ARPE-19 cells (obtained from Guangzhou Saiku Biotechnology Co., Ltd.) were cultured in DMEM/F12 medium containing 1% penicillin-streptomycin and 10% inactivated fetal bovine serum (FBS). The cells were then placed in a cell culture incubator at 37 °C with a 5% CO2 atmosphere. After overnight incubation, the cells fully adhered to the wall. Subsequently, the cells were pre-treated with different concentrations (0, 6.25, 12.5, 25, 50 μM) of luteolin (Solarbio) for 48 h, and then starved overnight in serum-free medium. Subsequently, they were induced with 10 ng/mL of TGF-β2 in DMEM/F12 medium containing 1% FBS for an additional 48 h to induce EMT. Cell proliferation and migration were evaluated at different time points (0, 24, 48, and 72 h) after the cells reached a final density of 80%–90% (Chen et al., 2014; Chen et al., 2022).

When the cells reached 80%–90% confluence, they were enzymatically digested, collected, and centrifuged. The cell density was adjusted to 1 × 10⁵ cells/mL, and the cells were seeded in a 96-well plate. After 24 h, different concentrations of luteolin were added to the 96-well plates and incubated in a 37 °C, 5% CO2 incubator. After incubating for 0, 24, 48, or 72 h, the original culture medium was discarded, replaced with 10% CCK-8 reagent (40203ES80; Shanghai Yisheng Biotechnology Co., Ltd.), and incubated for 1‒4 h. The absorbance was measured at 450 nm using an enzyme marker (Thermo Fisher Scientific) (Zhang et al., 2023). This process was repeated three times with five replicates for each concentration.

Two lines were drawn on the back of the 6-well plate in advance; each line was the size of the well diameter to ensure that all subsequent photographs were taken in the same position. When the cells reached 80%–90% confluence, they were enzymatically digested, centrifuged, and collected. The cell density was adjusted to 1 × 10⁷/mL, and the cells were evenly seeded in a 6-well plate. Before the cells reached 80%–90% confluence, they were starved for 24 h and then marked with a 20–200 μL yellow gun tip perpendicular to the horizontal line on the back. Photographs were taken and recorded at fixed positions at 0, 12, 24, 36, and 48 h after marking. The migrated area was calculated using the following formula: (initial area - area at each time point)/initial area. Statistical analysis was performed to evaluate the differences in the migrated area over time (Shi et al., 2021). This process was repeated three times.

After treatment with different cell concentrations of luteolin, the cells were digested, centrifuged, and collected. We added 500–600 μL of prepared DMEM/F12 medium containing 10% fetal bovine serum to the lower chamber and 200 μL of cell suspension without serum to the upper chamber. Incubation was continued for 24 h. The following day, the cells were fixed with 4% paraformaldehyde for 30 min, stained outside the chamber with 0.1% crystal violet for 20 min, and counted. The migration of ARPE-19 cells was observed under an inverted microscope, randomly selecting five fields of view (including top, bottom, left, right, and center) with a ×20 objective lens, pictures taken, and the counts analyzed using ImageJ (Zhang et al., 2023). This process was repeated at least three times.

After sample processing, the cells were washed with PBS, fixed overnight in pre-cooled 75% ethanol. Subsequently, the cells were washed twice with PBS. For staining, 0.5 mL of PI staining solution was added to each sample. The cells were gently pipetted to ensure proper mixing, incubated for 15 min in the dark, and analyzed on a flow cytometry machine as soon as possible. The DNA content of the cells at each stage was determined using flow cytometry, and the resulting data were analyzed using FlowJo v10 software (Saika, 2006; Zhou et al., 2020). This process was repeated at least three times.

Proteins were extracted from ARPE-19 cells and mouse retinas, and the protein concentrations were determined using the BCA method. The experiment was performed with α-SMA (1:5000, Proteintech), vimentin (1:2000, CST), GAPDH (1:5000, Proteintech), p-ERK1/2 (1:2000, Immunoway), goat anti-rabbit antibody (1:5000, Boster), and goat anti-mouse antibody (1:5000, Boster). The target proteins were detected using an ECL chemiluminescence kit (Advansta). The blots were visualized using a Syngene G-Box Gel Imaging System (Liu et al., 2024; Xu et al., 2025).

ARPE-19 cells were seeded in a 6-well plate and allowed to adhere for 48 h. Subsequently, the cells were washed with PBS, fixed with 4% paraformaldehyde for 15 min, permeabilized using 0.3‰ Triton X, and washed three times with PBS for 5 min each. The sections were blocked with 10% goat serum at room temperature for 1 h and incubated with vimentin (1:200, CST) at 4 °C overnight. The next day, the cells were washed three times with PBS for 5 min each and incubated with goat anti-rabbit antibody at room temperature in the dark for 1 h. After staining with DAPI for 5 min, an anti-fluorescence quencher was added, and the cells were observed under an inverted fluorescence microscope (Ferro Desideri et al., 2024). The entire process was repeated at least three times.

Male C57BL/6J mice aged 6–8 weeks were purchased from Nanjing Kerui Animal Co., Ltd. All experiments complied with the “ARVO Statement for the Use of Animals in Ophthalmic and Vision Research” and were approved by the Institutional Animal Care and Use Committee of Nanchang University (License No.: SYXK (Gan) 2021-0001). Twenty-four mice were divided equally into three groups: normal, ARPE-19 cells + platelet-rich plasma (PRP) + PBS, and ARPE-19 cells + PRP + Luteolin groups, all operated on in the right eye. Gatifloxacin eye drops was used 3 days before the operation. Mice were anesthetized via intraperitoneal injection of 3.6% concentration and 0.1 mL/10 g dose of chloral hydrate before the operation. The PVR model was established by the intravitreal injection of ARPE-19 cells and PRP using a 30G needle inserted through the sclera, 1 mm behind the corneal-scleral junction. Subsequently, 0.75 μL of ARPE-19 cells (approximately 1 × 10⁵/mL) and 0.75 μL of PRP were loaded into a 10 μL Hamilton syringe (Hamilton, Reno, NV), and the needle was removed to label the mice. Three days later, PBS or luteolin was injected into the vitreous cavities of the different groups. After 28 days, the PVR model was established, and the effects of luteolin on PVR in mice and the related molecular mechanisms were determined (Ferro Desideri et al., 2024; Hombrebueno et al., 2014; Pastor, 1998).

The tablets were baked at 85 °C for 35 min and dewaxed with an environmental dewaxing solution for 30 min. They were then washed with a graded alcohol to water solution and rinsed with PBS for 10 min. To facilitate antigen retrieval, high-temperature repair was performed by immersing the tablets in an antigen retrieval solution for 8 min, followed by an 8-minute cooling period. This cycle was repeated once, and the tablets were then cooled to normal temperature. Next, permeabilization was achieved by treating the tablets with 0.5% Triton X-10 for 30 min. The eyeball was circled with a histochemical pen and blocked for more than 1 h. The vimentin protein was incubated overnight at 4 °C. The next day, the cells were incubated with goat anti-rabbit secondary antibody (1:200) for 1 h in the dark and was stained with DAPI for 5 min. Anti-fluorescence quencher was added to seal the slides, and the cells were observed and photographed under an inverted fluorescence microscope (Shi et al., 2021; Zhang et al., 2023). The entire process was repeated thrice.

The slices were baked in an oven at 85 °C for 35 min, dewaxed with the environmental dewaxing solution for 30 min, and rinsed with a graded alcohol to water. The sections were stained with hematoxylin for 1 min and then rinsed with distilled water (ddH2O). After staining with eosin for 30 s, the sections were rinsed again with ddH2O. To remove water marks, the slides were wiped in a dark box, focusing on the back and periphery of the slide. After thorough air-drying, the slides were sealed with neutral resin and a cover glass was placed over the eyeball area. The images were observed under an inverted microscope, and photographs were taken for documentation purposes (Zhang et al., 2023).

The above experiments were repeated at least three times, and the data obtained were plotted and statistically analyzed using GraphPad Prism software (version 8.0). The data are presented as the mean ± SD. Statistical analysis was performed using ANOVA (analysis of variance) followed by Tukey’s post hoc test for multiple comparisons. A value of P < 0.05 was considered statistically significant (Buonato and Lazzara, 2014).

The OD value was measured using the CCK-8 assay, and the half-inhibitory concentration (IC50) of luteolin treatment for 48 h was calculated to be 37.54 μM. Among the tested concentrations, treatment with 25 μM luteolin for 48 h was identified as the optimal condition to achieve approximately 50% inhibition of cell viability without prolonging exposure. Cell viability decreased with increasing luteolin concentration and treatment time (Figure 1). To further analyze DNA cycle distribution, flow cytometry was performed on ARPE-19 cells treated with different concentrations of luteolin (0, 12.5, 25 μM) for 48 h. As luteolin concentration increased, the proportion of cells in G1 phase decreased while those in S phase increased (Figure 2). This suggests that luteolin interferes with normal cell cycle progression, leading to S-phase accumulation and reduced cell proliferation. When TGF-β2 was applied for 48 h, the number of G1-phase cells increased and S-phase cells decreased, consistent with EMT induction and cell cycle arrest. In contrast, in cells pretreated with luteolin, the proportion of G1-phase cells decreased while the S-phase fraction increased significantly (Figure 3). These findings indicate that luteolin mitigates TGF-β2–induced EMT in ARPE-19 cells, possibly by restoring normal cell cycle progression and maintaining proliferative balance.

Scratch and Transwell migration assays were used to assess the migration ability of ARPE-19 cells treated with different concentrations of luteolin. The scratch assay showed that untreated ARPE-19 cells had strong migration ability and almost completely closed the scratch after 48 h. However, after treatment with 6.25 μM, 12.5 μM, and 25 μM luteolin for 12 h, cell migration was inhibited. The cells’ migration ability did not show significant improvement with increased time (Figure 4A). The Transwell assay showed that after 24 h of treatment, the number of cells that migrated in the luteolin treatment group was significantly lower than that in the control group, and this effect was concentration-dependent (Figure 4B). The results from both assays confirmed that luteolin had a significant inhibitory effect on both the horizontal and vertical migration of ARPE-19 cells.

In this study, TGF-β2-induced EMT was verified by evaluating cell morphology and the expression of mesenchymal-related proteins. The inhibitory effect of luteolin on this process was further examined by treating cells with different concentrations of luteolin. ARPE-19 cells, a human RPE cell line with the typical cobblestone-like morphology, underwent a marked morphological change after exposure to 10 ng/mL TGF-β2, showing elongated and spindle-shaped features. However, after luteolin treatment, cells gradually regained an epithelial-like appearance, with many cells recovering a hexagonal or cobblestone-like shape and only a few remaining elongated (Figure 5A). Western blot analysis revealed that α-SMA and vimentin protein levels were significantly increased after 48 h of TGF-β2 treatment, whereas 12.5 μM and 25 μM luteolin markedly reduced this upregulation in a concentration-dependent manner (Figure 5C). Immunofluorescence staining for vimentin showed strong fluorescence after TGF-β2 stimulation, indicating high expression compared with the control group. In contrast, pretreatment with 12.5 μM and 25 μM luteolin significantly reduced vimentin fluorescence intensity compared with the TGF-β2 group (Figure 5B). These findings confirm that TGF-β2 successfully induces EMT in ARPE-19 cells and that luteolin effectively suppresses this transition by downregulating mesenchymal markers such as vimentin and α-SMA.

As mentioned above, 10 ng/mL TGF-β2 successfully induced EMT. Scratch and transwell migration assays were used to study the effect of luteolin on the migration of TGF-β2-treated ARPE-19 cells. In the scratch assay, TGF-β2 alone promoted wound closure in a time-dependent manner. After pre-treatment with 6.25 µM or 12.5 µM luteolin, it inhibited the wound closure of TGF-β2-treated ARPE-19 cells (Figure 6A). In the Transwell assay, the TGF-β2 pre-treatment group promoted the vertical migration of ARPE-19 cells (P < 0.0001). Pre-treatment with different concentrations of luteolin before TGF-β2 treatment significantly reduced the migration ability of ARPE-19 cells, and the degree of inhibition increased with increasing luteolin concentration (P < 0.0001) (Figure 6B). These results indicated that luteolin inhibited the horizontal and vertical migration of TGF-β2-treated ARPE-19 cells.

Starting from the day of induction (day 0), a slit-lamp examination was performed weekly to monitor the progression of PVR in the anterior segment and fundus of the mouse eye. Throughout the experiment, no accidental deaths, vitreous hemorrhages, cataracts, or infectious endophthalmitis were observed in any of the experimental animals. After 28 days of PVR induction, the eyeballs were removed, and paraffin sections were prepared and stained with hematoxylin and eosin (H&E). In the control group, the retinal layers were clear, the vitreous was transparent, the retinal structure was normal, and there were no signs of retinal folds or detachment points. Additionally, the vascular system was normal. Conversely, in the PBS group, the arrangement of the retinal cells was disrupted, the retina was funnel-shaped and detached, and the optic nerve was pulled with a large amount of traction membrane on the surface of the vitreous and retina, a large number of retinal folds, and swelling of the ganglion cell layer. In contrast, the luteolin group showed significant improvements in retinal morphology, with a significant reduction in the traction membrane on the surface of the vitreous and retina compared to the PBS group, and only partial folding and separation were observed (Figure 7).

At the same time, we further verified the results by performing immunofluorescence staining of mesenchymal proteins on paraffin sections and protein blotting experiments on mouse retinas. The immunofluorescence experiments showed that the fluorescence intensity of vimentin in the PBS group significantly increased, whereas that of vimentin in the luteolin group significantly decreased (Figure 8A). Western blot analysis showed that the protein expression of α-SMA and vimentin in the PBS group was significantly higher than that in the normal group (P < 0.0001), whereas the protein expression of α-SMA and vimentin in the luteolin group was significantly lower than that in the model group (P < 0.001) (Figure 8B). These results indicate that the PVR model, induced by injecting ARPE-19 cells combined with PRP into the vitreous cavity of mice, was successfully established and that luteolin injection into the vitreous cavity could reduce the severity of PVR.

The protein levels of phosphorylated ERK1/2 (p-ERK1/2) and total ERK1/2 were determined via Western blot analysis. Compared with the control group, the expression of p-ERK1/2 was markedly increased in the PBS-treated PVR model group (P < 0.001), while total ERK1/2 levels remained unchanged. In contrast, luteolin treatment significantly reduced the expression of p-ERK1/2 compared with the model group (P < 0.01), without affecting total ERK1/2 expression (Figure 9). These findings suggest that luteolin inhibit the progression of proliferative vitreoretinopathy (PVR) in mice by modulating ERK1/2 pathway activation rather than altering total protein expression.