Current Computer - Aided Drug Design - Volume 11, Issue 3, 2015

A study is presented on how well possible drug-molecules can be predicted with respect to their function and binding to a selection of neuro-receptors by the use of artificial neural networks. The ligands investigated in this study are chosen to be corresponding to the G protein-coupled receptors μ-opioid, serotonin 2B (5-HT2B) and metabotropic glutamate D5. They are selected due to the availability of pharmacological drug-molecule binding data for these receptors. Feedback and deep belief artificial neural network architectures (NNs) were chosen to perform the task of aiding drugdesign. This is done by training on structural features, selected using a “minimum redundancy, maximum relevance”-test, and testing for successful prediction of categorized binding strength. An extensive comparison of the neural network performances was made in order to select the optimal architecture. Deep belief networks, trained with greedy learning algorithms, showed superior performance in prediction over the simple feedback NNs. The best networks obtained scores of more than 90 % accuracy in predicting the degree of binding drug molecules to the mentioned receptors and with a maximal Matthew's coefficient of 0.925. The performance of 8 category networks (8 output classes for binding strength) obtained a prediction accuracy of above 60 %. After training the networks, tests were done on how well the systems could be used as an aid in designing candidate drug molecules. Specifically, it was shown how a selection of chemical characteristics could give the lowest observed IC50 values, meaning largest bio-effect pr. nM substance, around 0.03-0.06 nM. These ligand characteristics could be total number of atoms, their types etc. In conclusion, deep belief networks trained on drug-molecule structures were demonstrated as powerful computational tools, able to aid in drug-design in a fast and cheap fashion, compared to conventional pharmacological techniques.

Withania somnifera commonly known as Ashwagandha in India is used in many herbal formulations to treat various cardiovascular diseases. The key metabolite of this plant, Withaferin A was analyzed for its molecular mechanism through docking studies on different targets of cardiovascular disease. Six receptor proteins associated with cardiovascular disease were selected and interaction studies were performed with Withaferin A using AutoDock Vina. CORINA was used to model the small molecules and HBAT to compute the hydrogen bonding. Among the six targets, β1- adrenergic receptors, HMG-CoA and Angiotensinogen-converting enzyme showed significant interaction with Withaferin A. Pharmacophore modeling was done using PharmaGist to understand the pharmacophoric potential of Withaferin A. Clustering of Withaferin A with different existing drug molecules for cardiovascular disease was performed with ChemMine based on structural similarity and physicochemical properties. The ability of natural active component, Withaferin A to interact with different receptors associated with cardiovascular disease was elucidated with various modeling techniques. These studies conclusively revealed Withaferin A as a potent lead compound against multiple targets associated with cardiovascular disease.

Ligand bound beta 2 adrenergic receptor (β2AR) crystal structures are in use for screening of compound libraries for identifying inducers and blockers. However, in case of G protein coupled receptors (GPCR), docking and binding energy (BE) calculations are not enough to discriminate agonist and antagonists. Absence of a reliable model for discriminating β2AR antagonist is still a major hurdle. Docking of known antagonists and agonists into activated and ground state β2AR revealed several key intermolecular interactions and H-bonding patterns, which in combination, emerged as a model for prediction of antagonists. Present study identifies an alternative binding orientation, within the binding pocket, for blockers and a minimum grid size to lessen the false positive predictions. Cluster analysis revealed structural variability among the antagonists and a conserved pattern in case of agonists. A combination of docking and structure activity relationship (SAR) model reliably discriminated antagonists. Based on key intermolecular interactions, a set of agonists and antagonists useful to SAR, quantitative structure activity relationship (QSAR) and pharmacophore modeling, has also been proposed for identifying antagonists.

The design of inhibitors specific for one relevant carbonic anhydrase isozyme is the major challenge in the new therapeutic agents development. Comparative computational chemical structure and biological activity relationship studies on a series of carbonic anhydrase II inhibitors, benzenesulfonamide derivatives, bearing pyrimidine moieties are reported in this paper using docking, Linear Interaction Energy (LIE), Metadynamics and Quantitative Structure Activity Relationship (QSAR) methods. The computed binding affinities were compared with the experimental data with the goal to explore strengths and weaknesses of various approaches applied to the investigated carbonic anhydrase/inhibitor system. From the tested methods initially only QSAR showed promising results (R2=0.83-0.89 between experimentally determined versus predicted pKd values.). Possible reasons for this performance were discussed. A modification of the LIE method was suggested which used an alternative LIE-like equation yielding significantly improved results (R2 between the experimentally determined versus the predicted ΔGbind improved from 0.24 to 0.50).

An in-house library of 200 molecules from natural plant products was designed in order to evaluate their binding to Plasmodium ACP enoyl reductase (ENR), a promising biological target for antimalarial chemotherapeutics. The binding site of PfENR was explored computationally and the molecules were docked using AutoDock. Furthermore, the top-ranked scaffolds from docking studies were also compared with known PfENR inhibitors using 3D-QSAR. To this effect, a 3D-QSAR model was derived from a set of experimentally established PfENR inhibitors, using Comparative Molecular Force Field Analysis (CoMFA) and Comparative Molecular Similarity Indices Analysis (CoMSIA). The best optimum CoMFA model exhibited a leave-one-out correlation coefficient (q2) and a noncross- validated correlation coefficient (r2) of 0.630 and 0.911, respectively. The result of this cumulative approach proposed five structurally distinct natural products as potent PfENR inhibitors. This study may lay a stepping stone towards Functional oriented synthesis (FOS) of novel PfENR inhibitors in future.

The activation of tumor suppressor p53 protein through inhibition of its interaction with the oncogenic Murine Double Minute 2 (MDM2) protein presents a novel therapeutic strategy against cancer. Accordingly, several small-molecule inhibitors have been developed that mimic three hydrophobic groups of p53 involved in p53-MDM2 binding and thus block the p53-binding pocket on MDM2. Interestingly, presence of a fourth, solvent-exposed hydrophilic moiety in these MDM2 inhibitors is shown to enhance their binding to MDM2 by protecting the inhibitor-MDM2 binding interface from surrounding solvent. In this context, we hypothesized that vitamin folic acid (FA) may prove to be suitable as the hydrophilic cover for enhancing activity of present MDM2 inhibitors. The proposed conjugation of FA to MDM2 inhibitors may also lead to their enhanced and selective uptake by cancer cells, owing to significantly higher expression of the FA receptors on cancer cells compared to normal cells. Therefore, based on our novel hypothesis we designed FA-conjugated MDM2 inhibitors and investigated their binding with MDM2 protein as well as the FA receptor. Specifically, a molecular modeling approach combining flexible receptor docking and molecular mechanics energy minimization calculations revealed highly favorable interactions of FA-conjugated MDM2 inhibitors with both MDM2 protein and the FA receptor as compared to native crystal ligands. Furthermore, these binding interactions were found to be stable using 50,000 ps molecular dynamics simulations. In summary, the newly-designed molecules of this kind, with better MDM2 target binding and enhanced cellular uptake potential, may prove highly useful against cancer and thus warrant further experimental investigations.

Fern plants are known for their ethnomedicinal applications. Huge amount of fern medicinal plants information is scattered in the form of text. Hence, database development would be an appropriate endeavor to cope with the situation. So by looking at the importance of medicinally useful fern plants, we developed a web based database which contains information about several group of ferns, their medicinal uses, chemical constituents as well as protein/enzyme sequences isolated from different fern plants. Fern ethnomedicinal plant database is an all-embracing, content management web-based database system, used to retrieve collection of factual knowledge related to the ethnomedicinal fern species. Most of the protein/enzyme sequences have been extracted from NCBI Protein sequence database. The fern species, family name, identification, taxonomy ID from NCBI, geographical occurrence, trial for, plant parts used, ethnomedicinal importance, morphological characteristics, collected from various scientific literatures and journals available in the text form. NCBI’s BLAST, InterPro, phylogeny, Clustal W web source has also been provided for the future comparative studies. So users can get information related to fern plants and their medicinal applications at one place. This Fern ethnomedicinal plant database includes information of 100 fern medicinal species. This web based database would be an advantageous to derive information specifically for computational drug discovery, botanists or botanical interested persons, pharmacologists, researchers, biochemists, plant biotechnologists, ayurvedic practitioners, doctors/pharmacists, traditional medicinal users, farmers, agricultural students and teachers from universities as well as colleges and finally fern plant lovers. This effort would be useful to provide essential knowledge for the users about the adventitious applications for drug discovery, applications, conservation of fern species around the world and finally to create social awareness.

Small natural or synthetic peptides have been reported to exhibit potent inhibitory capability against trypsin, some of which were also found to have antibacterial potency. Here, we described a successful application of in silico-in vitro integrated approach to rationally design and optimize bifunctional peptides with both trypsin inhibitory and antimicrobial activities. In the procedure, computer-aided methods including protein docking, peptide redocking, molecular dynamics simulations and binding free energy calculations were employed to model and analyze the intermolecular interaction between human trypsin (hT) and natural trypsin inhibitors (TIs). Based on the modeled hT–TI complex structures a number of promising peptide fragments were derived from the trypsin inhibitory loop of TIs, which were then tested experimentally to determine their inhibitory potency on recombinant hT protein as well as their antibacterial potency against three clinical strains. Consequently, few peptides were found to possess a good profile of trypsin inhibitory and antibacterial bi-functionality. Structural visualization and noncovalent examination of hT complex with a potent peptide revealed that the hydrophobic forces and van der Waals contacts between the peptide nonpolar residues and the hydrophobic pocket around hT active site confer significant stability to the complex architecture, while few specific hydrogen bonds and cation-π interactions at the complex interface contribute to peptide selectivity for hT.

Hepatocellular Carcinoma is the most common primary malignant tumor of the liver. Development of multidrug resistance is the main obstacle to the success of anticancer drugs. In this study, designing and docking study of thiosemicarbazide hybrids with amino acids or peptides against hepatocellular carcinoma was performed since hybrids of biologically active compounds with amino acids or peptides may show target specificity and lower toxicity. All the structures were drawn in 2D platform and converted to the 3D platform using ChemDraw 10.0. Evaluations of ADME properties were done by using QikProp 3.0 to check for the possibility of oral delivery. In silico prediction of LD50 values were performed using Pro-Tox webserver. Interestingly, it was found that conjugation with amino acids decreases toxicity and increases the therapeutic index of thiosemicarbazide. Finally, all the compounds were docked to the crystal structure of the Vascular Endothelial Growth Factor Receptor-2 and Checkpoint kinase-1 utilizing Glide 5.0, Schrödinger 8.5, to understand the interaction of ligands with the receptor. A significant number of derivatives have been found active in both the receptors and also displayed multikinase inhibitory activity similar to Sorafenib, against hepatocellular carcinoma. Further, wet lab synthesis, in vitro ADMET and biological screening studies need to be performed to prove that designed compounds are effective against hepatocellular carcinoma as predicted by molecular modeling. However, as predicted by molecular modeling, the efficacy of designed compounds against hepatocellular carcinoma, needs to be confirmed by wet lab synthesis, in vitro ADMET and biological screening studies.