Locating Large, Flexible Ligands on Proteins

Abstract

Many biologically important ligands of proteins are large, flexible, and in many cases charged molecules that bind to extended regions on the protein surface. It is infeasible or expensive to locate such ligands on proteins with standard methods such as docking or molecular dynamics (MD) simulation. The alternative approach proposed here is scanning of a spatial and angular grid around the protein with smaller fragments of the large ligand. Energy values for complete grids can be computed efficiently with a well-known fast Fourier transform-accelerated algorithm and a physically meaningful interaction model. We show that the approach can readily incorporate flexibility of the protein and ligand. The energy grids (EGs) resulting from the ligand fragment scans can be transformed into probability distributions and then directly compared to probability distributions estimated from MD simulations and experimental structural data. We test the approach on a diverse set of complexes between proteins and large, flexible ligands, including a complex of sonic hedgehog protein and heparin, three heparin sulfate substrates or nonsubstrates of an epimerase, a multibranched supramolecular ligand that stabilizes a protein-peptide complex, a flexible zwitterionic ligand that binds to a surface basin of a Kringle domain, and binding of ATP to a flexible site of an ion channel. In all cases, the EG approach gives results that are in good agreement with experimental data or MD simulations.

Computational model predicts protein binding sites of a luminescent ligand equipped with guanidiniocarbonyl-pyrrole groups

“…The van der Waals radii of ligand atoms were added automatically by OpenBabel. The 14-3-3ζ structure (PDB 4IHL, [ 20 ]) was refined with MODELLER [ 31 ] as described in reference [ 26 ]. Charges, van der Waals radii and missing hydrogen atoms were added by PDB2PQR v2.0.0 [ 27 – 28 ] at pH 6.5 with the Amber force field option (values are provided as a table in Supporting Information File 2 and in Supporting Information File 3 , respectively).…”

“…The 14-3-3ζ electrostatic field was calculated by solving the non-linear Poisson–Boltzmann equation with APBS version 1.5 [ 32 ] with ionic concentrations of 0.1 mol/L NaCl and 0.01 mol/L MgCl 2 and relative dielectric permittivities ε r protein = 2 and ε r water = 79. To convert the electrostatic potential grid into energy grids suitable for the Hamiltonian, chemically relevant polymer fragments were used as molecular probes in Epitopsy [ 26 ]. Epitopsy is a tool designed to calculate the electrostatic energy of a protein–ligand system from the protein potential grid and ligand charge distribution; this approach captures ion-size effects and yields energy grids that are directly comparable to molecular dynamics simulation data.…”

“…To validate our method, we applied it to the QQJ-096/14-3-3/c-Raf complex for which the potential binding site was available from all-atom molecular dynamics simulations in a previous study [ 26 ]. Our results for this system are reported in the first part of the Supporting Information File 1 .…”

“…Therefore, standard docking methods are not an option. Since this class of large, flexible, charged ligands is biologically important, we have previously devised a method, Epitopsy [ 26 ], to identify binding sites of fragments of such ligands on proteins. In this work we go a significant step further: to identify the putative binding sites of the full compound 1 on 14-3-3ζ, and possibly to guide further experiments, we, first, developed a computational model of the molecular system consisting of 14-3-3ζ, 1 , and an implicit solvent.…”

Computational model predicts protein binding sites of a luminescent ligand equipped with guanidiniocarbonyl-pyrrole groups

“…Van der Waals radii of ligand atoms were added automatically by OpenBabel. The 14-3-3 structure (PDB 4IHL, [18]) was refined with MODELLER [23] as described in reference [24]. Charges, van der Waals radii and missing hydrogen atoms were added by PDB2PQR v2.0.0 [19,20] at pH 6.5 with the Amber force field option (values are provided in Table S3 in the Supporting Information File 2 and Table S4 in the Supporting Information File 3).…”