The study with human PVR tissues was approved by the ethics committee of Zhongshan Ophthalmic Center (2020KYPJ016) and followed the Declaration of Helsinki. All patients provided informed consent. Proliferative membranes were discarded tissues during vitrectomy from PVR patients following RRD, immediately fixed in 4% paraformaldehyde for 24 h, and followed by immunofluorescence and analysis.

The human RPE cell line (ARPE-19) and the HEK-293T cell line were obtained from the American Type Culture Collection (ATCC, Manassas, VA, United States). Cells were cultured in DMEM/F12 or DMEM medium (Gibco, United States) with 10% fetal bovine serum and 1% Penicillin-Streptomycin at 37 °C in 5% CO2. 10 ng/mL TGF-β1 (Novoprotein, Suzhou, China) was used to induce EMT for 48 h (Ma et al., 2023). All cell lines were validated by STR analysis and tested for mycoplasma contamination.

Lentivirus was packaged in 293T cells by transfecting with psPAX2, pMD2. G plasmids, and shRNAs targeting ARL13B or a scramble control (Transheepbio, Suzhou, China) with TurboFect (Thermo, United States), collected, and used to infect ARPE-19 cells in 10 μg/mL Polybrene (Solarbio, Beijing, China). 48 h later, 2 μg/mL puromycin (Solarbio, Beijing, China) was applied to select infected cells.

Cells were seeded in 96-well plates at a density of 1 × 10⁴ and treated with TGF-β1. After 2 h of EdU incubation (Ribobio, Guangzhou, China), cells were fixed, permeabilized, and stained with Apollo and Hoechst. EdU-positive cells were visualized using a fluorescence microscope (Nikon, Japan) and quantified using ImageJ (NIH, United States) by calculating the percentage of EdU-positive cells (EdU stained/Hoechst stained × 100%).

Cells were seeded in 6-well plates at a density of 0.6 × 10⁶. A 200 µL plastic pipette tip was used to create a scratch at 80% confluence. Scratch images at 0 and 72 h with TGF-β1 were captured using an optical microscope (Nikon). The scratched areas were measured using ImageJ software (NIH, United States), and the wound closure rate was calculated as (A0 − A72)/A0 × 100%, where A0 is the area at 0 h and A72 is the area at 72 h.

200 µL of cells at a density of 1 × 10⁵ with TGF-β1 were seeded in the transwell insert (Corning). The lower chambers were filled with 600 µL medium containing 10% FBS. After 24 h, cells that migrated to the lower chamber were fixed, stained with crystal violet (Solarbio, China), and quantified using ImageJ. Migrated cells were imaged using an optical microscope (Nikon).

Coverslips with cells, PVR clinical specimens, or mouse eye sections were fixed, blocked, and incubated with primary antibodies overnight at 4 °C. Followed by secondary antibodies (anti-mouse Alexa 488, 555; anti-rabbit Alexa 555, 647; Thermo Fisher Scientific, United States) and DAPI staining. Images were captured using a confocal microscope (Zeiss, Germany). Primary antibodies included ARL13B (1:200, 17711-1-AP, Proteintech, China), α-SMA (1:200, A2547, Siama, United States), and Smad3 (1:200, 9523, CST, United States).

Total proteins were extracted using lysis buffer (Keygentec, China) and quantified via BCA assay (Thermo Fisher Scientific, United States). After SDS-PAGE (8% or 10%), proteins were transferred to PVDF membranes (Millipore, United States), blocked, and incubated with primary antibodies overnight at 4 °C, followed by HRP-conjugated secondary antibodies (CST, United States). Chemiluminescent substrate (Millipore, United States) was used for detection, and blots were visualized. Primary antibodies: ARL13B (1:1000, 17711-1-AP, Proteintech), Fibronectin (1:1000, 26836, CST, United States), Vimentin (1:1000, 5741, CST, United States), ZO-1 (1:10000, 33-9100, Thermo, United States), phospho-Smad3 (1:1000, 9520, CST, United States), Smad3 (1:1000, 9523, CST, United States), and GAPDH (1:1000, 2118, CST, United States). Band intensity was analyzed using ImageJ (NIH).

Total RNA was extracted according to the manufacturer’s instructions (Esscience, China) and synthesized to cDNA (Takara, Japan). Real-time PCR was performed using SYBR Green Master Mix (Roche, Switzerland) on the Roche 480 system (Switzerland). Primer sequences are listed in Supplementary Table S1. Relative gene expression was calculated using the 2^−ΔΔCt method, with GAPDH as an internal control.

Total RNA from cells was isolated (Esscience, China).

Total RNA from cells was isolated (Esscience, China). Enriched for mRNA and sequenced on the Illumina NovaSeq 6000 platform. FPKM values were calculated using StringTie, and differentially expressed genes were identified with the edgeR package. All RNA-seq samples were processed and sequenced in the same batch, using identical protocols and reagents. Genes with a log2 (fold change) > 0.5 and adj. p.value <0.05 were considered statistically significant. Gene set enrichment analysis (GSEA) was conducted using DAVID software.

All animal procedures were approved by the Animal Care and Use Committee of Sun Yat-sen University (2020-092) and followed the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. C57BL/6 J mice (6–8 weeks, 30 ± 10 g) from Charles River Laboratories (Beijing Vital River Laboratory Animal Technology Co., Ltd.) were used. The fundus was checked before the experiments. Mice were anesthetized, and pupils were dilated before intravitreal injection of 1 × 10⁴ ARPE-19 cells. Fundus images were captured with a retinal imaging system (Phoenix Micron IV). PVR stages were assessed at 3, 7, and 14 days using Fastenberg’s classification: stage 0: no proliferative response; stage 1: intravitreal proliferation; stage 2: focal traction with localized vascular changes; stage 3:localized detachments; stage 4:extensive retinal detachment; stage 5: total retinal detachment (Fastenberg et al., 1982).

Mouse eyeballs were fixed, embedded, and sectioned into 5 µm slices. After deparaffinization and rehydration, tissues were stained with hematoxylin, differentiated in hydrochloric ethanol, and stained with eosin. Samples were dehydrated, cleared in xylene, and imaged using a light microscope. (Nikon).

Statistical analyses were performed using GraphPad Prism 9.0 (GraphPad Software, Inc.). Comparisons were made using the Mann-Whitney U test and the unpaired Student’s t-test. Data are presented as mean ± standard deviation (SD) from at least three independent experiments, with p < 0.05 considered statistically significant.

After RRD, detached RPE migrates to epiretinal or subretinal and becomes the major cell component of the proliferative membrane, where it undergoes EMT with molecular features switching from epithelial to mesenchymal (Chiba, 2014). To investigate the expression of ciliary protein ARL13B in PVR specimens, we performed immunofluorescence staining and found that ARL13B expression was significantly lower in α-SMA + cells as compared to the Pan-CK + cells (Figures 1A,B). Accordingly, shortened primary cilia were found in α-SMA + cells as compared to the Pan-CK + cells (Figure 1A,C). To further investigate the role of ARL13B in EMT, we used TGF-β1 to induce EMT of ARPE-19 in vitro. The Western blot showed a significant reduction of ARL13B with TGF-β1 treatment in ARPE-19 cells (Figures 1D,E), accompanied by shortened primary cilia in immunofluorescence staining (Figures 1F,G). These results were further confirmed by using iPSC-RPE cells (Supplementary Figure S1). Together, these findings suggest that reduced ciliary protein ARL13B RPE is associated with RPE-EMT during PVR.

ARL13B is an essential component of primary cilia and is involved in cilia structure maintenance in different cell types (Larkins et al., 2011). To further explore the effect of ARL13B on the cilia structure of RPE cells, we knocked down ARL13B in ARPE-19 cells by using ARL13B shRNA lentivirus. Western blot confirmed a robust reduction of ARL13B in knockdown (ARL13B KD) cells compared to scramble shRNA controls (Figures 2A,B). Notably, immunofluorescence staining demonstrated a significant shortening of primary cilia in ARL13B KD cells (Figures 2C,D), indicating that loss of ARL13B can lead to cilia deficiency in ARPE-19 cells.

To explore the role of ARL13B in TGF-β1-induced EMT in vitro, we first investigated morphological changes of ARL13 B-deficient RPE cells. As the phase image results showed, the loss of ARL13B alone resulted in a cellular morphological change from oval to spindle-shaped. This morphological change was more pronounced upon TGF-β1 (Figure 3A), indicating that loss of ARL13B facilitates the transition from an epithelial to a mesenchymal phenotype for RPE cells. Consistently, Western blot showed that loss of ARL13B led to upregulation of mesenchymal markers fibronectin and vimentin, accompanied by downregulation of the epithelial marker ZO-1, especially under the treatment of TGF-β1. (Figures 3B–F). These findings suggest that loss of ARL13B promotes EMT in RPE cells.

To gain a view of the global impact of ARL13B on the EMT of RPE cells, we conducted a bulk RNA-seq (Figure 4A). The volcano plots showed that compared to scramble controls, ARL13B knockdown cells had 85 upregulated and 31 downregulated differentially expressed genes (Figure 4B). The number of upregulated and downregulated differentially expressed genes increased to 265 and 156, respectively, upon TGF-β1 treatment (Figure 4C). GSEA analysis of the differentially expressed genes showed EMT as the top significantly enriched processes in ARL13B knockdown cells with or without TGF-β1 treatment (Figures 4D,E). The GSEA further confirmed a global activation of EMT-related genes in ARL13B KD cells, especially with TGF-β1 (Figures 4F,G). These findings indicate that loss of ARL13B enhances the EMT in RPE cells.

The formation of proliferative membranes relies on RPE cell proliferation (Chiba, 2014). To investigate the role of ARL13B in RPE proliferation, we conducted an EdU assay. The results showed that ARL13B knockdown alone increased the number of EdU-labeled cells, and this effect was further enhanced upon TGF-β1 treatment (Figures 5A,B), indicating that ARL13B loss promotes RPE cell proliferation both in the absence and presence of TGF-β1. We further examined the expression of cyclin D1, a key regulatory protein controlling the transition from the G1 phase to the S phase. The Western blot revealed that ARL13B knockdown increased cyclin D1 expression, with or without TGF-β1 treatment (Figures 5C,D). Those results suggest that loss of ARL13B promotes RPE proliferation at least partially by upregulating cyclin D1 expression.

To investigate the role of ARL13B in RPE migration, we performed transwell and scratch assays. The transwell assay showed that ARL13B knockdown enhanced RPE cell migration, particularly upon TGF-β1 treatment (Figures 6A,B). Consistently, the scratch assay showed that ARL13B knockdown accelerated wound closure compared to the scramble group, both in the absence and presence of TGF-β1 (Figures 6C,D).

Cell migration relies heavily on a dynamic rearrangement of the actin cytoskeleton (Thapa et al., 2023). To investigate the cytoskeletal dynamics, we visualized the F-actin (filamentous actin) using phalloidin staining. The immunofluorescence staining revealed increased phalloidin upon ARL13B knockdown, with more pronounced changes in TGF-β1 treatment (Figures 7A,C). This cytoskeletal remodeling was correlated with the shortening of cilia length (Figure 7A,B). These data indicate that the loss of ARL13B enhances the migration of RPE cells.

Previously, we have shown that loss of ARL13B promotes EMT, proliferation, and migration in vitro. To further investigate the role of ARL13B in PVR formation, we injected ARL13B KD and scrambled ARPE-19 cells into the vitreous of mice as described in our previous study (Ma et al., 2023). We then assessed PVR progression via fundus photography with Fastenberg classification on days 3, 7, and 14. On day 3, both ARL13B KD and scrambled groups showed similar cell distribution across the retinal surface (Figures 8A,B). On day 7, retinal detachment was observed in both groups, but more severe in the ARL13B KD group, the majority of which progressed to stage 3, while scrambled controls remained at stage 2 (Figures 8A,C). On day 14, the ARL13B KD group showed more extensive detachment, reaching stages 4 or 5, compared to stages 3 or 4 for the majority of scrambled controls. (Figures 8A,D). Histological analysis on day 14 confirmed more severe retinal detachment and increased α-SMA expression in the ARL13B KD group (Figures 8E,F). Together, these findings indicate that loss of ARL13B promotes PVR formation.

To explore the molecular mechanism by which ARL13B regulates EMT, we further analyzed bulk RNA-seq data described earlier in results 3.3, and the GSEA analysis found that ARL13B knockdown significantly upregulated TGF-β1 signaling (Figure 9A). To further investigate the effect of ARL13B on TGF-β1 signaling, we first examined the canonical TGF-β1 signaling pathway by checking phosphorylated Smad3 (pSmad3) levels. As the Western blot results showed, pSmad3 levels increased starting at 5 min after TGF-β1 treatment, with significantly higher expression in the ARL13B KD group, indicating enhanced Smad3 activation (Figures 9B,C). However, there was no significant difference in the ratio of phosphorylated and total Smad3 between the two groups (Figure 9E). Notably, total Smad3 levels were elevated in the ARL13B KD group (Figures 9B,D), suggesting that increased pSmad3 resulted from elevated total Smad3. Consistently, the immunofluorescence staining showed increased nuclear translocation of Smad3 in the ARL13B KD group in both the absence and presence of TGF-β1 (Figures 9F,G). Meanwhile, we also detected the pSmad3 by immunofluorescence staining, the result showed that ARL13B knockdown enhanced nuclear translocation of Smad3 (Supplementary Figure S2). In parallel, we examined the involvement of the non-canonical TGF-β1 signaling pathway, like the ERK pathway. Western blot analysis revealed no significant differences in total ERK and phosphorylated ERK (p-ERK) levels between ARL13B KD and scramble cells, and the p-ERK/ERK ratio remained unchanged (Supplementary Figure S3). These findings suggest that the ERK pathway does not participate in the EMT induced by ARL13B deficiency. To further explore the regulation of ARL13B on Smad3, we performed real-time PCR analysis, which revealed that ARL13B deficiency did not affect Smad3 mRNA expression (Supplementary Figure S4). These results imply that ARL13B may regulate Smad3 levels at the protein translation or stability instead of the transcriptional level. Given that the cilia signaling, such as the Hedgehog pathway, is known to be involved in the EMT (Wang et al., 2024), we next investigated whether the Hedgehog pathway contributes to the ARL13B deficiency–mediated enhancement of EMT. Western blot analysis revealed no significant changes in PTCH2 and SHH protein expression, indicating that ARL13B deficiency regulates EMT independently of cilia-associated signals such as the Hedgehog pathway (Supplementary Figure S5). Taken together, loss of ARL13B may enhance RPE-EMT by upregulating Smad3 expression through post-transcriptional regulation.