sensitivity of Ca²⁺ inhibition (Eq. 4). The parameter f is the apparent IP₃ affinity of the channel (Eq. 4). The Hill coefficient in Eq. 2 reflects the degree of cooperativity of Ca²⁺ binding. Identification of the factors that affect these parameters will facilitate understanding of the mechanism for generating diversity in the Ca²⁺ sensitivity of the channel.

What can we extrapolate about the gating mechanism of the IP₃R channel from the phenomenological model? The first term of Eq. 2 contains K(IP₃) that corresponds to the apparent sensitivity of the FRET signal to Ca²⁺ in the presence of IP₃. Because K(IP₃) is a function of [IP₃] (Fig. 4D), it is possible to consider that IP₃ binding reduces the intrinsic affinity of Ca²⁺ binding sites on the IP₃R molecule for Ca²⁺, as proposed previously (12). However, this relatively simple interpretation is unlikely because the Ca²⁺-dependent activation of the channel occurs within a range of [Ca²⁺] where the FRET signal is almost constant in the presence of IP₃. For example, the IP₃R channel was markedly activated with 0.13 μM Ca²⁺ in the presence of 1 μM IP₃ (7), even though 0.13 μM Ca²⁺ is well below the EC₅₀ value of FRET change in the presence of 1 μM IP₃ (∼0.38 μM; Fig. 3A). In addition, if the IP₃ binding selectively reduces the intrinsic Ca²⁺ affinity of the binding sites responsible for Ca²⁺-dependent inactivation without changing the affinity of the sites responsible for activation (12) or uncovers a hidden high-affinity Ca²⁺ binding site for channel activation (13), changes in the Hill coefficient (Fig. 4B) and the maximal FRET signal (Fig. 4A) must be altered depending on [IP₃]. Therefore, we have formulated the alternate hypothesis of a dual-ligand regulation mechanism of IP₃R channel gating.

Fig. 5D shows a state model that can reproduce the results of the FRET measurements from this study. In this model, four Ca²⁺ ions and a single IP₃ molecule can bind to a single tetrameric channel. To reproduce the steep Ca²⁺ dependence (Hill coefficient = 4) (Fig. 4B), two Ca²⁺ ions are assumed to bind instantaneously. IP₃ binding increases the FRET signal according to Eq. 3, irrespective of the binding of the first two Ca²⁺ ions (R₀₀₁ and R₂₀₁; Fig. 5D). The binding of two Ca²⁺ ions alone does not change the FRET signal (R₂₀₀, Fig. 5D), whereas the binding of the subsequent two Ca²⁺ ions decreases the FRET signal to −26.7% (R₂₂₀ in Fig. 5D). This relatively simple model can reproduce the experimental results of the FRET measurements made in this study (Fig. 5E). Key features of our model are: (i) IP₃ binding does not change the intrinsic affinity of IP₃R for Ca²⁺; (ii) channels occupied with three and four Ca²⁺ ions (R₂₁₀ and R₂₂₀, respectively) do not bind IP₃; (iii) IP₃ binding prevents the transition of the channel to the R₂₁₀ and R₂₂₀ states; (iv) Ca²⁺ binding is a sequential process, and the binding of the first and second Ca²⁺ ions is necessary for the binding of third and fourth Ca²⁺ ions; and (v) the Ca²⁺ binding affinity for the third and fourth Ca²⁺ ions is higher than that of the first and second Ca²⁺. The third feature is the cause of the apparent IP₃-induced reduction in the Ca²⁺ sensitivity of the FRET signal change observed in the absence of any change in the intrinsic Ca²⁺ binding affinity. The fifth feature is essential to reproduce the steep dependence of FRET changes on cytosolic Ca²⁺, and it has not been reported previously. Our model is similar to the previous four-state model, which can reproduce frequency encoding by Ca²⁺ oscillations and Ca²⁺ wave propagation (30–32). Our model, however, can reproduce IP₃-dependent reduction of apparent Ca²⁺ sensitivity of the channel, whereas the previous four-state models were constructed based on the steady-state bell-shaped Ca²⁺ dependence of IP₃R at a fixed concentration of IP₃ (2 μM). Remarkably, the number of states in our model is drastically smaller than that of previous models, which can reproduce IP₃-dependent reduction of Ca²⁺ sensitivity, composed of 4,096 (11) or 3,750 (14) states.

We show here that the subtracted FRET signal reflects the activity of the IP₃R channel. Which states are active in the model proposed? Foskett and colleagues (14) showed that IP₃R exhibits IP₃-independent spontaneous opening in the absence of Ca²⁺. These observations suggest that R₀₀₀ and R₀₀₁ possess an indistinguishable low open probability. We found that the subtracted FRET signal is a good approximation of the sum of the fraction of the state of R₂₀₁ and 0.075-fold of the fraction of the states of R₀₀₀ and R₀₀₁ (Fig. S6). These results indicate that (i) R₂₀₁ is the main conducting open state; (ii) R₀₀₀ and R₀₀₁ are open states with a low open probability; and (iii) R₂₂₀ is an inactivated state.

The state in which all of the ligand binding sites are occupied, R₂₂₁, is not present in the model shown in Fig. 5D. Therefore, high concentrations of Ca²⁺ prevent IP₃ binding to the receptor in this model. This specific property is consistent with the previous experimental observations in which IP₃ binding to recombinant IP₃R1 expressed in Sf9 cells was examined under various concentrations of Ca²⁺ (23, 33). Our model shows that IP₃ binding prevents binding of Ca²⁺ to the inactivation sites and that, reciprocally, Ca²⁺ binding to the inactivation sites prevents IP₃ binding. Thus, there is no direct transition between the active R₂₀₁ state and the inactive R₂₂₀ state. We propose that this dual-ligand competition is the main mechanism underlying the IP₃-dependent regulation of the bell-shaped relationship between IP₃R gating and cytosolic Ca²⁺. This dual-ligand competition model reproduces the experimental results obtained by the FRET measurement without assuming a complex allosteric regulation of the affinity and function of Ca²⁺ binding sites as proposed previously (14).

Single particle analysis has suggested that Ca²⁺ binding induces structural transition from the tight square form to the relaxed windmill form (25, 27), but the relationship between the receptor structure and its function has not been examined. In this study, we found that the relaxed state (R₂₂₀, which may correspond to the windmill form) is an inactivated state, whereas the square form contains the active conducting state, R₂₀₁ (Fig. 5D). There are multiple Ca²⁺ binding sites within a single IP₃R subunit (4, 5). The Ca²⁺ sensitivity of the FRET signal of the E2100Q mutant was reduced to 4.07 μM (from 122 nM in the wt channel) (Fig. 3D). The Hill coefficient was also reduced to 1.2, from 4 in the wt channel. However, the maximal FRET signal change of E2100Q was not altered (Fig. 3D). These results suggest that E2100 is involved in the Ca²⁺ binding site responsible for the first two Ca²⁺ ions that are involved in channel activation (Fig. 5D). Our analysis unveiled that there are two different types of Ca²⁺ binding site within IP₃R: low-affinity sites responsible for channel activation and high-affinity sites responsible for channel inactivation. The four Ca²⁺ binding sites proposed in this study are the minimal requirement. The measurement of the FRET signal used in this study will be useful for the identification of Ca²⁺ binding sites involved in both the channel activation and inactivation and will further our understanding of the molecular basis of IP₃R gating.

Materials and Methods

HeLa cells expressing cR and vR were treated with 10 μM U73122 for 5 min and then with 1 μM thapsigargin for 5 min in balanced salt solution (BSS; 115 mM NaCl, 5.4 mM KCl, 1 mM MgCl₂, 10 mM glucose, and 20 mM Hepes–KOH, pH 7.4 at 37 °C) at 37 °C. After washing three times with BSS containing 5 mM EGTA, cells were permeabilized with 60 μM β-escin in the internal solution (19 mM NaCl, 125 mM KCl, 10 mM Hepes–KOH, pH 7.4 at 37 °C) containing 5 mM EGTA for 3–5 min. Permeabilized cells were gently washed with the internal solution containing 5 mM EGTA. Free Ca²⁺ concentrations in the internal solutions were adjusted with K₂HEDTA and CaHEDTA at 37 °C according to the described method (28). Fluorescent signals were acquired with an IX-71 or IX-81 inverted microscope (Olympus), a cooled CCD camera ORCA-ER (Hamamatsu Photonics), and a 40× (n. a., 1.35) objective lens (Olympus), as described (34). A 425–445 nm excitation filter and a pair of 460- to 510-nm (ECFP) and 525- to 5,650-nm (Venus) emission filters were used. The images were captured at every 2–30 s with an exposure time of 100–150 ms. The emission ratio was calculated after subtraction of the background fluorescence. The Venus/ECFP emission ratio was defined as R, and ΔR was defined as R − R₀, where R₀ is the basal level. Cells showing the initial ratio of 1.50 ± 0.37 (from 0.90 to 3.54; n = 3,295) were used for FRET measurements. The data acquisition was performed with TI Workbench and MetaMorph/MetaFluor software (Molecular Devices). Off-line analysis was performed with TI Workbench and Igor Pro software.

Other methods are provided in SI Materials and Methods.