Characterisation of the Binding Pocket and Drug Repositioning to Inhibit the Interaction Between the RBD Spike Protein of SARS-CoV-2 Omicron BA.1 and Integrins

MOTIVATION Severe COVID-19 exhibits angiogenic characteristics, including intussusceptive angiogenesis, endothelialitis, and activation of procoagulant pathways. This pathological condition can be attributed to the direct infection of human lung endothelial cells (ECs) by SARS-CoV-2. We showed that SARS-CoV-2 can infect ACE2-negative primary human lung microvascular endothelial cells (HL-mECs), leading to a shift in cell remodelling toward a pro-inflammatory and pro-angiogenic phenotype. This infection occurs via the RGD (R403, G404, and D405) motif located on the Spike protein, which is known to bind to different integrins, such as the αvβ3 integrin present on HL-mECs [1]. The RGD motif mutation D405N hinders Omicron BA.5 infection in HL-mECs and their dysfunction due to the lack of Spike/integrins interaction (2), demonstrating the direct involvement of the RGD in SARS-CoV-2 entry. A drug repositioning strategy was thus employed to find approved drugs able to inhibit this interaction. METHODS From the Spike crystallographic data available in the PDB, 7TGW was selected for the subsequent computational studies. The modelling of the missing residues of the structure was performed using the web server SWISS Model. The construction of the glycans around the Spike PDB structure was performed using the GlyCam web server. The simulation to assess the stability of the glycans and the following Gaussian Accelerated Molecular Dynamics studies were performed with Amber24. The ligands for drug repositioning were built and energy minimized using RDKit and XTB software. Docking studies were carried out by Autodock Vina, using Vinardo scoring functions. RESULTS Our findings show that the isolated Receptor Binding Domain (RBD) of the Spike protein, containing the RGD motif, interacts with αvβ3 integrin. In contrast, the glycosylated trimeric Spike protein requires a cofactor, such as heparin, to enable this interaction [2,3]. Classical and Gaussian molecular dynamics studies were performed to characterise the RBD domain active pocket engaged with integrin, in comparison to the inactive Spike pocket in the whole protein. The RBD domain’s active pocket features a larger volume and a greater solvent-accessible surface area (SASA) compared to the Spike protein inactive pocket. The conformation of the RGD motif, which interacts with αvβ3, shows a similar distance between the Cβ atoms of Arg and Asp residues (values 7,15 Å for the RBD and 6,99 Å for the Spike protein), as well as the angle of Cβ, Cα, Cβ of R403, G404, and D405 residues (values 94,98 ° for RBD and 92,95 ° for Spike protein). Furthermore, the SASA values of the nitrogen atoms in the Arg side chain and the oxygen atoms in the Asp side chain that directly interact with αvβ3 are strikingly similar. The active pocket is defined by specific values of volume and solvent accessibility, enabling interaction with αvβ3. To validate this hypothesis, a drug database has been repositioned at the active pocket of the RBD domain to identify small molecules that may inhibit the integrin/RBD interaction. Two molecules of particular interest have been selected for in vitro testing. Additionally, studies are ongoing to predict the Spike protein-heparin complex, able to interact with αvβ3, to define the active binding pocket involved in this binding process. This interaction enables the SARS-CoV-2 virus to infect endothelial cells, triggering a pro-inflammatory and pro-angiogenic phenotype.