3.1. Software Tools Used for Epitope and Ligand Binding Site Determination

In silico prediction of the linear hPSA epitope RHSLFHP was carried out using the consensus result of 10 servers presented in Table 2: ExPASy–ProtScale (Kyte & Doolittle hydropathy scale, window size 9), BcePred (antigenic propensity method), BepiPred-2 (random forest regression algorithm), LBtope (support vector machine models), SCRATCH–COBEpro (support vector machine based propensity score; Scheme S1), Sann and RVP-net (solvent accessibility algorithms), NetTurnP and LEPS (β-turn determination) and the informational spectrum method based on electron-ion interaction potential (ISM-EIIP) [27,28,29,30,31,32,33,34,35,36,37,38,39,40]. The hPSA sequence is given in Scheme S1.

The 3D structures of the hPSA molecule and its binding sites were predicted by coupled Phyre2–3DLigandSite servers for protein modeling, prediction and analysis (Figure 1a) [41,42]. 100% of hPSA residues were modeled at >90% confidence, the 3D structure template, obtained with X-ray diffraction, was protein data bank (PDB) file 2ZCH chain P (Scheme S2) [41]. The results obtained by Phyre2–3DLigandSite-based prediction were confirmed by RaptorX web portal Binding Prediction (Figure 1b) [44].

The antisense peptide, i.e., paratope AVRDKVG, was designed by the translation of the RHSLFHP epitope in a 3′ to 5′ direction (Scheme 1) [19,20]. Potential paratopes to hPSA_53–59 epitope were selected using BLAST, with the blastp quadripeptide option (protein–protein BLAST) [19,20,45].

The 3D structures of the sense epitope RHSLFHP (Scheme S3) and its antisense paratope AVRDKVG Scheme S4), presented in Figure 2, were modeled using PEPstrMOD method [46]. The PDB files were visualized using CCP4 software version 7.0 [62].

Protein–peptide interactions between hPSA and its antisense peptide AVRDKVG were investigated using the pepATTRACT and CABS-dock web servers [47,48]. Both web services required the hPSA (protein) input file in PDB format (Scheme S2) and docking peptide sequence (AVRDKVG).

The pepATTRACT server is a novel fully blind docking protocol that does not require any information about the binding site [47]. In a very short time (~10 min) this web server returns an analysis of the most prevalent protein–peptide contacts among the top 50 generated models [47]. The pepATTRACT protocol is well tuned to assist users to identify the binding site, performs large-scale in silico experiments, and represents a useful starting point to rationalize the design for further protein–peptide docking experiments [47]. Model No. 24, presented in Figure A1a, is a precise description of hPSA–AVRDKVG docking. Results are given in PDB format as above (Scheme S5).

The CABS-dock server enables the modeling of protein–peptide interactions through an efficient method for the flexible docking of peptides to proteins without pre-defining the localization of the binding site [48]. The CABS-dock protocol was tested over the largest dataset of non-redundant protein–peptide interactions available to date, which includes bound and unbound docking cases [48]. Using a ligand RMSD cutoff of 5.5 Å, the best 10 models are selected with reference to the quality of docking models, which represent low-to-medium-accuracy models [48]. Model No. 5, presented in Figure A1b and Table A1, gives a precise description of hPSA–AVRDKVG docking. This result was given in the PDB format as Scheme S6.