Protein-ligand binding is important for early stage drug discovery. Structure-based drug design/discovery (SBDD), a rational and efficient strategy, aims to predict the binding mode and the affinity of protein with the ligand. For decades, researchers are on the way to improve the accuracy of SBDD and the great potentials and success of such method have been seen in the drug discovery field.
On a recent paper a FEP (Free-energy perturbation) protocol was released that enabled highly accurate affinity predictions across a broad range of ligands and target classes. Researchers applied this method in several drug discovery projects and got a high level of accuracy, indicating the ability of this approach to drive decisions in lead optimization.
This method is based on an improved force field OPLS2.1, which incorporates a robust model for non-bonded interactions in conjunction with extensive training of torsional and covalent parameters against more than 10,000 representative organic compounds. OPLS2.1 displays a lower error statistics than some other force fields, such as OPLS2005. In addition, the Desmond program is employed to run FEP simulations and molecular dynamics/replica exchange capabilities are augmented in Desmond with the newly developed FEP/REST (free energy perturbation/replica exchange with solute tempering) algorithm. Desmond with FEP/REST is implemented to run on graphics processing units (GPUs).
The researchers applied this methods into several studies. First, FEP/REST methodology was used to validate eight targets and related ligands. They evaluated the data of those eight sets produced from FEP methodology and got a scatter plot of predicted versus experimental binding energies for the entire data set (Fig.1).
In order to further validate the ability of FEP methodology on lead optimization, they examined several representative compounds with the targets by both computational and wet lab measurements; the results were corresponded to each other and structure-activity relationships successfully captured by FEP (Fig 2).
This technology was then employed into eight active drug discovery projects. On project one, several hundred compounds had been synthesized and the target biochemistry affinity and some other biological properties had been achieved. Next, FEP technology was used to optimize compounds, selecting compounds with better biological properties while maintaining a high affinity. 195 compounds were scored by FEP and 22 were synthesized and assayed. By compared the experimental and predicted data, the true negative rate was 93% and the true positive rate was 71%, which indicated a high accuracy of FEP technology (Fig.3).
Over the past decades, SBDD methods such as molecular docking, pharmacophore modeling and mapping, structure-based virtual screening have been greatly improved. On one recent study, SBDD was successfully applied to discover inhibitors of Human Helicase DDX3. Based on the structural information of DDX3, a homology of a closed conformational model was built and then a hit was then docked into the predicted RNA binding pocket for pharmacophore building. Commercial compounds databases were screened by virtual screening, followed by docking experiment via GOLD software. Selected small molecules were validated by DDX3 helicase activity assay and an impressive hit rate of 40% was obtained. The success of SBDD in drug discovery field is not only dependent on the constantly changing technology, but also about the detailed structural knowledge of the target macromolecules, which are mainly obtained from crystal structures, NMR data or homology models. In the future, it is still necessary to increase the accuracy and effectiveness of existing technologies in computational drug discovery field, but the most important tendency will be the integration of computational chemistry and biology together with chemoinformatics and bioinformatics.
Blog written by Xiangrong CHEN
- Wang, L., et al., Accurate and reliable prediction of relative ligand binding potency in prospective drug discovery by way of a modern free-energy calculation protocol and force field. Journal of the American Chemical Society, 2015. 137(7): p. 2695-2703.
- Fazi, R., et al., Homology Model-Based Virtual Screening for the Identification of Human Helicase DDX3 Inhibitors. Journal of chemical information and modeling, 2015. 55(11): p. 2443-2454.
- Chen, L., et al., From laptop to benchtop to bedside: structure-based drug design on protein targets. Current pharmaceutical design, 2012. 18(9): p. 1217.
- Ou-Yang, S.-s., et al., Computational drug discovery. Acta Pharmacologica Sinica, 2012. 33(9): p. 1131-1140.