Three-Dimensional Localization of Serine 2808, a Phosphorylation Site in Cardiac Ryanodine Receptor*

Type 2 ryanodine receptor (RyR2) is the major calcium release channel in cardiac muscle. Phosphorylation of RyR2 by cAMP-dependent protein kinase A and by calmodulin-dependent protein kinase II modulates channel activity. Hyperphosphorylation at a single amino acid residue, Ser-2808, has been proposed to directly disrupt the binding of a 12.6-kDa FK506-binding protein (FKBP12.6) to RyR2, causing a RyR2 malfunction that triggers cardiac arrhythmias in human heart failure. To determine the structural basis of the interaction between Ser-2808 and FKBP12.6, we have employed two independent approaches to map this phosphorylation site in RyR2 by three-dimensional cryo-electron microscopy. In one approach, we inserted a green fluorescent protein (GFP) after amino acid Tyr-2801, and mapped the GFP three-dimensional location in the RyR2 structure. In another approach, the binding site of monoclonal antibody 34C was mapped in the three-dimensional structure of skeletal muscle RyR1. The epitope of antibody 34C has been mapped to amino acid residues 2,756 through 2,803 of the RyR1 sequence, corresponding to residues 2,722 through 2,769 of the RyR2 sequence. These locations of GFP insertion and antibody binding are adjacent to one another in domain 6 of the cytoplasmic clamp region. Importantly, the three-dimensional location of the Ser-2808 phosphorylation site is 105-120Å distance from the FKBP12.6 binding site mapped previously, indicating that Ser-2808 is unlikely to be directly involved in the binding of FKBP12.6 to RyR2, as had been proposed previously.

natural death in the United States, producing about 330,000 deaths each year (1). SCD occurs when the electrical system in the heart malfunctions, leading to fatal arrhythmias. Most cases of SCD can be explained by cardiovascular abnormalities identifiable at autopsy, but some SCD is associated with structurally normal hearts (2). In some such cases, abnormal calcium release through a dysfunctional cardiac muscle ryanodine receptor (type 2 RyR, or RyR2) has been implicated, and molecular genetic analysis has demonstrated that mutations in the RyR2 gene can be linked to about 14% of these structurally normal heart SCD cases (2).
RyR2 functions as a calcium release channel in the heart, where it plays a crucial role in excitation-contraction coupling. To date, more than 60 naturally occurring mutations in RyR2 have been linked to two genetic forms of ventricular arrhythmias: catecholaminergic polymorphic ventricular tachycardia (CPVT) and arrhythmogenic right ventricular dysplasia type 2 (ARVD2; Refs. [2][3][4][5][6]. These mutations are largely clustered in three regions of the RyR2 amino acid sequence: region 1 is near the N terminus, from residues 77 to 466; region 2 is near the middle of the sequence, spanning residues 1,724 through 2,958; and region 3 includes both cytoplasmic and membrane-associated regions of the sequence and comprises residues 3,778 through 4,959. Intriguingly, the same three regions are homologous to the three malignant hyperthermia and central core disease mutation regions in skeletal isoform RyR (type 1 RyR or RyR1, Ref. 7), suggesting that the three regions represent structural domains critical for the regulation of both RyR1 and RyR2, and suggesting also that the molecular pathologies underlying these RyR-linked skeletal and cardiac muscle diseases involve similar mechanisms.
The causal relationships among RyR2 mutations, the abnormal Ca 2ϩ release function, and physiological manifestation is under investigation. Some mechanisms underlying RyR2 channel dysfunction in SCD have been proposed, including channel instability because of destabilization of the RyR2-FKBP12.6 interaction (8,9) or domain-domain interaction (10,11); reduced threshold for store overload-induced Ca 2ϩ release (12); altered channel sensitivity to cytoplasmic calcium (13); and reduced Mg 2ϩ inhibition (14). FKBP is an isomerase that is widely expressed and involved in many cell functions. Two isoforms, FKBP12 and FKBP12.6, bind tightly to RyR1 and RyR2, respectively, and they are considered to be integral subunits of RyR (15,16). FKBP12 and FKBP12. 6 have been shown to stabilize the closed state of the RyR1 and RyR2 Ca 2ϩ channels, thereby preventing inappropriate spontaneous channel openings (17,18). Marx et al. (8) have shown that protein kinase A (PKA) phosphorylation of RyR2 dissociated FKBP12.6 and affected the channel open probability, and that hyperphosphorylation of RyR2 at a single amino acid residue Ser-2809 (in rabbit RyR2 sequence, corresponding to Ser-2808 in mouse) leads to defective channel function in failing heart. Subsequently, Wehrens et al. (9) have shown that disease-causing mutations in RyR2 (three CPVT mutations: S2246L, R2474S, and R4497C) reduced the binding affinity of FKBP12.6 for RyR2, and increased the channel leakage that can trigger fatal cardiac arrhythmias under physical and emotional stresses.
However, the mechanisms proposed by Marx, Wehrens, and colleagues are not consistent with other laboratory findings. Li et al. (19) did not find any affect of PKA phosphorylation of the RyR2 on calcium sparks in mouse ventricular myocytes. Jiang et al. (20) did not observe dissociation of FKBP12.6 from RyR2 in cardiac microsomal membranes treated with PKA. Stange et al. (21) made site-directed substitutions of RyR2 at Ser-2809 (with Asp or Ala), and showed that mutant RyR2s neither abolished FKBP12.6 binding nor substantially changed channel functional properties. Xiao et al. (22) found that FKBP12.6 can bind to both the Ser-2808 (mouse) phosphorylated and non-phosphorylated forms of RyR2, and that a S2808D phosphomimetic mutant retained the ability to bind FKBP12.6. Furthermore, complete phosphorylation at Ser-2808 by exogenous PKA did not disrupt the FKBP12.6-RyR2 complex. George et al. (23) have demonstrated that CPVT RyR2 mutations augmented Ca 2ϩ release, yet the RyR2-FKBP12.6 complex remained intact.
The binding sites for FKBP12 and FKBP12.6 have been mapped in the three-dimensional structures of RyR1 and RyR2, respectively (24,25). However, no information regarding the phosphorylation site Ser-2808 in the RyR three-dimensional architecture has been reported. We believe that a direct structural approach to localize the Ser-2808 in the three-dimensional structure of RyR is relevant to testing the proposed FKBP12.6 dissociation mechanism. Recently, we have mapped two mutation hotspots in the three-dimensional structure of RyR2 by green fluorescent protein (GFP) labeling at specific amino acid residues, combined with cryo-electron microscopy (cryo-EM) and single-particle image processing methods (26,27), these studies provided some structural insights into the mechanism of RyR2 mutations. In the present study, we employed the same structural approach to determine the Ser-2808 phosphorylation site in the three-dimensional structure of RyR, thereby directly testing the mechanism of PKA phosphorylation-induced FKBP12.6 dissociation that was proposed by Marks and colleagues. We have used two independent approaches to identify the three-dimensional location of the phosphorylation site. In the first approach, we inserted GFP after amino acid Tyr-2801, and mapped its three-dimensional location in the RyR2 structure. In the second approach, the binding site of monoclonal antibody 34C (mAb34C) was mapped in the three-dimensional structure of RyR1. The epitope of the 34C antibody has previously been determined to include amino acid residues 2,756 through 2,803 of RyR1, corresponding to residues 2,722 through 2,769 of RyR2 (28). We found that the mapped GFP insertion site and the antibody binding site are adjacent to one another in domain 6 of the cytoplasmic clamp region. However, neither site is close to the FKBP12.6 binding site mapped previously.
Cell Culture and cDNA Transfection-HEK293 cells were maintained in Dulbecco's modified Eagle's medium as described previously (31). HEK293 cells grown in 100-mm tissue culture dishes for 18 -20 h after subculture were transfected with 12 g of RyR A4860G or RyR Y2801-GFP cDNAs using Ca 2ϩ phosphate precipitation.
Expression and Purification of RyR2 A4860G and RyR2 Y2801-GFP -HEK293 cells grown for 24 -26 h after transfection were washed three times with phosphate-buffered saline plus 2.5 mM EDTA and were harvested in the same solution by centrifugation for 8 min at 700 ϫ g in an IEC Centra-CL2 centrifuge. The cell pellets were solubilized in a lysis buffer containing 25 mM Na-PIPES (pH 7.2), 140 mM NaCl, 5 mM EGTA, 2.5 mM dithiothreitol, 1% CHAPS, 0.5% egg lecithin, and a protease inhibitor mixture (2 mM benzamidine, 4 g/ml leupeptin, 2 g/ml pepstatin A, 4 g/ml aprotinin, and 0.75 mM phenylmethylsulfonyl fluoride) and were incubated on ice for 1 h. Insoluble material was removed by centrifugation at 16,000 ϫ g twice in a microcentrifuge at 4°C for 30 min, and the cell lysates were collected. Sucrose (400 mM final concentration) and NaCl (400 mM final concentration) were added to the cell lysates. Glutathione-Sepharose beads (100 l) containing 400 g of bound glutathione S-transferase (GST)-FKBP12.6 fusion protein were added to the cell lysates, and the mixture was incubated at 4°C with rotation for 2 days. The beads were collected by centrifugation and washed four times with lysis buffer containing 0.5% CHAPS and 0.25% egg lecithin. The RyR2 proteins were eluted from the beads by incubation with 100 mM glutathione in 25 mM Na-PIPES (pH 7.2), 400 mM NaCl, 400 mM sucrose, 1 mM EGTA, 0.2 mM CaCl 2 , 2.5 mM dithiothreitol, 0.4% CHAPS, 0.16% egg lecithin, 200 mM Tris/HEPES (pH 7.4), and protease inhibitor mix-ture for 15 min. The eluate was aliquoted, frozen in liquid nitrogen, and stored at Ϫ90°C.

Ca 2ϩ Release Measurement and [ 3 H]Ryanodine
Binding Assay-The concentration of free cytosolic Ca 2ϩ in the transfected HEK293 cells was measured with the fluorescence Ca 2ϩ indicator dye fluo-3 as described previously (31). Equilibrium [ 3 H]ryanodine binding to cell lysates was carried out as described previously (32).
Western Blotting and Fluorescence Imaging-The purified RyR2 A4860G and RyR2 Y2801-GFP proteins were subjected to 6% SDS-PAGE and were detected by Western blotting using anti-RyR antibody (1:1000) and anti-GFP antibody (1:1000) (Affinity BioReagents, Golden, CO) as described previously (30). For detection of the green fluorescence of RyR2 Y2801-GFP , HEK293 cells were grown on glass coverslips placed in a 100-mm tissue culture dish and transfected with RyR2 Y2801-GFP as described above. 24 h after transfection, the coverslips were washed three times with phosphate-buffered saline, mounted on a glass slide, and visualized with a Leica DMRB fluorescent microscope using a ϫ40 objective.
mAb34C Fab Purification and Activity Test-The mAb34C hybridoma cell line was injected into mice to produce ascites fluid. MAb34C IgG was purified from 1.2 ml of ascites fluid by affinity chromatography on immobilized protein A using the Pierce ImmunoPure IgG purification kit as described by the manufacturer (Perbio Science, Rockford, IL). Fab fragments were obtained from purified IgG by papain digestion using Pierce ImmunoPure Fab preparation kit. To test the binding activity of Fab, a precipitation experiment was performed: 40 g of GST-FKBP12 with glutathione-Sepharose beads, and 12 g of purified 34c Fab with or without 20 g of RyR1, were incubated separately for 30 min at room temperature, and the two complexes were then mixed together and incubated for a further 30 min. The whole protein complexes were eluted and then analyzed by SDS-PAGE.
Purification of RyR1 and Preparation of RyR1 ϩ mAb34c Fab Complexes-RyR1 was isolated from terminal cisternae vesicles from rabbit skeletal muscle as previously described (33). The RyR1 ϩ mAb34c Fab complexes were formed by addition of mAb34c Fab fragments to purified RyR1 with a molar ratio 8:1 (8 Fab fragments per RyR monomer), and incubated for 30 min.
Cryo-electron Microscopy and Image Processing-EM grids were prepared for cryo-EM by an FEI Vitrobot computer-controlled freeze-plunging instrument (FEI Company, Hillsboro, OR). Micrographs were recorded using low-dose protocols on an FEI Tecnai F20 field emission gun transmission electron microscope operated at 200 kV, equipped with an Oxford CT3500 cryo-transfer holder (Gatan, Inc., Warrendale, PA). The temperature of the grids was maintained at around Ϫ170°C. The defocus of the micrographs ranged between 1.5 and 4.5 m, at a magnification of 50,760 ϫ (Ϯ2%). To obtain more side-view information of the RyR molecules, we tilted the grids between 0°and 50°during data collection. Each exposure corresponded to an electron dose of ϳ10 e Ϫ /Å 2 . Micrographs were checked for drift, astigmatism, and the presence of Thon rings by optical diffraction. Selected electron micrographs were digitized on a Zeiss/Imaging scanner (Z/I Imaging Corporation, Huntsville, Alabama) with a step size of 14 m.
Images were processed using the SPIDER/WEB software package (34). For two-dimensional multi-reference classification analysis, each particle image was assigned to one group of a set of different orientations after the first round of projection matching, and two-dimensional averages for each group were computed for both the control RyR1 and the RyR1 ϩ mAb34C Fab complex. A set of multi-reference images was generated by mixing the control RyR1 and RyR1 ϩ mAb34C Fab complex two-dimensional averages (with 0 to 4 Fab fragments occupied the binding sites). The particles from the RyR1 ϩ mAb34C Fab complex dataset were subjected to a supervised classification employing the multi-reference image templates, and particles were assigned to a class based on the highest similarity (judged by the cross-correlation coefficient, Ref. 35). Any particles without 4 Fab fragments bound were then eliminated from the dataset in the following image processing.
Three-dimensional reconstructions were obtained from images of particles lying in all available orientations, through a projection matching procedure (36). The final three-dimensional reconstructions of RyR2 A4860G , RyR2 Y2801-GFP , control RyR1, and the RyR1 ϩ mAB34C Fab complex were computed from 7,591, 6,086, 16,043, and 16,782 particle images, respectively, ensuring that the particle orientations covered the entire angular space, and that each view would be represented by almost equal numbers of particles. 4-fold symmetry was enforced in all three-dimensional reconstructions. The final resolutions were estimated to be 27 Å by the Fourier shell correlation with a cut-off of 0.5 (37) for both RyR2 A4860G and RyR2 Y2801-GFP , and 25 Å for both control RyR1 and the RyR1 ϩ mAB34C Fab complex.

Construction, Expression, Functional Characterization, and
Purification of RyR2 A4860G and RyR2 Y2801-GFP -We have previously shown that mutation A4860G, located in the pore inner helix of RyR2, displays a reduced basal channel activity and a reduced sensitivity to activation by caffeine (29). These observations suggest that the A4860G mutation stabilizes the closed state of the channel. We reasoned that such a mutant channel would be superior to the wild-type channel for structural studies. Indeed, we found that the A4860G mutant protein can be readily purified and visualized by cryo-EM as intact, squareshaped particles, and there is no structural difference between the wild-type and A4860G RyR2s. Accordingly, the A4860G mutant RyR2 (RyR2 A4860G ) was subsequently used as the control for our structural studies. To determine the location of the phosphorylation site, Ser-2808, in the three-dimensional structure of RyR2, GFP was inserted after Tyr-2801 in the RyR2 A4860G (Fig. 1A). The resulting GFP-RyR2 fusion protein, RyR2 Y2801-GFP , was expressed in HEK293 cells. As shown in Fig.  1B, HEK293 cells expressing RyR2 Y2801-GFP were readily detected by their green fluorescence, indicating proper folding of the inserted GFP. To determine whether RyR2 Y2801-GFP functions as a Ca 2ϩ release channel, we measured intracellular Ca 2ϩ release, using the fluorescent Ca 2ϩ indicator dye fluo-3 acetoxymethyl ester (fluo-3 AM) in HEK293 cells transfected with RyR2 A4860G or RyR2 Y2801-GFP . As shown in Fig. 1C, HEK293 cells transfected with RyR2 Y2801-GFP cDNA displayed multiple Ca 2ϩ release events in response to repeated stimulation by caffeine, a well known RyR activator. Interestingly, the addition of ryanodine led to a diminished response to multiple caffeine stimulations (Fig. 1D). This diminution most likely resulted from modulation of the channel by ryanodine, which is known to bind only to the open state of the channel, and to dramatically increase the channel open probability. As a consequence of binding, the ryanodine-modified channel no longer responded to further caffeine stimulation (30). These results suggest that RyR2 Y2801-GFP is sensitive to caffeine and ryanodine modification. Importantly, the Ca 2ϩ release responses of RyR2 Y2801-GFP to caffeine and ryanodine were similar to those observed for the control RyR2 A4860G . Furthermore, RyR2 Y2801-GFP and RyR2 A4860G showed similar Ca 2ϩ dependence of [ 3 H]ryanodine binding (Fig. 1E). The EC 50 for Ca 2ϩ of RyR2 A4860G and RyR2 Y2801-GFP are 0.32 Ϯ 0.01 M and 0.31 Ϯ 0.02 M (mean Ϯ S.E., n ϭ 6), respectively. These data suggest that the RyR2 Y2801-GFP protein retains proper folding and channel function.
The RyR2 A4860G and RyR2 Y2801-GFP proteins were purified from cell lysates by affinity chromatography, with GST-FKBP12.6 used as the affinity ligand (30). The purified protein was analyzed by SDS-PAGE and Western blotting. A single high molecular mass band, which migrated at a slightly slower rate than the RyR2 A4860G band did (as expected, due to the additional mass of GFP), was detected in the purified sample of RyR2 Y2801-GFP . This band was recognized by both the anti-RyR and anti-GFP antibodies. On the other hand, the purified RyR2 A4860G protein was recognized only by the anti-RyR antibody (Fig. 1F).
Three-dimensional Localization of Y2801-GFP-The RyR2 A4860G and RyR2 Y2801-GFP samples were diluted to a suitable concentration, applied to carbon-coated EM grids, blotted, and rapidly frozen, and imaged by electron microscopy. Fig. 2, A and B shows typical electron micrographs in which individual RyR2 A4860G and RyR2 Y2801-GFP channels are visible, most of these channels show standard RyR morphology (38).
Image processing was performed using the SPIDER/WEB software package (34). Three-dimensional reconstructions were obtained by a projection matching procedure (36) . Fig. 2, C and D show surface representations of the three-dimensional reconstructions of RyR2 A4860G and RyR2 Y2801-GFP in three principal orientations. For both the reconstructed structure consists of two major components: a large cytoplasmic assembly composed of at least 10 distinct domains (labeled by numerals 1-10; Refs. 38 and 39), and a smaller transmembrane assembly (labeled "TA"). For both of the reconstructions, the final resolution was estimated to be 27 Å using the Fourier shell correlation criterion with 0.5 as the cutoff value (37). Overall, the two three-dimensional reconstructions are very similar, but there are still some differences. The most noticeable difference was found in the region between domains 5 and 6; specifically, the volume of the bridge linking domains 5 and 6 of RyR2 Y2801-GFP appears to be larger than the corresponding bridge volume in RyR2 A4860G (highlighted in red circles). To more precisely delineate the differences, we generated a three-dimensional difference map by subtracting the volume of RyR2 A4860G from that of RyR2 Y2801-GFP . The difference regions were displayed in green and superimposed on the three-dimensional reconstruction of RyR2 A4860G (Fig. 2E). This difference map clearly indicates one significant difference, located in each bridge connecting domains 5 and 6 in the cytoplasmic assembly. GFP was inserted into each RyR2 monomer and, because RyR2 is a homo-tetramer composed of four identical monomers, the difference is expected to repeat four times in the three-dimensional difference map.
We are confident that the extra density in the map represents the excess mass contributed by the GFP insertion in RyR2 Y2801-GFP , because it is the only significant difference that is seen when the three-dimensional difference map is displayed with the same density threshold as we used to image the RyR2 Y2801-GFP and RyR2 A4860G reconstructions, no additional filtering or masking was need. Furthermore, the calculated volume of one of the four repeated difference in Fig. 2E corresponds to a molecular mass of 28 kDa, assuming a protein density of 1.37g/cm 3 ; this value agrees well with the molecular mass of GFP (38). Other minor differences, only visible at a much lower density threshold, are not consistent with the inserted GFP in terms of either volume or density. In summary, Tyr-2801, as indicated by the inserted GFP, is located in the bridge between domains 5 and 6, in each of the corner (clamp) regions of RyR2.
Three-dimensional Localization of mAb34C Fab Fragment-In the second approach, we localized the binding site of a sequence specific monoclonal antibody, mAb34C, on skeletal isoform RyR1 (Fig. 3A). MAb34c is an antibody that is commercially available and is widely used to detect RyR in biochemical assays; it recognizes all three isoforms of RyR. The epitope of mAb34C has previously been determined to involve amino acid residues 2,756 through 2,803 of RyR1, corresponding to residues 2,722 through 2,769 of RyR2 (28), about 40 amino acid residues away from the RyR2 PKA phosphorylation site Ser-2808. The primary sequence and three-dimensional structure are highly conserved between RyR1 and RyR2, and the binding site of FKBP12 in RyR1 is located in the same region as is the FKBP12.6 binding site in RyR2 (24,25). Thus, through comparison of the three-dimensional locations of the mAb34C binding site and the FKBP binding site, we will be able to assess the likelihood of interaction between FKBP and the PKA phosphorylation site.
A mAb34C hybridoma cell line was use to produce milligram amounts of antibodies. mAb34C IgG was purified from mouse ascites, and Fab fragments were obtained from purified IgG by papain digestion. We used Fab fragments instead of IgGs in our single particle cryo-EM, because one IgG molecule has two antigen-binding sites, and could thus cross-link two RyR mol-ecules. The purified mAb34c IgG and Fab were checked by SDS-PAGE. The binding activity of the mAb34C Fab was tested by coprecipitation with RyR1. Fusion protein GST-FKBP12 binds specifically to RyR1, but not to mAb34C Fab. MAb34C Fab was incubated with or without RyR1 before mixing with GST-FKBP12, and it was then precipitated by glutathione-Sepharose FIGURE 3. Binding site, purification, and binding activity test for mAb34C. A, schematic illustration of the mAb34C Fab fragment binding site in the RyR1 amino acid sequence. The epitope of mAb34C comprises amino acid residues 2,756 through 2,803 of RyR1, corresponding to residues 2,722 through 2,769 of RyR2, thus the mAb34C binding site is close to the PKA phosphorylation site, Ser-2808 (purple circle). B, SDS-PAGE of mAb34C ascites, IgG, and Fab fragment. IgG was purified from mouse ascites, and Fab fragments were obtained from purified IgG by papain digestion. C, co-precipitation of mAb34C Fab fragments and RyR1 by GST-FKBP12 pull-down. Sup, supernatant; Pell, pellet.
beads. After precipitation, the supernatant and pellet were analyzed by SDS-PAGE. Fig. 3C shows that in the presence of RyR1 (lanes 1 and 2), the mAb34C Fab was pulled down by GST-FKBP12 and glutathione-Sepharose beads, but it was not pulled down in the absence of RyR1 (lanes 3 and 4). This precipitation test clearly demonstrates that the purified Fab fragment of mAb34C can bind to detergent-solubilized RyR1 specifically.
The purified mAb34C Fab fragments were incubated with RyR1 for complex formation. Both the mAb34C Fab ϩ RyR1 complexes and control RyR1 (without added Fab fragments) were imaged by cryo-EM. Fig. 4, A and B shows typical electron micrographs in which individual particles of control RyR1 and mAb34C Fab-RyR1 complex are visible.
A problem inherent in this approach is that among the more than ten thousand receptor images that were selected for computation of the three-dimensional structures of mAb34C Fab ϩ RyR1 complex, some receptors will have less than the full complement of four bound Fab molecules. (RyR1 is a homo-tetramer and thus has four Fab binding sites). This situation contrasts with that for the GFP insertion approach, in which cDNA of GFP is inserted into that of RyR2, and thus all of the expressed receptors have the GFP insertion (HEK293 cells do not possess endogenous RyR). In the antibody approach, since the binding affinity between RyR1 and mAb34C Fab is estimated to be in the micromolar range, a small proportion of RyR1 subunits are expected to be free of bound Fab. A simple calculation, using an estimated binding affinity of 1 M, reveals that 98.8% of RyR1 subunits are bound to Fab under our experimental conditions. To estimate the actual percentage of RyR1 that are bound to Fab molecules, we performed a two-dimensional classification analysis (multi-reference-based alignment, followed by a variation statistical analysis, Ref. 35) at the beginning of image processing, to assess the extent of this heterogeneity. Our result indicated that about 93% of the RyR1 particles have 4 Fab fragments bound, whereas 7% RyR1 have fewer. Most of the latter Fab-depleted RyR1 particles were then eliminated from the dataset in the following image processing. However, this two-dimensional classification is accurate for low-tilt angle data (below 30°), so we cannot rule out that a small percentage of Fab-depleted RyR1 from high-tilted angle data remain in the final dataset (less than 2% of the total number of particles). Because the three-dimensional reconstruction is calculated from over 16,000 particle images, a strong density feature due to bound Fab fragment is expected, and in fact observed, at our current level of resolution (see results below).  Fab ϩ RyR1 complex in three views. The final resolution was estimated to be 25 Å. Overall, the two three-dimensional structures are also very similar, the differences appear mainly in domain 6, as shown in the side-views in the two panels (highlighted by red circles), where domain 6 of the mAb34C Fab ϩ RyR1 complex appears "taller" at its outer margin. We generated a three-dimensional difference map by subtracting the volume of the control RyR1 from the volume of the mAb34C Fab ϩ RyR1 complex. The difference regions, displayed in orange, are superimposed on the three-dimensional reconstruction of control RyR1 (Fig. 4E). This difference map clearly indicates that there is one difference per RyR1 subunit, located at the periphery of domain 6.

DISCUSSION
FKBP12.6 binds tightly to cardiac muscle RyR2, and is thought to modulate RyR2 channel gating by stabilizing the closed state of the channel, and by promoting transitions between closed and open states (i.e. eliminating sub-conductance states; Ref. 16). A highly homologous protein, FKBP12, binds to skeletal muscle RyR1 and stabilizes RyR1 in a similar manner (17). Dissociation of FKBP from RyR by drugs (e.g. FK506) or antibiotics (e.g. rapamycin) causes an increased channel open probability, higher frequency of sub-conductance states, and sarcoplasmic reticulum calcium leakage (18). The Marks group (8) has proposed that PKA hyperphosphorylation of a unique amino acid, Ser-2809 in RyR2, dissociates FKBP12.6, and that such a dissociation affects RyR2 functioning, thereby contributing to the ventricular arrhythmias that often occur during heart failure. In a later report from the same group, it was proposed that FKBP12.6 dissociation occurs through an electrostatic repulsion between the negatively charged phosphate group at Ser-2809 of RyR2 and a negatively charged residue, Asp-37, of FKBP12.6 near the hydrophobic binding pocket (9). Mutations in RyR2 that cause CPVT and AVRD2 are hypothesized to weaken the affinity of RyR2 for FKBP12.6 (9). As a result, disruption of the FKBP12.6-RyR2 interaction at Ser-2809 was proposed as a novel mechanism for cardiac muscle dysfunction in sudden cardiac death and heart failure.
However, the proposed mechanisms are not consistent with other laboratory findings (19 -23), and our present results contribute to the growing body of evidence that disfavors the proposal that PKA hyperphosphorylation at Ser-2809 (or Ser-2808 in different species) in RyR2 induces dissociation of FKBP12.6. The FKBP12 and FKBP12.6 binding sites were previously mapped to the same region in the three-dimensional structures of RyR1 and RyR2, at or near the junction of domains 3 and 9 (red in Fig. 5A and Refs. 24 and 25). In the present study, we used two independent approaches to localize two regions within the RyR2 amino acid sequence that are near the phosphorylation site in the three-dimensional structure of RyR. The GFP insertion at Tyr-2801, which is seven amino acid residues upstream of Ser-2808, is located in the bridge between domains 5 and 6 (green in Fig. 5A). The binding site for the Fab fragment derived from monoclonal antibody mAb34C, whose epitope is about 40 amino acid residues away from Ser-2808, is located in domain 6 (orange in Fig. 5A). These two sites are adjacent to one another in each of the clamp structures in the RyR cytoplasmic region, and their center to center separation is about 40 Å. The three-dimensional location of the phosphorylation site Ser-2808 is also close to a previously mapped mutation site at Ser-2367, which is located in the bridge between domains 5 and 6 (pink in Fig. 5A, Ref. 26). Considering the GFP mapping to be the most accurate determination of the position of Ser-2808 in the three-dimensional structure of RyR, we conclude that Ser-2808 is not close to the binding site of FKBP12.6. Within the homo-tetramer structure of RyR (with its four phosphorylation sites and its four FKBP12.6 binding sites), we consider the two FKBP12.6 binding sites that are nearest to the phosphorylation site, as shown in Fig. 5A, the FKBP12.6 binding site in the same clamp region is about 105 Å apart (green double-ended arrow) from the Ser-2808 site, while the other one in the adjacent clamp regions is about 120 Å away (red double-ended arrow).
In both skeletal and cardiac muscle, RyRs self-assemble into two-dimensional ordered arrays at the terminal cisternae membrane of the sarcoplasmic reticulum. One question arises: can PKA phosphorylation of a Ser-2808 in one RyR disrupt the binding of FKBP12.6 to a neighboring RyR in the arrays. Paolini et al. (40) have defined the orientation of the two-dimensional arrays of RyR, and we have superimposed our three-dimensional mapping of Ser-2808 and FKBP12.6 binding site onto their structural model (Fig. 5B). Even in two-dimensional arrays, the distance between the phosphorylation site Ser-2808 (which takes the place of Tyr-2801) and the closest FKBP12.6 binding site in the neighboring RyR exceeds 130 Å (yellow double-ended arrow).
In addition to the direct structural mappings, we have found two lines of indirect evidence to corroborate that the phosphorylation site Ser-2808 is not close to the FKBP binding site. First, in the purification of RyR2 Y2801-GFP , we used glutathione-Sepharose beads containing pre-bound GST-FKBP12.6 fusion protein as the affinity binding ligand. The successful purification indicated that the insertion of GFP near Ser-2808 did not prevent the binding of FKBP12.6 to RyR2. Second, in the coprecipitation test of mAb34C Fab fragments, we used glutathione-Sepharose beads with pre-bound GST-FKBP12 to pulldown the mAb34C Fab ϩ RyR1 complex. The positive results further indicate that the binding of FKBP12 does not interfere with the binding of mAb34C Fab fragments.
The site of the GFP insertion is only seven amino acid residues away from Ser-2808 in the primary sequence, but the rather large distance between the GFP site and the actual FKBP12.6 binding location as found by three-dimensional cryo-EM (Ͼ100 Å) is not consistent with the idea that the phosphorylation site is directly involved in the binding of FKBP12.6. Other recent studies have also raised doubts about the role of this phosphorylation site in the FKBP12.6-RyR2 interaction. A C-terminal truncation analysis demonstrated that a region between residues 1815 and 1855, near divergent region 3, is essential for FKBP12.6 interaction (41). Consistent with this finding, we used cryo-EM to map a GFP insertion at Thr-1874 (in the middle of divergent region 3) to domain 9, a position adjacent to the FKBP12.6 binding site (41). In contrast to the FKBP dissociation mechanism (8,9), recent studies have found that PKA phosphorylation of Ser-2808 in RyR2 does not result in dissociation of FKBP12.6, because FKBP12.6 can bind to both the Ser-2808 phosphorylated and non-phosphorylated forms of RyR2 (22). RyR2s with the mutation S2808D or S2808A at this conserved phosphorylation site also retained the ability to bind FKBP12.6 (21).
RyR2 is known to be phosphorylated at Ser-2808 by PKA, CaMKII, and PKG, and at Ser-2814 by CaMKII. Phosphorylation of RyR2 at these residues modulates the activity of the channel. However, little is known about the structural basis of regulation of RyR2 by phosphorylation. Our data clearly demonstrate that the phosphorylation motif that encompasses the Ser-2808 and Ser-2814 is far away from the FKBP12.6 binding site, thus excluding the possibility that the phosphorylation of RyR2 at Ser-2808 directly disrupts the binding of FKBP12.6, causing its dissociation from the channel, as proposed previously. On the other hand, it is of interest to note that the three-dimensional location of the Ser-2808 site is close to a previous mapped Ser-2367 site located in the central disease-causing mutation hot spot (pink in Fig. 5A; Ref. 26), suggesting that the effect of phosphorylation on the FIGURE 5. The PKA phosphorylation site serine 2808 is not close to the FKBP12.6 binding site. A, three-dimensional reconstruction of RyR2 is shown in blue, the site of the GFP insertion after Tyr-2801 is shown in green, and the binding site of mAb34C is shown in orange. The two locations are close to a previously mapped mutation site Ser-2367 (in pink, visible in middle panel). If the GFP site is considered to be a marker of the Ser-2808 location, the distance to the FKBP12.6 site on the same clamp structure is 105 Å (green double-headed arrow), and the distance is 120 Å to the FKBP12.6 bound to an adjacent clamp structure (red double-ended arrow). B, Ser-2808 is not close to the FKBP12.6 binding site in the nearest neighbor RyR. The three-dimensional mapping of Ser-2808 and the FKBP12.6 binding site are superimposed onto the structural model of two-dimensional ordered arrays of RyR. The distance between Ser-2808 and the FKBP12.6 in the nearest neighbor RyR is over 130 Å (yellow double-ended arrow). function of RyR2 may be similar to that of the disease-causing mutations.
There are three diseases-causing mutation hot spots in the RyR2 sequence that harbor diseases-causing mutations, region 1 is near the N terminus of the channel, from residues 77 to 466; region 2 is near the middle of the sequence, spanning residues 1,724 through 2,958; and region 3 includes part of the cytoplasmic-and membrane-associated regions of the channel and comprises residues 3,778 to 4,959. Two mutation hotspots lie in the cytoplasmic regions of RyR2, the N-terminal region and the central region, and are well separated in the primary sequence (Ͼ1,200 amino acids); however according to the hypothesis proposed by Ikemoto and Yamamoto (10), these two regions occur in structural domains that physically interact with each another in the three-dimensional space, and changes in the strength of their interaction due to disease-causing mutations can affect channel gating. More specifically, the interactions between the N-terminal and central regions (domain zipping) stabilize RyR in the closed state, and reduction in the affinity of the interaction causes separation of these regions (domain unzipping) and thereby causing RyR to open more frequently. In support of this hypothesis, we have mapped a mutation in the N-terminal region (Ser-437) into a region between domains 5 and 9 (27), and a mutation in the central region (Ser-2367) to a site between domains 5 and 6 (26). These findings suggest that the N-terminal mutations are located in domain 5, while the central mutations are in domain 6, and that there are domaindomain interactions between them to stabilize the RyR channel (27). Because the three-dimensional location of the Ser-2808 PKA phosphorylation site is located near to the domain-domain boundary, it is possible that PKA hyperphosphorylation at Ser-2808 may alter the domain-domain interactions between the N-terminal and the central regions of RyR2, and thus the gating properties of the channel.
In summary, evidence is continuing to mount against the proposal that PKA hyperphosphorylation at Ser-2808 induces dissociation of FKBP12.6 from RyR2 via a direct steric repulsion between the negatively charged phosphate group at Ser-2808 and the negatively charged residue Asp-37 of FKBP12.6. If the FKBP12.6-RyR2 interaction were weakened by PKA hyperphosphorylation, this weakening action would have to occur via an indirect mechanism, such as an allosteric effect operating over a distance of tens of Angstroms (i.e. a long-range conformational change), or alternating an effect induced by one or more RyR regulatory factors, including small molecules like Ca 2ϩ and Mg 2ϩ , and large proteins like calmodulin, PKA, CaMKII, protein phosphatase 1, protein phosphatase 2a, and DHPR.