Protein preparation wizard in maestro

For the purpose study, the 2.83 Å resolution crystal structure of human IKKβ (hIKKβ), which is partially phosphorylated and bound to the staurosporine analog K252a (PDB ID: 4KIK) (Liu et al., 2013 (link)) was used. The structure was optimized using the Protein Preparation Wizard in Maestro (Schrödinger, 2017 ) adding bond orders and hydrogen atoms to the crystal structure using the OPLS3 force field. Prime was used to fix missing residues or atoms in the protein and to remove co-crystallized water molecules. PROPKA was used to check for the protonation state of ionizable protein groups (pH = 7.0). The hydrogen bonds were optimized through the reorientation of hydroxyl bonds, thiol groups, and amide groups. In the end, the system was minimized with the value of convergence of the RMSD of 0.3 Å. Indicaxanthin and staurosporine analog K252a were prepared using LigPrep The force field adopted was OPLS3 (Jorgensen et al., 1996 (link)) and Epik 3.9 (Schrödinger, 2017-1) was selected as an ionization tool at pH 7.2 ± 0.2. Tautomers generation was unflagged and the maximum number of conformers generated was set at 32.

The structures of the NITVK and VQDLL motifs of SspB were taken directly from the Streptococcus gordonii SspB C-terminal domain crystal structure (Protein Data Bank entry 2WZA) (31 (link)). The structure was processed using the Protein Preparation Wizard in Maestro (Schrödinger release 2018-1; Schrödinger, LLC, New York, NY). The similarity searches for the NITVK and VQDLL motifs were performed with Surflex-sim version 2.601 (32 (link)) using two approaches. The first approach was to use all atoms of the peptide structures for the NITVK and VQDLL motifs as the hypothetical ligand. The second approach was to use the side chains of residues for the NITVK motif and the side chains of residues VLL for the VQDLL motif. The screened libraries were created from the ZINC (28 (link)) drug-like library (ZINC 2014 version) containing 24,877,119 compounds and the ZINC 15 (29 (link)) drug-like library (ZINC 2016 version) containing 17,244,856 compounds. The results were ranked, and the top 500 compounds of each screen were retained. The selection of compounds was based on compound score; diversity, by eliminating compounds that were structurally similar to a higher scoring compounds; and finally, compounds that were commercially readily available for purchase. Seventeen compounds and 16 compounds were purchased (MolPort SIA, Riga, Latvia) for the NITVK and VQDLL motifs, respectively.

RORγt structures with a full orthosteric inverse agonist bound were obtained from the Protein Data Bank (PDB; https://www.rcsb.org/; obtained 1 June 2019) [51 (link),52 (link)]. The structures were superimposed using VERTAA in the BODIL molecular modeling environment [53 (link)], and the validity of side-chain conformations in the ligand-binding site was checked against the electron densities with COOT (version 0.8.9.1) [48 (link)]. Two protein 3D structures were selected for VS (PDB IDs: 5NTW [32 (link)] and 5VB6 [42 (link)]). While 5NTW is a good representative of most inverse agonist bound RORγt structures, 5VB6 presents an exceptional conformation state of the LBD. This unique state is explained by the crystallization process, during which RORγt was covalently tethered to a cofactor peptide stabilizing its structure thermodynamically, which in turn provides higher conformational flexibility. Hence, focusing on 5VB6 may provide new atomic insight into the RORγt inverse agonism [42 (link)]. The two structures were prepared with Protein Preparation Wizard in Maestro (Schrödinger Release 2018-2; Schrödinger, LLC, New York, NY, USA, 2018; https://www.schrodinger.com) for molecular docking and NIB modeling. Protonation states were assigned using PROPKA with pH 7.4 ± 0.0, and added hydrogens were optimized with the OPLS3 force field [54 (link)].