Comparative study of the anti-inflammatory activity of hyperoside and its isomer in the treatment of diabetic retinopathy: From network pharmacology and bioinformatics analysis to experimental verification

INTRODUCTION

DR is one of the common complications of diabetes and the primary cause of blindness in working-age people worldwide (1, 2). Furthermore, a notable percentage of patients with DR, ranging from 3.4 % to 12.3 %, experience progression to severe vision-threatening conditions annually, resulting in substantial burdens on patients and society (3). The current therapeutic strategies for DR mainly focus on the treatment of advanced microvascular complications. Anti-vascular endothelial growth factor (VEGF) therapy is currently the primary treatment option for managing DR in clinical practice. However, anti-VEGF drugs have a relatively short half-life, requiring monthly or bi-monthly injections to maintain efficacy. Furthermore, approximately 15–20 % of DR patients do not respond or have a partial response to anti-VEGF treatment. In addition, several studies have indicated that prolonged use of anti-VEGF treatment can lead to thinning of the retina, choroid, and sclera, as well as retinal pigment epithelial tears, which may be caused by chronic suppression of VEGF function, leading to neuronal injury (4, 5). Recent studies have shown that DR is a neurovascular disorder driven by the dysfunction of the retinal neurovascular unit, including neuroinflammation, neuron loss, glial cell specific changes and vascular dysplasia (6, 7). Therefore, it is crucial to further investigate and develop new drugs and treatment approaches, with a particular focus on neuroprotection, as it may represent a promising strategy for the prevention and management of DR.

With advancements in natural pharmaceutical chemistry, numerous natural compounds or drugs derived from natural sources have been successfully applied in clinical settings. For example, paclitaxel has been utilized in the treatment of ovarian, breast and lung cancer (8). Hyperoside and isoquercitrin, both flavonol glycosides, are primarily found in plants such as Abelmoschus manihot (L.) Medik., Hypericum monogynum, Crataegus pinnatifida and Polygonum aviculare (9). Previous studies have demonstrated that hyperoside exhibits a wide range of pharmacological properties, including anti-inflammatory, antioxidant, vascular protection, and neuroprotective effects (10). Similarly, isoquercitrin, an isomer of hyperoside, has demonstrated significant pharmacological activities, such as anticancer, anti-inflammatory, anti-allergic, antioxidant, and neuroprotective effects (11, 12). In DR, hyperoside demonstrates a protective effect by mitigating oxidative stress and preventing high glucose-induced cell damage and apoptosis (13). However, the protective effect and mechanism of isoquercitrin on DR has not been elucidated.

This study aims to explore the similarities and differences in the protective effects and potential mechanisms of hyperoside and its isomer isoquercitrin on DR using network pharmacology and bioinformatics. We investigated the structural differences between isoquercitrin and hyperoside as isomers to identify similarities and differences in potential binding targets. The variations in the mechanisms of action in DR resulting from these target differences were further analysed using GO and KEGG. Simultaneously, to explore the expression patterns of these presumed drug targets in the context of DR, we conducted differential expression gene (DEG) analysis by integrating transcriptomes from macular and peripheral samples of non‑proliferative DR (NPDR) patients in GSE160306, identifying potential hub targets. We assessed the binding activities of isoquercitrin and hyperoside with these potential hub targets through molecular docking. Finally, further experiments were performed, providing evidence and guidance for hyperoside and isomers as the potential protective drug against DR (Fig. 1).

Fig. 1. Work flow chart of this study.

EXPERIMENTAL

Screening for isoquercitrin, hyperoside and DR targets

We used the pharmmapper (https://www.lilab-ecust.cn/pharmmapper/) to collect and predict targets of isoquercitrin and hyperoside. Our next step was to search for targets of disease using the search term “diabetic retinopathy” by the GeneCards Database (https://www.genecards.org/).

Network construction

Cytoscape 3.9.1 was utilized to import the active compounds and targets of the drugs to construct compound-target networks. In addition, we obtained protein interactions using STRING (https://cn.string-db.org/) to generate protein-protein interaction (PPI) networks using Cytoscape 3.9.1.

Differentially expressed genes screening

The "Limma" package in R was utilized to perform differential expression analysis on macular and periphery samples. Genes exhibiting a p-value < 0.05 and an absolute log2 fold change (log₂FC) > 1 were identified as differentially expressed genes (DEGs). Visualization of these DEGs was achieved through a heatmap generated using the "pheatmap" package and a volcano plot created with the "ggplot2" package.

Functional enrichment analysis

Biological functions of DEGs were analysed by GO and KEGG. GO and KEGG enrichment analyses were visualized with Sangerbox and were considered statistically significant at p < 0.05 (14). Enriched terms were ranked by p value in ascending order, and the top 10 terms were selected for visualization.

Molecular docking

The chemical structures of the compounds were initially drawn using ChemDraw. Subsequently, these structures were imported into Chem3D software to generate their corresponding 3D models. In parallel, the 3D structures of the target proteins were retrieved from the RCSB Protein Data Bank (PDB) database (https://www.rcsb.org/). Then, the compounds and targets were imported into AutoDock4.2.6 software for processing and molecular docking. The binding modes with the lowest binding energy in the molecular docking results were screened and visualized with PyMOL2.5 software.

Cell culture and treatment

The BV-2 murine microglia cell line was cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) with low glucose (1.0 g L^–1 D-glucose; Gibco, USA), supplemented with 10 % fetal bovine serum (Invitrogen, USA), and maintained at 37 °C in a humidified incubator with 5 % CO₂. Upon reaching the fourth passage, the cells were treated with high glucose complete medium (4.5 g L^–1 D-glucose, 10 % fetal bovine serum) and sodium palmitate (100 μmol L^–1; Sigma-Aldrich, USA) for 24 hours to proceed with further experiments.

RNA isolation and real-time quantitative reverse transcription polymerase chain reaction (RT-qPCR)

Total RNA was extracted from the specimens using TRIzol reagent (Thermo Fisher Scientific, USA), following the manufacturer’s protocol. Reverse transcription was performed using the PrimeScript™ RT reagent Kit (Takara Biomedical Technology, China). Quantitative PCR (qPCR) was carried out using SYBR Green Master Mix (Beyotime Biotechnology, China). The expression data were normalized using Actin as an internal reference. The mouse primer sequences were as follows: IL-1β forward: GCTTCAGGCAGGCAGTATCA, reverse: AATGGGAACGTCACACACCA, TNFα forward: ACAGAAAGCATGATCCGCGA, reverse: TTGCTACGACGTGGGCTAC, AR_forward: CTGGGAAGGGTCTACCCAC, reverse: GGTGCTATGTTAGCGGCCTC, LGALS3_forward: AGACAGCTTTTCGCTTAACGA, reverse: GGGTAGGCACTAGGAGGAGC, ACTIN_forward: TGAGCTGCGTTTTACACCCT, reverse: TTTGGGGGATGTTTGCTCCA.

All bioinformatics analyses were carried out with the software R4.1.1. Statistical analyses were performed using GraphPad Prism. Data are presented as mean ± standard error of the mean (SEM). The Shapiro-Wilk test was used to assess the normality of the data distribution. For independent groups meeting parametric assumptions, ordinary one-way ANOVA followed by Dunnett's multiple comparisons test was used to compare each treatment group with the HG+PA model group. Where homogeneity of variances was violated, Brown-Forsythe and Welch ANOVA tests were used, with Dunnett's T3 test for appropriate multiple comparisons. Two-sided multiplicity-adjusted p values < 0.05 were considered statistically significant.

RESULTS AND DISCUSSION

Functional enrichment analysis and compound-target-pathway network construction of potential targets for isoquercitrin/hyperoside in DR

A total of 285 possible targets for isoquercitrin and 286 for hyperoside were predicted by PharmMapper. Using "Diabetic retinopathy" as a keyword in GeneCards, 1315 targets related to DR were identified. Through the intersection of hyperoside and DR-related targets, 86 overlapping genes were obtained (Fig. 2a). The PPI network of hyperoside-related genes was built using the STRING database and visualized with Cytoscape software, consisting of 86 nodes and 920 edges. Similarly, the intersection of isoquercitrin and DR-related targets resulted in 81 overlapping genes (Fig. 2b). The 81 essential genes related to isoquercitrin were input into STRING to obtain the PPI network, which includes 81 nodes and 817 edges.

In GO analysis of biological processes (BP), both hyperoside and isoquercitrin were associated with multiple stress-related processes, including response to organic substance, response to oxygen-containing compound, cell death, regulation of biological quality, cellular response to chemical stimulus, cellular response to oxygen-containing compound, response to chemical, and response to stress. Specific functions of hyperoside include apoptotic process (Fig. 2c), while isoquercitrin was involved in the regulation of response to stress (Fig. 2d).

Pathway enrichment analysis of hyperoside and isoquercitrin included Pathways in cancer, Proteoglycans in cancer, Relaxin signalling pathway, FoxO signalling pathway, Prostate cancer, Endocrine resistance, AGE-RAGE signalling pathway in diabetic complications, Prolactin signalling pathway, Adherens junction. Additionally, KEGG enrichment analysis showed that EGFR tyrosine kinase inhibitor resistance was specifically enriched for hyperoside (Fig. 2e). Isoquercitrin was associated with MAPK signalling pathway (Fig. 2f). The variations in targets between these two compounds, along with disparities in the enrichment of GO and KEGG, may constitute crucial factors contributing to the differences in bioactivity observed both in vivo and in vitro.

Based on these results, we employed Cytoscape to construct the Compound-target-pathway network diagram. The network analysis function in the software was used to analyse the network topology parameters. The results of hyperoside network map highlighted genes with higher degrees, including MAPK1, HRAS, RAF1, AKT1, AKT2, EGFR, IGF1R, suggesting their biological importance (Fig. 2g). The isoquercitrin network map revealed higher gene degrees for MAPK1, HRAS, AKT1, EGFR, MAPK10, MAPK14, MAPK8, IGF1R, indicating their potential significance in biological processes (Fig. 2h).

Fig. 2. Functional enrichment analysis and compound-target-pathway network construction of overlapping genes: a) Venn diagram, screening for overlapping genes of DR and hyperoside. The green colour represented hyperoside, the pink colour represented DR; b) Venn diagram, screening for overlapping genes of DR and isoquercitrin. The green colour represents isoquercitrin, the pink colour represents DR; c) biological process analysis of GO enrichment of hyperoside and DR overlapping genes; d) biological process analysis of GO enrichment of isoquercitrin and DR overlapping genes; e) KEGG pathway plot of overlapping genes between hyperoside and DR; f) KEGG pathway plot of overlapping genes between isoquercitrin and DR; g) hyperoside-target-pathway network. Hyperoside network map highlighted genes with higher degrees, including MAPK1, HRAS, RAF1, AKT1, AKT2, EGFR, IGF1R; h) Isoquercitrin-target-pathway network. The isoquercitrin network map revealed higher gene degrees for MAPK1, HRAS, AKT1, EGFR, MAPK10, MAPK14, MAPK8, IGF1R.

Identification and analysis of DEGs in DR

To investigate the expression patterns of potential drug targets in DR, we performed DEG analysis by integrating transcriptomic data from macular and peripheral samples of NPDR patients in the GEO dataset GSE160306. The PCA plot illustrates relatively good repeatability within the two sample groups, with very similar sample data and clear discrimination between the groups (Fig. 3a). Consequently, 540 DEGs were identified in macular samples, including 410 upregulated and 130 downregulated genes. Volcano maps (Fig. 3b) and heat maps (Fig. 3c) were generated to visually represent the distribution of DEGs.

Fig. 3. Identification and analysis of DEGs in DR: a) PCA plots, distribution in macular and periphery groups; b) volcano plot of DEGs, the red colour represents upregulated genes and blue colour represents downregulated genes; c) heatmaps of the top fifty differential genes in the macular and periphery groups.

Identification hub potential targets in DR

By overlapping hyperoside or isoquercitrin with 540 DEGs to obtain potential drug therapeutic targets, we identified 11 genes in macular samples (Figs. 4a,b). Finally, 4 potential DR-related targets were obtained by overlapping differential genes with DR-related genes in GeneCards. Further GO enrichment analysis showed that its biological function was mainly related to regulation of protein localization, regulation of cell proliferation, cell proliferation, regulation of protein localization to plasma membrane, negative regulation of extrinsic apoptotic signalling pathway, negative regulation of cellular component organization, regulation of protein localization to cell periphery, protein localization, regulation of extrinsic apoptotic signalling pathway, regulation of cellular localization (Fig. 4c). The expression levels of the four hub genes in the GSE160306 dataset are displayed in Fig. 4d. AR, DPP4(dipeptidyl peptidase 4), and HMGCR (3-hydroxy-3-methylglutaryl-coenzyme A reductase) were increased in the macular, while LGALS3 expression was decreased in the macular.

Fig. 4. Identification and analysis of hub potential targets in DR: a) Venn plot of DEGs, DR and hyperoside; b) Venn plot of DEGs, DR and isoquercitrin; c) biological process analysis of GO enrichment of overlapping genes; d) expression of hub potential targets in GSE160306.

Molecular docking

To evaluate the binding affinities of the four potential target proteins for hyperoside and isoquercitrin, molecular docking was performed using AutoDock. AR and LGALS3 showed relatively favourable binding affinities for both compounds, supporting their potential interactions with hyperoside and isoquercitrin (Table I, Fig. 5).

Table I. Molecular docking binding energies of the hub target proteins with hyperoside and isoquercitrin

Fig. 5. Molecular docking maps of hub targets with hyperoside and isoquercitrin: a) AR-hyperoside; b) AR-isoquercitrin; c) LGALS3-hyperoside; d) LGALS3-isoquercitrin.

Experimental validation in BV-2 cells

Finally, the biological activities of hyperoside and isoquercitrin were experimentally verified in BV-2 cells. To examine the cellular effects of hyperoside and isoquercitrin, we measured IL-1β, TNF-α, AR, and LGALS3 mRNA expression in BV-2 cells exposed to high glucose and palmitate (HG+PA) (Fig. 6). Compared with the HG+PA group, hyperoside at 20 μmol L^–1 reduced IL-1β mRNA expression (p = 0.0343), while isoquercitrin at 10 μmol L^–1 also reduced IL-1β expression (p = 0.0306; Fig. 6a). Hyperoside at 20 μmol L^–1 decreased TNF-α mRNA expression (p = 0.0435), whereas no significant decrease was indicated for either tested concentration of isoquercitrin (Fig. 6b). AR mRNA expression was lower following treatment with hyperoside at 20 μmol L^–1 (p = 0.0116) and 10 μmol L^–1 (p = 0.0029), as well as isoquercitrin at 10 μmol L^–1 (p = 0.0051; Fig. 6c). In contrast, LGALS3 mRNA expression did not differ significantly between the HG+PA group and the control group (Fig. 6d). Thus, at the tested concentrations, hyperoside at 20 μmol L^–1 was associated with decreases in all three measured inflammatory or candidate-target transcripts (IL-1β, TNF-α, and AR), whereas isoquercitrin at 10 μmol L^–1 was associated with decreases in IL-1β and AR.

Fig. 6. Experimental validation in BV-2 cells: a) IL-1β expression was validated by RT-qPCR in BV-2 (HG and PA (100 μmol L^–1)); b) TNF-α expression was validated by RT-qPCR in BV-2 (HG and PA (100 μmol L^–1)); c) AR expression was validated by RT-qPCR in BV-2 (HG and PA (100 μmol L^–1)); d) LGALS3 expression was validated by RT-qPCR in BV-2 (HG and PA (100 μmol L^–1)). Values are shown as mean ± SEM (n = 6 independent biological replicates). Compared with the HG+PA group, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

DR is the main cause of adult-acquired blindness, and there is still a lack of effective drugs to treat it (15). Many studies have reported that natural products and herbal formulas can effectively treat DR. Hyperoside, a flavonoid, has been shown to exert protective effects against DR; however, the protective potential of its isomer isoquercitrin remains largely unexplored. More importantly, whether the structural differences between these two isomers contribute to differences in their biological activities and protective effects against DR remains unclear.

In this study, we used network pharmacology and bioinformatics to investigate the potential therapeutic mechanisms and targets of hyperoside and isoquercitrin for DR. First, we used PharmMapper to predict the target, screen out the genes related to DR, and construct the disease-drug-target-mechanism network. Further functional enrichment of these genes showed that hyperoside and isoquercitrin have different mechanisms of action on DR. For GO analysis of biological process function, hyperoside was specifically associated with the apoptotic process, whereas isoquercitrin was involved in the regulation of the stress response. Additionally, KEGG functional enrichment analysis indicated that hyperoside significantly enriched pathways related to EGFR tyrosine kinase inhibitor resistance, while isoquercitrin was linked to the MAPK signalling pathway. These differences suggest that the two isomers may influence the development of DR through different mechanisms.

To investigate the expression patterns of potential drug targets in DR, we employed bioinformatics analysis on the dataset GSE160306. This involved examining gene expression differences between macular and peripheral tissues in patients with DR, resulting in the identification of 540 differentially expressed genes. Finally, four potential therapeutic targets for DR were obtained by further intersection of DR related genes. Molecular docking further demonstrated favourable binding affinities of AR and LGALS3 for hyperoside and isoquercitrin. However, due to the different binding activity to the target protein, the biological activity may be different. AR regulates lipocalin production, maintains or disrupts energy metabolism and glucose homeostasis, and is a potential target for the prevention of androgen-related metabolic disorders (16). AR is widely distributed in the retina; however, its role in DR remains to be elucidated. Clinical studies have determined that blood LGALS3 levels are closely related to diabetes and its complications (17). Studies have shown that LGALS3 has a pro-inflammatory effect, causing degeneration of retinal nerve fibers and cell bodies, and is a key target of diabetic induced neurodegeneration of the mouse visual system. Knockout of LGALS3 effectively reduced apoptosis of ganglion cell layer (18). These findings suggest that AR and LGALS3 may be involved in the pathogenesis of DR and may represent potential targets of hyperoside and isoquercitrin. Modulation of AR and LGALS3 expression may therefore contribute to the protective effects of these compounds against DR.

The involvement of microglia in the development of DR has been well-established, mainly focusing on their contribution to inflammation, which is still an indispensable part of the pathogenesis of the disease, especially in the late stage of the disease (19). In the present study, hyperoside at 20 μmol L^–1 significantly reduced IL-1β, TNF-α, and AR mRNA expression in HG+PA-treated BV-2 cells, whereas isoquercitrin at 10 μmol L^–1 significantly decreased IL-1β and AR expression. These findings suggest that hyperoside may exert broader anti-inflammatory effects under the tested conditions, although no statistically significant superiority over isoquercitrin has been established. The observed concentration-related responses were not consistently monotonic, potentially reflecting complex cellular responses to flavonoids, as biphasic effects have been reported for natural antioxidants and flavonoids (20, 21). Furthermore, differences in compound stability, solubility, and cellular exposure may influence their observed activities under in vitro conditions (22, 23). However, these possibilities remain speculative and require further investigation using a broader concentration range and appropriate mechanistic assays.

Several limitations should be acknowledged. First, the computational analyses were primarily based on human gene and protein databases, whereas experimental validation was performed in murine BV-2 cells, potentially limiting cross-species extrapolation. Second, AR and LGALS3 were assessed only at the mRNA level without protein-level or functional validation, and molecular docking alone cannot confirm direct target engagement or causal mechanisms. Third, the BV-2 metabolic-stress model does not fully reproduce the complex retinal microenvironment of diabetic retinopathy. Future studies incorporating additional independent biological replicates, protein-level analyses, target-specific functional experiments, and human retinal or in vivo models are needed to validate these findings and clarify the underlying mechanisms.

CONCLUSIONS

In conclusion, integrated network pharmacology, transcriptomic analysis, and molecular docking revealed shared and potentially distinct mechanisms of hyperoside and isoquercitrin in diabetic retinopathy. Hyperoside was primarily associated with apoptotic processes and EGFR tyrosine kinase inhibitor resistance, whereas isoquercitrin was linked to stress response regulation and MAPK signalling, with AR and LGALS3 identified as potential targets. Experimental validation showed that hyperoside at 20 μmol L^–1 significantly reduced IL-1β, TNF-α, and AR mRNA expression in HG+PA-treated BV-2 cells, whereas isoquercitrin at 10 μmol L^–1 significantly decreased IL-1β and AR expression. These findings suggest that hyperoside may exhibit broader anti-inflammatory activity under the tested conditions and highlight AR as a potential regulatory target. However, further protein-level validation, mechanistic investigations, and retinal disease models are required to clarify their therapeutic potential in diabetic retinopathy.

Acronyms, abbreviations, symbols. – AR – androgen receptor, BP – biological process, DPP4 – dipeptidyl peptidase 4, DR – diabetic retinopathy, DEGs – differential expression genes, GEO – Gene Expression Omnibus, GO – gene ontology, HG – high glucose, HMGCR – 3-hydroxy-3-methylglutaryl-coenzyme A reductase, KEGG – Kyoto Encyclopedia of Genes and Genomes, PA – palmitate, PCA – principal component analysis, PDB – Protein Data Bank, PPI – protein-protein interaction, LGALS3, galectin-3, RT-qPCR – reverse transcription quantitative polymerase chain reaction, SEM – standard error of the mean.

Conflict of interest. – The author(s) report no conflicts of interest in this work. During the preparation of this work, ChatGPT (5.0) was used for language improvement. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Funding. – This work was funded by Natural Science Foundation of Jiangsu Province (BK20250551), Medical Research Projects of Jiangsu Provincial Health Commission (MQ2024034), Nantong Municipal Commission of Health and Family Planning (QN2025017, MS2024021), The high-level science and technology project Nurturing fund of Nantong First People's Hospital (YBSJJ2023003).

Author’s contribution. – Conceptualization, Y.G., X.W., and T.C.; methodology, T.C.; validation, Y.G. and X.W.; formal analysis, visualization, Y.G. and T.C.; writing, original draft preparation, Y.G. and X.W.; writing, review and editing, T.C. and D.Z.; funding acquisition, T.C. and D.Z. All authors have read and agreed to the published version of the manuscript.

REFERENCES

1. N. Bayan, N. Yazdanpanah and N. Rezaei, Role of toll-like receptor 4 in diabetic retinopathy, Pharmacol. Res. 175 (2022) Article ID 105960 (10 pages); https://doi.org/10.1016/j.phrs.2021.105960

2. K. Y. Zou, X. Li, B. B. Ren, F. Cheng, J. Ye and Z. L. Ou, Single-cell analysis identifies MKI67⁺ microglia as drivers of neovascularization in proliferative diabetic retinopathy, J. Transl. Med. 23(1) (2025) Article ID 10 (15 pages); https://doi.org/10.1186/s12967-025-06320-w

3. C. Sabanayagam, R. Banu, M. L. Chee, R. Lee, Y. X. Wang, G. Tan, J. B. Jonas, E. L. Lamoureux, C.-Y. Cheng, B. E. K. Klein, P. Mitchell, R. Klein, C. M. G. Cheung and T. Y. Wong, Incidence and progression of diabetic retinopathy: A systematic review, Lancet Diabetes Endocrinol. 7(2) (2019) 140–149; https://doi.org/10.1016/s2213-8587(18)30128-1

4. Y. Z. Le, VEGF production and signalling in Muller glia are critical to modulating vascular function and neuronal integrity in diabetic retinopathy and hypoxic retinal vascular diseases, Vision Res. 139 (2017) 108–114; https://doi.org/10.1016/j.visres.2017.05.005

5. J. O. Wallsh and R. P. Gallemore, Anti-VEGF-resistant retinal diseases: A review of the latest treatment options, Cells 10(5) (2021) Article ID 1049 (23 pages); https://doi.org/10.3390/cells10051049

6. L. Ji, H. Tian, K. A. Webster and W. Li, Neurovascular regulation in diabetic retinopathy and emerging therapies, Cell. Mol. Life Sci. 78(16) (2021) 5977–5985; https://doi.org/10.1007/s00018-021-03893-9

7. S. Sivaprasad, T. Y. Wong, T. W. Gardner, J. K. Sun and N. M. Bressler, Diabetic retinal disease, Nat. Rev. Dis. Primers 11(1) (2025) Article ID 62 (18 pages); https://doi.org/10.1038/s41572-025-00646-x

8. K. Wang, H. Zhang, L. Yuan, X. Li and Y. Cai, Potential implications of hyperoside on oxidative stress-induced human diseases: A comprehensive review, J. Inflamm. Res. 16 (2023) 4503–4526; https://doi.org/10.2147/jir.S418222

9. S. Xu, S. Chen, W. Xia, H. Sui and X. Fu, Hyperoside: A review of its structure, synthesis, pharmacology, pharmacokinetics and toxicity, Molecules 27(9) (2022) Article ID 3009 (26 pages); https://doi.org/10.3390/molecules27093009

10. X. Chen, A. C. Famurewa, J. Tang, O. O. Olatunde and O. J. Olatunji, Hyperoside attenuates neuroinflammation, cognitive impairment and oxidative stress via suppressing TNF-α/NF-κB/caspase-3 signalling in type 2 diabetes rats, Nutr. Neurosci. 25(8) (2021) 1774–1784; https://doi.org/10.1080/1028415x.2021.1901047

11. X. Luo, Y. Gong, Q. Jiang, Q. Wang, S. Li and L. Liu, Isoquercitrin promotes ferroptosis and oxidative stress in nasopharyngeal carcinoma via the AMPK/NF-kappa B pathway, J. Biochem. Mol. Toxicol. 38(1) (2023) Article ID e23542 (10 pages); https://doi.org/10.1002/jbt.23542

12. Q. Zhang and Y. Yan, The role of natural flavonoids on neuroinflammation as a therapeutic target for Alzheimer’s disease: A narrative review, Neural Regen. Res. 18(12) (2023) 2582–2591; https://doi.org/10.4103/1673-5374.373680

13. W. Wu, Z. Xie, Q. Zhang, Y. Ma, X. Bi, X. Yang, B. Li and J. Chen, Hyperoside ameliorates diabetic retinopathy via anti-oxidation, inhibiting cell damage and apoptosis induced by high glucose, Front. Pharmacol. 11 (2020) Article ID 797 (12 pages); https://doi.org/10.3389/fphar.2020.00797

14. W. Shen, Z. Song, X. Zhong, M. Huang, D. Shen, P. Gao, X. Qian, M. Wang, X. He, T. Wang, S. Li and X. Song, Sangerbox: A comprehensive, interaction-friendly clinical bioinformatics analysis platform, iMeta 1(3) (2022) Article ID e36 (6 pages); https://doi.org/10.1002/imt2.36

15. X. Ai, P. Yu, Y. Hou, X. Song, J. Luo, N. Li, X. Lai, X. Wang and X. Meng, A review of traditional Chinese medicine on treatment of diabetic retinopathy and involved mechanisms, Biomed. Pharmacother. 132 (2020) Article ID 110852 (11 pages); https://doi.org/10.1016/j.biopha.2020.110852

16. S. Sakkiah, T. Wang, W. Zou, Y. Wang, B. Pan, W. Tong and H. Hong, Endocrine disrupting chemicals mediated through binding androgen receptor are associated with diabetes mellitus, Int. J. Environ. Res. Public Health 15(1) (2017) Article ID 25 (17 pages); https://doi.org/10.3390/ijerph15010025

17. Y. Li, T. Li, Z. Zhou and Y. Xiao, Emerging roles of galectin-3 in diabetes and diabetes complications: A snapshot, Rev. Endocr. Metab. Disord. 23(3) (2022) 569–577; https://doi.org/10.1007/s11154-021-09704-7

18. H. R. Mendonca, R. Carpi-Santos, K. da Costa Calaza and A. M. Blanco Martinez, Neuroinflammation and oxidative stress act in concert to promote neurodegeneration in the diabetic retina and optic nerve: Galectin-3 participation, Neural Regen. Res. 15(4) (2020) 625–635; https://doi.org/10.4103/1673-5374.266910

19. P. Quiriconi, V. Hristov, M. Aburaya, U. Greferath, A. I. Jobling and E. L. Fletcher, The role of microglia in the development of diabetic retinopathy, npj Metab. Health and Dis. 2(1) (2024) Article ID 7 (8 pages); https://doi.org/10.1038/s44324-024-00009-2

20. U. Barreiro-Sisto, S. Fernández-Fariña, A. M. González-Noya, R. Pedrido and M. Maneiro, Enemies or allies? Hormetic and apparent non-dose-dependent effects of natural bioactive antioxidants in the treatment of inflammation, Int. J. Mol. Sci. 25(3) (2024) Article ID 1892 (19 pages); https://doi.org/10.3390/ijms25031892

21. E. J. Calabrese, P. Pressman, A. W. Hayes, G. Dhawan, R. Kapoor, E. Agathokleous and V. Calabrese, Rutin, a widely consumed flavonoid, that commonly induces hormetic effects, Food Chem. Toxicol. 187 (2024) Article ID 114626 (9 pages); https://doi.org/10.1016/j.fct.2024.114626.

22. L. Mahmutović, A. Sezer, E. Bilajac, A. Hromić-Jahjefendić, V. N. Uversky and U. Glamočlija, Polyphenol stability and bioavailability in cell culture medium: Challenges, limitations and future directions, Int. J. Biol. Macromol. 279 (2024) Article ID 135232 (8 pages); https://doi.org/10.1016/j.ijbiomac.2024.135232

23. L. Hu, Y. Luo, J. Yang and C. Cheng, Botanical flavonoids: Efficacy, absorption, metabolism and advanced pharmaceutical technology for improving bioavailability, Molecules 30(5) (2025) Article ID 1184 (35 pages); https://doi.org/10.3390/molecules30051184