Nesrin Gökhan-Kelekçi

Janus kinases (JAKs) are central mediators of cytokine-driven signal transducer and activator of transcription (STAT) signaling, and dysregulation of the JAK–STAT axis is implicated in inflammatory, autoimmune, and neoplastic diseases. To identify new small-molecule inhibitors of Janus kinase 1 (JAK1) and Janus kinase 2 (JAK2), we implemented an integrated virtual screening workflow combining structure-based docking and ligand-based prioritization using the InterBioScreen (IBS) compound library. A drug-likeness filter reduced 521,627 IBS molecules to 116,064 candidates, which were then docked into the ATP-binding sites of JAK1 (PDB ID: 4EI4) and JAK2 (PDB ID: 6VGL) using Glide SP; compounds were prioritized using docking score thresholds of <− 8.5 kcal/mol (JAK1) and <− 9.0 kcal/mol (JAK2), yielding 407 JAK1- and 298 JAK2-focused hits. Subsequent analysis shortlisted 42 candidates for enzymatic evaluation. In vitro kinase assays identified multiple nanomolar inhibitors, including compound 1–3 (IC50 = 0.032 μM) among the most potent for JAK1 and compound 2–8 (IC50 = 0.026 μM) for JAK2. Finally, 100 ns molecular dynamics (MD) simulations supported stable binding for representative complexes, with persistent hinge-region interactions for compound 1–3 in JAK1 (e.g., Glu957/Leu959) and compound 2–8 in JAK2 (e.g., Glu930/Leu932), consistent with a stable interaction network in the active site. Collectively, these results define validated IBS-derived hit scaffolds for further optimization and selectivity profiling toward JAK-targeted therapeutics.

Alzheimer’s disease (AD) benefits from multitarget-directed ligands (MTDLs) that can enhance cholinergic transmission while attenuating monoamine-oxidase-linked oxidative stress. Here, we report a benzoxazolinone-based propionyl thiosemicarbazide series, synthesized and fully characterized by infrared (IR) spectroscopy, 1H nuclear magnetic resonance (NMR), and high-resolution mass spectra (HRMS). The compounds showed consistent submicromolar inhibitory activity across AChE, BChE, MAO-A, and MAO-B in vitro. Several AChE potencies approached the reference donepezil, and selected BChE activities were within an order of magnitude of tacrine. Notably, 4bk′ (5-Me/benzyl) inhibited three targets (IC50: 0.029 ± 0.001 μM for AChE, IC50: 0.071 ± 0.003 μM for BChE, IC50: 0.048 ± 0.002 μM for MAO-B), and 4af′ (5-Cl/phenyl) showed a balanced profile (IC50: 0.025 ± 0.001 μM for AChE, IC50: 0.056 ± 0.002 μM for BChE, IC50: 0.095 ± 0.003 μM for MAO-A), 4ac′ (5-Cl/propyl) combined potent AChE and MAO-B inhibition (IC50: 0.035 ± 0.001 μM for AChE, IC50: 0.045 ± 0.002 μM for MAO-B), whereas 4ag′ (5-Cl/4′-Cl-phenyl) was strongly MAO-B-selective (IC50: 0.041 ± 0.001 μM for MAO-B). Antioxidant capacity was pronounced for para-substituted analogues. The efficient compounds 4bk′, 4af′, 4ac′, and 4ag′ presented quite low toxicity on healthy cells (cell survival % was above %70 for 4bk′, 4af′, and 4ac′, while it was around 64% for 4ag′) even when they were applied at 100 times higher concentrations than their IC50 values, which indicates that they are safe at effective doses. Moreover, assessment of the compounds at a concentration of 10 μM demonstrated no cytotoxic effects on either healthy BV-2 microglial cells or H9c2 rat myoblastoma cells. Docking and 100 ns molecular dynamics (MD) simulations (AChE: 4EY7; MAO-B: 2V5Z) supported stable binding for dual-active representatives (RMSD ∼1.5–2.8 Å). In silico ADME (QikProp) indicated compliance with Lipinski’s Rule of Five and Jorgensen’s Rule of Three. Collectively, this scaffold is tunable from MAO-B-selective to balanced MTDL profiles suitable for further AD-relevant optimization.

A series of novel chiral urea and thiourea compounds were designed as multitarget‐directed ligands for Alzheimer's disease and evaluated against hAChE, hBChE, and hMAO‐B, with exploratory assessment of HSD10. These chiral compounds were synthesized and fully characterized, and their enantiopurity was confirmed via HPLC. Biological screening revealed that several of the 38 single‐enantiomer derivatives exhibited potent triple inhibition of hAChE, hBChE, and hMAO‐B. Biological screening revealed that a significant number of the 38 synthesized single‐enantiomer urea and thiourea derivatives potent inhibition across the primary target panel. Among them, compounds 14 S ( h AChE: 0.08 μM, h BChE: 0.93 μM, h MAO‐B: 0.02 μM), 23 R ( h AChE: 0.06 μM, h BChE: 0.74 μM, h MAO‐B: 0.08 μM), 6 S ( h AChE: 0.08 μM, h BChE: 0.65 μM, h MAO‐B: 0.08 μM), and 30 S ( h AChE: 0.04 μM, h BChE: 0.74 μM, h MAO‐B: 0.19 μM) showed the most promising multitarget profiles. Compounds 6 S and 23 R also displayed anti‐Aβ aggregation activity, whereas 12 S combined dual hMAO‐B/hAChE inhibition with antioxidant activity. Compounds 3 R and 11 R were found to be highly active against h MAO‐B (0.03 μM and 0.06 μM, respectively) and h BChE (0.68 and 0.72 μM). HSD10 inhibition was limited overall, with only compound 26 S showing moderate activity (52.2%) at higher micromolar concentration. Molecular docking and molecular dynamics simulations supported the proposed ligand‐target interactions for the most active compounds. In addition, selected derivatives showed no acute cytotoxicity under the tested conditions in HepG2 cells and high permeability potential in the PAMPA‐BBB assay. Based on their in vitro profiles, seven compounds ( 3 R , 6 S , 11 R , 12 S , 14 S , 23 R , and 30 S ) were advanced to in vivo evaluation, and compounds 6 S , 12 S , and 23 R , 30 S produced significant improvements in the Morris Water Maze model. Collectively, these findings identify chiral urea and thiourea derivatives as promising lead structures for further anti‐Alzheimer drug discovery.

Artificial intelligence (AI)-based approaches have drawn significant attention for their potential to address major challenges in drug discovery processes, such as time constraints, high costs, and low success rates. Specifically, machine learning (ML) and deep learning (DL) algorithms are effectively utilized in various stages of drug development, including target identification, molecular screening, lead compound selection, optimization, and ADMET prediction. In this study, the integration of current AI models into pharmaceutical R&D processes is examined from an interdisciplinary perspective, and their application domains are evaluated through relevant case studies in the literature. It has been observed that ML-based methods can yield successful results even with limited data, while DL architectures offer advantages in modeling complex molecular relationships. Furthermore, the architectural frameworks, training strategies, and diversity of ML and DL algorithms are comprehensively discussed within the scope of the study. It is demonstrated that traditionally experience-based decision processes such as retrosynthetic planning and formulation development can be accelerated and made more sustainable through data-driven systems. Additionally, AI-assisted predictions are shown to reduce the experimental burden and enhance research efficiency in preclinical and clinical stages. These evaluations suggest that AI technologies are not merely supportive tools but also strategic components at the core of innovative drug discovery approaches.

Thirty compounds having 1-[2-(5-substituted-2-benzoxazolinone-3-yl) acetyl]-3,5-disubstitutedphenyl-2-pyrazoline structure and nine compounds having N′-(1,3-disubstitutedphenylallylidene)-2-(5-substituted-2-benzoxazolinone-3-yl)acetohydrazide skeleton were synthesized and evaluated as monoamine oxidase (MAO) inhibitors. All of the compounds exhibited selective MAO-A inhibitor activity in the nanomolar or low micromolar range. The results of the molecular docking for hydrazone derivatives supported the in vitro results. Five compounds, 6 (0.008 μM, Selectivity Index (SI): 9.70 × 10^–4), 7 (0.009 μM, SI: 4.55 × 10^–5), 14 (0.001 μM, SI: 8.00 × 10^–4), 21 (0.009 μM, SI: 1.37 × 10^–5), and 42 (0.010 μM, SI: 5.40 × 10^–6), exhibiting the highest inhibition and selectivity toward hMAO-A and nontoxic to hepatocytes were assessed for antidepressant activity as acute and subchronic in mice. All of these five compounds showed significant antidepressant activity with subchronic administration consistent with the increase in the brain serotonin levels and the compounds crossed the blood–brain barrier according to parallel artificial membrane permeation assay. Compounds 14, 21, and 42 exhibited an ex vivo MAO-A profile, which is highly consistent with the in vitro data.

A new racemic pyrazoline derivative was synthesized and resolved to its enantiomers using analytic and semipreparative high‐pressure liquid chromatography. The absolute configuration of both fractions was established using vibrational circular dichroism. The in vitro monoamine oxidase (MAO) inhibitory profiles were evaluated for the racemate and both enantiomers separately for the two isoforms of the enzyme. The racemic compound and both enantiomers were found to inhibit hMAO‐A selectively and competitively. In particular, the R enantiomer was detected as an exceptionally potent and a selective MAO‐A inhibitor ( K _i = 0.85 × 10 ⁻³ ± 0.05 × 10 ⁻³ μM and SI: 2.35 × 10 ⁻⁵ ), whereas S was determined as poorer compound than R in terms of K _i and SI (0.184 ± 0.007 and 0.001). The selectivity of the enantiomers was explained by molecular modeling docking studies based on the PDB enzymatic models of MAO isoforms.

A series of 1-[2-((5-methyl/chloro)-2-benzoxazolinone-3-yl)acetyl]-3,5-diaryl-4,5-dihydro-1H-pyrazole derivatives were prepared by reacting 2-((5-methyl/chloro)-2-benzoxazolinone-3-yl)acetylhydrazine with appropriate chalcones. The chemical structures of all compounds were confirmed by elemental analyses, IR, ¹H NMR and ESI–MS. All the compounds were investigated for their ability to selectively inhibit monoamine oxidase (MAO) by in vitro tests. MAO activities of the compounds were compared with moclobemide and selegiline and all the compounds were found to inhibit human MAO-A selectively. The inhibition profile was found to be competitive and reversible for all compounds by in vitro tests. Among the compounds examined, compounds 5ae, 5af and 5ag were more selective than moclobemide, with respect to the K _i values experimentally found. In addition, the compound 5bg showed MAO-A inhibitor activity as well as moclobemide. A series of experimentally tested compounds (5ae–5ch) were docked computationally to the active site of the MAO-A and MAO-B isoenzyme. The AUTODOCK 4.01 program was employed to perform automated molecular docking.