Screening Active Compounds Against HIV-1 Reverse Transcriptase By Molecular Modeling Technology

Object ive:1.To construct a non-nucleoside HIV-1 reverse transcriptase inhibitors (NNRTIS) database incuding coumarin natural products as the virtual screening reference molecule database by network database technology.2.To conduct the virtual screening of 240,950 compounds from SPECS database including 16062 natural products, SYSU database, TCM Bank database(traditional chinese medicine chemical composition database) and NNRTIS database by molecular docking techniques.3.To detect the biologically activity of 25 candidate compounds by the anti-HIV-1 activity assay method in vitro and verify the predictive ability of the virtual screening method.Methods:This paper studied on the screening of non-nucleoside antive compounds against HIV-1 reverse transcriptase by the following methods:1. Based on the database resource and data mining results, a non-nucleoside HIV-1 reverse transcriptase inhibitors database was established by the web network technology for storing and querying the information of the compounds and the corresponding protein target as the virtual screening reference molecules database. The build of the web database is conducted by the currently popular LAMP website structure, including the Linux operating system, Apache HTTP Server, MySQL database and PHP scripting language.2. A candidate virtual molecule library was builted by SPECS database including 16062 natural products, SYSU database, TCM Bank database (traditional chinese medicine chemical composition database) and NNRTIS database.3. The ligand of the HIV-1 reverse transcriptase crystal structure form the RCSB PDB (Protein Data Bank) database was docked to itself by the molecular docking methods of the four drug design commercial softwares of MOE, Sybyl, Schrodinger, Discovery Studio. According to the docking conformation, the crystal structure resolution and the ligand structure characteristics, the suitable protein crystal structure was chosen to participate in the virtual screening.4. Based on molecular docking technology and the DUD (A Directory of Useful Decoys) database library resources,43 non-nucleoside reverse transcriptase inhibitors and 430 structurally similar non-active compounds were docked to the binding pocket of the HIV-1 reverse transcriptase crystal structure by the docking modules of four drug design commercial softwares. According to the docking scores, the ROC curve model was established and the appropriate docking module was chosen as a virtual screening approach.5. The molecular low-energy conformations of candidate molecules was searched by the energy optimized module of MOE software, and its three-dimensional similarity alignment conformation with the ligand of the protein crystal structure was gotten by WEGA molecular three-dimensional similarity program. The conformation was searched for the low energy conformation and the alignment conformation respectively by Discovery Studio software as the condidate molecule conformation.6. The conformation of the candidate molecules was docked to the binding pocket of HIV-1 reverse transcriptase crystal structure by appropriate molecular docking module. According to the docking scoring statistical analysis, the compound possessing the high score and good interaction model was chosen as the biological test candidate molecule.7. The anti-HIV-1 activity of candidate molecules in vitro was detected by the internationally accepted test methods. The 50% concentration of each compound to inhibit the formation of the synplasm of the lymphocytes cell infected by the HIV-1IIIB C8166 (EC50),50% toxic concentration to the normal C8166 T lymphocytes (CC50), and the therapeutic index TI (EC5o/CC50) was obtained by the method.Results:1. A non-nucleoside HIV-1 reverse transcriptase inhibitors database for storing and querying the information of the compounds and the corresponding protein target was established successfully by the network database technology using the LAMP web structure. The structure information of 1340 compounds and the corresponding 205 target protein were stored by the database. Meanwhile the database can be retrieved by the four methods including the compound name, compound SMILE format, target protein name and target protein PDB ID.2. The candidate molecules library of 240,950 compounds, including 201762 compounds of SPECS database,7097 compounds of SYSU database,30751 compounds of TCM Bank database, and 1340 compounds of NNRTIS database was established.3. The 2ZD1 PDB ID of HIV-1 reverse transcriptase was selected for virtual screening by the analysis and itself docking of the protein crystal structure.4.The ROC curve model was established based on the docking of the 473 compounds from the DUD database. According to the size of the AUC, The Glide SP docking module of Schrodinger drug design software was selected as a virtual screening method.5. The low-energy conformation and the alignment conformation of each small molecule was generated by the MOE software and the WEGA programme packages separately. The conformation was searched by the Discovery Studio software and the 20 conformations of each compound will be obtained.6. The conformation of 240,950 compounds was successfully docked to the binding pocket of the 2ZD1 crystal structure of HIV-1 reverse transcriptase by the Glide SP docking module. By analyzing the screening results,25 compounds was tested by the anti-HIV-1 bioassay in vitro finally.7. Virtual screening results showed that in the 240,950 compounds the score distribution curves of the inhibitors and the unknown activity molecules was obviously seperated and the score mean value was -9 and -7 respectively, however, in the 165 compounds whose score was greater than 11.0, the score distribution curves of the inhibitors and the unknown activity molecules curve was obviously close and the score mean value was -11.32 and -11.38 respectively. These results fully validating the virtual screening method has some predictive capacity and the compounds possessing the high score have the greater possibility to inhibit against the HIV-1 reverse transcriptase.8. Virtual screening results showed that in the 165 compounds whose score was greater than 11.0, the number of the inhibitors was accounting for 27.20%, the Chinese herbal medicine of Inphyllum P whose score ranking 98 from TCM Bank database was reported that the EC50 value of the inhibitory against HIV-1 reverse transcriptase was 0.13uM, these results fully validating the virtual screening method has some reliability and accuracy.9. Virtual screening results show the compounds possessing the higher scores, formed hydrogen bonds with the residues Lys101 and π-π stacking interactions with the residues TYR181, TRP229, and horseshoe shaped docking conformation molecules have the greater possibility to inhibit against the HIV-1 reverse transcriptase.10. In this paper, the 25 compounds was tested by the anti-HIV-1 bioassay in vitro, and the results showed that the 25 compounds have anti-HIV-1 activity, of which the EC50 of the 10 compounds was less than 10ug/ml, and the TI of 1-(4-fluorophenyl)-3-[2-(1H-indol-3-yl)ethyl]thiourea and 6-[[5-(4-chloro anilino)-1,3,4-thiadiazol-2-yl]methoxy]-4-methylchromen-2-one was greater than 10.11. The virtual screening and bioassay test results showed that the docking conformation of 1-(4-fluorophenyl)-3-[2-(1H-indol-3-yl)ethyl]thiourea was well aligned with the ligand of the crystal complex of 2ZD1 and presented the U shaped structure. It formed the hydrogen bonds with the key residue Lys101 and the π-π stacking interactions with the conservative residue Trp229 and the aromatic residue Tyrl81 in the binding pocket of HIV-1 reverse transcriptase were also observed. The bioassay in vitro showed The CC50 value of the compound was greater than 100, its EC50 value was 4.54ug/ml, and its therapeutic index (TI) is greater than 18.35. The docking conformation of 6-[[5-(4-chloroanilino)-1,3,4-thiadiazo1-2-y1]methoxy]-4-methylchromen-2-one was presented the horseshoe shaped structure and well aligned with the ligand in the crystal complex. It formed the hydrogen bonds with the key residue Lys101、Lys103 and the π-π stacking interactions with the conservative residue Trp229 and the aromatic residue Tyr181 in the binding pocket of HIV-1 reverse transcriptase were also observed. The bioassay in vitro showed the CC50 value of the compound was greater than 100, the EC50 was 9.85ug/ml, and its therapeutic index (TI) is greater than 10.15. The conclusion that the virtual screening results are in good agreement with the bioassay experiment fully provides the reliability of the virtual screening methodConclusion:1. From the compound structure, the flexible U and horseshoe shaped structure of non-nucleoside reverse transcriptase inhibitors can adapt to the bingding pocket change of the HIV-1 reverse transcriptase by adjusting its structure. This is a important factor for HIV-1 reverse transcriptase inhibition activity.2. From the residues of the reverse transcriptase, the hibitors formed the hydrogen bonds with HIV-1 reverse transcriptase important residues Lys101 and the π-π stacking interactions with the conserved residues Trp229, the aromatic residues Tyr181. This interaction model make the binding of the inhibitors to the reverse transcriptase more stable.3. From the anti-HIV activity of the compound, the 25 compound molecules selected for the bioassay test have the anti-HIV activity, however the activity was weak except the two compounds whose TI value was greater than 10. This results indicates that the virtual screening method conducted in this paper was the high-throuthput screening method and have the predictive capability and reliability. This method can narrow the scope of screening active compounds and should combine with the molecular dynamics simulation for the depth and accuracy molecules research.4. From the screening methods, the combination of network database, molecular modeling and bioassay test can quickly conduct the active compounds screening. Valuable data resources was provided by Network database resources for virtual screening of compounds. Network database technology can be used to store, query structure information and construct the candidate molecule libraries for molecular modeling. Molecular modeling techniques can be used to predict the interaction model of protein with the compound and reduce the scope of the active compound screening. In vitro anti-HIV-1 activity assay can quickly verify the ability to predict the virtual screening method.