Borrelidin Binding Site Of Target ThrRS And Virtual Screening For Its Inhibitors

Phytophthora sojae (P. sojae for short) is an oomycete and a soil-borne plant pathogen that causes stem and root rot of soybean. This is a prevalent disease in most soybean growing regions, and a major cause of crop loss. In our attempts to search for potent agricultural antibiotics using traditional screening, we found that Borrelidin has high antifungal activity against P. sojae with an IC50 value of 0.0056 μg/mL, which was at least 20-fold lower than those of the antifungal activity of Borrelidin against other phytopathogenic fungi (including Metalaxyl). And further research had revealed that Borrelidin inhibits P. sojae ThrRS (PsThrRS for short) and Borrelidin is a noncompetitive inhibitor of PsThrRS. To identify Borrelidin binding site of PsThrRS, and to discover more selective and high active lead compouds against PsThrRS using virtual screening, which will be used to control the diseases caused by P. sojae, in this paper, Borrelidin binding site of EThrRS was studied at first by the combination of molecular and site-directed mutagensis. Then, the 3D structure of PsThrRS was generated by homology modeling. Borrelidin binding site of PsThrRS was studied by molecular docking, agnain, because of its great accuracy. To discover more important amino acid residues of PsThrRS involved in Borrelidin binding, each of amino acid residue involved in Borrelidin binding was mutated virtually. And the changes of the binding free energies of mutants were evaluated by docking. At last, to discover the potential PsThrRS inhibitors, two virtual screening methods based on molecular docking and pharmacophore model were performed, respectively. The results were as follows:1) Research on Borrelidin binding site of EThrRSIn this segment,3D structures of EThrRS and Borrelidin were obtained from PDB and NCBI compounds database. Then, Borrelidin was docked into active center of EThrRS using AutoDock Vina. The amino acid residues around Borrelidin, which had interctions with Borrelidin, were viewed using Discovery Studio Viewer 3.5. Docking results showed that all the amino acid residues associated with borrelidin binding were outside but adjacent to the active site, which were consisted of Y313, R363, R375, P424, E458, G459, and K465. It should be noted that the binding free energy of borrelidin (-9.1 kcal/mol) was higher than that of ATP (-9.6 kcal/mol), suggesting that the binding of ATP is easier than that of borrelidin. It means that ATP binds to EThrRS previous to borrelidin, and ATP locates deeper than borrelidin in the active pocket of EThrRS. Therefore, we speculated that borrelidin can induce the cleft closure to block the release of Thr-AMP and PPi, rather than inhibits the binding of ATP and threonine. To find more important residues binding to Borrelidin, the amino acid sequences of ThrRSs from E. coli, Borrelidin-resistant species (M. jannaschii and A. fulgidus), and Borrelidin-susceptible species (S. solfataricus and P. sojae) were aligned. Sequence alignment results showed that Y313, P424, E458, and G459 may be the more important residues involving the binding of borrelidin, and any change of these residues may affect the sensitivity of ThrRS to Borrelidin. In order to verify the results predicted by computer, the mutant P424K, E458A, and G459A was subsequently constructed by using site-directed mutagenesis, respectively, and their activities were compared to that of the wild-type ThrRS by enzyme kinetics (for instance, kcat, and Km) using high performance liquid chromatography (HPLC). Ki was measured by enzyme inhibitory assay to compare the affinity of wild type EThrRS and mutants. And the structures of wild type EThrRS and mutants were compared using circular dichroism. At last, stopped-flow fluorescence was used to compare the apparent first-order rate constants (Kapp) of wild type EThrRS and mutants. The circular dichroism results showed that their secondary structures were not changed comparing to wild type ThrRS. Enzyme activities assay results showed that activities of mutants were not changed comparing to wild type ThrRS, while their affinities reduced and their apparent rate constants decreased. It may suggest that Y313, P424, E458, and G459 may be the more important residues involving the binding of Borrelidin to EThrRS.2) Research on 3D structure of PsThrRSIn this segment, the amino acid sequence of PsThrRS was aligned to the sequence of ThrRSs from PDB.1000 models were generated by Homology modeling using program Modeller 9.10 with the parameter of "very fast molecular dynamics optimization". Sequence alignment results showed that Residues 1-76 and 95-152 of PsThrRS were well aligned with residues 371-445 and 462-525 of H. sapiens ThrRS (4HWT), respectively, while residues 72-110 and 147-744 of PsThrRS were well aligned with residues 4-41 and 42-642 of E. coli ThrRS (1QF6), respectively. So the corresponding structures in H. sapiens and E. coli ThrRS molecules were used as template. Then, the 1000 models were evaluated by the build-in software of Modeller, and top 10 models were selected. After the 10 models were evaluated by program MolProbity 3.15, one of the 10 models was selected. Model 393 was selected as the best model, since it contained the fewest violations in bond angles and lengths, atom overlaps, and phi/psi angles. Ramachandran plot showed that 95.96% (711/742) of all residues were in favored regions, with 99.7% (740/742) in allowed regions. There were only two residues in disallowed regions of Model 393. These results indicated that model 393 was reasonable. To find the potential binding site of PsThrRS, a 20 ns molecular dynamic was performed using gromacs 5.0. And we found that there were not more Borrelidin binding site of PsThrRS, which may suggest that Borrelidin binding site of PsThrRS may be similar to those of EThrRS.3) Research on Borrelidin binding site of PsThrRSIn this segment, Borrelidin was docked into the active center of PsThrRS using AutoDock Vina. Then, all the amino acid residues around Borrelidin within 5 A were viewed using Discovery Studio Viewer 3.5. Docking results showed that Borrelidin binding site of PsThrRS was similar to that of EThrRS, and it was also outside but adjacent to the active site of PsThrRS. So the binding site of PsThrRS was similar to that of EThrRS, too. Residue His-412, Tyr-416, Met-435, Cys-437, Arg-466, Arg-478, Gln-484, Asp-486, Ala-564, Tyr-566, Lys-569, Gln-583, Thr-586, Gln-588, and His-616 were found around Borrelidin within 5 A. And each residue selected was substituted using other 19 residues, respectively. The models of all the mutants were generated using Modeller 9.10 and docked by Borrelidin using AutoDock Vina. The maximum and minimum of binding free energies corresponding to each 19 residue mutants were calculated to compare the change range in each residue. More change range, more important. Results of the amino acid sequence alignment, virtual mutagenesis, and docking revealed that residue Tyr-416, Met-435, Cys-437, Arg-466, Gln-484, Asp-486, Gln-583, Thr-586, Gln-588, His-616, and Ala-618 were more important for Borrelidin binding, since their energy range were not less that 1.5 kcal/mol.4) Virtual screening of PsThrRS inhibitorsIn this segment,2141 Borrelidin derivatives were selected from ZINC database, which contains more than 35 million compounds based on the rule that the compounds should share at least 50% similarity to the structure of Borrelidin. Then,23 compounds with lower binding free energies than Borrelidin were selected using virtual screening based on docking all the 2141 compounds to PsThrRS. To verify the relationship between the affinity of compound with PsThrRS and their structure, all the Borrelidin derivatives were sorted based on their binding free energies and filtered into 32 compounds with different energy. The models of pharmacophore of PsThrRS inhibitors were generated based on the 32 compounds using Discovery Studio 3.5. Hypo 1 was the best model, since it was characterized by the highest cost difference (116.338), best correlation coefficient (0.992), and lowest root-mean-square (RMS) deviation (0.325807), which suggested that Hypo 1 had good performance in prediction. Hypo 1 was used to virtual screening against the 2141 Borrelidin derivatives, again, and previous 23 compounds with high affinities were selected consuming only 35 min, which was just one percent of the time (about 60h) using docking by AutoDock Vina at the same computer. It suggested that virtual screening based on model of pharmacophore Hypo 1 was much faster than that of docking. Therefore, a bigger database drugbank (contained 6858 compounds) was screened using Hypo 1, and 10 compounds were obtained. The 10 compounds were identified again by molecular docking using AutoDock Vina, and 6 compounds were found with the lower binding free energy than Borrelidin. The 10 compounds were not same as previous 23 compounds. It suggested that the hit rate of virtual screening of PsThrRS inhibitors based on Hypo 1 was about 60% (6/10). Moreover, the number of inhibitors obtained from 6858 compounds was not more than that from 2141 Borrelidin derivatives, which may suggest that PsThrRS inhibitors were enriched by searching ZINC database using Borrelidin as a leading compound, and it was because of the special structure of Borrelidin that Borrelidin could inhibit PsThrRS well.In summary, this study elaborated Borrelidin binding site of EThrRS using the combination of bioinformatics and molecular biology; The 3d structure model of PsThrRS was generated using homology modeling at first time; Borrelidin binding site of PsThrRS were elaborated using the combination of molecular dynamics simulation, virtual mutagensis, and molecular docking; PsThrRS potental inhibitors were discovered using docking-based and structure fit virtual screening, respectively, which will be used to control the diseases caused by P. sojae.