Research Articles

Structural basis for the gating mechanism of the type 2 ryanodine receptor RyR2

See allHide authors and affiliations

Science  21 Oct 2016:
Vol. 354, Issue 6310, aah5324
DOI: 10.1126/science.aah5324

Gating a calcium channel

The type 2 ryanodine receptor (RyR2) controls the release of calcium ions from the sarcoplasmic reticulum in cardiac cells—the initiating step in cardiac muscle contraction. Mutations in RyR2 are associated with cardiac diseases. Peng et al. used single-particle electron cryomicroscopy to determine the structure of RyR2 from porcine heart at 4.4-Å resolution with the calcium channel closed and at 4.2-Å resolution with the calcium channel open. The structures reveal how interdomain motions result in a conformational change in the cytoplasmic region of RyR2 that is transduced by a central domain to cause motions that open or close the channel.

Science, this issue p. 301

Structured Abstract

INTRODUCTION

Ryanodine receptors (RyRs) are intracellular Ca2+ channels that control the release of Ca2+ from the sarco(endo)plasmic reticulum. Among the three mammalian RyR isoforms, RyR2 is primarily expressed in the heart and brain and is activated by Ca2+ influx by a mechanism known as calcium-induced calcium release. This Ca2+ release is fundamental to cellular processes ranging from muscle contraction to learning and memory.

RyRs are the largest known ion channels with a molecular mass of > 2 megadalton for a homotetramer. Each RyR protomer consists of a cytoplasmic region of over 4500 residues and a carboxyl terminal transmembrane (TM) domain. The four identical TM segments enclose a central ion-conducting pore, whereas the cytoplasmic regions serve as a scaffold for interactions with diverse ligands and protein modulators.

RyR channel gating involves a long-range allosteric mechanism. The structures of rabbit skeletal muscle RyR1 were determined at 3.8-Å resolution for the closed state and various lower resolutions for the potentially open states. Elucidating the gating mechanism of RyR requires structural determination of the open states at high resolution.

RATIONALE

RyR2 harbors more than 150 mutations associated with cardiac disorders, such as catecholaminergic polymorphic ventricular tachycardia type 1, idiopathic ventricular fibrillation, and sudden cardiac death. Structural elucidation of RyR2 may provide the molecular basis for understanding disease mechanisms and for the development of potential novel therapeutics. We purified endogenous RyR2 from porcine heart using glutathione S-transferase–fused FKBP12. To obtain the structure of RyR2 in the closed state, 5 mM EDTA was included throughout purification. To capture an open RyR2, the protein was purified in the presence of 20 μM Ca2+ and the compound 2,2′,3,5′,6-pentachlorobiphenyl (PCB95) that can stabilize RyR1 in the open state.

RESULTS

The electron microscopy (EM) maps for RyR2 were reconstructed to 4.4- and 4.2-Å resolutions for the closed and open states, respectively. Compared to the structure of RyR1, a number of armadillo repeats in the C terminus of the helical domain 2 were invisible in RyR2, likely due to intrinsic flexibility. At 20 μM Ca2+, the cytoplasmic gate in the closed structure is dilated by approximately 8 Å, resulting in the shift of the constriction site from Ile4868 to Gln4864. The four Gln4864 residues enclose a gate with a diameter of approximately 4 Å, allowing Ca2+ passage in a single file.

CONCLUSION

Structure comparison of the open and closed RyR2 shows little intradomain rearrangement of the armadillo-containing cytoplasmic domains including the amino terminal domain, the Handle domain, and the Helical domain. Relative shifts between these domains result in the breathing motion of the periphery of the cytoplasmic canopy and the rotation of the Central domain. The horseshoe-shaped Central domain, with its convex side interacting with the three armadillo domains and its concave side wrapping around the cytoplasmic O-ring of the channel domain, serves as the primary transducer that integrates and translates the conformational changes of the cytoplasmic domains to channel gating. However, the mechanism of Ca2+ sensing and activation of RyR2 remains to be elucidated.

Cryogenic EM structures of RyR2 from porcine heart in both the closed and open states at near-atomic resolutions.

(Top) Representative two-dimensional class averages of electron micrographs of the closed and open RyR2. (Bottom) The two structures are superimposed relative to the transmembrane domain. The blue arrows indicate the overall shifts of the cytoplasmic region from the closed state to the open state. SR, sarcoplasmic reticulum.

Abstract

RyR2 is a high-conductance intracellular calcium (Ca2+) channel that controls the release of Ca2+ from the sarco(endo)plasmic reticulum of a variety of cells. Here, we report the structures of RyR2 from porcine heart in both the open and closed states at near-atomic resolutions determined using single-particle electron cryomicroscopy. Structural comparison reveals a breathing motion of the overall cytoplasmic region resulted from the interdomain movements of amino-terminal domains (NTDs), Helical domains, and Handle domains, whereas almost no intradomain shifts are observed in these armadillo repeats–containing domains. Outward rotations of the Central domains, which integrate the conformational changes of the cytoplasmic region, lead to the dilation of the cytoplasmic gate through coupled motions. Our structural and mutational characterizations provide important insights into the gating and disease mechanism of RyRs.

Ryanodine receptors (RyRs) are intracellular ion channels responsible for the rapid release of calcium (Ca2+) ions from the sarcoplasmic/endoplasmic reticulum (SR/ER) into the cytoplasm, a key event in excitation-contraction coupling in skeletal and cardiac muscles (15). Among the three mammalian RyR isoforms, RyR2 is primarily expressed in the heart and brain and is activated by Ca2+ influx through a mechanism known as calcium-induced calcium release (CICR) (610). The RyR2-mediated Ca2+ release is fundamental to a number of cellular processes ranging from muscle contraction to learning and memory (11, 12). Mutations in RyR2 are associated with cardiac disorders, such as catecholaminergic polymorphic ventricular tachycardia type 1 (CPVT1) and idiopathic ventricular fibrillation (IVF) (1317).

Owing to their physiological and pathological importance, RyRs represent important targets for structural investigation. Electron microscopy (EM) has been the primary tool for structural elucidation of RyRs. The architecture and details of RyR1 have been gradually unveiled by EM analysis with increasing resolutions from below 20 to 3.8 Å (1826). The near-atomic resolution structure of RyR1, which was captured in a closed state, revealed a hierarchical organization of the 2.2-megadalton tetrameric channel (26). Nine domains were resolved in the cytoplasmic region in each protomer, including four armadillo repeats–containing domains, the amino terminal domain (NTD), the Handle domain, the Helical domain, and the Central domain, three intertwined SPRY domains (the structural domains identified in SplA and RyRs), and two RyR domains designated P1 and P2 domains (Fig. 1A) (26). The armadillo repeats in the NTD, the Handle domain, and the Central domain together constitute a superhelical assembly in each protomer, which extensively interacts with the other super spiral formed by the Helical domain. In essence, the scaffold of the cytoplasmic domain is established by eight interconnected superhelical assemblies. Classification of the closed RyR1 structures (24, 25) and the potentially open RyR1 structures at low resolutions (25, 27, 28) revealed thermal or “breathing” motions of this cytoplasmic canopy.

Fig. 1 The cryo-EM structure of RyR2 from porcine heart (pRyR2) in a closed state.

(A) Domain organization of a pRyR2 protomer. The domain boundaries are indicated, with the reliably assigned boundaries labeled red. (B) The overall structure of the closed pRyR2 at a nominal resolution of 4.2 Å. The tetrameric structure of pRyR2 is domain-colored with the same scheme as in (A). (C) Comparison of the EM maps (low-pass filtered to 6-Å resolution) of pRyR2 and rRyR1 (rabbit RyR1). The FKBP12 and the carboxyl terminal half of the HD2 domain, shown in magenta, were resolved in rRyR1 but are invisible in pRyR2. (D) Structural comparison of pRyR2 (yellow) with rRyR1. The structures of rRyR1 (PDB code: 3J8H) and pRyR2 are superimposed relative to the channel domain. All structure figures were prepared with PyMol (60), and the EM maps were generated in Chimera (61). Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

In side views, the NTDs, the Central domains, and the channel domain build up a central tower (26). The tetrameric channel domain contains a transmembrane region of the characteristic voltage-gated ion channel superfamily fold and an extended cytoplasmic O-ring in each protomer consisting of the elongated S6 segments (S6Cyt), the zinc finger–containing carboxyl terminal domain (CTD), and a small globular cytoplasmic domain between S2 and S3 in the voltage sensor–like (VSL) domain, which we refer to as the VSLCyt domain hereafter. The Central domain contains a unique U-motif that hooks into the O-ring, whereas the concave face of the armadillo repeats in the Central domain wraps around the O-ring. We proposed that the Central domain may represent the primary transducer that integrates and translates the conformational changes of the cytoplasmic sensors induced by a variety of stimuli to the gating of the central pore (26, 27).

Notwithstanding these exciting advances in the structural and mechanistic understanding of RyR1, a high-resolution structure in the open state is necessitated to reveal the molecular basis for the long-range allosteric gating mechanism. Furthermore, high-resolution structures of RyR2 are also required to better understand the mechanism and to develop novel therapeutics for the treatment of RyR2-associated diseases. Here, we report the cryogenic EM (cryo-EM) structures of porcine RyR2 (pRyR2) in both the open and closed states. Structural comparison and structure-guided mutational analysis provide important insights into the gating and disease mechanism of RyRs.

Structure determination of RyR2

Following the successful strategy for the purification of RyR1 and Cav1.1 complex from rabbit skeletal muscles (26, 29, 30), we used glutathione S-transferase (GST)–fused FKBP12 to purify RyR2 from porcine heart. Please refer to the supplementary materials for details of protein purification and cryosample preparation (fig. S1).

To obtain the structure of RyR2 in the closed state, 5 mM EDTA was included throughout purification. To capture an open RyR2, 20 μM Ca2+ was included throughout the purification (23, 31). It was reported that the compound 2,2′,3,5′,6-pentachlorobiphenyl (PCB 95) stabilized RyR1 in the open state and increased [3H]ryanodine binding to both RyR1 and RyR2 (23, 32, 33); we also added PCB 95. Despite the addition of excess GST-FKBP12 before size exclusion chromatography (SEC), the complex of GST-FKBP12 and RyR2 fell apart during SEC purification, indicating a weaker association of FKBP12 with RyR2 than with RyR1 (fig. S1, A and B).

The structures of RyR2 were determined at overall resolutions of 4.4 Å for the closed state and 4.2 Å for the open state (Fig. 1B, table S1, and figs. S1 and S2). Consistent with their 70% sequence similarity, the overall architecture and domain organization of RyR2 are identical to those of RyR1 (26) (Fig. 1, A and B, and fig. S4). As expected, no density was observed for FKBP12. In addition, a number of armadillo repeats in the C terminus of the Helical domain 2 (HD2) were invisible, likely due to intrinsic flexibility in the absence of FKBP12 (Fig. 1, C and D, and fig. S1, G and H).

Dilation of the S6 bundle leads to channel opening

In the presence of 20 μM Ca2+ and PCB 95, the dilation of the central pore was evident even during two-dimensional (2D) class average compared with the images obtained in the presence of 5 mM EDTA (Fig. 2A). When the pore domains of the two structures are superimposed, the conformation of structural elements on the luminal side, including the selectivity filter (SF), the pore helices, and the supporting helices on S5 and S6 (the turret, residues 4790 to 4840), remains nearly unchanged (Fig. 2B). The deviation occurs at residue Phe4854, beyond which the four S6 segments tilt outward, resulting in the dilation of the intracellular gate by ~8 Å (Fig. 2B and Movie 1). Consequently, the pore constriction site is shifted from Ile4868 in the closed RyR2 to Gln4864 in the open structure. The four Gln4864 residues enclose a gate with a diameter of ~4 Å, allowing passage of Ca2+ ions in single file (Fig. 2, C to E). Therefore, the new structure obtained in the presence of 20 μM Ca2+ and PCB95 (2,2′,3,5′,6-pentachlorobiphenyl) represents an open conformation. Consistent with its critical location in the conduction pathway, mutating Gln4864 markedly or completely abolished channel activation by caffeine (fig. S5).

Fig. 2 Structural comparison of the open and closed RyR2.

(A) Representative 2D class averages of cryo-EM images of the closed and open RyR2. Note the marked difference of the central pore. (B) Conformational changes of the S6 bundle of the channel domain in the open and closed RyR2 structures. The two structures are superimposed relative to the selectivity filter and its supporting turrets (residues 4790 to 4840). In this figure, the SR luminal side is placed on top, opposite to the presentations in other figures. (C) Pore radii of RyR2 in the open and closed states. (Left) The pore radii along the ion-conducting pathway, calculated by HOLE (62), are tabulated. The position of the S4-S5 helical axis is set as the origin of the y axis. Whereas Ile4868 in each protomer constitutes the intracellular gate in the closed RyR2, the constriction site is formed by Gln4864 in the open structure. The corresponding residue numbers in human RyR2 (hRyR2) are shown in brackets. The same applies to other figures. (Right) The channel passage is indicated by yellow (closed) and cyan (open) dots in the center of two diagonal protomers in each channel. (D) The dilation of the intracellular gate in the open RyR2. Shown here is the cytoplasmic view of the intracellular gate. The constriction residues in the closed and open states, Ile4868 and Gln4864, respectively, are shown as spheres and colored yellow in the closed and green in the open structures. (E) In the open pRyR2 structure, Gln4864 residues form the constriction site that allows passage of Ca2+ ion. Shown here are the cytoplasmic views of the pore domain. The surface electrostatic potential is calculated in PyMol. Please refer to fig. S3B for the EM maps of the discussed structural elements and fig. S5 for mutational characterizations of Gln4864.

Movie 1 Conformational changes of RyR2 between the closed and open states.

The morph was generated using the cryo-EM structures of the closed and open pRyR2. The structure is domain-colored as in Fig. 1B. To generate the morph, the two structures were superimposed in PyMol relative to the channel domain. Forty intermediates were then generated using the crystallography and NMR system (63). The morphs were merged and displayed in PyMol using the “mset” command. The same was applied to Movies 2 to 4 and movies S1 to S4. Specific views are shown in each movie. In the side views, the luminal side is placed at the bottom.

Internal stability and interdomain shifts of cytoplasmic domains

Determination of the pRyR2 structure in both the closed and open states offers the opportunity to characterize the structural elements responsible for the unique long-range allosteric gating mechanism of RyRs. When the two structures are superimposed relative to the channel domain, the overall structure undergoes intricate shifts (Fig. 3A, fig. S6, and Movie 1).

Fig. 3 Structural shifts of the cytoplasmic domains between the closed and open RyR2.

(A) Overall conformational changes between the open and closed RyR2. The two structures are superimposed relative to the channel domain. The cyan arrows indicate the overall shifts of the cytoplasmic domains from the closed to the open state. (B) Almost no intradomain rearrangements of the NTD, the Handle domain, and the HD1 between the open and closed RyR2. The domains in the open RyR2 are colored the same as in Fig. 1B, whereas those in the closed structure are colored yellow. (C) The motions of NTD and HD1 relative to the Handle domain. The structures are superimposed relative to the Handle domain. The arrows indicate the motion directions of NTD and HD1. (D) Movement of NTDs within the same protomer and from the neighboring protomer relative to HD1. The structures are superimposed against HD1. Please refer to Movie 1 for the morph illustrating the conformational changes between the open and closed RyR2.

In the cytoplasmic region, the resolution of the peripheral structures, such as the three SPRY domains and P1 and P2 domains, is insufficient for detailed analysis, whereas the secondary structural elements of other domains are well-resolved. The NTD, the Handle domain, and the HD1 domain exhibit nearly no intradomain rearrangements between the closed and open RyR2 structures when compared individually (Fig. 3B). Therefore, the overall motion of the cytoplasmic region may result from domain-wise displacement and relative shifts between domains.

It has been previously analyzed in detail that each interface in the cytoplasmic region generally involves more than two domains (26). Such clustered interactions may be important for the coupled conformational changes among domains. From the closed to open state, the periphery of the cytoplasmic canopy constituted by the Helical domains, the SPRY domains, and the P1 and P2 domains undergoes a combination of downward rocking toward the SR lumen and clockwise rotation when viewed from the cytoplasmic side. On the contrary, the inner circle enclosed by the NTDs appears to rotate upward and counterclockwise (Fig. 3C; fig. S6, A and B; and Movie 1). The opposing motion directions between the outer and inner rings are in a sense achieved by twisting the two ends of the helical assembly formed by the NTD, Handle domain, and Helical domain in each protomer (Fig. 3C and Movie 1).

One HD1 domain simultaneously interacts with two NTDs from neighboring protomers (Fig. 3D). The four NTDs and four HD1 domains also complete a closed ring. In the cytoplasmic view, the combined relative motions of NTDs and HD1s result in contraction and expansion of the ring accompanying pore closure and opening, respectively (Fig. 3D; fig. S6, A and B; and Movie 1). Despite the pronounced overall structural shifts, there are only minor relative motions between the NTD, Handle domain, and HD1 when the domains are compared pairwise (Fig. 3, C and D). The major structural rearrangements occur at the interface between these domains and the Central domain.

Twisting of the Central domain

When the structures of the Central domain in the closed and open RyR2 are superimposed, the backbone of the armadillo repeats remains nearly unchanged. The auxiliary motifs, including the preceding helix α0, a capping helix α4 that is located on top of the concave face of the armadillo repeats, and the U-motif on the C-terminal end, slightly fold toward the center of the concave surface (Fig. 4A).

Fig. 4 Intra- and interdomain shifts of the Central domain.

All the structural comparisons in this figure are made relative to the Central domain. (A) Intradomain shifts of the Central domain. The backbone of the horseshoe-shaped armadillo repeats within the Central domain remains nearly unchanged, whereas the edges exemplified by the U-motif and α4 helix fold toward the concave side. (B) The conformational changes of NTD and the Handle domain relative to the Central domain. (C) The shifts of the Handle and HD1 domains relative to the Central domain. (D) The motions of the NTDs from two adjacent protomers and the Handle domain within the same protomer relative to the Central domain. Please refer to Movie 2 and movies S1 to S3 for the morphs illustrating the conformational changes and relative displacement of domains described in this figure when the structures are compared relative to the channel domain.

Whereas the concave surface of the Central domain wraps up the cytoplasmic O-ring of the channel domain, the two edges of the convex side contact two neighboring NTDs and one Helical domain, respectively. In addition, despite the distance in the primary sequence, the N-terminal helical repeats of the Central domain appear to be the continuation of the C-terminal repeats of the Handle domain in the 3D structure, and the flanking α0 helix is sandwiched by two helices in the Handle domain.

When the structures are superimposed relative to the Central domain, the consecutive helical repeats of the NTD and the Handle domain appear to rock around the interface between the Handle domain and the Central domain. Consequently, the NTD, which is on the distal end, undergoes a larger degree rotation (Fig. 4B and movie S1). Importantly, there are only insignificant changes of the interface between HD1 and Central domain, suggesting coupled motions of these domains during channel opening (Fig. 4C and movie S2). This observation implicates that structural shifts of the Helical domain may directly induce the concordant movement of the Central domain.

In the context of the RyR2 tetramer, the Central domain appears to slightly swirl around an axis encompassed by its horseshoe-shaped armadillo repeats. The clockwise screw of the Central domain in the cytoplasmic view, which is concordant with the clockwise rotation of the Helical domain, synchronizes with the expansion of the central tower consisting of the channel domain, the Central domain, and the NTD during channel opening (Movies 1 and 2). Meanwhile, the interface between the Central domain and the two neighboring NTDs undergoes marked rearrangements (Fig. 4D and movie S3).

Movie 2 Conformational changes of the “central tower” between the open and closed RyR2.

The rotations of the Central domains are evident in the luminal view.

Movie 3 Conformational changes of the channel domain and the Central domain during pore opening.

For visual clarity, only two diagonal protomers are shown.

Movie 4 Potential rearrangement of interactions between the S6Cyt residues from adjacent protomers.

The resolutions do not allow accurate assignment of the side chains of negatively charged residues. These side groups are modeled to facilitate discussion of the potential rearrangement of the interactions during pore opening. It remains to be elucidated whether Ca2+ ions play a role in the rearrangement of the interactions among the charged and polar residues in the S6Cyt segment.

Coupled changes between the Central domain and channel domain

The swirl of each Central domain results in the outward motion of the S6Cyt segments through intricate interactions between the channel domain and the Central domain (Fig. 5A and Movie 3). The U-motif of the Central domain hooks into the cytoplasmic O-ring of the channel domain. Consistent with their extensive interactions, there is no relative motion between the U-motif and the CTD. The displacement of the U-motif thus leads to coupled outward shifts of the CTD and its preceding S6Cyt (Fig. 5B and Movie 3). Accompanying this dilation, the central pore axis is shortened by approximately 3 Å in the open RyR2 (Fig. 5A).

Fig. 5 The carboxyl terminal domain (CTD) couples the conformational changes of the Central domain and channel domain of RyR2.

(A) The rotations of individual Central domains pull the S6Cyt segments outward through the interactions between the U-motif and CTD. (B) The helical hairpin in the U-motif of the Central domain forms a stable structural unit with the CTD. The structures are superimposed relative to CTD. (C) The VSL domain remains almost rigid during pore opening. The superimposition of the VSL domain of the open and closed structures shows little intradomain rearrangement, whereas the S4-S5 linker undergoes lateral shifts that relax the constriction to the S6 bundle at the cytoplasmic gate in the closed RyR. (D) The outward expansion of the CTDs and U-motifs accompanies rearrangements of the interactions between adjacent CTDs and between CTD and the U-motif from a neighboring protomer. (E) The integrity of the zinc finger in CTD is essential for RyR2 function. Mutations C4888H, C4891H, and H4908C completely abolish the activation of RyR2 by caffeine, whereas the H4913C mutant remains sensitive to caffeine activation (residue numbering of human and mouse RyR2). See fig. S7 for details of the mutational analysis and Movie 3 and movie S4 for the conformational changes discussed in this figure when the structures are compared relative to the channel domain.

There is a minor relative movement between VSLCyt and CTD, whereas the backbone of VSL remains rigid during channel opening (Fig. 5, B and C). In addition to the contacts between VSLCyt and the U-motif, the cytoplasmic tip of VSL may contact the EF-hand motif of the neighboring Central domain in the closed structure. This interaction no longer exists in the open structure (movie S4). In the closed RyR1 and RyR2 structures, the S4-S5 linker constrains the cytoplasmic gate formed by the S6 tetrahelical bundle at the boundary between the SR membrane and cytoplasm, reinforcing the closure of the channel (26). In the open RyR2, the S4-S5 linker moves closer to the VSL domain, resulting in an expansion of the S4-S5 ring that releases the constriction to the cytoplasmic gate (Fig. 5, B and C). We predicted that interactions among the four CTDs may facilitate closure of the S6 bundle (26). Indeed, the adjacent CTDs no longer interact with each other in the open RyR2, whereas the U-motif from the neighboring Central domain still contacts the CTD (Fig. 5D and movie S4).

The concerted movements of U-motif, CTD, and S6Cyt segment requires the rigidity of CTD, which is provided by the zinc finger motif at the joint between CTD and S6 segment (26). Supporting this notion, single point mutations of the zinc-coordinating residues, C4888H, C4891H, and H4908C, all abolished the response of RyR2 to caffeine activation (Fig. 5E and fig. S7). Only H4913C retains the sensitivity to caffeine as wild type (WT), suggesting that the 4913 position is relatively more adaptable than other positions in coordinating Zn binding in the CTD (the corresponding residue numbers in human RyR2 are labeled and referred to in all functional assay-related figures and texts).

Discussion

The recently reported 3.8-Å-resolution cryo-EM structure of RyR1-FKBP12 complex in the closed state revealed domain organization and atomic details of the largest ion channels (26). Comparison of the near-atomic-resolution structures of the open and closed RyR2 presented here identifies the Central domain as the primary transducer of conformational changes that controls the opening and closure of the channel domain, confirming our previous conjecture (26, 27). The Central domain is surrounded by a second tier of transmitters, including NTDs and the Handle and Helical domains, which provide a scaffold for interactions with a multitude of peripheral domains and channel modulators. The Central domains undergo both intradomain contraction and domain-wise rotation that collectively result in the outward tilting of the pore-forming S6 segments through extensive interactions between the Central domain and channel domain, primarily through the U-motif and the CTD. Consistent with the pivotal role of the Central domain in channel gating, a single point mutation E3987A in mouse RyR2 or E3885A in rabbit RyR3, located within the Central domain, reduced the sensitivity of single RyR channels to cytosolic Ca2+ activation by a factor of 1000 to 10,000 (34, 35). However, deleting the entire EF-hand motif in the Central domain does not affect cytosolic Ca2+ activation of single RyR2 channels, suggesting that the EF-hand is unlikely to be the cytosolic Ca2+ sensor (36). Detailed understanding of the role of the Central domain in cytosolic Ca2+ activation of RyR2 will have to await atomic-resolution structures of RyRs, which may unveil the binding sites for Ca2+.

One unexpected feature is the little intradomain shifts of the armadillo repeats–containing domains. The VSL domain also remains rigid during channel opening. Furthermore, the U-motif of the Central domain and the CTD form a stable structural unit that has literally no internal rearrangement during pore opening. In a sense, these rigid domains represent individual modules, whereas the Central domain is a relatively flexible joint. The overall conformational changes are achieved through the relative motions between these modules. This analysis predicts that some ligands and modulators may target interfaces of these modules to induce conformational changes, a scenario that awaits further structural and biochemical characterizations.

Mutations in human RyR2 are associated with a range of debilitating diseases such as CPVT1, IVF, and sudden cardiac death (15, 17). More than 100 disease-associated missense mutations and deletions can be mapped onto all major domains in the structure of RyR2 (fig. S8A). Mutations that are located on the surface may disrupt interactions between RyR and its modulators (fig. S8A). The HD1, the Central domain, and the channel domain represent major disease hot spots in addition to NTD (5, 26, 37) (fig. S8, A and B). Supporting the importance of interactions among domains, one type of mutations and deletions clusters at domain interfaces. These mutations may alter the ability of RyRs to sense chemicals/modulators or to propagate conformational changes that are required for ion channel regulation and gating (fig. S8, C to F).

On the other hand, the coupled motion of the Helical domain and Central domain suggests that shifts of the Helical domain may be directly translated to the structural changes of the Central domain (movie S2). As the Helical domains are positioned on the edge of the cytoplasmic canopy and contact multiple peripheral domains (Fig. 1B), conformational changes of the Helical domains can be achieved through mechanical forces exerted by the upstream Cav1.1 complex or the neighboring RyR molecules, as in the case of RyR1 (3843). This analysis provides a clue to understanding the molecular basis for excitation-contraction coupling in skeletal muscles.

The S6Cyt segment contains a large number of negatively charged or polar residues (Glu, Asp, and Gln) and one positively charged residue (R4875) (Fig. 6A). Mutational studies suggested that mouse RyR2-R4874 may form intersubunit salt bridges with D4868 and E4872 (Fig. 6B) (44). Mutating these key residues dramatically diminished the activation of RyR2 by caffeine (Fig. 6C and fig. S9A). It has recently been shown that the E4872A mutation selectively and completely abolishes luminal, but not cytosolic, Ca2+ activation of mouse RyR2. Further, introducing metal binding His residues at this site converts RyR2 from a luminal Ca2+-gated channel to a luminal Ni2+-gated channel (44). These data strongly suggest that a S6Cyt region encompassing the E4872 residue contains a cation binding pocket that may contribute to RyR2 activation by luminal Ca2+. However, the composition of this putative cation binding pocket is unknown. Despite the fact that the densities for the negatively charged residues are invisible in the cryo-EM structures, those of the backbones and of the side chain of R4875 in pRyR2 (corresponding to R4874 in human and mouse RyR2) are discernible. Structural comparison of the closed and open states reveals that residue R4874 moves away from E4872, while E4878 may move close to E4872 in the open state (these residue numbers corresponding to those in mouse and human RyR2) (Fig. 6B). This suggests that, in addition to E4872, E4878 may also be involved in luminal Ca2+ activation of RyR2. Indeed, mutating mouse RyR2-E4878 markedly reduced luminal Ca2+ activation of RyR2 (Fig. 6D and fig. S9B). Atomic resolution structures of RyR2 in the presence of high concentrations of Ca2+ will be required to understand the structural basis of luminal Ca2+ activation of RyR2.

Fig. 6 The putative luminal Ca2+ activation sites.

(A) The S6Cyt segments are enriched in charged and polar residues. The densities for the side groups of the negatively charged residues are largely invisible due to radiation damage. They are modeled in this figure to facilitate discussion. (B) Potential rearrangement of interactions between two neighboring S6Cyt helices in the open and closed pRyR2 structures. The putative hydrogen bonds are represented by red dotted lines. (C) Ca2+ release–cumulative caffeine concentration relationships. The amplitude of caffeine-induced Ca2+ release at each caffeine concentration was normalized to that of the maximum release for each experiment. Data shown are mean ± SEM (n = 3 separate experiments). (D) Single point mutation E4878A (corresponding to E4879 in pRyR2) leads to decreased sensitivity to luminal Ca2+ activation. The relationships between the open probability (Po) and luminal Ca2+ concentrations (pCa) of single RyR2-WT (solid circles) and E4878A mutant channels (open circles) are shown. Data points shown are mean ± SEM from 5 to 7 channels.

A number of questions remain to be investigated. The structure of the open pRyR2 was obtained in the presence of Ca2+ and PCB 95. However, no densities were identified for these ligands in the reconstruction. The bottleneck in the improvement of resolution in part resides in the flexibility of the protein, particularly in the absence of FKBP12 or FKBP12.6. Comparison of RyR1 and RyR2 suggests highly conserved sequences of the structural elements involved in interactions with FKBP12 in the NTD, SPRY1, and Handle domains (fig. S10) (26, 4547). Therefore, the basis underlying the markedly different affinities of FKBP12 binding to RyR1 and RyR2 remains to be determined (48, 49).

Materials and methods

Expression and purification of GST-FKBP12 and GST-FKBP12.6

Both FKBP12 and FKBP12.6 can bind to RyR2, but the two isoforms show distinct binding affinities, particularly to the dog RyR2 (48). As the sequence of porcine FKBP12 or FKBP12.6 is not available in public domain, we used human FKBP12 and FKBP12.6 for affinity purification of pRyR2 based on the previous protocol for purification of rabbit RyR1 (rRyR1) (26). The cDNAs encoding human FKBP12/12.6 were cloned into pGEX-4T-2 vector with a C-terminal 6×His tag. The plasmids were transformed into BL21 (DE3) strain for protein overexpression. When OD600 reached 1.0, the temperature was adjusted to 16°C, and 0.2 mM IPTG was added for induction overnight. Cells collected by centrifugation were resuspended in lysis buffer solution (25 mM Tris-HCl, pH 8.0, and 150 mM NaCl). After removal of cell debris by centrifugation at 22,000 g for 1 hour, the supernatant was loaded to Ni2+-NTA resin (Qiagen). Sequentially washed by the W1 buffer containing 25 mM Tris-HCl, pH 8.0 and 300 mM NaCl, and W2 buffer containing 25 mM Tris-HCl, pH 8.0 and 20 mM imidazole, the target protein was eluted by the buffer containing 25 mM Tris-HCl, pH 8.0 and 300 mM imidazole. Anion exchange chromatography (SOURCE 15Q, GE Healthcare) was performed for further purification. As GST-FKBP12 appeared to have a better solution behavior, we used the GST-FKBP12-His protein as bait to pull-down pRyR2.

Preparation of sarcoplasmic reticulum membrane from porcine heart

A single porcine heart was diced into small pieces, and resuspended in five volumes of homogenization buffer containing 25 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, and the protease inhibitor cocktail (0.2 mM PMSF, 1.3 μg/ml aprotinin, 0.7 μg/ml pepstatin, and 5 μg/ml leupeptin). Fifteen cycles of homogenization were performed with a blender (JYL-C010, Joyoung). The debris was removed by low-speed centrifugation (6000 g) for 6 min. The supernatant was then centrifuged at high speed (20,000 g) for 1 hour. The pellet was resuspended in two volumes of homogenization solution and flash frozen in liquid nitrogen.

Purification of pRyR2 by GST-FKBP12

The pRyR2 was purified following the same strategy for the purification of rabbit RyR1 (26) and the Cav1.1 complex (29). The porcine sarcoplasmic reticulum membrane (1/4 of total membrane from a single heart) was solubilized at 4°C for 2 hours in the homogenization buffer (mentioned above) plus 4% CHAPS, 1% soybean lecithin, and 2 mM DTT. Approximately 3-4 mg of GST-FKBP12 was added to the extraction system. After ultrahigh speed centrifugation (200,000 g), the supernatant was applied to GS4B (GE Healthcare) column. To obtain pRyR2 in closed state, the resin was washed with buffer similar to the homogenization buffer, except that the NaCl concentration was changed to 1 M, and 2 mM DTT and 0.1% Digitonin were supplemented. The complex was eluted by 75 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10 mM GSH, 0.1% Digitonin, 2 mM DTT, 5 mM EDTA, and protease inhibitors. The eluted protein was concentrated before size exclusion chromatography (SEC, Superose 6 10/300 GL, GE Healthcare) in the buffer containing 25 mM Tris, pH 7.5, 300 mM NaCl, 0.1% Digitonin, 2 mM DTT, 5 mM EDTA, and protease inhibitors. The pRyR2 fractions were concentrated to approximately 0.1 mg/ml for EM sample preparation.

To capture pRyR2 in an open state, a similar protocol was used, except that EDTA was omitted. Instead, 20 μM CaCl2 and 20 μM PCB 95 were supplemented to GS4B eluent. After SEC, additional 10 μM PCB 95 was added before concentrating the pRyR2 fractions.

Cryo-EM image collection

Cryo-EM grids were prepared using Vitrobot Mark IV (FEI). Aliquots (3 μl each) of pRyR2 protein were placed on glow-discharged copper grids covered with continuous carbon film (Zhongjingkeyi Technology Co. Ltd.). Grids were immediately blotted for 1.5 s and flash-frozen in liquid ethane. Cryo-EM images were acquired with a Titan Krios (FEI) operated at 300 kV. FEI Falcon-II direct electron detector was mounted and the pixel size was 1.05 Å at a nominal magnification of 75,000. Defocus value varied from –1.5 to –3.0 μm. Each image was dose-fractionated to 19 frames with a total electron dose of 44 e2 for a total exposure time of 1.2 s. EPU package (FEI) and AutoEMation package (50) upgraded recently were used for automated data collection.

Image processing

Frames of each image were aligned and summed using whole-image motion correction program MotionCorr (51). CTFFIND3 was adopted to calculate the contrast transfer function parameters (52). Particles were auto-picked by RELION1.4 and rechecked manually (53). Two- and three-dimensional classifications and refinements were all performed using RELION1.4 (53). In total, 213,595 particles from 4,439 micrographs were selected for the closed pRyR2. After 2D and 3D classification with C1 symmetry, 91,076 particles were selected for refinement with RyR1 (EMD accession number 2807) (26) as initial model. Combined with local angular search, a second round of 3D classification with C4 symmetry was performed using generated pRyR2 model from the prior refinement. 48,454 particles were finally chosen for refinement, which gave rise to an electron density map of 4.4 Å according to the gold-standard Fourier shell correlation (FSC) 0.143 criterion. A local mask of channel domain improved the resolution to 4.2 Å. Local resolution was estimated using ResMap (54).

With regard to the reconstruction of open pRyR2, 589,270 particles were picked from 9662 micrographs. Subsequent procedures were similar to those described above, and 133,196 particles were used for final refinement. The calculated resolution is 4.5 Å and 4.3 Å (with a local mask of channel domain). Particle polishing was performed for particle based beam-induced motion correction and radiation-damage weighing. Subsequent refinement using the polished particles further improved the resolution of 3D reconstruction to 4.2 Å and 4.1 Å (with a local mask of channel domain). Please refer to fig. S2 for the workflow of image processing.

Model building and structure refinement

Sequence alignment was done at www.ebi.ac.uk/Tools/msa/clustalo. The output file was then further processed at http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi (55). The structure of RyR1 (PDB code 3J8H) (26) was used for homologous model building of the closed pRyR2 and manually adjusted in COOT (56). The open pRyR2 was built based on the structure of closed pRyR2. Structure refinement was performed using PHENIX (57) in real space with secondary structure and geometry restrained.

Caffeine-induced Ca2+ release in HEK293 cells

Mouse RyR2 cDNA was used for transfection of HEK293 cells and electrophysiological studies of RyR2 (the corresponding residue numbers in human RyR2 are labeled instead in all functional assay figures for clarity). Cytosolic Ca2+ concentration in transfected HEK293 cells was measured using the fluorescence Ca2+ indicator dye Fluo-3 AM (Molecular Probes). HEK293 cells grown for 18 to 20 hours after subcultures were transfected with 12 μg of WT or mutant RyR2 cDNAs. Cells grown for 18 to 20 hours after transfection were washed four times with PBS and incubated in KRH (Krebs-Ringer-Hepes, 125 mM NaCl, 5 mM KCl, 1.2 mM KH2PO4, 6 mM glucose, 1.2 mM MgCl2, 2 mM CaCl2 and 25 mM HEPES, pH 7.4) buffer without MgCl2 and CaCl2 at room temperature for 40 min and at 37°C for 40 min. After being detached from culture dishes by pipetting, cells were collected by centrifugation at 1000 rpm for 2 min in a Beckman TH-4 rotor. Cell pellets were loaded with 10 μM Fluo-3 AM in high-glucose Dulbecco’s Modified Eagle Medium at room temperature for 60 min, followed by washing with KRH buffer plus 2 mM CaCl2 and 1.2 mM MgCl2 (KRH+ buffer) three times and resuspended in 150 μl KRH+ buffer plus 0.1 mg/ml BSA and 250 μM sulfinpyrazone. The Fluo-3 AM loaded cells were added to 2 ml (final volume) KRH+ buffer in a cuvette. The fluorescence intensity of Fluo-3 AM at 530 nm was measured before and after repeated additions of various concentrations of caffeine (0.025 to 5 mM) in an SLM-Aminco series 2 luminescence spectrometer (SLM Instruments) with 480 nm excitation at 25°C. The peak levels of each caffeine-induced Ca2+ release were determined and normalized to the highest level (100%) of caffeine-induced Ca2+ release for each experiment.

Single channel recordings in planar lipid bilayers

Recombinant RyR2 WT and E4878A mutant channels were purified from cell lysate prepared from HEK293 cells transfected with the RyR2 WT or the E4878A mutant cDNA by sucrose density gradient centrifugation, as described previously (35, 58). Heart phosphatidylethanolamine (50%) and brain phosphatidylserine (50%) (Avanti Polar Lipids), dissolved in chloroform, were combined and dried under nitrogen gas and resuspended in 30 μl of n-decane at a concentration of 12 mg lipid per ml. Bilayers were formed across a 250-μm hole in a Delrin partition separating two chambers. The trans chamber (800 μl) was connected to the head stage input of an Axopatch 200A amplifier (Axon Instruments, Austin, TX). The cis chamber (1.2 ml) was held at virtual ground. A symmetrical solution containing 250 mM KCl and 25 mM Hepes (pH 7.4) was used for all recordings, unless indicated otherwise. A 4-μl aliquot (~1 μg protein) of the sucrose density gradient-purified recombinant RyR2 WT or E4878A mutant channels was added to the cis chamber. Spontaneous channel activity was always tested for sensitivity to EGTA and Ca2+. The chamber to which the addition of EGTA inhibited the activity of the incorporated channel presumably corresponds to the cytosolic side of the Ca2+ release channel. The direction of single channel currents was always measured from the luminal to the cytosolic side of the channel, unless mentioned otherwise. Recordings were filtered at 2500 Hz. Data analyses were carried out using the pclamp 8.1 software package (Axon Instruments). Free Ca2+ concentrations were calculated using the computer program Fabiato and Fabiato (59).

Supplementary Materials

www.sciencemag.org/content/354/6310/aah5324/suppl/DC1

Figs. S1 to S10

Table S1

Movies S1 to S4

References

References and Notes

  1. Acknowledgments: We thank J. Lei and X. Li for technical support. We thank the Tsinghua University Branch of the China National Center for Protein Sciences (Beijing) for providing facility support. The computation was completed on the “Explorer 100” cluster system of Tsinghua National Laboratory for Information Science and Technology. This work was supported by funds from the Ministry of Science and Technology of China (2015CB9101012014, 2016YFA0500402, and ZX09507003006), the National Natural Science Foundation of China (project 31321062), the Canadian Institutes of Health Research (MOP-123506 to S.R.W.C.), and the Heart and Stroke Foundation of Canada (G-16-00014214 to S.R.W.C.). W.G. was supported by the Alberta Innovates–Health Solutions Graduate Studentship Award. S.R.W.C. is an Alberta Innovates–Health Solutions Scientist and was supported in part by the Heart and Stroke Foundation/Libin Cardiovascular Institute Professorship in Cardiovascular Research. The research of N.Y. was supported in part by an International Early Career Scientist grant from the Howard Hughes Medical Institute and an endowed professorship from Bayer Healthcare. The atomic coordinates of pRyR2 in the closed and open states have been deposited in the Protein Data Bank with the accession codes 5GO9 and 5GOA, respectively. The corresponding maps have been deposited in the Electron Microscopy Data Bank with the accession codes EMD-9528 and EMD-9529, respectively.
View Abstract

Navigate This Article