This Article Abstract Full Text (PDF) All Versions of this Article: biophysj.106.099028v1 92/10/3541    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Laver, D. R. Search for Related Content PubMed PubMed Citation Articles by Laver, D. R.

Ca²⁺ Stores Regulate Ryanodine Receptor Ca²⁺ Release Channels via Luminal and Cytosolic Ca²⁺ Sites

School of Biomedical Sciences, University of Newcastle and Hunter Medical Research Institute, Callaghan, New South Wales, Australia

Correspondence: Address reprint requests to Dr. Derek Laver, Tel.: 61-2-4921-8732; E-mail: derek.laver{at}newcastle.edu.au.

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
The free [Ca²⁺] in endoplasmic/sarcoplasmic reticulum Ca²⁺ stores regulates excitability of Ca²⁺ release by stimulating the Ca²⁺ release channels. Just how the stored Ca²⁺ regulates activation of these channels is still disputed. One proposal attributes luminal Ca²⁺-activation to luminal facing regulatory sites, whereas another envisages Ca²⁺ permeation to cytoplasmic sites. This study develops a unified model for luminal Ca²⁺ activation for single cardiac ryanodine receptors (RyR₂) and RyRs in coupled clusters in artificial lipid bilayers. It is shown that luminal regulation of RyR₂ involves three modes of action associated with Ca²⁺ sensors in different parts of the molecule; a luminal activation site (L-site, 60 µM affinity), a cytoplasmic activation site (A-site, 0.9 µM affinity), and a novel cytoplasmic inactivation site (I₂-site, 1.2 µM affinity). RyR activation by luminal Ca²⁺ is demonstrated to occur by a multistep process dubbed luminal-triggered Ca²⁺ feedthrough. Ca²⁺ binding to the L-site initiates brief openings (1 ms duration at 1–10 s^–1) allowing luminal Ca²⁺ to access the A-site, producing up to 30-fold prolongation of openings. The model explains a broad data set, reconciles previous conflicting observations and provides a foundation for understanding the action of pharmacological agents, RyR-associated proteins, and RyR₂ mutations on a range of Ca²⁺-mediated physiological and pathological processes.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
In cardiac muscle, contraction is mediated by depolarization and subsequent opening of voltage-dependent, L-type Ca²⁺ channels in the cell membrane (sarcolemma). Opening of these channels allows Ca²⁺ influx into the cell, which in turn opens Ca²⁺ release channels in the sarcoplasmic reticulum (SR) membrane. SR Ca²⁺ release can be a small highly localized increase in Ca²⁺ concentration called a spark, or can propagate along the cell in a Ca²⁺ wave, or can occur simultaneously along the length of the cell resulting in a global increase of Ca²⁺ called a transient (1). Ca²⁺ sparks are believed to be the elementary SR Ca²⁺ release event and Ca²⁺ waves and transients are thought to be the summation of spark events (2).

It has long been known that the excitability of Ca²⁺ release from muscle SR is substantially increased by its luminal Ca²⁺ load ([Ca²⁺]_L) (3,4). More recently, it has been shown that the excitability of neuronal endoplasmic reticulum (ER) is regulated in the same way, suggesting that modulation of Ca²⁺ signaling by [Ca²⁺]_L is a general phenomenon (5). However, the basis for load-dependent excitability remains controversial. Experiments on muscle cells and isolated SR vesicles have found that the effect of [Ca²⁺]_L on Ca²⁺ release could not be explained entirely by the associated Ca²⁺ gradient across the SR and that [Ca²⁺]_L must somehow control the Ca²⁺ permeability of the membrane (6,7). This was confirmed when activity of isolated RyRs in artificial bilayers was found to be modulated by [Ca²⁺]_L (8). At issue is the mechanism underlying the activation of RyRs by luminal Ca²⁺ being attributed to two quite different processes. The true-luminal hypothesis attributes luminal Ca²⁺-activation and inhibition to Ca²⁺ regulatory sites on the luminal side of the RyR (9). The feedthrough hypothesis proposes that luminal Ca²⁺ permeates the pore and binds to cytoplasmic activation and inhibition sites (10–12). The latter is supported by the close correlation between the effect of [Ca²⁺]_L and Ca²⁺ flux (lumen to cytoplasm), and by the observation that luminal regulation is dependent on Ca²⁺ buffering on the cytoplasmic side of the membrane (11). However, there is an increasing body of evidence suggesting that there are sensing sites for Ca²⁺ regulation on the luminal side of the protein. For example, it has been shown that luminal Ca²⁺-activation is abolished by tryptic digestion of the luminal side of cardiac ryanodine receptors (RyR₂) (13). Although a recent bilayer study has presented evidence that both mechanisms somehow control skeletal ryanodine receptors (RyR₁) (14), it is not understood how cytoplasmic and luminal Ca²⁺ sites explain current observations or reconcile conflicting interpretations in the literature. In this study, measurements of the gating kinetics of isolated RyR₂ in lipid bilayers underlie the proposal of a novel mechanism for activation of RyRs by luminal-triggered Ca²⁺ feedthrough. This model predicts that in addition to cytoplasmic sites for Ca²⁺-activation and low affinity Ca²⁺/Mg²⁺-inhibition, RyRs possess a luminal Ca²⁺ binding site and a high affinity cytoplasmic Ca²⁺ inactivation site (Fig. 1). Ca²⁺ binding at the luminal site is sufficient to activate channel openings whereupon Ca²⁺ flows from lumen to cytoplasm (feedthrough) either enhances activation or causes inactivation.

View larger version (19K): [in this window] [in a new window]   FIGURE 1  The luminal-triggered Ca2+ feedthrough model for [Ca2+]L regulation of RyRs. (A) Schematic of luminal-triggered Ca2+ feedthrough. Ca2+ binding to the L-site causes channel opening whereupon luminal Ca2+ has access to the cytoplasmic Ca2+ activation (A-site) and inactivation sites (I2-site). The I1-site that mediates the low affinity Ca2+/Mg2+ is not included in the model (see text for details). (B) Kinetic schemes for Ca2+ binding at the L-, A-, and I2-sites (Schemes I–III) and the overall scheme (Scheme IV) resulting from the combined action of Ca2+ at all three sites. Schemes I–III are representations of multistep processes. Hence some reaction rates have complex dependencies on [Ca2+] and these are given in the equations listed in Table 1. Asterisks indicate sites occupied with Ca2+. Reaction rates that depend on the Ca2+ current have subscripts o and c to indicate rates associated with open and closed channels, respectively. The open and closed status of the channel associated with each kinetic state is indicated by the subscripts O and C, respectively.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
SR vesicles (containing RyR₂) were obtained from sheep hearts and were reconstituted into artificial lipid bilayers as previously described (18). Lipid bilayers were formed from phosphatidylethanolamine and phosphatidylcholine (8:2 wt/wt) in n-decane (50 mg/ml). During experiments, the cis (cytoplasmic) and trans (luminal) solutions contained 250 mM Cs⁺ (230 mM CsCH₃O₃S, 20 mM CsCl) and various concentrations of CaCl₂. The trans solution was altered by addition of aliquots of stock solutions and the cis solution by local perfusion which allowed solution exchange within 1 s (19).

Solutions were pH-buffered with 10 mM n-tris[Hydroxymethyl]methyl-2-aminoethanesulfonic acid and solutions were titrated to pH 7.4 using CsOH. Free [Ca²⁺] up to 100 nM was estimated using published association constants (20) and the program Bound and Determined (21) and concentrations higher than this were measured using a Ca²⁺ electrode (Radiometer, Copenhagen, Denmark). [Ca²⁺] below 10 µM was buffered with 4.5 mM 1,2-bis(o-aminophenoxy)ethane-n,n,n',n'-tetraacetic acid (4K⁺) (BAPTA) and titrated with CaCl₂. [Ca²⁺] in the range 10–50 µM was buffered with either sodium citrate (up to 6 mM) or dibromo BAPTA (up to 2 mM).

Acquisition and analysis of ion channel recordings
Bilayer apparatus and data recording methods have been described previously (14). Electrical potentials are expressed using standard physiological convention (i.e., cytoplasmic side relative to the luminal side at virtual ground). Measurements were carried out at 23 ± 2°C. Before analysis, the current signal was digitally filtered at 1 kHz with a Gaussian filter and sampled at 5 kHz. Single channel properties were measured using Channel2 software (P. W. Gage and M. Smith, Australian National University, Canberra). Open probability (P_o) and open and closed durations were calculated from single channel records using a threshold discriminator at 50% of channel amplitude. For experiments in which bilayers contained several RyRs, P_o was calculated from the time-averaged current divided by the unitary current and the number of channels (n). Both methods of calculating P_o gave similar results. The mean channel open and closed durations (_o and _c) could also be calculated from mean open and closed durations in multichannel recordings (_o(n) and _c(n)), provided that multiple openings were rare: _o = _o(n) and _c = _c(n) x n. The number of channels in each experiment was determined during periods of strong activation, which were usually achieved by turning off the local perfusion and exposing the RyRs to the cis bath. The values _o and _c were calculated mainly from the recording of 100–1000 opening events. However, under conditions that produced extremely low channel activity, the mean durations were obtained from as few as 40 events covering >400 s of recording (sampling error <40^–, i.e., <16%). Dwell-time frequency histograms of channel open and closed events were obtained exclusively from single channel recordings and were displayed as probabilities (counts/total number of events). These were plotted using the log-bin method suggested by Sigworth and Sine (22), which displays exponentials as peaked distributions centered at their exponential time constant. Sampling bins were equally spaced on a log scale with 3.5 or 7 bins per decade.

On rare occasions (<5% of experiments), RyRs appeared to form coupled clusters in the bilayer, similar to that seen for RyR₁ (14). The experiments on coupled and uncoupled RyRs were analyzed and presented separately. Channel gating in coupled clusters was analyzed using the Hidden Markov Model (23), as described previously (14). The algorithm calculated the transition probability matrix of the underlying Markov process from the raw signal using maximum likelihood criteria. The mean channel opening and closing rates were calculated from the transition probability matrix.

Ca²⁺-dependencies of P_o, _o, and opening rate were characterized by fitting these data with Hill curves using the following equations for activation and inactivation (shown here for the case of P_o, similar equations apply to _o and opening rate):

The expressions P_min and P_max are the activities of the minimally and maximally activated channel, K_a and K_i are the [Ca²⁺] for half-activation and inhibition, and n_a and n_i are the corresponding Hill coefficients. The theory was fitted with the data using the method of least-squares.

The luminal-triggered Ca²⁺ feedthrough model: Kinetic schemes and equations
RyRs are known to possess cytoplasmic sites for Ca²⁺-activation (A-sites 1 µM affinity) and low-affinity Ca²⁺/Mg²⁺-inhibition (I₁-site, 10 mM affinity, previously named the I-site). Based on novel data presented below, the model includes a cytoplasmic Ca²⁺-inactivation site (the I₂-site, which causes partial inactivation with 1 µM affinity) and a luminal activation site (L-site, 60 µM affinity). Fig. 1 A shows a schematic representation of the locations of these sites on the RyR protein. It is envisaged that the channel can open if Ca²⁺ is bound to either the A- or L-sites. Thus luminal Ca²⁺ can open the channel by binding to the L-site, whereupon the flow of Ca²⁺ through the pore can increase the cytoplasmic [Ca²⁺] in the vicinity of the pore mouth ([Ca²⁺]_P) and reinforce channel activation (increasing _o) by binding to the A-site or inactivate the channel by binding to the I₂-site (decreasing _o).

Kinetic schemes for Ca²⁺ binding at the L-, A-, and I₂-sites are shown in Fig. 1 B (I–III) and the overall regulation of RyRs by the three Ca²⁺ sites is described by an eight-state model (Scheme IV) presenting the amalgamation of Schemes I–III. The equations describing the action of luminal and cytoplasmic Ca²⁺ on channel activity and the model parameters are listed in Table 1. The I₁-site was not considered in the model calculations because Ca²⁺ feedthrough is unlikely to increase [Ca²⁺]_P to levels where significant Ca²⁺ binding will occur at this site.

The first step in developing the luminal-triggered Ca²⁺ feedthrough model was to formulate a phenomenological description of RyR₂ regulation by cytoplasmic Ca²⁺. The kinetic schemes used previously to describe cytoplasmic Ca²⁺-activation of RyRs are complex and involve many closed and open states (24,25). To avoid this level of complexity the whole cytoplasmic Ca²⁺-activation process was lumped into two single-step Ca²⁺ binding schemes associated with the A- and I₂-sites (Schemes II and III in Fig. 1 B). Thus, the parameters , , , , , and K_I in Eqs. 4–10 were adjusted until the model fitted the cytoplasmic Ca²⁺ concentration ([Ca²⁺]_C)-dependencies of _o and _c (Figs. 5 and 10). The second step was to ascertain the gating properties associated with the L-site (Scheme I). This was done by fitting the voltage/luminal Ca²⁺ concentration ([Ca²⁺]_L)-dependence of _c (Fig. 6) using the parameters , ß, K_L in Eqs. 2 and 3. The final step was to calculate the effects of Ca²⁺ feedthrough on the voltage/[Ca²⁺]_L-dependencies of _o, _c, and their associated dwell-time distributions from theoretical estimates of [Ca²⁺]_P and compare these estimates with the data.

View larger version (14K): [in this window] [in a new window]   FIGURE 5  The [Ca2+]L-dependence of RyR2 Po. (A–D) RyRs were activated at the voltage indicated, by 2 mM ATP and 100 nM [Ca2+]C. Data points show the mean ± SE of 3–18 measurements. Solid curves show the fit to the data of the luminal-triggered Ca2+ feedthrough model using the parameters listed in Table 1.

View larger version (20K): [in this window] [in a new window]   FIGURE 10  Recordings from an experiment with six RyRs in the bilayer showing coupled gating. At +20 mV (top trace), the channels appeared to gate independently. The dashed lines indicated the current levels associated with various numbers of open channels. The closed current level is labeled C. At –40 mV (bottom trace), the opening of a single channel (transitions from C to O1) was followed closely by openings to higher levels. Only four levels are apparent in this section (labeled O1–O4). The baths contained symmetric 250 mM Cs+ solutions with 2 mM ATP, [Ca2+]C = 100 nM, and [Ca2+]L = 1 mM.

View larger version (19K): [in this window] [in a new window]   FIGURE 6  The effects of membrane potential on the [Ca2+]L-dependent gating of RyR2. (A) Mean channel open times at three membrane voltages and (B) channel opening rates at two voltages. Rates were calculated from the inverse of the mean of closed times. The data points show mean ± SE of 3–14 measurements. In panels A and B, the dashed curves show Hill fits to the data (parameter values plotted in C and D) and the solid curves show the fits of the luminal-triggered Ca2+ feedthrough model using the parameters listed in Table 1. (C) Half-activating and inactivating [Ca2+]L, Ka (•), and Ki () for o. (D) The Ka for RyR opening rate (•) and maximum opening rate (Vmax ). The Hill coefficient for opening rate was 1.6 ± 0.3. In panels C and D, the solid and dashed curves show the fit to the data of the luminal-triggered Ca2+ feedthrough model. The line (long dashes) shows the voltage-dependence in Ka expected from translocation of Ca2+ through the membrane voltage (i.e., where ).

(1)

For a given distance from the pore, [Ca²⁺]_P has a linear dependence on I_Ca, which is encapsulated in the parameters X in Eq. 7b and Y in Eq. 10b (see Table 1). I_Ca was calculated using a rate theory model of RyR conductance (27). This part of the model has only two adjustable parameters, X and Y.

Model predictions were made by simulating the time series of channel gating events and using these to generate theoretical open and closed dwell time distributions, _o, _c, and P_o. For each set of experimental conditions, the model reaction rates were calculated using the equations in Table 1. The sojourn of the simulated channel through its various states was driven by a random number generator in conjunction with the transition probability matrix that was derived from the model reaction rates. Sequential open and closed events were marked by transitions between open and closed states in the model. The effects of limited time resolution in the data were accommodated by amalgamating simulated events shorter than <200 µs.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
The effects of cytoplasmic and luminal Ca²⁺ on RyR₂ in the presence of ATP are presented here first because 1), it is recognized that the effects of luminal Ca²⁺ require the presence of cytoplasmic agonists such as ATP, caffeine, or sulmazole (8,28,29); and 2), ATP is the physiological coagonist. Data are presented that define the affinity and gating properties of one luminal and two cytoplasmic Ca²⁺ sites on RyR₂. These properties were used to construct the luminal-triggered Ca²⁺ feedthrough model. The results were then fitted by adjusting the model parameters associated with the proximity of the two cytoplasmic Ca²⁺ binding sites to the pore mouth. The model is then applied to elucidating the effects of ATP and ion channel coupling on RyR regulation by Ca²⁺.

Activation of RyR₂ by cytoplasmic Ca²⁺ in the presence of ATP
It is shown that the activity of RyR₂ was dependent on both [Ca²⁺]_C (Fig. 2 A) and [Ca²⁺]_L (Fig. 2 B). The experiments in this section focus on the properties of the A- and I₂-sites as revealed by the effects of [Ca²⁺]_C on RyR₂ activity. Fig. 2 A highlights the gradual transition in the gating properties that occurs as the dominant cytoplasmic agonist switches from ATP to Ca²⁺. RyRs activated by ATP (100 nM [Ca²⁺]_C, Fig. 2 A, top three traces) had relatively long open and closed events compared to RyRs activated by high [Ca²⁺]_C (with 2 mM ATP, Fig. 2 A, bottom two traces). To probe the gating mechanisms for this behavior, RyR gating was analyzed by compiling frequency histograms of open and closed times. These could be described by the sum of two exponential decays (these appear as peaked distributions when using the log-bin method in Fig. 3). The time-constants and relative weighting of the exponential components depended on [Ca²⁺]_C and [Ca²⁺]_L. Time-constants within each open time distribution had a 3–10-fold separation and lay within the range 1–50 ms (e.g., dashed lines in Fig. 3 A). The closed time distributions varied considerably with [Ca²⁺]_C. In the presence of 0.1 µM [Ca²⁺]_C, two widely separated exponentials were clearly resolved (open arrows in Fig. 3 B) in which >80% of the events were associated with the longer time-constant. Raising [Ca²⁺]_C progressively reduced the longer time constant until both were quite similar at 10 µM (solid arrows in Fig. 3 B). The dwell-time distributions are frequently represented here by their means. The value _o gave a good representation of the average time-constant of the open time distribution and the value for _c lay within 20% of the longer closed time constant.

View larger version (17K): [in this window] [in a new window]   FIGURE 2  The effects of [Ca2+]L and [Ca2+]C on the activity of RyR2 (–40 mV). (A) The effect of [Ca2+]C on the activity of a RyR in the presence of 100 µM [Ca2+]L. Increasing [Ca2+]C increases Po in association with a decrease in duration of mean open and closed events. Channel openings are downward current jumps from the baseline (indicated with a dash). (B) The effect of [Ca2+]L on RyR activity in the presence of 0.1 µM [Ca2+]C and 2 mM ATP.

View larger version (12K): [in this window] [in a new window]   FIGURE 3  Dwell-time probability distributions of channel open (A) and closed (B) events. The data are plotted using the log-bin method of Sigworth and Sine (22). Event duration distributions were compiled from a single RyR, which was activated by 2 mM ATP and the indicated [Ca2+]C in the presence of 100 µM [Ca2+]L (–40 mV). An increase in [Ca2+]C from 0.1 to 10 µM shifted both the open and closed distributions to shorter times. Solid curves show double-exponential fits to the data and dashed curves show individual exponential components for the fits to the open circles. When using the log-bin method the exponential components within these distributions appear as peaks centered at their respective time-constants. The arrows in panel B show the time constants associated with the data at 0.1 µM [Ca2+]C (, open arrows) and 10 µM [Ca2+]C (•, solid arrows).

View larger version (7K): [in this window] [in a new window]   FIGURE 4  The effect of [Ca2+]C on the activity of RyR2 at –40 mV. The values Po, o, and c were measured in the presence of 2 mM ATP at [Ca2+]L <10 µM (•), 100 µM (), or 1000 µM (). (A) Mean open probability Po mean ± SE. The numbers of experiments and the Hill parameters are listed in Table 2 (Hill fits to the data are not shown). (B, C) The mean ± SE of three-to-eight measurements of o and c. The curves [Ca2+]L <10 µM (solid), 100 µM (short dashes), and 100 µM (short/long dashes) show the fit to the data of the luminal-triggered Ca2+ feedthrough model using the parameters in Table 1.

View larger version (6K): [in this window] [in a new window]   FIGURE 8  The effect of ATP and [Ca2+]C on the activity of RyR2 at +40 mV. The values Po, o, and c were measured in the presence of 2 mM ATP (•) and in its absence (x). (A) Mean open probability Po mean ± SE. The numbers of experiments and the Hill parameters are listed in Table 2 (Hill fits to the data are not shown). (B, C) The mean ± SE of 3–9 measurements of o and c. The curves (2 mM ATP, solid; and absence of ATP, dashed) show the fits of the luminal-triggered Ca2+ feedthrough model using the parameters in Table 1.

Bell-shaped dependence of RyR₂ activity on [Ca²⁺]_L in the presence of ATP
Although several studies have reported a bell-shaped [Ca²⁺]_L-dependence of RyR₁ P_o in the presence of ATP, there has been no equivalent demonstration of this in RyR₂. In the absence of luminal Ca²⁺ the activity of RyR₂ was virtually zero (P_o = (8 ± 4) x 10^–4 at [Ca²⁺]_L 0.1 µM, n = 6) indicating a near-absolute requirement for luminal Ca²⁺. Increasing [Ca²⁺]_L (from 1 to 100 µM at –40 mV) substantially increased the activity of the RyRs, whereas a further increase to 1 mM decreased channel activity (Fig. 2 B). At –60 mV and –40 mV, voltages favoring the flow of Ca²⁺ from luminal to cytoplasmic baths, the [Ca²⁺]_L for half-activation of P_o, was 50 µM (K_a for the left side of the bell) and [Ca²⁺]_L for half-inactivation was 1 mM (K_i for the right side of the bell, Fig. 5, A and B). Both K_a and K_i increased as the bilayer voltage became more positive (Fig. 5, C and D), and at positive voltages the decline in P_o was beyond the experimental range of [Ca²⁺]_L (K_i could no longer be monitored).

Luminal Ca²⁺ would not be expected to influence the channel-opening rate (1/_c) other than by affecting a luminal site because cytoplasmic sites are inaccessible to luminal ions when the channel is closed. Therefore, analysis of _o and _c should provide valuable information of the effect of Ca²⁺ feedthrough in channel activation. The results show that [Ca²⁺]_L had very different effects on _o and channel opening rate (Fig. 6). The value _o had a bell-shaped dependence on [Ca²⁺]_L, which is shown at several voltages in Fig. 6 A (the bell is most clearly seen at –20 mV). The K_a and K_i values for the [Ca²⁺]_L-dependence of _o over the entire experimental voltage range is shown in Fig. 6 C. The values K_a and K_i increased with voltage in a way expected for Ca²⁺ ions, which must cross the membrane to reach their effector site (long-dashed line, Fig. 6 C). These results indicate that luminal Ca²⁺ activation and inactivation of _o are mediated by cytoplasmic facing sites. Moreover, the nonadditive effects of luminal and cytoplasmic Ca²⁺ on channel activation and inhibition indicate that the luminal and cytoplasmic actions are meditated by common binding sites.

Luminal Ca²⁺ activated RyR₂ by increasing their opening rate and approached a maximum (V_max) at high concentrations (Fig. 6 B). The voltage-dependence of K_a and V_max are shown in Fig. 6 D. The K_a indicates the involvement of a Ca²⁺ binding site with an affinity (K_L) 60 µM and the V_max indicates that the Ca²⁺-bound site can trigger channel openings at a rate of 1–10 s^–1. The K_a for opening rate did not show a significant voltage-dependence (P > 0.15, t-test) indicating that luminal Ca²⁺ ions do not move through the trans-membrane electric field to reach the site. V_max varied threefold over the experimental voltage range and this could reflect an intrinsic voltage-dependence of the RyR opening mechanism. Taken together, these results indicate that [Ca²⁺]_L-activation of opening rate is mediated by luminal facing sites of action.

The properties of the L-site are derived from the data in Fig. 6, B and D, as follows: RyR₂ opening rate associated with the L-site (' in Scheme I) has an asymptotic dependence on [Ca²⁺]_L (Fig. 6 B) that is characteristic of a two-step mechanism [(L_C + 2Ca²⁺) L*_C L*_O] in which Ca²⁺ binding (which is relatively fast) leads to channel opening. In the absence of Ca²⁺ feedthrough, the mean duration of these openings (1 ms) is determined by the closing rate, ß'. The Hill coefficient for the [Ca²⁺]_L-dependence of opening rate was 1.6, suggesting a second-order dependence of ' on [Ca²⁺]_L in Eq. 2. The voltage-dependence of V_max is accommodated by imposing a voltage-dependence on '.

Characteristics of the luminal-triggered Ca²⁺ feedthrough model
The model accounts for the [Ca²⁺]_L-, [Ca²⁺]_C-, and voltage-dependencies of _o and _c over the range –60 mV to +20 mV (Figs. 4–6, curves, the model discrepancy at +40 mV is discussed in the section on model limitations). The model predictions of the [Ca²⁺]_C-dependencies of _o and _c, shown in Fig. 4, B and C, can be understood in terms of the various reaction steps in Scheme IV. In the absence of luminal Ca²⁺, _o is primarily determined by the rates and of the steps LAI_2C LA*I_2O LA*I*_2C. These rates have opposite [Ca²⁺]_C-dependencies and their effects on the [Ca²⁺]_C-dependence of _o tend to cancel, leaving a small increase in _o with increasing [Ca²⁺]_C (Fig. 4 B, solid curve). When [Ca²⁺]_L = 100 µM the L-site is mostly bound to Ca²⁺ so that _o is primarily determined by the inactivation transitions in L*AI*_2C L*AI_2O L*A*I_2O L*A*I*_2C. Under these conditions the [Ca²⁺]_C-dependent decrease in _o (Fig. 4 B, short dashes) reflects the Ca²⁺-inactivation rate, . Fig. 4 C shows the model predictions of the [Ca²⁺]_C-dependence of _c. At low [Ca²⁺]_{C c} is primarily determined by ' in the reaction step LAI_2CL*AI_2O so that _c depends on [Ca²⁺]_L (compare curves at 10 µM and 100 µM [Ca²⁺]_L in Fig. 4 C). Increasing [Ca²⁺]_C decreases _c as '_c increasingly determines the channel opening rate via the reaction step LAI_2C LA*I_2O. When [Ca²⁺]_C 3 µM, '_c becomes very large, then ' in the reactivation step LA*I*_2C LA*I_2O determines the lower limit for _c at high [Ca²⁺]_C. When [Ca²⁺]_L = 1000 µM the Ca²⁺ feedthrough is large enough to cause saturation of the I₂-site so that [Ca²⁺]_C is unable to modulate inactivation (Fig. 4 B, long/short dashes).

The effects of [Ca²⁺]_L and voltage on _o and _c are compared with the model in Fig. 6, A and B, respectively. In the absence of Ca²⁺ feedthrough (low [Ca²⁺]_L or large positive voltage), _o is determined by ß' in the reaction LAI_2C L*AI_2O. The contribution of Ca²⁺ feedthrough is modeled by an extension of the former scheme: LAI_2C L*AI_2O L*A*I_2O L*A*I*_2C. Increasing Ca²⁺ feedthrough biases the state occupancies to the right so that the closing step on the left becomes less frequent. Hence, _o increases with [Ca²⁺]_L until the Ca²⁺-dependent closing step on the right becomes fast enough to decrease _o. According to the conductance model, the curvature in the voltage-dependencies in the log-plots in Fig. 6 C comes from saturation of I_Ca as the driving force on Ca²⁺ becomes large. Fig. 6 B shows the corresponding model predictions of channel opening rate at 100 nM [Ca²⁺]_C. At these low [Ca²⁺]_C, the channel opening rate reflects ' in the luminal activation step LAI_2CL*AI_2O. The luminal site affinity (K_L) determines the [Ca²⁺]_L range for the rising phase in the opening rate.

In addition to _o and _c, it was found that the model could also account for the effect of [Ca²⁺]_L on the shape of the dwell-time distributions in the presence of low [Ca²⁺]_C. Fig. 7 shows the distribution of channel open and closed times at three [Ca²⁺]_L representing the subactivating (10 µM), maximal-activating (100 µM), and inactivating regions (1 mM) of the bell-shaped dependence of RyR₂ activity on [Ca²⁺]_L (–40 mV, [Ca²⁺]_C = 100 nM). The shape of the closed time distributions varied considerably with [Ca²⁺]_L (Fig. 7, A, C, and E). As [Ca²⁺]_L increased, the time-constant of the main exponential component decreased from 10 s to 200 ms and the distribution changed from unimodal to bimodal with the appearance of another time constant of 1 ms. Open time distributions (Fig. 7 B, D, and F) were double exponentials as described in Fig. 3. Although the peak of the log-binned open distributions depended on [Ca²⁺]_L, the shape of the distributions was similar over the experimental range. The model faithfully reproduced the effect of [Ca²⁺]_L on these dwell time distributions (Fig. 7, curves). According to the model, closure of the channel causes termination of Ca²⁺ feedthrough which, at low [Ca²⁺]_C, leads to deactivation of the RyR via Ca²⁺ dissociation from the A-site, underlying the long closed events. By the same process, channel closures allow Ca²⁺ dissociation from the I₂-site, which permits reopening of the channel. The 1-ms time constant in the closed dwell-time distributions is associated with instances of recovery from I₂-site inactivation that occurred before the channel had time to deactivate.

View larger version (16K): [in this window] [in a new window]   FIGURE 7  The effect of [Ca2+]L on closed (A, C, and E) and open (B, D, and F) dwell-time distributions. The data are plotted using the log-bin method of Sigworth and Sine (22). Event duration distributions were compiled from a single RyR, which was activated by cytoplasmic 2 mM ATP (100 nM [Ca2+]C and voltage = –40 mV) and the indicated [Ca2+]L. The data are plotted using log-bins with 3.5 per decade. Solid curves show simulated distributions generated from the luminal-triggered Ca2+ feedthrough model using the parameters listed in Table 1.

Fig. 9 compares the [Ca²⁺]_L-dependent properties of RyR₂ in the presence and absence of ATP. Also, as previously reported, ATP markedly amplified the effects of [Ca²⁺]_L on P_o (at low [Ca²⁺]_C, Fig. 9 A). In the presence of ATP (–40 mV), _o showed a bell-shaped dependence on [Ca²⁺]_L whereas in the absence of ATP, _o remained relatively constant (Fig. 9 B). It is interesting to note that ATP had similar effects on the [Ca²⁺]_L- and [Ca²⁺]_C-dependencies of _o in that, in both cases, an increase in Ca²⁺ could only increase _o when ATP was present. This is consistent with common sites of action for cytoplasmic and luminal Ca²⁺. K_a for opening rate was not significantly affected by ATP (Fig. 9 B), indicating that ATP did not modify the affinity of the L-site. Rather, ATP appeared to stabilize the open state of the channel and destabilize the closed state.

View larger version (6K): [in this window] [in a new window]   FIGURE 9  The effects of ATP on the [Ca2+]L-dependent gating of RyR2. (A) The open probability of RyRs (100 nM [Ca2+]C and voltage = –40 mV) in the presence of 2 mM ATP (•) and in its absence (). Also shown are the corresponding o (B) and opening rates (C). Solid and dashed curves show the fit to the data of the luminal-triggered Ca2+ feedthrough model using the parameters listed in Table 1. The data points show mean ± SE of 3–18 measurements. Hill fits (not shown) to the opening rate reveal that 2 mM ATP increases Vmax from 0.8 ± 0.1 to 4.0 ± 0.4 without significantly changing Ka (Ka = 45 ± 8 µM and 60 ± 20 µM in the absence and presence of ATP, respectively).

The model predictions for the [Ca²⁺]_L-dependencies of _o and _c in the absence of ATP can be explained along similar lines to those presented above for ATP-activated RyRs. The contribution of Ca²⁺ feedthrough to _o depends on the steps LAI_2C L*AI_2O L*A*I_2O L*A*I*_2C. However, in the absence of ATP the states L*A*I_2O and L*AI_2O have approximately the same average duration. Therefore, although Ca²⁺ feedthrough biases the state occupancies to the right, in the absence of ATP this causes no significant increase in _o. Fig. 9 B shows the corresponding model predictions of channel opening rate. The decreased [Ca²⁺]_L-dependent opening rate in the absence of ATP was accommodated by a decrease in ' in the luminal activation step LAI_2CL*AI_2O.

Ca²⁺ feedthrough couples RyR₂ in lipid bilayers
In six instances, vesicle fusion events incorporated clusters of 3–6 RyR₂ into the bilayer. At subactivating [Ca²⁺]_C (100 nM Ca²⁺ and 2 mM ATP) and when conditions favored Ca²⁺ feedthrough, the opening of one RyR tended to promote the opening of other RyRs (i.e., channels displayed coupled gating). Fig. 10 shows a recording of six RyRs at positive and negative bilayer potentials. The current trace shows transitions between the current baseline (labeled C for closed) and equally spaced levels corresponding to 1–6 open channels. At –40 mV, channel openings were clearly grouped into bursts and the weighting of current levels in these records markedly deviated from a binomial distribution (not shown). Hidden Markov Model analysis of these records (see Materials and Methods) showed that the mean RyR opening rate associated with transitions between the current baseline and level O1 was 10x slower than opening rates associated with transitions between higher levels (Fig. 11 A, solid circle). Channel closing rates did not depend on the number of open RyRs (Fig. 11 B). Thus, RyR coupling appeared to be mediated by the opening rates. The degree of channel coupling was substantially reduced by decreasing [Ca²⁺]_L from 1 mM to 0.1 mM or by application of positive bilayer potentials (Fig. 11 A), conditions that oppose Ca²⁺ feedthrough. These results are consistent with the notion that the opening of one channel in the bilayer permits Ca²⁺ feedthrough that can trigger the opening of neighboring RyRs via their A-sites. The luminal-triggered Ca²⁺ feedthrough model was extended to include this possibility by introducing another parameter, Z, which describes the proportionality between I_Ca and its effect on [Ca²⁺]_P at a neighboring channel. The predicted opening and closing rates for the first opening in a cluster were calculated the same way as for single RyRs. The rates associated with the n^th channel opening were calculated using a value for [Ca²⁺]_P in Eqs. 7b and 10b that incorporated the additional term, .

View larger version (10K): [in this window] [in a new window]   FIGURE 11  The dependence of mean channel opening rate (A) and closing rate (B) on the number of open channels. Rates were measured in the presence of 2 mM ATP and 100 nM [Ca2+]C. Opening rates were significantly increased (P < 0.05, t-test) by the presence of other open channels under conditions that favored Ca2+ feedthrough (•, –40 mV [Ca2+]L = 1 mM, six measurements). When Ca2+ feedthrough was relatively low, this did not occur (, +40 mV [Ca2+]L = 1 mM; x, –40 mV [Ca2+]L = 0.1 mM, three measurements each). Closing rates did not significantly depend on the presence of other open channels in the bilayer. The data are compared with predictions of the luminal-triggered Ca2+ feedthrough model: •, solid line; x, long dashes; and , dashed line.

Model limitations
The luminal-triggered Ca²⁺ feedthrough model accounts for the Ca²⁺ regulation of RyR₂ by the actions of three Ca²⁺ binding sites that are linked by Ca²⁺ flowing through the pore. However, other Ca²⁺ regulation mechanisms could also play a role in controlling Ca²⁺ release from the SR. For example, the model does not include the low affinity Ca²⁺/Mg²⁺ inhibition mechanism (18). In addition, the [Ca²⁺]_C-activation mechanism is known to be complex and involves several Ca²⁺-dependent steps (24,25). Simplification of this into A- and I₂-site gating mechanisms should make the model unsuitable for predicting the complex shape of the dwell-time distributions. With this in mind, it is surprising that the model accounts so well for the [Ca²⁺]_L-dependent dwell-time distributions at subactivating [Ca²⁺]_C (Fig. 7). There is also the possibility that the L-site modulates the RyR in other ways than merely causing channel opening. An example of this has recently been reported in RyR₁ (14) where a luminal Ca²⁺ binding site was shown to modulate the A-site affinity for Mg²⁺ (preliminary data indicates that this phenomenon does not occur in RyR₂).

As yet, there has been no direct measure of the Ca²⁺ current under experimental conditions. Therefore, I_Ca is estimated from rate theory calculations based on an energy barrier model of the RyR pore (27). A possible failure of the energy barrier model to accurately predict I_Ca under some conditions could underlie the deviation of the model from the data (see the voltage-dependence in _o that occurred between +20 mV and +40 mV in Fig. 6, A and C). The rate theory model provided a good prediction of the total current in the channel (Cs⁺ plus Ca²⁺ in this instance, not shown). However, at +40 mV, I_Ca is relatively small and only represents 1% of the total current. Therefore, it is possible for the model to accurately predict the total current and yet be an order-of-magnitude out in estimating I_Ca. Just a twofold error in I_Ca would account for the deviations between the model and the data at +40 mV. Interestingly, _o at positive voltages could be better explained by a voltage-dependent equilibrium between Ca²⁺ on either end of the pore in which (Fig. 6 C, long dashes). In any case, the luminal-triggered Ca²⁺ feedthrough model provides a good fit to the data between –60 mV and +20 mV, which encompasses the membrane potential of the SR (0 mV, (32)).

The extension of the luminal-triggered Ca²⁺ feedthrough model to coupled RyR clusters involved a number of assumptions that are yet to be validated. In the model, coupling occurs between nearest neighbors, although it could also occur between more remote pores. It is not clear that, in the bilayer, RyRs do form the same arrays as they do in muscle. The model assumes that the action of Ca²⁺ feedthrough on neighboring channels can be predicted by the steady-state action of raised [Ca²⁺]_C. However, neighboring RyRs would be stimulated by a rapid stepwise increase in [Ca²⁺]_C and it is known that RyRs can exhibit gating properties that are peculiar to non-steady-state situations (33,34).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
The results obtained here support the proposition that RyR activation by Ca²⁺ stores is due to Ca²⁺ binding sites on both the luminal and cytoplasmic sides of the channel. This study shows that store regulation of RyRs and ER/SR excitability involves three modes of action associated with different parts of the RyR molecule; the L-site, A-site, and I₂-site (Fig. 1). The L- and I₂-sites have not been previously identified and this study makes the first measurement of their properties. The effects of [Ca²⁺]_L on opening rate in Fig. 6, B and C, highlight the existence of a luminal-facing Ca²⁺ activation site (L-site) with an affinity of 60 µM. The binding of Ca²⁺ to the L-site on its own can activate channel openings of 1 ms duration at rates of 1–10 s^–1. Once the channel is open, the flux of Ca²⁺ from the luminal to cytoplasmic sides of the channel increases the cytoplasmic [Ca²⁺] near the A-site and produces up to 30-fold prolongation of channel openings. Thus, a large component of the RyR activation by luminal Ca²⁺ (up to 97%) is due to the effects of Ca²⁺ feedthrough. However, without the luminal site to trigger the initial channel openings, activation by [Ca²⁺]_L would not occur. Therefore, the proposed mechanism for store activation of RyRs is luminal-triggered Ca²⁺ feedthrough, incorporating both the true luminal and feedthrough hypothesis. This hybrid mechanism would explain the apparently contradictory findings whereby enzyme digestion of luminal RyR domains abolishes luminal Ca²⁺ activation (13), but there is a close correlation between RyR activity and Ca²⁺ flux through the pore (10).

Ca²⁺-inactivation
This study makes the first demonstration in single channel recording of a high affinity, [Ca²⁺]_C-dependent inactivation in RyRs. It is manifest as a Ca²⁺-dependent reduction in channel _o (Figs. 2 A and 4 B). The affinity of the I₂-site is 1.2 µM, which is similar to that of the A-site (0.9 µM) so that both activation and inactivation occur over the same range of [Ca²⁺]_C. [Ca²⁺]_C-activation overrides the effects of the I₂-site, which causes 20% reduction in P_o in response to cytoplasmic Ca²⁺. However, the I₂-site causes up to 98% reduction in P_o in response to feedthrough of luminal Ca²⁺ in the presence of physiological (100 nM) [Ca²⁺]_C. This is because channel closures block Ca²⁺ feedthrough, leading to deactivation via the A-site and substantial lengthening of closed events.

It is well known that cytoplasmic Ca²⁺ and Mg²⁺ cause identical inhibition of RyRs at high concentrations (1 mM for RyR₁ and 10 mM for RyR₂ (35)). This inhibitory action is mediated by a low affinity nonspecific cation site (I₁-site, previously named the I-site). It was thought that inactivation by high [Ca²⁺]_L was mediated by Ca²⁺ feedthrough to the I₁-site (11) because until now, the I₁-site was the only Ca²⁺-dependent inhibition mechanism that had been clearly identified in RyRs. It is unlikely that the I₁-site causes inactivation by high [Ca²⁺]_L because inactivation has a similar [Ca²⁺]_L sensitivity in RyR₁ and RyR₂ (see (11) and (10)) whereas the Ca²⁺ sensitivity of the I₁-site differs 10-fold between the two isoforms. Moreover, the I₁-site shows no specificity between Ca²⁺ and Mg²⁺, whereas the luminal inhibition via the I₂-site is 100-fold less sensitive to Mg²⁺ than to Ca²⁺ (unpublished data).

Accessibility of A- and I₂-sites to luminal Ca²⁺
It is noteworthy that even though the A- and I₂-sites have similar affinity for cytoplasmic Ca²⁺, the effects of luminal Ca²⁺ on these sites occur over quite different ranges of [Ca²⁺]_L. The model indicates that the feedthrough effects of Ca²⁺ are 40-fold larger for the A-site than for the I₂-site (see X and Y in Table 1), which suggests a marked difference in the accessibility of these sites to luminal Ca²⁺. This difference is readily explained by the relative proximity of these sites to the Ca²⁺ pore (10). Ca²⁺ emanating from the pore will diffuse into the cytoplasm and be sequestered by buffering molecules (4.5 mM BAPTA). This leads to a decline in [Ca²⁺] with distance from the pore (see Eq. 1). The luminal-triggered Ca²⁺ feedthrough model predicts that each pA of Ca²⁺ current through the channel causes a 15 µM increase in [Ca²⁺] at the A-site and a 0.35 µM increase at the I₂-site (e.g., I_Ca = 4.9 pA at –40 mV when [Ca²⁺]_L = 1 mM hence [Ca²⁺]_P = 73 µM at the A-site and 1.7 µM at the I₂-site). This places the A- and I₂-sites at 11 nm and 26 nm from the pore, respectively. Given that the furthest point on the RyR from the pore is 20 nm (36), it would seem that the I₂-sites are located at the periphery of the protein or perhaps on an adjacent inhibitory protein. The relative proximity of the A- and I₂-sites to the pore is in accord with the observation that Ca²⁺ buffering of the cytoplasmic bath affects inactivation much more strongly than activation (10).

Cytoplasmic agonists and luminal Ca²⁺ activation
Several studies have shown that [Ca²⁺]_L-dependent activation of RyRs is mainly seen in the presence of agonists such as ATP, caffeine, or sulmazole (8,28,29). Previously it was proposed that these agonists unmask a luminal Ca²⁺ sensing site (6). The luminal-triggered Ca²⁺ feedthrough model points to a different mechanism in which the L-site affinity is not altered by agonists. Rather, they stabilize the channel open conformation, destabilize the closed conformation, and, in conjunction with Ca²⁺ feedthrough, markedly increase channel activation in response to [Ca²⁺]_L.

One of these agonists, ATP, is shown here to enhance [Ca²⁺]_L-activation of RyR₂ by three modes of action. Firstly, ATP increases the rate of opening of RyRs that occurs in response to Ca²⁺ binding at the L-site (Fig. 9 C). Secondly, in accord with previous findings (37), ATP is a cofactor that causes _o to increase in response to Ca²⁺ binding at the A-site. Therefore, Ca²⁺ feedthrough should increase _o in the presence of ATP, whereas in the absence of ATP, _o should not be altered. The results show this to be the case (Fig. 9 B). Thirdly, ATP decreases the rate of inactivation via the I₂-site. In regard to the last two actions, the effects of ATP on the [Ca²⁺]_L-dependence of _o were entirely consistent with Ca²⁺ feedthrough and the observed effects of ATP on the [Ca²⁺]_C-dependence of _o (Fig. 9, dashed and solid curves).

Previous studies (30,38–40) have shown that cytoplasmic agonists such as caffeine and ATP have quite different effects on RyR gating than Ca²⁺. Caffeine and ATP produce a much lower P_o and they generate much longer channel openings and closures than Ca²⁺ (Fig. 2 A, see top trace showing an RyR activated by ATP with bottom trace of an RyR activated by Ca²⁺ and ATP). The different forms of activation by Ca²⁺ and ATP can be understood in terms of the luminal-triggered Ca²⁺ feedthrough model. Firstly, cytoplasmic Ca²⁺ activation via the A-site can cause much faster activation rates than ATP, which is limited by the triggering of openings by the L-site (hence the longer _c and the lower P_max with ATP relative to Ca²⁺). Secondly, channels activated by cytoplasmic Ca²⁺ are more affected by I₂-site mediated inactivation than ATP-activated RyRs (hence _o is longer ATP than with Ca²⁺).

Another prediction of the model is that cardiac and skeletal RyR isoforms are differently regulated by luminal Ca²⁺. It is known that there are marked differences in the ways that RyR₁ and RyR₂ are regulated by the cytoplasmic milieu (41). Of particular relevance is that RyR₁ and RyR₂ are differently regulated by cytoplasmic ATP. RyR₁ can be activated by ATP in the absence of cytoplasmic and luminal Ca²⁺ (10,14). However, as shown here and elsewhere (28,42), ATP acts as a cofactor on RyR₂, which increases its activation in response to Ca²⁺ but does not in itself trigger channel openings. This difference in ATP regulation underlies important differences in the way that RyR₁ and RyR₂ respond to luminal Ca²⁺ (see Scheme V). It is proposed here that opening of RyR₂ at 100 nM [Ca²⁺]_C is primarily triggered by Ca²⁺ binding to the L-site whereas Ca²⁺ feedthrough serves to reinforce stability of channel opening. However, in RyR₁, cytoplasmic ATP will trigger channel openings so that the L-site is bypassed, resulting in a mechanism that is identical to that previously proposed for RyR₁ (10). It is possible that in the absence of ATP, RyR₁ would default to the RyR₂ mechanism for luminal Ca²⁺ regulation provided that RyR₁ also has an L-site.

Scheme V

More generally, this model predicts that any cofactor that prolongs channel openings triggered by cytoplasmic Ca²⁺ will promote RyR activation by luminal Ca²⁺