Crystal structures of ryanodine receptor SPRY1 and tandem-repeat domains reveal a critical FKBP12 binding determinant

Ryanodine receptors (RyRs) form calcium release channels located in the membranes of the sarcoplasmic and endoplasmic reticulum. RyRs play a major role in excitation-contraction coupling and other Ca2+-dependent signalling events, and consist of several globular domains that together form a large assembly. Here we describe the crystal structures of the SPRY1 and tandem-repeat domains at 1.2–1.5 Å resolution, which reveal several structural elements not detected in recent cryo-EM reconstructions of RyRs. The cryo-EM studies disagree on the position of SPRY domains, which had been proposed based on homology modelling. Computational docking of the crystal structures, combined with FRET studies, show that the SPRY1 domain is located next to FK506-binding protein (FKBP). Molecular dynamics flexible fitting and mutagenesis experiments suggest a hydrophobic cluster within SPRY1 that is crucial for FKBP binding. A RyR1 disease mutation, N760D, appears to directly impact FKBP binding through interfering with SPRY1 folding.

Ambiguity in the RyR1 cryo-EM maps. Details of the 4.8Å (panel A) and 3.8Å (panel B) maps of RyR1 showing two alternative pathways of connectivity after residues 590-600. Shown are cartoon representations for the corresponding deposited models, showing the previously crystallized domain C (red), and the following helices in the armadillo repeats (orange). The two possible pathways, which are visible in both maps, are indicated by red and blue arrows. Following the pathway to the left (red arrow) leads to the SPRY1 domain in the density on the left (panel A, cyan). Following the pathway downwards leads to extra helices in the region 600-630, and the SPRY1 domain near the bottom of panel B (cyan). In the latter case, the SPRY3 domain ends up in the density on the left (yellow). C, D Cryo-EM density of a domain next to FKBP12.6 or FKBP12 in the 4.8Å and 3.8Å maps, respectively. The dotted circle indicates an individual domain likely corresponding to a SPRY domain. In both cases, de novo tracing of the structure is clearly not possible. The fitted crystal structures for FKBP12.6 and FKBP12 are shown in cartoon representation (beige).
Cross-eyed stereo views of the RyR2 SPRY1 domain backbone trace (A) and 2mFo-DFc electron density (B), the RyR1 Repeat12 domain backbone trace (C) and 2mFo-DFc electron density (D). Densities are shown at 1.5σ cut-off values. Coloring scheme in panels A and C is the same as for Figures 2 and 3 in the main manuscript.
A. Superposition of the RyR2 SPRY1 domain (colors) with the RyR1 SPRY2 domain (grey). Although the core strands are preserved, the loops vary in length and conformation, and the SPRY2 domain lacks the finger. B. Superposition of the RyR1 Repeat12 domain (colors) and the RyR1 Repeat 34 (i.e. phosphorylation) domain (grey). The superposition is done using the first repeats only, indicating the different angles between the repeats. The U-lid and three-stranded β-sheet are not present in the phosphorylation domain, indicating that the latter is not a good template to produce homology models of the Repeat12 domain.

Supplementary Figure 4
Comparison of the Van der Waals surfaces of SPRY1 versus SPRY2 (A) and Repeat12 versus Repeat34 (B). The electrostatic potentials are shown, with negative potentials in red, and positive potentials in blue. Flexible loops, for which no electron density was observed, are not included. Overall, there are distinct shape differences between SPRY1 and SPRY2, which are the result of several loops with different lengths and conformations. Repeat34 forms a horseshoe-shaped structure, which is not observed for Repeat12 due to the presence of a three-stranded β-sheet that fills up the space. The repeat domains are predominantly positively charged.

Supplementary Figure 5
Structure-based sequence alignment of the SPRY1 and SPRY2 domains. Secondary structure elements are indicated above and below. Stretches that are part of the crystallized constructs but that displayed no density are shown as dotted lines. The 'finger', formed by two antiparallel β-strands (red) pointing away from the SPRY core is unique for SPRY1.
Structure-based sequence alignment of the RyR1 Repeat12 and Repeat34 (phosphorylation) domains. Secondary structure elements are shown above and below. The first repeat of each domain corresponds quite well, but there is almost no structural homology within the inter-repeat linker, and the second repeat in Repeat12 has a shortened α1' helix, followed by a unique three-stranded β-sheet (strands β 4-6 ).
A, Positions of SPRY1 (cyan), Repeat12 (brown) and SPRY2 (magenta) in the 4.8Å cryo-EM map of CIPtreated rabbit RyR1 with bound FKBP12.6 (EMDB 6107). Also shown are the positions for the N-terminal domains A (blue), B (green) and C (red), as well as FKBP12.6 (tan). B, Normalized correlation coefficients for the top ten hits for docking of the crystal structures of SPRY1, SPRY2, and Repeat12.
A Positions of the SPRY1 (cyan), Repeat12 (brown) and SPRY2 (magenta) in the 6.1Å cryo-EM map of rabbit RyR1 (EMDB 2751). Also shown are the positions for the N-terminal domains A(blue), B (green) and C (red). No FKBP was observed in this map. B Normalized correlation coefficients for the top ten hits for docking of the crystal structures of SPRY1, SPRY2, and Repeat12.
A, Local fit of the Repeat12 domains in the corner of the 3.8Å map. There is a visible mismatch between the crystal structure and the electron density, which does not show the three-stranded β-sheet. B, Docked position of the top hit for the Repeat34 (phosphorylation) domain in the 3.8Å map. C,D Cross correlation coefficients (CCC) for Repeat12 and Repeat34 in the corner (C) and turret (D) positions of the 3.8Å map, per secondary structure element as calculated by VMD. The location of the β-sheet, which has the lowest CCC in the corner position, is indicated. Figure 10 A B

C D
A,B, Side-by-side comparison of fits for crystal structures of SPRY1 (cyan, left) and SPRY2 (magenta, right) into the globular density next to FKBP12.6 (beige). The electron density is for the 4.8Å map (EMDB 6107), which shows the clearest density for a loop pointing to FKBP12.6 (dotted cyan line). The SPRY1 domain visually fits much better in the density than SPRY2. Arrows on the right panel show several loops that do not fit well for SPRY2 (arrows). Most importantly, a flexible loop in SPRY1 has clear density in the map next to FKBP12 (dotted cyan lines in the left panel), suggesting it becomes ordered upon FKBP12/12.6 binding. The corresponding loop in SPRY2 has a minimal length to link the β-strands and cannot extend to the FKBP12 or FKBP12.6 surface. C,D, Correlation coefficients per secondary structure element for fits of SPRY1 and SPRY2 into the position next to FKBP, for the 3.8Å (C) and 4.8Å (D) maps. SPRY2 loops that do not fit well are highlighted. Overall CCC values determined by VMD are also shown.
Functional characterization of His 10 -tagged full-length RyR1 constructs using Fluo-4 based intracellular Ca 2+ imaging. This figure is related to Figure 5 in the main manuscript. A, Representative caffeineinduced intracellular Ca 2+ transients for indicated constructs expressed in HEK-293T cells are shown. A graded series of caffeine concentrations was perfused at the time intervals indicated by the black bars. B, The concentration dependence of caffeine activation of the indicated RyR1 constructs is shown. Values represent mean + SEM. C, Summary of mean EC 50 and 95% confidence interval (C.I.) of the mean for the number of cells analyzed (n). No significant differences in mean EC 50 were observed as determined by one-way ANOVA followed by a Dunnett's post-test (p<0.05). FRET-based measurement of intramolecular distances between FKBP and SPRY1 in full-length RyRs. This figure is related to Figure 5 in the main manuscript. A, The Cy3NTA concentration dependence of FRET from AF488-D44-FKBP12.6 to Cy3NTA bound to His 655 is shown. Fractional occupancy of 3 M Cy3NTA binding (85%) was used to correct static FRET measurements (Fig. 5)  FKBP binding to full-length GFP-RyR1 fusion constructs with mutations in the SPRY1 domain. A, HEK293T cells expressing GFP-tagged RyRs containing His 10 or scrambled sequences in the SPRY1 domain are shown after equilibration in saturating (10 nM) D49-AF568-FKBP (F-FKBP). GFP (top panels), F-FKBP fluorescence (middle) and a merged image of the two channels (bottom) are shown. B, Summary data corresponding to F-FKBP binding analyses from Fig. 7 are shown. F-FKBP:RyR1 B max for each construct is normalized to the B max of WT GFP-RyR1. K d values were not determined for the His 675 , 675 loop scramble, and F674A/L675A constructs, where no saturable binding was observed (Fig. 7).