Skip to main content
RSC Advances logoLink to RSC Advances
. 2026 Nov 16. Online ahead of print. doi: 10.1039/d6ra09569k

From wastes to therapeutic treasure: phytochemistry, molecular docking, and DFT-based reactivity profiling of walnut green husk and shell

Abdullahi Ibrahim Uba a,†,✉, Mehmet Veysi Cetiz b,†, Gunes Ak c, Enver Saka c, Awad Alzebair d, Omayma A Eldahshan e,f, Abdel Nasser B Singab e,f, Mohamed M M Abdelrazek g, Sara Saeed Kotb g, Ismail Koyuncu b, Sakina Yagi h, Ilhami Gulcin i,j, Ismail Yapıcı k, Meng-Yao Li l,✉, Gokhan Zengin c
PMCID: PMC13772000  PMID: 43042029

Abstract

Agricultural by-products have attracted increasing scientific interest because of their rich phytochemical profiles and potential for use in various applications. The present study investigated the phytochemical composition, antioxidant activity, enzyme-inhibitory properties, and cytotoxic effects of the green husk and shell of Juglans regia L. (Juglandaceae). Ethyl acetate (EA), acetone, 70% acetone (hydroacetone), and aqueous extracts were obtained by maceration. Among the extracts examined, the hydroacetone extract of the shell contained the highest total phenolic content (67.43 mg gallic acid equivalent (GAE) per g), whereas the acetone extract of the green husk showed the greatest total flavonoid content (21.33 mg rutin equivalent (RE) per g). HPLC/MS profiling identified a diverse array of bioactive constituents like caffeic acid, protocatechuic acid, p-coumaric acid derivative, juglanins B and C, and hydrojuglone glucoside, and their distribution varied according to both the walnut fruit part and the extraction solvent. The hydroacetone shell extract exhibited the strongest antioxidant performance in the radical-scavenging activities, as well as reducing capacities. The aqueous shell extract and hydroacetone green husk extract showed the highest metal chelating activity, with values of 16.20 and 16.02 mg EDTAE per g, respectively. Similarly, the hydroacetone and aqueous shell extracts showed the greatest total antioxidant capacity (2.46 and 2.34 mmol trolox equivalent (TE) per g, respectively). Regarding enzyme inhibition, the hydroacetone and acetone extracts of the green husk recorded the highest acetylcholinesterase (2.77 mg galantamine equivalent (GALAE) per g) and butyrylcholinesterase (3.39 mg GALAE per g) inhibitory activity. The EtOAc shell extract revealed the highest tyrosinase (65.80 mg kojic acid equivalent (KAE) per g) and α-glucosidase (7.37 mmol acarbose equivalent (ACAE) per g) inhibitory activities, as well as cytotoxicity against the colorectal carcinoma (HCT-116) cell line (IC50 93.50 µg mL−1). To complement the experimental findings, network pharmacology, machine-learning analysis, and molecular docking were employed to explore compound–target associations and provide a computational assessment of the potential interactions of selected phytochemicals with relevant biological targets. These findings support the potential valorisation of these underutilized walnut by-products as natural sources of biologically active compounds for future food, pharmaceutical, and other biotechnological applications.


The walnut by-products can be valuable sources of biologically active compounds for future food, pharmaceutical, and other biotechnological applications.graphic file with name d6ra09569k-ga.webp

1. Introduction

Juglans regia L. (Juglandaceae), commonly referred to as the English or Persian walnut, is believed to have originated in regions of Central Asia and parts of the Balkan Peninsula. It is now cultivated extensively worldwide, particularly across Europe, America, and Asia, primarily for its edible kernel.1,2 The walnut fruit is generally spherical and enclosed within a hard shell characterized by a yellowish-brown outer surface with irregular markings. Structurally, the fruit consists of four principal components: the kernel, seed coat, shell, and pericarp, the latter commonly referred to as the green husk.3 Walnut is recognized as a nutritionally valuable dry fruit because of its high-quality proteins, minerals, dietary fiber, and carotenoids. Its oil is particularly rich in unsaturated fatty acids and vitamin E, contributing to its nutritional and health-promoting properties. In traditional medicine, different parts of J. regia have been used in the management of a wide range of conditions, including helminthiasis, diarrhea, sinusitis, stomach disorders, arthritis, asthma, eczema, scrofula, skin diseases, diabetes mellitus, anorexia, thyroid disorders, cancer, and infectious diseases.4 Numerous pharmacological properties have also been documented, including antimicrobial, antioxidant, anti-inflammatory, analgesic, antitumor, and antidiabetic activities.5,6 Phytochemical investigations have identified diverse bioactive constituents in different parts of the plant, including diarylheptanoids, quinones and their derivatives, phenolic compounds, and terpenes.5,6

Commercial walnut production is primarily focused on the edible kernel and, to a lesser extent, the wood, whereas the shell and green husk are generally regarded as processing by-products.7 The hard shell is generated in substantial quantities during kernel processing and is commonly discarded, incinerated, or used as a low-value fuel source. Such practices represent both a loss of potentially valuable resources and a potential environmental concern.7 Agricultural by-products have consequently attracted increasing research interest because their accumulation and uncontrolled disposal can contribute to environmental pollution. At the same time, the growing demand for naturally derived ingredients for applications in food, nutrition, cosmetics, pharmaceuticals, and other industrial sectors has encouraged the exploration of these materials as alternative sources of valuable compounds.8 In general, the reuse of inexpensive residues generated by food, forestry, and agricultural industries offers both environmental and economic advantages by reducing waste while creating opportunities for the recovery of value-added compounds.9

Although the green husk is considered a by-product of walnut production, it has a long history of traditional use. It has been employed in folk medicine for pain relief and in the treatment of conditions such as cancer, diabetes, microbial infections, and skin and cardiovascular disorders.3,10 In addition, the green husk is traditionally used in the preparation of walnut liqueur and has been shown to contain a diverse mixture of phenolic compounds.11 Walnut shell has also been investigated for non-pharmaceutical applications, including its use as a filtration material for the separation of crude oil from water.12 Recent research has further highlighted the potential of walnut shell and green husk as sources of naturally occurring bioactive compounds with nutritional and health-related applications.7,13 Previous investigations have reported antioxidant activity in walnut shells and have indicated that the material contains nutritionally relevant minerals such as phosphorus and iron.14 Furthermore, polysaccharides isolated from walnut shells and their derivatives have demonstrated antioxidant properties.1 Chemically, walnut shell contains several classes of compounds, including phenolic acids, saponins, flavonoids, esters, and polysaccharides.15 The chemical characteristics of walnut shell have also encouraged its potential incorporation into value-added products, including antibacterial soaps, masks, and other household or personal-care products. The green husk has likewise been extensively investigated for its biological activities. Studies have demonstrated antimicrobial, antioxidant, anti-inflammatory, antidiabetic, and cytotoxic effects.9,16–18 Phytochemical analyses have identified a broad spectrum of phenolic constituents in the green husk, including benzoic acid derivatives, hydroxycinnamic acids, naphthoquinones, flavonoids, tannins, and diarylheptanoids.7,19,20

Molecular docking and machine-learning approaches have emerged as valuable tools in natural product research, enabling the prediction of compound–target interactions, identification of potential bioactive molecules, and prioritization of promising candidates. Their integration enhances the efficiency of bioactive compound screening, provides insights into potential mechanisms of action, and supports the identification of lead compounds for drug discovery.21,22 Here, network pharmacology, machine-learning approaches, and molecular docking further explored the biological relevance of the identified constituents. Collectively, these findings support the potential of walnut shell and green husk as underutilized sources of bioactive compounds and provide a rationale for their further investigation and valorisation rather than treating them solely as agricultural waste. Accordingly, this study aimed to provide a comprehensive phytochemical characterization of extracts from the green husk and shell and to evaluate their antioxidant, enzyme-inhibitory, and cytotoxic activities.

2. Materials and methods

2.1. Plant collection

Walnut samples were collected in July 2025 from Taşkent, Konya, Türkiye. The collected walnut was identified by Evren Yildiztugay from the Biotechnology Department of Selcuk University, Konya, Türkiye. Following fruit collection, the walnut green skin was immediately separated. Then, both the walnut green skin and shells were dried at 40 °C for 48 hours. Once fully dried, the plant material was ground into a fine powder. The powdered samples were stored in light-proof containers under constant conditions to keep them chemically stable and prevent spoilage.

2.2. Plant extract preparation

Five solvents covering a broad range of polarities were used for extraction: ethyl acetate, acetone, 70% acetone/water (70%), and water. For each solvent, 10 g of dried plant material was combined with 200 mL of solvent, though the extraction method varied by solvent type. For the four organic solvents, extraction was done by maceration—soaking the plant material at room temperature for 24 hours. The water extraction (10 g sample) instead used a 15 minute hot-water infusion for 15 min. Afterward, the organic extracts were concentrated by rotary evaporation under reduced pressure, while the aqueous extract was freeze-dried. All extracts were kept at 4 °C until analysis.23

2.3. Total phenolic and flavonoid contents

Total phenolic24 and total flavonoid25 contents were measured by established colorimetric methods. Gallic acid and rutin served as the reference standards for the phenolic and flavonoid calibration curves, respectively, and the results were expressed as mg gallic acid equivalents (GAE) and mg rutin equivalents (RE) per gram of extract.

2.4. HPLC/MS analysis

Using a high-performance liquid chromatography (HPLC) analysis in conjunction with an ESI/MS spectrometer, the phytochemical analysis of walnut extracts was evaluated in accordance with the previously published methodology.26 Using the molecular weights, this method enabled the tentative identification of phytoconstituents. After dissolving the extract (100 µg mL−1) in HPLC-grade methanol, it was filtered through a membrane disc (0.20 µm). After that, 10 µL of the filtrate was added to the HPLC-ESI-MS/MS. The specs of the utilized HPLC apparatus are as follows: Waters® is equipped with a reversed-phase C-18 column (ACQUITY UPLC-BEH C-18, dimensions = 2.1 × 50 mm, particle size ∼1.7 µm). The mobile phase was sonicated and passed through a membrane disc filter (0.2 µm) before injection. Using gradient elution (water and methanol acidified with 0.1% formic acid) with a flow rate of 0.2 mL min−1, the elution run took 35 minutes. Positive and negative ions were obtained using ESI-MS on an XEVO TQD triple quadrupole instrument. The mass spectrometer and HPLC unit were provided by Waters® Corporation, Milford, MA 01757, U.S.A. The vacuum pump was supplied by Edwards®, U.S.A., at desolvation temperatures of 150 and 440 °C. The software MassLynx 4.1 was used to obtain the mass spectra in an ESI range of 100–1000 m/z. The acquired mass spectra were tentatively identified by comparison with published data.

2.5. Antioxidant assays

The antioxidant methods were described in a previous study.27 Reducing power was determined by the FRAP and CUPRAC assays, and radical scavenging activity by the DPPH and ABTS assays. Results from these four assays were expressed as mg Trolox equivalents per gram of dry extract (mg TE per g). Total antioxidant capacity was measured by the phosphomolybdenum (PBD) method and reported as mmol TE per g extract. Metal chelating ability was also tested, and the results were given as mg EDTA equivalents per gram of extract (mg EDTAE per g).

2.6. Enzyme inhibitory activity assays

The enzyme inhibitory activity of the extracts was tested against acetylcholinesterase (AChE), butyrylcholinesterase (BChE), tyrosinase, and α-amylase using colorimetric methods described previously.27 Results were expressed relative to standard inhibitors: cholinesterase inhibition as mg galanthamine equivalents per gram of extract (mg GALAE per g), α-amylase inhibition as mg acarbose equivalents per gram (mg ACAE per g), and tyrosinase inhibition as mg kojic acid equivalents per gram (mg KAE per g). The extracts were additionally evaluated for their inhibitory effects on human carbonic anhydrase isoenzymes I and II (hCA I and hCA II).28

3. Cytotoxic effect

3.1. Cell culture

Three human cancer cell lines, HeLa (cervical adenocarcinoma), A549 (lung adenocarcinoma) and HCT-116 (colorectal carcinoma), and one non-cancerous line, HEK-293 (human embryonic kidney), were obtained from ATCC and kept in liquid nitrogen until use. Cells were grown in DMEM/F12 or RPMI-1640 supplemented with 10% fetal bovine serum (FBS), 100 IU mL−1 penicillin, and 100 µg mL−1 streptomycin and maintained at 37 °C in a humidified atmosphere with 5% CO2.

3.2. Cell viability assay

Cytotoxicity of the extracts was measured with the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay. HeLa, A549, HCT-116 and HEK-293 cells were seeded in sterile 96-well plates at 1 × 104 cells per well and allowed to attach for 24 h. The medium was then replaced with medium containing the extracts at 0, 2,5, 5, 10, 25, 50, 100 and 200 µg mL−1, and the cells were incubated for a further 24 h. Next, 10 µL of MTT solution (0.5 mg mL−1) was added to each well. After 4 h, the medium was removed and the formazan crystals were dissolved in 100 µL of DMSO. Absorbance was read at 570 nm, with 690 nm as the reference wavelength, on a Multiskan GO microplate reader (Thermo Fisher Scientific, USA). Viability was expressed as a percentage of the untreated control, and IC50 values were calculated by nonlinear regression using GraphPad Prism 9.2.29

3.3. Selectivity index of the extracts in cancer cells compared with normal cells

IC50 values obtained for each cancer cell line were compared with those obtained for the non-cancerous HEK-293 cells. The selectivity index (SI) was calculated as the IC50 in HEK-293 cells divided by the IC50 in the cancer cell line; higher values indicate greater selectivity for cancer cells. SI values were determined for all eight plant extracts in HeLa, A549, and HCT-116 cells.

3.4. Apoptotic effect of the ethyl acetate extracts on HCT-116 cells: AO/EB staining

Apoptosis was assessed morphologically in HCT-116 cells treated with the ethyl acetate extracts of the green husk and shell at 100 µg mL−1. After incubation, the cells were washed with PBS and fixed in 70% ethanol. Fixed cells were rinsed with distilled water, stained with an acridine orange/ethidium bromide working solution (A6014 and E1510; Sigma-Aldrich, Germany), and imaged under a fluorescence microscope.

3.5. Apoptotic effect of the ethyl acetate extracts on HCT-116 cells: annexin V assay

Apoptosis was also quantified with the FITC Annexin V Apoptosis Detection Kit I (BD Biosciences, New Jersey, USA), following the manufacturer's protocol. HCT-116 cells were seeded in 6-well plates at 5 × 105 cells per well and incubated for 24 h, then treated with the ethyl acetate extracts (100 µg mL−1) for another 24 h. Cells were harvested by trypsinization, washed, and resuspended in 1X binding buffer at 1 × 106 cells per mL. Aliquots of 100 µL were transferred to flow cytometry tubes, and 5 µL of FITC Annexin V and 5 µL of propidium iodide were added to each. The tubes were incubated for 15 min at room temperature in the dark, after which 400 µL of 1X binding buffer was added. Samples were analyzed on a BD FACSVia flow cytometer (BD Biosciences, New Jersey, USA).

3.6. Network pharmacology

Canonical SMILES representations of the investigated phytochemicals were retrieved from PubChem.30 Potential protein targets were subsequently inferred using a locally implemented target-prediction pipeline based on the ChEMBL multitask model. Before prediction, the chemical structures were standardized with RDKit and encoded as Morgan molecular fingerprints.31 These fingerprints were used as input to the ChEMBL multitask ONNX model to generate candidate ChEMBL target identifiers. The resulting identifiers were annotated against ChEMBL v36 to obtain the corresponding target names, official gene symbols, UniProt accessions, target classifications, and species information.32,33

For each compound, predictions achieving a model score of at least 0.50 were considered eligible, while no more than 50 targets were retained per compound. The dataset was subsequently restricted to Homo sapiens targets for which an official gene symbol could be assigned. These curated targets were used for subsequent intersection analyses with term-specific disease-associated gene sets.

Gene sets related to cell death, necroptosis, and apoptosis were independently retrieved from the GeneCards database (https://www.genecards.org/). Within each functional module, only protein-coding genes with a GeneCards relevance score of ≥15 were retained. The resulting disease-associated gene sets were compared with the predicted human target lists to identify common genes. Protein–protein interaction (PPI) relationships among the shared genes were investigated using STRING v12.0 (https://string-db.org/).34 A minimum interaction confidence score of 0.4 was applied to retain interactions while maintaining adequate network connectivity. The resulting networks were subsequently imported into Cytoscape v3.10.3 for network inspection and topological visualization.35

3.7. Compound–target and PPI network construction

For each biological term, compound–target relationships and corresponding PPI networks were constructed from the intersecting gene sets. Compound–target associations were first restricted to genes simultaneously identified among the human targets predicted by the ChEMBL-based workflow and the disease-related genes obtained from GeneCards. The resulting ligand–target relationships were represented in Cytoscape v3.10.3 as bipartite networks, with phytochemicals and target genes assigned to separate node categories and predicted interactions represented by connecting edges.

For the PPI analysis, the shared genes identified for each disease-related module were independently queried against STRING v12.0 with Homo sapiens specified as the organism. The interaction confidence cutoff was maintained at 0.4, and the analysis was performed without introducing additional proteins beyond the submitted gene set. The resulting interaction networks were exported to Cytoscape v3.10.3 and used to characterize the connectivity and network organization of the candidate targets associated with each biological term.32–35

3.8. Functional annotation of common genes

Functional enrichment analysis was conducted using ToppFun to characterize the biological functions and signaling pathways represented by the overlapping target genes. Gene Ontology (GO) analysis was restricted to the Biological Process (BP) category, and pathway-level enrichment was assessed against the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Enriched categories were considered statistically significant when the corresponding p-value was below 0.05.36 For graphical presentation, the enriched results were ranked according to the number of genes represented in each category. The five highest-ranking GO-BP terms and ten most represented KEGG pathways were subsequently selected and displayed using bar plots.37 All figures were generated in R v4.3.3.

3.9. External transcriptomic validation

External transcriptomic validation of the eight colorectal cancer-related common targets (FAP, CD274, NOS2, BCL2L1, PIK3R2, PIK3CG, TNIK, and PARP1) was performed using the GSE44076 dataset. Primary colorectal tumor samples and colon mucosa samples obtained from healthy donors were retained, whereas adjacent-normal and other non-relevant samples were excluded. The final dataset comprised 98 tumor and 50 control samples (n = 148) (Table S2). Expression data were processed in R using GEOquery/bioconductor and annotated using the corresponding GPL13667 platform information. Probe-level measurements mapping to the same gene were averaged to obtain gene-level expression values. The expression distribution was examined before analysis, and no additional log2 transformation was applied. All eight predefined targets were successfully mapped and were retained irrespective of their individual differential-expression results. Tumor-control differences were evaluated using the limma package, with Benjamini–Hochberg correction for multiple testing.38–40

3.10. Machine-learning analysis

The discriminative capacity of the predefined eight-gene signature was evaluated using ElasticNet logistic regression in R. The same eight genes were used as predictors without further feature selection. Samples were divided by stratified sampling into training (n = 119) and held-out test (n = 29) sets at an approximately 80 : 20 ratio. Model tuning was performed by repeated cross-validation using only the training data, and the test set was reserved for final evaluation. Performance was assessed using ROC-AUC, accuracy, balanced accuracy, sensitivity, specificity, and F1 score. Gene importance derived from the fitted ElasticNet model was used only to describe the relative contribution of individual genes and was not used to redefine the original network-derived target set (Table S3).41,42

3.11. Independent TCGA validation

The eight-gene signature was further evaluated in an independent RNA-sequencing cohort comprising TCGA colon adenocarcinoma (TCGA-COAD) and rectal adenocarcinoma (TCGA-READ) samples. The combined dataset contained 647 tumor and 51 normal samples (n = 698). Differential expression between tumor and normal tissues was assessed using DESeq2, followed by Benjamini–Hochberg correction. For each of the eight targets, log2 fold changes and adjusted P values were compared with the GSE44076 results to evaluate the concordance of tumor-associated expression direction between the GEO and TCGA cohorts.

3.12. Molecular docking

The predominant metabolites reported for Juglans regia L. green husk and shell were subjected to structure-based molecular docking against the selected protein targets. Metabolites were considered for docking only when unambiguous canonical SMILES strings could represent their chemical structures. Three-dimensional structures of the target proteins were retrieved from the Protein Data Bank (PDB).43 Information on the proteins used, the selected docking regions, and the corresponding search-space coordinates and dimensions is provided in Table S4. Ligand preparation was performed from the canonical SMILES representations using Gypsum-DL. The procedure was used to generate three-dimensional ligand geometries while considering protonation and tautomeric states relevant to pH 7.4. The generated structures were exported as SDF files and subsequently examined with RDKit. For compounds for which the MMFF94 force field was applicable, geometry optimization was performed for a maximum of 2000 iterations. The resulting valid conformer with the lowest calculated energy was selected for the subsequent docking calculations. Ligands were then prepared for AutoDock Vina using Meeko, which generated the required PDBQT representations and defined the relevant molecular properties, including atom types, hydrogen atoms, partial charges, rotatable bonds, and torsional degrees of freedom.

The potential binding cavities of the protein structures were identified independently using POCASA v1.1. The coordinates of the predicted cavities were subsequently used to establish the corresponding docking search spaces. Docking calculations were performed with AutoDock Vina v1.1.2 (ref. 44) using an exhaustiveness setting of 32. For each protein–ligand combination, the generated poses were evaluated on the basis of their predicted binding scores, and the highest-ranked geometrically acceptable pose was retained for further structural analysis. Protein–ligand contacts were subsequently examined using the Protein–Ligand Interaction Profiler (PLIP).45 This analysis was used to identify the principal non-covalent interactions formed within the predicted binding region, including hydrogen bonds, hydrophobic contacts, aromatic interactions, and salt bridges.

3.13. DFT calculations and setup

All calculations were carried out with the ORCA program 6.1.1.46 The geometry optimizations and frequency analyses were carried out at the B3LYP/6-31G(d,p) level of theory, using the Becke three-parameter Lee–Yang–Parr hybrid functional with the Pople-type 6-31G(d,p) basis set. Solvent effects have not been taken into account in all calculations. We converged to optimality based on the tight self-consistent field (TightSCF) criteria. At the optimized geometries, Harmonic frequency analyses were performed to ensure that all structures are minima (all positive frequencies) on the potential energy surface. Analysis included zero-point vibrational energy corrections (ZPE).47,48

3.14. Statistical analysis

Each experimental measurement was obtained from three independent replicates. Differences among the extract groups were evaluated by one-way analysis of variance (ANOVA) using GraphPad Prism 9.2. When the overall ANOVA indicated significant differences, Tukey's multiple-comparison post hoc test was applied to identify differences between individual groups. Statistical significance was defined as p < 0.05.

4. Results and discussion

4.1. Total phenolic (TPC) and flavonoids (TFC) contents

The TPC and TFC of different extracts from green husk and shell of J. regia were determined, and results are presented in Table 1. The TPC ranged between 16.73 and 67.43 mg GAE per g, with the hydroacetone extract from the shell exhibiting the highest content. In fact, the acetone, hydroacetone, and water extracts of the shell were 1.1, 3.3, and 3.6 times richer than their corresponding extracts from the green husk. The TFC ranged from 1.36 to 21.33 mg RE per g, with the acetone extract from the green husk showing the highest content. Unlike the pattern observed for TPC, the acetone and EtOAc extracts of green husk were 6.1 and 6.3 times higher than their respective shell extracts. Jadhav et al.14 reported TPC and TFC values of 121 and 90 µg mL−1, respectively, in the methanolic extract of J. regia shell flour at a concentration of 100 mg mL−1. In contrast, the methanolic extract of green husk from Moroccan J. regia exhibited higher phenolic and flavonoid levels than those obtained in the present study.18 Consistent with this observation, Kithi et al.7 reported substantial variation in the phytochemical composition of J. regia green husk, with phenolic content ranging from 3.72 to 122.26 mg GAE per g and flavonoid content from 0.34 to 66.07 mg RE per g. The observed variability may be related to cultivar, geographical location, sampling period within the growing season, and differences in analytical procedures.

Table 1. Extraction yields (%), total phenolic and flavonoid contents in green husk and shell extracts of Juglans regiaa.

Parts Extracts Extraction yields (%) TPC (mg GAE per g)b TFC (mg RE per g)c
Green husk EA 2.54 27.38 ± 0.63e 17.84 ± 0.21b
Acetone 6.58 37.09 ± 1.10d 21.33 ± 0.27a
Acetone/water 12.74 20.20 ± 0.03g 1.93 ± 0.10e
Water 10.25 16.73 ± 0.11h 1.36 ± 0.15f
Shell EA 4.56 23.74 ± 0.33f 2.83 ± 0.04d
Acetone 7.97 41.02 ± 0.61c 3.47 ± 0.06c
Acetone/water 14.18 67.43 ± 0.47a 3.52 ± 0.21c
Water 13.51 59.48 ± 0.29b 3.32 ± 0.16c
a

Values are presented as mean ± SD of three independent measurements (n = 3). Means within the same column followed by different superscript letters (a–d) differ significantly at p ≤ 0.05.

b

mg gallic acid equivalents per gram of extract.

c

mg rutin equivalents per gram of extract.

4.2. HPLC/MS phytochemical analysis

LC-MS/MS analysis in negative ionization mode identified fourteen compounds in the walnut green husk and shell extracts (Table 2). Each compound was identified from its deprotonated molecular ion [M–H]−, which appeared one mass unit below its molecular weight, together with its MS/MS fragments and comparison with published data. The compounds were phenolic acids, an organic acid, naphthoquinone-related glycosides (tetralone and hydrojuglone derivatives), flavonoid glycosides, and methoxylated flavones. The number of compounds detected depended on both the plant part and the extraction solvent. The ethyl acetate extract of the green husk and the acetone/water extract of the shells gave the richest profiles, with nine compounds each. The green husk acetone extract contained seven compounds. The green husk acetone/water, shell ethyl acetate, and shell acetone extracts contained five each. The water extracts were the poorest, with only two compounds in the green husk and three in the shells.

Table 2. Chemical composition of green husk and shell extracts of Juglans regia.

m/z [M–H]−
SN Identified compounds R t M.W. Green husk-EA Green husk-acetone Green husk-acetone/water Green husk-water Shells-EA Shells-acetone Shells-acetone/water Shells-water MS/MS fragment m/z M.F. Ref.
1 Caffeic acid 0.85 180 179.3844 — — — — — — — 133.1968 C9H8O4 118
2 Malic acid 1.15 134 133.1968 133.1940 133.1786 133.1811 133.1910 133.1891 133.1721 133.1739 65.7034 C4H6O5 118
119
3 Protocatechuic acid 1.90 154 — — — — — — 153.2321 — 109–93 C7H6O4 5
4 3-Caffeoylquinic acid (chlorogenic acid) 1.96 354 — — — — — — 353.4973 — 353–191 ([M–H]−-162) quinic acid-179 C16H18O9 120
121
5 p-Coumaric acid derivative 4.75 282 281.5666 281.5327 — — — — — — 166–133.5–117 5
6 Hydrojuglone glucoside 5.67 338 — 337.5338 — — 337.5839 337.5276 337.4871 — 271–175 C16H18O8 5
7 Dihydroxytetralone galloyl hexoside isomer 6.30 492 491.6119 491.6515 491.5944 491.6003 — — — — 341–301–88 C23H24O12 118
8 A pentosyl kaempferol derivative 6.89 418 — — — — 417.6843 417.7032 417.7085 417.6684 285 120
122
9 Quercetin 3-rhamnoside (Quercitrin) 6.90 448 — — — — — — 447.5749 — 301–255–243–179 C21H20O11 120
121
119
5
118
10 Dihydroxy tetralone hexoside 9.59 340 339.5529 339.6600 — — 339.6667 339.5854 — — 161 C16H20O8 122
11 Trihydroxy-dimethoxy flavone 10.1 330 329.7228 — 329.6598 — — 329.7026 329.6760 329.6688 314–227–195–140 C17H14O7 119
12 Juglanin B 10.54 328 327.6046 327.5992 327.5665 — 327.5940 — 327.5549 — 312–253–185 C20H24O4 5
119
13 Juglanin C 11.91 326 325.5809 325.5834 325.5384 — — — 325.5257 — 310–253 C20H22O4 123
14 Tetra-methoxy flavone-O-deoxy hexoside 12.21 488 487.8209 — — — — — — — 341–326–311–271 119

Malic acid was the only compound found in all eight extracts. Several compounds appeared in only one extract. Caffeic acid and tetra-methoxy flavone-O-deoxyhexoside were detected only in the green husk ethyl acetate extract. Protocatechuic acid, chlorogenic acid, and quercitrin occurred only in the shell acetone/water extract. Some compounds were specific to one plant part. Dihydroxytetralone galloyl hexoside isomer was present in all four green husk extracts but absent from the shells. The pentosyl kaempferol derivative was present in all four shell extracts but absent from the green husk. Hydrojuglone glucoside was found mainly in the shells, in the ethyl acetate, acetone, and acetone/water extracts, and also in the green husk acetone extract. Juglanin B and juglanin C were more common in the green husk. Each was present in three green husk extracts and in the shell acetone/water extract, and juglanin B was also present in the shell ethyl acetate extract. Dihydroxy tetralone hexoside was detected in the ethyl acetate and acetone extracts of both materials. Trihydroxy–dimethoxy flavone was detected in five extracts, and it was the only flavone found in the shell water extract.

The fragmentation patterns supported these assignments. Chlorogenic acid gave a fragment at m/z 191, which matches deprotonated quinic acid after the caffeoyl unit (162 Da) is lost. Quercitrin lost a rhamnose residue (146 Da) to give the quercetin aglycone at m/z 301, which then fragmented further to m/z 255, 243, and 179. The kaempferol derivative produced the kaempferol aglycone ion at m/z 285 after losing a pentose (132 Da). Protocatechuic acid fragmented to m/z 109 through decarboxylation (loss of 44 Da). The methoxylated compounds each lost a methyl radical (15 Da): trihydroxy–dimethoxy flavone went from m/z 329 to 314, juglanin B from 327 to 312, and juglanin C from 325 to 310. Tetra-methoxy flavone-O-deoxyhexoside first lost its deoxyhexose to give m/z 341, then lost methyl groups in turn to give m/z 326 and 311. The dihydroxytetralone galloyl hexoside isomer produced fragments at m/z 341, 301, and 88. Dihydroxy tetralone hexoside gave a main fragment at m/z 161. These results show that ethyl acetate recovered the less polar, methoxylated constituents of the green husk most efficiently. The aqueous acetone mixture extracted a wider range of phenolic acids and flavonoid glycosides from the shells. Water alone was a poor solvent for most of the identified compounds. Most of these compounds have also been identified in different parts of J. regia in the literature.5,6

4.3. Antioxidant activity

The antioxidant activity of different extracts from green husk and shell of J. regia was determined using six complementary assays. The DPPH and ABTS assays assess the free radical scavenging capacity of the extracts, whereas the FRAP and CUPRAC assays determine their reducing power. The phosphomolybdenum assay measures the ability of the extracts to reduce Mo(vi) to Mo(v), reflecting their total antioxidant capacity. In addition, metal chelating activity was assessed through the determination of iron-chelating ability. Results are presented in Table 3. Consistent with its high TPC, the hydroacetone extract of the shell demonstrated the highest antiradical activity (DPPH = 670.82 mg TE per g; ABTS = 1152.73 mg TE per g) and ion-reducing capacity (CUPRAC = 396.52 mg TE per g; FRAP = 471.78 mg TE per g). Notably, its values were 2.5-, 3.0-, 1.7-, and 1.8-fold higher in the DPPH, ABTS, CUPRAC, and FRAP assays, respectively than those of the second highest-performing extract, the acetone extract of shell. Also, the hydroacetone and aqueous extracts of the shell displayed the highest total antioxidant activity (2.46 and 2.34 mmol TE per g, p ≥ 0.05), followed by the acetone extract of the green husk (1.87 mmol TE per g). The best metal chelating power was recorded from the aqueous extract of the shell and hydroacetone extract of the green husk (16.20 and 16.02 mg EDTAE per g, respectively; p ≥ 0.05), followed by the acetone extract of both the shell and green husk (14.67 and 14.19 mg EDTAE per g, respectively; p ≥ 0.05). Overall, although the green husk extracts revealed considerable antioxidant activity in all assays, the shell extracts had stronger antioxidant activity.

Table 3. Antioxidant properties of green husk and shell extracts of Juglans regiaa.

Parts Extracts DPPH (mg TE per g)b ABTS (mg TE per g)b CUPRAC (mg TE per g)b FRAP (mg TE per g)b PBD (mmol TE per g)c MCA (mg EDTAE per g)d
Green husk EA 79.41 ± 1.15d 169.85 ± 1.63d 95.49 ± 4.56d 97.74 ± 3.49f 1.67 ± 0.10c 2.08 ± 0.54e
Acetone 98.57 ± 0.32c 215.93 ± 0.16c 149.12 ± 1.77c 157.27 ± 1.39d 1.87 ± 0.12b 14.19 ± 0.43bc
Acetone/water 101.25 ± 0.33c 210.62 ± 1.45c 90.41 ± 0.48d 110.37 ± 1.26e 0.88 ± 0.01e 16.02 ± 0.40a
Water 85.04 ± 0.66d 162.58 ± 1.30de 68.22 ± 0.36e 85.89 ± 1.01g 0.93 ± 0.03e 9.34 ± 0.20d
Shell EA 79.03 ± 0.86d 144.56 ± 1.29e 90.12 ± 4.42de 91.31 ± 4.14fg 1.45 ± 0.04d 9.85 ± 0.58d
Acetone 267.25 ± 1.59b 388.67 ± 14.30b 238.16 ± 4.78b 267.54 ± 8.91b 1.46 ± 0.09d 14.67 ± 0.14b
Acetone/water 670.82 ± 10.37a 1152.73 ± 14.43a 396.52 ± 20.48a 471.78 ± 3.89a 2.46 ± 0.02a 13.39 ± 0.10c
Water 98.82 ± 0.21c 214.95 ± 0.17c 229.68 ± 1.29b 247.42 ± 0.35c 2.34 ± 0.07a 16.20 ± 0.04a
a

Values are presented as mean ± SD of three independent measurements (n = 3). Means within the same column followed by different superscript letters differ significantly at p ≤ 0.05.

b

mg trolox equivalents per g of extract.

c

mmol trolox equivalents per g of extract.

d

mg EDTAE per g of extract.

Studies investigating the antioxidant properties of methanolic extracts from different parts of J. regia have demonstrated considerable variation in activity among plant tissues. Bourais et al.18 reported that the green husk exhibited substantially greater antioxidant activity than the kernel, as evidenced by its stronger DPPH radical-scavenging capacity (IC50 = 32.27 µg mL−1) and ferric-reducing activity (EC50 = 10.45 µg mL−1), compared with the kernel (DPPH IC50 = 84.22 µg mL−1; FRAP EC50 = 39.36 µg mL−1). These findings are consistent with previous reports describing significant DPPH and/or ABTS radical-scavenging activities in J. regia husk extracts.49–52 Similarly, Jadhav et al.14 reported an 84% DPPH radical-scavenging activity for a methanolic extract of J. regia shell flour at a concentration of 100 mg mL−1. The antioxidant potential of these by-products may be associated with their rich composition of phenolic and other bioactive constituents. LS-MS analyses have identified several compounds well known for their antioxidant properties, including gallic acid,53 protocatechuic acid,54 chlorogenic acid,55p-coumaric acid derivatives,56 and quercitrin.57

Nevertheless, the present study extends the existing knowledge by evaluating additional antioxidant-related properties that have received comparatively limited attention in J. regia by-products. In particular, it is the first study to investigate Cu2+-reducing capacity, metal-chelating activity, and total antioxidant capacity using the phosphomolybdenum assay. Furthermore, the influence of extraction solvent was systematically assessed. The results demonstrated that hydroacetone and acetone were the most effective solvents for recovering antioxidant compounds, indicating that solvent polarity plays an important role in determining the antioxidant yield and composition of extracts. Overall, these findings further support the potential of walnut green husk and shell as valuable sources of antioxidant compounds with possible applications in food, nutraceutical, and other value-added products.

4.4. Enzyme inhibitory activity

Recently, inhibiting key enzymes has been shown to alleviate the symptoms of certain global health problems. For instance, acetylcholinesterase is a target enzyme in the management of Alzheimer's disease (AD) and inhibiting it can increase the level of acetylcholine in the synaptic cleft, thereby enhancing memory function in AD patients.58 Similarly, inhibiting amylase and glucosidase can hinder the increase in glucose levels in diabetic patients after a carbohydrate-rich diet.59 Tyrosinase is also targeted to control hyperpigmentation problems; it is a key enzyme in melanin synthesis.60 Several compounds have been produced as enzyme inhibitors in the pharmaceutical industry, but most of them have unpleasant side effects.61–63 Therefore, safe, natural and effective inhibitors need to replace synthetic ones.

The enzyme inhibitory potential of extracts derived from the green husk and shell of J. regia was assessed against AChE, BChE, tyrosinase (Tyr), α-amylase, and α-glucosidase; the results are presented in Table 4. The enzyme inhibitory activity varied according to the tested part of the fruit, extract, and tested enzyme. The anti-AChE activity ranged between 1.90 and 2.77 mg GALAE per g with the hydroacetone and acetone extracts from the green husk, in addition to the hydroacetone extract of the shell, displaying the highest inhibitory effect. The anti-BChE activity ranged between 0.52 and 3.39 mg GALAE per g, with the acetone extract of the green husk revealing the highest activity, while the other organic extracts from the two tested parts demonstrated relatively similar effects (p ≥ 0.05). A previous study on aqueous extract from the leaf demonstrated significant inhibition against AChE (40.09%) and BuChE (56.23%) at 100 mg mL−1.64 The anti-Tyr activity of 16.13 and 65.80 mg KAE per g with the EtOAc extract of the shell revealed the highest inhibitory effect. Overall, the three organic extracts of the shell recorded higher anti-Tyr activity than the corresponding extracts obtained from the green husk (p < 0.05). The anti-Tyr activity of the green husk was previously determined on the aqueous extract IC50 10.134 mg mL−1.65

Table 4. Enzyme inhibitory effects of green husk and shell extracts of Juglans regiaae.

Parts Extracts AChE (mg GALAE per g)b BChE (mg GALAE per g)b Tyrosinase (mg KAE per g)c Amylase (mmol ACAE per g)d Glucosidase (mmol ACAE per g)d
Green husk EA 1.90 ± 0.08e 2.28 ± 0.45b 56.24 ± 1.06d 0.87 ± 0.01a 5.40 ± 0.23c
Acetone 2.66 ± 0.03ab 3.39 ± 0.16a 55.87 ± 1.08d 0.63 ± 0.01d 6.00 ± 0.24b
Acetone/water 2.77 ± 0.01a 2.09 ± 0.08bc 55.39 ± 0.19d 0.36 ± 0.01e na
Water 2.61 ± 0.02b 1.57 ± 0.10c 16.13 ± 0.56f 0.08 ± 0.01g 3.51 ± 0.01d
Shell EA 2.10 ± 0.08d 2.27 ± 0.24b 65.80 ± 0.74a 0.79 ± 0.01b 7.73 ± 0.11a
Acetone 2.44 ± 0.04c 2.32 ± 0.30b 61.21 ± 0.27c 0.70 ± 0.02c 6.04 ± 0.06b
Acetone/water 2.69 ± 0.01ab 2.24 ± 0.02b 63.40 ± 0.48b 0.62 ± 0.01d na
Water 1.95 ± 0.04e 0.52 ± 0.12d 24.27 ± 0.34e 0.13 ± 0.01f na
a

Values are presented as mean ± SD of three independent measurements (n = 3). Means within the same column followed by different superscript letters differ significantly at p ≤ 0.05.

b

mg galantamine equivalents per g of extract.

c

mg kojic acid equivalents per g of extract.

d

mmol acarbose equivalents per g of extract.

e

na – not active.

Regarding the enzymes involved in carbohydrate metabolism and diabetes management, the extracts from both fruit parts exhibited relatively weak inhibitory activity against α-amylase, with values ranging from 0.08 to 0.87 mmol ACAE per g. A different pattern was observed for α-glucosidase, for which several extracts demonstrated substantial inhibitory activity. The EtOAc extract of the shell exhibited the highest activity (7.37 mmol ACAE per g), followed by the acetone extracts of the shell (6.04 mmol ACAE per g) and green husk (6.00 mmol ACAE per g), which showed similar inhibitory potential. The EtOAc extract of the green husk (5.40 mmol ACAE per g) and its aqueous extract (3.51 mmol ACAE per g) also showed notable activity, whereas the remaining extracts did not exhibit detectable inhibition of α-glucosidase. The present findings differ from those reported by Bourais et al.,18 who found that methanolic extracts of J. regia kernel and green husk were more effective against α-amylase, with inhibitory values of 37.37 and 75.42 µg mL−1, respectively, than against α-glucosidase, for which values of 978.92 and 789.46 µg mL−1 were recorded. Similarly, Drioiche et al.5 reported that a decoction prepared from the green husk exhibited considerable inhibitory activity against both α-amylase and α-glucosidase, with EC50 values of 104.798 and 12.12 µg mL−1, respectively. The differences among these studies may be related to variations in plant material, extraction procedures, solvent polarity, and the concentration and composition of bioactive constituents recovered during extraction. The extracts were further evaluated for their inhibitory effects on the human carbonic anhydrase isoforms hCA I and hCA II (Table 5). Overall, extracts exerted moderate inhibitory activity against both isoforms, and the acetone extracts of both fruit parts revealed the best inhibitory effect against hCAI (IC50 24.75 and 25.20 µg mL−1 for green husk and shell, respectively). In contrast, the EtOAc extract of the green husk showed the best effect towards hCAII with an IC50 value of 87.64 µg mL−1.

Table 5. Carbonic anhydrase inhibitory effects of green husk and shell extracts of Juglans regia.

Parts Extracts CA I CA II
IC50 (µg mL−1) R 2 IC50 (µg mL−1) R 2
Green husk EA 154.00 0.9528 87.72 0.9626
Acetone 24.75 0.9958 93.64 0.9733
Acetone/water 67.28 0.9982 115.50 0.9805
Shell Water 73.72 0.9958 135.88 0.9990
EA 231.00 0.9747 216.56 0.9996
Acetone 25.20 0.9951 147.44 0.9830
Acetone/water 45.29 0.9827 113.61 0.9954
Water 157.50 0.9980 99.00 0.9932
Asetazolamid (standart inhibitör) (ng mL−1) 2.88 0.9870 1.01 0.9879

4.5. Cell-death, cytotoxicity and cell cycle analysis in Juglans regia green husk and shell

The cytotoxic activities of the green husk and shell extracts of Juglans regia, prepared using ethyl acetate, acetone, acetone/water, and water, were evaluated against different cancer cell lines. The cells were treated with the green husk and shell extracts at concentrations ranging from 0 to 200 µg mL−1 for 24 h, and the IC50 values are presented in Table 6. Among the tested samples, the ethyl acetate extracts of the green husk and shell exhibited the strongest cytotoxic activity. Their IC50 values against HCT-116 cells were 104.86 and 93.50 µg mL−1, respectively (Fig. S1 and S2). Notably, the shell extract showed greater cytotoxic activity than the green husk extract. Interestingly, even though a greater number of compounds were detected in the green husk extract, the shell extract evinced a lower IC50 value. This observation suggests that cytotoxic activity cannot be explained solely by the number of diverse bioactive compounds. Instead, the observed differences may be related to the relative abundance and intrinsic biological activity of each compound. However, the current phytochemical analysis alone is insufficient to confirm the presence of synergistic or antagonistic interactions among these compounds.

Table 6. Cytotoxic effects of green husk and shell extracts of Juglans regia (IC50 (µg mL−1)).

Parts Extracts HELA A549 HCT-116 HEK-293
Green husk EA 174.24 154.18 104.86 131.36
Acetone 258.21 202.21 121.14 133.47
Acetone/water 591.78 618.12 692.71 601.6
Water 672.78 701.36 639.31 651.92
Shell EA 109.43 104.31 93.50 94.3
Acetone 147.46 136.66 108.07 100.42
Acetone/water 721.96 801.18 764.63 695.16
Water 525.43 612.27 524.31 505.26

To determine whether the observed cytotoxic effects were selective toward cancer cells rather than non-cancerous cells, the selectivity index (SI) was calculated for all extracts (Table 7). SI values above 1.00 suggest that the tested extracts showed greater selectivity for cancer cells compared with normal cell.66,67 In this point, the highest SI values were observed for HCT-116 cells, indicating that the ethyl acetate extracts of J. regia may demonstrate comparatively greater selectivity to this cell line.

Table 7. Selective index evaluation of cancerous cells compared to HEK-293 normal cells of the tested extracts.

Parts Extracts HELA A549 HCT-116
Green husk EA 0.8 0.9 1.3
Acetone 0.5 0.7 1.1
Acetone/water 1.0 1.0 0.9
Water 1.0 0.9 1.0
Shell EA 0.9 0.9 1.0
Acetone 0.7 0.7 0.9
Acetone/water 1.0 0.9 0.9
Water 1.0 0.8 1.0

The literature also contains findings that are comparable with this study. For instance, one study reported cytotoxic effects of different walnut leaf extracts (hexane, chloroform, ethyl acetate, and methanol) on BHY, MCF7, and HT-29 cells.68 The results indicated that fraction of chloroform showed the best inhibitory activity against the MCF7 cell line. Another study observed the cytotoxic effects of walnut seed, green husk, and leaves of J. regia methanolic extracts against colon cancer cell lines and reported that the inhibition of cell growth ranged from 0–23%–13–78%.69 Similar findings were also noted in a study investigating the cytotoxic effects of J. regia methanolic extracts against three cancer cell lines (DLD-1, CaCO-2, MCF-7). The reported IC50 values were 21.11, 250.03, and 255.99 µg mL−1 respectively.70 Taken together, these findings suggest that J. regia extracts can exhibit considerable cytotoxic activity against several cancer cell lines. But their effectiveness may depend on the extraction method, solvent selectivity, extract concentration, phytochemical constituents, and the specific cell line tested.

Fluorescence microscopy combined with specific dyes is widely used to determine morphological features associated with different processes of cell death.71,72 Among these dyes, acridine orange binds to nucleic acids of cells, so viable cells emit green fluorescence, while ethidium bromide enters cells and stains their nuclei orange to red. The early apoptotic cells typically evince green fluorescence with visible fragmentation. In addition, late apoptotic cells, characterized by membrane blebbing or nuclear shrinkage, exhibited orange-red fluorescence as a result of compromised membrane integrity. Moreover, the necrotic cells also display orange-red or red fluorescence due to severe membrane damage. However, these necrotic cells can be distinguished from apoptotic cells based on their characteristic (including cellular and nuclear swelling) morphology.72–74 In this study, the AO/EB double staining marked morphological and fluorescence alterations in HCT-116 cells following treatment with J. regia green shell and husk extracts at 100 µg mL−1 (Fig. 1). Untreated control cells exhibited predominantly green fluorescence at high cell density, indicating that most cells remained viable. In contrast, treatment with both extracts resulted in a pronounced reduction in cell density and clear alterations in cellular morphology.

Fig. 1. AO/EB staining after green husk (A) ethyl acetate (100 µg mL−1) and shell (B) ethyl acetate (100 µg mL−1) applied to HCT-116 cell.

Fig. 1

To further confirm and quantitatively assess the live, apoptotic, and necrotic cells, Annexin V/PI (propidium iodide) double staining was performed. During the early stages of apoptosis, phosphatidylserine is translocated from the inner to the outer leaflet of the plasma membrane, and it acts as an “eat me” signal for phagocytic cells.75 Annexin V dye specifically binds to phosphatidylserine, which enables the detection of apoptotic cells including early apoptotic and live cells.76 In contrast, propidium iodide (PI) is impermeable to cells with intact plasma membranes but enters cells with compromised membrane integrity, thereby allowing the identification of late apoptotic and necrotic populations.77

Understanding the principles of double staining is important for interpreting changes in the cell membrane associated with different modes of cell death.77 For instance, if Annexin V is negative and propidium iodide is positive, the cells are generally considered necrotic, whereas cells positive for both Annexin V and PI are known as late apoptotic or secondary necrotic. Our findings were consistent with these staining patterns. Treatment with the ethyl acetate extract of the green husk markedly increased the PI-positive necrotic cell population from 1.2% to 43.7%. The proportion of late apoptotic cells also increased from 1.7% to 7.0%, accompanied by a substantial decrease in viable cells from 96.4% to 49.1% (Fig. 2). A similar pattern was observed with the ethyl acetate shell extract, although the changes were less pronounced. The late apoptotic population increased from 1.4% to 3.8%, while necrotic cells increased from 0.8% to 34.7% (Fig. 3). Accordingly, the proportion of viable cells decreased from 96.9% to 60.6%. When comparing apoptosis and necrosis, apoptosis is generally considered a more favorable mode of cell death in anticancer therapy because it is a tightly regulated process. In contrast, necrosis is typically associated with loss of membrane integrity and the release of cellular components, which can trigger inflammation. But the induction of non-apoptotic cell death mechanisms can provide an alternative strategy for eliminating apoptosis-resistant tumor cells.78

Fig. 2. (A) Annexin-V/PI staining results after applying green husk ethyl acetate (100 µg mL−1) applied to HCT-116 cell. (B) Annexin-V/PI staining results after green husk ethyl acetate (100 µg mL−1) applied to HCT-116 cell. Blue: live-[(FITC−)/(PI−)]; Green: early apoptotic [(FITC+)/(PI−)]; Red: late apoptotic [(FITC+)/(PI+)]; Black: shows necrotic [(FITC+)/(PI+)] cells.

Fig. 2

Fig. 3. (A) Annexin-V/PI staining results after applying shell ethyl acetate 100 µg mL−1 to HCT-116 cell. (B) Annexin-V/PI staining results after shell ethyl acetate 100 µg mL−1 to HCT-116 cell. Blue: live-[(FITC−)/(PI−)]; Green: early apoptotic [(FITC+)/(PI−)]; Red: late apoptotic [(FITC+)/(PI+)]; Black: shows necrotic [(FITC+)/(PI+)] cells.

Fig. 3

In addition to these findings, flow cytometric cell cycle analysis was performed to further investigate the effects of the extract (ethyl acetate extract of green husk) on cell cycle progression in HCT-116 cells. The cell cycle consists of G1 (gap1), S (DNA synthesis), G2 (gap2), and M (the phase of mitosis) phases. As a result of our findings, cell cycle analysis revealed a marked accumulation of HCT-116 cells in the G0/G1 phase after treatment with the extract at 100 µg mL−1. The rate of cells in the G0/G1 phase increased from 39.0% to 69.7%, whereas the S phase rate decreased from 26.2% to 9.4% and the G2/M phase population waned from 31.9% to 19.6%. These findings indicate that the ethyl acetate extract predominantly induced cell cycle arrest at the G0/G1 phase and blocked progression from G1 to S phase. The increase in the G0/G1 population was accompanied by a marked decrease in the S phase population. These results suggest that the extract restricted the transition from G1 to S phase. Entry into the S phase is essential for DNA replication and cell proliferation.79 Therefore, inhibition of the G1/S transition may reduce the proliferation of HCT-116 cells. This effect may contribute to the antiproliferative activity of the extract on the HCT-116 cells. The literature also has comparable results with a recent study. A study conducted by Salimi et al. has reported that the J. regia leaves chloroform fraction is effective against human oral cancer, colon adenocarcinoma, and breast adenocarcinoma cell lines, and these extracts also induce cell cycle arrest in the G0/G1 phase after 24 hours.80 Another study has observed the methanolic extract of Juglans green fruit pericarp against BC cell lines. Results indicate significant cell cycle arrest in the G0/G1 phase in cancer cells post-treatment with the methanol extract.81

The anticancer activity of J. regia extracts has largely been linked to their effective chemical composition. In our study, chemical profiling of the extract revealed several potentially bioactive molecules, including juglanin B and C, caffeic acid, quercetin derivatives, and hydrojuglone glycoside. Among these compounds, juglanin B has previously been associated with activity against colorectal cancer cells. Sun et al. isolated juglanin B from green walnut husks and reported a marked inhibitory effect on colorectal cancer (IC50: 25.4).82 These findings suggest that juglanin B may be one of the compounds contributing to the cytotoxic activity observed in our study.

Another important compound detected in our extract was hydrojuglane glycoside. In plants, phenolic compounds are often stored in glycosylated forms, which generally improves their solubility and stability and facilitates their transport within plant tissues. In several Juglans species, hydrojuglone glucoside is found as a glycosylated precursor of juglone. This storage mechanism enables the accumulation of juglone-related metabolites and limits their potential autotoxic effects on plant tissues.83 Therefore, the presence of hydrojuglone-related compounds in our extract can be relevant to its observed cytotoxic effects.

Juglone has been investigated for its anticancer properties in different cancer cell lines, including HCT-15 (human colon carcinoma), HL-60 (human leukemia), SCG-7901 (human gastric cancer), and SKOV3 (ovarian cancer) cells. Relevant to our results, a study has reported that juglone can interfere with S phase in HCT-15 cells. This observation is consistent with the changes detected in our cell cycle analysis and may indicate a similar mechanism underlying the response to the ethyl acetate extract. Previous studies have also reported the cytotoxic effects of juglone, with apoptosis induction and increased intracellular reactive oxygen species levels.84

Considering both our chemical profiling and the available literature, the anticancer effect of the ethyl acetate extract is unlikely to derive from a single compound. Instead, compounds such as caffeic acid, juglanin B, hydrojuglone glycoside, and malic acid may contribute to the observed cytotoxic and cell-cycle effects. However, further studies using isolated compounds are needed to determine their individual contributions and to establish whether interactions among these constituents play a role in the overall anticancer activity of the extract.

4.6. Network pharmacology analysis

4.6.1. Compound–target network

Target prediction for the eight selected phytochemicals yielded 298 compound–target associations involving 90 unique human single-protein targets at a prediction score cut-off of 0.50. The cytoscape network comprised 98 nodes and 298 edges, with an average number of neighbors of 6.08, a characteristic path length of 2.60, and a network density of 0.063 (Fig. 4C). Malic acid had the largest number of predicted targets (42), followed by juglanin C (41) and caffeic acid (40). In contrast, hydrojuglone glucoside was associated with 33 targets. Several proteins, including HTR3A, CYP11B2, PER2, PDE2A, SCN9A, SLC16A1, TRPV1, and MME, were shared by all eight compounds. The marked overlap among the predicted targets indicates that the compounds may influence related molecular processes while retaining distinct target profiles.

Fig. 4. Colorectal cancer-focused network pharmacology analysis. (A) Colorectal cancer PPI network. (B) Disease enrichment analysis. (C) Compound–target network. (D) Target intersection. (E) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. (F) Gene Ontology (GO) enrichment analysis.

Fig. 4

4.6.2. Disease enrichment analysis

Disease enrichment analysis of the predicted targets showed significant associations with several disease categories, including a number of colorectal cancer-related terms (Fig. 4B). Rectal carcinoma was associated with 10 predicted genes, while hereditary nonpolyposis colorectal cancer involved six genes. Significant enrichment was also observed for metastatic colorectal carcinoma, secondary malignant neoplasm of the colon and/or rectum, adenocarcinoma of the colon, and different stages of colon cancer (Table S1). Although colorectal cancer was not the only disease group represented in the enrichment profile, the recurrence of several CRC-related terms supported further analysis of the predicted targets in a colorectal cancer-specific context.

4.6.3. Colorectal cancer-related target intersection

The colorectal cancer-associated gene set retrieved from GeneCards was further examined using STRING to evaluate the functional connectivity of the disease-related proteins. Of the genes submitted to STRING, 487 were mapped to the interaction network, forming 17 413 protein–protein interactions (Fig. 4A). The dense connectivity of this network reflects the broad molecular complexity of colorectal cancer and provides the disease-specific background for comparison with the phytochemical target set. The intersection of the predicted compound targets with the GeneCards-derived colorectal cancer genes identified eight common targets: FAP, CD274, NOS2, BCL2L1, PIK3R2, PIK3CG, TNIK, and PARP1 (Fig. 4D). CD274 showed the widest compound coverage and was predicted for seven of the eight phytochemicals, whereas FAP was associated with four compounds. Caffeic acid covered the largest number of colorectal cancer-related targets. High prediction scores were also observed for quercetin 3-rhamnoside-FAP, quercetin 3-rhamnoside-CD274, 3-caffeoylquinic acid-FAP, and 3-caffeoylquinic acid-PIK3CG (Fig. 4C). The shared targets represent different but interconnected aspects of colorectal cancer biology. CD274, encoding PD-L1, is involved in immune-checkpoint regulation in the CRC microenvironment, while PI3Kγ/PIK3CG has been linked to tumor-associated macrophage-mediated immune regulation.85,86FAP is predominantly associated with activated fibroblasts and the stromal compartment of colorectal tumors.87TNIK is an essential component of Wnt/β-catenin-dependent transcription, whereas PIK3R2 encodes the p85β regulatory subunit of PI3K and has been associated with PI3K pathway activation and tumor progression.88,89BCL2L1 encodes the anti-apoptotic protein BCL-XL and contributes to survival of subsets of colorectal cancer cells, while PARP1 participates in DNA-damage recognition and repair.90,91NOS2 further links the network to nitric oxide-dependent inflammatory and redox signaling in gastrointestinal carcinogenesis.92 Together, these associations support a multi-target profile involving immune regulation, stromal signaling, cell survival, DNA-damage responses, and inflammatory pathways. As these interactions were computationally predicted, structural and experimental validation is required to establish their biological relevance.

4.7. Functional enrichment analysis

4.7.1. KEGG pathway enrichment

KEGG analysis identified 39 enriched pathways, of which the 15 highest-ranked terms are shown in Fig. 4E. The enrichment profile was dominated by cancer- and survival-related pathways, including small-cell lung cancer, pancreatic cancer, chronic myeloid leukemia, apoptosis, pathways in cancer, and JAK-STAT signaling. The mTOR signaling and colorectal cancer pathways were also among the prominent terms. The colorectal cancer pathway involved PIK3R2 and PIK3CG, whereas the broader pathways in cancer additionally included BCL2L1 and NOS2. The recurrent involvement of PIK3R2 and PIK3CG across these pathways is consistent with the established role of PI3K/AKT/mTOR signaling in colorectal cancer progression and cell survival93–95

4.7.2. Gene ontology enrichment

GO analysis identified 487 enriched terms, comprising 389 Biological Process (BP), 91 Molecular Function (MF), and 7 Cellular Component (CC) terms; the five highest-ranked terms from each category are presented in Fig. 4F. BP enrichment was mainly associated with regulation of programmed cell death and apoptosis, response to biotic stimulus, and regulation of protein localization, involving CD274, BCL2L1, NOS2, PIK3CG, PARP1, and FAP. CC terms were concentrated in class I phosphatidylinositol 3-kinase complexes, particularly the class IA and IB complexes, largely through PIK3R2 and PIK3CG. MF enrichment was similarly centered on phosphatidylinositol-3-kinase activity, phosphatidylinositol kinase activity, lipid kinase activity, and protein kinase binding. Overall, the GO findings further support a functional convergence on apoptosis regulation and PI3K-dependent signalling, in agreement with the KEGG analysis.93–95

4.7.3. External transcriptomic validation

The eight CRC-related common targets were examined in GSE44076, which included 98 primary colorectal tumors and 50 healthy colon mucosa samples. All eight genes were represented in the expression matrix (Table 8). Seven showed significant tumor-control differences after Benjamini–Hochberg correction. FAP displayed the strongest increase in tumor tissue (log2FC = 2.60), followed by PARP1 (1.04), BCL2L1 (0.64), PIK3R2 (0.39), and NOS2 (0.30). In contrast, PIK3CG (−0.40) and TNIK (−0.90) were significantly decreased, whereas CD274 remained essentially unchanged (FDR = 0.431) (Fig. 5A & B). The composition of GSE44076 and the GPL13667 Affymetrix Human Genome U219 Array platform are consistent with the original GEO annotation. The marked increase in FAP is noteworthy because FAP is predominantly associated with activated fibroblasts in the tumor stroma and has been linked to colorectal cancer progression and adverse clinicopathological features. The increase in BCL2L1 is also compatible with its established anti-apoptotic role; BCL-XL dependence has been demonstrated in a subset of colorectal tumors. Similarly, the expression pattern of PIK3R2 is consistent with the strong representation of PI3K-related terms in the GO and KEGG analyses. PI3K/AKT/mTOR signaling is frequently altered in CRC and contributes to proliferation, survival, metastasis, and therapeutic resistance.

Table 8. Cross-cohort transcriptomic validation of the eight shared CRC targets.
Gene GSE44076 log2FC GEO FDR TCGA log2FC TCGA FDR Concordant Significant in both
FAP 2.60 3.99 × 10−21 3.05 1.10 × 10−40 Yes Yes
CD274 −0.02 0.431 0.19 0.396 No No
NOS2 0.30 0.030 0.09 0.763 Yes No
BCL2L1 0.64 9.12 × 10−14 1.00 7.36 × 10−24 Yes Yes
PIK3R2 0.39 1.46 × 10−7 1.25 2.11 × 10−10 Yes Yes
PIK3CG −0.40 1.77 × 10−14 −1.74 1.06 × 10−25 Yes Yes
TNIK −0.90 3.33 × 10−11 −0.55 2.95 × 10−4 Yes Yes
PARP1 1.04 2.06 × 10−37 0.28 4.34 × 10−5 Yes Yes
Fig. 5. Transcriptomic and machine-learning validation of the eight-gene CRC signature. (A–C) GEO and TCGA transcriptomic validation. (D) ElasticNet classification performance. (E) ElasticNet feature importance.

Fig. 5

4.7.4. Independent validation in TCGA-COAD/READ

The expression pattern was subsequently examined in an independent TCGA-COAD/READ cohort comprising 647 tumor and 51 normal samples. Six genes retained both statistical significance and the same direction of change observed in GSE44076. FAP, BCL2L1, PIK3R2, and PARP1 were significantly upregulated in both cohorts, whereas PIK3CG and TNIK were consistently downregulated. NOS2 showed the same positive direction in both datasets but did not reach significance in TCGA. CD274 was not significantly altered in either cohort (Fig. 5C). Overall, seven of the eight targets showed concordant expression direction between GEO and TCGA, while six remained significant in both datasets. This cross-cohort agreement strengthens the transcriptomic support for the network-derived target set and reduces the likelihood that the observed pattern is specific to a single microarray cohort. The absence of a reproducible bulk-expression change for CD274 does not exclude its biological relevance. PD-L1 expression in CRC is spatially heterogeneous and varies with immune-cell infiltration, microsatellite status, and tumor site, with discordance also reported between primary and metastatic lesions. The reduction in TNIK expression should likewise be interpreted cautiously. TNIK is a functional component of the TCF4/β-catenin transcriptional machinery and has been shown experimentally to support Wnt-dependent transcription in colorectal cancer. Thus, differential mRNA abundance and functional pathway relevance need not necessarily change in the same direction.96,97

4.7.5. Machine-learning validation

The predefined eight-gene panel was further evaluated by ElasticNet logistic regression without re-selecting genes according to their GEO significance. The model was trained on 119 samples and evaluated on a held-out test set of 29 samples. The classifier achieved an ROC-AUC of 1.00, with accuracy, balanced accuracy, sensitivity, specificity, and F1 score all equal to 1.00 (Fig. 5D). Feature-importance analysis ranked PARP1 highest, followed by FAP, PIK3CG, and BCL2L1 (Fig. 5E). Importantly, these values were used only to interpret the fitted model; the original eight-gene panel was retained throughout the analysis. This distinction is important because the machine-learning step was intended to test the discriminatory information contained in the network-derived signature rather than to generate a new set of targets. The complete separation observed in the held-out samples indicates that the eight-gene expression profile captures a strong tumor-control signal in GSE44076. However, the test set contained only 29 samples. The perfect performance should therefore be interpreted as a dataset-level validation result rather than evidence of established diagnostic accuracy. Independent prospective cohorts would be required before any clinical classification claim could be made (Table S3).

Taken together, the transcriptomic analyses narrowed the biological support for the original network-derived targets without redefining the target set. Six genes: FAP, BCL2L1, PIK3R2, PIK3CG, TNIK, and PARP1, showed reproducible and significant changes in both GEO and TCGA, whereas NOS2 showed partial cross-cohort support and CD274 lacked a consistent bulk-expression signal. The resulting pattern is compatible with several processes already highlighted by the network and enrichment analyses, including stromal remodeling, apoptosis, PI3K-related signaling, Wnt regulation, and DNA-damage responses. PARP1 is particularly relevant to the latter, given its established role in DNA repair and the continuing interest in DNA-damage-response targeting in CRC.98,99 Rather than pointing to a single dominant mechanism, these findings support a multi-target model in which the selected phytochemicals may act across several interconnected components of colorectal cancer biology. This interpretation is also more consistent with recent integrative network-pharmacology studies in CRC, which typically combine target prediction, functional enrichment, transcriptomic validation, machine learning, and structure-based analyses instead of relying on a single computational layer.100,101

4.7.6. Molecular docking analysis

Molecular docking was performed to investigate the potential interactions of the identified phenolic compounds with AChE, amylase, BChE, glucosidase, hCA I, hCA II, and tyrosinase. Predicted binding energies varied substantially among both compounds and target enzymes, indicating that ligand recognition was strongly target-dependent (Table S5). Among the evaluated compounds, quercetin 3-rhamnoside exhibited the most favorable and consistently broad docking profile, with binding energies of −8.6, −9.2, −10.4, −8.2, −9.8, and −9.0 kcal mol−1 against AChE, amylase, BChE, glucosidase, hCA I, and hCA II, respectively (Fig. 6A). Its interaction with tyrosinase was comparatively weaker (−5.1 kcal mol−1), whereas 3-caffeoylquinic acid showed the most favorable predicted interaction with tyrosinase (−6.0 kcal mol−1). This target-dependent pattern suggests that the structural features governing recognition of quercetin 3-rhamnoside differ among the investigated enzyme binding sites.

Fig. 6. Molecular docking analysis of selected compounds against target enzymes. (A) Docking scores (kcal mol−1) across the enzyme panel. (B–D) Predicted binding modes and key intermolecular interactions of quercetin 3-rhamnoside with BChE, hCA I, and amylase, respectively.

Fig. 6

The strongest predicted interaction was observed for quercetin 3-rhamnoside-BChE, with a docking score of −10.4 kcal mol−1 (Fig. 6A & B). The three-dimensional representation of this complex (Fig. 6B) shows that the ligand is accommodated within the BChE binding cavity and stabilized by a combination of hydrogen bonding, hydrophobic contacts, and π-stacking interactions. In particular, the interaction with Phe329 provides an additional aromatic contribution to ligand stabilization. The interaction plot below the structural representation further illustrates the contribution of H-bonding, hydrophobic contacts, and π-stacking to the predicted complex. Such multipoint interactions are consistent with previous studies showing that flavonoids can interact with key residues within cholinesterase binding sites and exhibit inhibitory activity against AChE and BChE.102–104

Quercetin 3-rhamnoside also showed a favorable predicted interaction with AChE (−8.6 kcal mol−1) (Fig. 6A). Although no individual AChE complex is illustrated in the structural panels of Fig. 4, the docking matrix indicates that its predicted affinity was stronger than that of most other compounds tested. This finding is consistent with previous computational and experimental evidence demonstrating that quercetin can occupy functionally relevant regions of AChE and form hydrogen-bond and van der Waals interactions with residues within the enzyme gorge.103

The interaction with amylase was also highly favorable (−9.2 kcal mol−1) (Fig. 6A & D). The three-dimensional representation in Fig. 4D shows quercetin 3-rhamnoside positioned within the enzyme binding region, where it forms multiple hydrogen-bond and hydrophobic interactions with surrounding residues. The interaction diagram indicates H-bond contacts involving residues including Gln, Asp, and Ala, together with hydrophobic contacts and a salt-bridge interaction. The extensive polar interaction network may be facilitated by the numerous hydroxyl groups of the flavonoid glycoside. A previous computational study has similarly demonstrated interactions between quercetin/flavonoid derivatives and α-amylase, supporting the plausibility of the predicted binding mode observed here.105

A particularly notable finding was the strong predicted binding of quercetin 3-rhamnoside to hCA I (−9.8 kcal mol−1) (Fig. 6A & C). The structural representation in Fig. 6C shows the ligand positioned within the hCA I binding region and surrounded by residues capable of hydrogen bonding, hydrophobic interactions, and salt-bridge formation. The interaction plot indicates several hydrogen-bond contacts, together with hydrophobic and electrostatic interactions, suggesting that binding is supported by a combination of complementary non-covalent forces rather than a single dominant interaction. The strong docking score is also consistent with previous experimental studies demonstrating that flavonoids and phenolic compounds can inhibit hCA I and hCA II, although their inhibitory mechanisms and isoform selectivity can differ substantially.106

The hCA II interaction (−9.0 kcal mol−1) was similarly favorable according to the docking matrix (Fig. 6A), although the corresponding structural complex is not separately illustrated in the figure. Together, the hCA I and hCA II results suggest that quercetin 3-rhamnoside may possess structural complementarity toward both carbonic anhydrase isoforms. However, the docking results should not be interpreted as evidence of equivalent experimental inhibitory potency because docking scores are approximate scoring-function outputs and do not fully account for protein flexibility, solvent effects, entropy, or other contributions to ligand binding.107

For glucosidase, quercetin 3-rhamnoside displayed a favorable docking score of −8.2 kcal mol−1 (Fig. 6A). The magnitude of this score, together with the favorable profiles observed against amylase, supports the possibility that the compound can interact with carbohydrate-hydrolyzing enzymes. Previous studies have reported that quercetin and related natural products can interact with both α-amylase and α-glucosidase, although the magnitude and mechanism of inhibition depend on the specific ligand and enzyme system.108,109

In contrast, tyrosinase showed a distinct docking pattern (Fig. 6A). Whereas quercetin 3-rhamnoside produced a relatively weak score of −5.1 kcal mol−1, 3-caffeoylquinic acid exhibited the most favorable interaction (−6.0 kcal mol−1). This difference suggests that the relatively large, highly hydroxylated flavonoid glycoside does not necessarily achieve the same degree of complementarity within the tyrosinase binding environment as it does within the other enzyme targets.110,111 The result also demonstrates why docking profiles should be interpreted on a target-by-target basis rather than assuming that a compound with favorable docking across several proteins will necessarily display comparable affinity toward all targets.

4.8. DFT analysis

4.8.1. Frontier orbital energies and band gap analysis

Geometry optimization and frontier molecular orbital analysis of all eight polyphenolic compounds were carried out. The HOMO and LUMO energies are particularly important indicators of chemical reactivity and can be used to gain insight into the capacity to donate or accept electrons of each compound.112,113 The energy difference between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) is referred to as the HOMO–LUMO gap (ΔE_HOMO–LUMO) and is used to indicate the stability and resistance of a molecule to electron-transfer reactions.114

The computed frontier molecular orbital energies (HOMO and LUMO) along with the resulting band gaps (ΔEHOMO–LOMO) for all eight polyphenolic compounds are compiled in Table S6. The corresponding orbital energy level shifts are plotted in Fig. S3, highlighting the progressive electronic stabilization across the series.

The results reveal significant variations in the energy of the frontier orbitals among the eight compounds, which are intricately linked to their molecular structures and the extent of conjugation. Malic acid exhibits the highest HOMO–LUMO gap (7.07 eV), attributed to its saturated aliphatic structure devoid of aromatic character and the absence of extensive π-electron delocalization. Conversely, hydrojuglone glucoside presents the smallest gap (3.74 eV), characterized by enhanced electronic delocalization and elevated reactivity. The intermediate gaps (4.17–5.23 eV) are observed in polyphenolic compounds possessing extended aromatic systems, renowned for their biological activity and antioxidant properties.

The HOMO–LUMO gap analysis yields a quantitative reactivity ranking. The compounds with smaller gaps (3.74–4.29 eV) are highly reactive, and those with larger gaps (5.03–7.08 eV) are more stable. This ordering is directly related to the chemical hardness (resistance to deformation) and softness (chemical reactivity) of the materials. The systematic trend found in this analysis allows rational selection and prioritization of polyphenolic compounds with respect to their potential antioxidant and redox properties.115 Fig. S4 presents a comparative analysis of reactivity rankings, utilizing frontier orbital energy gaps to differentiate between reactive and stable species. Fig. 7 elucidates the spatial distributions and phase characteristics of the corresponding frontier orbitals, thereby demonstrating electron wavefunction delocalization across the phenolic frameworks.

Fig. 7. HOMO–LUMO diagrams of the compounds (isovalue = ±0.02 a.u.) calculated at the B3LYP/6-31G(d,p) level. Green and red lobes represent the positive and negative phases of the wavefunction, respectively.

Fig. 7

4.8.2. Global reactivity descriptors

Global reactivity descriptors derived from frontier orbital energies provide quantitative measures of chemical properties including electronegativity (χ), chemical hardness (η), and electrophilicity (ω). These parameters enable prediction of molecular reactivity patterns and relative susceptibility to oxidation.116 The evaluated conceptual DFT descriptors, including electronegativity (χ), chemical hardness (η), and electrophilicity index (ω), together with the resulting qualitative reactivity classifications, are summarized in Table S7.

The global reactivity descriptors have a similar trend to frontier orbital analysis. The highest electronegativity (3.83 eV) and hardness (3.54 eV) of malic acid also prove its excellent chemical stability. The lowest hardness is 1.87 eV for hydrojuglone glucoside which implies that it exhibits the highest chemical softness and inherent reactivity. These descriptors correspond to the HSAB (Hard-Soft Acid-Base) theory prediction and can be used to validate the correlation of these polyphenolic compounds with their known biological activity.

4.8.3. Dipole moment components and analysis

Molecular dipole moment is one of the basic properties that describe the distribution of charge in a molecule, which is directly affected by its geometry and electronic structure. Using ORCA 6.1.1 at the B3LYP/6-31G(d,p) level of theory, the dipole moment (DM) components (µx, µy, µz) in atomic units (a.u.) and the total dipole moment (DM) magnitude in Debye units (D) were computed for all eight polyphenolic compounds.117 The calculated Cartesian dipole moment components (µx, µy, µz) in atomic units and total dipole moment magnitudes (µtotal) in Debye for all eight compounds are listed in Table S8.

The analysis of dipole moments shows that there is great variation in charge distribution among the eight compounds. Hydrojuglone glucoside has the highest total dipole moment (6.24 D) which means it has a large degree of charge separation and high polarity. Malic acid, on the other hand, has the lowest dipole moment (2.46 D), due to its simpler and more symmetric structure. The dipole moments of polyphenolic compounds are intermediate (2.51–6.24 D) and are proportional to the molecule's complexity and hydroxyl group distribution. The three-dimensional components (µx, µy, µz) show that there is not a uniform distribution of charge along one of the molecular axes, which suggests that the molecular geometry is asymmetric. The dipole moment of a molecule is generally higher if the molecule is more polar, which allows it to interact with polar biological environments and enzyme binding sites, thus contributing to enhanced biological activity.

4.8.4. Three-dimensional molecular electrostatic potential analysis

The molecular electrostatic potential (MEP) surface displays the distribution of electron density around molecules in three dimensions,117 with regions that are electron-rich (red) and electron-deficient (blue). MEP analysis is important for polyphenolic compounds to understand their antioxidant reactivity mechanisms by locating sites of preferential electron donation to free radicals and to predict intermolecular interaction patterns. The spatial distribution and magnitude of MEP extrema are directly linked to their bioavailability, radical scavenging efficiency, and potential therapeutic efficacy. The 3D molecular electrostatic potential (MEP) surfaces for all eight compounds are displayed in Fig. S5, showing electron-rich nucleophilic domains in red and electron-deficient electrophilic domains in blue. Detailed, high-resolution MEP maps for the three most nucleophilic candidates are further highlighted in Fig. S6 to identify preferential radical attack and electron donation sites.

The analysis of the compounds showed that the most nucleophilic compounds, with the greatest electron-rich regions, are glycosylated polyphenols (hydrojuglone glucoside, µ = 6.24 D; quercetin 3-rhamnoside, µ = 4.44 D) compared to the smaller phenolic acids (caffeic acid, µ = 2.51 D; malic acid, µ = 2.46 D). The direct correlation between the molecular size/complexity and the intensity of the MEP suggests that the larger polyphenolic scaffolds may possess more hydroxyl groups that confer more potent antioxidant reactivity sites and biological activity in terms of radical scavenging capability.

5. Conclusion

The present investigation examined the phytochemical composition as well as the antioxidants, enzyme-inhibitory, and cytotoxic activities of various extracts obtained from the green husk and shell of J. regia. Both fruit parts exhibited notable antioxidant and enzyme-inhibitory effects, although the magnitude of these activities differed depending on the extraction solvent and the assay employed. Overall, the shell showed stronger antioxidant activity, with the hydroacetone extract being particularly effective, while its inhibitory effects against tyrosinase and α-glucosidase were most pronounced in the EtOAc extract. In contrast, the green husk demonstrated greater inhibitory activity toward AChE and BChE, particularly in the acetone and hydroacetone extracts. Both fruit components displayed relatively weak inhibitory effects against the human carbonic anhydrase in forms hCA I and hCA II. However, the EtOAc extracts exhibited appreciable cytotoxicity against the HCT-116 cell line. LC-MS-based chemical profiling, confirmed the presence of a wide range of potentially bioactive compounds in both walnut fruit parts. Network pharmacology and machine-learning analyses provided additional insight into potential compound–target relationships, while molecular docking was used to examine the possible interactions of selected constituents with relevant protein targets. Taken together, these results indicate that walnut green husk and shell could serve as valuable natural sources of antioxidant and other biologically active compounds. Moreover, their status as largely underexploited by-products suggest considerable potential for the recovery and utilization of high-value bioactive molecules. Further studies focusing on the isolation and structural identification of the compounds responsible for the observed biological effects, as well as clarification of their underlying mechanisms of action, are warranted.

Author contributions

Abdullahi Ibrahim Uba: investigation, methodology, formal analysis, writing – original draft, writing – review and editing. Mehmet Veysi Cetiz: investigation, methodology, visualization, formal analysis, writing – original draft, writing – review and editing. Gunes Ak: investigation, methodology, visualization, formal analysis, writing – original draft, writing – review and editing. Enver Saka: investigation, methodology, formal analysis, writing – original draft, writing – review and editing. Awad Alzebair: investigation, methodology, writing – original draft, writing – review and editing. Omayma A. Eldahshan: investigation, methodology, writing – original draft, writing – review and editing. Abdel Nasser B. Singab: investigation, methodology, resources. Mohamed M. M. Abdelrazek: investigation, methodology, visualization, writing – original draft, writing – review and editing. Sara Saeed Kotb: investigation, methodology, visualization, writing – original draft, writing – review and editing. Ismail Koyuncu: investigation, methodology, writing – original draft, writing – review and editing. Ismail Yapıcı: investigation, methodology, visualization, formal analysis. Ilhami Gulcin: investigation, methodology, formal analysis, writing – original draft, writing – review and editing. Sakina Yagi: investigation, methodology, writing – original draft, writing – review and editing. Meng-Yao Li: investigation, methodology, writing – original draft, writing – review and editing. Gokhan Zengin: investigation, methodology, formal analysis, supervision, writing – original draft, writing – review and editing.

Conflicts of interest

There are no conflicts to declare.

Supplementary Material

RA-OLF-D6RA09569K-s001

Acknowledgments

Generative artificial intelligence (AI) tools were employed during manuscript preparation exclusively for language refinement. In addition, the authors used Google Gemini for designing Graphical Abstract. Their use did not extend to the generation, analysis, or interpretation of any experimental data. The authors retain full responsibility for all content presented in the final manuscript.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6ra09569k.

References

  1. Zhou S. Huang G. Chem. Biol. Technol. Agric. 2023;10:75. doi: 10.1186/s40538-023-00447-y. [DOI] [Google Scholar]
  2. Iordănescu O. A. Radulov I. Buhan I. P. Cocan I. Berbecea A. A. Popescu I. Poşta D. S. Camen D. Lalescu D. Agronomy. 2021;11:1092. doi: 10.3390/agronomy11061092. [DOI] [Google Scholar]
  3. Jahanban-Esfahlan A. Ostadrahimi A. Tabibiazar M. Amarowicz R. Int. J. Mol. Sci. 2019;20:3920. doi: 10.3390/ijms20163920. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Panth N. Paudel K. R. Karki R. J. Integr. Med. 2016;14:359–373. doi: 10.1016/S2095-4964(16)60274-1. [DOI] [PubMed] [Google Scholar]
  5. Drioiche A. Al Kamaly O. Baammi S. Zibouh K. Amaiach R. Bennani L. Abdelhakim E. O. L. El Imache A. El Makhoukhi F. Sahpaz S. Front. Chem. 2025;13:1638205. doi: 10.3389/fchem.2025.1638205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bi D. Zhao Y. Jiang R. Wang Y. Tian Y. Chen X. Bai S. She G. Nat. Prod. Commun. 2016;11:1934578X1601100643. doi: 10.1177/1934578X1601100643. [DOI] [PubMed] [Google Scholar]
  7. Kithi L. Lengyel-Kónya É. Berki M. Bujdosó G. Horticulturae. 2023;9:782. doi: 10.3390/horticulturae9070782. [DOI] [Google Scholar]
  8. Mourtzinos I. and Goula A., in Polyphenols in Plants, Elsevier, 2019, pp. 23–44 [Google Scholar]
  9. Fernández-Agulló A. Pereira E. Freire M. S. Valentão P. Andrade P. B. González-Álvarez J. Pereira J. Ind. Crops Prod. 2013;42:126–132. doi: 10.1016/j.indcrop.2012.05.021. [DOI] [Google Scholar]
  10. La Torre C. Caputo P. Plastina P. Cione E. Fazio A. Fermentation. 2021;7:305. doi: 10.3390/fermentation7040305. [DOI] [Google Scholar]
  11. Stampar F. Solar A. Hudina M. Veberic R. Colaric M. Food Chem. 2006;95:627–631. doi: 10.1016/j.foodchem.2005.01.035. [DOI] [PubMed] [Google Scholar]
  12. Srinivasan A. Viraraghavan T. Bioresour. Technol. 2008;99:8217–8220. doi: 10.1016/j.biortech.2008.03.072. [DOI] [PubMed] [Google Scholar]
  13. Chamorro F. Carpena M. Lourenço-Lopes C. Taofiq O. Otero P. Cao H. Xiao J. Simal-Gandara J. Prieto M. A. Biol. Life Sci. Forum. 2022;12(1):35. [Google Scholar]
  14. Jadhav R. Patil N. Saha B. Int. Res. J. Environ. Sci. 2024;13:1–8. [Google Scholar]
  15. Adamovic M. Adamovic A. Andjic M. Dimitrijevic J. Zdravkovic N. Kostic O. Pecarski D. Pecarski T. Obradovic D. Tomovic M. Cosmetics. 2024;11:163. doi: 10.3390/cosmetics11050163. [DOI] [Google Scholar]
  16. Aithal K. B. Kumar S. M. Rao N. B. Udupa N. Rao S. B. Cell Biol. Int. 2009;33:1039–1049. doi: 10.1016/j.cellbi.2009.06.018. [DOI] [PubMed] [Google Scholar]
  17. Beiki T. Najafpour G. D. Hosseini M. Color. Technol. 2018;134:71–81. doi: 10.1111/cote.12322. [DOI] [Google Scholar]
  18. Bourais I. Elmarrkechy S. Taha D. Badaoui B. Mourabit Y. Salhi N. Alshahrani M. M. Al Awadh A. A. Bouyahya A. Goh K. W. Molecules. 2022;27:8989. doi: 10.3390/molecules27248989. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Sheng F. Hu B. Jin Q. Wang J. Wu C. Luo Z. Molecules. 2021;26:3013. doi: 10.3390/molecules26103013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Cosmulescu S. N. Trandafir I. Achim G. Botu M. Baciu A. Gruia M. Not. Bot. Horti Agrobot. Cluj-Napoca. 2010;38:53–56. doi: 10.15835/nbha3814624. [DOI] [Google Scholar]
  21. Atanasov A. G. Zotchev S. B. Dirsch V. M. Orhan I. E. Banach M. Rollinger J. M. Barreca D. Weckwerth W. Bauer R. Bayer E. A. Majeed M. Bishayee A. Bochkov V. Bonn G. K. Braidy N. Bucar F. Cifuentes A. D’Onofrio G. Bodkin M. Diederich M. Dinkova-Kostova A. T. Efferth T. El Bairi K. Arkells N. Fan T.-P. Fiebich B. L. Freissmuth M. Georgiev M. I. Gibbons S. Godfrey K. M. Gruber C. W. Heer J. Huber L. A. Ibanez E. Kijjoa A. Kiss A. K. Lu A. Macias F. A. Miller M. J. S. Mocan A. Müller R. Nicoletti F. Perry G. Pittalà V. Rastrelli L. Ristow M. Russo G. L. Silva A. S. Schuster D. Sheridan H. Skalicka-Woźniak K. Skaltsounis L. Sobarzo-Sánchez E. Bredt D. S. Stuppner H. Sureda A. Tzvetkov N. T. Vacca R. A. Aggarwal B. B. Battino M. Giampieri F. Wink M. Wolfender J.-L. Xiao J. Yeung A. W. K. Lizard G. Popp M. A. Heinrich M. Berindan-Neagoe I. Stadler M. Daglia M. Verpoorte R. Supuran C. T. The International Natural Product Sciences. Nat. Rev. Drug Discovery. 2021;20:200–216. doi: 10.1038/s41573-020-00114-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Yang C. Chen E. A. Zhang Y. Molecules. 2022;27:4568. doi: 10.3390/molecules27144568. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Sut S. Dall'Acqua S. Flores G. A. Cusumano G. Koyuncu İ. Yuksekdag O. Emiliani C. Venanzoni R. Angelini P. Selvi S. Food Sci. Nutr. 2025;13:e70053. doi: 10.1002/fsn3.70053. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Slinkard K. Singleton V. L. Am. J. Enol. Vitic. 1977;28:49–55. doi: 10.5344/ajev.1977.28.1.49. [DOI] [Google Scholar]
  25. Zengin G. Aktumsek A. Afr. J. Tradit. Complement. Altern. Med. 2014;11:481. doi: 10.4314/ajtcam.v11i2.37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Nilofar N. Zengin G. Acar M. Bouyayha A. Youssra A. Eldahshan O. Fayez S. Fahmy N. Chem. Biodiversity. 2024;21:e202301651. doi: 10.1002/cbdv.202301651. [DOI] [PubMed] [Google Scholar]
  27. Grochowski D. M. Uysal S. Aktumsek A. Granica S. Zengin G. Ceylan R. Locatelli M. Tomczyk M. Phytochem. Lett. 2017;20:365–372. doi: 10.1016/j.phytol.2017.03.005. [DOI] [Google Scholar]
  28. Zengin G. Terzić M. Abul N. Gulcin I. Koyuncu I. Basarali M. K. Đorđević T. Cziáky Z. Jekő J. Cespedes- Acuna C. L. Food Biosci. 2024;60:104280. doi: 10.1016/j.fbio.2024.104280. [DOI] [Google Scholar]
  29. Koyuncu I. Temiz E. Seker F. Balos M. M. Akkafa F. Yuksekdag O. Yılmaz M. A. Zengin G. Chem. Biodiversity. 2024;21:e202301747. doi: 10.1002/cbdv.202301747. [DOI] [PubMed] [Google Scholar]
  30. Kim S. Chen J. Cheng T. Gindulyte A. He J. He S. Li Q. Shoemaker B. A. Thiessen P. A. Yu B. Nucleic Acids Res. 2023;51:D1373–D1380. doi: 10.1093/nar/gkac956. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Rogers D. Hahn M. J. Chem. Inf. Model. 2010;50:742–754. doi: 10.1021/ci100050t. [DOI] [PubMed] [Google Scholar]
  32. Mayr A. Klambauer G. Unterthiner T. Steijaert M. Wegner J. K. Ceulemans H. Clevert D.-A. Hochreiter S. Chem. Sci. 2018;9:5441–5451. doi: 10.1039/C8SC00148K. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Zdrazil B. Felix E. Hunter F. Manners E. J. Blackshaw J. Corbett S. De Veij M. Ioannidis H. Lopez D. M. Mosquera J. F. Nucleic Acids Res. 2024;52:D1180–D1192. doi: 10.1093/nar/gkad1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Szklarczyk D. Kirsch R. Koutrouli M. Nastou K. Mehryary F. Hachilif R. Gable A. L. Fang T. Doncheva N. T. Pyysalo S. Nucleic Acids Res. 2023;51:D638–D646. doi: 10.1093/nar/gkac1000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Shannon P. Markiel A. Ozier O. Baliga N. S. Wang J. T. Ramage D. Amin N. Schwikowski B. Ideker T. Genome Res. 2003;13:2498–2504. doi: 10.1101/gr.1239303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Balkrishna A. Devi K. Arya V. Dabas A. S. Afr. J. Bot. 2024;175:122–141. doi: 10.1016/j.sajb.2024.10.010. [DOI] [Google Scholar]
  37. Cetiz M. V. Yagi S. Kurt U. Koyuncu I. Yuksekdag O. Caprioli G. Acquaticci L. Angeloni S. Senkardes I. Zengin G. eFood. 2024;5:e70026. doi: 10.1002/efd2.70026. [DOI] [Google Scholar]
  38. Davis S. Meltzer P. S. Bioinformatics. 2007;23:1846–1847. doi: 10.1093/bioinformatics/btm254. [DOI] [PubMed] [Google Scholar]
  39. Benjamini Y. Hochberg Y. J. R. Stat. Soc., B: Stat. Methodol. 1995;57:289–300. doi: 10.1111/j.2517-6161.1995.tb02031.x. [DOI] [Google Scholar]
  40. Ritchie M. E. Phipson B. Wu D. Hu Y. Law C. W. Shi W. Smyth G. K. Nucleic Acids Res. 2015;43:e47. doi: 10.1093/nar/gkv007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Friedman J. H. Hastie T. Tibshirani R. J. Stat. Softw. 2010;33:1–22. [PMC free article] [PubMed] [Google Scholar]
  42. Kuhn M. J. Stat. Softw. 2008;28:1–26. [Google Scholar]
  43. Berman H. M. Nucleic Acids Res. 2000;28:235–242. doi: 10.1093/nar/28.1.235. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Trott O. Olson A. J. J. Comput. Chem. 2009;31:455–461. doi: 10.1002/jcc.21334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Salentin S. Schreiber S. Haupt V. J. Adasme M. F. Schroeder M. Nucleic Acids Res. 2015;43:W443–W447. doi: 10.1093/nar/gkv315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Neese F. Wiley Interdiscip. Rev.: Comput. Mol. Sci. 2025;15:e70019. [Google Scholar]
  47. Barone V. J. Chem. Phys. 2004;120:3059–3065. doi: 10.1063/1.1637580. [DOI] [PubMed] [Google Scholar]
  48. Khvostichenko D. Choi A. Boulatov R. J. Phys. Chem. A. 2008;112:3700–3711. doi: 10.1021/jp076979t. [DOI] [PubMed] [Google Scholar]
  49. Soto-Maldonado C. Vergara-Castro M. Jara-Quezada J. Caballero-Valdés E. Müller-Pavez A. Zúñiga-Hansen M. E. Altamirano C. Electron. J. Biotechnol. 2019;39:1–7. doi: 10.1016/j.ejbt.2019.02.001. [DOI] [Google Scholar]
  50. Rahmani F. Dehganiasl M. Heidari R. Rezaee R. Darvishzadeh R. Int. Food Res. J. 2018;25:35–42. [Google Scholar]
  51. Ghasemi K. Ghasemi Y. Ehteshamnia A. Nabavi S. M. Nabavi S. F. Ebrahimzadeh M. A. Pourmorad F. J. Med. Plant Res. 2011;5:1128–1133. [Google Scholar]
  52. Zhang Z. Liao L. Moore J. Wu T. Wang Z. Food Chem. 2009;113:160–165. doi: 10.1016/j.foodchem.2008.07.061. [DOI] [Google Scholar]
  53. Badhani B. Sharma N. Kakkar R. RSC Adv. 2015;5:27540–27557. doi: 10.1039/C5RA01911G. [DOI] [Google Scholar]
  54. Lai C. Xia H. Zhang Y. He Y. Wu X. Ye B. Yang H. Zhang B. Life. 2026;16:1206. doi: 10.3390/life16071206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Wang S. Li Y. Meng X. Chen S. Huang D. Xia Y. Zhu S. Food Chem. 2021;357:129904. doi: 10.1016/j.foodchem.2021.129904. [DOI] [PubMed] [Google Scholar]
  56. Halpani C. G. Mishra S. Pharm. Sci. Adv. 2024;2:100023. doi: 10.1016/j.pscia.2023.100023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Li X. Jiang Q. Wang T. Liu J. Chen D. Molecules. 2016;21:1246. doi: 10.3390/molecules21091246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Reynoso-García M. F. Nicolás-Álvarez D. E. Tenorio-Barajas A. Y. Reyes-Chaparro A. Int. J. Mol. Sci. 2025;26:3781. doi: 10.3390/ijms26083781. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Liu X. Nie Y. H. Wang B. B. Chen X. H. Zhang X. Y. Chen C. J. Food Sci. Technol. 2026;63:1064–1072. doi: 10.1007/s13197-025-06446-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Kang W. Tong H. H. Y. Li S. ACS Omega. 2025;10:38922–38932. doi: 10.1021/acsomega.5c04807. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Bold R. Valkanova V. BJPsych Adv. 2026;32:280–283. doi: 10.1192/bja.2026.10230. [DOI] [Google Scholar]
  62. Peng Z. Zhao L. He Y. Wang G. J. Agric. Food Res. 2026;26:102692. [Google Scholar]
  63. Zilles J. C. Dos Santos F. L. Kulkamp-Guerreiro I. C. Contri R. V. Exp. Dermatol. 2022;31:1500–1521. doi: 10.1111/exd.14662. [DOI] [PubMed] [Google Scholar]
  64. Karakaya S. Koca M. Yeşilyurt F. Hacımüftüoğlu A. Ankara Üniversitesi Eczacilik Fakültesi Dergisi. 2019;43:230–238. doi: 10.33483/jfpau.530099. [DOI] [Google Scholar]
  65. Akin M. Arabaci G. Saki N. Asian J. Chem. 2013;25(16):9337–9340. doi: 10.14233/ajchem.2013.15526. [DOI] [Google Scholar]
  66. Tronina T. Bartmańska A. Popłoński J. Rychlicka M. Sordon S. Filip-Psurska B. Milczarek M. Wietrzyk J. Huszcza E. Int. J. Mol. Sci. 2023;24:7408. doi: 10.3390/ijms24087408. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Klimek K. Tyśkiewicz K. Miazga-Karska M. Dębczak A. Rój E. Ginalska G. Molecules. 2021;26:2366. doi: 10.3390/molecules26082366. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Salimi M. Majd A. Sepahdar Z. Azadmanesh K. Irian S. Ardestaniyan M. H. Hedayati M. H. Rastkari N. Pharm. Biol. 2012;50:1416–1422. doi: 10.3109/13880209.2012.682118. [DOI] [PubMed] [Google Scholar]
  69. Carvalho M. Ferreira P. J. Mendes V. S. Silva R. Pereira J. A. Jerónimo C. Silva B. M. Food Chem. Toxicol. 2010;48:441–447. doi: 10.1016/j.fct.2009.10.043. [DOI] [PubMed] [Google Scholar]
  70. Żurek N. Pycia K. Pawłowska A. Potocki L. Kapusta I. T. Molecules. 2023;28:1989. doi: 10.3390/molecules28041989. [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Liu K. Liu P. C. Liu R. Wu X. Med. Sci. Monit. Basic Res. 2015;21:15–20. doi: 10.12659/MSMBR.893327. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Baeza-Morales A. Pascual-García S. Martínez-Peinado P. Navarro-Sempere A. Segovia Y. Medina-García M. Pujalte-Satorre C. García M. M. Martínez-Espinosa R. M. Sempere-Ortells J. M. Sci. Rep. 2025;15:23485. doi: 10.1038/s41598-025-06999-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Veetil V. T. Jayakrishnan V. Aravindan V. Rajeeve A. D. Koolath S. Yamuna R. Sci. Rep. 2024;14:27269. doi: 10.1038/s41598-024-79003-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Abdelhakim H. K. Hassan S. S. Ismail K. M. Sci. Rep. 2026;16:23653. doi: 10.1038/s41598-026-63591-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Segawa K. Nagata S. Trends Cell Biol. 2015;25:639–650. doi: 10.1016/j.tcb.2015.08.003. [DOI] [PubMed] [Google Scholar]
  76. Koopman G. Reutelingsperger C. P. M. Kuijten G. A. M. Keehnen R. M. J. Pals S. T. van Oers M. H. J. Blood. 1994;84:1415–1420. doi: 10.1182/blood.V84.5.1415.bloodjournal8451415. [DOI] [PubMed] [Google Scholar]
  77. Alshehade S. A. Almoustafa H. A. Alshawsh M. A. Chik Z. Heliyon. 2024;10:e33665. doi: 10.1016/j.heliyon.2024.e33665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Teoh W. Y. Sim K. S. Moses Richardson J. S. Abdul Wahab N. Hoe S. Z. Evid.-Based Complement. Altern. Med. 2013;2013:958407. doi: 10.1155/2013/958407. [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. Ligasová A. Frydrych I. Koberna K. Int. J. Mol. Sci. 2023;24(4):3674. doi: 10.3390/ijms24043674. [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Salimi M. Majd A. Sepahdar Z. Azadmanesh K. Irian S. Ardestaniyan M. H. Hedayati M. H. Rastkari N. Pharm. Biol. 2012;50:1416–1422. doi: 10.3109/13880209.2012.682118. [DOI] [PubMed] [Google Scholar]
  81. Mirzad E. Javanmardi M. Haidarizadeh M. Abdi M. Jentashapir J. Cell. Mol. Biol., 2024;15:e151858. [Google Scholar]
  82. Sun D. X. Guo X. F. A L. T. Ma X. L. Wei H. Y. Ma G. X. Shi L. L. Zhang J. Zhongguo Zhong Yao Za Zhi. 2019;44:2278–2282. doi: 10.19540/j.cnki.cjcmm.20190321.201. [DOI] [PubMed] [Google Scholar]
  83. Widhalm J. R. Rhodes D. Hortic. Res. 2016;3:16046. doi: 10.1038/hortres.2016.46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  84. Ahmad T. Suzuki Y. J. Antioxidants. 2019;8:91. doi: 10.3390/antiox8040091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  85. Chen X. Chen L.-J. Peng X.-F. Deng L. Wang Y. Li J.-J. Guo D.-L. Niu X.-H. Transl. Oncol. 2024;40:101851. doi: 10.1016/j.tranon.2023.101851. [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Lee Y. S. Song S. J. Hong H. K. Oh B. Y. Lee W. Y. Cho Y. B. Exp. Mol. Med. 2020;52:815–831. doi: 10.1038/s12276-020-0436-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  87. Kalaei Z. Manafi-Farid R. Rashidi B. Kiani F. K. Zarei A. Fathi M. Jadidi-Niaragh F. Cell Commun. Signal. 2023;21:139. doi: 10.1186/s12964-023-01151-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  88. Mahmoudi T. Li V. S. Ng S. S. Taouatas N. Vries R. G. Mohammed S. Heck A. J. Clevers H. EMBO J. 2009;28:3329. doi: 10.1038/emboj.2009.285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Cortés I. Sánchez-Ruíz J. Zuluaga S. Calvanese V. Marqués M. Hernández C. Rivera T. Kremer L. González-García A. Carrera A. C. Proc. Natl. Acad. Sci. U. S. A. 2012;109:11318–11323. doi: 10.1073/pnas.1118138109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  90. Zhang H. Xue J. Hessler P. Tahir S. K. Chen J. Jin S. Souers A. J. Leverson J. D. Lam L. T. Mol. Cancer. 2015;14:126. doi: 10.1186/s12943-015-0397-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  91. Catalano F. Borea R. Puglisi S. Boutros A. Gandini A. Cremante M. Martelli V. Sciallero S. Puccini A. Cancers. 2022;14:1388. doi: 10.3390/cancers14061388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  92. de Oliveira G. A. Cheng R. Y. Ridnour L. A. Basudhar D. Somasundaram V. McVicar D. W. Monteiro H. P. Wink D. A. Antioxid. Redox Signal. 2017;26:1059–1077. doi: 10.1089/ars.2016.6850. [DOI] [PMC free article] [PubMed] [Google Scholar]
  93. Leiphrakpam P. D. Are C. Int. J. Mol. Sci. 2024;25:3178. doi: 10.3390/ijms25063178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  94. Zhong J. Ding S. Zhang X. Di W. Wang X. Zhang H. Chen Y. Zhang Y. Hu Y. Front. Biosci. 2023;28:37. doi: 10.31083/j.fbl2802037. [DOI] [PubMed] [Google Scholar]
  95. Danielsen S. A. Eide P. W. Nesbakken A. Guren T. Leithe E. Lothe R. A. Biochim. Biophys. Acta - Rev. Cancer. 2015;1855:104–121. doi: 10.1016/j.bbcan.2014.09.008. [DOI] [PubMed] [Google Scholar]
  96. Wei X.-L. Wu Q.-N. Chen D.-l. Zeng Z.-L. Lu J.-B. Liu Z.-X. Ju H.-Q. Ren C. Pan Z.-Z. Wang F.-H. Xu R.-H. J. Cancer. 2018;9:4325–4333. doi: 10.7150/jca.27735. [DOI] [PMC free article] [PubMed] [Google Scholar]
  97. Mahmoudi T. Li V. S. W. Ng S. S. Taouatas N. Vries R. G. J. Mohammed S. Heck A. J. Clevers H. EMBO J. 2009;28:3329–3340. doi: 10.1038/emboj.2009.285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  98. Wu F.-H. Wei H.-Z. Deng H.-Y. Xiao G.-H. Zhang Y.-C. Neoplasma. 2023;70:1–14. doi: 10.4149/neo_2022_220724N745. [DOI] [PubMed] [Google Scholar]
  99. Catalano F. Borea R. Puglisi S. Boutros A. Gandini A. Cremante M. Martelli V. Sciallero S. Puccini A. Cancers. 2022;14:1388. doi: 10.3390/cancers14061388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  100. Raheem K. Li J. Zhang J. Ke J. Ren Z. Feng C. PLoS One. 2026;21:e0344478. doi: 10.1371/journal.pone.0344478. [DOI] [PMC free article] [PubMed] [Google Scholar]
  101. Zhao X. Xiu J. Yang H. Han W. Jin Y. Int. J. Mol. Sci. 2025;26:930. doi: 10.3390/ijms26030930. [DOI] [PMC free article] [PubMed] [Google Scholar]
  102. Khan M. T. H. Orhan I. Şenol F. S. Kartal M. Şener B. Dvorská M. Šmejkal K. Šlapetová T. Chem.-Biol. Interact. 2009;181:383–389. doi: 10.1016/j.cbi.2009.06.024. [DOI] [PubMed] [Google Scholar]
  103. Liao Y. Mai X. Wu X. Hu X. Luo X. Zhang G. Molecules. 2022;27:7971. doi: 10.3390/molecules27227971. [DOI] [PMC free article] [PubMed] [Google Scholar]
  104. Katalinić M. Rusak G. Domaćinović Barović J. Šinko G. Jelić D. Antolović R. Kovarik Z. Eur. J. Med. Chem. 2010;45:186–192. doi: 10.1016/j.ejmech.2009.09.041. [DOI] [PubMed] [Google Scholar]
  105. Bitew M. Desalegn T. Demissie T. B. Belayneh A. Endale M. Eswaramoorthy R. PLoS One. 2021;16:e0260853. doi: 10.1371/journal.pone.0260853. [DOI] [PMC free article] [PubMed] [Google Scholar]
  106. Koz Ö. Ekinci D. Perrone A. Piacente S. Alankuş-Çalışkan Ö. Bedir E. Supuran C. T. J. Enzyme Inhib. Med. Chem. 2012;28:412–417. doi: 10.3109/14756366.2011.651464. [DOI] [PubMed] [Google Scholar]
  107. Pantsar T. Poso A. Molecules. 2018;23:1899. doi: 10.3390/molecules23081899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  108. Guo F. An J. Wang M. Zhang W. Chen C. Mao X. Liu S. Wang P. Ren F. Foods. 2023;12:3415. doi: 10.3390/foods12183415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  109. Lam T.-P. Tran N.-V. N. Pham L.-H. D. Lai N. V.-T. Dang B.-T. N. Truong N.-L. N. Nguyen-Vo S.-K. Hoang T.-L. Mai T. T. Tran T.-D. Nat. Prod. Bioprospecting. 2024;14:4. doi: 10.1007/s13659-023-00424-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  110. El-Nashar H. A. S. El-Din M. I. G. Hritcu L. Eldahshan O. A. Molecules. 2021;26:7546. doi: 10.3390/molecules26247546. [DOI] [PMC free article] [PubMed] [Google Scholar]
  111. Terzi H. Yalçın H. Yıldız M. Zengin G. Pehlivan E. Uba A. I. S. Afr. J. Bot. 2025;180:811–819. doi: 10.1016/j.sajb.2025.03.044. [DOI] [Google Scholar]
  112. Azhagiri S. Jayakumar S. Gunasekaran S. Srinivasan S. Spectrochim. Acta, Part A. 2014;124:199–202. doi: 10.1016/j.saa.2013.12.106. [DOI] [PubMed] [Google Scholar]
  113. Choudhary V. Bhatt A. Dash D. Sharma N. J. Comput. Chem. 2019;40:2354–2363. doi: 10.1002/jcc.26012. [DOI] [PubMed] [Google Scholar]
  114. Amorim J. Altamirano L. De Souza D. S. Li C. Ejaz S. A. Arévalo J. C. Sci. Rep. 2025;16:4. doi: 10.1038/s41598-025-31230-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  115. Nacer H. Priyankha S. Hema G. Zerouali A. Haddag H. Mogren Al Mogren M. Prakash M. Hochlaf M. Adv. Theory Simul. 2026;9:e01415. doi: 10.1002/adts.202501415. [DOI] [Google Scholar]
  116. Bhatia M. Comput. Toxicol. 2024;29:100295. doi: 10.1016/j.comtox.2023.100295. [DOI] [Google Scholar]
  117. Liu X., Nanda K. D., Liang J. and Head-Gordon M., arXiv, preprint, arXiv:2608.24655, 2026
  118. HosseiniHashemi S. K. Baseri H. Mehjabin J. J. Alam M. K. HossinAshrafi S. K. Lotfi M. BioResources. 2025;20:7265–7277. [Google Scholar]
  119. Osztie R. Czeglédi T. Ross S. Stipsicz B. Kalydi E. Béni S. Boldizsár I. Riethmüller E. Bősze S. E. Alberti Á. Int. J. Mol. Sci. 2024;25:6930. doi: 10.3390/ijms25136930. [DOI] [PMC free article] [PubMed] [Google Scholar]
  120. Amaral J. S. Seabra R. M. Andrade P. B. Valentao P. Pereira J. A. Ferreres F. Food Chem. 2004;88:373–379. doi: 10.1016/j.foodchem.2004.01.055. [DOI] [Google Scholar]
  121. Liu P. Li L. Song L. Sun X. Yan S. Huang W. Food Chem. 2019;286:669–677. doi: 10.1016/j.foodchem.2019.02.054. [DOI] [PubMed] [Google Scholar]
  122. Medic A. Jakopic J. Hudina M. Solar A. Veberic R. J. King Saud Univ., Sci. 2022;34:101890. doi: 10.1016/j.jksus.2022.101890. [DOI] [Google Scholar]
  123. Liu JunXi L. J. Meng Min M. M. Li Chen L. C. Huang XinYi H. X. Di DuoLong D. D. J. Chromatogr. A. 2008;1190:80–85. doi: 10.1016/j.chroma.2008.02.083. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

RA-OLF-D6RA09569K-s001

Data Availability Statement

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6ra09569k.


Articles from RSC Advances are provided here courtesy of Royal Society of Chemistry

RESOURCES