Machine learning-based drug design for identification of thymidylate kinase inhibitors as a potential anti-Mycobacterium tuberculosis

Artificial Intuition and accelerating the process of antimicrobial drug discovery

New drug development is a very challenging, expensive, and usually time-consuming process. This issue is very important with regard to antimicrobials, which are affected by the global issue of the development and spread of resistance. This framework underscores the urgency of accelerating drug development processes while reducing their costs. In this context, new bioinformatics tools can provide important support for drug development by limiting and shortening in vitro evaluation of the best outcomes, thereby minimizing costs. Recently, new Artificial Intelligence (AI)-based tools have been developed for de novo design of new molecules, or for the identification of features of inhibitors among a large set of molecules that can guide rational design. With this work, we present an Artificial Intuition (AI4)-based pharmacological analysis of a series of antimicrobial compounds that are known to be active against Mycobacterium tuberculosis. The compounds have been subjected to Molecular Dynamic Simulation (MDS), and the respective outputs processed with a Quantitative Complexity Management (QCM) tool in order to determine the corresponding complexity profiles. The comparison of different analogues in each series revealed a relationship between the complexity of the various chemical moieties and their importance for the biological activity of each compound, suggesting that QCM may be a useful tool in guiding the optimization process. This first attempt to apply the tool in the field of drug development has yielded interesting results, indicating that QCM, which powers AI4, can be implemented for rational drug design in the near future.

Explainable AI-driven prediction of APE1 inhibitors: enhancing cancer therapy with machine learning models and feature importance analysis

The viability of cells and the integrity of the genome depend on the detection and repair of damaged DNA through intricate mechanisms. Cancer treatment employs chemotherapy or radiation therapy to eliminate neoplastic cells by causing substantial damage to their DNA. In many cases, improved DNA repair mechanisms lead to resistance to these medicines; therefore, it is essential to expand efforts to develop drugs that can sensitise cells to these treatments by inhibiting the DNA repair process. Multiple studies have demonstrated a correlation between the overexpression of Apurinic/Apyrimidinic Endonuclease (APE1), the primary mammalian enzyme responsible for excising apurinic or apyrimidinic sites in DNA, and the resistance of cells to cancer therapies; in contrast, APE1 downregulation increases cellular susceptibility to DNA-damaging agents. Thus, the effectiveness of existing therapies can be improved by promoting the targeted sensitization of cancer cells while protecting healthy cells. The current study aims to employ explainable artificial intelligence (XAI) to enhance the accuracy and reliability of machine learning models for the prediction of APE1 inhibitors. Various ML-based regression models are employed to predict the pIC50 value of different medicines. Bayesian optimization and the Permutation Feature Importance (PFI) approach are employed to determine the best hyperparameters of machine learning models and to discover the most significant features for recognizing drug candidates that target APE1 enzymes, respectively. To acquire comprehensive elucidations for the predictive models in our research, two XAI methodologies, namely SHAP and LIME, are used. The SHAP analysis reveals that the features 'C1SP2' and 'ASP-2' are essential in influencing the model's predictions. The SHAP values demonstrate variability for features such as 'maxHBint2' and 'GATS1s,' signifying that their impact is dependent on specific instances within the dataset. The LIME study corroborates these findings, demonstrating that 'C1SP2' and 'ASP-2' are the most significant positive contributors, whereas features like 'SHCHnX,' 'nHdCH2,' and 'GATS1s' result in a decrease in the predicted values. Due to the limited sample size of the APE1 dataset, direct training on this dataset posed challenges in model generalization and reliability. To overcome this limitation, the BACE-1 dataset is leveraged for model training, enabling the ML models to learn from a more extensive and diverse chemical space. Among the tested algorithms, XGBoost demonstrated superior predictive performance, achieving R2 = 0.890, MAE = 0.186, and RMSE = 0.245, significantly surpassing state-of-the-art methods, such as LightGBM and QSAR-ML, which attained R2 scores of 0.798 and 0.630, respectively. These results highlight the robustness of our approach, demonstrating its enhanced generalization capability and superior predictive accuracy compared to existing methodologies.

Identification of mycobacterial Thymidylate kinase inhibitors: a comprehensive pharmacophore, machine learning, molecular docking, and molecular dynamics simulation studies

Thymidylate kinase (TMK) is a pivotal enzyme in Mycobacterium tuberculosis (Mtb), crucial for phosphorylating thymidine monophosphate (dTMP) to thymidine diphosphate (dTDP), thereby playing a critical role in DNA biosynthesis. Dysregulation or inhibition of TMK activity disrupts DNA replication and cell division, making it an attractive target for anti-tuberculosis drug development. In this study, the statistically validated pharmacophore mode was developed from a set of known TMK inhibitors. Further, the robust pharmacophore was considered for screening the Enamine database. The chemical space was reduced through multiple molecular docking approaches, pharmacokinetics, and absolute binding energy estimation. Two different molecular docking algorithms favor the strong binding affinity of the proposed molecules towards TMK. Machine learning-based absolute binding energy also showed the potentiality of the proposed molecules. The binding interactions analysis exposed the strong binding affinity between the proposed molecules and active site amino residues of TMK. Several statistical parameters from all atoms MD simulation explained the stability between proposed molecules and TMK in the dynamic states. The MM-GBSA approach also found a strong binding affinity for each proposed molecule. Therefore, the proposed molecules might be crucial TMK inhibitors for managing Mtb inhibition subjected to in vitro/in vivo validations.

Identification of new potent NLRP3 inhibitors by multi-level in-silico approaches

Nod-like receptor protein 3 (NLRP-3), is an intracellular sensor that is involved in inflammasome activation, and the aberrant expression of NLRP3 is responsible for diabetes mellitus, its complications, and many other inflammatory diseases. NLRP3 is considered a promising drug target for novel drug design. Here, a pharmacophore model was generated from the most potent inhibitor, and its validation was performed by the Gunner-Henry scoring method. The validated pharmacophore was used to screen selected compounds databases. As a result, 646 compounds were mapped on the pharmacophore model. After applying Lipinski's rule of five, 391 hits were obtained. All the hits were docked into the binding pocket of target protein. Based on docking scores and interactions with binding site residues, six compounds were selected potential hits. To check the stability of these compounds, 100 ns molecular dynamic (MD) simulations were performed. The RMSD, RMSF, DCCM and hydrogen bond analysis showed that all the six compounds formed stable complex with NLRP3. The binding free energy with the MM-PBSA approach suggested that electrostatic force, and van der Waals interactions, played a significant role in the binding pattern of these compounds. Thus, the outcomes of the current study could provide insights into the identification of new potential NLRP3 inflammasome inhibitors against diabetes and its related disorders.

Machine learning and molecular simulation-based protocols to identify novel potential inhibitors for reverse transcriptase against HIV infections

Acquired immunodeficiency syndrome (AIDS) is a potentially fatal condition affecting the human immune system, which is attributed to the human immunodeficiency virus (HIV). The suppression of reverse transcriptase activity is a promising and feasible strategy for the therapeutic management of AIDS. In this study, we employed machine learning algorithms, such as support vector machines (SVM), k-nearest neighbor (k-NN), random forest (RF), and Gaussian naive base (GNB), which are fast and effective tools commonly used in drug design. For model training, we initially obtained a dataset of 5,159 compounds from BindingDB. The models were assessed using tenfold cross-validation to ensure their accuracy and reliability. Among these compounds, 1,645 compounds were labeled as active, having an IC50 below 0.49 µM, while 3,514 compounds were labeled "inactive against reverse transcriptase. Random forest achieved 86% accuracy on the train and test set among the different machine learning algorithms. Random forest model was then applied to an external ZINC dataset. Subsequently, only three hits-ZINC1359750464, ZINC1435357562, and ZINC1545719422-were selected based on the Lipinski Rule, docking score, and good interaction. The stability of these molecules was further evaluated by deploying molecular dynamics simulation and MM/GBSA, which were found to be -38.6013 ± 0.1103 kcal/mol for the Zidovudine/RT complex, -59.1761 ± 2.2926 kcal/mol for the ZINC1359750464/RT complex, -47.6292 ± 2.4206 kcal/mol for the ZINC1435357562/RT complex, and -50.7334 ± 2.5713 kcal/mol for the ZINC1545719422/RT complex.

Exploring potent aldose reductase inhibitors for anti-diabetic (anti-hyperglycemic) therapy: integrating structure-based drug design, and MMGBSA approaches

Aldose reductase (AR) is an important target in the development of therapeutics against hyper-glycemia-induced health complications such as retinopathy, etc. In this study, we employed a combination of structure-based drug design, molecular simulation, and free energy calculation approaches to identify potential hit molecules against anti-diabetic (anti-hyperglycemic)-induced health complications. The 3D structure of aldoreductase was screened for multiple compound libraries (1,00,000 compounds) and identified as ZINC35671852, ZINC78774792 from the ZINC database, Diamino-di nitro-methyl dioctyl phthalate, and Penta-o-galloyl-glucose from the South African natural compounds database, and Bisindolylmethane thiosemi-carbazides and Bisindolylme-thane-hydrazone from the Inhouse database for this study. The mode of binding interactions of the selected compounds later predicted their aldose reductase inhibitory potential. These com-pounds interact with the key active site residues through hydrogen bonds, salt bridges, and π-π interactions. The structural dynamics and binding free energy results further revealed that these compounds possess stable dynamics with excellent binding free energy scores. The structures of the lead inhibitors can serve as templates for developing novel inhibitors, and in vitro testing to confirm their anti-diabetic potential is warranted. The current study is the first to design small molecule inhibitors for the aldoreductase protein that can be used in the development of therapeutic agents to treat diabetes.

Structure-based virtual screening, molecular simulation and free energy calculations of traditional Chinese medicine, ZINC database revealed potent inhibitors of estrogen-receptor α (ERα)

Breast Cancer, a heterogeneous disease at the molecular level, is the most common cause of woman mortality worldwide. We used molecular screening and simulation approaches to target nuclear receptor protein-estrogen receptor alpha (Erα) protein to design and develop of specific and compelling drugs from traditional Chinese medicine (TCM), and ZINC database against pathophysiology of breast cancer. Using virtual screening, only six hits TCM22717, TCM23524, TCM31953, while ZINC05632920, ZINC05773243, and ZINC12780336 demonstrated better pharmacological potential than the 4-hydroxytamoxifen (OHT) taken as control. Binding mode of each of the top hit revealed that these compounds could block the main active site residues and block the function of Erα protein. Moreover, molecular simulation revealed that the identified compounds exhibit stable dynamics and may induce stronger therapeutic effects in experimental setup. All the complexes reported tighter structural packing and less flexible behaviour. We found that the average hydrogen bonds in the identified complexes remained higher than the control drug. Finally, the total binding free energy demonstrated the best hits among the all. The BF energy results revealed -30.4525 ± 3.3565 for the 4-hydroxytamoxifen (OHT)/Erα complex, for the TCM22717/Erα -57.0597 ± 3.4852 kcal/mol, for the TCM23524/Erα complex the BF energy was -56.9084 ± 3.3737 kcal/mol, for the TCM31953/Erα the BF energy was -32.4191 ± 3.8864 kcal/mol while for the ZINC05632920/Erα complex -46.3182 ± 2.7380, ZINC05773243/Erα complex -38.3690 ± 2.8240, and ZINC12780336/Erα complex the BF energy was calculated to be -35.8048 ± 4.1571 kcal/mol.Communicated by Ramaswamy H. Sarma.

Napthyridine-derived compounds as promising inhibitors for Staphylococcus aureus CrtM: a primer for the discovery of potential anti-Staphylococcus aureus agents

The disease-free existence of humans is constantly under attack by a variety of infections caused by a variety of organisms including bacteria. Notable among the bacteria is Staphylococcus aureus which is an etiological organism for infections including impetigo, folliculitis, and furuncles. The response of the human immune system against this disease is often neutralized by the production of a pigment called Staphyloxanthin (STX) via a series of reactions mediated by several enzymes. Among these enzymes, dehydrosqualene synthase, also known as CrtM, has emerged as a viable drug target due to its role in mediating the first step of the pathway. Consequently, this study employs molecular modeling approaches including molecular docking, quantum mechanical calculations, and molecular dynamics (MD) simulations among others to investigate the potential of napthyridine derivatives to serve as inhibitors of the CrtM. The results of the study revealed the high binding affinities of the compounds for the target as demonstrated by their docking scores, while further subjection to screening pipeline aimed at determining their fitness for development into drugs revealed just one compound namely 6-[[1-[(2-fluorophenyl) methyl]triazol-4-yl]methoxy]-4-oxo-1H-1,5-naphthyridine-3-carboxylic acid as the compound with good drug-like, pharmacokinetics, and toxicity properties profiles. A 100 ns-long MD simulation of the complexes formed after molecular docking revealed the stable interaction of the compound with the target. Ultimately, this study can be a promising outlet to discover a weapon to fight against clinically resistant bacteria, however, further experimental studies are suggested to carry out in the wet lab, pre-clinical, and clinical levels.

Structure based virtual screening and molecular simulation study of FDA-approved drugs to inhibit human HDAC6 and VISTA as dual cancer immunotherapy

Cancer immunotherapy has significantly contributed to the treatment of various types of cancers mainly by targeting immune checkpoint inhibitors (ICI). Among them, V-domain immunoglobulin suppressor of T cell activation (VISTA) has been explored as a promising therapeutic target. Besides, histone deacetylase 6 (HDAC6) has been demonstrated to be efficacious target for several cancers. The current theoretical work was performed to explore the virtual repurposing of the FDA-approved drugs as inhibitors against these two (VISTA and HDAC6) cancers therapeutic targets. The crystal structure of the two proteins were downloaded from PDB and subjected to virtual screening by DrugRep webserver while using FDA-approved drugs library as ligands database. Our study revealed that Oxymorphone and Bexarotene are the top-ranked inhibitors of VISTA and HDAC6, respectively. The docking score of Bexarotene was predicted as - 10 kcal/mol while the docking score of Oxymorphone was predicted as - 6.2 kcal/mol. Furthermore, a total of 100 ns MD simulation revealed that the two drugs Oxymorphone and Bexarotene formed stable complexes with VISTA and HDAC6 drug targets. As compared to the standard drug the two drugs Oxymorphone and Bexarotene revealed great stability during the whole 100 ns MD simulation. The binding free energy calculation further supported the Root Mean Square Deviation (RMSD) result which stated that as compared to the ref/HDAC6 (- 18.0253 ± 2.6218) the binding free energy score of the Bexarotene/HDAC6 was good (- 51.9698 ± 3.1572 kcal/mol). The binding free energy score of Oxymorphone/VISTA and Ref/VISTA were calculated as - 36.8323 ± 3.4565, and - 21.5611 ± 4.8581 respectively. In conclusion, the two drugs deserve further consideration as cancer treatment option.