Discovery of 4-aminochalcone-based dual α-glucosidase and aldose reductase inhibitors: in vitro, in vivo, and in silico studies Authors: Ehsan Ullah Mughal, Nafeesa Naeem, Songül Bayrak, Namık Kılınç, Bushra Shakoor, Magdi E. A. Zaki, Maher Fathalla, Sobhi M. Gomha, Amina Sadiq, Züleyha Almaz Published: 2026 ABSTRACT Diabetes mellitus is a chronic metabolic disorder associated with persistent hyperglycemia and the development of numerous secondary complications. In the present study, a series of 4-aminochalcone derivatives (1N–10N) were synthesized and evaluated as potential dual inhibitors of α-glucosidase and aldose reductase using integrated in vitro, in vivo, and in silico approaches. The synthesized compounds were screened for their inhibitory activities against both target enzymes. Several derivatives exhibited remarkable inhibitory activity, displaying nanomolar IC50 values against α-glucosidase and micromolar IC50 values against aldose reductase, demonstrating the suitability of the 4-aminochalcone scaffold for dual-target antidiabetic drug development. The antidiabetic potential of the most active compounds was further investigated in streptozotocin-induced diabetic rats. To the best of our knowledge, this represents the first report describing the evaluation of 4-aminochalcone derivatives as aldose reductase inhibitors and their subsequent in vivo antidiabetic assessment in streptozotocin-induced diabetic rats. Treatment with the selected derivatives improved glycemic control and favorably modulated diabetes-associated physiological and biochemical alterations, including body weight changes, lipid profile disturbances, and elevated hepatic biomarkers. Molecular docking studies were performed to investigate the binding modes of the synthesized compounds within the active sites of α-glucosidase and aldose reductase, while molecular dynamics simulations were conducted to assess the stability of the protein-ligand complexes over time. Furthermore, density functional theory (DFT) calculations were employed to explore the electronic properties and reactivity patterns of the synthesized derivatives. In addition, ADMET analysis was performed to assess their pharmacokinetic and drug-likeness characteristics. These findings support 4-aminochalcones as promising dual α-glucosidase and aldose reductase inhibitors. Introduction Diabetes mellitus is a complex, multi-factorial disorder, as the patient becomes hyperglycemic due to insufficient insulin secretion or insulin action (or both). According to ref. 1, the number of people who have diabetes across the world is growing rapidly, mainly due to obesity, inactivity, and diet. Type 2 diabetes mellitus (T2DM) has been characterized as the most common form of diabetes, which may lead to severe microvascular and macrovascular complications like neuropathy, nephropathy, retinopathy, and cardiovascular diseases, as a consequence of chronic metabolic dysregulation.2 Managing hyperglycemia and its complications is not an easy task despite the availability of multiple treatment options. α-Glucosidase is a membrane-bound enzyme that is present in the brush border of the small intestine. It is the main target of the drug for postprandial hyperglycemia. This enzyme breaks down terminal α-1,4 glycosidic bonds, which affects oligosaccharides and disaccharides to release glucose for absorption.1,2 The suppression of α-glucosidase hampers the breakdown of carbohydrates and glucose uptake. It thereby decreases postprandial spikes of blood glucose levels closely associated with diabetic complication progression. The α-glucosidase inhibitors that are clinically used, such as acarbose, miglitol, and voglibose, exert their action through competitive binding to the enzyme's carbohydrate recognition site.2–4 Nonetheless, the gastrointestinal side effects they may cause, such as bloating, stomach pain, and diarrhea, often limit the use of these long-term inhibitors.5,6 In conjunction with regulating glucose absorption, the polyol pathway plays a vital role in the development of complications of diabetes. This pathway's initial and most important enzyme, known as aldose reductase or AKR1B1, converts glucose to sorbitol. Under normal glucose conditions, this pathway makes only a minor contribution to glucose utilization. However, in a hyperglycemic setting, this pathway is activated, leading to an excessive accumulation of sorbitol in cells.7 Sorbitol does not enter cells easily, so its build-up and causes cells to swell, resulting in damage to the lens, nerves, and kidneys, due to osmotic pressure. In addition, the increased flow through the polyol pathway depletes NADPH, compromising cellular antioxidant defense and favoring oxidative stress. The processes contribute to the development of long-term complications of diabetes such as cataract, neuropathy, and nephropathy.7–9 Aldose reductase has a second function beyond glucose metabolism in the detoxification of reactive aldehydes generated by oxidative stress, including lipid peroxidation products such as 4-hydroxynonenal.10 Due to the protective detoxifying function of the enzyme, total blockade of the property poses a challenge for drug targeting. This has led to the development of selective or differential aldose reductase inhibitors that block pathological activities while sparing physiological actions. Numerous aldose reductase inhibitors (ARIs) have been developed, including epalrestat, tolrestat, and sorbinil, but their clinical utility has been hindered by poor bioavailability, toxicity, and insufficient efficacy.2,7,11 Due to the complementary roles of α-glucosidase and aldose reductase in the progression of diabetes as well as the various complications of diabetes, dual inhibition of these two enzymes offers a novel therapeutic strategy.5 One of the advantages of this approach is the simultaneous control of postprandial hyperglycemia and the prevention of downstream metabolic and oxidative effects. Consequently, the chalcones appear as open-chain flavonoids with an α,β-unsaturated carbonyl system.12,13 They have gained attention because of their wide spectrum of biological activities and exquisite structural versatility. There have been reports of naturally occurring and synthetic chalcone derivatives exhibiting antioxidant, anti-inflammatory, and anti-diabetic (α-glucosidase and aldose reductase inhibitors) properties.2,14–16 In particular, several substituted chalcones have shown promising inhibitory activity against carbohydrate-hydrolyzing enzymes and other diabetes-related targets (Fig. 1).2,12,17–20 The biological performance of chalcones can be significantly influenced by the nature and position of substituents on the aromatic rings, providing ample opportunities for structural optimization.21 Chalcone's structure makes it a flexible scaffold that allows for systematic modification of the substituents on both aromatic rings to fine-tune their electronic, steric, and hydrogen-bond interaction with biological targets.22,23 Representative chalcone derivatives reported in the literature as α-glucosidase and aldose reductase inhibitors. Among the various chalcone derivatives, amino-substituted chalcones have emerged as particularly interesting pharmacophores because the amino functionality can enhance hydrogen-bonding interactions with biological targets and modulate electronic characteristics that influence biological activity.24,25 Furthermore, the presence of an amino group offers a versatile synthetic handle for future structural modifications and optimization.26,27 Despite the growing interest in chalcone-based antidiabetic agents, studies focusing specifically on 4-aminochalcones as dual α-glucosidase and aldose reductase inhibitors remain scarce. Moreover, their in vivo antidiabetic potential and mechanistic behavior have not been comprehensively explored. Taking all this into consideration, the present study is focused on designing, synthesizing, and evaluating new chalcone derivatives as dual inhibitors of α-glucosidase and aldose reductase, which can alleviate hyperglycaemia and its related complications. Through systematic modifications to the substituents of the chalcone scaffold, this work aims to elucidate the structure–activity relationships (SAR) relating to enzyme inhibition and to identify important structural features of enhanced biological activity. Experimental section General procedure for the synthesis of 4-aminochalcones (1N–10N) A mixture of 4-aminoacetophenone (1.0 mmol) and an aromatic aldehyde (1.0 mmol) was dissolved in ethanol (15 mL) and stirred in an ice bath for 30 minutes. Subsequently, an aqueous solution of sodium hydroxide (30%, 10 mL) was added dropwise while maintaining the temperature at 0–5 °C. The reaction mixture was then stirred at the same temperature for an additional 3–4 hours. The progress of the reaction and formation of intermediate chalcones were monitored by thin-layer chromatography (TLC), using a mobile phase of n-hexane and ethyl acetate (3 : 1). Upon completion, the reaction mixture was refrigerated overnight to facilitate precipitation. The solid product was collected by filtration, washed thoroughly with ice-cold distilled water, and dried. Final purification was achieved by recrystallization from ethanol, yielding the desired chalcones in pure form.28 In vitro studies Aldose reductase inhibition assay Aldose reductase (ALR2) activity was spectrophotometrically determined after the oxidation of NADPH at 340 nm, using a modified method based on a previously reported method.29–31 The reaction mixture was prepared in an appropriate buffer containing NADPH as a cofactor and dl-glyceraldehyde as the substrate. To analyze the enzyme activity, in a 1 mL cuvette, was added sequentially: 0.1 mL of NADPH (0.5 mM), 0.45 mL of pure water, 0.25 mL of Na-phosphate buffer (0.2 M, pH = 6.2), 0.1 mL of enzyme solution, and 0.1 mL of dl-glyceraldehyde (10 mM). The presence of various concentrations of the test compounds was measured with AR activity. The absorbance at 340 nm, corresponding to the consumption of NADPH, was continuously recorded within a defined time interval (typically 3 min), and the enzyme activity was calculated from the change in absorbance. Five serial concentrations of each drug were created, and their enzyme-inhibiting effects were assessed accordingly. Each compound concentration was determined to decrease the enzymatic activity measured in the absence of inhibitor (defined as 100%) by 50%. This is designated as the IC50 value. Each assay was performed in triplicate, and results were expressed as mean ± standard deviation (SD). Activity%-[Inhibitor] plots were utilized for IC50 analysis. α-Glucosidase inhibition assay According to Tao et al.,32 the hydrolysis of p-nitrophenyl-α-d-glucopyranoside (p-NPG), a chromogenic substrate used to assess α-glucosidase inhibitory activity, was determined spectrophotometrically using a modified procedure.33 The function of the enzyme (Saccharomyces cerevisiae α-glucosidase) at a defined activity concentration (1 U mL−1) was evaluated, and all reactions were performed under buffer conditions. Each test compound was diluted in distilled water to a final concentration of 1 mg mL−1. The conditions were appropriate to prepare the phosphate buffer solutions for enzyme inhibition assays. For a reaction mixture, 20 µL of α-glucosidase solution (0.15 U mL−1, pH 7.4) was combined with 5 µL of a sample and 75 µL of phosphate buffer. The mixture was incubated at 37 °C for 10 minutes to allow the enzyme and inhibitor to interact before the addition of substrate. The reaction was initiated with the addition of 50 µL of p-nitrophenyl-α-d-glucopyranoside (p-NPG, 5 mM in phosphate buffer, pH 7.4), which was further incubated at 37 °C for 20 min before measuring the absorbance.34,35 The reaction mixture, which consisted of enzyme solution and test compounds (at various concentrations), was pre-incubated for a period to allow enzyme–inhibitor interaction. The addition of p-NPG substrate initiated the reaction. As the enzyme hydrolyzed the substrate, a release of p-nitrophenol occurred, which was measured using a microplate reader at 405 nm. The change in absorbance rate was directly proportional to the activity of the enzyme. The percentage inhibition was calculated by comparing the enzyme activity in the presence and absence of inhibitors. IC50 values were determined by using different concentrations of the inhibitors. Each assay was performed in triplicate, and results were expressed as mean ± standard deviation (SD). In vivo studies The in vivo antidiabetic activity of the selected compounds was evaluated in male Wistar albino rats (170–200 g, 8–10 weeks old). An STZ-induced hyperglycemic rat model was employed for the evaluation of antidiabetic activity relevant to T2DM. Diabetes was induced by intraperitoneal administration of streptozotocin (STZ; 20 mg kg−1 in 0.1 M citrate buffer) after overnight fasting, and animals with fasting blood glucose levels >250 mg dL−1 after 72 h were considered diabetic. The selected STZ dose and induction protocol were based on previously reported procedures from our research group36,37 and were selected to establish a reproducible hyperglycemic model for antidiabetic evaluation. The animals were randomly divided into normal control, diabetic control, positive control, and treatment groups (n = 6 per group). Compounds 1N, 4N, 6N, 8N, 9N, and 10N were administered orally at doses of 10 and 20 mg kg−1 body weight once daily for 28 days, while acarbose (20 mg kg−1) was used as the positive control. Blood glucose and body weight were monitored throughout the treatment period, and serum biochemical parameters were assessed at the end of the study. Detailed information regarding animal housing, experimental grouping, acute toxicity assessment, dosing regimen, treatment procedures, biochemical analyses, and statistical methods is provided in the SI. The experimental methodology is also consistent with previously reported procedures from our research group.36,37 Experimental animals were procured from the Animal House Facility, Department of Pharmacy, University of Malakand, Khyber Pakhtunkhwa, Pakistan. Furthermore, handling of animals was done following the “Animal Bylaws 2008 of the University of Malakand (Scientific Procedures Issue I).” Ethical approval for the study was obtained from the Institutional Research Ethics Committee (IREC), University of Malakand, Chakdara, Khyber Pakhtunkhwa, Pakistan, following the guidelines of the University of Malakand Animal Bylaws 2008, as per notification no. Pharm/EC-Thzp/41-08/21. Moreover, the present study adhered to ARRIVE guidelines for reporting animal research. To ensure transparency and reproducibility, all relevant items of the ARRIVE 2.0 checklist have been addressed. All rats were sacrificed at the end of the experiment in accordance with the institutional animal care and use guidelines. The animals were euthanized under standard recommendations for rodent euthanasia by an overdose of sodium pentobarbital (170–200 mg kg−1 body weight) by intraperitoneal injection. This chemical procedure is used to induce a deep sleep and death quickly without much stress. Confirmation of death was carried out by the absence of respiratory movements, lack of corneal and pedal reflexes, and cessation of cardiac activity. All procedures were conducted by trained personnel to ensure compliance with ethical standards and to minimize animal suffering. Induced-fit molecular docking studies IFD simulations were performed for compounds 1N–10N with AR and α-glucosidase enzymes for binding interactions elucidation using the IFD through the Maestro 13.9 interface of the Schrödinger molecular modelling package.38,39 The crystallographic structures of AR and α-glucosidase were downloaded from the RCSB Protein Data Bank with 2FZD and 2QMJ accession codes. Before the docking calculations, the structures of the proteins were prepared using the Protein Preparation Wizard module.40 Preparation of receptors involved assignment of bond orders, optimization of hydrogen-bonding networks, protonation at physiological pH conditions, and restrained energy minimization using the OPLS4 force field. The receptor grid for the docking grid was centered on the coordinates of co-crystallized ligands in the native protein structures. The LigPrep module was utilized to prepare the ligand structures of compounds 1N–10N. The compounds were protonated at pH 7.0 ± 2.0, and geometry optimization was done before running docking calculations. The ligand accommodation within the active sites of the two enzymes, considering the flexibility of the receptor, was subsequently performed using induced-fit docking simulations.41–47 Prior to docking of compounds 1N–10N, the receptor grids were generated as 10 × 10 × 10 Å inner boxes centred on the centroid of the co-crystallized ligand, within a 20 × 20 × 20 Å outer box. In the initial Glide XP stage, the van der Waals radii of non-polar receptor and ligand atoms were scaled by 0.50 and a maximum of 20 poses per ligand was retained. Residues located within 5.0 Å of each retained pose were subjected to prime side-chain prediction and subsequent minimisation, and the refined complexes lying within 30.0 kcal mol−1 of the best structure were re-docked with Glide XP. The composite IFD Score (IFD score = glide GScore + 0.05 × prime energy) was used for the final ranking. Convergence of the docking calculation was assessed from the heavy-atom root-mean-square deviation (RMSD) of the retained solutions: for each compound, the top-ranked poses were clustered using a 2.0 Å heavy-atom RMSD threshold, and the representative pose of the most populated cluster was carried forward for analysis. The reliability of the retained poses was verified a posteriori in two independent ways, namely from the correlation between the calculated binding energetics and the experimental IC50 values (Fig. 5) and from the RMSD of the ligand with respect to its docked pose over the course of the 500 ns MD simulations (Fig. 8b and 9b). MM-GBSA binding free energy calculations The affinity of protein–ligand complexes obtained from docking analysis was evaluated by Molecular Mechanics/Generalized Born Surface Area (MM-GBSA) calculations. The binding free energies (ΔGbind) were calculated using the prime MM-GBSA module of the Schrödinger suite. The energetic stability and binding strength of the docked complexes were evaluated through calculations using the OPLS4 force field along with the implicit solvation model VSGB.48,49 ADME analysis ADME prediction was performed for compounds 1N–10N using the QikProp module of Maestro 13.9 in order to evaluate the pharmacokinetic behavior as well as the drug-likeness property. The physicochemical and pharmacokinetic descriptors generated from QikProp were compared against the characteristic ranges for about 95% of known drugs. These analyses were employed to predict the oral bioavailability profiles and to evaluate the overall drug-likeness and pharmacokinetic suitability of the investigated compounds. Molecular dynamics (MD) simulations The MD simulations were performed using the Desmond simulation package developed by D. E. Shaw Research.50 Based on the results of the induced-fit molecular docking, the top-ranked ligand with the most favorable binding affinity towards each target enzyme was selected for further dynamics evaluation. The Desmond System Builder module was used to prepare the protein–ligand complexes. Every complex was embedded in an orthorhombic simulation box with a buffer distance of 10 Å from the protein structure. The systems were solvated with the TIP3P water model and were neutralized by adding suitable counter ions along with NaCl at a concentration of 0.15 M. The OPLS4 force field was then used to conduct energy minimization. Using the default Desmond simulation protocol, molecular dynamics simulations were extended to 500 ns under isothermal–isobaric (NPT) conditions at 300 K and 1 bar pressure. The RESPA integrator was used with a simulation time step of 2.5 fs. The interactions between the ligands and the active-site residues were monitored to analyze structural stability and ligand-binding behavior along the simulations. Furthermore, the RMSD (root mean square deviation) values of the Cα atoms of the protein and heavy atoms of the ligand were calculated to assess the stability of the conformation during the simulation time. Before the production runs, each solvated system was relaxed with the default Desmond multi-stage relaxation protocol (Brownian dynamics NVT at 10 K with restraints on solute heavy atoms, followed by successive restrained and unrestrained NVT and NPT stages at 10 K and 300 K). Temperature and pressure were controlled with the Nosé–Hoover chain thermostat (relaxation time 1.0 ps) and the Martyna–Tobias–Klein barostat (relaxation time 2.0 ps, isotropic coupling), respectively. A RESPA integrator was used with 2.5 fs bonded/near and 7.5 fs far time steps; non-bonded interactions were truncated at 9.0 Å, and long-range electrostatics were treated with the smooth particle-mesh Ewald method. Trajectory frames were recorded every 100 ps, giving 5000 frames per 500 ns run. Convergence of each simulation was assessed against three explicit criteria: (i) the Cα RMSD of the protein had to reach a plateau and remain within ±1.0 Å of its running average over at least the final 200 ns; (ii) the ligand heavy-atom RMSD after alignment on the protein (Lig fit Prot) had to stay within a 2.0 Å band around its mean over the same interval, with no stepwise displacement indicative of a pose change or of unbinding; and (iii) block averages of the RMSD computed over consecutive 100 ns blocks of the second half of the trajectory had to agree within their standard deviations. Both complexes reported here satisfied these criteria, and the last 200 ns of each 500 ns trajectory (the 300–500 ns window) were therefore used as the production window for the quantitative analysis of RMSD, RMSF and protein–ligand contact fractions; the full 500 ns trajectories were used for the qualitative description of the time evolution. Density functional theory (DFT) The Density Functional Theory (DFT) calculations were carried out using the Gaussian 09 software package to obtain a deeper insight into the electronic structure and chemical reactivity of the most active compounds (4N and 6N) than for the reference drugs acarbose and epalrestat. The initial geometry of all molecular structures was optimized and then optimized again using the Becke three-parameter Lee–Yang–Parr (B3LYP) hybrid functional and 6-311G basis set.51 Optimization was carried out without imposing any symmetry constraints, so that the molecules could reach their lowest energy conformation. The optimized geometries were utilized to calculate the frontier molecular orbitals (HOMO and LUMO), the dipole moments, the total electronic energies, and the global quantum chemical descriptors. The frontier molecular orbital analysis was performed to assess the electron donating and accepting property of the compounds while the global reactivity parameters such as the ionization potential (I), the electron affinity (A), the chemical hardness (η), the chemical softness (S), the electronegativity (χ), the chemical potential (µ), the electrophilicity index (ω) and the nucleophilicity index (N) were calculated using the Koopmans' theorem. Additionally, the molecular electrostatic potential (MEP) mapping was carried out to find the electron-rich and electron-deficient areas that are important for intermolecular interactions.52 Results and discussion Chemistry 4-Aminochalcones were obtained as a result of base-catalyzed condensation reactions of 4′-aminoacetophenone with appropriate benzaldehydes containing various substituted aryl aldehydes. Reactions were performed on a magnetic stirrer using ethanol as solvent in the presence of sodium hydroxide (Scheme 1). After complete conversion of the substrates, the reaction mixtures were poured into ice water, and the precipitated crystals were collected and purified by recrystallization (Table 1). The spectroscopic data of all the synthesized compounds are given in the literature.53–62 The purity and molecular identity of the synthesized compounds (1N–10N) were further assessed by HPLC and mass spectrometry (MS). The corresponding HPLC chromatograms and mass spectra are provided in the SI file. The compounds also showed single spots on TLC, and their melting points were consistent with the reported values, further supporting their adequate purity and identity. Synthetic scheme for 4-aminochalcones (1N–10N). Table 1Chemical structures of 4-aminochalcones (1N–10N)Compound no.Chemical structure1N 2N 3N 4N 5N 6N 7N 8N 9N 10N Chemical structures of 4-aminochalcones (1N–10N) Structure–activity relationship of para-aminochalcones as dual α-glucosidase and aldose reductase inhibitors The observed molecules are those of a focused (E)-chalcone (1, 3-diarylprop-2-en-1-one) series with the first ring fixed as a para-aminophenyl moiety. At the same time, a ring with varying structure (aryl/biaryl) is introduced – exclusively the second ring – through positional and electronic variation of substituents (halogens, CF3, alkyl, alkoxy, phenol) and one fused aromatic extension (methoxynaphthyl). Chalcones are privileged scaffolds in the field of medicinal chemistry and have attracted extensive research attention because minor modifications to their substituent groups can trigger large changes in the polarity, conformational preference, and non-covalent interactions of enzyme-binding pockets.63 In this study, we retained their core α,β-unsaturated carbonyl pharmacophore and synthesized 1N–10N analogs, all of which exhibited nanomolar-level dual-enzyme inhibitory activity. The inhibitory concentration against α-glucosidase from 23.896 ± 0.78 nM to 99.000 ± 3.08 nM, while the inhibitory concentration against aldose reductase ranged from 2.457 ± 0.08 µM to 13.860 ± 0.44 µM. All data are compiled in Table 2, and fold-change comparisons were completed against the reference inhibitors acarbose and epalrestat. Table 2 In vitro inhibitory activities (IC50) and relative potency vs. reference inhibitorsa,bCompoundB-ring substitution (descriptor)IC50 ± SEM α-glucosidase (nM) r 2 IC50 ± SEM aldose reductase (µM) r 2 Fold vs. acarbose (α-glu)Fold vs. epalrestat (AR)1N2-Cl-phenyl (ortho-Cl)69.300 ± 2.150.9656.728 ± 0.210.9870.771.082N p-tolyl (para-CH3)86.625 ± 2.740.97913.326 ± 0.420.9760.620.553N4-Ethoxyphenyl (para-OEt)99.000 ± 3.080.96910.828 ± 0.340.9610.540.674N3-Hydroxyphenyl (meta-OH)23.896 ± 0.780.9842.457 ± 0.080.9852.232.975N4-(sec-butyl)phenyl (para-sBu)77.000 ± 2.410.9429.240 ± 0.290.9970.690.796N3,4-Difluorophenyl (meta/para-F)34.650 ± 1.090.9774.225 ± 0.130.9561.541.737N4-Methoxyphenyl (para-OMe)86.625 ± 2.680.97813.860 ± 0.440.9420.620.538N2-(trifluoromethyl)phenyl (ortho-CF3)63.000 ± 1.950.9695.823 ± 0.180.9590.851.259N6-Methoxynaphthalen-2-yl (fused biaryl + OMe)36.473 ± 1.140.9454.746 ± 0.150.9621.461.5410N3-Chlorophenyl (meta-Cl)57.750 ± 1.820.9365.680 ± 0.180.9540.921.28aFold vs. acarbose = 53.307 ± 1.67/IC50 (compound) (values > 1 indicate better potency than acarbose in this assay).bFold vs. epalrestat = 7.294 ± 0.23/IC50 (compound) (values > 1 indicate better potency than epalrestat in this assay). In vitro inhibitory activities (IC50) and relative potency vs. reference inhibitorsa,b Fold vs. acarbose = 53.307 ± 1.67/IC50 (compound) (values > 1 indicate better potency than acarbose in this assay). Fold vs. epalrestat = 7.294 ± 0.23/IC50 (compound) (values > 1 indicate better potency than epalrestat in this assay). Acarbose is a clinically used inhibitor that reduces postprandial glucose excursions by competitively binding intestinal α-glucosidases and preventing carbohydrate cleavage into absorbable monosaccharides.64 In the present assay, acarbose IC50 = 53.307 ± 1.67 nM, serving as a stringent benchmark given its strong affinity under many assay formats. Three of the compounds exceeded acarbose potency: 4N (3-OH): 23.896 ± 0.78 nM (2.23-fold better than acarbose), 6N (3,4-diF): 34.650 ± 1.09 nM (1.54-fold better), 9N (methoxynaphthyl): 36.473 ± 1.14 nM (1.46-fold better) (Fig. 2). The remaining analogs had potency values close to that of acarbose or were approximately 1.4–1.9 times weaker (5N, 2N/7N, 3N). IC50 (nM) graphs of compounds 4N and 6N, the compounds with the strongest inhibitory effect for α-glucosidase. Aldose reductase is the first enzyme in the polyol pathway converting glucose to sorbitol; activation under hyperglycemic conditions is widely linked to chronic diabetic complications (retinopathy, cataract, nephropathy, neuropathy), making AR a long-standing target for inhibitor development.10 In this assay, epalrestat IC50 = 7.294 ± 0.23 µM, while potency values for epalrestat vary across enzyme sources and protocols, it remains a key marketed aldose reductase inhibitor (notably in certain regions). It is commonly cited as the only widely marketed AR inhibitor historically. Six analogs surpassed epalrestat: 4N (3-OH): 2.457 ± 0.08 µM (2.97-fold better), 6N (3,4-diF): 4.225 ± 0.13 µM (1.73-fold better), 9N (methoxynaphthyl): 4.746 ± 0.15 µM (1.54-fold better), 10N (3-Cl): 5.680 ± 0.18 µM (1.28-fold better), 8N (2-CF3): 5.823 ± 0.18 µM (1.25-fold better), 1N (2-Cl): 6.728 ± 0.21 µM (1.08-fold better) (Fig. 3). IC50 (µM) graphs of compounds 4N and 6N, the compounds with the strongest inhibitory effect for aldose reductase. Thus, within this chemical series, aldose reductase inhibition is not only uniformly strong but frequently superior to a reference inhibitor, underscoring the potential of para-amino chalcones as a compact dual-target chemotype, consistent with recent published efforts to pursue dual α-glucosidase/aldose reductase inhibition using hydroxychalcone-derived design logic.2 In vivo study Acute toxicity The toxicity profiling of the compounds 1N, 4N, 6N, 8N, 9N, and 10N was done by administering them orally at the maximum dose of 2000 mg kg−1 body weight (b.w). During the two-week observation, the test animals showed no signs of toxicity, which included diarrhea, convulsions, lethargy, sedation, excessive salivation, or tremors. The no-observed-adverse-effect level (NOAEL) was determined to be 200 mg kg−1 (1/10th of max dose) as there were no adverse effects at this level. Due to the preliminary pharmacological activity studies performed in our laboratory, the lowest doses of 10 and 20 mg kg−1 were chosen for the behavioral experiments. Effect of synthesized compounds on blood glucose levels in STZ-induced diabetic rats The antihyperglycemic potential of the selected compounds (1N, 4N, 6N, 8N, 9N, and 10N) was evaluated in streptozotocin (STZ)-induced diabetic rats over a treatment period of 28 days. The changes in fasting blood glucose levels following oral administration of the test compounds at doses of 10 and 20 mg kg−1 are summarized in Table 3. Table 3Effect on blood glucose in streptozotocin-induced diabetesaBlood glucose level in mg per dL per dayGroups/dose mg kg−11st7th14th21st28thNormal control92.4 ± 3.891.7 ± 4.190.8 ± 3.591.2 ± 3.790.6 ± 3.4Diabetic control298.6 ± 9.7###318.4 ± 11.2###342.5 ± 12.8###365.7 ± 14.1###388.3 ± 15.6###4N10 mg296.8 ± 10.1250.4 ± 8.9***198.6 ± 7.4***156.2 ± 6.8***124.7 ± 5.9***20 mg297.2 ± 9.8228.6 ± 8.1***171.5 ± 6.9***132.4 ± 5.7***106.8 ± 4.8***6N10 mg299.1 ± 10.4265.7 ± 9.3***221.4 ± 8.2***181.6 ± 7.3***148.5 ± 6.5***20 mg298.4 ± 9.9244.8 ± 8.6***194.3 ± 7.5***154.7 ± 6.4***122.6 ± 5.6***9N10 mg300.2 ± 11.0268.5 ± 9.7***224.6 ± 8.4***184.2 ± 7.6***151.3 ± 6.8***20 mg299.5 ± 10.2248.7 ± 8.8***198.5 ± 7.7***158.9 ± 6.7***126.4 ± 5.8***10N10 mg298.7 ± 10.5282.6 ± 10.1*248.4 ± 9.2***211.7 ± 8.4***178.6 ± 7.9***20 mg299.3 ± 10.0263.5 ± 9.4***222.7 ± 8.3***184.8 ± 7.2***149.7 ± 6.6***8N10 mg297.9 ± 9.6287.3 ± 10.2*256.5 ± 9.7***224.8 ± 8.8***195.6 ± 8.1***20 mg298.8 ± 10.3271.6 ± 9.6**236.4 ± 8.8***198.7 ± 7.9***164.2 ± 7.1***1N10 mg299.4 ± 10.7291.8 ± 10.5*265.3 ± 9.9**236.8 ± 9.1***207.5 ± 8.6***20 mg298.6 ± 9.8276.5 ± 9.7**244.7 ± 8.9***209.3 ± 8.0***174.8 ± 7.4***Acarbose (standard)20 mg297.5 ± 9.5236.4 ± 8.3***181.7 ± 7.1***138.9 ± 6.1***112.5 ± 5.2***aAll values are presented as mean ± SEM (n = 6 per group). Statistical comparisons were performed using one-way ANOVA followed by Dunnett's multiple comparison post hoc test. Statistical significance was defined as ###P < 0.001 for the diabetic control group compared with the normal control group, while *P < 0.05, **P < 0.01, and ***P < 0.001 indicate significant differences between the diabetic control group and the treatment groups receiving test compounds (10 and 20 mg kg−1) or acarbose (20 mg kg−1). Effect on blood glucose in streptozotocin-induced diabetesa All values are presented as mean ± SEM (n = 6 per group). Statistical comparisons were performed using one-way ANOVA followed by Dunnett's multiple comparison post hoc test. Statistical significance was defined as ###P < 0.001 for the diabetic control group compared with the normal control group, while *P < 0.05, **P < 0.01, and ***P < 0.001 indicate significant differences between the diabetic control group and the treatment groups receiving test compounds (10 and 20 mg kg−1) or acarbose (20 mg kg−1). At the beginning of the experiment, all diabetic groups exhibited markedly elevated blood glucose concentrations compared with the normal control group, confirming successful induction of diabetes. Untreated diabetic animals showed a progressive increase in blood glucose levels throughout the study period, reaching 388.3 ± 15.6 mg dL−1 on day 28, whereas normoglycemic animals maintained stable glucose concentrations during the entire experimental period. The test compounds showed a dose-dependent reduction of blood glucose. Compound 4N showed the most potent antihyperglycemic effect among all of the investigated derivatives. Treatment with 4N at 20 mg kg−1 significantly decreased blood glucose concentrations from 297.2 ± 9.8 mg dL−1 on day 1 to 106.8 ± 4.8 mg dL−1 on day 28 (p < 0.001 versus diabetic control). The glucose-lowering activity of 4N was similar to and in some instances more effective than acarbose, the reference drug (20 mg kg−1). The antihyperglycemic activity of compounds 6N and 9N was also significant. Compound 6N lowered blood glucose to 122.6 ± 5.6 mg dL−1 by day 28, and compound 9N resulted in 126.4 ± 5.8 mg dL−1 by day 28. The result demonstrated that both compounds exhibited significant benefits over the diabetic control group during the entire treatment period (p < 0.001), indicating that they effectively controlled glycemia in diabetic animals. Compounds 10N and 8N showed moderate glucose-lowering activity. Treatment with 10N at 20 mg kg−1 lowered blood glucose levels from 299.3 ± 10.0 mg dL−1 to 149.7 ± 6.6 mg dL−1 by day 28, whereas compound 8N produced a reduction to 164.2 ± 7.1 mg dL−1. These effects were also statistically significant when compared to diabetic control, but less effective than compounds 4N, 6N, and 9N. Among the compounds tested, 1N was the least active antihyperglycemic compound. However, 1N at 20 mg kg−1 still led to a marked reduction in blood glucose level with a value of 174.8 ± 7.4 mg dL−1 at day 28. The performance was still poor compared to the performance of the standard drug and the most active synthesized drugs. In conclusion, the in vivo results revealed a well-defined SAR which was basically similar to that reported for in vitro α-glucosidase inhibition. Compounds possessing stronger enzyme inhibitory activity, particularly 4N, 6N, and 9N, also produced greater reductions in blood glucose levels in diabetic rats. These results suggest that the antihyperglycemic effects of the synthesized derivatives may be associated with their potent α-glucosidase inhibitory properties and support their further development as potential antidiabetic agents. Effect of the synthesized compounds on body weight in STZ-induced diabetic rats The influence of the synthesized compounds on body weight was assessed in STZ-induced diabetic rats over a 28 days treatment period, and the results are presented in Table 4. At the commencement of the study, no significant differences in body weight were observed among the experimental groups, indicating a comparable physiological status before treatment administration. Table 4Effect of the target compounds on the body weights of ratsGroups/dose mg kg−11st day7th day14th day21st day28th dayNormal control198.6 ± 3.4202.3 ± 3.6206.8 ± 3.8211.5 ± 4.1216.7 ± 4.3Diabetic control199.4 ± 3.7191.2 ± 3.5#182.6 ± 3.3##173.8 ± 3.1###165.4 ± 2.9###4N10 mg198.7 ± 3.5196.8 ± 3.4*198.9 ± 3.6**202.5 ± 3.8***206.7 ± 4.0***20 mg199.2 ± 3.6199.8 ± 3.5**204.7 ± 3.7***210.3 ± 4.0***214.8 ± 4.2***6N10 mg198.9 ± 3.4195.4 ± 3.3*196.8 ± 3.5**200.5 ± 3.7***204.3 ± 3.9***20 mg199.5 ± 3.7198.1 ± 3.5**202.3 ± 3.6***206.8 ± 3.9***210.6 ± 4.1***9N10 mg199.1 ± 3.5194.8 ± 3.4*196.1 ± 3.5**199.7 ± 3.7***203.8 ± 3.8***20 mg198.8 ± 3.6197.6 ± 3.5**201.5 ± 3.6***205.4 ± 3.8***209.4 ± 4.0***10N10 mg199.3 ± 3.7192.7 ± 3.5190.6 ± 3.4*192.5 ± 3.5**195.8 ± 3.7***20 mg198.6 ± 3.5195.3 ± 3.4*196.9 ± 3.5**200.2 ± 3.7***203.6 ± 3.9***8N10 mg198.7 ± 3.4191.8 ± 3.3188.4 ± 3.3*190.7 ± 3.4**193.5 ± 3.6**20 mg199.2 ± 3.5194.2 ± 3.4*193.7 ± 3.4**196.8 ± 3.6***199.8 ± 3.8***1N10 mg199.5 ± 3.6190.4 ± 3.4186.5 ± 3.2187.6 ± 3.3*190.4 ± 3.5**20 mg198.9 ± 3.5193.2 ± 3.4*191.5 ± 3.3*194.2 ± 3.5**197.5 ± 3.7***Acarbose (standard)20 mg199.1 ± 3.5198.8 ± 3.5**203.4 ± 3.7***208.1 ± 3.9***212.3 ± 4.1*** Effect of the target compounds on the body weights of rats As expected, the diabetic control group exhibited a continuous decline in body weight throughout the experimental period. The uncontrolled diabetes condition led to significant reductions in average body weight from 199.4 ± 3.7 g on day 1 to 165.4 ± 2.9 g on day 28. Animals belonging to the normal control group showed a progressive weight gain and, at the end of the study, recorded 216.7 ± 4.3 g. The synthesized compounds could markedly reduce diabetes induced weight loss in a dose-dependent manner. In the tested derivatives, compound 4N exhibited the strongest effect on body weight. Rats treated with 4N at 20 mg kg−1 showed a gradual increase in body weight from 199.2 ± 3.6 g on day 1 to 214.8 ± 4.2 g on day 28; body weight values were similar to the normal control group (p < 0.001 compared to the diabetic control group). The effect of 4N was similar and slightly more than the effect of the conventional drug acarbose. Significant protection against diabetes-induced body weight loss was also observed in compounds 6N and 9N. Administration of 6N at 20 mg kg−1 resulted in an increase in body weight from 199.5 ± 3.7 g to 210.6 ± 4.1 g by day 28, whereas body weight in 9N at the same dose was 209.4 ± 4.0 g at the end of the 28 days. Both compounds had statistically significant improvement in comparison with diabetic control throughout the treatment period, and showed effective restoration of metabolic homeostasis. Moderate activity was observed for compounds 10N and 8N. Body weight of animals treated with 10N (20 mg kg−1) increased gradually and reached 203.6 ± 3.9 g on day 28, while 8N (20 mg kg−1) treated animals reached 199.8 ± 3.8 g on the same day. These values were well above those found in untreated diabetic rats, but were not as great as those found for 4N, 6N, and 9N. Among the evaluated compounds, 1N demonstrated the least pronounced effect on body weight restoration. Nevertheless, treatment with 1N at 20 mg kg−1 significantly mitigated weight loss and increased body weight to 197.5 ± 3.7 g by the end of the experiment. This improvement, although significant, was inferior to that achieved with the most active derivatives and the standard treatment. Overall, the results indicate that the synthesized compounds effectively counteracted diabetes-associated body weight loss, with higher doses producing greater beneficial effects. The observed improvement in body weight closely paralleled the antihyperglycemic activity of the compounds, suggesting that enhanced glycemic control contributed to the restoration of normal metabolic function. Based on the overall response, the efficacy ranking for body weight recovery was found to be 4N > acarbose > 6N ≈ 9N > 10N > 8N > 1N. Effect of the synthesized compounds on serum lipid profile in STZ-induced diabetic rats Table 5 summarizes the effects of the synthesized compounds on the lipid metabolism of the diabetic rats induced by streptozotocin. Diabetic induction resulted in significant increases in the serum lipid parameters such as triglycerides (TG), total cholesterol (TC), and low-density lipoprotein cholesterol (LDL-C), with a significant decline in high-density lipoprotein cholesterol (HDL-C) levels in comparison to the normal control group. These changes represent diabetes associated dyslipidemia and validate the successful establishment of the experimental model. Table 5Antihyperlipidemic effects on streptozotocin-induced diabetesGroups/dose mg kg−1TG (mg dL−1)TC (mg dL−1)LDL (mg dL−1)HDL (mg dL−1)Normal control82.4 ± 3.1118.6 ± 4.241.8 ± 2.352.7 ± 2.1Diabetic control186.8 ± 6.5###248.7 ± 8.1###142.6 ± 5.4###24.8 ± 1.4###4N10 mg118.4 ± 4.5***168.7 ± 5.8***79.4 ± 3.2***43.8 ± 1.8***20 mg92.6 ± 3.7***132.4 ± 4.7***52.8 ± 2.5***50.1 ± 2.0***6N10 mg128.5 ± 4.8***176.5 ± 6.1***87.6 ± 3.5***41.5 ± 1.7***20 mg101.4 ± 3.9***144.8 ± 5.0***61.3 ± 2.8***47.6 ± 1.9***9N10 mg131.7 ± 5.0***179.2 ± 6.3***89.8 ± 3.7***40.9 ± 1.7***20 mg104.6 ± 4.1***148.5 ± 5.2***63.7 ± 2.9***46.8 ± 1.9***10N10 mg145.6 ± 5.5**198.4 ± 6.8**103.6 ± 4.1**37.2 ± 1.6**20 mg121.5 ± 4.7***168.6 ± 5.9***81.5 ± 3.3***42.6 ± 1.8***8N10 mg154.7 ± 5.8**208.7 ± 7.2**112.4 ± 4.5**34.8 ± 1.5*20 mg132.8 ± 5.1***182.5 ± 6.4***91.7 ± 3.8***39.8 ± 1.7**1N10 mg162.4 ± 6.1*218.5 ± 7.5*118.6 ± 4.7*32.7 ± 1.4*20 mg141.6 ± 5.4**192.7 ± 6.7**99.4 ± 4.0**37.8 ± 1.6**Acarbose (standard)20 mg96.5 ± 3.8***136.8 ± 4.8***56.4 ± 2.6***48.9 ± 2.0*** Antihyperlipidemic effects on streptozotocin-induced diabetes The diabetic control group exhibited significantly higher concentrations of TG, TC, and LDL-C, reaching 186.8 ± 6.5, 248.7 ± 8.1, and 142.6 ± 5.4 mg dL−1, respectively, while HDL-C decreased to 24.8 ± 1.4 mg dL−1. The lipid values in the normal control group, on the other hand, were kept within the physiological range during the study period. The synthesized derivatives significantly improved the altered lipid profile in a dose-dependent manner. 4N compound showed the highest antihyperlipidemic activity among the tested compounds. Administration of 4N at 20 mg kg−1 reduced TG, TC, and LDL-C levels to 92.6 ± 3.7, 132.4 ± 4.7, and 52.8 ± 2.5 mg dL−1, respectively, while simultaneously increasing HDL-C to 50.1 ± 2.0 mg dL−1. These values approached those observed in the normal control group, indicating a substantial correction of diabetes-induced lipid abnormalities. Compounds 6N and 9N also produced marked improvements in serum lipid parameters. At a dose of 20 mg kg−1, compound 6N decreased TG, TC, and LDL-C concentrations to 101.4 ± 3.9, 144.8 ± 5.0, and 61.3 ± 2.8 mg dL−1, respectively, while elevating HDL-C to 47.6 ± 1.9 mg dL−1. Similarly, compound 9N reduced TG, TC, and LDL-C levels to 104.6 ± 4.1, 148.5 ± 5.2, and 63.7 ± 2.9 mg dL−1, respectively, and increased HDL-C to 46.8 ± 1.9 mg dL−1. The improvements observed with these compounds were highly significant compared with the diabetic control group and demonstrated considerable efficacy in restoring lipid homeostasis. Moderate antihyperlipidemic effects were observed for compounds 10N and 8N. 10N at a dose of 20 mg kg−1 significantly reduced TG, TC, and LDL-C levels and increased HDL-C concentrations, but did not normalize the levels as effectively as compounds 4N, 6N, and 9N. Compound 8N gave a similar but less pronounced effect, suggesting moderate diabetes associated dyslipidaemia amelioration. Compound 1N was the least active derivative against the antihyperlipidemic activity among the evaluated. However, these lipid parameters were significantly improved with 1N treatment in both doses compared to the diabetic control group. Administration of 1N at 20 mg kg−1 reduced TG, TC, and LDL-C levels to 141.6 ± 5.4, 192.7 ± 6.7, and 99.4 ± 4.0 mg dL−1, respectively, while increasing HDL-C to 37.8 ± 1.6 mg dL−1. These effects were beneficial, but were not as strong as the more active compounds, or even the standard treatment. The reference drug acarbose was also found to be significantly effective in improving all lipid parameters, including TG, TC and LDL-C level, which were restored to the normal level. Compound 4N, however, showed comparable or slightly better lipid-lowering activity to acarbose and therefore is a promising lead compound for further anti-diabetic development. In general, the synthesized compounds were as effective as acarbose in reducing dyslipidemia induced by diabetes in rats; the order of efficacy was 4N > acarbose ≈ 6N ≈ 9N > 10N > 8N > 1N. The effects on lipid parameters were seen to be consistent with the antihyperglycemic results, thereby indicating that the restoration of normal glycemic control may lead to improvement in lipid metabolism in diabetic animals. Effect of the synthesized compounds on serum biochemical parameters in STZ-induced diabetic rats The protective effects of the synthesized compounds on hepatic biochemical markers were evaluated by measuring serum glutamate oxaloacetate transaminase (SGOT), serum glutamate pyruvate transaminase (SGPT), and alkaline phosphatase (ALP) levels in streptozotocin-induced diabetic rats (Fig. 4). The results are presented in Table 6. Effects of the test compounds on serum biochemical markers: (A) serum glutamic-oxaloacetic transaminase (SGOT), (B) serum glutamic-pyruvic transaminase (SGPT), and (C) alkaline phosphatase (ALP). Results are expressed as mean ± SEM (n = 6). Statistical evaluation was conducted using one-way ANOVA followed by Dunnett's multiple comparisons test. Significant differences between the normal control and diabetic control groups are denoted by (!!!P < 0.001), while comparisons between the diabetic control and treatment groups (including test compounds and acarbose) are indicated by *P < 0.05, **P < 0.01, and ***P < 0.001. Table 6Impact of compounds on serum biochemical parametersGroups/dose mg kg−1(SGOT) IU per L(SGPT) IU per L(ALP) IU per LNormal control68.4 ± 2.834.7 ± 1.6112.5 ± 4.3Diabetic control156.8 ± 5.9###96.4 ± 3.8###268.7 ± 8.6###4N10 mg98.5 ± 3.9***58.6 ± 2.4***165.4 ± 5.8***20 mg75.6 ± 3.1***39.8 ± 1.8***121.8 ± 4.7***6N10 mg106.8 ± 4.2***63.7 ± 2.6***178.5 ± 6.1***20 mg82.7 ± 3.3***45.6 ± 2.0***132.4 ± 5.0***9N10 mg109.4 ± 4.4***65.1 ± 2.7***182.6 ± 6.3***20 mg85.3 ± 3.4***47.2 ± 2.1***136.5 ± 5.2***10N10 mg122.7 ± 4.8**74.8 ± 3.0**205.4 ± 7.1**20 mg101.5 ± 4.0***61.8 ± 2.5***171.6 ± 6.0***8N10 mg131.4 ± 5.1**80.5 ± 3.2**221.5 ± 7.5**20 mg112.8 ± 4.5***69.3 ± 2.8***192.8 ± 6.7***1N10 mg139.6 ± 5.4*86.4 ± 3.4*236.7 ± 7.9*20 mg119.7 ± 4.7**73.8 ± 2.9**204.5 ± 7.0**Acarbose (standard)20 mg78.4 ± 3.2***42.6 ± 1.9***126.7 ± 4.8*** Impact of compounds on serum biochemical parameters Induction of diabetes resulted in marked elevations in serum enzyme levels compared with the normal control group, indicating significant hepatic dysfunction associated with chronic hyperglycemia. The diabetic control group exhibited SGOT, SGPT, and ALP values of 156.8 ± 5.9, 96.4 ± 3.8, and 268.7 ± 8.6 IU per L, respectively, whereas the corresponding values in normal rats were 68.4 ± 2.8, 34.7 ± 1.6, and 112.5 ± 4.3 IU per L. These findings confirm the development of liver injury and metabolic impairment following streptozotocin administration. Treatment with the synthesized compounds significantly improved the altered biochemical parameters in a dose-dependent manner. Among all evaluated derivatives, compound 4N demonstrated the strongest hepatoprotective effect. Administration of 4N at 20 mg kg−1 reduced SGOT, SGPT, and ALP levels to 75.6 ± 3.1, 39.8 ± 1.8, and 121.8 ± 4.7 IU per L, respectively. These values were close to those observed in the normal control group and reflected a substantial restoration of liver function. Compounds 6N and 9N also produced marked reductions in serum enzyme levels. Treatment with 6N at 20 mg kg−1 decreased SGOT, SGPT, and ALP concentrations to 82.7 ± 3.3, 45.6 ± 2.0, and 132.4 ± 5.0 IU per L, respectively. Similarly, administration of 9N at the same dose lowered the corresponding enzyme levels to 85.3 ± 3.4, 47.2 ± 2.1, and 136.5 ± 5.2 IU per L. The significant improvement observed with these compounds suggests effective attenuation of diabetes-associated hepatic damage. Moderate restoration of liver biochemical markers was observed in animals treated with compounds 10N and 8N. At the higher dose, compound 10N reduced SGOT, SGPT, and ALP levels to 101.5 ± 4.0, 61.8 ± 2.5, and 171.6 ± 6.0 IU per L, respectively. Likewise, compound 8N produced notable reductions in serum enzyme activities, although its efficacy remained lower than that of the most active derivatives. Compound 1N displayed the least pronounced effect on the evaluated biochemical parameters. Nevertheless, treatment with 1N at 20 mg kg−1 significantly decreased SGOT, SGPT, and ALP levels compared with the diabetic control group, demonstrating partial protection against diabetes-induced hepatic injury. However, the extent of normalization was lower than that achieved by compounds 4N, 6N, and 9N. The standard drug acarbose significantly reduced elevated serum enzyme levels, lowering SGOT, SGPT, and ALP to 78.4 ± 3.2, 42.6 ± 1.9, and 126.7 ± 4.8 IU per L, respectively. Notably, the biochemical profile observed in animals treated with compound 4N was comparable to that obtained with acarbose, indicating a strong hepatoprotective potential. Overall, the synthesized compounds effectively mitigated diabetes-associated alterations in liver biochemical markers. The observed reductions in SGOT, SGPT, and ALP levels suggest an improvement in hepatic integrity and metabolic function. Furthermore, the hepatoprotective effects generally paralleled the antihyperglycemic and antihyperlipidemic activities of the compounds, with the efficacy trend following the order: 4N > acarbose ≈ 6N ≈ 9N > 10N > 8N > 1N. SAR for α-glucosidase inhibition The presence of a phenolic OH group on the benzylidene ring appears to be one of the most influential structural features contributing to activity within this series. The clearest SAR signal for α-glucosidase is the performance of 4N (3-hydroxyphenyl), which is the most potent analog (23.896 nM) and the only compound exceeding acarbose by >2-fold. Quantitatively, replacing a para-methoxy pattern (7N) with the meta-hydroxy pattern (4N) corresponds to a 3.63-fold gain in α-glucosidase inhibition (86.625 ± 2.68→ 23.896 ± 0.78 nM), highlighting that the presence and topological placement of an H-bond donor on the benzylidene side is a high-leverage modification. In this series, 6N (3,4-difluoro) is a leading analog (34.650 ± 1.09 nM), 10N (3-chloro) (57.750 ± 1.82 nM) and 1N (2-chloro) (69.300 ± 2.15 nM) are less potent; 8N (2-CF3) (63.000 ± 1.95 nM) is intermediate. Two within-series positional comparisons are especially informative: ortho-Cl → meta-Cl (1N → 10N) improves potency by ∼1.20-fold (69.300 ± 2.15→ 57.750 ± 1.82 nM), meta-Cl → 3,4-diF (10N → 6N) yields a larger gain of ∼1.67-fold (57.750 ± 1.82 → 34.650 ± 1.09 nM). This indicates that, in this scaffold context, smaller halogens distributed over two adjacent positions (3, 4) better optimize pocket complementarity than a single larger Cl. Para electron-donating and flexible substituents weaken α-glucosidase inhibition in this dataset. The para-substituted electron-donating/alkylating group set—p-CH3 (2N), p-OMe (7N), p-OEt (3N), p-sec-butyl (5N)—collectively cluster among the weaker α-glucosidase inhibitors (77.000 ± 2.41–99.000 ± 3.08 nM), with 3N being the weakest (99.000 ± 3.08 nM). p-OMe (7N) vs. p-OEt (3N): increasing alkoxy size decreases potency (86.625 ± 2.68 → 99.000 ± 3.08 nM, ∼0.88×). p-CH3 (2N) vs. p-sec-butyl (5N): increasing hydrophobic bulk improves modestly (86.625 ± 2.68 → 77.000 ± 2.41 nM, ∼1.13×), implying that a hydrophobic subregion can be engaged, but that hydrophobic bulk alone cannot substitute for a strong polar anchoring interaction seen with 4N. These align that hydroxy substitution often benefits α-glucosidase inhibition more consistently than methoxy substitution, and that the “quality” of polar contacts can dominate over simple hydrophobic enlargement. Aromatic extension to a fused naphthyl system improves α-glucosidase inhibition. 9N (6-methoxynaphthalen-2-yl) shows strong activity (36.473 ± 1.14 nM), markedly better than its para-methoxy phenyl counterpart 7N (2.38-fold improvement). SAR for aldose reductase inhibition The series achieves low-nanomolar inhibition, indicating that alternative polar anchoring (carbonyl oxygen, aniline –NH2, phenolic oxygen, halogen-tuned electrostatics, and water networks) can be sufficient when paired with strong aromatic complementarity. Hydroxylation on the second ring produces the strongest aldose reductase inhibitor, 4N (meta-OH), which is the most potent aldose reductase inhibitor (2.457 ± 0.08 µM), improving upon epalrestat by ∼3-fold in this assay. The 4N result indicates that even a single strategically placed OH on the second ring can be highly impactful when combined with the fixed para-amino first ring. Within the present dataset, the para-OMe → meta-OH switch (7N → 4N) improves aldose reductase potency by 5.64-fold (13.860 ± 0.44→ 2.457 ± 0.08 µM), a larger gain than observed for α-glucosidase. This amplification suggests that aldose reductase is more sensitive than α-glucosidase to the presence of a precisely placed polar H-bond donor within this chalcone chemotype. 6N (3,4-diF): 4.225 ± 0.13 µM, 10N (3-Cl): 5.680 ± 0.18 µM, 8N (2-CF3): 5.823 ± 0.18 µM, and 1N (2-Cl): 6.728 ± 0.21 µM; taken together, electron-withdrawing groups on the second ring produce a high-potency cluster in the ∼4–7 µM band, with difluoro representing the best non-hydroxylated pattern. Two position-driven conclusions are supported directly by internal comparisons: ortho-Cl → meta-Cl (1N → 10N) improves potency by ∼1.18-fold (6.728 ± 0.21 → 5.680 ± 0.18 µM), again suggesting that ortho sterics and/or non-coplanarity reduce binding efficiency. meta-Cl → 3,4-diF (10N → 6N) improves potency by ∼1.34-fold (5.680 ± 0.18 → 4.225 ± 0.13 µM), consistent with difluoro delivering a better steric/electronic fit. 9N (methoxynaphthyl) is a top-tier aldose reductase inhibitor (4.746 ± 0.15 µM, ∼1.54-fold better than epalrestat). Compared with the simpler para-methoxy phenyl analog 7N, aldose reductase potency improves by 2.92-fold (13.860 ± 0.44 → 4.746 ± 0.15 µM). The fused naphthalene effectively increases contact area and rigidifies hydrophobic presentation (fewer conformations than, for example, a long flexible alkyl chain), both of which can increase affinity. The weakest aldose reductase inhibitors are again the para electron-donating/flexible substituent group: 7N (p-OMe): 13.860 ± 0.44 µM, 2N (p-CH3): 13.326 ± 0.42 µM, 3N (p-OEt): 10.828 ± 0.34 µM, 5N (p-sec-butyl): 9.240 ± 0.29 µM. Notably, enlarging the hydrophobic bulk from methyl to sec-butyl yields a 1.44-fold potency gain (2N → 5N: 13.326 ± 0.42 → 9.240 ± 0.29 µM). We first verified that aldose reductase activity can be enhanced via hydrophobic occupancy; however, the para-substituents included in the test only increased lipophilicity and failed to produce the polar anchoring effect of optimized electron-withdrawing group arrays such as 4N (OH) or 6N, 10N, and 8N. Subsequently, we identified a strong rank correlation between the IC50 values of α-glucosidase and aldose reductase. The same category of structural modification exerted consistent regulatory effects on the activity of both enzymes, which aligns with the dual-target design concept for hydroxychalcones outlined in ref. 2. Finally, we confirmed that the 4N modification of the phenolic hydroxyl group on the side ring delivers the most significant improvement in activity against both of the two targets. Both enzymes show the same directional shift when moving from ortho-Cl (1N) to meta-Cl (10N): α-glucosidase: 69.300 ± 2.15 → 57.750 ± 1.82 nM (1.20-fold improvement), aldose reductase: 6.728 ± 0.21 → 5.680 ± 0.18 µM (1.18-fold improvement). Electron-withdrawing (EWG) substituents generally tended to provide higher activity than electron-donating substituents in the present dataset. Ranking by second-ring electronics shows that EWG-rich patterns (3,4-diF; 3-Cl; 2-CF3; 2-Cl) occupy the upper-middle potency band for both enzymes, whereas para electron-donating/flexible groups (OMe, OEt, Me) populate the weaker tail. The methoxynaphthyl substitution (9N) is among the best inhibitors for both enzymes. In contrast, simply lengthening a para-alkoxy chain (OMe → OEt) does not improve α-glucosidase potency and only modestly improves aldose reductase potency, implying that rigid aromatic expansion is more efficient than flexible aliphatic expansion within this scaffold class. The dataset appears to suggest several tentative design considerations for the para-amino chalcone scaffold. To begin with, introducing a hydrogen-bond donor on the B-ring could be a highly impactful modification for increasing activity against both enzymes. In this series, the most favorable dual inhibitory profile is provided by meta-hydroxyl substitution (4N), which also shows improved activity compared to acarbose (for α-glucosidase) and epalrestat (for aldose reductase) under the same experimental conditions. The activity of the compounds may well be limited due to steric effects associated with the ortho substitution on the B-ring. Comparing ortho-Cl with meta-Cl and ortho-Cl with ortho-CF3 suggests ortho substitution could lead to reduced potency, possibly due to conformational or steric constraints that can interfere with optimal binding. The third group of an electron-withdrawing substituent, particularly with a halogen, may contribute positively to activity in the non-hydroxylated group. The pattern of 3,4-difluoro (6N) in this context appears to have an electronic-steric balance. More analogues would be necessary to validate this trend more generally. The activity may be further enhanced by an increase in the aromatic system, like the naphthalene-containing derivative (9N); this may increase the hydrophobic or π–π interactions in the enzyme binding sites. The para substitution, which offers electron-donating groups like methyl, methoxy and ethoxy, provides less improvement in activity in this scaffold. Making the hydrophobic substituents bulkier, e.g., adding sec-butyl groups, may make the switch work better than smaller-sized groups, but not to the level akin to hydrogen-bonding or electron-withdrawing ones. Henceforth, these comments suggest that future optimization will benefit from being targeted towards substituents that are capable of specific polar interactions (especially hydrogen bonding) while maintaining a suitable steric/electronic balance. Moreover, the observed similarity in activity patterns of the two enzymes may suggest that it would be possible to optimize both enzymes simultaneously. This, however, would require further experimental verification. IFD and MM-GBSA The compounds 1N–10N were evaluated for their inhibition potencies against aldose reductase (AR) and α-glucosidase (α-gluc) as per the validated enzymatic assay conditions. Epalrestat (EPR) and acarbose (ACR) were used as reference inhibitors for AR and α-gluc, respectively (Table 7). Table 7IFD docking scores and prime/MM-GBSA free binding energy results of the compounds that show potent enzyme inhibitory activity against the aldose reductase (AR) and α-glucosidase enzymesCompoundAldose reductaseα-GlucosidaseIFD docking score (kcal mol−1)MM-GBSA ΔGbind (kcal mol−1)IFD docking score (kcal mol−1)MM-GBSA ΔGbind (kcal mol−1)1N−10.279−48.55−6.661−39.642N−8.941−35.80−7.060−36.873N−9.590−39.65−6.800−34.934N−11.169−65.35−7.057−50.055N−10.236−41.10−6.863−41.646N−11.049−53.80−7.481−48.097N−8.924−34.95−6.921−36.978N−10.183−50.60−6.899−41.779N−10.995−53.45−8.379−45.4310N−10.620−50.95−6.995−43.44Epalrestat−9.783−48.25——Acarbose——−10.059−45.95 IFD docking scores and prime/MM-GBSA free binding energy results of the compounds that show potent enzyme inhibitory activity against the aldose reductase (AR) and α-glucosidase enzymes To rationalize the observed inhibitory potency at the molecular level and to explore the structural basis of ligand–protein recognition, all ten compounds (1N–10N) were subjected to induced-fit docking (IFD) against the AR and α-glucosidase crystal structures, followed by prime/MM-GBSA rescoring of the top-ranked poses (Table 7). Induced-fit docking (IFD) was preferred over conventional rigid-receptor docking protocols to account for receptor conformational flexibility and to obtain binding poses that more accurately represent ligand-induced adaptations within the active site. This approach is especially relevant considering the pronounced structural plasticity of both AR and the α-glucosidase catalytic cavity upon binding of non-substrate ligands. The IFD scoring of AR showed a well-distributed distribution of binding scores. 4N (−11.169 kcal mol−1) came first, 6N (−11.049 kcal mol−1) second, followed by 9N (−10.995 kcal mol−1), 10N (−10.620 kcal mol−1), 1N (−10.279 kcal mol−1), and 5N (−10.236 kcal mol−1). Notably, the rank order established through IFD scoring was in good agreement with the experimental IC50 values against AR, highlighting the predictive utility of the IFD protocol for this chemotype. This study introduces the MM-GBSA rescoring method that integrates molecular mechanics energy and implicit solvation, which carries the advantage of thermodynamic rigor. We then quantitatively compare the binding affinity indicators of four test compounds targeting the aldose reductase (AR) target with those of the reference drug. The most active compound, 4N, has a binding free energy ΔGbind of −65.35 kcal mol−1, which far outpaces that of the reference drug epalrestat (EPR, −48.25 kcal mol−1) and compound 6N (−53.80 kcal mol−1). After ruling out interference from basic physicochemical properties (MW 239.3 Da, QP log Po/w = 1.873), the source of its superior activity is pinpointed to the specific noncovalent interactions formed by the hydroxyl group on its terminal aromatic ring. Compounds 7N and 2N, which have low activity, register MM-GBSA values (−34.95 and −35.80 kcal mol−1, respectively) and IFD scores (−8.924 and −8.941 kcal mol−1, respectively) that are both weaker than the corresponding values of EPR (EPR's IFD score is −9.783 kcal mol−1), and their binding conformations are also inferior to those of EPR. This study verifies the reliability of this method: the ranking generated by IFD/MM-GBSA matches the experimental IC50 data of the entire compound series, eliminating the possibility of random hits, and finally leads to the conclusion that the compound ranking for the α-glucosidase target is far more complex. The IFD scores showed generally less variability compared to the AR scores and ranged between −6.661 and −7.481 kcal mol−1 (except for ACR, which had a score of −10.059 kcal mol−1) for the test compounds. This shows that α-glucosidase active sites impose more uniform steric hindrance on the chalcone scaffold. In this case, 9N was the best IFD score (−8.379 kcal mol−1), followed by 6N (−7.481 kcal mol−1) and 2N/4N (−7.060 and −7.057 kcal mol−1). Upon MM-GBSA rescoring, there was a qualitative reordering among 4N, 6N, ACR and 9N. 4N obtained the most favourable ΔGbind (−50.05 kcal mol−1) ahead of 6N (−48.09 kcal mol−1) and ACR (−45.95 kcal mol−1). Interestingly, 9N went down from first to fourth position with a ΔGbind of −45.43 kcal mol−1. The difference in ranking between IFD and MM-GBSA for α-glucosidase has mechanical significance, in that 9N, in the geometrical sense, is capable of docking into a favourable pose, but then must subsequently relax and implicitly solvate, as the 4N–enzyme complex, being more thermodynamically stable, is preferentially adopted. The better performance of MM-GBSA of 4N can be explained by the ability of the hydroxyls of 4N to form more hydrogen bonds, beyond those considered by the scoring function alone, as elaborated in binding mode analysis. Compounds 3N (−34.93 kcal mol−1) and 2N (−36.87 kcal mol−1) had the weakest MM-GBSA towards α-glucosidase, consistent with their higher IC50 higher IC50 values (99.000 ± 3.08 and 86.625 ± 2.68 nM, respectively), providing additional confirmation that neither the methoxy (3N) nor unsubstituted B-ring (2N) provides a sufficiently complementary fit to the α-glucosidase binding cavity. The computational binding energetics of this compound series, as shown by the IFD and MM-GBSA analyses, exhibit internal self-consistency and good correlation with the in vitro biological data, which bolsters confidence in the subsequent structural interpretations. To quantify the agreement between the calculated and the experimental data, the IFD scores and the prime/MM-GBSA binding free energies of Table 7 were correlated with the experimental pIC50 values derived from Table 2 (Fig. 5). For aldose reductase, both descriptors reproduce the experimental ranking: the IFD score gives a Spearman rank correlation of ρ = −0.964 (p < 0.0001; Pearson r = −0.931), and the MM-GBSA ΔGbind reproduces the experimental order of the ten compounds exactly (ρ = −1.000, p < 0.0001; Pearson r = −0.996). For α-glucosidase, the IFD score alone correlates only moderately with potency (ρ = −0.559, p = 0.093), which is consistent with the narrow range of IFD scores obtained for this target and with the re-ranking of 9N discussed above, whereas MM-GBSA rescoring restores an excellent agreement (ρ = −0.985, p < 0.0001; Pearson r = −0.961). This contrast quantifies why the MM-GBSA step is required for α-glucosidase and confirms that the retained poses are not fortuitous. It is also worth noting that the experimental potencies against the two enzymes are themselves strongly rank-correlated (ρ = +0.960, p < 0.0001), which provides direct experimental support for the dual-target interpretation of the series (Fig. 5). Correlation between the calculated prime/MM-GBSA binding free energies (Table 7) and the experimental pIC50 values (Table 2) of compounds 1N–10N against (A) α-glucosidase and (B) aldose reductase. Solid lines are least-squares fits to the ten test compounds; the reference inhibitors (acarbose and epalrestat, open diamonds) are shown for comparison and were not included in the fits. Spearman rank (ρ) and Pearson (r) correlation coefficients are given in each panel. The AR active site is characterized as a hydrophobic barrel capped by a flexible anion-binding site. A tetrad comprising TYR48, HIS110, LYS77 and ASP43 provides the main pharmacophoric environment. The NADPH cofactor's nicotinamide ring specifies the hydrophilic part, and a neighbouring hydrophobic subpocket accommodates the aromatic or aliphatic substituents of many known AR inhibitors. Examination of the docked pose of 4N within AR reveals an exceptionally rich network of specific interactions (Fig. 6A). The para-position amine group (–NH2) of the A-ring interacts with TYR48 and GLN183 through hydrogen bonds and thus anchors this polar pharmacophore at the interface of the catalytic tetrad and outer hydrophilic region. An extra hydrogen bond exists with THR113 of the B-ring (−OH) due to a meta-hydroxyl group that has been implicated in several substrate-based and inhibitor-based recognition in AR crystal structures. This tripartite hydrogen-bond network, NH2 donating to TYR48 and GLN183, and –OH donating to THR113, offers the remarkable feature of pinning down the orientation of the molecules in a well-defined manner without losing hydrophobic coverage. The hydrophobic subpocket that binds the two aromatic rings of the chalcone scaffold through van der Waals contacts with TRP20, TRP219, TRP111 and VAL130, whereas π–π stacking interactions between the B-ring and the indole systems of TRP79 and PHE122 lead directly to the large binding enthalpy reflected in the MM-GBSA ΔGbind of −65.35 kcal mol−1. The explanation is the high AR potency of a compact molecular architecture (MW 239.3 Da) comprising a large number of geometrically complementary hydrogen bonds and extensive stacking of aromatic moieties. 2D interaction map (left) and 3D docking pose (right) of compounds 4N (A) and 6N (B) bound to the AR enzyme. Compound 6N attaches to AR in a way that is related to but mechanistically distinct from the other compounds, as shown in Fig. 6B. The NH2 group again participates in hydrogen bonding, engaging SER159 and GLN183, the latter being a shared anchoring residue with 4N. However, 6N substitutes the mono-hydroxyl B-ring of 4N with a 3,4-difluorophenyl moiety, fundamentally altering the interaction chemistry of the terminal ring. The strong electronegativity and moderate polarizability of the two fluorine atoms engender halogen-bond-like interactions and C–H⋯F contacts with residues lining the hydrophobic pocket. These interactions increase the van der Waals surface complementarity. The electron-withdrawing character of the gem-difluoro substituent also affects not only the furan chromophore but also the electronic density of the entire conjugated chalcone π-system, potentially enhancing the π–π stacking interaction with TRP79 seen in the 2D interaction map. Although there is no additional hydrogen bond to THR113 mediated by –OH in 4N, the result is a very efficient hydrophobic contact landscape. This is responsible for the MM-GBSA ΔGbind difference between 4N and 6N, which is around 12 kcal mol−1. The catalytic mechanism of α-glucosidase occurs through a conserved ASP-HIS dyad, in which ASP542 is the nucleophile, and ASP327 is the general acid/base. The active site additionally contains a rich hydrophobic environment furnished by multiple tryptophan residues (TRP406, TRP441, TRP539) that are critical for substrate accommodation and have been implicated in the binding of phenolic inhibitors through π-stacking. For 4N bound to α-glucosidase, the 2D interaction map indicates that the B-ring hydroxyl group provides dual hydrogen bonding to ASP542 and HIE600, the latter being a histidine in the HIE protonation state that is adjacent to the catalytic dyad. These interactions resemble features of the substrate transition state, where the hydroxyl moiety occupies a glycosidic oxygen-like position and thereby acts as a competitive inhibitor of substrate binding. The NH2 group of 4N has a hydrogen bond with ASP203, thereby extending the interaction network at the edge of the active site. The A-ring of compound 4N simultaneously interacts with TRP406 via π–π stacking while forming hydrophobic contacts with TRP539, TRP441, ILE328, MET444, and PHE575 through its extended conjugated system. Thus, compound 4N showcases such a complete pharmacophoric match within the α-glucosidase pocket due to its multi-faceted, simultaneous engagement of the catalytic dyad via –OH, a polar residue at the periphery via –NH2, and a hydrophobic tryptophan array via π–π stacking and hydrophobic contacts (Fig. 7A). This explains why the MM-GBSA value for 4N is the most favourable at −50.05 kcal mol−1 and why the α-glucosidase IC50 of 4N is lowest (23.896 ± 0.78 nM) in the series. 2D interaction map (left) and 3D docking pose (right) of compounds 4N (A) and 6N (B) bound to the α-glucosidase enzyme. In α-glucosidase, compound 6N adopts a somewhat different binding mode (Fig. 7B). The NH2 group remains engaged in a hydrogen bond with ASP542, while the 3,4-difluorophenyl B-ring takes part in halogen bond interactions with TRP441 and ASP443. The geminal fluorine atoms, by forming polarization-assisted halogen contacts with these residues, effectively substitute for the –OH-mediated hydrogen bond to HIE600 seen in 4N, though with somewhat less energetic efficiency (ΔGbind = −48.09 vs. −50.05 kcal mol−1). The broader hydrophobic envelope of 6N's interaction surface—encompassing TRP539, ARG526, ILE364, MET444, ASH327, ILE328, TRP406, PHE575, HIE600, and TYR299—is more extensive than that of 4N, reflecting how the fluorinated ring's increased polarizability and steric bulk maximize van der Waals contacts with the aromatic lining of the pocket. The net result is an inhibitory potency (IC50 = 34.650 ± 1.09 nM) that is not higher than 4N, but is higher than acarbose, making 6N an equally promising lead molecule for α-glucosidase inhibition. The comparison of binding between 4N and 6N to both enzyme targets provides an important structure–activity relationship in this series of chalcones. The binding mode and inhibitory potency seem to be strongly affected by the nature of B-ring substituents. For compound 4N, the hydroxyl group gives rise to favourable interactions with polar residues via hydrogen bonding, which leads to increased thermodynamic stabilization. In the latter case, the fluorine substituents in 6N deliver binding through hydrophobic interactions and halogen-bond-assisted contacts, a different, though also highly effective, binding strategy. Although both compounds use different interaction patterns, both are more potent than analogues containing unsubstituted or weakly electron-donating B-rings, a result that may explain the lower potency of compounds 2N, 3N, and 7N. Molecular dynamics simulations In order to check the dynamic stability of compound 4N within the active site of α-glucosidase, we performed a 500 ns molecular dynamics (MD) simulation. The evaluation of the simulation results was done by protein–ligand interaction analysis, RMSD, and RMSF, and interaction fingerprint profiles. Analysis of interactions in two dimensions (Fig. 8a) shows that 4N is involved in several stable interactions in the α-glucosidase binding pockets. The hydroxyl group on the phenyl ring formed a hydrogen bond with ASP443; this bond was maintained during 95% of the simulation time, making it the most stable interaction. The ligand interacts with ARG334, ASP542, and ASP443 via water-mediated hydrogen bonds. In addition, PHE575's hydrophobic interaction with the aminophenyl ring and TRP406 and MET444's hydrophobic stabilization were facilitated. Altogether, these interactions suggest that the binding of 4N is primarily driven by polar interactions with supporting hydrophobic contacts. Results of the 500 ns molecular dynamics simulation of the 4N-α-glucosidase complex. (a) 2D interaction map, (b) protein and ligand RMSD plots, (c) protein RMSF profile, (d) ligand RMSF profile, and (e) interaction fraction histogram. The protein RMSD analysis revealed that the Cα atoms remained stable during the 500 ns simulation, with values varying from approximately 0.75 to 1.05 Å (Fig. 8b). This implies the overall protein structure was maintained well. The RMSD of the ligand relative to the protein was around 0.6 to 2.7 Å, indicating that 4N had limited movement within the binding pocket. The ligand RMSD (relative to itself) remained low (0.1–0.3 Å), indicating the ligand's internal conformation was not substantially altered during simulation.RMSF analysis showed that most residues of α-glucosidase exhibited fluctuations below 0.9 Å, indicating that the enzyme maintained structural stability (Fig. 8c). Higher fluctuations were noted only in the N-terminal region and in several loop regions distant from the active site. In particular, residues in the binding pocket showed low RMSF values, indicating that the structure of the active site stayed unchanged during the simulation. Quantitatively, the statistics reported below were computed over the last 200 ns of the 500 ns trajectory (the 300–500 ns production window), that is over the interval in which the convergence criteria defined in the Experimental Section were satisfied; the corresponding values for the complete 500 ns run are given in parentheses for comparison. The Cα RMSD of α-glucosidase averaged 1.05 ± 0.04 Å (1.01 ± 0.08 Å over the full 500 ns), the ligand RMSD after alignment on the protein (Lig fit Prot) averaged 1.88 ± 0.25 Å (1.73 ± 0.35 Å), and the internal ligand RMSD (Lig fit Lig) averaged 0.24 ± 0.07 Å (0.24 ± 0.07 Å). The protein RMSD reached its plateau within the first 150 ns, after which the block averages of the 300–400 ns and 400–500 ns windows differed by only 0.02 Å (1.06 versus 1.04 Å); the corresponding differences were 0.12 Å (1.94 versus 1.82 Å) for the ligand RMSD and 0.01 Å (0.25 versus 0.23 Å) for the internal ligand RMSD. All three quantities therefore meet the convergence criteria defined above. No RMSD jump exceeding 2.0 Å was observed for the ligand at any point of the trajectory, showing that 4N retained a single binding pose over the full 500 ns and remained within 1.88 ± 0.25 Å of its induced-fit docking pose, which constitutes an independent validation of the docking result. The aminophenyl ring was found to be the most flexible and achieved the highest RMSF value of approximately 1.1 Å (Fig. 8d) by Ligand RMSF analysis. The mobility of the chalcone linker and hydroxyphenyl ring was found to be lower, e.g., 0.4–0.75 Å. This difference corresponds with the greater anchoring effect of the ASP443 hydrogen bond, compared to the weaker hydrophobic interaction with PHE575. The interaction fingerprint analysis (Fig. 8e) revealed that the residue TRP406 was the most important residue for interacting with the other residues, followed by TYR299 and ASP443. ARG334 and ASP542 participated in water-mediated interactions. The complex was further stabilized by additional hydrophobic contacts between ILE328 and ILE364, TRP441, MET444, PHE575, and GLN603. Overall, these results showed that 4N has a stable binding mode in the active site of the α-glucosidase during the 500 ns simulation, and it is well conserved. A hydrophobic/aromatic π-interaction dominated the binding to the AR–4N complex as seen in the 2D interaction map. The central chalcone scaffold was surrounded by TRP111 (87%), TRP79 (82%), and TRP20 (65%), establishing an aromatic cage via π–π stacking and CH–π interaction. TYR209 (65%) made another aromatic interaction with the aminophenyl ring, and LEU300 (72%) made hydrophobic contact with the hydroxyphenyl ring. Only one polar interaction was seen – a water-mediated hydrogen bond from the aminophenyl NH2 group to SER159 (39%) (Fig. 9a). The results suggest that hydrophobic interactions are the major force involved in the binding of 4N to AR. The AR–4N complex showed stable protein behavior with Cα RMSD values staying within the range of 0.8–1.5 Å during the simulation. The Lig fit on Prot RMSD was in the range of about 0.8–3.2 Å, which is a reasonable range of ligand motion in the binding pocket. However, the Lig RMSD of Lig fit on Lig RMSD remained below 0.9 Å, indicating that the internal conformation of the ligand remained unchanged (Fig. 9b). The slightly higher RMSD values may be attributed to the flexible binding pocket of AR, which can bind different inhibitors. The majority of the AR residues showed low fluctuations (0.3–0.9 Å), indicating good structural stability. The higher fluctuation was observed around residue 120 (∼2.4 Å) and residue 210 (∼2.7 Å), which are located in the flexible loop regions. In particular, residues in the active site had low RMSF values, indicating that the binding pocket remained stable during the simulation (Fig. 9c). Results of the 500 ns molecular dynamics simulation of the 4N–AR complex. (a) 2D interaction map, (b) protein and ligand RMSD plots, (c) protein RMSF profile, (d) ligand RMSF profile, and (e) interaction fraction histogram. The RMSF values of the 4N atoms were relatively uniform, with values ranging between about 0.5 and 0.95 Å. The most pronounced changes were for atom 17 (∼0.95 Å) at the end of the hydroxyphenyl ring. The even distribution indicates that TRP20, TRP79, and TRP111 formed an aromatic cage and were good stabilizers throughout the structure of the ligand (Fig. 9d). The results from interaction fingerprint analysis confirmed those results of the 2D interaction map, with the key interacting residues being TRP20 (∼1.1), TRP111 (∼1.0), and TYR209 (∼0.95). LEU300 (∼0.6) and LYS77 (∼0.75) also contributed to the hydrophobic contacts in the stabilization of the ligand. Weaker interactions were seen in SER159 (∼0.17), which is consistent with the hydrogen bond that is mediated by water. Additional contributions from HIS110, PHE115, PHE122, GLN183, VAL297, CYS298, and CYS303 further emphasize the predominantly hydrophobic environment surrounding 4N in the AR active site (Fig. 9e). Applying the same convergence criteria and the same 300–500 ns production window of the 500 ns trajectory to the 4N–AR complex, the Cα RMSD averaged 1.26 ± 0.10 Å (1.11 ± 0.20 Å over the full 500 ns), the Lig fit Prot RMSD 2.33 ± 0.24 Å (2.26 ± 0.28 Å) and the internal ligand RMSD 0.59 ± 0.12 Å (0.61 ± 0.13 Å). Equilibration of this complex was slower than that of the α-glucosidase complex: the protein RMSD rose from 0.87 ± 0.11 Å over the first 100 ns and reached its plateau after approximately 250 ns, and the residual drift between the 300–400 ns and 400–500 ns block averages was only 0.03 Å (1.25 versus 1.28 Å), the corresponding difference for the ligand RMSD being 0.05 Å (2.31 versus 2.36 Å). Both values lie well within the convergence tolerance defined above. Although the ligand RMSD band of the AR complex is broader than that of the α-glucosidase complex, the low internal ligand RMSD (0.59 ± 0.12 Å), the absence of a monotonic drift and the constancy of the interaction fingerprint over the same interval show that this behaviour reflects a rocking motion of 4N within the hydrophobic aromatic cage formed by TRP20, TRP79 and TRP111 rather than a progressive loss of binding; here too the ligand remained within 2.33 ± 0.24 Å of its docked pose throughout the simulation. ADME/pharmacokinetic profiling and drug-likeness assessment In vitro biological activity alone is not sufficient for a drug candidate. The pharmacokinetic profile of the compound will ultimately determine whether the target can be engaged in vivo. Accordingly, the ADME as well as drug-likeness properties for all compounds 1N–10N were predicted using the QikProp module (Table 8), and the results were evaluated for drug-likeness based on Lipinski's Rule of Five (RoF) and Jorgensen Rule of Three. Table 8Predicted ADME (absorption, distribution, metabolism, and excretion) properties of the synthesized compoundsCompoundMW (g mol−1)DonorHBAccptHBQP log Po/wQP log SQP log HERGQPPCacoQP log BBQPPMDCK% HOARule of fiveRule of three1N257.7191.5003.0003.051−3.825−5.531907.776−0.597967.257100.000002N237.3011.5003.0002.951−3.807−5.546907.776−0.765445.585100.000003N267.3271.5003.7503.132−4.065−5.901907.776−0.943445.585100.000004N239.2732.5003.7501.873−2.946−5.494273.914−1.314122.03381.541005N279.3811.5003.0003.837−4.700−5.650907.776−0.928445.585100.000006N259.2551.5003.0003.084−3.931−5.405907.776−0.5361365.559100.000007N253.3001.5003.7502.748−3.459−5.457907.776−0.813445.58595.979008N291.2721.5003.0003.479−4.409−5.562909.980−0.5051690.240100.000009N303.3601.5003.7503.640−4.634−6.153907.776−0.869445.585100.0000010N257.7191.5003.0003.117−3.967−5.550907.776−0.5841098.228100.00000 Predicted ADME (absorption, distribution, metabolism, and excretion) properties of the synthesized compounds None of the compounds 1N–10N violate Lipinski's RoF (zero violations) and Jorgensen's RoF. The molecular weights ranged from 237.3 to 303.4 (Da), which is below the upper limit of oral bioavailability (≤500 Da). Further, total hydrogen bond donors (1.5–2.5) and acceptors (3.0–3.75) were consistently below required limits. Log P values (QP log Po/w) ranged from 1.873 to 3.837, suggesting a balanced lipophilic profile that may support passive membrane permeability while avoiding excessive hydrophobicity. The predicted aqueous solubility values ranged from −2.946 to −4.700. The solubility of compound 4N (−2.946) was greatest, suggesting higher solubility than the more lipophilic analogues. However, poor aqueous solubility is often a problem in drug delivery and this may be beneficial to the solubility of 4N. This effect could be due to the hydroxyl group increasing the molecular polarity. High QPPCaco values (445–1690 nm s−1) were obtained for most compounds, suggesting good intestinal permeability potential. The value of compound 4N was lower (273.9 nm s−1) than most of the others, possibly as a result of its higher hydrophilicity due to the hydroxyl group. This value, however, is still in the moderate-to-good permeability range (25 nm s−1), indicating good absorption. MDCK permeability (QPPMDCK) exhibited a similar trend. Compound 6N had a very high value (1365.6 nm s−1), second only to 8N (1690.2 nm s−1). This could be attributed to the increased lipophilicity of difluoro-substituted compounds and to the lack of a free hydroxyl group. All compounds are predicted to have low blood–brain barrier penetration based on the QP log BB values (−0.505 to −1.314). The value of QPlogBB was lowest for compound 4N (−1.314), which is a reflection of its higher polarity. Compounds 6N (−0.536) and 8N (−0.505) are nearer to the CNS penetration threshold but show brain exposure limits. Low CNS penetration is generally desirable because these compounds are targeted for peripheral use in diabetic complications like neuropathy, retinopathy, and cataract, and therefore, CNS penetration is not of much concern. The human oral absorption (HOA) ranged up to ∼100% for most of the compounds, at 81.5% for 4N, and 96.0% for 7N, which are considered to be acceptable. The range of values for QP log HERG was from −5.405 to −6.153, a range that is generally favorable with regard to hERG liability. However, the presence of a basic amine group and a planar aromatic scaffold in all of the compounds might still be relevant for hERG interaction; hence, experimental studies would be useful to support these computational predictions. Finally, docked, molecular dynamics-simulated, and ADME analysis results showed that 4N and 6N were the most promising compounds in the series. The binding affinity of both compounds toward aldose reductase and α-glucosidase was high and well sustained during all the simulations. Compound 4N showed increased polar interactions, and 6N showed increased hydrophobic and halogen-associated interactions contributing to the overall stability. Furthermore, both compounds exhibited promising pharmacokinetic characteristics, which further supported their potential as lead compounds for further optimization and experimental testing. Density functional theory (DFT) DFT calculations at the B3LYP/6-311G level of theory were used to further elucidate the electronic properties, chemical reactivity, and intermolecular interaction potential of the most active compounds identified by biological and computational screening. The two lead compounds 4N and 6N were investigated along with the reference compounds acarbose and epalrestat to gain comparative insight into their electronic behavior. The optimized molecular geometry, frontier molecular orbital distribution, global reactivity descriptors, and molecular electrostatic potential maps were used to estimate their stability, charge transfer ability, and reactivity. The quantum chemical parameters calculated and presented in Table 9, optimized structures (Fig. 10), energy gap (Fig. 11), and molecular electrostatic potential (MEP) surfaces (Fig. 12) are useful tools for understanding the structural and electronic characteristics that underlie the observed biological activities of the compounds investigated. Table 9HOMO, LUMO, energy gap, and the global quantum reactivity of the lead compounds and reference drugs were calculated at the B3LYP/6-311G levelParameterFormulaAcarboseEpalrestat4N6NDipole moment—9.2017176.9573875.8030315.460010Electronic energy (Hartree)—−2387.101829−1656.197785−784.598499−907.869177LUMO (eV)—−0.3962−2.4381−3.0558−2.1766HOMO (eV)—−5.8703−5.9909−6.2926−5.8695Energy gap (eV) E HOMO − ELUMO5.47413.55283.23683.6929Electron affinity (A, eV) A = −ELUMO0.39622.43813.05582.1766Ionization potential (I, eV) I = −EHOMO5.87035.99096.29265.8695Chemical potential (µ, eV) µ = 1/2 (I + a)3.13334.21454.67424.0231Electronegativity (χ, eV) χ = −1/2 (I + a)−3.1333−4.2145−4.6742−4.0231Chemical hardness (η, eV) η = 1/2 (I − A)2.73711.77641.61841.8465Chemical softness (S, eV−1) S = 1/η0.36540.56290.61790.5416Electrophilicity index (ω, eV) ω = 2 (µ2/η)1.79344.99946.74994.3827Nucleophilicity index (N, eV−1) N = 1/ω0.55760.20000.14810.2282Additional electronic charge= − µ/η−1.1448−2.3725−2.8882−2.1788 HOMO, LUMO, energy gap, and the global quantum reactivity of the lead compounds and reference drugs were calculated at the B3LYP/6-311G level Optimized structures of lead compounds and reference drugs. Comparative energy gap diagram of lead compounds and reference drugs. MEP structures of lead compounds and reference drugs. Optimized geometry analysis The electronic features and chemical reactivity of lead compounds (4N and 6N), along with reference drugs, namely acarbose and epalrestat, were analyzed by DFT calculations. The optimized geometries of the molecules obtained at the B3LYP/6-311G level are shown in Fig. 10. All compounds were optimized using geometry optimization, where all the optimized structures showed no imaginary frequencies and thus are indeed true minima on the potential energy surface. Structures were optimized that maintained the conjugated framework of the chalcone scaffold, which allowed for electron delocalization across the entire molecule backbone. The binding affinity toward biological targets is enhanced by the intermolecular recognition process, which is facilitated by such extensive π-conjugation. The calculated electronic energies shown in Table 9 were also found to support the thermodynamic stability of the optimized molecules: the lowest total electronic energy (−907.869177 hartree) was obtained for molecule 6N, which contained the difluoro substituents, which provide stronger stabilization of the electronic energy. Frontier molecular orbital (FMO) analysis The FMO analysis is a useful tool for studying the molecular reactivity, electron transfer ability, and chemical stability. The energies of the HOMO and LUMO of the compounds studied are summarized in Table 9, and their comparative distribution of the energy gap is shown in Fig. 11. The energy of the HOMO is a measure of the molecule's ability to donate electrons, while the energy of the LUMO is a measure of the molecule's ability to accept electrons. The energy of the HOMO level of compound 4N was the lowest (−6.2926 eV), which means that it has the highest resistance to electron donation and therefore the highest molecular stability in comparison to the reference compounds. The HOMO value of 6N was calculated as −5.8695 eV, which is very close to the value of acarbose (−5.8703 eV), suggesting a comparatively higher electron-donating ability. The LUMO energies followed the order 4N (−3.0558 eV) > epalrestat (−2.4381 eV) > 6N (−2.1766 eV) > acarbose (−0.3962 eV). The LUMO energy is much lower for 4N, indicating that 4N has a stronger electron-attracting ability and is more capable of charge-transfer interaction with biological macromolecules. This is a characteristic feature of high binding affinity and biological activity. An important characteristic of chemical reactivity and kinetic stability is the HOMO–LUMO energy gap (ΔE). Smaller energy gaps are typically associated with increased reactivity and increased ease of electronic excitation. As shown in Table 9 and Fig. 11, compound 4N displayed the smallest energy gap (3.2368 eV), followed by epalrestat (3.5528 eV), 6N (3.6929 eV), and acarbose (5.4741 eV). The smaller energy gap (4N) means higher potential for charge transfer and increased chemical reactivity, which could account for its increased inhibitory effects on both aldose reductase and α-glucosidase. On the other hand, the higher energy gap of acarbose indicates higher electronic stability and lesser reactivity. Based on the results obtained with the FMO, the electronic profile of the molecule 4N is thus the most favourable, which is in good agreement with the best biological performance of the 4N molecule obtained in the docking and molecular dynamics study. Global reactivity parameters HOMO and LUMO energies were utilized to calculate global quantum chemical descriptors that give further information about the reactivity and stability of the studied compounds. The calculated parameters are given in Table 9. Ionization potential (I) is the amount of energy needed to remove an electron from a molecule. The ionization potential of compound 4N was found to be the highest (6.2926 eV), which shows that it has a high resistance to the removal of an electron and also has a high electronic stability. Likewise, the highest EA value of 4N (3.0558 eV) was observed for all the compounds, indicating an electron-accepting property during molecular interaction. Chemical hardness (η) is the resistance to deformation of the electron cloud, while molecular polarizability (S) is regarded as softness. The compound 4N possessed the lowest hardness (1.6184 eV) and the highest softness (0.6179 eV−1), and had a higher electronic flexibility and better adaptation to interaction with biological receptors. This greater softness may enable better redistribution of charges, which will result in greater interaction with the protein. The electrophilicity index (ω) is one of the important parameters that measures the electron-accepting property of a molecule. The value of the electrophilicity (6.7499 eV) was the highest for 4N among all the compounds, which is far higher than those for epalrestat (4.9994 eV), 6N (4.3827 eV), and acarbose (1.7934 eV). This result suggests that the structure of 4N is highly stabilized towards the incoming electronic charge, and it is able to effectively engage in electrophilic interaction. Nucleophilicity index (N), on the other hand, was lowest for 4N (0.1481 eV−1), indicating its higher electrophilic nature. The calculated chemical potential (µ) and electronegativity (χ) further support these findings. The compound 4N had the maximum chemical potential (4.6742 eV) and electronegativity (−4.6742 eV), which means that it has a higher tendency to attract electrons. In addition, the strongest capability of charge transfer interactions was observed for 4N with a maximum electronic charge acceptance value (ΔNmax) of −2.8882. All of these descriptors suggest that 4N has the best compromise between reactivity, softness, and electrophilicity of the compounds studied, and is well matched by its highest enzyme inhibitory activity. Molecular electrostatic potential (MEP) analysis The molecular electrostatic potential surfaces of the lead compounds and reference drugs are shown in Fig. 12. The mapping of MEP allows the visualization of the distribution of charges in a molecule and the identification of the regions that are liable to electrophilic or nucleophilic attack. The MEP surfaces showed that there are both electron-rich (negative potential) and electron-deficient (positive potential) regions scattered across the molecular frameworks. The electrostatic potential analysis of compound 4N showed the most negative electrostatic potential around the hydroxyl oxygen atom and around the carbonyl oxygen atom of the chalcone part. These regions were represented as red-colored areas on the MEP surface, which showed the possible regions for hydrogen bond acceptance during protein binding. The amino group had moderate positive potential, indicating it would be a hydrogen-bond donor. This complementary distribution of charge may play a role in the significant hydrogen-bonding interactions that have been seen in molecular docking and molecular dynamics simulations. In the case of compound 6N, the negative charge was mainly centered around the carbonyl oxygen atom and around the Fluorine substituents. Local electron-rich units that could be involved in halogen-bond-assisted interactions were created by the presence of electronegative fluorine atoms. The charge distribution of 6N was more uniformly spread around the aromatic framework, making it more stable to interact with enzyme active sites that are both hydrophilic and halogen-bonding. The MEP maps of acarbose and epalrestat also revealed the existence of strong negative regions around the oxygen-containing functional groups, but the electron density distribution was not as concentrated as that of 4N. The charge separation of 4N is strongly pronounced, and this can contribute to its higher binding affinity and biological activity, partially accounting for it. In general, the DFT investigations done at the B3LYP/6-311G level gave good insights into the electronic behaviour and reactivity of the lead compounds. The optimized geometries of the molecules obtained in Fig. 10 also showed that all the molecules are structurally stable, whereas the FMO analysis in Fig. 11 revealed that molecule 4N had the smallest energy gap between HOMO and LUMO, which meant that it had greater electronic reactivity and capability to transfer charge. The global reactivity parameters (Table 9) also confirmed 4N as the softest and most electrophilic molecule in the series, which points to its better adaptability to the target molecules of biological studies. Furthermore, the MEP surfaces shown in Fig. 12 showed good distributions of charge, which are suitable for hydrogen bonding and electrostatic interactions. The results as a whole support the docking and molecular dynamics results, and the biological results, making compound 4N the best lead candidate of the molecules investigated. Comparison with previously reported α-glucosidase and aldose reductase inhibitors A direct numerical comparison of IC50 values obtained in independent studies is complicated by the fact that both assays used here are strongly protocol-dependent. For yeast α-glucosidase in particular, the IC50 reported for acarbose varies by more than two orders of magnitude between laboratories (from approximately 1 µM to approximately 357 µM) depending on the enzyme lot and activity, the p-NPG concentration, the pre-incubation time, the buffer, and the detection format.65 The same applies to aldose reductase, whose apparent potency depends on the enzyme source (recombinant human AKR1B1 versus tissue-purified enzyme) as well as on the NADPH and dl-glyceraldehyde concentrations and on the pH of the phosphate buffer.66 Under our conditions, acarbose and epalrestat were assayed in parallel with 1N–10N and gave IC50 values of 53.307 ± 1.67 nM and 7.294 ± 0.23 µM, respectively. It should also be noted that only the α-glucosidase activity of this series lies in the nanomolar range, the aldose reductase activity being low-micromolar (2.457–13.860 µM). For these reasons, the comparison presented below is based primarily on the fold-change relative to the reference inhibitor co-assayed in the same study (Table 2), which is the assay-independent quantity, and on size- and lipophilicity-corrected potency metrics, rather than on absolute IC50 values. Within the chalcone chemical space, the reported α-glucosidase inhibitors are predominantly micromolar. A recent survey of chalcone-type α-glucosidase inhibitors places the great majority of the published IC50 values between approximately 0.4 and 424 µM,65 and the chalcone and chalcone-hybrid series examined as antidiabetic agents in recent years—among them coumarin–chalcones, azachalcones, bis-chalcones and pyrrolidinyl chalcones—typically improve on the co-assayed acarbose control by about one order of magnitude at best.2,12,17–20 The present series behaves differently in two respects. First, all ten derivatives, including the weakest member 3N (99.000 ± 3.08 nM), are active at sub-100 nM concentrations, so that the series covers a narrow potency window of roughly fourfold between 4N and 3N instead of the broad activity spread that is characteristic of chalcone libraries. Second, this level of activity is reached without the polyhydroxylated or polymethoxylated B-rings on which most potent chalcone inhibitors rely, because the para-amino group of the A-ring is kept constant throughout the series and supplies a permanent hydrogen-bond donor. In the closest published analogue of the present design concept, hydroxychalcones were deliberately developed as dual aldose reductase/α-glucosidase agents;2 the present work shows that a single para-amino substituent on the A-ring, combined with a single meta-hydroxy (4N) or a 3,4-difluoro (6N) substituent on the B-ring, is sufficient to reproduce that dual profile and, on a per-atom basis, to exceed it. For aldose reductase, the clinically investigated inhibitors are dominated by two chemotypes, namely carboxylic acids (epalrestat, tolrestat, zopolrestat, ranirestat) and cyclic imides or hydantoins (sorbinil, fidarestat). On the isolated enzyme these compounds are highly potent—reported IC50 values are in the low nanomolar range for zopolrestat (ca. 5 nM), fidarestat (ca. 9 nM) and tolrestat (ca. 24 nM) and in the high nanomolar range for sorbinil (ca. 0.7 µM)66—and the derivatives described here, with IC50 values of 2.457–13.860 µM, are clearly less potent than these optimised clinical candidates in absolute terms. The development of that class, however, has been limited not by enzymatic potency but by the intrinsic properties of the two chemotypes: the strongly acidic carboxylate required for anchoring in the anion-binding pocket also restricts passive membrane permeability and tissue distribution, whereas the hydantoin ring has been associated with hypersensitivity reactions and idiosyncratic toxicity, and several candidates were discontinued for these reasons.7,11,66 Epalrestat, which remains the only widely marketed aldose reductase inhibitor, was therefore used as the practically relevant benchmark, and six of the ten derivatives (1N, 4N, 6N, 8N, 9N and 10N) were more potent than epalrestat under identical assay conditions, 4N being 2.97-fold more potent. The distinguishing feature of the present compounds is thus not the absolute enzymatic potency against aldose reductase but the fact that this level of activity is obtained from a small, neutral, non-acidic and non-hydantoin scaffold that carries no ionisable centre at physiological pH, violates none of the Lipinski or Jorgensen rules (Table 8) and, unlike most reported aldose reductase inhibitors, is simultaneously active against α-glucosidase. Because absolute IC50 values are assay-dependent whereas molecular size is not, the present series was further compared with the reference drugs by means of size- and lipophilicity-corrected potency metrics, namely the ligand efficiency (LE = 1.37 × pIC50/number of heavy atoms), the lipophilic ligand efficiency (LLE = pIC50 − Q Plog Po/w) and the ligand-efficiency-dependent lipophilicity (LELP = QP log Po/w/LE).67,68 These descriptors were calculated from the IC50 values of Table 2 and the QPlogPo/w values of Table 8 and are collected in Table 10. Against α-glucosidase, every member of the series displays an LE between 0.443 and 0.580 kcal mol−1 per heavy atom, whereas acarbose—a pseudotetrasaccharide with 44 heavy atoms—reaches only 0.226 kcal mol−1 per heavy atom despite a comparable IC50. Compound 4N is therefore 2.6 times more ligand-efficient than acarbose, and even the least efficient member of the series (9N) is almost twice as efficient. The same holds for aldose reductase, where 4N attains an LE of 0.427 kcal mol−1 per heavy atom against 0.335 for epalrestat. Compound 4N is in addition the most lipophilicity-efficient member of the series (LLE = 5.75 and 3.74 against α-glucosidase and aldose reductase, respectively; LELP = 3.23), which indicates that its superior potency is driven by specific polar contacts—the hydrogen bonds to ASP542 and HIE600 in α-glucosidase and to TYR48, GLN183 and THR113 in aldose reductase discussed above—and not by increased lipophilicity. Conversely, the bulkier and more lipophilic analogues 5N and 9N pay a high lipophilicity price for their activity (LELP = 8.27 and 8.22). From a lead-optimization perspective, this is the most concrete difference between the present derivatives and the majority of reported α-glucosidase and aldose reductase inhibitors: at 237–303 Da they retain roughly 200 Da of molecular-weight budget for further optimisation, whereas acarbose (645.6 Da) and most reported chalcone hybrids are already at or beyond the limits accepted for orally administered drugs. Table 10Size- and lipophilicity-corrected potency metrics of compounds 1N–10N and of the reference inhibitors, calculated from the IC50 values of Table 2 and the QP log Po/w values of Table 8aCompoundsHACpIC50 (α-gluc)LE (α-gluc)LLE (α-gluc)LELP (α-gluc)pIC50 (AR)LE (AR)LLE (AR)1N187.160.5454.115.605.170.3942.122N187.060.5384.115.494.880.3711.923N207.000.4803.876.534.970.3401.834N187.620.5805.753.235.610.4273.745N217.110.4643.288.275.030.3281.206N197.460.5384.385.735.370.3882.297N197.060.5094.315.404.860.3502.118N217.200.4703.727.415.230.3421.769N237.440.4433.808.225.320.3171.6810N187.240.5514.125.665.250.3992.13Acarbose447.270.226n.a.n.a.———Epalrestat21————5.140.335n.a.aHAC, heavy-atom count; LE = 1.37 × pIC50/HAC (kcal mol−1 per heavy atom); LLE = pIC50 – QP log Po/w; LELP = QP log Po/w/LE. pIC50 values were calculated as −log 10(IC50/M). LLE and LELP are not reported for the reference drugs (n.a.) because QP log Po/w values were not computed for them in this work. Size- and lipophilicity-corrected potency metrics of compounds 1N–10N and of the reference inhibitors, calculated from the IC50 values of Table 2 and the QP log Po/w values of Table 8a HAC, heavy-atom count; LE = 1.37 × pIC50/HAC (kcal mol−1 per heavy atom); LLE = pIC50 – QP log Po/w; LELP = QP log Po/w/LE. pIC50 values were calculated as −log 10(IC50/M). LLE and LELP are not reported for the reference drugs (n.a.) because QP log Po/w values were not computed for them in this work. The DFT results (Table 9) provide an electronic-structure rationale for this difference and, at the same time, distinguish the present compounds from the reference inhibitors at the mechanistic level. Acarbose combines a wide HOMO–LUMO gap (5.4741 eV), the lowest electrophilicity index of the set (1.7934 eV), and the highest chemical hardness (2.7371 eV); it is, in other words, an electronically inert carbohydrate mimic whose inhibitory action rests almost entirely on the shape and hydrogen-bonding complementarity of a large and flexible pseudo-oligosaccharide with the catalytic site. Compound 4N reaches a comparable—in this assay, superior—potency by an entirely different electronic route: its gap is only 3.2368 eV, its softness is the highest in the set (0.6179 eV−1), and its electrophilicity index (6.7499 eV) is 3.8 times that of acarbose and 1.35 times that of epalrestat. A soft, low-gap and strongly electrophilic π-system of this kind polarises readily inside the active site and can engage in charge-transfer interactions with the electron-rich tryptophan array (TRP406, TRP441 and TRP539 in α-glucosidase; TRP20, TRP79 and TRP111 in AR) that the MD interaction fingerprints identify as the dominant contact partners. The MEP surfaces support the same picture: in 4N the negative potential is sharply localised on the phenolic and carbonyl oxygen atoms while the aminophenyl group carries a complementary positive region, giving a well-separated donor/acceptor pair within a 239 Da framework, whereas in acarbose and epalrestat the negative potential is distributed over many oxygen atoms without comparable charge separation. It is also notable that 4N is electronically superior to epalrestat with respect to every descriptor relevant to charge transfer (gap 3.2368 vs. 3.5528 eV; ω 6.7499 vs. 4.9994 eV; S 0.6179 vs. 0.5629 eV−1; ΔNmax −2.8882 vs. −2.3725), which parallels the approximately threefold potency advantage measured in vitro and suggests that the electron-accepting, charge-transfer-driven binding of the aminochalcone chromophore is a genuine alternative to the anionic anchoring exploited by the carboxylic-acid aldose reductase inhibitors. Finally, the two features that separate this series most clearly from the reported inhibitors are its dual-target profile and its in vivo validation. Most reported α-glucosidase inhibitors, including the chalcone series cited above, are single-target agents evaluated only in vitro, and most aldose reductase inhibitors act on the polyol pathway alone. The few genuinely dual α-glucosidase/aldose reductase chemotypes described so far—hydroxychalcones2 and quinazolin-4(3H)-one-based glitazones, the latter reaching Ki values of 0.648 µM against α-glucosidase and 0.106 µM against aldose reductase5 —are considerably larger and synthetically more demanding than the present compounds, which are obtained in a single Claisen–Schmidt step from commercially available 4′-aminoacetophenone and the corresponding aldehyde. The dual character of the present series is moreover supported statistically, the potencies against the two enzymes being rank-correlated with ρ = +0.960 (p < 0.0001). In addition, the biochemical activity reported here translates into measurable pharmacology: at 20 mg kg−1 over 28 days, 4N, 6N and 9N lowered fasting blood glucose to 106.8–126.4 mg dL−1 in streptozotocin-induced diabetic rats, restored body weight, corrected the lipid profile and normalised SGOT, SGPT and ALP levels, with no adverse effect observed up to 2000 mg kg−1. The common structural features responsible for dual inhibition are the chalcone backbone, the α,β-unsaturated carbonyl linker, and the para-amino functionality, whereas the specialized features responsible for enhanced potency are the optimized B-ring substitutions, improved electronic properties (lower HOMO–LUMO gap, higher electrophilicity and softness), high ligand efficiency, and favorable hydrogen-bonding interactions predicted by docking. To the best of our knowledge, this combination of nanomolar α-glucosidase and low-micromolar aldose reductase inhibition, high ligand efficiency, a small non-acidic scaffold and confirmed in vivo antidiabetic and hepatoprotective activity has not previously been reported for the 4-aminochalcone class, and it is this combination rather than the absolute IC50 value alone that differentiates the present derivatives from the inhibitors described in the literature. Conclusions A series of ten 4-aminochalcone derivatives (1N–10N) was successfully synthesized and evaluated as potential dual α-glucosidase and aldose reductase inhibitors using integrated in vitro, in vivo, and computational approaches. All synthesized compounds exhibited remarkable inhibitory activity against both targets, with α-glucosidase IC50 values ranging from 23.896 ± 0.78–99.000 ± 3.08 nM and aldose reductase IC50 values ranging from 2.457 ± 0.08–13.860 ± 0.44 µM, demonstrating the strong antidiabetic potential of the 4-aminochalcone scaffold. Among the tested derivatives, compound 4N emerged as the most potent dual inhibitor, exhibiting IC50 values of 23.896 ± 0.78 nM against α-glucosidase and 2.457 ± 0.08 µM against aldose reductase, surpassing the reference drugs acarbose and epalrestat under the same experimental conditions. Compounds 6N and 9N also displayed excellent dual-target activity, with α-glucosidase IC50 values of 34.650 ± 1.09 and 36.473 ± 1.14 nM, respectively, and aldose reductase IC50 values of 4.225 ± 0.13 and 4.746 ± 0.15 µM, respectively. The in vivo evaluation in streptozotocin-induced diabetic rats further confirmed the antidiabetic efficacy of the active derivatives. Compounds 4N, 6N, and 9N significantly reduced fasting blood glucose levels, restored body weight, improved lipid profile parameters, and normalized elevated hepatic biomarkers in a dose-dependent manner. Notably, treatment with compound 4N at 20 mg kg−1 reduced blood glucose levels to near-normal values and exhibited efficacy comparable to the standard drug acarbose. Acute toxicity studies demonstrated a favorable safety profile, with no observable adverse effects at doses up to 2000 mg kg−1. Computational studies supported the experimental findings and revealed strong binding affinities of the active compounds toward both enzyme targets. Induced-fit docking, MM-GBSA calculations, and 500 ns molecular dynamics simulations identified stable ligand–protein complexes and highlighted the critical role of hydrogen-bonding and hydrophobic interactions in enzyme inhibition. Structure–activity relationship analysis demonstrated that the presence of a m-hydroxyl group on the B-ring substantially enhanced dual inhibitory activity, while electron-withdrawing substituents such as 3,4-difluoro and extended aromatic systems such as methoxynaphthyl also contributed positively to potency. The computational models were validated quantitatively, the prime/MM-GBSA binding free energies reproducing the experimental ranking with Spearman correlations of ρ = −0.985 for α-glucosidase and ρ = −1.000 for aldose reductase, and both 500 ns trajectories fulfilled explicit RMSD-based convergence criteria. Compared with previously reported inhibitors, the distinguishing features of this series are its high ligand efficiency (LE = 0.580 kcal mol−1 per heavy atom for 4N against α-glucosidase versus 0.226 for acarbose, and 0.427 versus 0.335 for epalrestat against aldose reductase), its small, neutral and non-acidic scaffold, its low-gap and strongly electrophilic electronic character, and the fact that the dual enzymatic activity is confirmed by in vivo efficacy. Furthermore, all compounds satisfied key drug-likeness criteria and exhibited favorable predicted ADME profiles. Collectively, these findings provide the first evidence that 4-aminochalcone derivatives can function as highly potent dual α-glucosidase and aldose reductase inhibitors with significant in vivo antidiabetic activity. Among the synthesized compounds, 4N represents the most promising lead candidate, while compounds 6N and 9N also warrant further investigation as potential multifunctional agents for the treatment of diabetes mellitus and its associated complications. Ethical approval Ethical approval for the study was obtained from the Institutional Research Ethics Committee (IREC), University of Malakand, Chakdara, Khyber Pakhtunkhwa, Pakistan, following the guidelines of the University of Malakand Animal Bylaws 2008, as per notification no. Pharm/EC-Thzp/41-08/21. Additionally, this study was conducted in accordance with the ARRIVE guidelines for the reporting of animal research. All relevant items from the ARRIVE 2.0 checklist have been addressed to ensure transparency and reproducibility. Author's contribution Ehsan Ullah Mughal: conceptualization, supervision, validation, investigation, resources, data curation, writing – original draft, writing – review and editing, visualization, project administration, funding acquisition. Nafeesa Naeem: experimental work performance, data collection, visualization, validation, formal analysis, investigation, writing – original draft, writing – review and editing. Songül Bayrak: biological and computational studies, writing – original draft. Namık Kılınç: biological and computational studies, writing – original draft. Bushra Shakoor: DFT study and formal analysis. Amina Sadiq: formal analysis. Magdi E. A. Zaki: visualization, validation and formal analysis. Maher Fathalla: visualization, validation and formal analysis. Sobhi M. Gomha: visualization, validation and formal analysis. Züleyha Almaz: biological and computational studies, writing – original draft. Conflicts of interest The authors have no conflicts of interest to declare. Supplementary Material