Discovery of a novel cardiac-specific myosin modulator using artificial intelligence-based virtual screening

Figures

• 

  Fig. 1. Artificial Intelligence-based high throughput screen for cardiac myosin modulators. a Cardiac myosin heads can either undergo their mechano-chemical cycle that produces force or muscle shortening (top), or can be sequestered into the functional OFF state (interacting heads motif, bottom). b Surface structure of human cardiac myosin S1 bound to Omecamtiv Mecarbil (OM, purple). Individual sub-domains of the myosin catalytic domain are labeled accordingly. c Artificial intelligence-based virtual screen for novel cardiac myosin modulators targeting the Omecamtiv Mecarbil-binding site. Several million compounds were virtually screened against the OM-binding site using AtomNet® technology. Top ranking 84 compounds were chosen for further investigation.

• 

  Fig. 2. Identification of F10 as a cardiac-specific allosteric myosin inhibitor. a Single dose acto-myosin S1 ATPase activity screen of the 84 top ranking compounds from AI-based virtual screen. Hit thresholds are shown as red dashed lines. Means ± s.e.m. for n = 4 independent repeats, except for compounds #46 and #49 with n = 3 independent repeats. b Dose-response analysis for hit compound F10 on acto-myosin S1 ATPase activity. Means ± s.e.m., n = 6 independent repeats. c Chemical structure and physico-chemical properties of F10 (MW – molecular weight; log PO/W – partition coefficient octanol/water; LE – ligand efficiency). d F-Actin concentration-dependent bovine cardiac myosin S1 ATPase activity in the absence (black, vehicle control) and in the presence of 100 μmol/L F10 (red). Means ± s.e.m., n = 3 independent repeats for control and n = 9 independent repeats in the presence of F10. e ATPase activity of myofibrils isolated from cardiac, and fast and slow skeletal muscle in the absence (black) and in the presence of 20 μmol L-1 F10. Means ± s.e.m. for n = 20-23 experiments from n = 3 independent myofibril preparations. Statistical significance between control and drug treatment group…

• 

  Fig. 3. F10 slows nucleotide release from bovine β-cardiac myosin. a Left: SDS-PAGE of isolated bovine cardiac myosin S1 with essential light chain (ELC) and myosin heavy chain (MHC) labeled accordingly. Middle: Representative traces of mant-ATP chase experiments in the presence of vehicle control (gray), 2 μmol L-1 Mavacamten (purple) or 20 μmol L-1 F10 (red). Bi-exponential fits to data are shown as black continuous lines. Right: Fraction of slow phase nucleotide release for the three conditions. Means ± s.e.m., n = 5 independent repeats for control, and n = 6 independent repeats in the presence of F10 or Mavacamten. b Left: SDS-PAGE of isolated bovine cardiac myosin with regulatory light chain (RLC), essential light chain (ELC) and myosin heavy chain (MHC) labeled accordingly. Middle: Representative traces of mant-ATP chase experiments in the presence of vehicle control (gray), Mavacamten (purple) and F10 (red). Bi-exponential fits to data are shown as black continuous lines. Right: Fraction of slow phase nucleotide release. Means ± s.e.m., n = 5 independent repeats for control, and n = 7 independent repeats in the presence of F10 or Mavacamten. c Left: SDS-PAGE of isolated bovine…

• 

  Fig. 4. Effect of F10 on cardiac muscle mechanics and thick filament structure. a Isometric force of demembranated rat ventricular trabeculae during relaxing conditions (pCa 9) and full Ca2+ activation (pCa 4.5) in the absence (gray) and in the presence of 20 μmol L−1 F10. Means ± s.e.m., n = 5 trabeculae. b Order parameter <P2> determined from polarized fluorescence of BSR-cRLC-E exchanged demembranated rat trabeculae during relaxing conditions (pCa 9) and full Ca2+ activation (pCa 4.5) in the absence (gray) and in the presence of 20 μmol L−1 F10. Means ± s.e.m., n = 3 trabeculae. Pictograms indicate the orientation of the BSR probe (red double arrow) and myosin heads (CD, catalytic domain; LCD, light chain domain, BSR – Bifunctional sulforhodamine) with respect to the thick filament axis (black arrow). c Normalized force-pCa relation of demembranated trabeculae in the absence (gray) and in the presence of F10 (red). Inset: Extracted pCa50 values before and after F10 treatment. Means ± s.e.m., n = 3 trabeculae. d Example traces of force re-development of demembranated trabeculae following a slack-stretch protocol in the absence (gray) and in the presence of F10. Inset: Rates of force…

• 

  Fig. 5. Effect of F10 and Mavacamten on haemodynamic parameters of Langendorff-perfused rat hearts. a Left ventricular systolic pressure (LVSP), b left ventricular end-diastolic pressure (LVEDP), c maximum and minimum rate of pressure development (dP dt−1max and dP dt-1min, respectively), d heart rate (HR) and e coronary flow (CF) in the absence of compounds (open circles) and in the presence of either 20 μmol L−1 F10 (left) or 1 μmol L-1 Mavacamten (right), and after washout. Data represent means ± s.e.m. with n = 4 rat hearts for F10 and n = 5 rat hearts for Mavacamten. Source data are provided as a Source Data file.

• 

  Fig. 6. Structure-activity relationship of F10. a Left: Top five highest scoring docking poses of F10 (stick representation) in the OM-binding site of human cardiac myosin S1 created using AutoDock Vina. The N-terminal domain (NTD), converter and lower 50 kDa domain (L50D) are shown in green, pink and yellow, respectively. Right: Schematic showing the proposed main interactions of F10 within the myosin catalytic domain. Potential hydrogen bonds are indicted by black dashed lines. b Chemical structures of tested F10 derivatives. c Dose-response analysis for the effect of F10 derivatives on bovine cardiac myofibrillar ATPase activity. Means ± s.e.m., n = 7 independent experiments. Source data are provided as a Source Data file.

• 

  Fig. 7. Proposed mechanism for the effect of F10 on cardiac myosin function. F10 accelerates myosin head detachment from actin and prevents the transition from weakly- to strongly-bound crossbridges, and stabilizes its structural (interacting heads motif) and functional OFF state (super-relaxed state).