Reliability of AlphaFold2 Models in Virtual Drug Screening: A Focus on Selected Class A GPCRs

Abstract

Protein three-dimensional (3D) structure prediction is one of the most challenging issues in the field of computational biochemistry, which has overwhelmed scientists for almost half a century. A significant breakthrough in structural biology has been established by developing the artificial intelligence (AI) system AlphaFold2 (AF2). The AF2 system provides a state-of-the-art prediction of protein structures from nearly all known protein sequences with high accuracy. This study examined the reliability of AF2 models compared to the experimental structures in drug discovery, focusing on one of the most common protein drug-targeted classes known as G protein-coupled receptors (GPCRs) class A. A total of 32 representative protein targets were selected, including experimental structures of X-ray crystallographic and Cryo-EM structures and their corresponding AF2 models. The quality of AF2 models was assessed using different structure validation tools, including the pLDDT score, RMSD value, MolProbity score, percentage of Ramachandran favored, QMEAN Z-score, and QMEANDisCo Global. The molecular docking was performed using the Genetic Optimization for Ligand Docking (GOLD) software. The AF2 models' reliability in virtual drug screening was determined by their ability to predict the ligand binding poses closest to the native binding pose by assessing the Root Mean Square Deviation (RMSD) metric and docking scoring function. The quality of the docking and scoring function was evaluated using the enrichment factor (EF). Furthermore, the capability of using AF2 models in molecular docking to identify hits with key protein-ligand interactions was analyzed. The posing power results showed that the AF2 models successfully predicted ligand binding poses (RMSD < 2 Å). However, they exhibited lower screening power, with average EF values of 2.24, 2.42, and 1.82 for X-ray, Cryo-EM, and AF2 structures, respectively. Moreover, our study revealed that molecular docking using AF2 models can identify competitive inhibitors. In conclusion, this study found that AF2 models provided docking results comparable to experimental structures, particularly for certain GPCR targets, and could potentially significantly impact drug discovery.

Keywords: AlphaFold2; G protein-coupled receptor; molecular docking; structure validation; virtual screening.

Figure 1

The biological hierarchy of class A GPCRs selected for this study. The hierarchy diagram includes the 20 protein subfamilies and the protein’s IUPHAR codes.

Figure 2

The superposition of X-ray structure (purple-blue color) and the bound ligand (green color) with their corresponding AF2 model (cyan color). (A) The binding site of Opioid (δ); (B) the binding site of Chemokine (CCR2); (C) the binding site of A Orphans (GPR52); and (D) the binding site of Somatostatin (SST2).

Figure 3

The docking results of the crystal ligand of each receptor docked into the experimental structures (X-ray and Cryo-EM), and the corresponding AF2 models. The results of the selected 32 receptors were divided into two figures, (A,B), for clarity. The error bars represent the standard deviations.

Figure 4

The docking results of 76 ligands docked into the experimental structures (X-ray and Cryo-EM), and the corresponding AF2 models. The results of the selected 32 receptors were divided into two figures, (A,B), for clarity. All color dots are considered outliers.

Figure 5

Cumulative distribution plots for docking of protein–ligand complexes. (A) When crystal ligands of each receptor were used as input; and (B) when the 76 active ligands were used as input. The dotted lines indicate a 2.0 Å RMSD cutoff.

Figure 6

Posing accuracy of the 76 docked active ligands. (A) The success rate (%) for the top-scored active ligands docked into X-ray structures; (B) docked into Cryo-EM structures; (C) docked into the corresponding AF2 models. (D) The heatmap of docking success rates for the 76 active ligands with different numbers of rotatable bonds. A docking pose is considered successful if RMSD between the docking pose and the experimental conformation of the crystal ligand is <2.0 Å. Otherwise, a docking pose is considered to have failed. However, an RMSD of 2–3 Å is annotated as ‘Fail-marginal’.

Figure 7

The AF2 binding site of top-scored ligands for selected targets, along with their 2D protein–ligand interaction. AF2 models are displayed in cyan, and the corresponding X-ray structure is displayed in green for comparison. (A,B) The binding of the top-scored ligand against Cannabinoid (CB2); (C,D) the binding of the top-scored ligand against Dopamine (D3); (E,F) the binding of the top-scored ligand against Tachykinin (NK1); (G,H) the binding of the top-scored ligand against Adrenoceptors (β1); (I) represent the 2D protein-ligand interactions that explain (B,D,F,H).