Tool For Drug Design Research Articles

Antimalarial Agents Derived from Metal-Amodiaquine Complexes with Activity in Multiple Stages of the Plasmodium Life Cycle†.

Quinolines are excellent ligands for metal coordination and are deployed as drugs for malaria treatment. There is a growing body of evidence indicating that metal complexes can be conjugated with antimalarial quinolines to be used as chemical tools to overcome the disadvantages of quinolines, improving their bioactive speciation, cellular distribution, and subsequently broadening the spectrum of activity to multiple stages of the complex Plasmodium life cycle. In this study, four novel complexes of ruthenium(II) and gold(I) containing amodiaquine (AQ) were synthesized and a careful chemical characterization revealed the precise coordination site of AQ to the metals. Their speciation in solution was investigated, demonstrating the stability of the quinoline-metal bond. Ru(II)- and Au(I)-AQ complexes were demonstrated to be potent and efficacious in inhibiting parasite growth in multiple stages of the Plasmodium life cycle assayed in vitro and in vivo. These properties could be attributed to the ability of the metal-AQ complexes to reproduce the suppression of heme detoxification induced by AQ, while also inhibiting other processes in the parasite life cycle, which can be attributed to the action of the metallic species. These findings indicate that metal coordination is a potential chemical tool for drug design and discovery in malaria.

DCAMCP: A deep learning model based on capsule network and attention mechanism for molecular carcinogenicity prediction.

The carcinogenicity of drugs can have a serious impact on human health, so carcinogenicity testing of new compounds is very necessary before being put on the market. Currently, many methods have been used to predict the carcinogenicity of compounds. However, most methods have limited predictive power and there is still much room for improvement. In this study, we construct a deep learning model based on capsule network and attention mechanism named DCAMCP to discriminate between carcinogenic and non-carcinogenic compounds. We train the DCAMCP on a dataset containing 1564 different compounds through their molecular fingerprints and molecular graph features. The trained model is validated by fivefold cross-validation and external validation. DCAMCP achieves an average accuracy (ACC) of 0.718 ± 0.009, sensitivity (SE) of 0.721 ± 0.006, specificity (SP) of 0.715 ± 0.014 and area under the receiver-operating characteristic curve (AUC) of 0.793 ± 0.012. Meanwhile, comparable results can be achieved on an external validation dataset containing 100 compounds, with an ACC of 0.750, SE of 0.778, SP of 0.727 and AUC of 0.811, which demonstrate the reliability of DCAMCP. The results indicate that our model has made progress in cancer risk assessment and could be used as an efficient tool in drug design.

Small molecule antagonists of PTPmu identified by artificial intelligence-based computational screening block glioma cell migration and growth.

PTPmu (PTPμ) is a member of the receptor protein tyrosine phosphatase IIb family that participates in both homophilic cell-cell adhesion and signaling. PTPmu is proteolytically downregulated in glioblastoma generating extracellular and intracellular fragments that have oncogenic activity. The intracellular fragments, in particular, are known to accumulate in the cytoplasm and nucleus where they interact with inappropriate binding partners/substrates generating signals required for glioma cell migration and growth. Thus, interfering with these fragments is an attractive therapeutic strategy. To develop agents that target these fragments, we used the AI-based AtomNetⓇ model, a drug design and discovery tool, to virtually screen molecular libraries for compounds able to target a binding pocket bordered by the wedge domain, a known regulatory motif located within the juxtamembrane portion of the protein. Seventy-four high-scoring and chemically diverse virtual hits were then screened in multiple cell-based assays for effects on glioma cell motility (scratch assays) and growth in 3D culture (sphere assays), and PTPmu-dependent adhesion (Sf9 aggregation). We identified three inhibitors (247678835, 247682206, 247678791) that affected the motility of multiple glioma cell lines (LN229, U87MG, and Gli36delta5), the growth of LN229 and Gli36 spheres, and PTPmu-dependent Sf9 aggregation. Compound 247678791 was further shown to suppress PTPmu enzymatic activity in an in vitro phosphatase assay, and 247678835 was able to inhibit the growth of human glioma tumors in mice. We propose that these three compounds are PTPmu-targeting agents with therapeutic potential for treating glioblastoma.

FFLOM: A Flow-Based Autoregressive Model for Fragment-to-Lead Optimization.

Recently, deep generative models have been regarded as promising tools in fragment-based drug design (FBDD). Despite the growing interest in these models, they still face challenges in generating molecules with desired properties in low data regimes. In this study, we propose a novel flow-based autoregressive model named FFLOM for linker and R-group design. In a large-scale benchmark evaluation on ZINC, CASF, and PDBbind test sets, FFLOM achieves state-of-the-art performance in terms of validity, uniqueness, novelty, and recovery of the generated molecules and can recover over 92% of the original molecules in the PDBbind test set (with at least five atoms). FFLOM also exhibits excellent potential applicability in several practical scenarios encompassing fragment linking, PROTAC design, R-group growing, and R-group optimization. In all four cases, FFLOM can perfectly reconstruct the ground-truth compounds and generate over 74% of molecules with novel fragments, some of which have higher binding affinity than the ground truth.

Docking Studies on Biomolecules from Marine Microalga Skeletonema costatum Against Hemolysin Protein of Bioluminescence Disease-Causing Vibrio harveyi.

Grow-out and hatchery units of shrimps are being impacted by disease-causing bacterial pathogens and predominantly marine Vibrios. The use of chemicals for governing bacterial pathogens in the aquaculture practices is developing resistance to bacteria. Henceforth, the application of bio-therapeutic agents from marine resources for controlling pathogens is most vital to be considered. Molecular docking is computer-assisted drug design tool to detect and counteract for drug-receptor interaction for known target protein of diseases. Therefore, an effort was made with the extract of the marine micro alga Skeletonema costatum against hemolysin protein of pathogenic bacteria Vibrio harveyi. The extract of S. costatum was tested against growth and virulence produced by V. harveyi during larviculture of Penaeus monodon. The extract was analyzed for phyto-constituents through GC-MS and used them as ligand molecule in docking. S. costatum extract at 200µgmL-1 was found to decrease 35.20% of cumulative percentage mortality (CPM) in postlarvae of P. monodon against V. harveyi infections. The biomolecule Docasane, an alkane from the extract of S. costatum, exposed highest binding interaction than other compounds during docking analysis. The level of significance (P < 0.05) was found in CPM, growth, and virulence factors of V. harveyi studies. Thus, the present finding predicts that extract of S. costatum containing biomolecules can be recommended for use in the shrimp culture-based grow-out and hatchery units for eliminating bioluminescent V. harveyi.

LS-MolGen: Ligand-and-Structure Dual-Driven Deep Reinforcement Learning for Target-Specific Molecular Generation Improves Binding Affinity and Novelty.

Deep learning-based molecular generative models have garnered considerable interest in the field of de novo drug design. However, most extant models focus on either ligand-based or structure-based strategies, thereby failing to effectively harness the combined knowledge derived from both ligands and the structure of the binding target. In this article, we introduce LS-MolGen, a novel ligand and structure-integrated molecular generative model. This model synergistically combines representation learning, transfer learning, and reinforcement learning. The targeted knowledge assimilation from transfer learning, coupled with an advanced exploration strategy in reinforcement learning, empowers LS-MolGen to efficiently generate novel and high-affinity molecules efficiently. The comparable performance of our model is affirmed through multiple evaluations, including EGFR, DRD3, CDK2, AA2AR, ADRB2, and a dedicated case study of inhibitor design for SARS-CoV-2 Mpro. The results indicate that LS-MolGen performs better than other ligand-based or structure-based generative models in de novo designing promising compounds with novel scaffolds and high binding affinity. This proof-of-concept study signifies the potential of our ligand- and structure-based generative model, LS-MolGen, as a promising new tool for target-specific molecular generation and drug design.

DrugEx: Deep Learning Models and Tools for Exploration of Drug-Like Chemical Space.

The discovery of novel molecules with desirable properties is a classic challenge in medicinal chemistry. With the recent advancements of machine learning, there has been a surge of de novo drug design tools. However, few resources exist that are user-friendly as well as easily customizable. In this application note, we present the new versatile open-source software package DrugEx for multiobjective reinforcement learning. This package contains the consolidated and redesigned scripts from the prior DrugEx papers including multiple generator architectures, a variety of scoring tools, and multiobjective optimization methods. It has a flexible application programming interface and can readily be used via the command line interface or the graphical user interface GenUI. The DrugEx package is publicly available at https://github.com/CDDLeiden/DrugEx.

QSPR Analysis of Some Novel Drugs Used in Blood Cancer Treatment Via Degree Based Topological Indices and Regression Models

Topological indices (TIs) have several biological applications in the treatment of blood cancer. TIs can be used to predict the efficacy of drugs in cancer treatment by providing information about the molecular structures of the drugs and their related properties. By analyzing the degree-based TIs of different drugs, researchers can identify the most effective drugs for treating specific types of blood cancer. Additionally, TIs can also be used to predict the toxicity of drugs, which is essential for developing safe and effective treatments. Furthermore, TIs can aid in the discovery and design of new drugs by providing insights into the relationship between molecular structure and biological activity. This article focuses on the applications of topological indices (TIs) in predicting the physical and biological properties of novel drugs used in the treatment of blood cancer. The drugs glasdegib, palbociclib, and daunorubicin are analyzed using degree-based topological indices computed through edge partitioning. A quantitative structure-property relationship (QSPR) model is then developed using linear regression to predict the properties such as boiling point (BP), flash point (FP), molar volume (MV), molecular weight, and complexity. The results demonstrate the potential of topological indices as a tool for drug discovery and design in the field of cancer treatment.

Understanding and targeting resistance mechanisms in cancer.

Resistance to cancer therapies has been a commonly observed phenomenon in clinical practice, which is one of the major causes of treatment failure and poor patient survival. The reduced responsiveness of cancer cells is a multifaceted phenomenon that can arise from genetic, epigenetic, and microenvironmental factors. Various mechanisms have been discovered and extensively studied, including drug inactivation, reduced intracellular drug accumulation by reduced uptake or increased efflux, drug target alteration, activation of compensatory pathways for cell survival, regulation of DNA repair and cell death, tumor plasticity, and the regulation from tumor microenvironments (TMEs). To overcome cancer resistance, a variety of strategies have been proposed, which are designed to enhance the effectiveness of cancer treatment or reduce drug resistance. These include identifying biomarkers that can predict drug response and resistance, identifying new targets, developing new targeted drugs, combination therapies targeting multiple signaling pathways, and modulating the TME. The present article focuses on the different mechanisms of drug resistance in cancer and the corresponding tackling approaches with recent updates. Perspectives on polytherapy targeting multiple resistance mechanisms, novel nanoparticle delivery systems, and advanced drug design tools for overcoming resistance are also reviewed.

Deep Generation Model Guided by the Docking Score for Active Molecular Design.

A deep generation model, as a novel drug design and discovery tool, shows obvious advantages in generating compounds with novel backbones and has been applied successfully in the field of drug discovery. However, it is still a challenge to generate molecules with expected properties, especially high activity. Here, to obtain compounds both with novelty and high activity to a target, we proposed a conditional molecular generation model COMG by considering the docking score and 3D pharmacophore matching during molecular generation. The proposed model was based on the conditional variational autoencoder architecture constrained by the pharmacophore matching score. During Bayesian optimization, the docking score was applied to enhance the target relevance of generated compounds. Furthermore, to overcome the problem of high structural similarity caused by Bayesian optimization, the idea of the scaffold memory unit was also introduced. The evaluation results of COMG show that our model not only can improve the structural diversity of generated molecules but also can effectively improve the proportion of target-related drug-active molecules. The obtained results indicate that our proposed model COMG is a useful drug design tool.

Modeling the SARS-CoV-2 receptor ACE2 indicates that protein-bound Zn affects distant protein stability and interactions

Abstract Molecular Dynamics (MD) is a widely used drug design tool capable of rapidly screening the binding capacity of many candidate inhibitors in a cost-effective manner. This approach has been leveraged to search for molecules that can bind to angiotensin converting enzyme 2 (ACE2) and prevent SARS-CoV-2 viral entry to host cells. However, ACE2 is difficult to model due to an embedded zinc ion (Zn), which introduces complex charge transfer and polarization. Since the embedded Zn is distant from the site of ACE2 binding, many groups have assumed Zn does not impact protein interactions and have excluded Zn while modeling ACE2. Here, we evaluated the impact of Zn by performing MD simulations of Zn-bound and Zn-free versions of the ACE2-spike protein (S1) and ACE2-monoclonal antibody (mAb) systems . We found that excluding Zn had a significant effect on total protein stability and the stability of interacting residues, indicating that Zn can impact distant protein sites. Additionally, we discovered that including protein-bound Zn in MD simulations improved the binding free energy of both systems by −3.26 and −14.8 kcal/mol in the ACE2-S1 and ACE2-mAb systems, respectively. These data suggest that excluding Zn may alter inhibitor selection decisions and should be included for accurate modeling. Interestingly, ACE2-mAb is particularly sensitive to changes in receptor structure, likely reflecting its generation against in vivo Zn-bound ACE2. Collectively, our data demonstrates that excluding Zn significantly impacts MD simulation outcomes which inform inhibitor selection, particularly when modeling receptor-mAb complexes.

Pharmacophore Modeling Guided by Conformational Dynamics Reveals Potent Anticancer Agents

Targeting the interaction between tumor suppressor p53 and murine double minute 2(MDM2) has been an attractive therapeutic strategy of recent cancer research. There are a few number of MDM2-targeted anticancer drug molecules undergoing clinical trials, yet none of them have been approved so far. In this study, a new approach is employed in which dynamics of MDM2 obtained by elastic network models are used as a guide in the generation of the ligand-based pharmacophore model prior to virtual screening. Hit molecules exhibiting high affinity to MDM2 were captured and tested by rigid and induced-fit molecular docking. The knowledge of the binding mechanism was used while creating the induced-fit docking criteria. Application of Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) method provided an accurate prediction of the binding free energy values. Two leading hit molecules which have shown better docking scores, binding free energy values and drug-like molecular properties were identified. These hits exhibited extra intermolecular interactions with MDM2, indicating a stable complex formation and hence would be further tested in vitro. Finally, the combined computational strategy employed in this study can be a promising tool in drug design for the discovery of potential new hits.

Prediction of KRASG12C inhibitors using conjoint fingerprint and machine learning-based QSAR models

Kirsten rat sarcoma virus G12C (KRASG12C) is the major protein mutation associated with non-small cell lung cancer (NSCLC) severity. Inhibiting KRASG12C is therefore one of the key therapeutic strategies for NSCLC patients. In this paper, a cost-effective data driven drug design employing machine learning-based quantitative structure-activity relationship (QSAR) analysis was built for predicting ligand affinities against KRASG12C protein. A curated and non-redundant dataset of 1033 compounds with KRASG12C inhibitory activity (pIC50) was used to build and test the models. The PubChem fingerprint, Substructure fingerprint, Substructure fingerprint count, and the conjoint fingerprint—a combination of PubChem fingerprint and Substructure fingerprint count—were used to train the models. Using comprehensive validation methods and various machine learning algorithms, the results clearly showed that the XGBoost regression (XGBoost) achieved the highest performance in term of goodness of fit, predictivity, generalizability and model robustness (R2 = 0.81, Q2CV = 0.60, Q2Ext = 0.62, R2 – Q2Ext = 0.19, R2Y-Random = 0.31 ± 0.03, Q2Y-Random = −0.09 ± 0.04). The top 13 molecular fingerprints that correlated with the predicted pIC50 values were SubFPC274 (aromatic atoms), SubFPC307 (number of chiral-centers), PubChemFP37 (≥1 Chlorine), SubFPC18 (Number of alkylarylethers), SubFPC1 (number of primary carbons), SubFPC300 (number of 1,3-tautomerizables), PubChemFP621 (N–C:C:C:N structure), PubChemFP23 (≥1 Fluorine), SubFPC2 (number of secondary carbons), SubFPC295 (number of C-ONS bonds), PubChemFP199 (≥4 6-membered rings), PubChemFP180 (≥1 nitrogen-containing 6-membered ring), and SubFPC180 (number of tertiary amine). These molecular fingerprints were virtualized and validated using molecular docking experiments. In conclusion, this conjoint fingerprint and XGBoost-QSAR model demonstrated to be useful as a high-throughput screening tool for KRASG12C inhibitor identification and drug design.

QSAR via Multisite λ-Dynamics in the Orphaned TSSK1B Kinase.

Multisite λ-dynamics (MSλD) is a novel method for the calculation of relative free energies of binding for ligands to their targeted receptors. It can be readily used to examine a large number of molecules with multiple functional groups at multiple sites around a common core. This makes MSλD a powerful tool in structure-based drug design. In the present study, MSλD is applied to calculate the relative binding free energies of 1296 inhibitors to the testis specific serine kinase 1B (TSSK1B), a validated target for male contraception. For this system, MSλD requires significantly fewer computational resources compared to traditional free energy methods like free energy perturbation or thermodynamic integration. From MSλD simulations, we examined whether modifications of a ligand at two different sites are coupled or not. Based on our calculations, we established a quantitative structure-activity relationship (QSAR) for this set of molecules and identified a site in the ligand where further modification, such as adding more polar groups, may lead to increased binding affinity. This article is protected by copyright. All rights reserved.

Steered Molecular Dynamics Simulations Study on FABP4 Inhibitors

Ordinary small molecule de novo drug design is time-consuming and expensive. Recently, computational tools were employed and proved their efficacy in accelerating the overall drug design process. Molecular dynamics (MD) simulations and a derivative of MD, steered molecular dynamics (SMD), turned out to be promising rational drug design tools. In this paper, we report the first application of SMD to evaluate the binding properties of small molecules toward FABP4, considering our recent interest in inhibiting fatty acid binding protein 4 (FABP4). FABP4 inhibitors (FABP4is) are small molecules of therapeutic interest, and ongoing clinical studies indicate that they are promising for treating cancer and other diseases such as metabolic syndrome and diabetes.

Deep Learning Model for Efficient Protein-Ligand Docking with Implicit Side-Chain Flexibility.

Protein-ligand docking is an essential tool in structure-based drug design with applications ranging from virtual high-throughput screening to pose prediction for lead optimization. Most docking programs for pose prediction are optimized for redocking to an existing cocrystallized protein structure, ignoring protein flexibility. In real-world drug design applications, however, protein flexibility is an essential feature of the ligand-binding process. Flexible protein-ligand docking still remains a significant challenge to computational drug design. To target this challenge, we present a deep learning (DL) model for flexible protein-ligand docking based on the prediction of an intermolecular Euclidean distance matrix (EDM), making the typical use of iterative search algorithms obsolete. The model was trained on a large-scale data set of protein-ligand complexes and evaluated on independent test sets. Our model generates high quality poses for a diverse set of protein and ligand structures and outperforms comparable docking methods.

EXTRACTION, PURIFICATION, AND CRYSTALLIZATION OF GTPASE ERA FROM STAPHYLOCOCCUS AUREUS

Protein crystal structure studies are an important tool for drug design. The growth of high-quality crystals suitable for X-ray diffraction is the limiting factor and the bottleneck in obtaining the structural data. Here we report the extraction, purification, and crystallization of the protein GTPase Era from the pathogenic bacterium Staphylococcus aureus. In bacterial cells, GTPase Era acts as a ribosome assembly factor. This enzyme is responsible for the cell growth and division. However, its structure is poorly understood. We obtained crystals of Staphylococcus aureus GTPase Era, which can be used in further structural studies by single-crystal X-ray diffraction analysis.

Computer-Aided Drug Design Project for Introductory High School Students

The pharmaceutical industry has been a major player in improving life expectancies and driving chemistry employment. While computational chemistry tools increasingly guide pharmaceutical research decisions, high school and undergraduate chemistry curricula do not reflect the prevalence of such techniques. Incorporation of computational tools into chemistry education is warranted to support industry-relevant learning experiences, student motivation to pursue chemistry careers, and public understanding of STEM careers and medicines. Therefore, this paper outlines how a computer-aided drug design (CADD) tool was introduced to high school students as a classroom project in tandem with traditional course material. Materials are provided to assist instructors in replicating this project in their own courses.

SHAMAN: A mixed solvent molecular dynamics protocol to identify small-molecules binding sites on RNA structural ensembles.

The majority of in silico binding site identification tools for computational drug design employs static structures of the target. However, such structures do not contain information about transient opening of cavities that could act as binding sites, a phenomenon often enhanced by the presence of a ligand. To address this limitation, a technique that makes use of Molecular Dynamics (MD) simulations of the target with selected co-solvent probes was introduced. Free-energy grids built from the probes occupancy provide an effective description of transiently formed pockets not observed in the resolved conformation. This framework was applied only very recently to highly flexible RNA molecules. Here, we present a protocol to detect binding sites on RNAs, called SHAdow mixed-solvent MetAdyNamics (SHAMAN). In SHAMAN, a “mother” unbiased MD simulation of a target RNA is run in parallel with “daughter” simulations, one for each selected probe. RNAs in the daughter simulations are restrained, like shadows, to explore the conformational space of the mother and, concurrently, the sampling of the probes is accelerated by Metadynamics. SHAMAN was tested on a set of molecules of therapeutic interest, composed of riboswitches and viral RNAs, such as HIV1 TAR, IAV promotor and IRES RNAs. Selected probes included aromatic rings, like benzene and imidazole, and other moieties, like formamide and acetate. Starting from both ligand-bound and apo systems, we were able to detect experimentally resolved pockets and to rank them among the most populated by at least one probe. SHAMAN enables identifying pockets in RNA dynamic conformational ensembles and opens new possibilities for fragment-based drug design on flexible RNA using ensemble virtual screening techniques. The generic version of the shadowing technique is implemented in the open-source PLUMED library, freely available at www.plumed.org.

Prediction of antioxidant peptides using a quantitative structure−activity relationship predictor (AnOxPP) based on bidirectional long short-term memory neural network and interpretable amino acid descriptors

Antioxidant peptides can protect against free radical-mediated diseases, especially food-derived antioxidant peptides are considered as potential competitors among synthetic antioxidants due to their safety, high activity and abundant sources. However, wet experimental methods can not meet the need for effectively screening and clearly elucidating the structure-activity relationship of antioxidant peptides. Therefore, it is particularly important to build a reliable prediction platform for antioxidant peptides. In this work, we developed a platform, AnOxPP, for prediction of antioxidant peptides using the bidirectional long short-term memory (BiLSTM) neural network. The sequence characteristics of peptides were converted into feature codes based on amino acid descriptors (AADs). Our results showed that the feature conversion ability of the combined-AADs optimized by the forward feature selection method was more accurate than that of the single-AADs. Especially, the model trained by the optimal descriptor SDPZ27 significantly outperformed the existing predictor on two independent test sets (Accuracy = 0.967 and 0.819, respectively). The SDPZ27-based AnOxPP learned four key structure-activity features of antioxidant peptides, with the following importance as steric properties > hydrophobic properties > electronic properties > hydrogen bond contributions. AnOxPP is a valuable tool for screening and design of peptide drugs, and the web-server is accessible at http://www.cqudfbp.net/AnOxPP/index.jsp.