QSAR studies in the discovery of novel type-II diabetic therapies

Artificial intelligence in antidiabetic drug discovery: The advances in QSAR and the prediction of α-glucosidase inhibitors

Artificial Intelligence is transforming drug discovery, particularly in the hit identification phase of therapeutic compounds. One tool that has been instrumental in this transformation is Quantitative Structure-Activity Relationship (QSAR) analysis. This computer-aided drug design tool uses machine learning to predict the biological activity of new compounds based on the numerical representation of chemical structures against various biological targets. With diabetes mellitus becoming a significant health challenge in recent times, there is intense research interest in modulating antidiabetic drug targets. α-Glucosidase is an antidiabetic target that has gained attention due to its ability to suppress postprandial hyperglycaemia, a key contributor to diabetic complications. This review explored a detailed approach to developing QSAR models, focusing on strategies for generating input variables (molecular descriptors) and computational approaches ranging from classical machine learning algorithms to modern deep learning algorithms. We also highlighted studies that have used these approaches to develop predictive models for α-glucosidase inhibitors to modulate this critical antidiabetic drug target.

FragGen: towards 3D geometry reliable fragment-based molecular generation

3D structure-based molecular generation is a successful application of generative AI in drug discovery. Most earlier models follow an atom-wise paradigm, generating molecules with good docking scores but poor molecular properties (like synthesizability and drugability). In contrast, fragment-wise generation offers a promising alternative by assembling chemically viable fragments. However, the co-design of plausible chemical and geometrical structures is still challenging, as evidenced by existing models. To address this, we introduce the Deep Geometry Handling protocol, which decomposes the entire geometry into multiple sets of geometric variables, looking beyond model architecture design. Drawing from a newly defined six-category taxonomy, we propose FragGen, a novel hybrid strategy as the first geometry-reliable, fragment-wise molecular generation method. FragGen significantly enhances both the geometric quality and synthesizability of the generated molecules, overcoming major limitations of previous models. Moreover, FragGen has been successfully applied in real-world scenarios, notably in designing type II kinase inhibitors at the ∼nM level, establishing it as the first validated 3D fragment-based drug design algorithm. We believe that this concept-algorithm-application cycle will not only inspire researchers working on other geometry-centric tasks to move beyond architecture designs but also provide a solid example of how generative AI can be customized for drug design.

Development of a Robust Read-Across Model for the Prediction of Biological Potency of Novel Peroxisome Proliferator-Activated Receptor Delta Agonists

A robust predictive model was developed using 136 novel peroxisome proliferator-activated receptor delta (PPARδ) agonists, a distinct subtype of lipid-activated transcription factors of the nuclear receptor superfamily that regulate target genes by binding to characteristic sequences of DNA bases. The model employs various structural descriptors and docking calculations and provides predictions of the biological activity of PPARδ agonists, following the criteria of the Organization for Economic Co-operation and Development (OECD) for the development and validation of quantitative structure-activity relationship (QSAR) models. Specifically focused on small molecules, the model facilitates the identification of highly potent and selective PPARδ agonists and offers a read-across concept by providing the chemical neighbours of the compound under study. The model development process was conducted on Isalos Analytics Software (v. 0.1.17) which provides an intuitive environment for machine-learning applications. The final model was released as a user-friendly web tool and can be accessed through the Enalos Cloud platform's graphical user interface (GUI).

Artificial intelligence (AI)-it's the end of the tox as we know it (and I feel fine)

The rapid progress of AI impacts diverse scientific disciplines, including toxicology, and has the potential to transform chemical safety evaluation. Toxicology has evolved from an empirical science focused on observing apical outcomes of chemical exposure, to a data-rich field ripe for AI integration. The volume, variety and velocity of toxicological data from legacy studies, literature, high-throughput assays, sensor technologies and omics approaches create opportunities but also complexities that AI can help address. In particular, machine learning is well suited to handle and integrate large, heterogeneous datasets that are both structured and unstructured-a key challenge in modern toxicology. AI methods like deep neural networks, large language models, and natural language processing have successfully predicted toxicity endpoints, analyzed high-throughput data, extracted facts from literature, and generated synthetic data. Beyond automating data capture, analysis, and prediction, AI techniques show promise for accelerating quantitative risk assessment by providing probabilistic outputs to capture uncertainties. AI also enables explanation methods to unravel mechanisms and increase trust in modeled predictions. However, issues like model interpretability, data biases, and transparency currently limit regulatory endorsement of AI. Multidisciplinary collaboration is needed to ensure development of interpretable, robust, and human-centered AI systems. Rather than just automating human tasks at scale, transformative AI can catalyze innovation in how evidence is gathered, data are generated, hypotheses are formed and tested, and tasks are performed to usher new paradigms in chemical safety assessment. Used judiciously, AI has immense potential to advance toxicology into a more predictive, mechanism-based, and evidence-integrated scientific discipline to better safeguard human and environmental wellbeing across diverse populations.

In silico approaches to develop new phenyl-pyrimidines as glycogen synthase kinase 3 (GSK-3) inhibitors with halogen-bonding capabilities: 3D-QSAR CoMFA/CoMSIA, molecular docking and molecular dynamics studies

Glycogen synthase kinase 3 (GSK-3) is involved in different diseases, such as manic-depressive illness, Alzheimer's disease and cancer. Studies have shown that insulin inhibits GSK-3 to keep glycogen synthase active. Inhibiting GSK-3 may have an indirect pro-insulin effect by favouring glycogen synthesis. Therefore, the development of GSK-3 inhibitors can be a useful alternative for the treatment of type II diabetes. Aminopyrimidine derivatives already proved to be interesting GSK-3 inhibitors. In the current study, comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA) have been performed on a series of 122 aminopyrimidine derivatives in order to generate a robust model for the rational design of new compounds with promising antidiabetic activity. The q2 values obtained for the best CoMFA and CoMSIA models have been 0.563 and 0.598, respectively. In addition, the r2 values have been 0.823 and 0.925 for CoMFA and CoMSIA, respectively. The models were statistically validated, and from the contour maps analysis, a proposal of 10 new compounds has been generated, with predicted pIC50 higher than 9. The final contribution of our work is that: (a) we provide an extensive structure-activity relationship for GSK-3 inhibitory pyrimidines; and (b) these models may speed up the discovery of GSK-3 inhibitors based on the aminopyrimidine scaffold. Finally, we carried out docking and molecular dynamics studies of the two best candidates, which were shown to establish halogen-bond interactions with the enzyme.Communicated by Ramaswamy H. Sarma.

In silico prediction of the possible antidiabetic and anti-inflammatory targets of Nymphaea lotus-derived phytochemicals and mechanistic insights by molecular dynamics simulations

Nymphaea lotus is used traditionally for the treatment of diabetes and its complications. However, the mode of action and the likely bioactive phytochemicals involved are not yet fully explored. GC-MS analysis was employed to identify the inherent compounds in N. lotus leaves. To gain an insight into the antidiabetic mode of action of this plant, the identified phytochemicals were subjected to computational studies against four molecular targets of diabetes, dipeptidyl peptidase-4, glycogen synthase kinase 3, NADPH oxidase (NOX), sodium-glucose co-transporter-2, and one target of inflammation, cyclooxygenase-2. Compounds with notable binding affinity were subjected to druggability test. Results from molecular docking showed that seven of the compounds investigated exhibited druggability properties and had outstanding binding affinity values for these targets relative to values obtained for the respective standards of each of the targets. Analysis of the MD trajectories from a 100 ns atomistic run shows that the integrities of the complex systems were more stable and preserved throughout the simulation than the unbound protein. These results indicated that the antidiabetic and anti-inflammatory effects of these compounds might be via the inhibition of these targets, laying the foundation for further studies, such as in vitro and in vivo studies to fully validate the anti-diabetic agents from this plant.Communicated by Ramaswamy H. Sarma.

Discovery of new potent lysine specific histone demythelase-1 inhibitors (LSD-1) using structure based and ligand based molecular modelling and machine learning

Lysine-specific histone demethylase 1 (LSD-1) is an epigenetic enzyme that oxidatively cleaves methyl groups from monomethyl and dimethyl Lys4 of histone H3 and is highly overexpressed in different types of cancer. Therefore, it has been widely recognized as a promising therapeutic target for cancer therapy. Towards this end, we employed various Computer Aided Drug Design (CADD) approaches including pharmacophore modelling and machine learning. Pharmacophores generated by structure-based (SB) (either crystallographic-based or docking-based) and ligand-based (LB) (either supervised or unsupervised) modelling methods were allowed to compete within the context of genetic algorithm/machine learning and were assessed by Shapley additive explanation values (SHAP) to end up with three successful pharmacophores that were used to screen the National Cancer Institute (NCI) database. Seventy-five NCI hits were tested for their LSD-1 inhibitory properties against neuroblastoma SH-SY5Y cells, pancreatic carcinoma Panc-1 cells, glioblastoma U-87 MG cells and in vitro enzymatic assay, culminating in 3 nanomolar LSD-1 inhibitors of novel chemotypes.

Advanced Bioinformatics Tools in the Pharmacokinetic Profiles of Natural and Synthetic Compounds with Anti-Diabetic Activity

Diabetes represents a major health problem, involving a severe imbalance of blood sugar levels, which can disturb the nerves, eyes, kidneys, and other organs. Diabes management involves several synthetic drugs focused on improving insulin sensitivity, increasing insulin production, and decreasing blood glucose levels, but with unclear molecular mechanisms and severe side effects. Natural chemicals extracted from several plants such as Gymnema sylvestre, Momordica charantia or Ophiopogon planiscapus Niger have aroused great interest for their anti-diabetes activity, but also their hypolipidemic and anti-obesity activity. Here, we focused on the anti-diabetic activity of a few natural and synthetic compounds, in correlation with their pharmacokinetic/pharmacodynamic profiles, especially with their blood-brain barrier (BBB) permeability. We reviewed studies that used bioinformatics methods such as predicted BBB, molecular docking, molecular dynamics and quantitative structure-activity relationship (QSAR) to elucidate the proper action mechanisms of antidiabetic compounds. Currently, it is evident that BBB damage plays a significant role in diabetes disorders, but the molecular mechanisms are not clear. Here, we presented the efficacy of natural (gymnemic acids, quercetin, resveratrol) and synthetic (TAK-242, propofol, or APX3330) compounds in reducing diabetes symptoms and improving BBB dysfunctions. Bioinformatics tools can be helpful in the quest for chemical compounds with effective anti-diabetic activity that can enhance the druggability of molecular targets and provide a deeper understanding of diabetes mechanisms.

To Probe Full and Partial Activation of Human Peroxisome Proliferator-Activated Receptors by Pan-Agonist Chiglitazar Using Molecular Dynamics Simulations

Chiglitazar is a promising new-generation insulin sensitizer with low reverse effects for the treatment of type II diabetes mellitus (T2DM) and has shown activity as a nonselective pan-agonist to the human peroxisome proliferator-activated receptors (PPARs) (i.e., full activation of PPARγ and a partial activation of PPARα and PPARβ/δ). Yet, it has no high-resolution complex structure with PPARs and its detailed interactions and activation mechanism remain unclear. In this study, we docked chiglitazar into three experimentally resolved crystal structures of hPPAR subtypes, PPARα, PPARβ/δ, and PPARγ, followed by 3 μs molecular dynamics simulations for each system. Our MM-GBSA binding energy calculation revealed that chiglitazar most favorably bound to hPPARγ (-144.6 kcal/mol), followed by hPPARα (-138.0 kcal/mol) and hPPARβ (-135.9 kcal/mol), and the order is consistent with the experimental data. Through the decomposition of the MM-GBSA binding energy by residue and the use of two-dimensional interaction diagrams, key residues involved in the binding of chiglitazar were identified and characterized for each complex system. Additionally, our detailed dynamics analyses support that the conformation and dynamics of helix 12 play a critical role in determining the activities of the different types of ligands (e.g., full agonist vs. partial agonist). Rather than being bent fully in the direction of the agonist versus antagonist conformation, a partial agonist can adopt a more linear conformation and have a lower degree of flexibility. Our finding may aid in further development of this new generation of medication.

The herbicide quinclorac as potent lipase inhibitor: Discovery via virtual screening and in vitro/in vivo validation

Lipolysis is primarily controlled by the stepwise action of hormone-sensitive lipase (HSL) and monoglyceride lipase (MGL) to release free fatty acids and glycerol. A high level of circulating free fatty acids is well-known to mediate insulin resistance. Thus, the need to discover lipase inhibitors against both enzyme systems remains urgent. Agrochemicals are tightly regulated chemicals and therefore are potential source of new medicinal agents. Accordingly, we implemented a computational workflow to search for new lipase inhibitory leads by virtually screening commercial agrochemicals against HSL and MGL employing binding pharmacophores and docking experiments. Ten agrochemicals were identified as potential lipase inhibitors, out of which quinclorac, a safe herbicide, achieved high-ranking score. Subsequent in vitro evaluation against rat epididymal lipase activity showed quinclorac to exhibit nanomolar anti-lipase IC₅₀ . Subsequent in vivo testing showed quinclorac to significantly decrease blood glycerol levels after acute exposure (150 mg/kg) and multiple dosing (50 or 25 mg/kg) (p < 0.05).

Combining Stochastic Deformation/Relaxation and Intermolecular Contacts Analysis for Extracting Pharmacophores from Ligand-Receptor Complexes

We previously combined molecular dynamics (classical or simulated annealing) with ligand-receptor contacts analysis as a means to extract valid pharmacophore model(s) from single ligand-receptor complexes. However, molecular dynamics methods are computationally expensive and time-consuming. Here we describe a novel method for extracting valid pharmacophore model(s) from a single crystallographic structure within a reasonable time scale. The new method is based on ligand-receptor contacts analysis following energy relaxation of a predetermined set of randomly deformed complexes generated from the targeted crystallographic structure. Ligand-receptor contacts maintained across many deformed/relaxed structures are assumed to be critical and used to guide pharmacophore development. This methodology was implemented to develop valid pharmacophore models for PI3K-γ, RENIN, and JAK1. The resulting pharmacophore models were validated by receiver operating characteristic (ROC) analysis against inhibitors extracted from the CHEMBL database. Additionally, we implemented pharmacophores extracted from PI3K-γ to search for new inhibitors from the National Cancer Institute list of compounds. The process culminated in new PI3K-γ/mTOR inhibitory leads of low micromolar IC₅₀s.

Erythropoietin and mTOR: A "One-Two Punch" for Aging-Related Disorders Accompanied by Enhanced Life Expectancy

Life expectancy continues to increase throughout the world, but is accompanied by a rise in the incidence of non-communicable diseases. As a result, the benefits of an increased lifespan can be limited by aging-related disorders that necessitate new directives for the development of effective and safe treatment modalities. With this objective, the mechanistic target of rapamycin (mTOR), a 289-kDa serine/threonine protein, and its related pathways of mTOR Complex 1 (mTORC1), mTOR Complex 2 (mTORC2), proline rich Akt substrate 40 kDa (PRAS40), AMP activated protein kinase (AMPK), Wnt signaling, and silent mating type information regulation 2 homolog 1 (Saccharomyces cerevisiae) (SIRT1), have generated significant excitement for furthering novel therapies applicable to multiple systems of the body. Yet, the biological and clinical outcome of these pathways can be complex especially with oversight of cell death mechanisms that involve apoptosis and autophagy. Growth factors, and in particular erythropoietin (EPO), are one avenue under consideration to implement control over cell death pathways since EPO can offer potential treatment for multiple disease entities and is intimately dependent upon mTOR signaling. In experimental and clinical studies, EPO appears to have significant efficacy in treating several disorders including those involving the developing brain. However, in mature populations that are affected by aging-related disorders, the direction for the use of EPO to treat clinical disease is less clear that may be dependent upon a number of factors including the understanding of mTOR signaling. Continued focus upon the regulatory elements that control EPO and mTOR signaling could generate critical insights for targeting a broad range of clinical maladies.

Simulated annealing molecular dynamics and ligand-receptor contacts analysis for pharmacophore modeling

AIM

Ligand-based pharmacophore modeling requires long list of inhibitors, while pharmacophores based on single ligand-receptor crystallographic structure can be too restricted or promiscuous.

METHODOLOGY

This prompted us to combine simulated annealing molecular dynamics (SAMD) with ligand-receptor contacts analysis as means to construct pharmacophore model(s) from single ligand-receptor complex. Ligand-receptor contacts that survive numerous heating-cooling SAMD cycles are considered critical and are used to guide pharmacophore development.

RESULTS

This methodology was implemented to develop pharmacophores for acetylcholinesterase and protein kinase C-θ. The resulting models were validated by receiver-operating characteristic analysis and in vitro bioassay. Assay identified four new protein kinase C-θ inhibitors among captured hits, two of which exhibited nanomolar potencies.

CONCLUSION

The results illustrate the ability of the new method to extract valid pharmacophores from single ligand-protein complex.

Computational modeling of the bat HKU4 coronavirus 3CLpro inhibitors as a tool for the development of antivirals against the emerging Middle East respiratory syndrome (MERS) coronavirus

The Middle East respiratory syndrome coronavirus (MERS‐CoV) is an emerging virus that poses a major challenge to clinical management.

The 3C‐like protease (3CL^pro) is essential for viral replication and thus represents a potential target for antiviral drug development. Presently, very few data are available on MERS‐CoV 3CL^pro inhibition by small molecules. We conducted extensive exploration of the pharmacophoric space of a recently identified set of peptidomimetic inhibitors of the bat HKU4‐CoV 3CL^pro. HKU4‐CoV 3CL^pro shares high sequence identity (81%) with the MERS‐CoV enzyme and thus represents a potential surrogate model for anti‐MERS drug discovery. We used 2 well‐established methods: Quantitative structure‐activity relationship (QSAR)‐guided modeling and docking‐based comparative intermolecular contacts analysis. The established pharmacophore models highlight structural features needed for ligand recognition and revealed important binding‐pocket regions involved in 3CL^pro‐ligand interactions. The best models were used as 3D queries to screen the National Cancer Institute database for novel nonpeptidomimetic 3CL^pro inhibitors. The identified hits were tested for HKU4‐CoV and MERS‐CoV 3CL^pro inhibition. Two hits, which share the phenylsulfonamide fragment, showed moderate inhibitory activity against the MERS‐CoV 3CL^pro and represent a potential starting point for the development of novel anti‐MERS agents. To the best of our knowledge, this is the first pharmacophore modeling study supported by in vitro validation on the MERS‐CoV 3CL^pro.

Highlights

MERS‐CoV is an emerging virus that is closely related to the bat HKU4‐CoV.

3CL^pro is a potential drug target for coronavirus infection.

HKU4‐CoV 3CL^pro is a useful surrogate model for the identification of MERS‐CoV 3CL^pro enzyme inhibitors.

dbCICA is a very robust modeling method for hit identification.

The phenylsulfonamide scaffold represents a potential starting point for MERS coronavirus 3CL^pro inhibitors development.

Comparison of the molecular interactions of 7'-carboxyalkyl apigenin derivatives with S. cerevisiae α-glucosidase

As one of the most investigated flavonoids, apigenin, is considered to be a strong α-glucosidase inhibitor. However, the clinical utility of apigenin is limited due to its low solubility. It was reported that the solubility and biological activity can be improved by introducing sole carboxyalkyl group into apigenin, especially the 7'-substitution. With the increase of length of the alkyl chain in carboxyalkyl group, B ring of the apigenin derivative is embedded much more deeply into the binding cavity while the carboxyalkyl stretches to the neighboring cavity. All of the terminal carboxyl groups form hydrogen bonding interactions easily with the surrounding polar amino acids, such as His239, Ser244, Arg312 and Asp349. Thus, the electron density values of the carbonyl in the carboxyl group become higher than the solution status due to the strong molecular interactions. In fact, electron densities of most of the chemical bonds are decreased after molecular docking procedure. On compared with the solution phase, however, dipole moments of most of these molecules are increased, and their vectors are reoriented distinctly in the active sites. It is noticed that all of the Highest Occupied Molecular Orbital (HOMO) and Lowest Unoccupied Molecular Orbital (LUMO) are distributed throughout the whole parent apigenin ring in solution phase, whereas the disappeared situation happened on the B rings of some molecules (II-IV) in the active site, leading to higher energy gaps.

Discovery of novel potent nuclear factor kappa-B inhibitors (IKK-β) via extensive ligand-based modeling and virtual screening

Inhibitor kappa-B kinase-beta (IKK-β) controls the activation of nuclear transcription factor kappa-B and has been linked to inflammation and cancer. Therefore, inhibitors of this kinase should have potent anti-inflammatory and anticancer properties. Accordingly, we explored the pharmacophoric space of 218 IKK-β inhibitors to identify high-quality binding models. Subsequently, genetic algorithm-based quantitative structure activity relationship (QSAR) analysis was employed to select the best possible combination of pharmacophoric models and physicochemical descriptors that explain bioactivity variation among training compounds. Three successful pharmacophores emerged in 2 optimal QSAR equations (r₁²₁₇₅  = 0.733, r₁²_LOO  = 0.52, F1 = 65.62, r₁²_PRESS against 43 test inhibitors = 0.63 and r₂²₁₇₅  = 0.683, r₂²_LOO  = 0.52, F2 = 72.66, r₂²_PRESS against 43 test inhibitors = 0.65). Two pharmacophores were merged in a single binding model. Receiver operating characteristic curve validation proved the excellent qualities of this model. The merged pharmacophore and the associated QSAR equations were applied to screen the National Cancer Institute list of compounds. Ten hits were found to exhibit potent anti-IKK-β bioactivity, out of which, one illustrates IC50 of 11.0nM.

Use of machine learning approaches for novel drug discovery

INTRODUCTION

The use of computational tools in the early stages of drug development has increased in recent decades. Machine learning (ML) approaches have been of special interest, since they can be applied in several steps of the drug discovery methodology, such as prediction of target structure, prediction of biological activity of new ligands through model construction, discovery or optimization of hits, and construction of models that predict the pharmacokinetic and toxicological (ADMET) profile of compounds.

AREAS COVERED

This article presents an overview on some applications of ML techniques in drug design. These techniques can be employed in ligand-based drug design (LBDD) and structure-based drug design (SBDD) studies, such as similarity searches, construction of classification and/or prediction models of biological activity, prediction of secondary structures and binding sites docking and virtual screening.

EXPERT OPINION

Successful cases have been reported in the literature, demonstrating the efficiency of ML techniques combined with traditional approaches to study medicinal chemistry problems. Some ML techniques used in drug design are: support vector machine, random forest, decision trees and artificial neural networks. Currently, an important application of ML techniques is related to the calculation of scoring functions used in docking and virtual screening assays from a consensus, combining traditional and ML techniques in order to improve the prediction of binding sites and docking solutions.

Computer-aided discovery of new FGFR-1 inhibitors followed by in vitro validation

AIM

FGFR-1 is an oncogenic kinase involved in several cancers. FGFR1-specific inhibitors have shown promising results against several human cancers prompting us to model this interesting target. Toward the end, we implemented elaborate ligand-based and structure-based computational workflows to explore the pharmacophoric requirements for potent FGFR-1 inhibitors. Results & methodology: Structure-based and ligand-based modeling applied on 59 diverse FGFR-1 inhibitors yielded novel pharmacophore and quantitative structure-activity relationship models that were used to scan the National Cancer Institute's structural database for novel leads. Four potent hits were captured, with the most active having IC₅₀ of 426 nM. Identities and purities of active hits were established using nuclear magnetic resonance and mass spectroscopy.

CONCLUSION

Elaborate ligand-based (pharmacophore/quantitaive structure-activity relationship) and structure-based (docking-based comparative intermolecular contacts analysis) modeling provided deep understanding of ligand binding within FGFR-1 as evidenced by the virtually captured new potent leads.

Innovative computer-aided methods for the discovery of new kinase ligands

Recent evidence points to significant roles played by protein kinases in cell signaling and cellular proliferation. Faulty protein kinases are involved in cancer, diabetes and chronic inflammation. Efforts are continuously carried out to discover new inhibitors for selected protein kinases. In this review, we discuss two new computer-aided methodologies we developed to mine virtual databases for new bioactive compounds. One method is ligand-based exploration of the pharmacophoric space of inhibitors of any particular biotarget followed by quantitative structure-activity relationship-based selection of the best pharmacophore(s). The second approach is structure-based assuming that potent ligands come into contact with binding site spots distinct from those contacted by weakly potent ligands. Both approaches yield pharmacophores useful as 3D search queries for the discovery of new bioactive (kinase) inhibitors.