Dynamics of the β-cardiac myosin auto-inhibited state explain cardiomyopathy pathogenesis

Figures

• 

  Fig. 1. Wild-type and DCM E525K CarIHMs have distinct dynamics. A Regulation of cardiac myosin and effects of mutations. B The left panel shows superimposition of WTConfA and WTConfB focused on the BH motor domain (residues 1-710). A ~20° twist in the S2 position creates different interactions between the S2 and the BH. On the right panel, a cartoon illustrating that in WT, a low proportion of particles are in IHM, with two conformations solved at high-resolution. C High-resolution structure of E525K-CarIHM. The top panel presents the full structure, the middle panel zooms in on the BH/FH interface, and the bottom panel zooms in on the catalytic site of the BH. D E525K-CarIHM is highly similar to WTConfA: both structures are aligned over the entire CarIHM, (RMSD 0.95 Å calculated from all Cα atoms of the IHM heavy chains. On the right panel, a cartoon illustrates that, in E525K, only one main CarIHM conformation is observed and that a high proportion of particles adopt the IHM. (E) Comparison of the interaction between the BHA-loop and the BHLoop-2 with the S2 in the three structures. The color code is the same as in (B, D). IHM interacting-heads motif, A-loop activation-loop, BH blocked…

• 

  Fig. 2. Dynamics of the S2 coiled-coil on the surface of the blocked head (BH) within the conformational ensembles of Wild-type (WT) and mutant CarIHM. The interaction of S2 with the BH is mediated by two clusters (1 and 2). A–G show the two charge clusters at the BH/S2 interface in the EM density map of Wild-type conformation A (WTConfA), conformation B (WTConfB), and E525K-CarIHM. A selection of residues are labeled when HCM or DCM mutations have been reported as pathogenic or likely pathogenic in ClinVar. HCM and DCM mutations are indicated in red and light blue, respectively. Mutations for R904 can either lead to HCM or DCM. In Cluster 2, charge of the residues at position 525 and 903 are shown to illustrate charge repulsions or attractions at the BH/S2 interface. H Results of the three all-atom MD simulations (WT-ConfA, DCM-E525K and HCM-E903K CarIHM) are shown by monitoring the 525Cα-903Cα distance over time. I The position of the S2 at distinct representative times of the simulations are shown to illustrate the drastic differences in S2 stability when the missense E525K and E903K mutations are introduced starting from a WTConfA structure. The two times chosen for each condition,…

• 

  Fig. 3. The dynamic Cluster 3 predicted by all-atom MD simulations and confirmed by CryoEM. A, B, two representative time points of the MD simulations of WTConfA and E525K are shown. The distal S2 (925–963) residues make no interaction at 0 ns. Transient interactions occur during the simulations (Supplementary Table 3 and Supplementary Video 8). Here, some interactions occurring at 92 ns for WTConfA and 41 ns for E525K are shown as example of Cluster 3 interactions. C The E525K CryoEM dataset was reprocessed with a series of local refinements and classifications (see “Methods” and Supplementary Fig. 10). One of the clusters from 3DVA was separated into 8 classes that show different conformations for S2, FHLoop-2, and FHHCM Loop. Orange rings outline Cluster 3 in each map. The resolution and the number of particles per class are shown. D Close-up of the EM density for Cluster 3 for the class boxed with pink dotted lines in (C). The distal S2, FHLoop-2, and FHHCM loop can be modeled to visualize the interactions that are formed. The ELCs are colored in pink, the RLCs are colored in yellow.

• 

  Fig. 4. Asymmetry of S2 and the Hook in CarIHM. A The asymmetries of the RLC/RLC interface and of the coiled-coil induce different environments for each RLC. Model and map showing the Hook region and the proximal part of S2. The E525K-CarIHM map, enhanced in the RLC/RLC region (see “Methods”), is shown. In CarIHM, the BHS2 residues are placed towards the C-terminus compared to the equivalent FH residues (see reporter residues represented by colored stars). The N-terminal extensions of the RLCs (RLCNterm, residues 1-20) are not in the core of the interface. B Zoom-in of the residues of Cluster 0, illustrating the asymmetry of the internal interactions stabilizing the S2 bend (see reporter residues represented by colored stars, each pair of residues). C Pathogenic and likely pathogenic mutations (Clinvar) are labeled according to whether they cause hypertrophic or dilated cardiomyopathy (HCM or DCM). CM: undifferentiated cardiomyopathy. Note that several Variable of Uncertain Significance have also been identified in S2 (see Supplementary Fig. 12).

• 

  Fig. 5. Docking of Cr1 and Cr3 and overall view of the refined atomic model of the thick filament. Density of Crown 1 (Cr1) (A) and Crown 3 (Cr3) (B) from the EMD-29734 map24 with the result of a rigid body fit of the CarIHM structure in which both FHHCM and FHLoop-2 are involved in interaction in Cluster 3 (as shown in Fig. 3D). A good agreement was obtained for the heads of the CarIHM motif for both Crown 1 and Crown 3 (correlation coefficient ~40%). For each fit, a zoom in Cluster 3 indicates that interactions do not occur in Cr1 (A) but are present in Crown 3 (B). A kink occurs in Cr3S2 at position 917 but not in Cr1S2 (dashed purple circle), allowing the distal part of Cr3S2 to engage in interactions with FH. C Refined atomic model of the human relaxed cardiac thick filament describing how selected CarIHM conformations are able to dock in the Crown 1 and Crown 3 environments.

• 

  Fig. 6. Interfaces stabilizing the crowns in the human cardiac relaxed filament model. A Overall model of the cardiac relaxed filament showing Crown 1 and Crown 3 (Cr1 and Cr3) in their environment. B–G Selected views of interfaces in the filament and how these were defined in the former PDB 8G4L24 (in transparent gray). Residues directly involved in interfaces are shown as spheres. B–D Interfaces that were not described in PDB 8G4L, now included in the filament model thanks to the approach combining the high-resolution structures of isolated CarIHM and short-term MD simulations. The phosphorylatable S15 from the N-terminal extension of the BHRLC (BH-RLCNTE) is outlined in (C). This regulatory residue is not directly involved in interactions stabilizing the CarIHM, but is rather involved in the intercrown interface. E–G Examples of optimized interfaces that were previously suggested in PDB 8G4L. The filament model thus gains completeness and accuracy through descriptions of realistic interfaces at the atomic level. In (G), FHLoop-2 does not form Cluster 3 interaction for Cr1 and S2 is involved in interactions with titin at this location. More distal residues of S2 (red star, 939–981),…

• 

  Fig. 7. Dynamics of the human cardiac relaxed thick filament. All-atom MD experiments were performed during 140 ns starting from the near-atomic model of the filament containing Mg.ADP.Pi (see “Methods”). A shows different conformations of the N-terminal extension of the BHRLC (BH-RLCNTE) of Crown 3 (Cr3). B shows how interactions between the Cr3-FH-RLC and MyBP-CC9 domain can form during the MD simulation when Cr3 slightly reorients as it is maintained close to the surface by intercrown interactions. C dynamic Cr3-FHLoop-1/MyBP-CC10 interface. D displays the putty representation of the root mean square fluctuation (RMSF) of the blocked head (BH) and the free head (FH) of each crown (Cr1 and Cr3) during the course of the simulation. The mobility of the BHLever arm of Crown 1 (Cr1) is much greater than that of Cr3, due to interactions involving the Cr3-FHRLC (intercrown and MyBP-CC9).

• 

  Fig. 8. Proposed model for the formation of CarIHM and role of its dynamics for docking on distinct environment of the thick filament. Near-atomic resolution cryoEM structures indicate that the RLC dimerization drives IHM formation by imposing mechanical constraints that orient the lever arms in such a way to promote interactions of the two heads in an asymmetric fashion to form head/head interactions. Subsequent formation of the Clusters 0 promotes interactions of the coiled-coil with the BH. ConfA and ConfB differ mainly by the interactions in Cluster 2, and both retain flexibility in the FH orientation as well as in the orientations of the Hook region compared to the rest of the IHM. Different IHM conformations are selected thanks to this flexibility to promote docking in distinct crown-specific environments. MYH7 mutations can displace the ON-OFF equilibrium, leading to hypercontractility (HCM variants) or hypocontractility (DCM variants). The stability of Crown 1 and Crown 3 anchoring to the filament seems coupled depending on RLC phosphorylation. Indeed, intercrown interactions (pink circle) mediated by the NTE of the FHRLC of Cr1 require that S15 is not phosphorylated so that it…