Research ArticlesBiochemistry

A cryo-EM–based model of phosphorylation- and FKBP12.6-mediated allosterism of the cardiac ryanodine receptor

See allHide authors and affiliations

Sci. Signal.  23 May 2017:
Vol. 10, Issue 480, eaai8842
DOI: 10.1126/scisignal.aai8842

More flexible with phosphorylation?

The type 2 ryanodine receptor (RyR2) mediates Ca2+ release from the sarcoplasmic reticulum of cardiomyocytes to initiate cardiac muscle contraction. Mutations in this intracellular Ca2+ channel are associated with cardiac diseases that may lead to heart failure. Dhindwal et al. used cryo-EM to determine the structure of rabbit RyR2 in complex with the regulatory protein FKBP12.6 in the closed state at 11.8 Å resolution. They found two conformations of RyR2, which may correspond to the extent of phosphorylation of a domain that harbors several disease-associated mutations. Because the more flexible conformation may correspond to phosphorylated RyR2, the authors suggest that phosphorylation may reduce the energy required for the Ca2+ channel to transition to an open state. These results provide a structural basis for understanding how phosphorylation may affect the activation of RyR2.

Abstract

Type 2 ryanodine receptors (RyR2s) are calcium channels that play a vital role in triggering cardiac muscle contraction by releasing calcium from the sarcoplasmic reticulum into the cytoplasm. Several cardiomyopathies are associated with the abnormal functioning of RyR2. We determined the three-dimensional structure of rabbit RyR2 in complex with the regulatory protein FKBP12.6 in the closed state at 11.8 Å resolution using cryo-electron microscopy and built an atomic model of RyR2. The heterogeneity in the data set revealed two RyR2 conformations that we proposed to be related to the extent of phosphorylation of the P2 domain. Because the more flexible conformation may correspond to RyR2 with a phosphorylated P2 domain, we suggest that phosphorylation may set RyR2 in a conformation that needs less energy to transition to the open state. Comparison of RyR2 from cardiac muscle and RyR1 from skeletal muscle showed substantial structural differences between the two, especially in the helical domain 2 (HD2) structure forming the Clamp domain, which participates in quaternary interactions with the dihydropyridine receptor and neighboring RyRs in RyR1 but not in RyR2. Rigidity of the HD2 domain of RyR2 was enhanced by binding of FKBP12.6, a ligand that stabilizes RyR2 in the closed state. These results help to decipher the molecular basis of the different mechanisms of activation and oligomerization of the RyR isoforms and could be extended to RyR complexes in other tissues.

INTRODUCTION

Each heartbeat is intricately controlled by intracellular Ca2+ released by a structure called a couplon (1) that consists of a voltage-gated Ca2+ channel (CaV1.2), which is also a dihydropyridine receptor (DHPR), a Ca2+ release channel known as ryanodine receptor isoform 2 (RyR2), and other junctional sarcoplasmic reticulum (SR) proteins (2). The DHPRs are embedded in the transverse tubules and peripheral junctions and lie in close proximity to the RyR2s in the membrane of the SR, the intracellular Ca2+ store (3). At each contraction, the depolarization phase of the cardiac action potential initiates the Ca2+ cycle with the entry of a small amount of extracellular Ca2+ through the DHPRs. The increased intracellular Ca2+ activates the RyR2 Ca2+ release channels, triggering a large-scale efflux of Ca2+ from the SR to the cytoplasm in a process known as calcium-induced calcium release (CICR) (4). This increase in cytoplasmic Ca2+ leads to the contraction of the myocardium or systole. The process by which the initial depolarization results in muscle contraction is called excitation-contraction coupling. The reuptake of Ca2+ back to the SR by Ca2+ ATPases (adenosine triphosphatases) causes muscle relaxation or diastole (5). The plasmalemmal Na+/Ca2+ exchanger corrects for the DHPR-mediated Ca2+ entry to prevent Ca2+ overload (6).

RyR2 channels are large homotetrameric proteins consisting of protomers each with a molecular mass of 565 kDa. RyR2s are a critical component of cardiac muscle function, and dysfunction of these calcium channels has been implicated in several cardiac pathological conditions. Anomalous RyR2 function can be caused by mutations in the RyR2 gene, causing catecholaminergic polymorphic ventricular tachycardia (CPVT) and arrhythmogenic right ventricular dysplasia type 2 (ARVD2), abnormal posttranslational modifications of RyR2 (such as hyperphosphorylation and irreversible oxidation), and possibly defective interaction with other regulatory proteins (7). Possibly in connection with the presence of RyR2 in the brain, patients with CPVT-associated mutations in RyR2 can also have seizures (8, 9). RyR2 is modulated directly or indirectly by various proteins such as the DHPR, FKBP12.6, calmodulin, calsequestrin, triadin, junctin, phosphatases, various kinases, and the small molecules Ca2+, Mg2+, and ATP (adenosine 5′-triphosphate) (10). Structural characterization is essential to understand the molecular mechanism that underlies the functioning of this complex Ca2+ channel.

The RyR isoform from skeletal muscle (RyR1) has been determined at high resolution (1113). Despite the high sequence similarity (80%) between both isoforms, RyR1 and RyR2 differ in their excitation-contraction coupling mechanism. In skeletal muscle, DHPRs physically interact with RyR1 to transduce the excitation-contraction coupling signal. In contrast, RyR2 and the cardiac DHPR do not interact, and thus, Ca2+ release from the SR is regulated through CICR (14). Structural comparison between RyR1 and RyR2 is important for inferring the molecular basis that underlies this and other functional differences between these two isoforms. This is critical not only in the context of muscle contraction but also in the wider context of intracellular Ca2+ signaling in other organs such as the brain, in which all RyR isoforms are present (RyR1, RyR2, and RyR3).

Here, we determined the three-dimensional (3D) structure of RyR2 from rabbit (Oryctolagus cuniculus) in complex with FKBP12.6 at 11.8 Å resolution by cryo-EM. At this resolution, we successfully built an atomic model and described in detail the domain-based architecture of RyR2, which is substantially different from that of RyR1. We also compared our RyR2 structure obtained from rabbit with the published high-resolution structure of RyR2 from pig (Sus scrofa) (15). Despite the high similarity between both RyR2 structures, they showed marked differences in one of its domains, which could be attributed to the presence of FKBP12.6 in our case and suggested a molecular mechanism for this important regulator of RyR2.

RESULTS

Purification and electron microscopic characterization of RyR2

For this study, rabbit RyR2 (which is 99% identical to human RyR2) from heart ventricle was obtained in complex with FKBP12.6. Our purification protocol yielded biochemically pure RyR2-FKBP12.6 (Fig. 1A) and structurally intact RyRs, as can be judged by negative stain and cryo-EM (Fig. 1, B and C), which showed RyRs with the typical fourfold morphology and central cross. From such cryo-EM images, we determined the 3D structure of RyR2 by single-particle analysis at a resolution of 11.8 Å at the 0.5 cutoff criterion and 9.1 Å at the 0.143 cutoff criterion (Fig. 1D). Each functional unit of RyR2 consisted of four identical subunits that tetramerized around the fourfold axis to form a central ion channel. Overall, RyR2 had a mushroom-shaped structure in which the cytoplasmic region resembled the cap and the transmembrane region constituted the stem (Fig. 2, A and B). The tetrameric RyR2 channel was 175 Å in height, and the cytoplasmic region had a scaffold-like structure forming a square prism with side and diagonal distances of 273 and 386 Å, respectively (Fig. 2, C and D). A square central rim of around 41 × 41 Å in size delimited a central depression. Multiple, well-defined domains could be distinguished within the cytoplasmic domain.

Fig. 1 Biochemical and electron microscopic characterization of RyR2.

(A) SDS–polyacrylamide gel electrophoresis (PAGE) of the purified RyR2-FKBP12.6 complex in lane 2. Molecular weight markers are in lane 1. (B) Representative image of the RyR2-FKBP12.6 complex obtained by negative stain EM. Scale bar, 100 nm. (C) Representative image of the RyR2-FKBP12.6 complex (shown in circles) embedded in ice obtained by cryo-EM. Scale bar, 100 nm. (D) Fourier shell correlation curve for the RyR2-FKBP12.6 structure indicating a resolution of 11.8 Å at the 0.5 cutoff criterion and 9.1 Å at the 0.143 cutoff criterion.

Fig. 2 Electron density map of the RyR2-FKBP12.6 complex shown in different views.

(A) Side view of the RyR2-FKBP12.6 complex. Different domains present in one subunit of RyR2 as well as the FKBP12.6 are colored and labeled. (B) Side view of the RyR2-FKBP12.6 complex cut centrally along the fourfold axis to show the architecture of the cytoplasmic domains, the central domain, and the transmembrane region. (C) Cytoplasmic view of the RyR2-FKBP12.6 complex (perpendicular to the fourfold axis). The different domains present in one subunit of RyR2 are colored and labeled. (D) RyR2-FKBP12.6 structure viewed from the lumen. Scale bar, 50 Å.

The transmembrane region was not as resolved as expected from this level of resolution. The use of submicromolar Ca2+ conditions in the buffer (2 mM EGTA; no Ca2+ added) meant that the 3D structure should correspond to the closed state of RyR2. Data set heterogeneity could blur structural details by mingling different conformations. Accordingly, further analysis and classification of the whole data set resulted in two classes revealing two conformations named C1 (or “upright”) and C2 (or “flexible”) that exhibited subtle but substantial differences.

Building an atomic model for the RyR2 structure

At the current resolution of RyR2, the boundaries of all 11 cytoplasmic domains per subunit [according to the numerical domain nomenclature for the RyRs (16)] were well resolved and had a clear equivalence with RyR1 domains at the same resolution. The 3D map was divided into its domain components and using the cryo-EM–based atomic structure of RyR1 and isolated fragments of RyR1 and RyR2 obtained by x-ray crystallography, we constructed an atomic model for RyR2 with FKBP12.6 bound (Fig. 3, A to C, and fig. S1A). This model was built independently of another published atomic model for RyR2 (15), as evidenced by the earlier deposition date of our model. The source of the starting models and the corresponding references are indicated in Table 1 for each domain. The model was built on the basis of the upright conformation that had better definition in several regions of the map. The different domains were docked using the six-dimension docking algorithm implemented in the Situs program (17) using the Laplacian filter option, which is more appropriate when a substantial amount of the domain surface is exposed to the solvent. One molecule of FKBP12.6 bound to each subunit was identified and modeled as well (Fig. 3A and fig. S1A). The docking of the atomic models in the EM density was adequate, judging by the close fitting of their respective outlines (Fig. 3, B and C). The unnormalized correlation coefficient of the top-scoring docking solution of these individual domains was within the 0.8 to 0.9 range (Table 1), as expected for appropriately docked densities. We modeled the less-resolved transmembrane region based on the structure of RyR1 and docked it into the transmembrane region of RyR2. Fitting of the entire (fourfold symmetric) atomic model into the RyR2 structure yielded an unnormalized correlation coefficient of 0.89 using the Situs program. Refinement of the final model with Refmac5 gave an average Fourier shell correlation of 0.93. Comparison of our atomic model of closed-state RyR2-FKBP12.6 with a published atomic model of closed-state RyR2 (15) indicated an overall root mean square deviation (RMSD) of 2.49 Å for a RyR2 monomer and a smaller RMSD when comparing each domain independently, ranging between 0.95 and 1.88 Å (Table 2).

Fig. 3 Atomic model of RyR2 and FKBP12.6 docked to the cryo-EM density map.

(A) Top view or the cytoplasmic scaffold of the RyR2 structure. The structure of the RyR2 conformation C1 is shown in semitransparency with the atomic model docked. The domains are color-coded as in Fig. 2 for all four subunits. Domains comprising one corner of RyR2 are labeled. (B) Same top view of the RyR2 structure is cut perpendicular to the fourfold axis, exposing the domains in the cytoplasmic scaffold. Domains comprising one corner of RyR2 structure are highlighted by a dotted square. (C) Side view of the RyR2 structure cut through the central plane. The modeled domains show a close fitting with the electron density, especially in the cytoplasmic scaffold. Positioning of the SPRY domains 1, 2, and 3 in the form of a triangle is highlighted by dotted lines.

Table 1 Domain sequence range, source of the structures used to build the atomic model of RyR2-FKBP12.6, and cross-correlation between the model and the cryo-EM density for each domain.
View this table:
Table 2 Comparison of the coordinates of our atomic map of RyR2-FKBP12.6 with those in a published high-resolution map of RyR2.
View this table:

Correspondence between structural domains and RyR2 sequence

Different domains within RyR2 were easily identifiable in the isosurface representation (Fig. 3, A to C) and in the slices of the volume. The different domains of RyR2 were numbered on the basis of the reasonable docking presented above and on the sequence alignment between RyR2 and RyR1 (fig. S1B), using residue numbers that refer to rabbit RyR2. The N terminus or N-terminal domain (NTD) of RyR2 (residues 1 to 642) has three subdomains: NTDA (residues 10 to 223), NTDB (residues 224 to 408), and NTDc (residues 409 to 642). The RyR2 protein then folds into three homologous domains: SPRY1 (called domain 9 in the previous low-resolution structures; residues 652 to 840, 1457 to 1484, and 1606 to 1625), SPRY2 (residues 841 to 856 and 1084 to 1254), and SPRY3 (called domain 5 in previous structures; residues 1255 to 1641) (16). A domain known as the RyR repeat 1-2 or P1 (called domain 10 in previous structures; residues 869 to 1605) is inserted between the SPRY1 and SPRY2 domains. The SPRY3 domain is followed by the Handle domain (called domain 3 in previous structures; residues 1642 to 2109) and two helical domains: HD1 (called domain 4 in previous structures; residues 2110 to 2679) and HD2 (called domains 7 and 8 in previous structures; residues 2982 to 3390). A second RyR repeat 3-4 or P2 (called domain 6 in previous structures; residues 2701 to 2907) is located between HD1 and HD2. HD2 is followed by a central domain (residues 3613 to 4207), which has two subdomains: EF-hand (called domain 11 in previous structures; residues 4027 to 4087) and U-motif (residues 4091 to 4207). Next, the sequence forms the transmembrane region (residues 4500 to 4888). RyR2 is a six-transmembrane cation channel consisting of transmembrane segments S1 to S6, the S4-S5 linker, the pore helix, the selectivity filter, and luminal loops (18). A cytoplasmic domain between S2 and S3 is known as the VSC domain (residues 4594 to 4719). The C-terminal region after the S6 helices (inner branches; residues 4840 to 4888) reenters the cytoplasmic region, forming a globular C-terminal domain (CTD) (residues 4889 to 4965).

Structural architecture of RyR2 and subunit interactions

According to our proposed atomic model, the overall RyR2 structure could be divided into a three-tier structure, starting from the N terminus that forms the large cytoplasmic scaffold or the peripheral region, a central region constituting the central domain, and a C-terminal region forming the channel domain that spans the membrane (movie S1). The cytoplasmic domains were arranged around a central vestibule formed by the NTDs (Fig. 3A) from the four subunits, each with three distinct subdomains. The subdomains NTDA and NTDB, each with a β trefoil fold, lay side by side and showed intersubunit interactions such that the NTDA of one subunit interacted with the NTDB from the neighboring subunit (Fig. 3, A and B). The structural features of subdomain NTDC were consistent with the previously determined all α-helical structure (19). Each corner of the RyR2, called the Clamp domain in previous structures, was made up of a P1 domain and three SPRY domains from one subunit, and two helical domains (HD1 and HD2) and a P2 domain from the neighboring subunit (Fig. 3B). The P1 domain was visualized as a small L-shaped structure pointing outward (Fig. 3, A and B). The three SPRY domains were arranged in the form of a triangle (Fig. 3C). The Handle domain was positioned such that it interacted with SPRY domains and NTD at one side, whereas the other side of the structure pointed toward the HD2 domain. The helical domains HD1 and HD2 formed two long concave surfaces and adopted a right-handed superhelix of repeating helix-turn-helix structural units (Fig. 3, A and B). In all four subunits, a density bulged out of the main structure at the boundary between the HD1 and HD2 domains (Fig. 3A). This density displayed some flexibility, as evidenced by its contour expanding more than the rest of the domains with a decrease in threshold. It corresponded to the P2 domain, a phosphorylation hotspot that harbors potential phosphorylation sites such as Ser2809 and Ser2815 (20, 21). The extra density in the periphery of the cytoplasmic scaffold next to the SPRY1 domain corresponded to the bound FKBP12.6 (Fig. 3, A and B).

Slices of the volume produced at different cutting planes (through the y and z axes) showed the modular distribution of the structure (Fig. 4, A to K). Structural details were also visible within several domains such as the low-density core of the β trefoils (Fig. 4, C and D). Visualization of the 11.8 Å repeating units of the HD1 domain validated the estimated resolution value calculated according to the 0.5 cutoff of the Fourier shell correlation curve. The central domain in the inner region between the cytoplasmic scaffold and the transmembrane domain (Fig. 4, E and F) did not show intersubunit interactions but was extensively involved in interactions with the cytoplasmic domains NTD, Handle, and HD1 and with the CTD and VSC domains of the transmembrane region. The central domain would thus be in a position to relay the signal from the cytoplasmic region to the transmembrane domain. In the atomic model, the arched architecture of the central domain ended with the density that was occupied by the U-motif that lay closer to the fourfold axis. Pointing away from the fourfold axis, a density coming out of the main structure extended toward the transmembrane region. According to the atomic model, this density would correspond to the fourth helix of the central domain. Finally, a strong density bulged out of the main curved structure and harbored a pair of EF-hands (Fig. 4G).

Fig. 4 Density map of the RyR2-FKBP12.6 complex rendered in solid style to show the structural details.

(A) Side view of the 3D structure cut through the central plane and displayed with two different thresholds. (B) Same slice as in (A) displayed in grayscale. (C to K) Successive planes perpendicular to the fourfold axis as indicated by the dashed lines in (A). The higher-density region within each domain is colored for one of the RyR2 subunits. (C) FKBP12.6 and the NTD, SPRY1, SPRY3, P1, HD1, HD2, and Handle domains. (D) NTD, SPRY1, SPRY2, P1, HD1, HD2, and Handle domains. Note the α-helical repeat of the Handle domain, with an interhelical spacing of 11.75 Å, which looks like a comb. (E) SPRY2, HD1, HD2, Handle, CD, and U-motif domains. (F) HD2, Handle, CD, U-motif, EF-hand, and CTD domains. (G) HD2, CD, U-motif, EF-hand, CTD, and VSC domains. (H) U-motif, EF-hand, VSC, and S6 domains. (I) S6 and VSC domains and the other transmembrane (TM) segments. (J to K) S6 helix and the other transmembrane segments.

According to sequence-based structural prediction (18, 22) and the cryo-EM structures of RyR1, the channel domain of RyR2 consists of six transmembrane helices per subunit, S1 to S6. The transmembrane region of RyR2 was not resolved in our reconstruction. Nevertheless, the VSC domain formed a visible bulge emerging into the cytoplasmic domain (Figs. 2A and 4G). The inner helices and inner branches (which, together, form S6) were also visible especially in the volume slices (Fig. 4, H to K). The helix S6 from all four subunits formed the central channel; it was a long helix with both transmembrane and cytoplasmic components. The transmembrane region of helix S6 lay diagonally across the membrane and bent near the S4-S5 linker extending in an almost vertical direction into the cytoplasm. Beyond the S6 helix, a small globular domain in the cytoplasmic region was assigned to the CTD, in which a small, elongated domain made up of five α helices docks within the U-motif of the central domain.

Characterizing different conformations in the data set

Classification of the whole data set discerned two conformations, C1 and C2 (Fig. 5, A to G). Although their corresponding 3D reconstructions are similar, overall they exhibit noticeable differences. Because the RyRs were purified directly from the heart tissue, it could be possible that these two conformations represented actual physiological states. Structural analysis of the two conformations of RyR2 revealed conformational differences that spanned from the peripheral domains to the central and transmembrane domains. In the Clamp domain, the P1 and SPRY3 domains were closer to the HD1 domain in the neighboring subunit in the C1 than in the C2 conformation (Fig. 5, A and B). In addition, the NTDs were arranged in an anticlockwise twist that placed NTDB closer to NTDA in the neighboring subunit in the C1 conformation (Fig. 5C). The P2 domain displayed more defined density in conformation C1 than in conformation C2 because a comparatively lower threshold was needed to reveal the P2 domain in the C2 conformation (Fig. 5, D and E). Changes in the orientation of NTDA and NTDB appeared to change the morphology of the arched framework of the central domain and slightly rotated the subdomains EF-hand and fourth helix of the central domain, such that in the C1 conformation these two domains could interact with the VSC domain, which is directly connected to the transmembrane assembly (Fig. 5, D and E). In addition, the HD2 domain was more curved toward the main body of the structure in the C2 conformation (movie S2).

Fig. 5 Comparison of the C1 and C2 conformations of RyR2.

(A) Surface representation of the cytoplasmic view (arrows highlight the larger separation of the Clamp domains in the C2 conformation). (B) Surface representation of the side view with arrows highlighting the separation of the Clamp domains. (C) Solid representation of the cytoplasmic view (arrows in a circle emphasize the final position of the NTDs in the C1 conformation). (D) RyR2 structure cut along a plane parallel to the fourfold axis. The insets show the P2 domain at a lower threshold. For the C2 conformation, the top inset shows the P2 domain at an equivalent threshold to that used for the C1 conformation, and the bottom inset shows the P2 domain at a lower threshold to bring up the same density as in the C1 conformation. (E) Solid representation of (D) with the difference in the arched architecture of the CD highlighted by dotted lines. Light dashed lines emphasize the difference in the conformation of the P2 domain. Arrow indicates the different orientation of the EF-hand. (F) Surface representation of the RyR2 structure cut centrally along the fourfold axis. (G) Solid representation of the RyR2 density map as seen from the cytoplasmic side, cut at a plane indicated in panel (F) that shows the different arrangement of the CTD in the C1 and C2 conformations.

Although the resolution was low, structural differences between the two conformations were also evident in the CTD and the inner branches (corresponding to the cytoplasmic sections of the S6 helices). In the C1 conformation, the CTDs were raised, and the four centrally located inner branches were better resolved. The densities of the centrally located inner branches from each protomer fused both at the level of the ion gate and also just below the CTD, displaying a cage-like structure, whereas the “flexible” or C2 conformation displayed a central collapse of the inner branches (Fig. 5F). In cross section, the four tips of the CTDs pointed toward the fourfold axis in the C1 conformation, whereas in the C2 conformation, the CTDs were positioned at an angle such that the tip of each CTD faced the edge of the adjacent protomer in an anticlockwise manner (Fig. 5G).

In summary, the cytoplasmic scaffold seemed to be in a more flexible state in the C2 conformation and in a more compact state in the C1 or upright conformation. The changes in the CTD, VSC, and central domains appeared to contribute to these long-range conformational pathways that correlated changes in the P2 domain in the cytoplasmic scaffold to changes in the S6 helix. The main differences were a tighter interaction of subunits both at the NTDs and the Clamp domains, greater definition of the P2 domain, a change in the structure of the central domain and CTDs, and a better definition of the inner branches in the C1 conformation (movie S2).

Heterogeneity in the basal extent of phosphorylation of RyR2

We determined the degree of heterogeneity in our data set with respect to phosphorylation by Western blot using an antibody against RyR2 phosphorylated at Ser2809. This residue in the P2 domain is a known target of the cyclic adenosine monophosphate (cAMP)–dependent protein kinase A (PKA) present in the heart (23). We used fully phosphorylated and dephosphorylated RyR2 as controls. The basal extent of phosphorylation of RyR2 Ser2809 was estimated to be between 40 and 64% (mean, 55%; SEM, 7.69) with respect to that of PKA-treated, hyperphosphorylated RyR2 (fig. S2, A and B).

Structural differences between RyR1 and RyR2

Although RyR1 and RyR2 share more than 79% similarity (66% identity) and overall similar structural arrangement (Fig. 6, A and B), we noted some divergences when superimposing the 3D structures of the closed channels (Fig. 6, C to F, and movie S1), particularly at the HD2 domain. In RyR1, the density for the HD2 domain was slender and elongated such that it extended to interact with the SR-facing region of the SPRY2 domain (from beneath) (Fig. 6, B and F). In contrast, in RyR2, the HD2 domain had a broader and shorter structure that did not reach the SPRY2 domain (Fig. 6, A and E).

Fig. 6 Comparison of the RyR2 and RyR1 isoforms.

(A) Side view of RyR2-FKBP12.6. (B) Side view of RyR1-FKBP12. (C) Side view of RyR2-FKBP12.6 with RyR1-FKBP12 superimposed in mesh display. The inset displays the difference in positioning of the P2 domain in both RyR1 and RyR2. To visualize the difference, RyR2 is set at a lower threshold than in the rest of the structure. The approximate locations of the DRs (Table 3) are indicated. (D) Side view of RyR1-FKBP12 with RyR2-FKBP12.6 superimposed in mesh display. (E) Luminal view of RyR2-FKBP12.6 with RyR1-FKBP12 superimposed in mesh display. (F) Luminal view of RyR1-FKBP12 with RyR2-FKBP12.6 superimposed in mesh display. The structure of RyR1-FKBP12 in the closed state corresponds to the EM database (EMDB) entry 1606 (64).

It was surprising that the HD2 domain appeared to have a different 3D structure, whereas it shared 83% sequence similarity between RyR1 and RyR2 and did not belong to one of the divergent regions (DRs). To help interpret the structural differences, we performed a sequence-based domain search using the CATH server (24), which classified the HD1 and HD2 domains as leucine-rich repeat variant structural motif (25). Leucine-rich repeat variant is a repetitive structural motif of alternating α and 310 helices arranged in a right-handed superhelix, with the 310 helices forming the outer circumference and the α helices lining the inner curvature. This superhelical architecture provides a versatile framework for protein-protein interactions. In both RyR1 and RyR2, the HD1 and HD2 domains had around 26 leucine repeats with the consensus sequence of LXXL. Our tentative explanation for the structural differences is that given the flexibility of the superhelical architecture, it is possible that the HD2 domain in RyR2 was more contracted and that HD2 in RyR1 was more extended, in the latter case perhaps through stabilizing interactions with the SPRY2 domain. This would affect the topology of this domain that is in the outer boundary of RyR and thus available to interact with regulatory proteins in the junctional space and with RyR itself.

A second important difference between RyR2 and RyR1 was in the P2 domain. At the isosurface threshold used to display the overall structure, the P2 domain was apparent in RyR1, whereas the density of the P2 domain in RyR2 appeared smaller (Fig. 6, A and B), suggesting that the P2 domain in RyR2 had either more conformational heterogeneity within the data set, higher flexibility, or both. In such scenarios, a lower threshold reveals the domain’s boundaries. For RyR2, lowering the threshold resulted in the appearance of the P2 domain (Fig. 6C, inset), which revealed a similar structure to that of the P2 domain of RyR1. Nevertheless, superimposing revealed that the P2 domain was positioned differently in RyR1 and RyR2 (Fig. 6C, inset). Because the P2 domain sequence is between the HD1 and HD2 domains in both RyR isoforms, its placement closer to the main body of the structure suggests that the HD1 domain could also be in a more contracted conformation in RyR2.

In addition, there were substantial structural differences in the transmembrane region of RyR2 and RyR1. When viewed from the luminal side (Fig. 6, E and F), the RyR2 structure was more flattened than RyR1 because the RyR2 structure did not show electron density corresponding to the luminal loops. A difference not related to the sequence was the “fatter” appearance of RyR2 when viewed from the side (Fig. 6, C and D). An extra density around the transmembrane domain appears in some reconstructions of membrane proteins and is related to the presence of detergents and possibly lipids surrounding the hydrophobic regions of the transmembrane domain (26, 27). The presence of this density in 3D reconstructions of RyR2 but not in RyR1 is probably related to the presence of lipids (in addition to detergent) that is usually required in RyR2 preparations.

Three DRs have been previously identified between RyR1 and RyR2 (28), which have the highest sequence variability between the two isoforms (Table 3). The information from the atomic model of RyR2 can be used to trace these locations in the structure of RyR2. DR1 (residues 4210 to 4562 in RyR2) spanned between the U-motif and the S1 helix in the transmembrane region. Our difference map showed that RyR2 had extra density in this region when compared to RyR1, which corresponded to the outer “crown” of the transmembrane domain portion that protrudes into the cytoplasm (Fig. 6, C and D). This extra mass lies close to the cytoplasmic VSC subdomain between helices S2 and S3, and thus, it appears that in RyR2, part of DR1 comes in contact with the more C-terminal VSC domain. DR2 (residues 1353 to 1396 in RyR2), although not assigned in the atomic model of RyR1, formed part of SPRY3. This region lay at the cytoplasmic scaffold, and the residues in this region might play a role in the interaction with DHPR in RyR1, which is absent in RyR2. DR3 (residues 1852 to 1891 in RyR2) was part of the Handle domain but was also not assigned in the atomic model of RyR1. We did not see major localized differences in the Handle domain, suggesting that the structures are similar in RyR1 and RyR2 despite diverging sequences.

Table 3 Divergent regions between RyR2 and RyR1 and their location in the RyR2 3D structure.
View this table:

Binding site of FKBP12.6 and associated conformational changes

We used FKBP12.6 as a means to purify RyR2, and thus, our RyR2s had FKBP12.6 bound. To ascertain the binding site of FKBP12.6, we compared our reconstruction of the RyR2-FKBP12.6 complex with a high-resolution structure of RyR2 without FKBP12.6 (Fig. 7, A and B) (15), both of which corresponded to RyR2 in the closed state. 3D difference mapping revealed that FKBP12.6 bound to the cradle formed by the Handle, SPRY1, and SPRY3 domains (Fig. 7C). The substructure of the 3D difference map corresponded to the molecular shape of FKBP12.6 and allowed further docking of the atomic coordinates of FKBP12.6 using a docking algorithm with a six-dimension search (17). A Laplacian filter option was introduced to enhance the contrast of the solvent-exposed regions. The solution with the highest cross-correlation value yielded an orientation (Fig. 7D) that was equivalent to that found for FKBP12 bound to RyR1 (29) and that is presented by our atomic model.

Fig. 7 3D location of FKBP12.6 and associated conformational changes.

(A) RyR2-FKBP12.6 in three orthogonal orientations. (B) RyR2 in the absence of FKBP12.6. (C) RyR2 in the absence of FKBP12.6 with the 3D difference map corresponding to FKBP12.6 (blue color) superimposed. The 3D location of FKBP12.6 in (C) and the associated conformational change in HD2 in (A) and (B) are highlighted within the squares. The blue mesh shows the volume of RyR2 displayed at lower threshold to expose the greater flexibility of HD2, which can occupy a larger region in space. (D) Enlargement of the region within the square in (C) showing the docking of the atomic coordinates of FKBP12.6 into the 3D difference map.

A second region of difference was found in the HD2 domain (which, as mentioned previously, is shorter and shows no apparent connection with the neighboring subunit when compared to the equivalent domain in the RyR1 isoform), which was even shorter in the high-resolution reconstruction of RyR2 without FKBP12.6, and twisted away from the P1 domain in the opposite subunit (squares in Fig. 7, A and B), resulting in an opening of the Clamp domain (Fig. 7, A and B, and movie S3). Such a difference in the HD2 domain was unexpected because both 3D reconstructions used native, full-length RyR2 protein isolated from natural sources, and although they are from different species (rabbit versus pig), the sequence identity between their respective HD2 domains is high (98%). This second region of difference, although similarly sized to the FKBP12.6 (squares in Fig. 7, A and B), did not show in the 3D difference map (Fig. 7C). Lowering the threshold revealed that in the case of RyR2 without FKBP12.6, the HD2 domain continued as a region of apparent lower density (blue mesh in Fig. 7B); the resulting smaller increment in density between the two reconstructions explained the lack of visible density in this region of the difference map. On the other hand, this suggested that the apparent smaller size of the HD2 domain resulted from flexibility in this domain in RyR2 without FKBP12.6.

DISCUSSION

Here, we reported the 3D structure of the cardiac isoform of the RyR Ca2+ release channel in the closed state at 11.8 Å resolution. At this resolution, we built an atomic model for this protein, which enabled us to analyze the modular organization of this channel. This atomic model was similar to another atomic map based on a high-resolution reconstruction (~4 Å), with RMSD for the different domains ranging between 0.95 and 1.88 Å. The slightly larger RMSD when comparing an entire subunit, 2.49 Å, could be interpreted in the context of the unusually large size of the protein. Regardless, the availability of two atomic maps of the same protein built independently from each other will provide an opportunity to examine the robustness of homology modeling strategies, starting from 3D reconstructions at different levels of resolution.

Natural heterogeneity was expected in the data set because the protein was obtained directly from heart tissue. Classification resolved two conformations within the RyR2 data set. The presence of 2 mM EGTA in the buffer sets the channel in a closed state and nullified the effect of any residual Ca2+ that could induce conformational changes in the structure and subsequent heterogeneity. As an additional test, we compared the two conformations (C1 and C2) and the published closed and open conformations of RyR2 (EMD-9528 and EMD-9529). Calculating the cross-correlation coefficient (using the “Fit in Map” command in Chimera) yielded higher cross-correlation values between any of the two conformations and the closed state (0.9309 for C1 and 0.9338 for C2) than between any of the two conformations and the open state (0.9197 for C1 and 0.9278 for C2), which reinforces the idea that both conformations C1 and C2 correspond to a closed state of RyR2. Within each category (open and closed), the C2 conformation (flexible) had slightly higher cross-correlation than the C1 (upright) conformation to either the open or closed state, probably owing to the more flexible appearance of the C2 conformation.

We further looked for structural differences between the two conformations that could be correlated to recognizable and physiologically important factors. Comparison of the two structural forms revealed a different appearance of P2, the density at the cytoplasmic site that bulges out of the main tetrameric structure. Although this domain overall exhibited very low resolution, pointing toward its highly dynamic nature, the C1 conformation had more structured density than the C2 conformation. The P2 domain has been studied as the target for phosphorylation by kinases and dephosphorylation by phosphatases. Its localization has been confirmed by four independently determined RyR1 structures (1113, 30). In addition, Zalk et al. reported structural heterogeneity in RyR1 that was attributed to phosphorylation (13). The large platform provided by the cytoplasmic domain of RyR2 provides an excellent framework in which several RyR modulators, including protein phosphatases and kinases, could bind and act, which, in turn, could modulate the function of the channel (21, 31). Activation of CAMKII (Ca2+/calmodulin-dependent protein kinase II) and PKA in the cardiomyocyte as a consequence of β-adrenergic stimulation results in the phosphorylation of RyR2. Several residues, especially Ser2809 and Ser2815 in the P2 domain, have been identified as the phosphorylation target site of these two enzymes (20, 21). Ser2809 is estimated to be phosphorylated in around 75% of RyR2 under physiological conditions (32). Phosphorylation of several other ion channels can mediate a global conformational change that can subsequently alter the gating properties of these channels (33, 34). In RyRs, the effect of phosphorylation has been implicated under several pathological conditions such as phosphorylation-induced calcium leak leading to arrhythmia, indicating that phosphorylation affects the function and regulation of RyR2 (35).

The extent of phosphorylation at the P2 domain measured for our sample of ~55% concurred with the observed structural differences. We could not exclude that other factors could also contribute to the observed heterogeneity. However, because of the conformational differences in the P2 domain, which is a phosphorylation hotspot in RyR2, we proposed that differences in phosphorylation could induce a series of structural changes in its neighboring domains. As mentioned previously, the main structural differences were observed in the Clamp domain. It appears that a more structured P2 domain in the C1 conformation could pull together domains HD1, SPRY3, and P1, as compared to the C2 conformation (Figs. 5A and 6A). This change in the Clamp domain could bring about a coordinated structural change that allosterically alters the channel domain and explain the observed variability. To ascertain the conformation that may correspond to a phosphorylated state, we compared the two structures for the RyR1 isoform, one with native degree of phosphorylation and one dephosphorylated [CIP (calf intestinal phosphatase)–treated] (13), and found that a higher degree of phosphorylation was associated with slightly larger separation between neighboring subunits at the Clamp domains and a slightly greater curvature of the HD2 domain toward the main body of the structure. These conformational changes would match the changes between RyR2 in the C1 and C2 conformations and suggest that the C2 (flexible) conformation corresponds to the phosphorylated form. The ability of phosphorylation to induce a conformational change could provide a structural basis to explain its activating effect, possibly by setting RyR2 in a conformation that needs less energy to transition to the open state.

The structural differences observed between RyR2 and RyR1 could be interpreted in the context of the functional differences between these two isoforms. The tip of the HD2 domain in RyR1 loops around the SPRY2 domain at residues ~1200 to 1225, whereas the 3D reconstruction of RyR2 showed that these two domains were separated by ~24 Å and thus too far away to interact. A functional difference between RyR1 and RyR2 that could correlate to the observed structural differences at the HD2 domain is their mode of interaction with the DHPR. DHPRs are physically coupled to RyR1 (36) but not to RyR2. Leong and MacLennan localized a stretch of 37 residues, Arg1076 to Asp1112, within RyR1 that could potentially interact with the DHPR II-III loop (37). Conversely, this interaction was not observed for the cardiac DHPR II-III loop and the residues corresponding to this stretch in RyR2. In both RyR isoforms, this stretch of 37 residues forms part of the SPRY2 domain (38). One possibility is that in RyR1, the connection between HD2 and SPRY2 is required for proper allosteric transmission of the DHPR signal, which is not required in RyR2 because it does not interact with DHPRs. Another nonmutually exclusive possibility is that the differences in HD2 underlie the different RyR-RyR interactions observed in skeletal and cardiac tissues. The typical RyR1-RyR1 “checkerboard” interactions take place through the HD2 domains of two interacting RyR1s (39). In contrast, RyR2 self-association requires SPRY1-SPRY1 or SPRY1-HD2 interactions but not HD2-HD2 interactions (40).

FKBP12.6 is an endogenous high-affinity ligand of RyR2 that allosterically modulates its gating properties. FKBP12.6 deficiency results in lower stability of the closed state, longer mean open lifetimes, and transition to subconductance states (41, 42), and can result in cardiac failure (43, 44). To establish the binding site of FKBP12.6 in RyR2, we subtracted a 3D reconstruction of RyR2 from that of the RyR2-FKBP12.6 complex, which revealed that FKBP12.6 bound to a hollow formed by the Handle, SPRY1, and SPRY3 domains. This location agrees with an earlier determination of the location of FKBP12.6 bound to RyR2 (45) and also with the location of FKBP12 bound to RyR1 (29). The atomic coordinates were then docked in the 3D difference map, revealing a relative orientation of FKBP12.6 with respect to the surface of RyR2 that matched that found for FKBP12 bound to RyR1 (29). The orientation did not match a previous proposition based on a low-resolution reconstruction (33 Å) (45); in this case, the relative orientation was based on biochemical information rather than structural features.

We also found that the HD2 domain was flexible in RyR2 not bound to FKBP12.6, which rendered the distal part of the domain invisible unless the 3D reconstruction of RyR2 was displayed at lower threshold. After discarding the possibility that this difference could arise from the different RyR2 isoforms used in the two studies (because rabbit and porcine RyR2 are highly homologous), we tentatively attributed the different conformation of HD2 to the presence of FKBP12.6, which was the major biochemical difference between the two samples. Thus, it would appear that FKBP12.6 could increase the ordering of the HD2 domain and, conversely, that lack of FKBP12.6 would result in an increase in flexibility of this domain. Because FKBP12.6 is a ligand that stabilizes RyR2 in the closed state, a role in the allosteric stabilization of the HD2 domain could be of functional relevance.

In conclusion, RyR2 is a prominent component of the cardiac couplon that mediates excitation-contraction coupling in cardiomyocytes. Aberrations in RyR2 function have been associated with several pathological conditions. Despite extensive biochemical and physiological characterization, many specific questions related to the function of this calcium channel remain unanswered. The structural information presented here for the RyR2-FKBP12.6 cardiac model will also be crucial and could be extended toward studies of the role of RyR2-mediated calcium signaling in the brain and other organs.

Amid available biochemical studies, there is disagreement on the role of phosphorylation in the physiological or pathological state of RyR2. The structural details presented here showed how changes in the phosphorylation domain of RyR2 could correlate with changes in its overall structure and provided a framework to design experiments to further understand the role that the extent of phosphorylation plays in the functional state of RyR2. The availability of atomic models of RyR2 provides the approximate location of important residues and their interaction with surrounding residues and will be helpful in understanding the impact of several mutations that contribute to various pathological conditions.

Finally, the structural study of RyR2, along with the structure of RyR1, provides a foundation for deciphering the mechanistic differences between these two isoforms and their evolutionary relationship. In this context, the HD2 domain of RyR2, which is apparently disconnected from the neighboring subunit and capable of flexibility, constituted a domain that conferred structural differentiation from RyR1, one that could underlie differences in signaling between the two isoforms.

MATERIALS AND METHODS

Reagents

CHAPS was purchased from EMD Chemicals (Merck Millipore), and dithiothreitol (DTT) was purchased from Calbiochem (EMD, Merck Millipore). All other chemicals with the highest available purity grade were purchased from either Thermo Fisher Scientific or Sigma-Aldrich. The PKA kinase enzyme system was purchased from Promega. Antibody against RyR2-phosphoSer2808 was purchased from Abcam (ab59225). Horseradish peroxidase (HRP)–conjugated secondary antibody was obtained from Abcam. Alkaline phosphatase from calf intestine was purchased from Calzyme. ECL Western blotting substrate was from Pierce.

Preparation of cardiac vesicles

Heart vesicles were prepared as previously described (46), with some modifications. All vesicle preparation steps were carried out at 4°C. Rabbit ventricles were homogenized in four volumes of buffer containing 5 mM imidazole (pH 7.4), 300 mM sucrose, and protease inhibitor cocktail (10 μM leupeptin, 0.77 μM aprotinin, and 10 μM Pefabloc) and were centrifuged at 13,200g for 20 min. The pellet was passed through a commercial mincer and homogenized again with four volumes of homogenization buffer. This homogenate was centrifuged at 16,300g for 20 min, and the supernatant was passed through six layers of nylon mesh. The filtrate was again centrifuged at 220,000g for 50 min. The pellet containing the microsomes was resuspended in the homogenization buffer using a Dounce homogenizer. The cardiac microsomes were aliquoted, flash-frozen using liquid nitrogen, and stored at −80°C. SDS-PAGE was used to check the electrophoretic profile of the resultant microsomal proteins. Quick Start Bradford Protein Assay was carried out to measure the protein concentration (Bio-Rad).

Purification of RyR2

The purification of RyR2 took advantage of the interaction between RyR2 and FKBP12.6. The plasmid containing His-tagged, SBP-fused human FKBP12.6 gene was obtained from T. Murayama (Juntendo University School of Medicine, Tokyo, Japan). The His-tagged FKBP12.6-SBP protein was expressed and purified as previously described (40). To purify RyR2 from cardiac vesicles, vesicles were first detergent-solubilized in 20 mM Na-MOPS buffer with 10.4% CHAPS, 2.2% phosphatidylcholine, 2.5% deoxycholate, 1 M NaCl, 2 mM DTT, and protease inhibitor cocktail for 45 min. After solubilization, the vesicles were centrifuged at 208,000g for 1 hour. The supernatant was diluted five-fold with dilution buffer containing 20 mM Na-MOPS, 2 mM DTT, and protease inhibitor cocktail. The diluted supernatant was passed through streptavidin beads (GE Healthcare) that were preequilibrated with SBP-FKBP12.6. The column was first washed with buffer solution (20 mM Na-MOPS, 200 mM NaCl, 2 mM DTT, 2 mM EGTA, protease inhibitor cocktail, and 0.5% CHAPS). RyR2 was then eluted in the above buffer solution with 5 mM biotin. The purity and quality of the RyR2-FKBP12.6 complex were assessed by 12.5% SDS-PAGE and by negative staining EM. The RyR2-FKBP12.6 complex was then dialyzed against 20 mM MOPS (pH 7.4), 200 mM NaCl, 2 mM DTT, 0.015% Tween 20, and 2 mM EGTA. All procedures were carried out at 4°C.

Cryo-electron microscopy

For preparation of the cryo-grids, the holey grids (Quantifoil, Cu 400 mesh) were coated with a thin layer of carbon using a Denton carbon evaporator. To these grids, 3.5 μl of aliquot of RyR2-FKBP12.6 complex (0.1 mg/ml) was adsorbed for 40 s. The excess of the protein solution was blotted off with Whatman 540 filter paper for 2 s, followed by vitrification by plunging the grid into liquid ethane using an FEI Vitrobot Mark IV cryo-plunger. Frozen-hydrated samples were imaged on a Titan Krios operated at 300 kV. Data were recorded with a Falcon II direct electron detector at 1.34 Å per pixel. Each image was obtained with an exposure time of 2 s with a dose rate of ~20 electrons/Å2, which was fractionated into seven frames per image. A total of 858 images were collected with a defocus value in the range of 2.5 to 4.0 micrometers at a nominal magnification of ×59,000.

Single-particle image processing

Inter-frame motion was corrected using MotionCorr (47). Around 1000 particles were picked manually using Boxer (48). Relion 1.3 was used for automated particle picking (49). First, a 2D classification was carried out on the manually picked particles to select the best classes. These classes were then given as input to the Relion’s Autopick module after optimizing the picking threshold and minimum interparticle distance so that only individual RyR2s not interacting with neighboring RyR2s were picked. All false positives picked by Relion were then removed manually for each micrograph. This yielded 13,600 particles. The particle stack was then normalized and submitted to FREALIGN v9 software (50) for refinement. The starting model was a 3D structure of RyR1 (EMDB: 1606) low pass–filtered to 80 Å to eliminate any bias. To calculate the resolution, the data set was split into two half data sets (odd and even), and the resolution was taken at the 0.5 cutoff of the Fourier shell resolution. To check for structural heterogeneity in the data, the data set was classified using Frealign into two classes, C1 and C2, which correspond to 52 and 48% of the data set, respectively. Classification into more than two classes yielded 3D reconstructions that did not give any further discernable structural feature than these seen in C1 and C2. For 3D difference mapping, the 3D structures of RyR2 with or without FKBP12.6 were filtered to 12 Å resolution. The 3D reconstruction of RyR2 was then directly subtracted from the 3D reconstruction corresponding to the RyR2-FKBP12.6 complex.

Construction of the atomic model of RyR2

The atomic structure was built for the cryo-EM map of the C1 or upright conformation of RyR2. Because of the fourfold symmetry, the model was built first for a single subunit. The model for each single subunit was then used to generate a tetrameric structure. First, molecular modeling was carried out starting from available homologous crystal structures (Table 1) using Modeller version 9.16 software (51). For the rest of the structure for which crystal structures were not available (corresponding to the Handle domain, the central domain, and the channel domain), the 3D atomic model of RyR1 from cryo-EM (PDB ID: 3J8H) was used. Next, the EM density map was segmented into regions corresponding to different domains using the Segger module in Chimera (52). The modeled domains were then docked into these segmented densities using the Colores six-dimension search module of Situs 2.8, using the Laplacian filter option (53). To connect the different domains with each other, especially for SPRY domains, the information from the cryo-EM–determined structure of RyR1 (PDB ID: 3J8H) was used. In the second step, after placing all the domains into the density map of RyR2, the refinement of the one complete subunit was carried out using REFMAC, using modified procedures for cryo-EM maps (54). Secondary structure restraints were generated using PROSMART (55). The refined subunit was then tetramerized, and refinement with REFMAC was carried for all the four subunits into the corresponding cryo-EM map. The structure was refined under symmetry restraints in REFMAC v5.8 (54). Figures and movies showing electron densities and atomic models were generated using UCSF Chimera (56) or PyMOL (www.pymol.org/) (57). Software was provided by SBGrid (58).

Western blot analysis

The extent of Ser2809 phosphorylation in purified rabbit RyR2-FKBP12.6 was estimated using an antibody specific to RyR2-phosphoSer2808 (equivalent to rabbit RyR2 Ser2809) (ab59225, Abcam) following a Western blot with a generic antibody against RyR (34C, ab2868, Abcam) to demonstrate equivalent protein loading. Band intensities of phosphorylated RyR2-Ser2809 or the total protein were measured, in parallel, in three independent samples: cAMP-dependent protein kinase (V5161, Promega)–treated, calf alkaline phosphatase (Calzyme)–treated, and untreated representing basal phosphorylation. The three samples were incubated for 4 hours at room temperature. The samples were applied to a 10% SDS-PAGE gel and transferred to a polyvinylidene difluoride membrane at 100 V for 2 hours, followed by blocking the nonspecific sites with 5% bovine serum albumin in tris-buffered saline (TBS) buffer [20 mM tris-HCl (pH 7.5) and 150 mM NaCl] for 1 hour at room temperature. The membrane was probed with the generic primary antibody in TBS buffer (1:2000 dilution) overnight at 4°C, followed by three washes with TBS-T buffer (TBS buffer and 0.1% Tween 20). The membrane was further incubated with an HRP-conjugated secondary antibody (1:5000 dilution) at room temperature for 1 hour, followed by visualization with enhanced chemiluminiscence. Images were recorded on a FluorChem E gel imager. The membrane was stripped of bound antibody in acidic glycine buffer [25 mM glycine (pH 2.2), 0.1% SDS, and 1% Tween 20] before the second Western blot to detect phosphorylation under identical conditions, except for the antibody dilution (1:1000). Densitometry analysis of the resulting bands was carried out using ImageJ software (NIH). The RyR2-Ser2809 phosphorylation peak area value for each band was normalized with its respective intensity obtained with the 34C antibody. The normalized value for the phosphatase-treated RyR2, considered the baseline, was then subtracted from both PKA-treated and untreated samples. The ratio of phosphoSer2809 found in untreated RyR2 was calculated with respect to the PKA-treated RyR2. The overall experiment was carried out in triplicate.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/10/480/eaai8842/DC1

Fig. S1. Atomic model of the structure of RyR2-FKBP12.6.

Fig. S2. Degree of phosphorylation of RyR2.

Movie S1. Overall structure of RyR2.

Movie S2. Morph between conformations C1 and C2 for RyR2.

Movie S3. Morph of RyR2 with and without FKBP12.6.

REFERENCES AND NOTES

Acknowledgments: We thank H. Wang and N. Scarsdale for expert computing support. We thank T. Murayama (Juntendo University School of Medicine, Tokyo, Japan) for providing the FKBP12.6 plasmid. Funding: This work was supported by the American Heart Association grant no. 14GRNT19660003 and the Muscular Dystrophy Association grant no. MDA352845. Cryo-EM sample preparation and screening were conducted at the Cryo-Electron Microscope Facility at the Virginia Commonwealth University (VCU), which is supported by the VCU School of Medicine and by Startup Funds to M. Samsó. Direct electron detector data collection was carried out at the Molecular Electron Microscopy Core facility at the University of Virginia (UVA), which is supported by the UVA School of Medicine and NIH grants S10-RR025067 and S10-OD018149. Author contributions: S.D., V.C., D.J.S., and A.R.N. performed the experiments. S.D., K.D., and M.S. collected the data. S.D., J.L., and M.S. analyzed the data. S.D., J.L., and M.S. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The cryo-EM map of the RyR2 structure was deposited in the 3D Electron Microscopy Data Bank under EMDB ID: 8303. The coordinates of the atomic model for RyR2 structure for conformation 1 were deposited in the Protein Data Bank under PDB ID: 5L1D.
View Abstract

Navigate This Article