Apocalmodulin and Ca 2 (cid:1) -Calmodulin Bind to Neighboring Locations on the Ryanodine Receptor*

Calmodulin (CaM) binds to the ryanodine receptor/ calcium release channel of skeletal muscle (RyR1), both in the absence and presence of Ca 2 (cid:1) , and regulates the activity of the channel activity by activating and inhibiting it, respectively. Using cryo-electron microscopy and three-dimensional reconstruction, we found that one apoCaM binds per RyR1 subunit along the sides of the cytoplasmic assembly of the receptor. This location is distinct from but close to the location found for Ca 2 (cid:1) CaM, providing a structural basis for efficient switching of CaM between these two positions with the oscillating intracellular Ca 2 (cid:1) concentration that generates muscle relaxation/contraction cycles. The locations of apoCaM and Ca 2 (cid:1) -CaM at a critical region for RYR1-dihydropy-ridine receptor interaction are suggestive of a direct role for CaM in the mechanism of excitation-contraction coupling. In skeletal muscle, the calcium release channel/ryanodine receptor (RyR1) one of the main regulators of cytoplasmic Ca 2 (cid:1) concentration, the signal for muscle contraction, allowing rapid translocation of Ca 2 (cid:1) from the sarcoplasmic retic-ulum to the cytoplasm the RyR1s to to be the receptors voltage with an existing three-dimensional reconstruction of RyR1 (38) as a reference. The con-tribution of 4-fold views was reduced with final numbers of particles of 2798 for RyR1 control and 2079 for RyR1 (cid:1) apoCaM. Two-dimensional averages and three-dimensional reconstructions were filtered to their limiting resolution as calculated according to the Fourier ring and shell correlation criteria, respectively, with a cutoff value of 0.5 (39). Three- dimensional reconstructions were displayed with isosurface represen-tations at a threshold that corresponds to the midpoint of the density profile of the reconstruction boundary.

In skeletal muscle, the calcium release channel/ryanodine receptor (RyR1) 1 functions as one of the main regulators of cytoplasmic Ca 2ϩ concentration, the signal for muscle contraction, by allowing rapid translocation of Ca 2ϩ from the sarcoplasmic reticulum to the cytoplasm (1)(2)(3). Some of the RyR1s are thought to contact directly and to be under the control of dihydropyridine receptors (DHPRs), voltage sensors situated on the cell membrane infoldings (T-tubules) that are juxtaposed to the sarcoplasmic reticulum. The DHPR senses the membrane depolarization wave associated with a neuronal impulse and undergoes a conformational change and intramembrane charge movement (4). This is apparently necessary and sufficient for RyR1 activation, which results in a global intracellular Ca 2ϩ increase from above pCa 7 to pCa 6 -4, a level of Ca 2ϩ that triggers contraction of the muscle fiber (5). RyR1 itself is regulated by Ca 2ϩ in a biphasic fashion with an activation peak at pCa ϳ5.5 and inhibition below pCa 3 or above 7 (6). In addition to its vital function in skeletal muscle, RyR1 is an abundant RyR isoform in cerebellar neurons (7) where it also appears to communicate with the plasma membrane DHPR (8). In the brain, RyRs have been hypothesized to play a role in synaptic plasticity and apoptosis (9).
As has been reported for some other Ca 2ϩ channels (10), RyR1 is further regulated by CaM, the main mediator of Ca 2ϩ signaling in eukaryotes. Modulation of RyR1 by CaM is complex. At pCa 4, Ca 2ϩ -saturated CaM (Ca 2ϩ -CaM) binds directly to the RyR1 with a stoichiometry of one mole of apoCaM per RyR1 subunit (i.e. four CaMs per tetrameric RyR1 complex) (11,12) and partially inhibits channel activity (6,13), but at pCa 7, Ca 2ϩ -depleted CaM (apoCaM) becomes an activator (14 -17). In both cases CaM reduces the direct effect produced by Ca 2ϩ alone acting on the RyR1. CaM also mediates RyR1's redox sensitivity to nitric oxide and oxygen species (18,19).
Other proteins are also modulated by Ca 2ϩ -CaM and apoCaM. Among these, either CaM is bound to the same sequence at both high and low [Ca 2ϩ ], or the effector protein has two or more binding sites with different requirements of [Ca 2ϩ ] for CaM binding (20). For RyR1, the number of CaM-binding sites as well as whether the two forms of CaM share common binding sites has been subject of study (12,14,15). Within RyR1's ϳ5000 amino acid residues, several putative CaM binding motifs have been proposed that could account for both Ca 2ϩ -dependent and -independent CaM binding (16,(21)(22)(23). Two recent independent reports suggest that there are two binding sites per subunit (one for apoCaM and one for Ca 2ϩ -CaM), both contained in a short segment (ϳ30 residues) of the RyR1 sequence (24,25).
Although the interactions of CaM with short peptides corresponding to the CaM-binding sites in several proteins have been characterized at atomic detail (26 -30), a structural model that includes the complete regulated protein is lacking but is necessary to understand the mechanism of action of CaM, particularly in those proteins that bind both apoCaM and Ca 2ϩ -CaM. Cryo-electron microscopy and image processing have been used to generate three-dimensional reconstructions of native isolated RyR1 (31). In this study, we have used these techniques to obtain three-dimensional reconstructions of RyR1 in the presence and absence of apoCaM. Difference mapping shows the locations where apoCaM binds to the RyR1, and incidentally, provides an independent means to the biochemical assays used previously to assess the stoichiometry of binding. The binding site for apoCaM found in this study is distinct from but close to the known location for Ca 2ϩ -CaM (11,32). We propose a model for CaM activity whereby this molecule acts as a mobile Ca 2ϩ -sensing subunit of RyR1. Furthermore, the locations of RyR1-bound CaM occur in a region implicated in interactions of RyR1 with DHPR, and on this basis we propose that CaM could be involved in the mechanism of excitationcontraction (E-C) coupling in skeletal muscle. Part of this study has been published previously in abstract form (33).

EXPERIMENTAL PROCEDURES
Preparation of RyR1⅐apoCaM Complexes-RyR1 was isolated as described previously (11). Phosphatidylcholine (PC) was purchased from * This work was supported by Grants AR 40615 and RR01219-19 from the National Institutes of Health and by a grant from the Ministry of Science and Technology of Spain (to M. S.). 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.
Avanti Polar Lipids, Inc.; brain CaM and all other chemicals were purchased from Sigma. RyR1⅐apoCaM complexes were prepared by mixing 1 volume of CaM-free RyR1 (in 10 mM Na-PIPES, pH 7.4, 0.45 M NaCl, 25 M EGTA, 20 M leupeptin, 1% CHAPS, 0.5% PC, 0.5 mM fresh dithiothreitol) and 1 volume of CaM in buffer C (20 mM Tris-HCl, pH 7.4, 0.15 M KCl, 2 mM EGTA, 20 M leupeptin) to achieve a CaM: RyR1 molar ratio of 18. The RyR1 control (no apoCaM) was also diluted with 1 volume of buffer C, and all mixtures were incubated for 20 -40 min at room temperature prior to cryo-plunging.
Cryo-electron Microscopy-PC and CHAPS concentrations were diluted to 0.09 and 0.2%, respectively, so as to allow sufficient image contrast, by adding buffer C during sample application to the electron microscopy grid (34). Grids were blotted with Whatman filter paper No. 540. Cryo-electron microscopy was performed on a Philips EM 420 transmission electron microscope operated at 100 kV under low dose conditions at a magnification of ϫ52,000, and micrographs were taken at 0°, 30°, and 50°tilt. High quality micrographs (71 for RyR1 control and 41 for RyR1⅐apoCaM) were digitized on a Hi-Scan (Eurocore) densitometer at a pixel size of 3.8 Å, which was binned down to 7.7 Å prior to three-dimensional reconstruction. Individual particles (7814 for RyR1 control and 4887 for RyR1⅐apoCaM) were selected from the micrographs and processed using the SPIDER/WEB software package (35) as described previously (36). 4-fold two-dimensional averages were generated using 900 particles. Three-dimensional reconstructions were computed using projection-matching algorithms (37) with an existing three-dimensional reconstruction of RyR1 (38) as a reference. The contribution of 4-fold views was reduced with final numbers of particles of 2798 for RyR1 control and 2079 for RyR1⅐apoCaM. Two-dimensional averages and three-dimensional reconstructions were filtered to their limiting resolution as calculated according to the Fourier ring and shell correlation criteria, respectively, with a cutoff value of 0.5 (39). Threedimensional reconstructions were displayed with isosurface representations at a threshold that corresponds to the midpoint of the density profile of the reconstruction boundary.

RESULTS
Two-dimensional Cryo-electron Microscopy-Electron micrographs of frozen hydrated specimens show the fine structure that is indicative of well preserved tetrameric RyR1s (Fig. 1A) and a high proportion of 4-fold-symmetric views, corresponding to RyR1s lying with their 4-fold symmetry axes normal to the carbon support film. Two-dimensional averages performed on such 4-fold views (resolution of 28 Å), prepared in the presence and absence of apoCaM, display the typical square shape with protruding corners (clamps) and a central cross (31,38) (Fig. 1, B and C). Direct comparison of the two two-dimensional averages reveals that in the presence of apoCaM, a small region in domain 3 (Fig. 1C, arrows) is increased in density, and a small protuberance appears on its external side. Subtraction of the RyR1 control average from that of the RyR1⅐apoCaM yields a difference map showing four discrete positive (white) densities (Fig. 1D) that are statistically significant at the 99.9% confidence level according to the t test (40). It is highly likely that these regions of excess density correspond to apoCaM bound to RyR1.
Three-dimensional Localization of apoCaM on the RyR1-The three-dimensional reconstructions of RyR1 with and without (control) apoCaM (Fig. 2, A and B) with resolutions of 28 and 26 Å, respectively, show a square prism-shaped cytoplasmic assembly with 10 distinguishable domains and a differentiated smaller transmembrane region, as was described previously (31,38,41). As anticipated from the two-dimensional analysis, the two reconstructions display a major difference at the outer face of domain 3: a concave region present in the control RyR1 reconstruction becomes protuberant when apoCaM is added (compare boxed areas in Fig. 2, A and B). Subtraction of the two volumes yields four ellipsoidal-shaped masses (Fig. 2C), each having its major axis approximately perpendicular to the plane of the square defined by the cytoplasmic assembly. We attribute each of these masses to an apoCaM molecule (i.e. one apoCaM per RyR1 subunit).
Consistent with this interpretation, we found that an atomic model of a peptide⅐apoCaM complex (42), following low pass filtration to the resolution of our three-dimensional reconstructions, closely matches in shape and size the ellipsoidal-shaped masses that form the difference map (not shown). To check the internal consistency of our results, the three-dimensional reconstructions and three-dimensional difference map were projected onto a plane normal to the 4-fold symmetry axis of the RyR1 to simulate the results of the two-dimensional analysis. These computed projections (Fig. 1, E-G) reproduce the main features of the two-dimensional analysis (Fig. 1, B-D).
An alternative but unlikely interpretation of the structural differences between RyR1 and RyR1⅐apoCaM is that they are due to conformational changes in RyR1 induced by apoCaM binding to RyR1. However, conformational differences should result in comparable levels of both positive and negative densities in difference images, whereas the three-dimensional difference map obtained from the RyR1⅐apoCaM and RyR1 threedimensional reconstructions does not display any significant negative density. Therefore, we conclude that the four additional ellipsoidal masses arise from four molecules of apoCaM bound to the surface of RyR1. This does not preclude the possibility of conformational changes at a smaller scale, which might not produce detectable differences at the level of resolution attained.
Relative Positioning of apoCaM and Ca 2ϩ -CaM on the RyR1-We compared the location of apoCaM to the location already found for Ca 2ϩ -CaM (11). Both forms of CaM contact domain 3 of each of the four RyR1 subunits. However, depending on the Ca 2ϩ concentration, CaM locates at distinct sites on the RyR1. ApoCaM is situated closer to the T-tubule face of RyR1 and protrudes more from the lateral surface of the structure, whereas Ca 2ϩ -CaM appears in each of the crevices formed by domains 3 and 6/8 (Fig. 3, A and B). The major axes of the Ca 2ϩ -CaM and apoCaM molecules form an angle of ϳ140° (Fig.  3C), and their geometric centers are 33 Ϯ 5 Å apart. If bound simultaneously, the volumes of Ca 2ϩ -CaM and apoCaM would interpenetrate partially (Fig. 3D). Therefore, simultaneous occupation of both sites seems unlikely.

DISCUSSION
Correlation with Biochemical Results-ApoCaM and Ca 2ϩ -CaM, which differ substantially in their structural and physicochemical properties as well as in their biological effects, have been spatially localized on the intact RyR1 tetrameric complex (summarized in Fig. 3, A and B). The two forms of CaM bind on the sides of the cytoplasmic assembly of RyR1 to distinct, overlapping locations with centers separated by 33 Ϯ 5 Å. The two sites are sufficiently close, and the orientations of the bound CaMs are such that simultaneous binding of apoCaM and Ca 2ϩ -CaM would appear to be unlikely due to steric interference. This is further evidence that only four CaM molecules could be bound to the RyR1 at a given time, regardless of the Ca 2ϩ concentration.
Although nine potential binding sites for CaM have been identified within the sequence of the RyR1 polypeptide (21)(22)(23)43), conflicting reports have appeared regarding the actual number of CaM-binding sites (12, 14, 15). Our findings, based upon three-dimensional structural determination, agree with the results of biochemical assays (12), which show that each Bound apoCaM or Ca 2ϩ -CaM protects the same two peptide bonds (between residues 3630 -3631 and 3637-3638) from proteolysis by trypsin (12), and further, a synthetic peptide containing these residues of the RyR1 subunit (amino acids 3614 -3643) was found to be capable of binding with high affinity either apoCaM or Ca 2ϩ -CaM (24). Additional experiments employing other related peptides, together with experiments using single point RyR1 mutants (25), were interpreted as supporting a model in which apoCaM and Ca 2ϩ -CaM bind to distinct but overlapping sites on the peptide. Our spatial localizations of the two forms of CaM are compatible with such results, although the 33 Ϯ 5 Å distance between the centers of the sites that we found is somewhat greater than might have been expected.
We propose several modes of CaM binding that could account for the ϳ33-Å separation between the centers of apoCaM and Ca 2ϩ -CaM while both forms of CaM still preserve interactions with amino acid residues contained within a short segment of the RyR1 sequence (Fig. 4). (i) For instance, the apoCaM and Ca 2ϩ -CaM sites could be related by a translation of the CaM along the linear sequence of the RyR1 such that the "protected region" is bound by or in the immediate neighborhood of one or the other of the two major domains of CaM (Fig. 4A). (ii) The domain of CaM not interacting with the protected sequence could even bind to a non-contiguous region of the RyR1 sequence (Fig. 4B), a situation that is reminiscent of a recently characterized mode of CaM binding to a target protein (30). (iii) The translation of CaM could involve a change in the polarity of the amino and carboxyl domains of CaM with respect to the target peptide (Fig. 4C). Although CaM interacts predominantly with its target peptides in an antiparallel fashion with the amino terminus of the peptide interacting with the C-lobe of Ca 2ϩ -CaM, the opposite orientation has also been described (29,44). (iv) Alternatively, conformational differences between RyR1⅐apoCaM and RyR1⅐Ca 2ϩ -CaM could contribute to the separation of CaM-binding sites at high and low [Ca 2ϩ ] (Fig. 4D).
RyR1 itself is sensitive to Ca 2ϩ , showing a bell-shaped activation curve with a maximum at pCa 4. Interestingly, CaM bound to RyR1 counteracts this effect partially, inhibiting the channel at millimolar Ca 2ϩ concentrations and activating it at submicromolar Ca 2ϩ concentrations. Because both CaM and RyR1 are affected by Ca 2ϩ , the RyR1-CaM interactions need to be interpreted in this context. We explored the possibility of a Ca 2ϩ -induced conformational change as a cause for the different locations of CaM at high and low [Ca 2ϩ ] by comparing three-dimensional reconstructions of RyR1 without added CaM at pCa Ͼ6 and at pCa 4 (11,36). Although there are some minor conformational differences in the vicinity of the CaM-binding regions, the relatively large distance between apoCaM and Ca 2ϩ -CaM is unlikely to be solely the result of a conformational change induced by Ca 2ϩ .
The target surface for both forms of CaM is situated ϳ10 nm away from the Ca 2ϩ -conducting pore contained in the transmembrane region, suggesting that there is long range allosteric coupling between domain 3 and the transmembrane domain. Similar results have been obtained with other RyR1 modulators, such as FK-506 binding protein and Imperatoxin A (IpTx a ) (11,36).
A Switching Mechanism for CaM-Tripathy et al. (15) found that the kinetics of apoCaM and Ca 2ϩ -CaM binding to RyR1enriched vesicles were sufficiently slow so as to preclude a mechanism of modulation by CaM in which it dissociates from one site and reassociates with the other with each cycle of contraction. To account for this result, they postulated that one of the CaM-binding sites present on RyR1 was occupied by both apoCaM and Ca 2ϩ -CaM. Our findings, as well as those of Rodney et al. (24) and Yamaguchi et al. (25), argue strongly for two distinct sites for apoCaM and Ca 2ϩ -CaM. Thus, it would appear that if the results of the kinetic studies and the evidence showing distinct sites for the two forms of CaM apply to the in vivo situation, then a mechanism for switching between the two sites must be non-dissociative in nature.
Based on the relatively close positioning of apoCaM and Ca 2ϩ -CaM reported in this study, we propose a model whereby CaM switches between these two positions with the oscillating [Ca 2ϩ ] responsible for the cycles of contraction-relaxation in skeletal muscle without dissociation from the RyR1. The necessity to capture another CaM molecule from the cytoplasmic pool would be avoided, allowing a fast response as has been reported (13). Given the intracellular CaM concentration (2 M in skeletal muscle (45) and 50 M in brain (1)) and the high affinity of RyR1 for apoCaM and Ca 2ϩ -CaM (K d values both in the nanomolar range (14,15)), RyR1 could exist primarily as a complex with CaM, even at resting Ca 2ϩ levels. Tripathy et al. (15) also suggested this scenario based upon the slow kinetics with which CaM dissociates from RyR1 relative to contraction/ relaxation cycles. Recent evidence supports the preassociation of apoCaM with voltage-gated Ca 2ϩ channels that are regu- FIG. 4. Four schematic models for CaM switching between the apoCaM and Ca 2ϩ -CaM locations. The two distinct locations of apoCaM and Ca 2ϩ -CaM can be accounted by either two distinct target sequences on the RyR1 with differing affinities for the two forms of CaM, a Ca 2ϩ -induced conformational change on the RyR1 that moves a unique CaM-binding site, or a combination of both. The relative positions for the two CaM domains and the target peptide at high and low Ca 2ϩ conditions are indicated for several models: translation to contiguous (A) or non-contiguous (B) sites, translation involving change in polarity of CaM relative to the target peptide (C), and Ca 2ϩ -induced conformational change of RyR1 (D). A conformational change as depicted in panel D could accompany the mechanisms shown in panels A-C. The vertical bar indicates the RyR1 peptide(s) containing the targets for CaM. The region containing two trypsin cleavage sites (residues 3630 -3638), which is protected from digestion by both forms of CaM (12), is shown in lighter gray. The circles in different shades of gray (one of them in semi-transparent display) represent the two domains of CaM. In panel B, the second vertical bar represents a region of RyR1 that is nearby spatially (but non-contiguous in sequence) to the segment containing residues 3630 -3638. lated by Ca 2ϩ -CaM, including the L-type channel (DHPR) that interacts with RyR1 at triad junctions (47). We propose that for pCaϾ6, apoCaM is bound to the apoCaM-binding site of RyR1. After Ca 2ϩ release, CaM would lose affinity for the apoCaM site and increase affinity for the neighboring Ca 2ϩ -CaM site of RyR1. Several characteristics of CaM are consistent with a non-dissociative movement of a single CaM molecule between the two CaM binding sites: two structural and functional modules capable of binding to their targets independently and with different affinities (46), the ability to form stable complexes with only one domain bound (28), and the high flexibility of the tether region that connects the two modules (26). These properties could contribute to a scenario whereby the Ca 2ϩ -induced translocation of CaM (e.g. such as in the models depicted in Fig.  4 A-C) is achieved by a two-step process with movement of one domain of CaM at a time. Thus, CaM could act as a mobile Ca 2ϩ -sensing subunit of RyR1 producing two alternative, opposite modulatory effects. Such coordinated CaM interactions could have a parallel in other proteins that bind apoCaM and Ca 2ϩ -CaM (20,47,48).
Does CaM Have a Role in E-C Coupling?-The currently favored mechanism of E-C coupling in skeletal muscle, first proposed by Schneider and Chandler (49), posits a "mechanical coupling" between RyR1 and the voltage sensor, identified as the DHPR (4). According to this mechanism, the DHPR undergoes a voltage-dependent conformational change that in turn induces a conformational change in RyR1 through a physical coupling of the two receptors (for reviews, see Refs. 50 -52). A cytoplasmic region of the DHPR known as the II-III loop has been strongly implicated in the interaction with the RyR1 that mediates E-C coupling (53). Previously we mapped the binding location on RyR1 of IpTx a , a peptide that mimics a portion of the II-III loop of DHPR but binds with much higher affinity to RyR1 than the II-III loop itself (54). IpTx a was found to bind within the crevices formed by RyR1's domains 3 and 6/8 (36), which are near to and perhaps overlapping the binding location Ca 2ϩ -CaM (Fig. 3, E and F). Most likely, additional regions of the RyR1, particularly within domain 3 and the clamps (domains 5-10), are involved in the interaction with DHPR (31,52,55,56). The existence of two closely spaced but distinct sites for apoCaM and Ca 2ϩ -CaM on RyR1, together with the proximity of these sites to a region of RyR1 that potentially interacts with DHPR, leads us to speculate that CaM might play a more direct role in the molecular mechanism of E-C coupling than previously considered (6,13,15,57).