Three-dimensional Structure of Ryanodine Receptor Isoform Three in Two Conformational States as Visualized by Cryo-electron Microscopy*

Using cryo-electron microscopy and single particle image processing techniques, we present the first three-dimensional reconstructions of isoform 3 of the ryanodine receptor/calcium release channel (RyR3). Reconstructions were carried out on images obtained from a purified, detergent-solubilized receptor for two different buffer conditions, which were expected to favor open and closed functional states of the channel. As for the heart (RyR2) and skeletal muscle (RyR1) receptor isoforms, RyR3 is a homotetrameric complex compris-ing two main components, a multidomain cytoplasmic assembly and a smaller ( ; 20% of the total mass) transmembrane region. Although the isoforms show structural similarities, consistent with the ; 70% overall sequence identity of the isoforms, detailed comparisons of RyR3 with RyR1 showed one region of highly significant difference between them. This difference indicated additional mass present in RyR1, and it likely corresponds to a region of the RyR1 sequence (residues 1303–1406, known as diversity region 2) that is absent from RyR3. The reconstructions of RyR3 determined under “open” and “closed” conditions were similar to each other in overall architecture. A difference map computed between the two reconstructions reveals subtle changes in conformation at several widely dispersed locations in the receptor, the most prominent of which

Intracellular calcium is essential for the regulation and function of many fundamental biological processes, and the release of calcium from intracellular storage compartments plays an essential role in modulating cytoplasmic Ca 2ϩ levels (1). Ryanodine receptors (RyRs) 1 and inositol 1,4,5-trisphosphate receptors are the two families of intracellular calcium release channels that have been characterized to date (reviewed in Refs. [2][3][4][5][6][7][8][9]. Both RyRs and inositol 1,4,5-trisphosphate receptors exist as homotetrameric protein complexes composed of an unusually large subunit, Ϸ560 kDa for RyRs and Ϸ313 kDa for inositol 1,4,5-trisphosphate receptors. Frequently an Ϸ12-kDa isoform of FK506-binding protein (FKBP, an immunophilin) binds with sufficient affinity to the receptors so as to co-purify with them, and the immunophilin may therefore be considered as a component of the receptors (10 -12).
Three different isoforms of RyRs, each encoded by a different gene, have been identified in mammals, designated as skeletal (RyR1), cardiac (RyR2), and brain (RyR3). Recent studies have shown that the receptors are much more widely expressed than is indicated by the nomenclature (13)(14)(15). For instance, all three isoforms are expressed in brain, but with distributions unique to each. RyR3, the subject of this study, is present at low levels in a variety of excitable and nonexcitable cells, the richest source being diaphragm muscle where it is co-expressed at about 2.5% of the level of RyR1 (16).
RyR1 and RyR2 are sufficiently enriched in skeletal and heart muscle, respectively, to have allowed their purification in sufficient quantities for biochemical and biophysical characterization (17)(18). RyR1 and RyR2 are key components of the multicomponent complexes that are responsible for excitationcontraction coupling in striated muscle (6). Our understanding of RyR3 has lagged behind that of the other isoforms because of difficulty in isolating sufficient quantities, and only recently have protocols been developed that allow commencement of its characterization (16, 19 -22). These studies have shown that RyR3 has similar, but not identical, conductance and pharmacological properties to those found for the other isoforms. For example, RyR3 is activated by Ca 2ϩ (micromolar) and by millimolar ATP and is inhibited by millimolar Mg 2ϩ and by Ca 2ϩ above millimolar levels. All of the isoforms bind with high affinity the plant alkaloid, ryanodine, which locks the receptors into similar subconductance states. As isolated, RyR1 and RyR2 contain FKBP12 and FKBP12.6, respectively. RyR1 and RyR3 are capable of binding both FKBP isoforms, whereas RyR2 binds only FKBP12.6. Purified RyR2 and RyR3 differ from RyR1 by their lack of modulation of channel activity by FKBP12 and FKBP12.6 (12,22,23). Despite the similarities of the isolated RyR isoforms, they are not functionally interchangeable in vivo, particularly with regard to their roles in e-c coupling (24 -28).
Comparison of the amino acid sequences of ryanodine receptor isoforms reveals three regions that are much more variable than the overall Ϸ70% identity among the isoforms (29). These hypervariable regions are thought to partially account for the functional differences among the isoforms. Indeed, two of these regions, termed D2 (residues 1303-1406 of RyR1) and D3 (res-* This work was supported by National Institutes of Health Grants HL32711 (to S. F.) and AR40615 (to T. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Three-dimensional structural information is essential to understanding how ryanodine receptors function but is difficult to obtain by X-crystallography owing to their large size and complexity, and because integral membrane proteins are inherently difficult to crystallize. Cryo-electron microscopy of purified, noncrystalline RyRs combined with three-dimensional image reconstruction offers a practical alternative to characterizing the three-dimensional structures of RyRs, albeit at less than atomic levels of resolution (34 -36). Here we describe the first such reconstructions of RyR3, obtained under conditions that favor two different structural configurations of the receptor. Comparisons with previous reconstructions of the other two isoforms show that all three have a highly conserved threedimensional organization in which about 80% of the protein mass forms a cytoplasmic assembly containing at least 10 distinguishable domains. One significant difference in morphology between RyR3 and RyR1 has been detected, and it likely corresponds to one of the regions of the sequence, the D2 region, that is hypervariable among the isoforms and is missing.

MATERIALS AND METHODS
Preparation of RyR3-RyR3 was obtained from bovine diaphragm sarcoplasmic reticulum as described previously elsewhere (16). RyR3 was immunoprecipitated from a CHAPS-solubilized fraction and eluted with the peptide epitope recognized by the immobilized antibody. The final elution of RyR3 was obtained in buffer A (20 mM PIPES (pH 7.2), 0.4 M NaCl, 5 mM Na 2 AMP, 0.1 mM EGTA, 0.2 mM CaCl 2 , 2.0 mM dithiothreitol, 0.6% CHAPS (w/v), and 0.3% soybean phospholipid, 1 g/ml leupeptin. The channels were stored in elution buffer (buffer A) that favors the open state of the channel (16).
Preparation of Grids for Cryo-electron Microscopy-Grids of frozen hydrated RyR3 were prepared as described previously for RyR2 (36), in which the final buffer conditions were attained by diluting the RyR3 (in buffer A) into the buffer of choice directly on the grid followed by rinsing the grid. The grid dilution buffers were similar to the above elution buffer A. Two buffer conditions were chosen, one which contained the channel activators, Ca 2ϩ and Na 2 AMP, and the other which contained no activators. We will refer to these two conditions as open and closed conditions, respectively.
To prepare open condition grids, 1.0 l of RyR3 (in buffer A) was injected into 4.0 l of buffer B (same as buffer A except containing no phospholipids and 1% CHAPS), which had been placed on the specimen grid. After a few seconds the grid was rinsed by inverting it and touching it to a droplet (50.0 l) of buffer C (same as buffer B except concentration of CHAPS was 0.5%).
Closed condition grids were prepared by injecting 0.5 l of RyR3 (in buffer A) into 4.5 l of buffer D (20.0 mM PIPES (pH 7.2), 0.4 M NaCl, 0.1 mM EGTA, 2.0 mM dithiothreitol, 1.0% CHAPS (w/v), and 1.0 g/ml leupeptin). The grid was rinsed as above except using buffer E (same as buffer D except CHAPS concentration was reduced to 0.5% (w/v)). We have found that RyR3 is structurally unstable when stored under closed conditions, and therefore it was important to dilute and rinse the grids as quickly as possible with closed condition buffer.
Cryo-electron Microscopy and Image Processing-A Philips EM 420 transmission electron microscope operated at 100 kV and equipped with a low dose kit and a GATAN model (model 626) cryo-transfer holder was used for cryo-electron microscopy. The net electron dose for each micrograph is estimated to be 10 electrons/Å 2 . The magnification was 38,600 as determined using tobacco mosaic virus as a calibration standard. Micrographs were recorded with the objective lens under focused by 2.0 -2.5 m. To obtain a sufficient number of orientations of RyR3 for three-dimensional reconstruction, it was also necessary to collect micrographs with the grid tilted by 30°. To minimize radiation exposure only one micrograph was recorded for each area of the specimen grid (with the grid either tilted or not tilted). Cryo-electron microscopy has been described in detail elsewhere (36). The selected micrographs were digitized using a Perkin-Elmer PDS 1010 A microdensitometer with a 20-m step size corresponding to the 5.33 Å on the object scale. Image processing was performed using the SPIDER software system (37) and employing a projection matching procedure for determining three-dimensional reconstructions as described previously (36,38). Resolutions were estimated by the Fourier shell correlation method using a cutoff value of 0.5 (see appendix to Ref. 39). The three-dimensional structure of RyR1 used for purposes of comparison in the present study was obtained by redetermining a reconstruction by the projection matching algorithm from the same data processed previously by the random conical tilt method (34). The resolution obtained for the new reconstruction was 34 Å. The number of particles used for three-dimensional reconstruction of RyR3 in the open buffer was 3493, and 1930 was used for the closed buffer reconstruction. The resolutions were estimated at 34 Å for the open and 40 Å for the closed reconstructions. The difference in resolution can be accounted for by the better contrast of micrographs for RyR3 in open buffer than in closed buffer. For comparison the open versus closed reconstructions were filtered to a common resolution of 40 Å. The reconstruction was repeated for RyR3 in open buffer with the same number of particles, and the same structure was reproduced.

RESULTS
Cryo-electron Microscopy of RyR3-RyR3 was purified by the procedure we described previously (16) in which the isolation buffer contains 0.1 mM Ca 2ϩ and 5 mM Na 2 AMP, conditions that were shown to favor the open state of the receptors when they are incorporated into lipid bilayers. Because one of our objectives was to compare the three-dimensional architecture of RyR3 with the structures of the other isoforms, which we determined previously under conditions that favor the closed state of the receptors (34,36), it was desirable to exchange the elution buffer for one that contained nanomolar levels of Ca 2ϩ and no nucleotide. We found that RyR3 was unstable when the Ca 2ϩ and nucleotide concentrations were reduced (as judged by the overall appearance of the receptor images in micrographs), and therefore it was necessary to rapidly change the buffer on the specimen grid itself and to freeze the grids within a few seconds (see "Materials and Methods"). Fig. 1, a and b, shows typical micrographs of frozen hydrated RyR3 obtained under the open and closed buffer conditions. No differences are apparent by direct inspection of the two micrographs ( Fig. 1, a and b), which is as expected because large changes in the structure are unlikely (40) and also because of the low signal-to-noise ratio of the images. One notable difference from comparable micrographs of the skeletal muscle isoform of the receptor (34) is that orientations other than the 4-fold symmetric one occur much more frequently (arrows in Fig. 1), obviating the need to tilt the specimen to high tilt angles for the determination of three-dimensional reconstructions.
Two-dimensional averaged images from cryo-microscopy of the three RyR isoforms in the 4-fold symmetric orientation have been described previously (16,34,36). Comparison of these averaged images indicated that RyR3 (under closed conditions) differed from the other isoforms principally by lacking mass density near the corners ("clamps" that refer to domains numbered 5-10 in the three-dimensional structure of RyR) of the square-shaped cytoplasmic assembly (Fig. 2, b and c). These differences were far greater than any other difference, and they are Ͼ98% statistically significant at Ͼ98% confidence level (Fig. 2d). As shown below, the three-dimensional reconstruction of RyR3 confirms this result and reveals more precisely the nature of this region of difference.
Three-dimensional Reconstruction of RyR3 in Buffers Favoring Open and Closed Conformational States of RyR3-Two three-dimensional structures of RyR3 were obtained, one under buffer conditions that favor the open state (Fig. 3a) of the receptor and the other conditions that favor the closed state (Fig. 3b). We will henceforth refer to these as open and closed RyR3 (see "Discussion").
The reconstructed closed and open three-dimensional reconstructions of RyR3 (Fig. 3) appear similar to each other as well as to previously published reconstructions of RyR1 (34) and RyR2 (36). The receptor comprises two major substructures: a large cytoplasmic assembly that is constructed from 10 or more distinct domains (indicated by the numerals in Fig. 3) and a smaller (Ϸ20% of total mass) base plate or transmembrane assembly attached to one of the overall square-shaped faces of the cytoplasmic assembly. The domains of the cytoplasmic assembly are arranged so as to form a complex surface topography containing several well defined, solvent-filled holes and deep crevices.
The three-dimensional reconstructions of open and closed RyR3 also show some structural differences. The most apparent of these occur in the transmembrane assembly, which in the open receptor appears to be larger and to be rotated counterclockwise by ϳ4°relative to its position in the closed receptor (as viewed from the face that interacts with the junctional face membrane of the sarcoplasmic reticulum (SR), cf. Fig. 3  (middle panels)). Other, more subtle differences occur in the cytoplasmic assembly; these are more clearly appreciated in a composite representation in which the closed RyR3 reconstruction is represented as a solid body (shown in shaded white color in Fig. 4A), and the open RyR3 is superimposed in transparent light blue. Two of the domains, numbered 6 and 10, that form the clamps (the assemblage of domains (numbers 5-10) at the corners of the cytoplasmic assembly) appear to be more elongated (Ϸ15 Å increase in height) when the RyR is viewed from the side (Fig. 4A, right panel). Domain 3, which connects adjacent clamps along the edges of the cytoplasmic assembly, appears to be slightly more extended in the open RyR3. The separation between domains 5 and 6 on the cytoplasmic facing (top) of the receptor is better defined in the open as compared with the closed RyR3 in which the two domains appear connected to one another (Fig. 3, left panels). Some of these structural changes are similar to results that have been reported for the skeletal isoform of the receptor, RyR1 (40, 42) (see "Discussion").
Sometimes comparisons of surface representations, such as are shown in Figs. 3 and 4, can give a misleading impression of the size and significance of differences between two independently determined reconstructions. Therefore, a more detailed comparative analysis of the RyR3 reconstructions is presented in Fig. 5, which shows difference maps obtained from the open and closed reconstructions superimposed on the reconstructions so that the locations and spatial extents of the most significant differences can be appreciated (i.e. those with the largest densities shown in red in Fig. 5). The differences shown in Fig. 5 support the interpretations made in the preceding paragraph. Reassuringly, nearly all of the differences occur on the surface of the receptor, rather than internally or outside of its boundaries, further indicating that they are well above the noise level.
Further clues to understanding the structural rearrangements that are responsible for the apparent enlargement of the transmembrane assembly in the open RyR3 can be inferred by viewing the reconstructions at a higher threshold level, as in Fig. 4B. Comparison of the two panels in Fig. 4B suggests that the there is a shift in density near the distal (lumen-facing) end of the transmembrane region away from the center (i.e. the symmetry axis) and toward the periphery in the open form of the receptor (Fig. 4B, left panel) relative to the closed form (right panel). It seems that the main globular regions of the transmembrane region, that become visible at the higher density threshold, may splay outwards in going from the closed to the open states.
Comparison of RyR3 and RyR1 Three-dimensional Structures-We have already remarked on the strong similarity of the three-dimensional architecture of RyR3 to those of the other isoforms characterized previously, and this similarity is consistent with the high degree of sequence identity among the three isoforms. Nevertheless, the isoforms are not functionally equivalent, and therefore, the structural differences are key to understanding the molecular basis of the differing functional roles of the receptors. Here we focus on comparing the structure of RyR3 with that of RyR1, the best characterized of the isoforms.
As discussed earlier, an initial comparison of averaged projection (two-dimensional) images of the two isoforms in the square-shaped, 4-fold symmetric orientation indicated that the most significant structural differences were located in the clamps at a region formed by domains 6 and 8. We therefore re-computed a three-dimensional reconstruction of RyR1 that was described previously using the same image processing methodology as was used for RyR3 (see "Materials and Methods"). The difference map for the three-dimensional volumes of RyR3 and RyR1 (both in the closed state) agrees with the two-dimensional results by showing a region of excess mass that is present in RyR1 but absent from RyR3. This difference is located on a portion of the surface of domain 6 that forms the exterior surface of the clamp assembly, and is shown superimposed (yellow region) on the solid body representation of RyR3 in Fig. 6. The density threshold used to display the difference is approximately equal to that used to display the surface of RyR3. The region of difference has the overall shape of a prolate ellipsoid of Ϸ35 ϫ 20 Å, a size that is consistent with an additional protein mass of Ϸ100 amino acids. Quite likely this region of difference between RyR1 and RyR3 corresponds to one of the highly divergent regions of the RyR amino acid sequence (the D2 region, residues 1303-1406 in RyR1) that is completely absent from RyR3. Another, much smaller region of excess mass in RyR1 is also present on the surface of domain 8 on the junctional sarcoplasmic-facing surface of the receptor (Fig. 6, middle panel).

DISCUSSION
Three-dimensional reconstructions of the RyR3 isoform in two different configurational states have been determined. The availability of detergent-solubilized RyR3 from bovine diaphragm terminal cisternae in sufficient amounts made these studies possible (16). Three-dimensional reconstructions in the 30 -40 Å resolution range are now available for all three RyR isoforms (34 -36).

Open versus Closed RyR3 Reconstructions
Reconstructions were determined for RyR3 under buffer conditions that should favor open and closed states of the receptor based upon conductance measurements on purified receptors following their incorporation into lipid bilayers (16,19,20,22). However, we cannot be certain that the solubilized RyR3 used for cryo-microscopy behaves exactly as do the receptors in a bilayer environment. Evidence that strongly suggests that solubilized receptors do indeed mimic reconstituted receptors when exposed to modulatory ligands comes from ryanodine binding experiments, which have been done using solubilized RyR3. Ryanodine is known to preferentially bind to the open form of the receptor over the closed form (41), and for solubilized RyR3, ryanodine binding is modulated by known receptor activators and inhibitors in much the same manner as channel gating in bilayer experiments (19,22).
It is reassuring that many of the differences that we found between the reconstructions of open and closed RyR3 are similar to changes described recently by another laboratory for closed and open RyR1 (40,42). The differences between open and closed RyR3 that are similar to those reported for RyR1 include the following.
Transmembrane Assembly-For RyR3 we found a 4°rotation of the transmembrane assembly in the open relative to the closed state when viewed from the SR junctional face (Fig. 3 (middle panels) and Fig. 4A (left panel)). For RyR1, a rotation of the same magnitude and handedness was seen for the ryanodine-induced state (40) but not when natural activators (nucleotide and Ca 2ϩ ) were used. We suggest that this rotation, as well as the apparent expansion in the overall size of the transmembrane region that we find for the open form of RyR3 (Fig.  3), should not be interpreted literally, rather, they result from underlying changes in the relative orientations of the transmembrane domains that are contributed by each of the four receptor subunits. At the limited resolution of the current three-dimensional reconstructions, these structural rearrangements themselves are not resolved but may still manifest themselves as visible changes, such as apparent rotations or changes in shape of the transmembrane region. Support for this interpretation is presented in Fig. 4B (and for RyR1 in the study by Serysheva et al. (42)) in which a high density threshold has been used to display the reconstructions of open and closed receptors. There it appears that the four transmembrane domains change their orientation with respect to the symmetry axis of the receptor so that regions at the distal end of the transmembrane assembly move away from the symmetry axis, perhaps thereby creating a pathway for ions to travel when the receptor is in its native environment. A similar, but not identical, interpretation has been described for RyR1 (42).
Cytoplasmic Assembly-For RyR1 a pronounced weakening or separation of the apparent connection between domains 9 and 10 in the clamp assemblies that form the corners of the cytoplasmic assembly was observed (40,42). For RyR3 we do not observe a distinct separation of these domains, although there is a slight weakening, of uncertain significance, of the connection between them (cf. left panels of Figs. 3, a and b, and  5). Perhaps this difference between the isoforms is related to the unique role that RyR1 plays in excitation-contraction coupling in mature skeletal muscle. In situ, RyR1 is thought to interact with the voltage sensing dihydropyridine receptors via the clamp regions of its cytoplasmic assembly (discussed in the following section), but RyR3 apparently cannot undergo this interaction.
Another change in the clamp regions of the cytoplasmic assembly is in the disposition of domain 6, which makes the cytoplasmic assembly appear somewhat thicker in the direction parallel to the 4-fold symmetry axis (Fig. 3, right panels) and reduces the apparent separation between domains 5 and 6 when the receptor is viewed onto the cytoplasmic face (cf. Fig.  3, left panels). Again, quite similar changes were reported for RyR1.
The remarkable similarities of RyR1 and RyR3 with regard to structural changes associated with the open and closed conditions are strong evidence that these changes are not artifactual and that they indeed correspond to authentic perturbations of the structure associated with receptor gating. An intriguing aspect of the structural differences between the open and closed states of both RyR1 and RyR3 is their global nature. The changes associated with the clamp assemblies are over 100 Å from the likely location of the channel gating mechanism in the transmembrane assembly. Previous work from our labora-tory has shown that proteins that are known to modulate channel gating of RyR1 (calmodulin, FK506-binding protein, and imperatoxin A) also bind far from the transmembrane region (10,43). Thus, very long range structural rearrangements, presumably underlying allosteric communication between functional sites, seem to be a general feature of ryanodine receptors.

Comparison of RyR3 and RyR1
Three-dimensional Architecture Overall RyR3 and RyR1 are nearly identical in structure at the resolutions attained in the three-dimensional reconstructions reported thus far. Analysis of the sequences has shown that one of the three hypervariable regions, named D2 (residues 1303-1406 in RyR1) by Sorrentino and Volpe (29) 6. Comparison of RyR3 and RyR1 reconstructions. A difference map was determined by subtracting the three-dimensional map of RyR3 from RyR1, obtained under similar closed buffer conditions. The difference map was density-thresholded so that only differences comparable in magnitude to the density level chosen to represent the three-dimensional reconstructions as solid bodies remained. These differences (yellow regions) are shown superimposed on the solid body representations of the RyR3 reconstruction (from left to right: cytoplasmic face, junctional face, and side view), and they correspond to excess mass that is present in RyR1 but absent in RyR3. Scale bar, 100 Å. TA, transmembrane assembly. completely absent from the RyR3 sequence. A deletion of this magnitude should be easily detectable in a comparison of the RyR1 and RyR3 reconstructions, assuming that in RyR1 the D2 region forms a rigid, globular fold and that other perturbations of the surrounding structure are minimal. Indeed, when we compared the structures of RyR1 and RyR3, which were determined under similar solution conditions and with identical methodology, a single region of excess mass present in RyR1 dominated all other differences (Fig. 6). Furthermore, the overall size of the region (Ϸ35 ϫ 20 Å) was reasonable for a segment containing Ϸ100 amino acid residues. The region is located in domain 6 along each of the exterior edges of the receptor.
Further support for our proposal that the major difference between the reconstructions of RyR1 and RyR3 represents the D2 region comes from the immuno-electron microscopy study by Murayama et al. (22), who visualized immunocomplexes of RyR1 and a D2-specific antibody in negative stain. They observed the antibody bound to the clamp regions of the receptor, in agreement with our interpretation of the major difference between the RyR1 and RyR3 reconstructions. Furthermore, we have analyzed the images described by Murayama et al. (22) in our own laboratory by digitizing them and then applying limited image averaging to them (data not shown), and we conclude that the D2-specific antibodies are bound along the edges of the receptor in the vicinity of domain 6, as we would predict based upon the findings described here. However, a threedimensional reconstruction of immunocomplexes containing a D2-specific antibody is required to prove that the binding site corresponds exactly to the region that we contend contains the D2 region, and we are currently working toward this goal.
The D2 region is currently of interest because of its perceived importance in excitation-contraction coupling in skeletal and cardiac muscle (27). A region nearby in the sequence involving residues 1076 -1122 appears also to be important and to play a role in the interaction of RyR1 with the voltage sensors/dihydropyridine receptors (32,33,44). Based largely upon freezefracture studies of skeletal muscle sarcolemma/transverse tubules and SR, Franzini-Armstrong and co-workers (46) have proposed a structural model for the triad junction, regions where the transverse tubules and SR form junctions to carry out excitation-contraction coupling. In this model the dihydropyridine receptors are arranged in groups of four (tetrads), which appear to be in register with the tetrameric RyR1 s in the opposing junctional SR (6,(45)(46)(47). Accordingly, it appears that each dihydropyridine receptor within a tetrad would interact with one of the clamp assemblies in the adjoining RyR1 (39,48,49). Thus, the location that we have assigned to the D2 region in the three-dimensional structure of RyR appears to be consistent with current models of the triad junction's architecture.
In conclusion, we have established that RyR3 is amenable to study by three-dimensional cryo-microscopy, and that RyR3, although similar in its overall three-dimensional architecture to the other RyR isoforms, does reveal at least one significant difference that is attributed provisionally to the D2 region of the amino acid sequence of the receptor. We also observed several structural differences at diverse locations between two conformational states of RyR3 that likely correspond to, or are related to, open and closed states of the receptor. Future studies at higher resolution will certainly resolve additional structural differences between the isoforms and clarify the structural basis of channel gating.