Docking data for "The evolution of the SARS-CoV-2 spike protein for differential usage of the host transmembrane serine proteases entry pathway"

The dataset includes predicted complexes of the SARS-CoV-2 Spike protein (specifically at the S2' cleavage site) with Hepsin and TMPRSS2 proteins. It contains data on three variants: Wuhan, Delta, and Omicron BA.1.

Compressed folders:

-357596-DeltaHepsin.tgz

-357597-DeltaTMPRSS2.tgz

-360039-WuhanHepsin.tgz

-360042-TMPRSSWuhan.tgz

-392981-TMPRSS-BA1_all.tgz

-392982-Hepsin-BA-all.tgz

Each compressed folder contains the following:

-Initial structures in pdb format

-Output complexes in pdb format

-Clusters in pdb format

-Protocols

-Parameters

-Scoring files

Protein-protein docking Molecular docking between the SARS-CoV-2 S protein of Wuhan, Delta (PDB: 7W92, [DOI: 10.1038/s41467-022-28528-w]), and BA.1 (PDB: 7XO5, [DOI: 10.1038/s41422-022-00672-4]) and the human proteases TMPRSS2 (PDB: 8HD8, [DOI: 10.1038/s41467-023-42527-5]) and Hepsin (PDB: 1Z8G, [DOI: 10.1042/BJ20041955]) was performed using the HADDOCK v2.5-2024.03 webserver ([DOI: 10.1021/ja026939x], [DOI: 10.1016/j.jmb.2015.09.014]). Missing loops in the protein structures were reconstructed using Modeller v10.5 ([DOI: 10.1006/jmbi.1993.1626]). Every heteroatom was removed from the reference structures. The relaxed atomistic coordinates for each S protein variant were derived via all-atom molecular dynamics (MD) simulations. These simulations were performed using AMBER22 with the FF19SB force fields and the pmemd.cuda module for enhanced performance ([DOI: 10.1021/acs.jcim.3c01153], [DOI: 10.1021/jz501780a], [DOI:10.1021/ct400314y]). For the Wuhan variant the S protein was retrieved from our previous modeling study [DOI: 10.1039/D0NR03969A] where for Delta and BA.1, ecah S protein was placed in a dodecahedral box, extending 20 Å beyond the solute in every cartesian direction, and solvated with the four-site OPC water model ([DOI: 10.1021/jz501780a]). The systems were neutralized with counterions, specifically one Cl− ion for the Delta variant and three Cl- ions for the BA.1 variant. To remove local clashes, a geometric optimization was performed using the steepest descent algorithm for 5000 cycles. The MD equilibration process consisted of several stages. First, temperature equilibration in the NVT ensemble was performed by gradually increasing the temperature through steps of 150, 200, 250, 300, and finally 310 K, each lasting 200 ps. During this phase, position restraints were applied to the heavy atoms of the proteins, with progressively decreasing spring constants of 5.0, 4.0, 3.0, and 1.0 kcal mol−1 Å−2, facilitating gradual relaxation. This was followed by a 1 ns equilibration at 310 K in the NPT ensemble without restraints. For production MD, the simulations were run in the NPT ensemble with periodic boundary conditions and Particle Mesh Ewald (PME) method ([DOI: 10.1063/5.0040966], [DOI: 10.1021/ct9001015]) using a grid spacing of 1.0 Å for long-range electrostatics. Non-bonded interactions were modeled with a Lennard-Jones potential using a 9Å cutoff. Temperature control was maintained using Langevin dynamics ([DOI: 10.1021/ct800573m]) with a collision frequency of 4.0 ps−1, and pressure control was managed by the Monte Carlo barostat ([DOI: 10.1016/j.cplett.2003.12.039]) with a 2.0 ps relaxation time at 1 bar. Bond constraints on hydrogen atoms were applied using the SHAKE algorithm ([DOI: 10.1016/0021-9991(77)90098-5]), and the hydrogen mass repartitioning scheme was applied via ParmEd ([DOI: 10.1371/journal.pcbi.1005659]), enabling a 4 fs integration time step ([DOI: 10.1021/ct5010406]). Each protein complex was simulated for a total of 20 ns. For the Wuhan variant, the 3D coordinates were retrieved from [DOI: 10.5281/zenodo.3817446].The active interaction region on the spike protein was defined as the cleavage site (residues P809-R815). For TMPRSS2 and Hepsin, the active sites were defined based on their catalytic residues: H296, D345, D435, S441, S460, and G462 for TMPRSS2, and H203, D257, D347, A348, and S353 for Hepsin. These specific regions were selected to guide the docking process and maximize biologically relevant interactions. Docking clusters were analyzed by selecting those with the lowest interaction energies for further structural analysis. To evaluate binding accuracy, native contacts between the S protein and proteases were computed using the contact map analysis based on the OV+rCSU method ([DOI: 10.12693/APhysPolA.145.S9, 10.1021/acs.jctc.6b00986]), which allows for a precise identification of critical stabilizing interactions, both specific and non-specifics. High-frequency contacts, defined as those appearing in over 70% of the generated models, were highlighted as key determinants of protein-protein recognition, providing insight into the most stable and consistent interactions across docking configurations.