Fig 1.
Simulation of SP4206 binding to IL2.
(A) Snapshots taken from Simulation 1, in which one of the three SP4206 ligands in the system reached the native binding pose with IL2. The crystal pose (PDB 1PY2) of the ligand (blue) is also shown for reference. (B) The simulation ligand-binding pose compared with the crystal pose. Also shown are conformations of the protein before (0 μs, inset) and after (0.5 μs) binding, in surface representation. Notably, the binding groove is not present at the initial conformation of the simulation. (C) Chemical structures of SP4206 and analogs. (D) Time series of the SP4206 RMSD with respect to the crystal binding pose in juxtaposition with the series of the volume of the binding groove. Note that at the SP4206-binding site, a transient groove that is comparable in size to the native binding groove emerged at approximately 0.02 μs, 0.08 μs, and 0.22 μs (marked by green arrows), prior to the ligand binding at 0.25 μs (gray area). (E) The binding process described by estimated binding energy (y-axis) and conformational fluctuation of the ligand, as measured by RMSD with respect to the conformation of the previous time step (x-axis). Additional analysis of the simulations is presented in panel A of Fig B in S1 Text.
Fig 2.
Simulations of binding of SP4206 analog molecules and free energy calculation.
(A) The binding poses of SP4206 and its analogs generated by spontaneous binding simulations. (B) The non-native binding poses of SP4206 and its analogs generated by simulations, shown in the structural context of IL2-IL2R association. The native binding pose is also shown with a mesh around the small molecule. The colors of the small molecules are as shown in (A). (C) The conformational fluctuations of SP4206 and analogs in simulation-generated binding poses (y-axis) compared to the dissociation constants of the compounds (x-axis). As shown, with one exception for SP4206, the fluctuation is smaller in the native binding pose than in the non-native ones for the same compound. (D) The FEP binding free energies of SP4206 and analogs in various simulation-generated binding poses (y-axis) compared to the dissociation constants (Kd) of the compounds (x-axis). The binding free energy was estimated to be 20.5 ± 0.38 or 19.7 ± 0.33 kcal mol−1 for SP4206, 13.58 ± 0.28 kcal mol−1 for SP4206-1, 11.59 ± 0.29 or 11.58 ± 0.28 kcal mol−1 for SP4206-2, and 12.39 ± 0.3 kcal mol−1 for SP4206-3. In the inset, the calculated binding energy (y-axis) is compared to the binding energies derived from Kd (x-axis). For a given compound, the calculated binding energy is consistently greater for the native binding pose than for the non-native ones.
Fig 3.
Binding pathway and the transition state.
(A) The binding process shown in Fig 1 in terms of SP4206 dipole directions, which are color coded according to simulation time. In the intermediate state (left cluster, circled in orange) the arrowheads tend to be pointing to the left, reflecting a general alignment of the dipole directions; the well-aligned arrowheads (indicated by the yellow arrow) just underneath this cluster show the native binding pose of SP4206. The surface of IL2 is colored by the local electrostatic properties of IL2. (B) Evolution of IL2-SP4206 contacts (defined in the inset) in the binding process. (C) Two conformations identified from binding simulations as members of the transition state ensemble. The native pose is also shown. (D) 10 simulations launched from the orange conformation in (C), of which 5 (black lines) quickly led to native binding. (E) A sketch of the energetic landscape and pathway of the binding.
Fig 4.
Emergence of the binding groove.
(A) Conformations of the IL2 backbone and Phe42 before and after SP4206 binding. Multiple conformations from the binding simulation discussed in Figs 1 and 3 are superimposed. The backbone of the residues lining the long-side edges of the SP4206 binding groove moved further apart, and Phe42 was stabilized in a single rotamer after binding. (B) Top panel: the time series of SP4206 RMSD with respect to the crystal binding pose (identical to the top panel of Fig 1D); middle panel: two running averages (of different widths of the averaging window) of the RMSD of the backbone of the residues lining the long-side edges of the binding groove (residues 32–44, 63–74) with respect to the SP4206-bound conformation; bottom panel: RMSD of Phe42 with respect to the SP4206-bound conformation. (C) The spatial occupancy (mesh) of the two conformers of Arg38 from a 5-μs simulation of apo IL2, in which SP4206 and other small molecules were not present.
Fig 5.
Binding of fragments of SP4206 to IL2.
The occupancy density maps of fragments of SP4206: S and T in (A) based on a 54-μs simulation; N, M, L in (B) based on a 37-μs simulation; and N, Q, R, P, L in (C) based on a 16-μs simulation. In (D), the occupancy density of 18 common drug fragments unrelated to SP4206 is shown, based on a 16.5-μs simulation. The native binding pose of SP4206 is shown, to mark the cryptic binding site and to indicate the locations of the fragments relative to a bound SP4206 molecule.
Fig 6.
Compound 43a binding to Bcl-xL.
(A) Positions of the small molecule in an 11-μs simulation (Simulation 1) in which it reached the native binding pose. Simulation time is color-coded from red, to gray, to blue. (B) The RMSD of the small molecule with respect to the crystal structure (PDB 2O2M) in the binding processes of Simulations 1 (green) and 2 (cyan). Also shown is the RMSD of the small molecule in a simulation starting from the bound structure. All three simulations eventually settled at the same RMSD region. (C) and (D) The binding poses generated by Simulations 1 and 2, respectively, compared with the crystal structure. Note that simulation-generated binding poses are very similar but not identical to the crystal structure. The relatively minor discrepancies may be attributable to the force field for the small molecule.