Page 89 - TD-4-1

Tumor Discovery Identification of a potential KRAS(G12C) inhibitor

2.2. Protein and ligand preparation to <10.0 kJ/mol, with a maximum duration of 100

The 174 compounds containing the acrylamide warhead picoseconds (ps) for the minimization steps. Subsequently,
were prepared and optimized for molecular docking using a 100 ps equilibration phase with position restraints on
LigPrep. 26,27 Compounds were obtained using Epik at the protein and ligand molecules was conducted using
pH 7.0 ± 1.0. The protein structure of the inactive GDP- NVT (Number of Volume and Temperature) and NPT
28
bound state of human KRAS(G12C) (PDB ID: 6OIM) was (Number of Pressure and Temperature) ensembles.
retrieved from the Protein Data Bank, with a resolution During this period, the systems were heated to 303.15 K
29
of 1.65 Å. Protein preparation was performed using the using a modified Berendsen thermostat (V-rescale), 37,38
30
protein preparation wizard in Maestro, which fixed all the and pressure was maintained using C-rescale pressure
39
missing residues and atoms of the protein structure. Several coupling set to a reference pressure of 1 bar. For energy
crucial steps in the preparation included introducing minimization, NVT, and NPT relaxation simulations,
hydrogen atoms, refining hydrogen bond networks, and short-range interactions utilized a smooth force switch
addressing missing side chains through the utilization of with a cut-off of 1.2 nm, and long-range electrostatics
Prime. Subsequently, water molecules were eliminated, and were computed using the Particle-Mesh-Ewald (PME)
40
a restrained minimization employing the OPLS3e force algorithm. In addition, hydrogen bonds were constrained
41
field was carried out. The resulting optimized protein by employing the Linear Constraint Solver algorithm.
31
structure was then utilized for the subsequent docking Finally, MD simulations were conducted for durations
process. of 100 ns, 300 ns, and 500 ns (Table 1) without applying
restraints. A 2 femtoseconds integration time step was
2.3. Covalent dockingbased virtual screening employed, and trajectory snapshots were captured at 1 ps
A set of 174 pharmacophores similar to Sotorasib was intervals.
employed in covalent docking-based virtual screening MDS analyses were performed using GROMACS, a
(CovDock-VS) against the KRAS(G12C) target (PDB Linux-×86_64-multicore CUDA-enabled program on an
32
ID: 6OIM). The reactive residue Cys12 was specifically NVIDIA GeForce GTX 1060 GPU and an 8-core Intel Core
selected, and grid generation for docking was created as i5-7000 central processing infrastructure. The computation of
a cubic grid box centered at the coordinates (x, y, and z) a single complex required approximately 100 ns of processing
of Sotorasib, with a length of 12 Å. The acrylamide time per simulation day. GROMACS modules were used to
electrophilic warhead group, identical to the template analyze MD trajectories, focusing on parameters such as
drug Sotorasib, was selected for the Michael addition – a root RMSD/mean-square deviation fluctuation (RMSF) and
reaction responsible for the covalent bonding of ligands hydrogen bond (H-bond) analysis. Graphs were generated
featuring an acrylamide functional group with the side using XMGrace and visual molecular dynamics (VMD). 43
42
chain of nucleophilic protein residues. Subsequently, the
ligands were selected and prioritized based on their Glide The calculation of binding free energy was performed
scores for the binding modes of pre-reactive complexes. 33,34 using the gmx_MMPBSA tool, which is based on Assisted
Model Building with Energy Refinement MM/PB(GB)SA
2.4. MD simulations approach. This tool is designed for conducting end-state
MD simulations were conducted on the protein-ligand free energy calculations using GROMACS trajectory files.
complexes derived from the covalent docking-based Specifically, the final 50 ns of the simulation trajectory were
virtual screening using GROningen MAchine for Chemical considered for the molecular mechanics with Generalized
Simulations (GROMACS) 2023.1. The CHARMM-GUI Born surface area (MM/GBSA) calculation.
35
(Chemistry at HARvard Macromolecular Mechanics
Graphical User Interface) server was used for solution Table 1. KRAS(G12C) complexes feature with their atom
building. The CHARMM36 force field was applied to numbers for duration 500 ns
36
define atom types and assign atomic partial charges for both Atoms KRAS(G12C)-Sotorasib KRAS(G12C)-C02b
the ligand and protein. Furthermore, the CHARMM-GUI
server was employed to generate the topology for both the Protein 2,704 2,704
KRAS(G12C) receptor and the promising inhibitors. For Ligand 72 87
the setup of MD simulations, the protein-ligand complexes Na + 14 13
were centered in an octahedral box and solvated with the Cl - 7 7
TIP3P water model. Neutralization of the complexes charge Water (TIP3P) 7,914 7,923
was achieved by adding 150 mM ions. The minimization Abbreviations: TIP3P: Transferable intermolecular potential three-
process concluded when the maximum force was reduced point; KRAS: Kirsten rat sarcoma viral oncogene homolog.

Volume 4 Issue 1 (2025) 81 doi: 10.36922/td.5163