Macromodel module

2-HTP, 6-OAU, and DL-175 were docked into the previously published hybrid template homology model (24 (link)) and the recently released AlphaFold model of the human GPR84 (37 (link)) using a standard precision docking protocol available in the Glide module of Schrodinger software (2020-1) (43 , 44 (link)). The protein structures were prepared with the Protein Preparation Wizard, whereas the agonists were generated with the Ligand Preparation module of Schrodinger software. The docking box was centered based on Tyr⁶⁹, Phe⁹⁶, Leu⁹⁹, Phe¹⁰¹, Arg¹⁷², Phe³³⁵, and Trp³⁶⁰ residues. The poses were selected based on the Glide docking score and taking into consideration experimental information. Minimization of docking complexes was performed using the MacroModel module of Schrodinger software. A default protocol with 1000 steps of minimization in implicit solvent was used to obtain the final complex. The OPLS_2005 force field was used in MacroModel calculations. The images for Figures 11 and 12 were created in Maestro 2020-1.

A MacroModel module of Schrodinger 2013 package was used for the conformational studies. The linear and the cyclic analogues MBP_82–98 (Ala⁹¹) were designed and minimized with an OPLS_2005 force field, the dielectric constant was set to 45, simulating the DMSO environment of the NMR solvent. Minimization was performed with the PRCG (Powell–Reeves conjugate gradient) algorithm, using 100,000 iterations and a convergence threshold of 0.01 kcal/mol Å. Energetically minimized conformations were further subjected to molecular dynamic simulations. During the molecular dynamics (MD) simulations, the method of stochastic dynamics was implemented; the simulation temperature was 600 K, the time step was equal to 1.5 fs, the equilibration time was equal to 2000 ps, and the simulation time was equal to 20,000 ps. Five hundred (500) conformations resulted after MD simulations, which were classified into 10 families (clusters), according to the rotation of dihedral angles φ (rotation: N-Ca), ψ (rotation: Ca-CO), and ω (rotation: CO-N). At a final stage, conformers that satisfy the distance constraints were minimized using the parameters mentioned above.

The FhCL3 homology model was built based on the crystal structure of FhCL1 previously reported by us (PDB code: 2O6X) [12 (link)] (sequence identity: 71%), using the Prime program of the Schrodinger software [46 ]. The ppFhCL3 homology model was built using the prosegment of the human pro-cathepsin L (PDB code: 1CJL) (sequence identity is 30%) by the Prime module. The initial coordinates of FhCL3-ppFhCL3 complex were obtained by superimposition with the complex of the human pro-cathepsin L (PDB code: 1CJL). The sequence identity between FhCL3 and human cathepsin L is 48%. Next, the FhCL3-ppFhCL3 complex was subjected to minimisation and molecular dynamic simulations the MacroModel module protocols of the Schrodinger software [46 ]. Default protocols with 5000 steps of minimisation and 5 ns of molecular dynamics simulations in implicit solvent at temperature 300 K were used to obtain the final complex. The OPLS_2005 force field was used in MacroModel calculations. The image with the molecular models was prepared with the Maestro program of Schrodinger software [46 ]. This analysis together with sequence alignment allowed the identification of key-residues participating in the binding and stabilisation of the ppFhCL3 with FhCL3 and narrowed the propeptide portion that possibly determines the blockage of the active site of the mature cathepsin (see Additional file 4).

LigPrep of Schrödinger software suit [43] was used for the preparation of ligands: generating 3D structures from 2D (SDF) representation, and performing a geometry minimization. The ligands were subjected to energy minimization using MacroModel module of Schrödinger with Merck Molecular Force Field (MMFFs). Truncated Newton Conjugate Gradient (TNCG) minimization method was used with 500 iterations and convergence threshold of 0.05 (kJ/mol). While Epik [44] was used to generate possible ionization states at pH 7.0±1.0.

Conformational analysis of compound 1 and 2 was assessed with Conformational Search tool from MacroModel module (Schrödinger LLC). Analysis was conducted with OPLS4^{69 (link)} force field force field in a water solvent was employed for the analysis. A mixed torsional/Low mode sampling approach was chosen with the default settings. The Polak-Ribier Conjugate Gradient⁷⁸ (PRCG) method with restarts every 3N iterations (maximum of 2500 iterations) was utilized for energy minimization, with a convergence threshold of 0.05. For compound 1, a total of 139 conformers were generated, while for compound 2, 278 conformers were generated. The ligand conformation that best matched the crystal structure was determined by SMARTS superimposition of the structure scaffolds with the reference ligand conformation. The selection of the final conformation was justified based on the lowest superimposition root mean square deviation (RMSD) values obtained from both conformational datasets.