G4941K substitution in the pore-lining S6 helix of the skeletal muscle ryanodine receptor increases RyR1 sensitivity to cytosolic and luminal Ca2+

The ryanodine receptor ion channel RyR1 is present in skeletal muscle and has a large cytoplasmic N-terminal domain and smaller C-terminal pore-forming domain comprising six transmembrane helices, a pore helix, and a selectivity filter. The RyR1 S6 pore-lining helix has two conserved glycines, Gly-4934 and Gly-4941, that facilitate RyR1 channel gating by providing S6 flexibility and minimizing amino acid clashes. Here, we report that substitution of Gly-4941 with Asp or Lys results in functional channels as indicated by caffeine-induced Ca2+ release response in HEK293 cells, whereas a low response of the corresponding Gly-4934 variants suggested loss of function. Following purification, the RyR1 mutants G4934D, G4934K, and G4941D did not noticeably conduct Ca2+ in single-channel measurements. Gly-4941 replacement with Lys resulted in channels having reduced K+ conductance and reduced selectivity for Ca2+ compared with wildtype. RyR1-G4941K did not fully close at nanomolar cytosolic Ca2+ concentrations and nearly fully opened at 2 μm cytosolic or sarcoplasmic reticulum luminal Ca2+, and Ca2+- and voltage-dependent regulation of RyR1-G4941K mutant channels was demonstrated. Computational methods and single-channel recordings indicated that the open G4941K variant results in the formation of a salt bridge to Asp-4938. In contrast, wildtype RyR1 was closed and not activated by luminal Ca2+ at low cytosolic Ca2+ levels. A model suggested that luminal Ca2+ activates RyR1 by accessing a recently identified cytosolic Ca2+-binding site in the open channel as the Ca2+ ions pass through the pore.

The skeletal muscle ryanodine receptor ion channel (RyR1) is a 2,200-kDa homotetrameric ion channel that releases Ca 2ϩ ions from an intracellular Ca 2ϩ store, the sarcoplasmic reticulum (SR) 4 (1)(2)(3)(4)(5). The large RyR1 cytoplasmic N-terminal "foot" structure forms the site of regulation by multiple factors that include Ca 2ϩ , Mg 2ϩ , ATP, and in vitro ligands such as ryanodine and caffeine. The C-terminal transmembrane domain forms a pore in the SR membrane that conducts monovalent and divalent cations (ϳ800 pS with 250 mM K ϩ as conducting ion, ϳ150 pS with 50 mM Ca 2ϩ ) but is selective for divalent cations (permeability ratio of Ca 2ϩ /K ϩ ϳ7:1). Mutations in RyR1 give rise to central core disease and malignant hyperthermia muscle diseases (6). Many of the naturally occurring central core disease mutations are in the C-terminal pore-forming region and impair RyR1 activity and ion conductance.
RyR1 has a pore structure characteristic of the voltage-gated ion channel family (7)(8)(9)(10)(11)(12). The RyR1 pore-forming region has an inner S6 helix, a pore helix, and a selectivity filter GGGIGDE motif (Fig. 1). Mutagenesis and single-channel measurements showed that negatively charged luminal residues Asp-4899 and Glu-4900 and the cytosolic residues Asp-4938, Glu-4942, and Asp-4945 in the pore-lining S6 helix impact RyR1 ion permeation and selectivity (13,14). The RyR1 S6 pore-lining helix has also two conserved glycines, Gly-4934 and Gly-4941, whose replacement with uncharged residues with an increased sidechain volume altered RyR1 channel gating and ion permeation (15). High-resolution cryoelectron microscopy (cryo-EM) using open RyR1 (Fig. 1) and closed RyR1 (16) and cardiac muscle RyR2 isoforms (17) provided structural insights in the mechanisms of channel opening and closing. Binding sites for channel activators Ca 2ϩ , ATP, and caffeine were identified (16). It was also shown that the RyR2 pore constriction site of Ile-4868 in the closed state is shifted to  in the open state (corresponding to Ile-4937 and Gln-4933 in RyR1, respectively) (17). Ion-pulling simulations that generated an open-channel conformation of RyR1 (18) from the 3.8-Å closed state of RyR1 (7) also showed that the pore constriction site was shifted from Ile-4937 in the closed channel to  in the open channel.
The premise of this study was that Gly-4934 and Gly-4941 have critical roles in the ion permeation and Ca 2ϩ regulation of RyR1. We addressed this by replacing the two glycines with two oppositely charged residues, aspartic acid and lysine. The results indicate that RyR1-G4934D, G4934K, and G4941D mutants were not gated by Ca 2ϩ , did not conduct Ca 2ϩ , and had reduced K ϩ conductances in single-channel measurements. In contrast to the closed wildtype (WT) channel, a K ϩ -conducting RyR1-G4941K mutant channel was recorded at nanomolar Ca 2ϩ concentrations. The mutant channel was nearly fully opened by SR luminal Ca 2ϩ in a voltage-dependent manner, which suggested that luminal Ca 2ϩ activated RyR1-G4941K by accessing a cytosolic Ca 2ϩ -binding site. Computational methods suggested that RyR1-G4941K formed a salt bridge with Asp-4938. Single-channel measurements of RyR1-G4941K/D4938N and RyR1-G4941K/D4945N mutants validated the computational predictions.

Expression and functional properties of RyR1-Gly-4934 and RyR1-Gly-4941 mutants
To assess the functional role of two conserved glycines in the S6 pore-lining helix in RyR1 gating and ion permeation, Gly-4934 and Gly-4941 were mutated to negatively charged aspartic acid and positively charged lysine. RyR1 mutant protein expression levels in HEK293 cells ranged from ϳ50 to 100% of WT (Table 1). In initial experiments, expression of functional channels was studied using two assays. A cellular Ca 2ϩ release assay indicated the number of HEK293 cells that expressed Ca 2ϩconducting WT and mutant channels, using the Ca 2ϩ -releasing drug caffeine. Retention of function in membrane isolates was determined in an in vitro ligand-binding assay using [ 3 H]ryanodine, a plant alkaloid commonly used to assess RyR activity and content (19). Caffeine-induced Ca 2ϩ release was observed in 30 -50% of HEK293 cells that expressed WT-RyR1 (Fig. 2). Cells expressing RyR1-G4934D or -G4934K showed a caffeine response near background levels, whereas replacement of Gly-4941 with aspartic acid or lysine showed a caffeine response similar to WT in ϳ75% of cells. This suggests a reduced number of HEK293 cells expressed caffeine-sensitive, Ca 2ϩ -conducting mutant channels. The variable responses may have resulted from the non-homogeneous exposure to caffeine and removal of Ca 2ϩ by cellular transport systems. The in vitro [ 3 H]ryanodine-binding assay in membrane isolates showed no binding for RyR1-G4934D to levels for RyR1-G4941K that exceeded WT (Table 1). This indicated variable levels of retention of function of the mutants compared with WT. Major differences in cellular caffeine-induced Ca 2ϩ release response and [ 3 H]ryanodine binding in membrane isolates were observed for RyR1-G4934K and -G4941D. These may have resulted from RyR1-G4934K having a reduced sensitivity to caffeine and RyR1-G4941D having lost function during isolation or in the binding assay. An alternative possibility for reduced sensitivity to caffeine and loss of high-affinity [ 3 H]ryanodine binding but not Ca 2ϩ -dependent channel activity is considered in single-channel measurements described below.

Single-channel measurements with WT-RyR1 and Gly-4934 and Gly-4941 mutants
Single channel measurements were performed with purified WT and mutant RyR1s to determine how mutagenesis of Gly-4934 and Gly-4941 affected channel gating and ion permeation. WT and mutant RyR1 channels were solubilized using the zwitterionic detergent CHAPS, purified on sucrose gradients, and reconstituted in proteoliposomes by removing the detergent by dialysis (20). In recording single channels, we took advantage of the finding that RyRs are Ca 2ϩ -gated channels impermeant to Cl Ϫ and conduct K ϩ more efficiently than Ca 2ϩ . The upper left trace in Fig. 3A shows a partially activated single WT-RyR1 channel recorded using 0.25 M KCl and 2 M Ca 2ϩ on both sides of the bilayer. K ϩ conductance was 820 pS (Fig. 3A, right panel, and Table 1). Reduction to 0.1 M cytosolic Ca 2ϩ decreased channel open probability (P o ) to near zero (Fig. 3A, 2nd trace on left, and Table 1). In the presence of 10 mM SR luminal Ca 2ϩ , a Ca 2ϩ current of Ϫ2.3 pA at 0 mV and reversal potential (E rev ) of 9.2 were obtained (Fig. 3A, 3rd trace, right panel). Applying constant field theory, E rev of 9.2 mV resulted in a permeability ratio of Ca 2ϩ over K ϩ (P Ca /P K ) of 6.6 for WT (Table 1).

Luminal RyR1 Ca 2؉ gating
reduced P o of RyR1-G4934K about 2-fold, whereas RyR1-G4934D and G4941D maintained a similar P o at 2 and 0.1 M cytosolic Ca 2ϩ . The three RyR1-G4934K, G4934D, and G4941D mutants exhibited a reduced K ϩ conductance and did not conduct Ca 2ϩ . We considered the possibility that exposure to detergent during purification caused loss of function. However, RyR1-G4934D, G4934K, and G4941D also had a reduced K ϩ conductance and did not conduct Ca 2ϩ when membrane fractions isolated from HEK293 cells were fused with lipid bilayers (data not shown). This suggested that RyR1-G4934D, G4934K, and G4941D mutations had a major impact on channel structure and impaired ion conductances in single-channel measurements.
RyR1-G4941K conducted Ca 2ϩ and had a reduced K ϩ conductance of 364 pS between Ϯ40 mV compared with 820 pS for WT with 0.25 M KCl on both sides of the bilayer (Fig. 3E and Table 1). Changes in KCl concentration indicated maximal K ϩ Table 1 Single channel measurements with wildtype and mutant RyR1s The following abbreviations are used: Ry, ryanodine; ND, not determined.    Table 1.

Luminal RyR1 Ca 2؉ gating
conductances for WT-RyR1 and RyR1-G4941K near 600 mosM K ϩ (Fig. 4). Maximal conductances and apparent K D values were 453 pS and 38 mM for G4941K versus 943 pS and 28 mM for WT. In the presence of 10 mM luminal Ca 2ϩ , the Ca 2ϩ current (Ϫ0.6 pA versus Ϫ2.3 pA for WT at 0 mV) and permeability ratio of Ca 2ϩ over K ϩ (P Ca /P K ϭ 2.1 versus 6.6 for WT) were greatly reduced (Table 1). Channel open probability of RyR1-G4941K was elevated at 2 M cytosolic Ca 2ϩ (0.89 versus 0.13 for WT) and 0.1 M Ca 2ϩ (0.56 versus 0.01 for WT) (Fig. 3E, Table 1). At 2 M Ca 2ϩ in the bath solution, the RyR1-G4941K mutation increased the duration of open events (34.6 ms versus 0.47 ms for WT) and decreased the duration of the closed events (0.32 ms versus 6.5 ms for WT) ( Table 2). The results suggest that RyR1-G4941K exhibited altered Ca 2ϩ -dependent gating and reduced ion conductances compared with WT. As a control, RyR1-G4941M was analyzed with methionine having a side-chain volume comparable with lysine. RyR1-G4941M exhibited caffeine-induced Ca 2ϩ release in HEK293 cells (Fig. 2) and bound [ 3 H]ryanodine in an in vitro ligandbinding assay, albeit at reduced levels compared with WT (Table 1). Single-channel measurements showed that replacement of a positively charged residue in G4941K with methionine with a comparable side-chain volume but uncharged side chain resulted in channel open probability, K ϩ conductance, and permeability ratios of Ca 2ϩ over K ϩ close to WT ( Fig. 3F and Table 1).

Activation of RyR1-G4941K by 2 M cytosolic or 2 M luminal Ca 2؉
Dependence of WT-RyR1 and RyR1-G4941K single-channel activities by 2 M cytosolic or 2 M luminal Ca 2ϩ was determined at Ϫ35 mV membrane potential, which promoted the movement of luminal Ca 2ϩ to the cis cytosolic side of the bilayer (Fig. 5). In the presence of 0.01 M cytosolic and luminal Ca 2ϩ , WT P o was essentially 0, and G4941K P o was 0.13. An increase in luminal or cytosolic Ca 2ϩ to 2 M Ca 2ϩ nearly fully activated RyR1-G4941K (P o ϭ 0.86 and 0.98, respectively), whereas WT was incompletely activated by 2 M cytosolic Ca 2ϩ (P o ϭ 0.16). The results indicate that in contrast to WT, (i) RyR1-G4941K had a P o Ͼ0.1 at low cytosolic and luminal Ca 2ϩ , which indicated Ca 2ϩ -independent open channel state(s), and (ii) was nearly fully opened (P o close to 1.0) by 2 M luminal or 2 M cytosolic Ca 2ϩ .

Altered regulation of RyR1-G4941K by cytosolic Ca 2؉ in the presence of 2 M luminal Ca 2؉
In agreement with rapid 45 Ca 2ϩ flux measurements of SR vesicles (21). WT-RyR1 channels were activated by micromolar cytosolic Ca 2ϩ and inhibited by millimolar cytosolic Ca 2ϩ at Ϫ35 mV (Fig. 6). Ca 2ϩ -dependent regulation of G4941K differed in several respects from WT. G4941K had elevated open probability at 0.01 M cytoplasmic Ca 2ϩ in the presence of 2 M luminal Ca 2ϩ . G4941K was also nearly fully activated at 2 Ϫ100  Table 1.

Luminal RyR1 Ca 2؉ gating
M cytosolic Ca 2ϩ and displayed elevated activity at 10 mM cytosolic Ca 2ϩ compared with WT.

RyR1-G4941K regulation by luminal Ca 2؉
RyR1 has been reported to be regulated by luminal Ca 2ϩ by binding to luminal channel sites (22) or cytosolic sites following passage of luminal Ca 2ϩ through the open channel (23). To distinguish these mechanisms, dependence of channel open probability on the luminal Ca 2ϩ concentration was probed at Ϫ35 and ϩ35 mV (Fig. 7). Cytosolic Ca 2ϩ was 0.01 or 2 M. In WT at 0.01 cytosolic Ca 2ϩ , an increase from 0.01 to 10 mM luminal Ca 2ϩ did not increase the low P o at Ϫ35 or ϩ35 mV ( Fig. 7A). At 2 M cytosolic Ca 2ϩ , an increase of luminal Ca 2ϩ from 0.01 to 10 mM significantly decreased P o at Ϫ35 mV but not at ϩ35 mV. Regulation of RyR1-G4941K by luminal Ca 2ϩ differed greatly from WT (Fig. 7B). At 2 M cytosolic Ca 2ϩ , RyR1-G4941K was fully activated at both membrane potentials at 0.01 M to 10 mM luminal Ca 2ϩ At 0.01 M cytosolic Ca 2ϩ , which kept the mutant channel open, luminal Ca 2ϩ was more effective in increasing P o at negative than positive membrane potentials. This suggested that luminal Ca 2ϩ flowing through the channel accessed cytosolic Ca 2ϩ activation sites in RyR1-G4941K. This was tested further by recording voltage dependence of RyR1-WT and -G4941K channel activities at 0.01 M cytosolic Ca 2ϩ and 0.01 or 2 M luminal Ca 2ϩ (Fig. 8). At 0.01 and 2 M luminal Ca 2ϩ , WT channels were closed, displaying no voltage-dependent activity. At 0.01 M luminal Ca 2ϩ , RyR1-G4941K displayed a weak voltage-dependent P o at membrane potentials ranging from Ϫ80 to ϩ80 mV. At 2 M luminal Ca 2ϩ , P o was higher at negative than positive membrane potentials. Thus, a larger number of Ca 2ϩ ions appeared to flow through the channel at negative than positive membrane potentials to activate RyR1-G4941K to a greater extent at negative potentials.

RyR1-WT and -G4941K activities in the presence of FKBP12 and in membrane isolates
We considered the possibility that elevated RyR1-G4941K channel open probability may result from the absence of FKBP12, a closely associated subunit of RyR1. Absence of FKBP12 in the WT-RyR1 macromolecular complex activates Ca 2ϩ release from the SR and increases channel activity in lipid bilayers (24,25). Proteoliposomes containing purified WT and RyR1-G4941K channels were therefore treated for 30 min with 5 M FKBP12 before addition to the cis bilayer chamber. Treatment of WT and RyR1-G4941K with FKBP12 did not significantly affect P o at 0.1 and 2 M Ca 2ϩ compared with channels recorded in the absence of FKBP12, respectively (Fig. 9, A and B). To increase   Luminal RyR1 Ca 2؉ gating the probability of preserving normal function, experiments were also performed with channels that were not subjected to detergent treatment. Fig. 9, A and B, shows that RyR1-G4941K channels obtained using membrane preparations isolated from HEK293 cells maintained an elevated P o at 0.1 and 2 M cytosolic Ca 2ϩ compared with WT, respectively. The results indicate that preincubation with FKBP12 and elimination of detergent treatment did not decrease the single-channel activities of RyR1-G4941K to those of WT.

Discussion
The principal finding of this study is that RyR1-G4941K exhibits a marked increase in the sensitivity of SR luminal Ca 2ϩ regulation in single-channel recordings. The RyR1-G4941K mutant exhibited a caffeine-induced Ca 2ϩ response in HEK293 cells and bound [ 3 H]ryanodine, but it had reduced Ca 2ϩ over K ϩ permeability ratio compared with WT. This was in contrast to the finding that RyR1-G4934D and -G4934K mutants lost caffeine-induced Ca 2ϩ response in HEK293 cells, had a pronounced decrease in [ 3 H]ryanodine binding, and Ca 2ϩ currents decreased to background levels. These findings suggest loss-offunction channels in HEK293 cells, membrane isolates, and single-channel measurements. By comparison, the G4941D mutation resulted in a more complex behavior. RyR1-G4941D responded to caffeine, which suggested the formation of a functional Ca 2ϩ -conducting channel in HEK293 cells. However, [ 3 H]ryanodine binding was low, and Ca 2ϩ currents decreased to background levels, which suggested loss of function on removal from HEK293 cells. The mutagenesis results support the modeling data described below that RyR1-G4941K forms a salt bridge to Asp-4938 and luminal Ca 2ϩ accesses a recently identified Ca 2ϩ -binding site in the open-channel conformation.
Our results suggest that the RyR1-G4941K mutant undergoes a structural change during extraction from HEK293 cells. The caffeine-induced Ca 2ϩ release suggested an intracellular Ca 2ϩ store in HEK293 cells in the absence of caffeine. In contrast, single-channel measurements suggested that RyR1-G4941K rapidly released Ca 2ϩ from the intracellular membrane compartment. Neither the potential loss of the FKBP12 subunit from the RyR1 channel complex during purification nor exposure to detergent was found to be responsible for the elevated channel activity in single-channel measurements. The results may be interpreted that the mutant channel underwent a structural change upon removal from the cellular environment.
Previous single-channel studies showed that mutagenesis of the two conserved glycines in the pore-lining S6 pore-lining helix altered RyR function. Substitution of RyR2-Gly-4864 (corresponding to RyR1-Gly-4934) with alanine resulted in no significant changes of RyR2 function, whereas replacement with valine and proline profoundly altered both channel gating and ion permeation (26). Single-channel studies were done using purified RyR2 channel preparations. We reported that replacement of RyR1-Gly-4934 and Gly-4941 with Ala altered RyR1 channel function in single-channel measurements using purified RyR1 preparations (15). A mutation further increasing the side-chain volume at Gly-4934 (G4934V) resulted in loss of function. In contrast, function was maintained when membrane fractions isolated from HEK293 cells were fused with lipid bilayers (18). In this study, loss of function by the G4934D and G4941D mutant channels was seen when proteoliposomes containing the purified mutants or membrane fractions were fused with lipid bilayers.
Aspartic acid has a smaller residue volume than valine (115 Å 3 versus 138Å 3) (27). This suggested that the introduction of additional negative charges near Asp-4938 and Glu-4942 rendered RyR1-G4934D structurally unstable in HEK293 cells and RyR1-G4941D when removed from the cellular environment.
Modeling the RyR1-G4941K mutation in the open-channel conformation revealed the preferential formation of an intersubunit salt bridge with Asp-4938 (Fig. 10) compared with interaction with Asp-4945. Among the four chains the interaction persists for 34% of the simulation, meaning that on average between one and two salt bridges with Asp-4938 are present at any point in the simulation. Single-channel recordings suggested that the formation of a salt bridge stabilizes the open channel state. In the G4941K/D4945N double mutant, we observed an increase in the G4941K and Asp-4938 interaction in the open-channel conformation and in single-channel measurements a prolonged duration of channel openings. As expected, mutating Asp-4938 to Asn attenuated salt-bridge formation with D4938N. G4941K now showed a preference for intrasubunit salt bridge formation with Asp-4945, and single- Time analysis of the single-channel data is in agreement with these conclusions. The RyR1-G4941K/D4945N double mutant showed a substantial increase in T o compared with RyR1-G4941K and D4945N, which indicated increased stability of the open-state or a barrier to return to the closed state. In contrast, the G4941K/D4938N double mutation resulted in an elevated T c compared with G4941K and D4938N, which implies that the closed state was more stable or that there was an additional barrier to opening. As a control, we modeled the open channel conformation of the G4941M mutation and monitored the distances between G4941M and Asp-4938 or Asp-4945 (Fig. 11). We observed closer contacts of G4941M with Asp-4945; however, neither showed significant contacts below the 3.5 Å range required for salt-bridge formation.
Single-channel measurements showed that replacement of glycine with positively charged lysine reduced K ϩ conductance and Ca 2ϩ selectivity of RyR1-G4941K, G4941K/D4938N, and G4941K/D4945N mutants to a greater extent than G4941M, D4938N, and D4945N mutants. Pore profiles for the openchannel RyR1-G4941K, G4941M, G4941K/D4938N, and G4941K/D4945N mutants are compared with WT in Fig. 12A.  5t9v, 5tal, and 5ta3). Distances less than 3.5 Å are considered as potential salt bridges. For G4941K, we observe a salt bridge preference between G4941K and Asp-4938. When Asp-4945 is mutated to Asn, we observe an increased preference for the G4941K-Asp-4938 salt bridge. As expected, mutating Asp-4938 to Asn attenuates the salt-bridge formation, and G4941K now shows a preference for Asp-4945.

Figure 11. G4941M contacts in the open channel.
A, as a control the conformations of G4941M are shown with respect to Asp-4938, Asp-4945, and the pore. G4941M is observed to move into an intersubunit pocket that is out of the pore and below Asp-4938 explaining contacts between Asp-4938 and G4941M and the increased pore radius of the G4941M mutation compared with G4941K. The two right panels monitor distances between G4941M and Asp-4938 (B) or Asp-4945 (C). We observe more close contacts between G4941M and Asp-4938 than between G4941M and Asp-4945; however, neither show significant contacts below 3.5 Å (the range expected for salt-bridge formation).

Luminal RyR1 Ca 2؉ gating
The four mutations had a pore radius comparable with WT at Gly-4894, a site of the selectivity filter, and Gln-4933, the pore constriction site of the open RyRs. In contrast, the G4941K mutation decreased the pore radius from ϳ5 Å in the WT open channel to ϳ3 Å. Combining the D4945N mutation with G4941K preserved the radius of RyR1-G4941K but decreased the variance of the radius, likely due to an increased preference for an intersubunit salt bridge between G4941K and Asp-4938. The G4941K/D4938N double mutation resulted in a small decrease in pore radius compared with G4941K, which was likely due to the decreased formation of intersubunit salt bridges. The decreased formation of intersubunit salt bridges caused G4941K to either protrude into the pore or form an intrasubunit salt bridge with Asp-4945, which caused the ␤, ␥, or ␦ carbons of lysine to protrude into the pore. RyR1-G4941M had a pore radius of ϳ4.5 Å. A move out of the pore into an intersubunit pocket may explain the larger pore radius for methionine compared with G4941K.
The data of Fig. 12A suggest that the flow of ions through the channel will be likely limited at two additional sites (RyR1-Gly-4894 and -Gln-4933). Therefore, the changes in pore size at RyR1-Gly-4941 should not be sufficient to account for the reduced ion conductances of RyR1-G4941K, G4941K/D4938N, and G4941K/D4945N. However, the introduction of positive charges from lysine must be also considered. The change in electrostatics was likely to have had a greater effect on Ca 2ϩ carrying a ϩ2 charge than on K ϩ carrying a ϩ1 charge. To estimate the electrostatics of the ions within the pore, we calculated the electrostatic Ca 2ϩ -and K ϩ -protein interaction energies using APBSmem software tool (28,29). Fig. 12, B and C, shows the interaction energies of the pore residues with Ca 2ϩ and K ϩ ions as the two ions move through the pore. The results show that the electrostatic interaction energies of WT varied between Ϫ5 and Ϫ2 kcal/mol for Ca 2ϩ and Ϫ2.5 and Ϫ1 kcal/mol for K ϩ . A rightward shift from the RyR1-G4941K position was observed for the energy maxima. The negative interaction energies of Ca 2ϩ and K ϩ are likely primarily due to a favorable interaction with negatively charged amino acid residues (RyR1-Asp-4899 and -Glu-4900) immediately after the GGGIG selectivity motif in the luminal vestibule and negatively charged residues in the cytoplasmic vestibule (Fig. 1A). Substitution of glycine in RyR1-G4941K with four positively charged lysine increased interaction energies for Ca 2ϩ and K ϩ to ϩ6 and ϩ3.3 kcal/mol, respectively. A further moderate increase in the interaction energies was obtained when the negatively charged aspartate residues in Gly-4941/D4938 and Gly-4941/ D4945 were replaced with asparagine. Together, the results of Fig. 12 suggest that the introduction of positive charges in G4941K and the further reduction of negative charges in the two double mutants introduced energetic and structural barriers, both of which likely contributed to the reduced ion conductances and selectivities of RyR1-G4941K, G4941K/D4938N, and G4941K/D4945N.
Recent cryo-EM studies identified a binding site for Ca 2ϩ at the interface of the RyR1 CSol and C-terminal domains (16). Earlier studies indicated that mutations at other sites affected Ca 2ϩ -gated channel activity. Substitution of RyR1-Glu-4032 by alanine and the corresponding amino acid residues in the RyR3 isoform reduced Ca 2ϩ -gated channel activity (30,31). RyR1-Glu-4032 is not part of the Ca 2ϩ -binding site but was suggested to stabilize the interaction between two RyR1 regions by H ϩ bonding to amide nitrogens (16). Mutations that give rise to central core disease in skeletal muscle exhibit loss of Ca 2ϩ -dependent channel activity and Ca 2ϩ conductance (32-34). Many Figure 12. Pore profiles and electrostatic interaction energies for WT and G4941 mutant channels. A, Gly-4941 mutations decreased the pore radius from ϳ5 Å in the WT open channel. The G4941K mutation shows a 3-Å pore radius with a significant amount of fluctuation. Introducing the D4945N mutation to the G4941K mutant preserves the radius of the G4941K mutant but significantly decreases the variance of the radius, likely due to increased preference for intersubunit salt bridges between G4941K and Asp-4938. The G4941K/D4938N double mutant shows a decreased pore radius compared with the G4941K mutant likely due to the formation of less salt bridges and the formation of the G4941K-Asp-4945 intrasubunit salt bridge. A move out of the pore may explain the larger pore radius for methionine compared with lysine in Gly-4941. B and C, Ca 2ϩ -and K ϩ -protein interaction energies in WT and mutants. Changes in the electrostatics of Ca 2ϩ (B) and K ϩ (C) ion passages through the pore in WT-RyR1, G4941K, G4941K/D4938N, and G4941K/ D4945N. The interaction energies were estimated by performing APBSmem calculations on 10 structures, which were sampled evenly over the last 5 ns of the respective trajectories. For each of these 10 structures, Ca 2ϩ or K ϩ were moved through the pore along the channel axis with a step size of 1.0 Å. Error bars represent the difference between the means with a 95% confidence interval.

Luminal RyR1 Ca 2؉ gating
of the central core disease mutations are located in the poreforming region of RyR1. To our knowledge, G4941K mutant has not been associated with a human pathology.
Other sites involved in the regulation of RyR1 by Ca 2ϩ include residues in S2 (35), the S4-S5 linker (36,37), and S6 of the cardiac RyR2 isoform corresponding to RyR1 amino acid residues Gln-4946 to Met-4954 of RyR1 (38). This study extends these findings and shows that sensitivity of RyR1-G4941K to cytosolic and luminal Ca 2ϩ increased compared with WT.
Regulation of the RyRs by luminal Ca 2ϩ has been attributed to luminal Ca 2ϩ -sensing channel sites (22, 39 -41), the binding to cytosolic sites following the passage of Ca 2ϩ through the open channel (23,42,43), or the involvement of Ca 2ϩ -sensing sites on both the luminal and cytosolic channel sites (44). To distinguish between luminal and cytosolic Ca 2ϩ -regulatory sites, we previously determined the voltage-and Ca 2ϩ -dependent regulation of WT-RyR1 (23). As shown in this study, closed WT-RyR1 channels are not activated by luminal Ca 2ϩ , and single channels were partially opened by cytosolic ATP in the presence of a low cytosolic Ca 2ϩ concentration. Channels were activated by 50 M luminal Ca 2ϩ to a greater extent at negative membrane potentials that favored Ca 2ϩ flow through the RyR1 pore than at positive membrane potentials that disfavored passage of Ca 2ϩ through RyR1. This suggested that luminal Ca 2ϩ bound to cytosolic channel sites following their passage through the open channel. A similar result was obtained when the partially open G4941K mutant channel was recorded at a low cytosolic Ca 2ϩ concentration. G4941K was activated to a greater extent by 2 M luminal Ca 2ϩ at negative than positive membrane potentials. These results are best rationalized by luminal Ca 2ϩ accessing cytosolic channel sites.
A model consistent with the results of this study suggests that cytosolic Ca 2ϩ ions access a recently identified Ca 2ϩ activation site located in the cytosolic foot structure of RyR1 (16). The model suggests that cytosolic Ca 2ϩ can access the Ca 2ϩ activation site when channels are closed or open (Fig. 13) (17). According to the model, luminal Ca 2ϩ activates RyR1 when gaining access to the Ca 2ϩ activation site following passage through the open-channel RyR1-Gln-4933 restriction site. RyR1-G4941K differs in two respects from WT in its regulation by Ca 2ϩ . First, the RyR1-G4941K mutant channel remains open at low cytosolic Ca 2ϩ concentrations; second, it exhibits an increased sensitivity to activation by cytosolic and luminal Ca 2ϩ compared with WT. Our results suggest that luminal Ca 2ϩ accesses Ca 2ϩ activation sites as they pass through the pore rather than traveling to openings that lie outside the pore.

Materials
[ 3 H]Ryanodine was obtained from PerkinElmer Life Sciences; protease and phosphatase inhibitors were from Sigma, and phospholipids were from Avanti Polar Lipids.

Preparation of wildtype and mutant channels
RyR1 mutations G4934D, G4934K, G4941D, and G4941K were introduced by Pfu polymerase-based chain reaction using mutagenic oligonucleotides and the QuikChange II site-directed mutagenesis kit (Stratagene, La Jolla, CA) (45). RyR1-G4941M, G4941K/D4938N, and G4941K/D4945N were prepared using a gene synthesis method and a proprietary protocol (Genewiz, Inc., South Plainfield, NJ). WT-RyR1 and mutants were transiently expressed in HEK293 cells using FuGENE 6 (Roche Applied Sciences) or jetPRIME (Polyplus, New York) according to the manufacturers' instructions. Cells were maintained in DMEM/F-12 (1:1) medium containing 10% fetal bovine serum at 37°C and 5% CO 2 and plated the day before transfection. Following transfection, cells were maintained at 35 or 37°C and harvested ϳ70 h after transfection. Crude membrane isolates (45) and proteoliposomes containing purified WT or mutant RyR1 channels were prepared as described (20).

Cellular Ca 2؉ release
Stored Ca 2ϩ release was determined as described (46). Briefly, HEK293 cells were grown on glass coverslips, and Ca 2ϩ transients were monitored with the fluorescence Ca 2ϩ indicator Fluo-4. The addition of ϳ8 mM caffeine induced cellular Ca 2ϩ release. Individ- Luminal RyR1 Ca 2؉ gating ual cells were monitored using the EasyRatioPro algorithm (Photon Technology International, Lawrenceville, NJ).

SDS-PAGE and immunoblot analyses
Proteins in crude membrane fractions (20 g of protein/ lane) were separated using 3-12% gradient SDS-PAGE, transferred overnight to nitrocellulose membranes (37), and probed with primary rabbit anti-RyR1 polyclonal antibody 6425 prepared by ⌿ProSci against RyR1 sequence FIKGLDSFSGK-PRGSG. Immunoblots were developed using enhanced chemiluminescence and quantified using the Bio-Rad ChemiDoc MP Imaging System and ImageQuantTL analysis software.

[ 3 H]Ryanodine binding
Ryanodine binds with high specificity to RyR1 and is widely used to probe RyR activity and content (19).

Single-channel recordings
Single-channel measurements were performed using planar lipid bilayers (14). Unless otherwise indicated, proteoliposomes containing purified recombinant RyR1s were added to the cis (cytosolic) chamber of a bilayer apparatus containing 0.25 M KCl in the cis chamber and 20 mM KCl in the trans (SR luminal) chamber in 20 mM KHEPES, pH 7.4, and 2 M Ca 2ϩ . After the appearance of channel activity, the trans KCl concentration was increased to 0.25 M KCl. A strong dependence of WT channel activity on cis Ca 2ϩ concentration indicated that the cytosolic region faced the cis chamber of the bilayer. The trans side of the bilayer was defined as ground. Electrical signals were filtered at 2 kHz (0.5 kHz for Ca 2ϩ currents at 0 mV), digitized at 10 kHz, and analyzed at 50% threshold setting (14). Data acquisition and analysis of 2-min recordings were performed using a commercially available software package (pClamp, Axon Instruments). Single-channel activities were also recorded in symmetrical 0.25 M KCl solution with 10 mM Ca 2ϩ on the trans side, and the reversal potential (E rev ) was measured to determine permeability ratios. The permeability ratio of Ca 2ϩ versus K ϩ (P Ca /P K ) was calculated using a modified form of the Goldman-Hodgkin-Katz Equation

Computational methods
To model the effects of the mutations, in silico mutations were performed for each of the three open-channel cryo-EM structures (PDB codes 5tal, 5t9v, and 5ta3). Amino acid substitutions were performed using Eris (47,48) followed by side-chain optimization in GROMACS 4.6.1 (49). Starting structures from Eris were obtained from 40 independent Monte-Carlo simulations with 20 steps each. The lowest energy structure among the 40 structures was selected as the starting point for GROMACS simulations. Each system was solvated using the TIP3P water model. Simulations were performed using the CHARMM36 force field (50, 51). The systems were equilibrated with 10,000 steps of steepest descent energy minimization followed by 200 ps of constant volume MD simulation performed at 310 K. Simulations were performed using a time step of 2 fs with all bonds constrained using the LINCS algorithm (52). Particle mesh Ewald was used for long-range electrostatic interactions. A 10-Å cutoff was used for nonbonded interactions. For production runs, 10 ns of MD simulations were performed at constant temperature and pressure of 310 K and 1 atm followed by 4 ns of annealing. For annealing simulations, pressure was held constant at 1 atm, whereas temperature was linearly decreased from 310 to 200 K. A 10,000 kJ/(mol⅐nm) potential was applied to fix the backbone atoms throughout the simulations and prevent large structural arrangements in the absence of the membrane. For analysis, 10 structures sampled evenly over the last 5 ns of the simulation were extracted. Pore profiles were calculated using the HOLE program (53). Data are the mean and standard error among 30 pore profiles computed from the sampled structures.
The Adaptive Poisson-Boltzmann Solver (APBS) as applied in APBSmem (28) was used to examine the electrostatics of Ca 2ϩ and K ϩ ion passage through the pore. APBSmem estimates the protein-ion electrostatic interaction energy by subtracting the energies of the membrane-embedded protein without the ion and the energy of the ion in bulk water from the total energy of the protein-ion assembly embedded in the lipid bilayer. For analysis, 10 WT and mutant structures sampled evenly over the last 5 ns of the respective simulations were extracted. Two focusing layers with a grid length of 90 Å and a total of 97 grid points in the x, y, and z dimensions were employed. Ion concentrations were set at 0.1 M, both for Ca 2ϩ and K ϩ ; the temperature was 298 K, and non-linear Poisson-Boltzmann method was engaged in Adaptive Poisson-Boltzmann Solver (29). To omit the membrane from the pore region, the upper and lower exclusions were set at 16 Å. Based on previously reported calculations of Ca 2ϩ solvation energy profile of the TRPV1 receptor (54), membrane thickness was 42.5 Å, and headgroup thickness was 7 Å. During the calculation, the dielectric constant for protein and membrane was set at 2.0, and the dielectric constant for headgroups and solvent was set at 80.0. The ion step size was 1.0 Å along the channel axis and the ionic radius used for Ca 2ϩ and K ϩ were 1.03 and 1.41 Å, respectively.

Biochemical assays and data analysis
Free Ca 2ϩ concentrations were determined by adding varying amounts of Ca 2ϩ to 1 mM EGTA solutions. Free Ca 2ϩ concentrations were measured using a Ca 2ϩ -selective electrode. Differences between samples were analyzed using Student's t Luminal RyR1 Ca 2؉ gating test or analysis of variance followed by Tukey's test, where p Ͻ 0.05 was considered significant.