Introduction

Age-related macular degeneration (AMD) is a progressive ophthalmic disorder that primarily affects the macular region of the retina, resulting in the deterioration of central visual function1. With the global population aging at an accelerating rate, the incidence of AMD continues to increase. This not only imposes substantial health-related challenges but also places a heavy economic burden on both individuals and society2,3. AMD is primarily classified into two main subtypes: dry and wet. Among these AMD cases, dry AMD accounts for approximately 85–90% of all AMD cases4,5.

Dry AMD is characterized by the deposition of drusen and pigmentary alterations in the fundus posterior pole. Concurrently, it features progressive atrophy of retinal pigment epithelial (RPE) cells and photoreceptor cells6. Oxidative stress is known to play a crucial role in the development of dry AMD7,8. Given their high metabolic rate, RPE cells are highly susceptible to generating reactive oxygen species (ROS). When ROS levels are imbalanced with those of the antioxidant system, oxidative stress ensues, inflicting damage to RPE cells9. Although the progression of dry AMD is relatively slow, it can progress to geographic atrophy in its late stage. In some cases, dry AMD may even transform into wet AMD, resulting in severe vision loss10,11. Currently, there is still a lack of effective therapeutic interventions for dry AMD4.

In recent years, natural compounds have emerged as promising candidates for the treatment of neurodegenerative and ocular diseases, drawing significant research interest12,13,14. Luteolin, a common flavonoid compound in the plant kingdom, typically exists as a glycoside. It is widely distributed across diverse plant sources, including vegetables (e.g., carrots, peppers, and celery), herbs (e.g., perilla and parsley), and flowers (e.g., honeysuckle and chrysanthemum)15. Luteolin exhibits a broad spectrum of bioactivities, such as antioxidant, anti-inflammatory, antitumor, and neuroprotective effects16,17,18. Mechanistically, luteolin inhibits protein acetylation and high glucose-induced oxidative stress in ARPE-19 cells through the SIRT1/P53 pathway19. In rat models of corneal alkali burns, luteolin administration effectively mitigates the cytokine- and chemokine-mediated inflammatory response, thereby alleviating corneal pathological damage20. When ARPE-19 cells are exposed to H2O2 treatment, luteolin significantly enhances the activities of superoxide dismutase and glutathione peroxidase while inhibiting epithelial-mesenchymal transition21. Moreover, in mouse models of depression-related dry eye, luteolin downregulates the production and release of proinflammatory cytokines in both hippocampal and corneal tissues22. Extracts of Lycium barbarum and Chrysanthemum morifolium protect mice against sodium iodate-induced retinal oxidative damage, alleviating retinal thinning and ROS levels while boosting superoxide dismutase levels. The active components responsible for this protective action are likely to be luteolin and Lycium barbarumpolysaccharides23. Intake of diverse dietary nutrients (e.g., minerals, antioxidants) slows AMD progression24,25, and luteolin’s significant antioxidant activity further supports its potential for dry AMD treatment. Taken together, luteolin exerts ocular protective effects via multiple mechanisms; however, the signaling pathways and multi-target interactions underlying these effects remain unclear.

Network pharmacology constructs a network model that connects drug components, targets, and diseases, thereby facilitating a comprehensive understanding of drug-disease interactions26,27. This research represents the first attempt to combine network pharmacology analysis with experimental validation, which is aiming to unravel the mechanism of action of luteolin in dry AMD. The procedural flowchart of the study is shown in Fig. 1.

Fig. 1
figure 1

Flowchart of the study.

Full size image

Methods

Related databases and platforms

TCMSP database (https://www.tcmsp-e.com/); GeneCards database (https://www.genecards.org/); OMIM database (https://omim.org/); DrugBank database (https://www.drugbank.com/); TTD database (https://db.idrblab.net/ttd/); CTD Database (http://ctdbase.org/); Super–PRED (https://prediction.charite.de/subpages/target_prediction.php); Swiss Target Prediction (www.swisstargetprediction.ch); PubChem database (https://pubchem.ncbi.nlm.nih.gov/); UniProt database (https://www.uniprot.org/); STRING 12.0 database (https://cn.string—db.org/); UniProt Protein Data Bank (https://www.uniprot.org/); PDB database (https://www.rcsb.org/); CB- DOCK2 (https://cadd.labshare.cn/cb-dock2); DAVID database (https://david.ncifcrf.gov/); KEGG PATHWAY database (https://www.genome.jp/kegg/pathway.html).

ADME properties of luteolin

The ADME characteristics of luteolin were determined by searching for the chemical term “Luteolin” in the TCMSP database, which provides information on its pharmacokinetic properties and other relevant attributes.

Target collection and construction of a “disease-target” network diagram for age-related macular degeneration

Disease-related targets associated with atrophic age-related macular degeneration, dry age-related macular degeneration, and non-exudative age-related macular degeneration were retrieved from the GeneCards, OMIM, DrugBank, and TTD databases. Targets with a relevance score ≥ 5 were selected from the GeneCards database. Subsequently, all database results were merged, and duplicate entries were excluded to obtain the final disease-related targets.

Target prediction of luteolin and construction of “Luteolin-target” network diagrams

Luteolin-related targets were retrieved from the CTD database. The canonical SMILES of luteolin were obtained from PubChem and imported into Super-PRED and SwissTargetPrediction for target prediction. In SwissTargetPrediction, predicted targets with a probability > 0.1 were retained. Target names were standardized to official gene symbols via UniProt, with duplicate, non-human, and non-standard entries excluded. Overlapping targets between luteolin and disease were identified via Venn diagram analysis.

Building the PPI network and analyzing key genes

The protein–protein interaction (PPI) network was constructed by inputting the intersecting targets into the STRING 12.0 database (Homo sapiens, confidence score ≥ 0.4). The resulting TSV file was subsequently imported into Cytoscape 3.10.1 for visualization. Topological metrics—Degree, Maximum Neighborhood Component (MNC), and Maximum Clique Centrality (MCC)—which reflect a node’s direct connection capability, participation in core functional modules, and impact on local network stability respectively, were calculated via the CytoHubba plugin, with the top 10 hub genes were obtained. Core targets were identified as the intersection of these three lists. The MCODE plugin was used with default parameters (degree cutoff = 2, node score cutoff = 0.2, K-score = 2, and Max depth = 100) to identify the most important target modules.

Molecular docking

To predict luteolin’s interactions with key proteins, its 3D structure (SDF format) was retrieved from PubChem. Core target proteins were sourced from UniProt and the Protein Data Bank (PDB), restricted to the human species, with PDB IDs selected where A ranges between 1 and 2. Corresponding 3D structures of core targets were obtained via the PDB IDs. Auto-blind docking of luteolin with target proteins was performed using the online CB-Dock2 database. Obtain PDB files of ligands pre- and post-docking, remove interfering structures, then use PyMOL for atomic alignment and minimum root mean square deviation (RMSD) calculation.

GO analysis and KEGG pathway enrichment analysis

The potential role of luteolin in dry AMD was explored via Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses28,29. Shared targets of luteolin and dry AMD were uploaded to the DAVID database for enrichment analyses, with the significance threshold set at P ≤ 0.05, False Discovery Rate (FDR) cut-off at ≤ 0.05, and “Homo sapiens” specified as the species. GO analysis covered biological processes (BP), cellular components (CC), and molecular functions (MF). KEGG pathways were annotated using pathway names and IDs from the KEGG PATHWAY database. The top 10 GO terms and 20 KEGG pathways were visualized.

Chemicals and reagents

DMEM/F12 medium was purchased from Servicebio (G4612, Wuhan, China). Fetal bovine serum (FBS) was obtained from Sijiqing (11011–8611, Zhejiang, China). Cell counting kit-8 (CCK-8) reagent (C0038), penicillin/streptomycin (C0222), and trypsin (C0201) were purchased from Beyotime (Shanghai, China). The Reactive Oxygen Species Assay Kit (CA1410) and Mitochondrial Membrane Potential Assay Kit with JC-1 (M8650) were obtained from Solarbio (Beijing, China). Sodium iodate was purchased from Macklin (7681-55-2, Shanghai, China), and luteolin was obtained from MedChemExpress (HY-N0162, Monmouth Junction, USA).

Cell culture and cell viability assays

The ARPE-19 cell line (GNHu45) was obtained from the Cell Bank of the Chinese Academy of Sciences. Cells were cultured in appropriate medium at 37 °C with 5% CO2. After being seeded in 96-well plates and reaching 70–80% confluence, the cells were treated with gradient concentrations of luteolin or sodium iodate for 24 h. Subsequent to treatment, 100 μL of medium containing 10% CCK-8 was added to each well, followed by incubation for 3 h. Absorbance was measured at 450 nm using a microplate reader to calculate cell viability.

Detection of intracellular ROS levels

Cells were randomly divided into three groups: control (Ctrl) group, sodium iodate (SI) group, and sodium iodate + luteolin (SI + LUT) group. In the SI + LUT group, cells were pretreated with luteolin for 24 h, after which the original medium was aspirated. Subsequently, both the SI + LUT group and the SI group were treated with sodium iodate simultaneously for 24 h.

2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) was diluted 1000-fold in serum-free medium to a working concentration of 5 μM. Cells were seeded in six-well plates. Following cell stimulation, the original medium was discarded, and freshly diluted DCFH-DA was added. Cells were incubated at 37 °C for 20 min, then washed three times to completely remove uninternalized DCFH-DA. Fluorescence was visualized and recorded using a fluorescence microscope.

Mitochondrial membrane potential level detection

Cells were treated as described in ROS levels detection. For each well in the six-well plate, the existing culture medium was aspirated. Subsequently, 1 mL of cell culture medium and 1 mL of 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolocarbocyanine iodide (JC-1) staining working solution were added. The mixture was thoroughly mixed, and the plate was incubated at 37 °C for 20 min. After incubation, the supernatant was aspirated, and the cells were washed twice with JC-1 staining buffer. Finally, the cells were examined using a fluorescence microscope.

Statistical analysis

Statistical analyses were performed using SPSS 27.0, and graphs were generated using GraphPad Prism 10.0. For GO and KEGG analyses, all results were false discovery rate (FDR)-corrected, with significance threshold set at FDR < 0.05. The experiment was independently replicated 3 times, with at least 3 replicates in each group (n = 3). Normality of the data was tested using the Shapiro–Wilk test, and homogeneity of variance was confirmed. Normally distributed data are expressed as mean ± standard deviation (x ± s) and analyzed for intergroup differences using one-way analysis of variance (ANOVA) followed by Dunnett’s post-hoc test. Non-normally distributed data are presented as median (interquartile range) [M (IQR)] and analyzed using the Kruskal–Wallis H test for multiple groups. Statistical significance was defined as P < 0.05, and 95% confidence intervals were calculated for key outcomes.

Results

ADME-related properties of luteolin

ADME characteristics of Luteolin were retrieved from TCMSP database (Table 1), including DL, OB, Caco2, BBB, and “Lipinski’s Rule of Five” (MW, AlogP, TPSA, Hdon, and Hacc).

Table 1 The ADME properties of luteolin.
Full size table

Dry AMD targets, luteolin-target genes, and intersection analysis

When searching for atrophic age-related macular degeneration in the GeneCards database, 2743 targets were retrieved, while 536 such targets were identified in the OMIM database. For dry age-related macular degeneration, using this as the search criterion, 2751 targets were obtained from the GeneCards database and 536 from the OMIM database. Additionally, 1 target was retrieved from the DrugBank database and 1 from the TTD database. When non-exudative age-related macular degeneration was used as the search criterion, 1 target was retrieved from the GeneCards database and 536 from the OMIM database. After combining all targets and removing duplicates, 4005 unique targets remained. The “disease-target” network graph contained 4006 nodes and 4005 edges. In the graph (Fig. 2A), red rectangles represent diseases, and blue rectangles represent disease targets.

Fig. 2
figure 2

Dry AMD and luteolin target genes. (A) A total of 4005 dry AMD-associated genes were retrieved from GeneCards, OMIM, DrugBank and TTD databases. Dry AMD is represented as a red rounded rectangle, and its associated genes as blue rounded rectangles. (B) From CTD, SuperPred and SwissTargetPrediction databases, 383 luteolin-targeted genes were obtained. Luteolin is shown as an orange rhombus, and its targeted genes as green circles. (C) Venn plot of common target genes between luteolin and dry AMD.

Full size image

A total of 383 target genes of luteolin were collected from various sources, including 215 from the CTD database, 107 from the Super-PRED database, and 100 from the SwissTargetPrediction database (Fig. 2B). Subsequently, intersection analysis identified 213 genes shared between luteolin and AMD (Supplementary Material 1, Fig. 2C). These shared genes were then used to further investigate the mechanism by which luteolin alleviates AMD.

PPI network analysis

The 213 intersecting targets were analyzed using the STRING database, generating a PPI network comprising 212 nodes and 5660 edges. In this network (Fig. 3A), rectangles represent intersecting targets, and edges indicate interactions between them. CytoHubba analysis was subsequently performed to identify eight core genes: TP53, TNF, IL6, AKT1, BCL2, STAT3, JUN, and CASP3 (Fig. 3B). The specific values of degree, MNC, and MCC are provided in Supplementary Material 2. Using the MCODE plugin in Cytoscape, 84 nodes with a score of 63.422 were identified in the PPI network. The identified module contained 2632 edges (Fig. 3C). Notably, the most significant module comprised the eight core genes mentioned above.

Fig. 3
figure 3

The PPI network and core genes of luteolin against dry AMD were analyzed via CytoHubba and MCODE. (A) The PPI network for luteolin against dry AMD: 212 blue rounded rectangles stand for the targets of luteolin in anti-dry AMD, and 5660 edges represent interactions between these targets. (B) CytoHubba sorted genes by degree, MNC, MCC and Venn diagram identified overlapping core genes. (C) MCODE was applied to analyze the most critical modules, where red rounded rectangles represent the overlapping genes (core genes) identified in (B).

Full size image

Molecular docking

For predicting interaction of luteolin with target proteins, three-dimensional structure of luteolin was acquired from PubChem database. Target proteins in PDB format were also retrieved: TP53 (PDB ID: 1AIE), TNF (PDB ID: 2R32), IL6 (PDB ID: 1ALU), AKT1 (PDB ID: 1UNQ), BCL2 (PDB ID: 5VAU), STAT3 (PDB ID: 6NJS), JUN (PDB ID: 5T01), and CASP3 (PDB ID: 1NMS). Subsequently, the core targets screened above were subjected to docking analysis with luteolin using CB-Dock2, and the resulting docking complexes were visualized (Fig. 4). Affinity reflects the binding capacity between a ligand and its receptor. When the affinity value is negative, a larger absolute value of this parameter indicates a stronger binding ability30. All complexes showed docking binding energies below − 5 kcal/mol, confirming the validity of the aforementioned screening results (Table 2). Among these, JUN exhibited the strongest docking activity with luteolin, with a binding energy of − 8.9 kcal/mol. Models with an RMSD ≤ 4 Å are generally regarded as reliable, whereas an RMSD ≤ 2 Å is indicative of accurate results31. All calculated RMSD values were found to be less than 2 Å (Table 2), thereby offering strong corroboration for the accuracy of the obtained docking results.

Fig. 4
figure 4

The molecular docking of luteolin with target proteins. The 3D images were derived from the online CB-Dock2 database (https://cadd.labshare.cn/cb-dock2). (A) Molecular docking of TP53 (PDB ID: 1AIE) with luteolin. (B) Molecular docking of TNF (PDB ID: 2R32) with luteolin. (C) Molecular docking of IL6 (PDB ID: 1ALU) with luteolin. (D) Molecular docking of AKT1 (PDB ID: 1UNQ) with luteolin. (E) Molecular docking of BCL2 (PDB ID:5VAU) with luteolin. (F) Molecular docking of STAT3 (PDB ID: 6NJS) with luteolin. (G) Molecular docking of JUN (PDB ID: 5T01) with luteolin. (H) Molecular docking of CASP3 (PDB ID: 1NMS) with luteolin.

Full size image
Table 2 Molecular docking analysis of luteolin binding energies with eight core anti-dry AMD targets.
Full size table

GO analysis and KEGG pathway enrichment analysis

A total of 213 luteolin target genes associated with AMD alleviation were input into the DAVID database, yielding 308 BP terms, 103 CC terms, and 109 MF terms. The P value indicated the significance of functional enrichment. Subsequently, the top 10 most significant entries were selected, and GO bar graphs were generated (Fig. 5A). In the BP category, prominent terms included response to xenobiotic stimuli, negative regulation of apoptotic processes, and positive regulation of gene expression. For CC, the terms encompassed cytoplasm, protein-containing complexes, extracellular space, and mitochondria. In the MF category, the terms involved protein binding, enzyme binding, and cytokine activity.

Fig. 5
figure 5

GO and KEGG Enrichment Analyses. (A) GO enrichment analysis bar graph. (B) Top-20 KEGG pathway enrichment analysis bubble diagram.

Full size image

KEGG analysis identified 152 pathways, with prominent ones including lipid and atherosclerosis, cancer-related pathways, and the AGE-RAGE signaling pathway in diabetic complications. These findings suggest that luteolin may act on these pathways to exert a therapeutic effect against AMD. For better visualization of significant pathways, Fig. 5B shows the top 20 significantly enriched KEGG signaling pathways.The Z-scores for all pathways, along with the heatmap of upregulated and downregulated pathways, are available in Supplementary Materials 3 and supplementary Fig. 1. Among these, the lipid and atherosclerosis pathway exhibited the most significant P value (Table 3).

Table 3 KEGG pathway enrichment analysis of luteolin against dry AMD.
Full size table

Network diagram of “luteolin-dry AMD-target-pathway”

Cytoscape 3.10.1 was used to visualize the “luteolin-target-pathway” network, with major biological mechanisms identified (Fig. 6). The analysis revealed that the network comprises 235 nodes and 1071 edges. These findings highlight the crucial signaling pathways through which luteolin exerts its therapeutic effects against AMD.

Fig. 6
figure 6

Network diagram of luteolin-dry AMD-target-pathway interactions.The inner blue rhombus is the relevant pathway and the outer green circle is the relevant target in the pathway.

Full size image

Pathway diagram of dry AMD and luteolin related targets

Analysis of the PPI network identified TP53, TNF, IL6, AKT1, BCL2, STAT3, JUN, and CASP3 as the eight key genes involved in luteolin’s therapeutic effects on AMD. This finding is consistent with the results of KEGG analysis, where the lipid and atherosclerosis pathway was the most significantly enriched. Notably, all eight aforementioned genes are associated with this pathway (Fig. 7).

Fig. 7
figure 7

The lipid and atherosclerosis pathway diagram, with red marking the intersection targets in this analysis.

Full size image

Luteolin attenuated sodium iodate-induced apoptosis in ARPE-19 cells

To assess the detrimental effects of sodium iodate on ARPE-19 cells, the cells were exposed to varying concentrations for 24 h. The findings indicated that at a sodium iodate concentration of 7.5 mM, the cell viability decreased to approximately 50% (Fig. 8A). Consequently, 7.5 mM was chosen as the concentration for model establishment.

Fig. 8
figure 8

Impact of luteolin on ARPE-19 cell viability. (A) Cells were incubated with varying concentrations of sodium iodate for 24 h. (B) The cells underwent treatment with varying concentrations of luteolin for a consistent 24 h period. (C) Cells underwent pretreatment with various luteolin concentrations prior to being exposed to 7.5 mM sodium iodate for 24 h. *P < 0.05, **P < 0.01, ***P < 0.001, n = 5.

Full size image

To evaluate the cytotoxicity of luteolin, the cells were treated with various concentrations of luteolin. CCK-8 assay results showed that luteolin is non-toxic to ARPE-19 cells and exerts a slight proliferative effect (Fig. 8B). To investigate luteolin’s protective effect against sodium iodate-induced damage, ARPE-19 cells were pretreated with different concentrations of luteolin for 24 h, followed by exposure to 7.5 mM sodium iodate. The results demonstrated that 40 μM luteolin exerted the most pronounced protective effect (Fig. 8C), suggesting that luteolin pretreatment can protect ARPE-19 cells from sodium iodate-induced damage.

Antioxidant properties of luteolin

Excessive production of ROS and impairment of the antioxidant defense mechanism lead to redox imbalance and increased oxidative stress. To assess the intracellular ROS levels, a DCFH-DA assay was used. Compared with the control group, the sodium iodate-treated group showed significantly elevated ROS levels; however, pretreatment with luteolin effectively reduced ROS levels (Fig. 9).

Fig. 9
figure 9

Luteolin’s effect on ROS levels in ARPE-19 cells. (A) DCFH-DA staining visualized intracellular ROS levels. Scale bar: 200 μm. (B) Fluorescence intensity quantification revealed luteolin suppressed ROS production. *P < 0.05, **P < 0.01, ***P < 0.001, n = 3.

Full size image

Protective effect of luteolin on mitochondrial membrane potentials

JC-1 staining was used to assess mitochondrial membrane potential across groups. An increase in green fluorescence signifies a reduction in mitochondrial membrane potential, suggesting that cells might undergo early apoptosis. Conversely, an increase in red fluorescence reflected a normal mitochondrial membrane potential.Compared with the control group, the sodium iodate-treated group showed a significant reduction in mitochondrial membrane potential. In contrast, pretreatment with luteolin increased the mitochondrial membrane potential in ARPE-19 cells (Fig. 10). These results indicate that luteolin can protect against sodium iodate-induced mitochondrial damage and alleviate early-stage cell apoptosis.

Fig. 10
figure 10

Effect of luteolin on mitochondrial membrane potential within ARPE-19 Cells. (A) JC-1 staining evaluated mitochondrial membrane potential. Red to green fluorescence shift indicates decreasing potential. Scale bar: 125 μm. (B) Fluorescence intensity analysis showed luteolin increased mitochondrial membrane potential. *P < 0.05, **P < 0.01, ***P < 0.001, n = 3.

Full size image

Discussion

Dry AMD is a common and complex ocular degenerative disease that poses a severe threat to the visual health of the elderly. Its pathological mechanisms include oxidative stress, inflammation, and abnormal activation of the complement system32,33. Luteolin, a natural flavonoid, has attracted extensive attention in recent years owing to its diverse biological activities, including antioxidant stress properties, attenuation of inflammatory responses, and regulation of cellular metabolism34. It exhibits promising potential for the prevention and treatment of dry AMD.

Network pharmacology analysis allows for the construction of network models, which facilitates exploring associations among targets, drugs, and diseases35,36. To identify the key targets of luteolinin the treatment of dry AMD, we employed a network pharmacology approach, yielding 213 overlapping targets of luteolin involved in the treatment of dry AMD. KEGG analysis suggested that luteolin may exert its beneficial effects primarily via the lipid and atherosclerosis pathway and cancer-related pathways. Notably, dry AMD patients often exhibit abnormal retinal lipid metabolism—specifically, accumulation of lipid-rich drusen (the hallmark lesion of dry AMD). Dry AMD is closely associated with lipid metabolism disorders37 and shares multiple risk factors with atherosclerosis38,39. Additionally, dry AMD patients who receive high-dose statin therapy have shown improvements in visual acuity40.The role of luteolin in atherosclerosis has been reported. Specifically, it influences atherosclerotic pathogenesis by inhibiting multiple steps of cholesterol biosynthesis to regulate lipid metabolism41. This lipid-regulating effect is dose-dependent, primarily achieved by modulating the expression of transcription factors and lipid synthases42. Additionally, luteolin exerts metabolic regulation in other disease models—for example, it effectively reduces serum triglycerides, total cholesterol in polycystic ovary syndrome rats43. AMD and cancer have common risk factors, with smoking standing out as a well-established one for both conditions5,44. Beyond this, potential risk factors linked to AMD, including inflammation-related genes (e.g., complement factor H polymorphism) and oxidative stress, are also associated with multiple cancers45,46,47. This suggests AMD and cancer may have a common pathogenic pathway. Notably, luteolin possesses well-documented antioxidant and antitumor properties48. It is reasonable to hypothesize that luteolin could exert a protective effect against AMD by modulating these cancer-related pathways.

PPI network analysis suggested that luteolin alleviates AMD symptoms by regulating multiple targets. CytoHubba analysis was then performed to screen eight core gene targets: TP53, TNF, IL6, AKT1, BCL2, STAT3, JUN, and CASP3. Docking results demonstrated that luteolin has a high affinity for JUN. c-Jun N-terminal kinase (JNK) acts as a critical activator of the JUN protein. Impairments in ocular microcirculation can induce activation of the JNK signaling pathway. Experimental evidence has shown that inhibiting JNK signaling alleviates symptoms in mouse models of dry AMD. These findings suggest that targeting the JNK signaling pathway may represent a promising therapeutic strategy for dry AMD49. TNF and IL6 are key inflammatory mediators that play pivotal roles in the inflammatory process. In the context of dry AMD, elevated plasma IL-6 levels show a strong correlation with the progression of geographic atrophy, a late-stage manifestation of the disease50.STAT3 is a pivotal protein involved in cell signaling pathways and the regulation of gene expression. It can be activated by inflammation-related cytokines such as IL651,52. Upon activation, STAT3 alleviates all-trans retinaldehyde-induced ferroptosis in 661W photoreceptor cells. This finding suggests that targeting STAT3 may serve as a potential therapeutic strategy for dry AMD53. In the dry AMD model, BCL2 expression is downregulated54,55, which leads to increased apoptosis and contributes to disease progression. Taken together, these results suggest that luteolin can alleviate dry AMD symptoms by targeting multiple proteins and intervening in various pathological processes.

Sodium iodate is a widely utilized stress inducer in both in vitro and in vivo models of dry AMD8,56. Its degeneration and atrophy-inducing mechanisms, such as oxidative stress and melanin-related pathways, closely align with dry AMD’s pathological processes57. When administered systemically, sodium iodate specifically targets RPE cells, causing focal loss and subsequent apoptosis of adjacent photoreceptors58. Prior to investigating the effects of luteolin, we evaluated its safety in ARPE-19 cells. The results of the safety assessment demonstrated that luteolin was noncytotoxic to ARPE-19 cells and had a slight proproliferative effect. We then found that sodium iodate exerted concentration-dependent cytotoxic effects on ARPE-19 cells. Notably, pretreatment with 40 μM luteolin effectively increased cell viability and counteracted the detrimental effects of sodium iodate, which is consistent with the effective concentrations of luteolin used in comparable in vitro models21. However, the in vivo effectiveness of this concentration remains unknown. This is because drugs undergo complex in vivo processes (e.g., absorption, distribution, metabolism, and excretion) that differ significantly from the in vitro scenario, where drugs interact directly with cells in a controlled environment. For instance, intravitreal injection of 1 μL of 10 μM luteolin in mice inhibits retinal neovascularization by blocking ROS generation59. Oral or intraperitoneal administration of 10 mg/kg luteolin reverses the elevation of proinflammatory factors in corneal tissues and reduces inflammatory factors in the aqueous humor22,60. Thus, additional verification through animal experimental models remains necessary for future research.

The pathogenesis of AMD is closely linked to oxidative stress in RPE cells61. RPE cells are equipped with endogenous antioxidant defense mechanisms to counteract oxidative damage. However, when the production of ROS exceeds the normal antioxidant capacity of cells and tissues, oxidative stress is triggered8,62. This, in turn, triggers a cascade of pathophysiological processes in cells and tissues. Mitochondria serve as the primary energy suppliers for cells. A reduction in mitochondrial membrane potential is indicative of mitochondrial dysfunction, which can lead to cellular damage and apoptosis63,64. The transition from red to green fluorescence in JC-1 staining clearly indicated a decrease in the mitochondrial membrane potential. These observations confirmed that sodium iodate treatment induced oxidative stress damage in ARPE-19 cells, leading to increased cellular ROS levels, reduced mitochondrial membrane potential, and ultimately cell apoptosis. In contrast, luteolin pretreatment reduced ROS production, restored mitochondrial membrane potential, and alleviated oxidative stress, thereby exerting a protective effect on ARPE-19 cells.

Luteolin exhibits strong antioxidant activity, with the ability to scavenge free radicals, inhibit lipid peroxidation, and protect cells from oxidative damage65. Its antioxidant mechanisms primarily encompass direct free radical scavenging, metal ion chelation, and upregulation of antioxidant enzyme activities66,67. Luteolin can upregulate the activity of antioxidant molecules such as superoxide dismutase by regulating the nuclear factor Nrf2 pathway in rats with polycystic ovary syndrome68. Similarly, in type 2 diabetic rats, luteolin treatment significantly increases superoxide dismutase, catalase, and glutathione levels versus untreated controls, indicating its effectiveness in alleviating oxidative stress69. Oxidative stress and inflammation are closely interrelated. Oxidative stress often serves as a trigger for chronic inflammatory diseases70,71. ROS can stimulate the production of inflammatory cytokines such as TNF-α and IL-6. These cytokines may trigger inflammation and cell death72, while luteolin’s anti-inflammatory properties include inhibiting ROS and inflammatory cytokine production73. Additionally, luteolin suppresses the generation of other inflammatory mediators, including prostaglandin E2 and nitric oxide74,75.

Luteolin, a natural flavonoid abundant in vegetables and fruits, has been shown to be safe for long-term low-dose intake76. This characteristic confers advantages for treating chronic diseases requiring lifelong management, such as dry AMD. Currently, the mainstream intervention for dry AMD involves antioxidant supplementation (e.g., the AREDS formula), but its efficacy remains limited77. In contrast, luteolin can synergize with existing antioxidants to enhance the protection of RPE cells through a multi-targeted approach, which holds promise for improving clinical outcomes. Diets combining flavonoids (luteolin, quercetin) with carotenoids exhibit strong antioxidant properties, with the synergistic antioxidant effect being more pronounced especially when flavonoids predominate78. However, the therapeutic application of luteolin is largely limited by its low bioavailability, which is attributed to its high lipophilicity, poor water solubility, extensive first-pass metabolism, and short half-life17,79. To address this issue, delivery strategies such as lipid carriers and nanoformulations have been developed. For instance, liposome-encapsulated luteolin exhibited enhanced antitumor activity against colorectal cancer CT26 cells compared with free luteolin80, while poly(propylene sulfide)-PEG nanoparticles (LUT-PPS-NPs) doubled the in vivo inhibition of tumor proliferation, migration, and invasion81. Future local delivery approaches, such as intravitreal or subconjunctival injection, may reduce systemic exposure while increasing ocular drug concentrations.

This study has several limitations. First, network pharmacology predictions rely on in silico simulations, with results derived from existing database data that may harbor errors or suffer from delayed updates. Second, the lack of in vivo validation means that the in vitro findings may not fully reflect luteolin’s effects in the complex physiological environment of a living organism. Third, the exclusive use of the ARPE-19 cell line, without incorporating primary human RPE cells or animal models, limits the generalizability of the results to authentic physiological or pathological contexts. Fourth, the investigation is limited to cellular-level endpoints, as functional and behavioral visual assessments have not been performed.

Conclusion

By integrating network pharmacology with experimental validation, this study demonstrates that luteolin exerts protective effects in dry AMD by modulating pivotal pathways such as “lipid metabolism and atherosclerosis” through core targets including TP53 and TNF. Molecular docking analyses further confirmed the high-affinity binding of luteolin to these core proteins. In vitro experiments demonstrated that luteolin significantly improved sodium iodate-induced loss of viability in ARPE-19 cells, reduced intracellular ROS accumulation, and stabilized mitochondrial membrane potential. These findings indicate that luteolin mitigates RPE cell injury via antioxidant activity and preservation of mitochondrial integrity, thereby providing a promising therapeutic candidate for dry AMD management.