Allosteric modulation of cardiac myosin dynamics by omecamtiv mecarbil

Figures

• 

  Fig 1. Structure of the myosin motor domain and acto-myosin cycle. A. Cartoon representation of the motor domain structure (PDB ID: 4PA0 [8]), with the subdomains highlighted in different colours and OM shown as purple spheres. The motor domain is connected to the rest of myosin through the lever arm and the regulatory domain (not shown). B. Simplified representation of the acto-myosin cycle, where myosin switches between actin-bound (bottom) and -unbound (top) states and between up and down conformations of the lever arm.

• 

  Fig 2. Mapping of Cα RMSF profiles onto the cMotorD structure. RMSF values from Apo and OM-bound simulations are colour mapped onto the cMotorD structure from blue (0 Å) to red (3.4 Å). The average structure is used for each simulation. The thickness of the tube representation is proportional to the RMSF value. High flexibility regions and the OM binding site are also labelled.

• 

  Fig 3. Collective motions in Apo and OM-bound simulations. A. Porcupine representation of PC1 (top panels) and PC2 (bottom) in ApoA1 (left panels) and OMA1 (right) simulations. The orange spikes show the direction and relative amplitude of motion of each residue along the PC. The approximate direction of the CLD hinge axis is also shown for Apo simulations (orange arrows). The two insets show the anti-correlated (Apo) and correlated (OM-bound) motions of the CLD (blue) and SH3 (green) subdomains. B. DynDom dynamic domain decomposition for ApoA1 PC1 (top) and the recovery stroke (bottom). The analysis was performed on the structures with minimum and maximum PC1 value from the MD simulation and on the experimental structures representing the pre-power stroke (PDB ID: 1QVI) and near-rigor (PDB ID: 1SR6) states for the recovery stroke. The fixed (white) and moving (yellow) domains identified by DynDom are shown, together with the hinge axis (orange) and the hinge regions (magenta).

• 

  Fig 4. Bending of the relay helix during the MD simulations. The relay helix bending was measured by calculating the angle formed by residues I478, F489 and E500 (Cα atoms only). The central plot shows the probability distribution of the angle values observed during Apo (green hues) and OM-bound (blue hues) simulations. Arrows indicate the approximate value of the angle measured for representative experimental structures of the pre-power stroke state (PDB IDs shown). Apo trajectories showed a higher propensity for bent conformations (lower angles) than OM-bound simulations. The insets show representative Apo bent (green, left) and OM-bound straight (cyan, right) structures. Experimental structures of the relay helix in the near-rigor (NR, white, PDB ID: 1SR6) and pre-power stroke (PPS, magenta, PDB ID: 1QVI) state from scallop myosin are also represented as reference.

• 

  Fig 5. OM-protein interactions. A. X-ray structure of the OM-binding site in cardiac myosin (PDB ID: 4PA0, chain A). Residues within contact distance from OM are shown as sticks and coloured according to the subdomain they belong to. B. Frequency of OM-residue contacts during OM-bound MD simulations. Each bar represents the fraction of the simulated time for which the corresponding residue was found in contact with OM (minimum distance between any non-hydrogen atom < 4 Å). Residues are labelled for frequencies larger than 0.5.

• 

  Fig 6. Network of inter-residue contacts in the OM-binding site. Red edges connect pairs of residues that are found in contact for at least 70% of the simulation. All the residues within 8 Å from OM were included in the analysis. Contacts are reported for replica 1 only (see S4 Fig for all the replicas). For each chain, pairs that were consistently found in contact in both OM-bound simulations and in none of the Apo ones are labelled.

• 

  Fig 7. Comparison of dynamic cross-correlation networks. Difference DCCM matrices (ΔDCCM) calculated between pairs of OM-bound and Apo simulations are mapped onto the initial structures of each OM-bound simulation. Edges connect residue pairs that have a positive (red) and a negative (green) ΔDCCM value, using a threshold of 0.16. Residues in the OM-binding site are represented as yellow spheres, while the CLD domain is coloured in blue.

• 

  Fig 8. Preferential connections in the local correlation network between the OM-binding site and functional regions in cMotorD. A. Plot of the difference between OM and Apo preferential connection scores (Δζ), calculated using fragment V698 as source site. The scores are derived from a consensus correlation matrix calculated over the four simulations of each binding state. A negative value indicates a stronger preferential connection to V698 in OM-bound simulations compared to Apo ones. Negative bars are coloured according to the subdomain. The position of DCM-related mutations is indicated with a yellow dot. Grey areas highlight the position of functional regions, namely the seven strands of the central β-sheet (b1 to b7), the P-loop (P), the F helix (FH), Loop 1 (L1), the G helix (GH), Switch 1 (SW1), the β-bulge (bb), the OH loop (OH), Switch 2 (SW2), the Strut (St) and the SH1 helix (SH1). The inset shows a close-up view of the fragments in the F-helix, the loop L1 and the G-helix (resid 185–228). B. Mapping of the Δζ scores onto the initial structures of OMA1. Only regions with negative values are coloured. The thickness of the tube is proportional to the magnitude of Δζ values. A…

• 

  Fig 9. Communication pathways from the OM-binding site in the network of local dynamic correlations. A. The shortest paths calculated from fragment V698 (magenta) to selected functional regions are reported as purple edges for Apo (top) and OM-bound (bottom) simulations. Fragments are mapped onto the structure by using the first residue in the fragment. The functional regions (F and G helix, β3–5, Switch 2 and SH1 helix) were selected among those with a stronger preferential connection to V698 in OM-bound simulations compared to the Apo ones. The nodes in the paths are represented as spheres coloured according to the subdomain they belong to and with a radius proportional to the number of paths going through them (so that larger spheres correspond to hubs). Edge thickness is proportional to the correlation value. B. Boxplot representation of the path lengths for the shortest paths reported in A for Apo (green) and OM-bound (cyan) simulations.

• 

  Fig 10. Pathways connecting the OM-binding site and the G-helix in the contact change network. The shortest paths connecting V698 (magenta) and G helix residues (red) in the network of OM-Apo contact changes are represented as yellow edges. The paths are calculated from a consensus contact change matrix calculated over all the simulations. Only contacts with a change in frequency larger than 0.1 were considered. The edge thickness is proportional to the difference between OM and Apo contact frequency. For each pair of endpoints, the top 5 shortest paths are represented. The nodes in the paths are represented as spheres coloured according to the subdomain they belong to and with a radius proportional to the number of paths going through them. The top 10 residues for number of paths are also labelled.