LPA1-Induced EMT of Retinal Pigment Epithelial Cells Promotes Subretinal Fibrosis via USP1-Mediated Deubiquitination and Stabilization of ZEB1

Abstract

Purpose

Subretinal fibrosis (SRF) is a vision-threatening complication of neovascular age-related macular degeneration (nAMD), yet effective antifibrotic strategies remain limited. This study investigated the role of lysophosphatidic acid receptor 1 (LPA1), a fibrosis-associated G protein–coupled receptor, in SRF progression and evaluated its therapeutic potential.

Methods

A two-stage laser injury model was used to induce SRF in C57BL/6J mice. LPA1 expression and its regulation of ubiquitin-specific peptidase 1 (USP1) and zinc finger E-box binding homeobox 1 (ZEB1) were assessed by western blotting and immunofluorescence. Retinal pigment epithelial (RPE) cell epithelial–mesenchymal transition (EMT) was evaluated by migration assays and fibrotic marker analysis. ZEB1 and USP1 loss-of-function experiments were performed to define their roles. Molecular docking and co-immunoprecipitation were used to examine USP1-mediated ZEB1 deubiquitination. The antifibrotic effects and the safety of LPA1 inhibition by BMS-986278 were assessed in vivo.

Results

LPA1 was upregulated in laser-induced SRF and was associated with increased fibrotic marker expression. LPA1 overexpression promoted RPE EMT and migration through the USP1/ZEB1 axis. Mechanistically, LPA1 increased USP1 expression, which stabilized ZEB1 and promoted its nuclear translocation by reducing ZEB1 ubiquitination. USP1 inhibition attenuated LPA1-induced EMT and fibrosis, similar to ZEB1 silencing. In vivo, BMS-986278 suppressed EMT and reduced SRF severity, with a stronger inhibitory effect on fibrotic lesions than on neovascular lesions.

Conclusions

LPA1 promotes SRF by driving RPE EMT through USP1-mediated deubiquitination and stabilization of ZEB1. Targeting the LPA1/USP1/ZEB1 axis may provide a promising therapeutic strategy for nAMD-associated SRF.

Age-related macular degeneration (AMD) stands as one of the most consequential causes of irreversible vision loss in elderly individuals worldwide, with global case numbers projected to reach 288 million by 2040.^1–3 The neovascular form of AMD (nAMD) is defined by abnormal choroidal neovascularization (CNV) that breaches Bruch's membrane, resulting in exudation and hemorrhage, ultimately damaging the retinal pigment epithelium (RPE) and photoreceptors.^4,5 Subretinal fibrosis (SRF), arising from aberrant wound repair during disease progression, further disrupts retinal architecture and contributes to permanent visual impairment. The occurrence of SRF in nAMD reflects a clear temporal increase, from 13.0% initially to 37.8% at 1 year.⁶ Although anti-vascular endothelial growth factor (VEGF) therapy effectively reduces vascular leakage in the majority of patients, a substantial proportion still progress to develop SRF.⁷ Pharmacological options specifically addressing this condition remain critically limited.

SRF is primarily characterized by the conversion of neovascular tissue into fibrovascular membranes.^8,9 During this process, myofibroblasts derived from cellular phenotypic transition secrete extracellular matrix (ECM), thereby accelerating fibrovascular membrane formation.⁴ Consequently, preventing the transition of lesion-associated cells, particularly RPE cells, into myofibroblast-like cells offers a promising strategy to attenuate SRF. Epithelial–mesenchymal transition (EMT) is a crucial event in organ fibrosis, and its occurrence in RPE cells is also important for SRF pathogenesis.^10–13 Retinal injury can trigger EMT in RPE cells that may present fibroblast-like features, including upregulated fibronectin and α-smooth muscle actin (α-SMA) abundance, thereby facilitating fibrotic remodeling.^14,15 Excessive ECM deposition by RPE cells contributes to pathological scar formation, causing photoreceptor dysfunction and, ultimately, irreversible vision loss. Accumulating evidence suggests regulation of this process by multiple molecular pathways, involving the TGF-β/Smad pathway and the phosphoinositide 3-kinase (PI3K)/Akt axis,^14,16 and these pathways are also considered key signaling pathways in organ fibrosis.^17–19 However, given the complexity of the molecular regulatory networks governing EMT in RPE cells, the underlying key mechanisms remain incompletely understood. Therefore, further investigation of RPE cell EMT is required to elucidate the pathogenesis of SRF and identify candidate targets for intervention.

Lysophosphatidic acid receptor 1 (LPA1) is an LPA-responsive G protein–coupled receptor (GPCR) implicated in fibrotic remodeling.²⁰ Elevated LPA1 expression has been associated with enhanced fibroblast recruitment and the progression of pulmonary fibrosis.²¹ Similarly, in Crohn's disease, LPA1 signaling promotes intestinal fibrotic remodeling through the TGF-β/Smad and mitogen-activated protein kinase kinase (MEK)/Erk axes.²² In bleomycin-induced pulmonary fibrosis, LPA1 is concomitantly upregulated with EMT markers during the intermediate and late stages of fibrosis, suggesting that LPA1 may participate in and promote EMT-associated fibrotic progression.²³ Notably, admilparant (BMS-986278), a selective LPA1 antagonist, has advanced to phase 2 clinical trials in pulmonary fibrosis.²⁴ The LPA1 signaling axis is increasingly recognized as an important mediator in ocular diseases, as well. Additionally, the LPA/LPA1 pathway has been implicated in inflammation, neovascularization, and fibrosis of proliferative vitreoretinal disorders.²⁵ Vitreous fluid collected from eyes with proliferative diabetic retinopathy exhibited significantly elevated LPA levels compared with those in non-diabetic controls.^26,27 In glaucoma, LPA1 has been linked to trabecular meshwork dysfunction and ECM deposition, enabling fibrotic remodeling.²⁸ Collectively, these findings indicate that LPA1 activation and upregulation may serve as upstream drivers of fibrotic remodeling in retinal disease. Nevertheless, the role of the LPA/LPA1 axis in AMD—particularly in fibrosis-related pathology—remains poorly characterized and warrants subsequent exploration. More importantly, although LPA1 has been implicated in systemic fibrotic diseases and several ocular pathological conditions, few studies have clarified whether LPA1 directly promotes RPE EMT during SRF progression, nor has the downstream mechanism linking LPA1 activation to fibrotic remodeling been established.

This study identified LPA1 as an important mediator of EMT and SRF in RPE cells. Under fibrotic conditions, LPA1 upregulation promoted EMT by enhancing ubiquitin-specific peptidase 1 (USP1)-mediated deubiquitination and stabilization of zinc finger E-box binding homeobox 1 (ZEB1), thereby facilitating its nuclear translocation. Pharmacological inhibition of LPA1 suppressed EMT in RPE cells and alleviated SRF in mice, supporting the therapeutic potential of targeting the LPA1/USP1/ZEB1 signaling axis.

Methods

Public Transcriptomic Data Analysis

To investigate transcriptomic signatures in the macular RPE–choroid complex, publicly available data were retrieved from the Gene Expression Omnibus (GEO) repository (accession GSE29801). Only macular RPE–choroid samples with matched gene-level expression data were retained for downstream analysis. In total, 91 macular RPE–choroid samples were included. In the transcriptomic analysis, LPA1 was represented by its encoding gene, LPAR1. Samples were therefore stratified into low-, middle-, and high-tertile groups according to LPAR1 expression, containing 31, 30, and 30 samples, respectively. Next, Gene Set Variation Analysis (GSVA) was conducted to assess the relative enrichment of fibrosis-related pathways among the tertile groups. To further quantify overall fibrotic activity, a composite fibrosis score was calculated based on three fibrosis-related GSVA signatures: HALLMARK_TGF_BETA_SIGNALING, REACTOME_EXTRACELLULAR_MATRIX_ORGANIZATION, and REACTOME_DEGRADATION_OF_THE_EXTRACELLULAR_MATRIX. For each pathway, raw GSVA scores were standardized across samples using z-score transformation, and the composite fibrosis score for each sample was defined as the mean of the three standardized pathway scores.

Cell Culture and Treatments

ARPE-19 cell line (GNHu45; Chinese Academy of Sciences Cell Bank, Shanghai, China) was propagated in a basal growth medium comprised of Gibco Dulbecco's Modified Eagle's Medium/Nutrient Mixture F-12 (DMEM/F-12; 1:1 v/v; A4192001; Thermo Fisher Scientific, Waltham, MA, USA), to which 10% fetal bovine serum (FBS; 12003C, Sigma-Aldrich, St. Louis, MO, USA) and a Gibco penicillin–streptomycin antibiotic solution (100 U/mL and 100 µg/mL; 15140122; Thermo Fisher Scientific) were added. Normoxic culture conditions were maintained at 37°C and 5% CO₂ in a humidified atmosphere. To establish a hypoxic state, cells were transferred to an airtight chamber continuously perfused with 1% O₂, 5% CO₂, and 94% N₂ for 24 hours. During hypoxia experiments, indicated agents were added at the onset of hypoxia and maintained unless otherwise specified.

Transwell Migration Assay

RPE cells were introduced into the upper compartments of Transwell inserts and allowed to migrate through the porous membrane at 37°C for 24 hours. Cells at the lower surface were subsequently fixed with 4% paraformaldehyde (PFA; C0121; Beyotime, Shanghai, China), visualized by crystal violet staining, and quantified microscopically.

Scratch Wound Healing Assay

RPE cells were cultivated in culture plates until the monolayer reached near-confluency. The cell layer was then mechanically disrupted with a sterile pipette tip to introduce a uniform wound gap. Following the removal of detached cellular debris by PBS washing, the cultures were replenished with fresh medium, which was prepared as described in the cell culture section. Wound images were recorded at the indicated time points, and the residual gap area was measured.

ARPE-19 Cell Transfection

ARPE-19 cells were prepared for transfection upon reaching approximately 60% to 70% confluence in culture plates. Sequentially, the cells were first transfected with either the LPA1 overexpression construct or the corresponding empty vector using Lipofectamine 3000 reagent (L3000015; Invitrogen, Carlsbad, CA, USA). After 24 hours, the culture medium was replaced with fresh complete medium, and cells were subjected to a second transfection with the indicated gene-specific small interfering RNAs (siRNAs) or a negative control siRNA. At the designated time points, samples were collected for downstream analyses. Overexpression and knockdown efficiencies were verified by western blotting (WB) and/or quantitative real-time PCR (RT-qPCR). The sequences of the LPA1 expression plasmid and siRNAs are listed in Supplementary Table S1.

RNA Extraction and RT-qPCR Analysis

Total cellular RNA was isolated using TRIzol LS Reagent (10296028; Thermo Fisher Scientific). For first-strand cDNA synthesis, the High-Capacity cDNA Reverse Transcription Kit (4368814; Applied Biosystems, Waltham, MA, USA) was employed. RT-qPCR was performed using SYBR Green PCR Master Mix (4309155; Applied Biosystems) on a 7500 Real-Time PCR System (Applied Biosystems). All primer sequences used for amplification are provided in Supplementary Table S2.

Co-Immunoprecipitation

Cell lysates were prepared in pre-chilled buffer containing 50-mM Tris-HCl (pH 7.4), 150-mM NaCl, 1-mM EDTA, 0.5% NP-40, and a protease inhibitor cocktail. Following centrifugation, clarified lysates were incubated overnight at 4°C with anti-ZEB1 (ab203829; Abcam, Cambridge, UK), anti-USP1 (14346-1-AP; Proteintech, Rosemont, IL, USA), or a non-specific control IgG antibody. Immune complexes were subsequently captured using Protein A/G agarose beads at 4°C for 2 hours. After extensive washing, the bead-bound proteins were eluted by heat denaturation in sodium dodecyl sulfate (SDS) sample buffer. To monitor protein interaction, precipitated complexes were analyzed by immunoblotting for ZEB1 and USP1. For ubiquitination assays, cells were pretreated with MG132 before harvest to inhibit proteasomal degradation and enrich ubiquitinated proteins. ZEB1 was then immunoprecipitated, and ubiquitinated ZEB1 was detected by immunoblotting with an anti-ubiquitin antibody (10201-2-AP; Proteintech).

Western Blotting

Proteins were isolated from ARPE-19 cells or mouse tissue lysates using pre-chilled radioimmunoprecipitation assay (RIPA) lysis buffer containing a cocktail of protease and phosphatase inhibitors. Nuclear fractions were extracted using a nuclear/cytoplasmic extraction kit (P0027; Beyotime). Protein concentrations were estimated by the bicinchoninic acid (BCA) method (P0010; Beyotime). Equal amounts of protein were resolved on SDS–polyacrylamide gel electrophoresis (PAGE) and subsequently electroblotted onto polyvinylidene fluoride (PVDF) membranes. Following blocking, membranes were treated overnight at 4°C with specific primary antibodies against fibronectin 1 (ab2413; Abcam), LPA1 (ab23698; Abcam), collagen I (ab270993; Abcam), ZEB1 (ab203829; Abcam), α-SMA (14395-1-AP; Proteintech), USP1 (14346-1-AP; Proteintech), GAPDH (60004-1-Ig; Proteintech), H3 (17168-1-AP; Proteintech), and E-cadherin (14472; Cell Signaling Technology, Danvers, MA, USA). Membranes were then incubated with horseradish peroxidase (HRP)-linked anti-rabbit IgG (SA00001-2; Proteintech) or anti-mouse IgG (SA00001-1; Proteintech). Immunoreactive signals were detected using an enhanced chemiluminescence (ECL) reagent (ab133406; Abcam), and band intensities were quantified by densitometric analysis in ImageJ (National Institutes of Health, Bethesda, MD, USA).

Molecular Docking

Protein–protein docking analysis was carried out on the Global RAnge Molecular Matching (GRAMM) platform under rigid-body settings. Three-dimensional structural models of the target proteins were obtained from publicly available repositories and uploaded to GRAMM with default parameters. The highest-scoring model was selected for subsequent evaluation. Interfacial features were then assessed using PDBePISA and rendered in PyMOL 3.1.

Animals

C57BL/6J mice, 6 to 8 weeks old, were sourced from Soochow University (Suzhou, China) and housed in a specific pathogen-free facility under a 12-hour light/dark schedule with ad libitum access to food and water. All animal procedures received prior approval from the Animal Ethics Committee of Shanghai East Hospital, Tongji University School of Medicine, and complied with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Subretinal Fibrosis Induction

A two-step laser injury model was adopted to establish SRF, as previously documented.²⁹ Mice were randomly assigned to the indicated treatment groups. Outcome assessment and image quantification were performed by investigators blinded to the treatment allocation. Laser burns were generated using an image-guided delivery system (Micron IV; Phoenix-Micron, Bend, OR, USA) with the following parameter settings: 647-nm wavelength, 50-µm spot size, 0.1-second exposure time, and 160-mW power. Four laser spots were created in the peripapillary area of each eye under slit-lamp visualization, with a coverslip serving as the contact interface. Bruch's membrane disruption was verified by the appearance of a subretinal bubble in the absence of marked hemorrhage. Lesions were excluded if laser application failed to produce an immediate bubble, if extensive hemorrhage obscured the lesion, or if adjacent lesions coalesced and could not be independently quantified. One day after the initial laser treatment, mice in the indicated groups received intravitreal administration of UCM-05194 (HY-167862A; MedChemExpress, Monmouth Junction, NJ, USA), BMS-986278 (HY-139853; MedChemExpress), or aflibercept (EYLEA; Bayer, Leverkusen, Germany) at a volume of 1 µL per eye, corresponding to a final dose of 40 µg per eye. A second laser treatment was applied to the same sites 7 days later to promote fibrotic lesion development. Lesions exhibiting hemorrhage or coalescence were omitted from subsequent evaluation. Ocular tissues were harvested on day 21, corresponding to 14 days after the second laser injury, for downstream analyses.

Immunofluorescence Staining

Following fixation, cells were permeabilized with 0.1% Triton X-100 in PBS and blocked with 5% BSA before overnight incubation with the indicated primary antibodies at 4°C. After rinsing, samples were incubated with Alexa Fluor 488–conjugated goat anti-rabbit IgG secondary antibody (A-11008; Invitrogen) at room temperature for 1 hour. Nuclear staining was performed with 4′,6-diamidino-2-phenylindole (DAPI; Beyotime), and fluorescence signals were recorded using a Leica fluorescence microscope (Leica Microsystems, Wetzlar, Germany).

Choroidal Flatmount Immunofluorescence

Mouse eyes were enucleated and immediately immersed in 4% PFA at 4°C. Following careful dissection to remove the anterior ocular segment and neural retina, the isolated RPE–choroid complex was processed through PBS washing and Triton X-100 permeabilization. Next, tissues were incubated with the specified primary antibodies at 4°C overnight and sequentially with fluorescent secondary antibodies at room temperature for 2 hours. The RPE–choroid complex was flatmounted onto glass slides and visualized under a scanning laser confocal microscope (Carl Zeiss Microscopy, Jena, Germany).

Fundus Fluorescein Angiography and Indocyanine Green Angiography

Fundus fluorescein angiography (FFA) and indocyanine green angiography (ICGA) were conducted 21 days after laser photocoagulation. Following anesthesia and pupil dilation, mice received intraperitoneal injections of fluorescein sodium (Fluorescite; Alcon, Fort Worth, TX, USA) and indocyanine green (Ruidu; Dandong Yichuang Pharmaceutical Co., Dandong City, China). FFA and ICGA images were acquired at 5 and 10 minutes after dye administration using a retina angiograph (Heidelberg Engineering, Heidelberg, Germany). Fluorescein leakage was graded based on previously established criteria,³⁰ and CNV size was quantified from ICGA images in ImageJ. Image quantification was performed in ImageJ using identical acquisition settings, background subtraction, consistent thresholds, and masked analysis.

Hematoxylin and Eosin Staining

Mouse ocular tissues and major organs were immersed in 4% PFA overnight, followed by dehydration through a graded ethanol series and paraffin embedding. Sections of 5-µm thickness were cut, deparaffinized, and subjected to hematoxylin and eosin (H&E) staining. Histologic images were acquired by light microscopy.

Statistical Analysis

All data are presented as mean ± SD. Each experiment was performed with at least three independent replicates (n ≥ 3). Statistical analyses were performed employing Prism 9.0 (GraphPad Software, Boston, MA, USA). Before parametric analyses, normality and homogeneity of variance were assessed using the Shapiro–Wilk test and Brown–Forsythe or Levene's test, respectively. Parametric tests, including unpaired two-tailed Student's t-test or one-way ANOVA, were applied when these assumptions were met; otherwise, corresponding nonparametric tests were used. Ordinal variables were analyzed using nonparametric tests. P < 0.05 was considered statistically significant.

Results

LPA1 Expression Is Elevated in Cell and Mouse Models of SRF

Previous studies have implicated LPA1 in fibrotic regulation across diverse experimental settings.²¹ Accordingly, LPA1 expression and fibrotic marker levels were evaluated in hypoxia-exposed ARPE-19 cells. RT-qPCR analysis revealed increased transcript levels of α-SMA, collagen I, and fibronectin 1; WB confirmed significant upregulation of these proteins, indicating hypoxia-induced fibrosis in ARPE-19 cells (Figs. 1A–C). Subsequent assessment of LPA1 expression reflected an appreciable enhancement in both mRNA and protein abundance under hypoxic conditions, suggesting a likely association between LPA1 upregulation and fibrosis (Figs. 1A–C). To further examine the translational relevance of these findings, human samples from the GEO database were analyzed to explore the relationship between LPA1 expression and fibrosis-related transcriptional signatures. Samples with higher LPA1 expression showed overall increased enrichment of fibrosis-related pathways, as visualized in the GSVA heatmap (Fig. 1D). Consistently, LPAR1 expression was positively correlated with the composite fibrosis score determined by Spearman correlation analysis (ρ = 0.229, P = 0.0291), suggesting that elevated LPAR1 expression is associated with increased fibrotic transcriptional activity in human macular RPE–choroid samples (Fig. 1E). These findings were further corroborated in a two-step laser-induced mouse model of SRF, with tissue samples harvested on day 21 (Fig. 1F). Immunofluorescence (IF) staining detected prominent α-SMA–positive fibrotic lesions, inferring successful establishment of the fibrotic model. Notably, LPA1 exhibited marked co-localization with α-SMA within these lesions, again emphasizing its key involvement in fibrotic remodeling (Fig. 1G). RT-qPCR analysis indicated increased transcript levels of α-SMA, collagen I, fibronectin 1, and LPA1 in the RPE–choroid complex, a trend concordantly observed at the protein level by WB (Figs. 1H–J). Collectively, these data suggest that LPA1 expression is substantially upregulated in the context of SRF.

Figure 1.

LPA1 expression is elevated in cell and mouse models of SRF. (A) Relative mRNA expression levels of fibronectin 1, collagen I, α-SMA, and LPA1 in ARPE-19 cells under normoxic and hypoxic conditions, as determined by RT-qPCR. (B, C) WB analysis of fibronectin 1, collagen I, α-SMA, and LPA1 protein expression in ARPE-19 cells exposed to normoxia or hypoxia. (D) GSVA heatmap illustrating fibrosis-associated pathway enrichment in human macular RPE–choroid samples stratified by LPA1 expression tertiles. (E) Spearman correlation analysis between LPA1 expression and the composite fibrosis score across human macular RPE–choroid samples. (F) Schematic overview of the two-step laser-induced mouse model of SRF. (G) IF staining images for LPA1 and α-SMA in the mouse RPE–choroid complex at day 21 post-laser treatment. Scale bar: 100 µm. (H) Relative mRNA expression of fibronectin 1, collagen I, α-SMA, and LPA1 in the mouse RPE–choroid complex isolated from normal control mice and laser-induced SRF mice at day 21. (I, J) Protein expression levels of fibronectin 1, collagen I, α-SMA, and LPA1 in the mouse RPE–choroid complex. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

LPA1 Promotes EMT in RPE Cells

EMT in RPE cells is regarded as an essential pathological event in SRF progression.^31,32 Because LPA1 was upregulated in fibrotic RPE cells, we further investigated whether increased LPA1 expression could promote EMT in RPE cells. LPA1-overexpressing ARPE-19 cells were generated, verified by both RT-qPCR and WB (Figs. 2A–C). Following LPA1 overexpression, ARPE-19 cells lost their characteristic cobblestone-like epithelial morphology and acquired an elongated, spindle-shaped mesenchymal appearance. Scratch-wound healing assays demonstrated that LPA1 overexpression significantly accelerated wound closure in ARPE-19 cells (Figs. 2D, 2E). Similarly, Transwell migration assays confirmed the enhanced migratory ability of LPA1-overexpressing cells relative to control cells (Figs. 2F, 2G). WB analysis of EMT-linked protein expression revealed that both LPA1 overexpression and hypoxia led to a decline in E-cadherin expression while upregulating fibrotic markers, including α-SMA, collagen I, and fibronectin 1 (Figs. 2H, 2I). Additionally, LPA1 protein levels were increased in both the hypoxia and LPA1-overexpression groups, with significant elevation observed in the combined group category (Fig. 2J). Notably, EMT-associated modifications were more pronounced under combined hypoxia and LPA1 overexpression, suggesting that LPA1 levels potentiate hypoxia-driven fibrotic transition in RPE cells (Figs. 2H, 2I). To extend these findings to an in vivo setting, the LPA1 agonist UCM-05194 was administered intravitreally in a mouse model of SRF. IF staining indicated that UCM-05194-treated mice developed larger fibrotic lesions than vehicle-treated control mice, suggesting that pharmacological LPA1 activation exacerbates SRF in vivo (Figs. 2K, 2L). Together, these data demonstrate the crucial contribution of LPA1 upregulation in promoting EMT in RPE cells and aggravating SRF.

Figure 2.

LPA1 promotes EMT in RPE cells. (A) Relative LPA1 mRNA levels in ARPE-19 cells transfected with an empty vector or the LPA1 overexpression plasmid, as measured by RT-qPCR. (B, C) WB analysis verifying LPA1 overexpression in ARPE-19 cells. (D, E) Scratch wound healing assays in ARPE-19 cells performed under the indicated conditions, photographed at 0 and 24 hours. (F, G) Transwell migration assays in ARPE-19 cells under the indicated conditions. Scale bar: 50 µm. (H–J) Protein levels of E-cadherin, fibronectin 1, collagen I, α-SMA, and LPA1 in ARPE-19 cells under the designated conditions. (K, L) IF staining for α-SMA in the RPE–choroid complex from control and UCM-05194–treated mice in the laser-induced SRF model. Scale bar: 100 µm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

LPA1 Promotes EMT by Inducing ZEB1 Nuclear Translocation

ZEB1 is a well-documented EMT-related transcription factor that has been linked to fibrotic remodeling covering a range of diseases.^33–35 Accumulating evidence suggests that ZEB1 is regulated by LPA1 signaling³⁶; however, whether this modulatory axis operates in RPE cells during SRF remains unclear. Therefore, the contribution of ZEB1 to LPA1-driven profibrotic responses was investigated in RPE cells. WB analysis revealed enhanced ZEB1 protein levels in ARPE-19 cells under both hypoxia and LPA1 overexpression (Figs. 3A, 3B). Because ZEB1 exerts transcriptional activity primarily in the nucleus, its subcellular distribution was examined. IF imaging detected increased nuclear localization of ZEB1 in the response to LPA1-overexpression, hypoxia, and combined groups (Figs. 3C, 3D). Consistently, nuclear-cytoplasmic fractionation followed by WB confirmed a concomitant rise in nuclear ZEB1 abundance and reduction in cytoplasmic ZEB1 levels under these conditions (Figs. 3E, 3F). These results indicate that LPA1 overexpression promotes ZEB1 nuclear translocation. To delineate the functional role of ZEB1 in LPA1-induced EMT, ZEB1 was selectively knocked down in LPA1-overexpressing RPE cells (Figs. 3G, 3H). ZEB1 knockdown markedly attenuated the EMT phenotype, highlighting that ZEB1 is required for LPA1-mediated EMT in RPE cells (Figs. 3I, 3J). Overall, these findings establish a critical role for ZEB1 in directing EMT in RPE cells driven by LPA1 overexpression and suggest its nuclear localization as a key contributor to the profibrotic phenotype.

Figure 3.

LPA1 promotes EMT by inducing ZEB1 nuclear translocation. (A, B) ZEB1 protein expression levels in ARPE-19 cells under different treatments. (C, D) IF staining depicting the subcellular localization of ZEB1 in ARPE-19 cells in response to the designated conditions. Nuclei counterstaining was achieved with DAPI. Scale bar: 20 µm. (E, F) Representative WB images and quantification of cytoplasmic and nuclear fractions illustrating ZEB1 distribution in ARPE-19 cells under different treatment conditions. (G, H) WB analysis confirming efficient ZEB1 knockdown in LPA1-overexpressing ARPE-19 cells. (I, J) Quantification of E-cadherin, fibronectin 1, collagen I, and α-SMA in control, LPA1-overexpressing, and ZEB1 knockdown (siZEB1) ARPE-19 cells. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

LPA1 Enhances ZEB1 Stability Through USP1-Mediated Deubiquitination

Given that LPA1 signaling has been linked to ubiquitin-dependent regulation of protein stability and that ZEB1 is reported to be controlled by ubiquitination,^37,38 we investigated whether ZEB1 is degraded through the ubiquitin–proteasome pathway in RPE cells. Cycloheximide chase assays revealed a relatively short half-life of ZEB1 in control ARPE-19 cells, whereas hypoxia or LPA1 overexpression delayed its degradation, prolonging the half-life (Figs. 4A, 4B). This stabilizing effect was most evident in the combined hypoxia and LPA1-overexpression group, indicating that both hypoxia and LPA1 upregulation cooperatively enhance ZEB1 protein stability in RPE cells (Figs. 4A, 4B). Assessing the ubiquitination status of ZEB1 further demonstrated that LPA1 overexpression significantly reduced ZEB1 ubiquitination in RPE cells, thereby highlighting LPA1-mediated stabilization of ZEB1 by suppressing its ubiquitin-mediated degradation (Fig. 4C). Previous studies have documented the role of USP1 in modulating ZEB1 ubiquitination.³⁹ Here, USP1 expression was markedly upregulated following LPA1 overexpression at both the transcript and protein levels (Figs. 4D–F). The role of USP1 was examined from two complementary perspectives: structural interaction and functional regulation. Structurally, molecular docking analysis predicted a favorable binding interface between USP1 and ZEB1 (Fig. 4G). This prediction was further validated by co-immunoprecipitation, confirming interaction between USP1 and ZEB1 in RPE cells (Fig. 4H). At the functional level, a USP1-knockdown RPE cell model was successfully established, as verified by WB (Figs. 4I, 4J). USP1 knockdown partially restored ZEB1 ubiquitination suppressed by LPA1 overexpression; pharmacologic inhibition of USP1 with ML323 yielded a comparable effect, inferring that USP1 is required for LPA1-induced ZEB1 deubiquitination (Fig. 4K). Importantly, both USP1 knockdown and ML323 treatment attenuated the increase in ZEB1 protein levels stimulated by LPA1 overexpression (Figs. 4L, 4M). In parallel, USP1 inhibition also blunted LPA1-driven EMT phenotype (Figs. 4N, 4O). Taken together, these results support USP1 as a critical intermediary linking LPA1 signaling to ZEB1 stabilization and EMT occurrence in RPE cells.

Figure 4.

LPA1 enhances ZEB1 stability through USP1-mediated deubiquitination. (A, B) Cycloheximide chase assays depicting ZEB1 protein degradation kinetics in ARPE-19 cells under the indicated conditions at 0, 2, 4, and 8 hours, as determined by WB. (C) Ubiquitination assay showing ZEB1 ubiquitination levels in control and LPA1-overexpressing ARPE-19 cells after MG132 treatment. ZEB1 was immunoprecipitated and immunoblotted with an anti-ubiquitin antibody. (D, E) USP1 protein levels in ARPE-19 cells under different treatments. (F) Relative USP1 mRNA expression in ARPE-19 cells under the indicated conditions, as estimated by RT-qPCR. (G) Predicted docking model of the interaction between USP1 and ZEB1. (H) Co-immunoprecipitation analysis confirming the interaction between USP1 and ZEB1 in ARPE-19 cells. (I, J) WB analysis confirmed USP1 knockdown in LPA1-overexpressing ARPE-19 cells transfected with three independent siRNAs. (K) Ubiquitination assay illustrating the effects of USP1 knockdown or ML323 treatment on ZEB1 ubiquitination levels in LPA1-overexpressing ARPE-19 cells. (L, M) ZEB1 protein expression in control, LPA1-overexpressing, siUSP1-treated, and ML323-treated ARPE-19 cells. (N, O) Protein expression levels of E-cadherin, fibronectin 1, collagen I, and α-SMA in ARPE-19 cells under the indicated conditions. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Targeting LPA1 Attenuates EMT in RPE Cells

To further evaluate the functional and therapeutic relevance of LPA1, an LPA1-knockdown hypoxic RPE model was constructed (Figs. 5A, 5B). Simultaneously, hypoxic RPE cells were exposed to the selective LPA1 inhibitor BMS-986278. Both interventions resulted in reduced USP1 expression compared to that observed in the hypoxia group (Figs. 5C, 5D). ZEB1 subcellular localization was subsequently examined by IF, revealing that both LPA1 knockdown and BMS-986278 treatment reduced nuclear accumulation of ZEB1 in hypoxic RPE cells (Figs. 5E, 5F). Consistently, nuclear–cytoplasmic fractionation and WB demonstrated a decline in nuclear ZEB1 abundance, alongside a lower nuclear/cytoplasmic index following either genetic or pharmacologic LPA1 inhibition (Figs. 5G, 5H). WB analysis further indicated that both interventions partly reversed the hypoxia-induced EMT phenotype, as evidenced by E-cadherin upregulation and concomitant decreases in fibronectin 1, collagen I, and α-SMA levels (Figs. 5I, 5J). Collectively, these results substantiate that LPA1 inhibition, by either genetic knockdown or pharmacologic blockade, suppresses hypoxia-driven EMT in RPE cells and is accompanied by lower USP1 expression and attenuated nuclear translocation of ZEB1.

Figure 5.

Targeting LPA1 attenuates EMT in RPE cells. (A, B) WB analysis confirming LPA1 knockdown in ARPE-19 cells under hypoxic conditions using three independent siRNAs. (C, D) USP1 protein expression in ARPE-19 cells under different treatments. (E, F) IF images of ZEB1 subcellular localization in ARPE-19 cells under the indicated conditions. Scale bar: 20 µm. (G) Cytoplasmic and nuclear fractions depicting ZEB1 distribution in ARPE-19 cells under different treatment conditions. (H) Quantification of the nuclear/cytoplasmic index of ZEB1. (I, J) Protein expression profiles of E-cadherin, fibronectin 1, collagen I, and α-SMA in ARPE-19 cells under the indicated conditions. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

BMS-986278 Attenuates SRF In Vivo

To further assess the therapeutic efficacy of BMS-986278 in vivo, a two-step laser-induced SRF model was established, and the effects of subsequent interventions were determined. Mice received a single intravitreal injection on day 1 after the initial laser treatment. On day 21, vascular leakage and CNV lesion area were partially reduced post-intravitreal BMS-986278 administration, as determined by FFA and ICGA, respectively. Compared with the control group, aflibercept produced a stronger inhibitory effect. Notably, combined treatment of BMS-986278 and aflibercept resulted in the most pronounced suppression of neovascularization (Figs. 6A–C). Consistent with these findings, choroidal flatmount IF detected a significant lowering in isolectin B4 (IB4)-positive neovascular lesions after BMS-986278 treatment, and the combination treatment further enhanced this effect (Figs. 6D, 6E). The effect of LPA1 inhibition on SRF was examined by IF staining, which revealed that BMS-986278 markedly suppressed α-SMA–positive fibrotic lesions and exerted a stronger antifibrotic effect than aflibercept alone; the combination treatment shrank the fibrotic area to a greater extent, although the additional downsizing did not reach statistical significance (Figs. 6D, 6F). In parallel, WB analysis demonstrated marked downregulation of fibrotic markers following BMS-986278 intervention, with the combination treatment yielding an elevated overall inhibitory effect (Figs. 6G, 6H). Additionally, histologic examination of the eye and major organs—including the heart, liver, spleen, lung, and kidney—indicated no evident toxic or pathological alterations, supporting the in vivo safety profile of BMS-986278 (Fig. 7). Together, these results point to an inhibitory effect of BMS-986278 on SRF, and its combination with aflibercept showed a trend toward broader suppression of neovascular and fibrotic changes, supporting a therapeutic strategy for clinical translation.

Figure 6.

BMS-986278 attenuates SRF in vivo. (A) FFA and ICGA images from normal control and treated mice in the indicated groups at day 21 post-CNV induction. (B) Quantification of fluorescein leakage grades based on FFA images in the indicated CNV groups. (C) Quantification of CNV lesion area based on ICGA images depicted in A. (D) IF images of choroidal flatmounts co-stained for IB4 and α-SMA in the indicated groups. Scale bar: 100 µm. (E, F) Quantification of the IB4-positive (E) and α-SMA–positive (F) lesion areas shown in D. (G, H) WB analysis of the protein levels of fibronectin 1, collagen I, and α-SMA in the RPE–choroid complex from the indicated treatment groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 7.

BMS-986278 exhibits no overt histologic toxicity in vivo. (A–F) Representative H&E staining images of the retina (A), heart (B), liver (C), spleen (D), lung (E), and kidney (F) from mice in the indicated treatment groups. No obvious histopathologic abnormalities were identified in ocular or major organ sections after treatment with BMS-986278, aflibercept, or their combination.

Discussion

As a prominent cause of vision loss in the elderly population, AMD—particularly nAMD—accounts for the preponderance of severe blindness cases worldwide.^2,40 Although anti-VEGF therapy has improved nAMD management, SRF remains a frequent end-stage complication that culminates in irreversible visual impairment, underscoring the critical need to precisely define its underlying pathogenic mechanisms.^7,41 The present study addressed this gap by identifying a possible LPA1/USP1/ZEB1 regulatory axis that promotes RPE-associated fibrotic remodeling through USP1-mediated deubiquitination and stabilization of ZEB1, thereby highlighting a post-translational mechanism of EMT regulation. Moreover, BMS-986278, an LPA1 antagonist, markedly mitigated these profibrotic alterations in both cellular and animal models. Collectively, these results suggest that the LPA1/USP1/ZEB1 cascade is an important mechanistic pathway in SRF pathogenesis and highlight LPA1 as a candidate therapeutic target for nAMD-associated fibrosis (Fig. 8).

Figure 8.

Schematic model of LPA1-mediated SRF. Under hypoxic conditions, LPA1 is transcriptionally upregulated in RPE cells and promotes EMT. Mechanistically, LPA1 augments USP1 expression, which in turn enhances ZEB1 deubiquitination, stabilizes ZEB1 protein, and facilitates its nuclear translocation and transcriptional activity. These molecular events drive the upregulation of fibrosis-related markers and contribute to SRF. Pharmacologic inhibition of LPA1 with BMS-986278 suppressed this signaling cascade and attenuated both EMT and SRF progression.

LPA1, a canonical LPA-responsive receptor within the GPCR family, mediates diverse lipid-driven cellular responses and has been implicated in inflammation, fibrosis, neural injury, and cancer.^42–44 More specifically, accumulating evidence has linked the LPA/LPA1 axis to key profibrotic events such as inflammatory signaling, fibroblast recruitment, tissue remodeling, and EMT, thereby supporting LPA1 as a potential antifibrotic target.^43,45,46 Building on these observations, the putative role of LPA1 in ocular fibrosis was explored by analyzing transcriptomic data derived from human RPE–choroid complexes. GSVA showed relatively higher enrichment of fibrosis-associated pathways in samples with elevated LPAR1 expression, whereas Spearman correlation analysis revealed a modest positive association between LPAR1 expression and the composite fibrosis score, suggesting that LPAR1 may be one of multiple factors associated with fibrotic activity. Aligning with these findings, LPA1 expression was also increased in hypoxia-exposed ARPE-19 cells and in a two-step laser-induced mouse model of SRF, with simultaneous upregulation of fibrotic markers. Functionally, LPA1 overexpression stimulated EMT-associated changes in RPE cells, whereas LPA1 inhibition attenuated fibrotic remodeling in vitro and in vivo. These findings support a functional contribution of LPA1 to EMT-associated SRF progression, although they do not establish LPA1 upregulation as the sole initiating cause of fibrosis.

ZEB1, a well-established transcriptional regulator of EMT, has been associated with fibrotic remodeling across multiple disease settings. In renal fibrosis, ZEB1 has been reported to drive tubular epithelial fibrosis and glycolytic reprogramming,^47,48 whereas in pulmonary fibrosis, its aberrant upregulation is closely linked to collagen deposition and disease severity.^34,49 Nonetheless, direct evidence for the involvement of ZEB1 in ocular fibrosis, particularly in SRF, remains limited. Here, the functional role of ZEB1 in inducing EMT and SRF in RPE cells was further delineated. Both hypoxia and LPA1 overexpression independently increased ZEB1 protein levels and promoted its nuclear accumulation, indicating concurrent regulation of ZEB1 abundance and subcellular localization under fibrotic conditions. Importantly, ZEB1 silencing moderately reversed the EMT phenotype induced by LPA1, establishing ZEB1 as a pivotal downstream effector of LPA1-driven fibrotic signaling in RPE cells.

Increasing evidence suggests that ubiquitination contributes to fibrosis by regulating key EMT-associated factors.⁵⁰ USP1, a classical deubiquitinase primarily recognized for its involvement in DNA damage repair,⁵¹ has recently been linked to fibrotic modulation in chronic pancreatitis and hepatic fibrosis.^52,53 However, the role of USP1 in ocular fibrosis, especially EMT induction in RPE cells, continues to be poorly understood. This study demonstrated that USP1 expression was elevated in LPA1-overexpressing RPE cells. Although the upstream mechanism of USP1 induction was not directly examined, classical LPA1 downstream pathways, including Rho/Rho-associated protein kinase (ROCK), PI3K/Akt, mitogen-activated protein kinase (MAPK)/Erk, and Yes-associated protein (YAP)/transcriptional coactivator with PDZ-binding motif (TAZ)-related signaling, may contribute to USP1 upregulation and EMT-like remodeling in RPE cells.^54–56 Moreover, modulation of USP1 activity markedly altered the profibrotic effects induced by LPA1, reflecting USP1 as a key intermediary between LPA1 signaling and fibrotic remodeling. Mechanistically, USP1 facilitated ZEB1 deubiquitination, thereby enhancing its stability and nuclear translocation and ultimately amplifying the EMT program. Corroborated by protein half-life analysis, ubiquitination assays, molecular docking, and co-immunoprecipitation, the current findings collectively indicate that LPA1 augments ZEB1 deubiquitination through USP1 upregulation. Importantly, genetic silencing of USP1 or pharmacologic inhibition with ML323 counteracted the LPA1–driven reduction in ZEB1 ubiquitination and mitigated the associated EMT phenotype in RPE cells. Together, these findings extend previous observations by identifying a protein-stability mechanism through which the LPA1/USP1 axis regulates ZEB1 in the context of SRF, thereby providing mechanistic insight into the molecular basis of fibrotic progression in RPE cells.

BMS-986278, clinically designated as admilparant, is a next-generation selective LPA1 inhibitor originally developed for pulmonary fibrosis and subsequently evaluated in phase 2 clinical studies in idiopathic and progressive pulmonary fibrosis.^20,24,57 Here, BMS-986278 exhibited a comparable antifibrotic effect in the ocular setting. In the present ocular fibrosis model, combination treatment with BMS-986278 and aflibercept showed a trend toward further suppression of neovascular and fibrotic lesions compared with either monotherapy. These findings suggest that LPA1 inhibition may provide an antifibrotic effect in SRF, but further studies with larger sample sizes are needed to determine whether combined LPA1 and VEGF inhibition offers a statistically robust therapeutic advantage. Recent studies have indicated that LPA1 exerts pro-angiogenic effects in several pathological events,^58,59 although direct mechanistic evidence in ocular neovascular disease is limited. The role of ZEB1 in angiogenesis appears to be context dependent, as ZEB1 suppression favors endothelial differentiation during vascular development,⁶⁰ whereas in wound healing and malignancy ZEB1 functions as a pro-angiogenic factor.^61,62 Notably, in oxygen-induced retinopathy (OIR) and CNV models, endothelial ZEB1 has been reported to promote pathological neovascularization under a hypoxic environment.⁶³ Combined with the current findings, therefore, BMS-986278 may reduce SRF both by directly suppressing RPE EMT and by indirectly limiting fibrosis through partial inhibition of CNV-associated leakage and inflammatory activation in this two-stage laser model. Overall, our results suggest that BMS-986278 exerts dual anti-neovascular and antifibrotic effects, offering a distinct treatment approach for SRF management in nAMD.

This study has several limitations that warrant acknowledgment. First, the in vivo data were derived primarily from a two-step laser-induced mouse model of SRF. Although this model recapitulates fundamental aspects of injury-associated fibrotic remodeling, it may not fully replicate the prolonged and progressive course of the human condition. It should also be acknowledged that, although our in vitro data support a direct role of LPA1 in RPE EMT, the in vivo antifibrotic effects of BMS-986278 may not be exclusively mediated by RPE cells and may also involve endothelial, inflammatory, fibroblast-like, and extracellular matrix-remodeling components. Second, despite BMS-986278 exhibiting promising antifibrotic effects and a favorable preliminary safety profile in this study, the current evidence is largely confined to animal experiments and histological assessment. Further investigations incorporating clinical samples and longitudinal validation will therefore be required to clarify the pathological significance and therapeutic potential of the LPA1/USP1/ZEB1 axis in SRF.

In summary, the present study identified the LPA1/USP1/ZEB1 cascade as a previously unrecognized mechanistic pathway driving EMT and SRF in RPE cells. Pharmacologic inhibition of LPA1 with BMS-986278 effectively attenuated fibrotic remodeling both in vitro and in vivo, highlighting this signaling axis as a candidate therapeutic avenue for nAMD-associated fibrosis progression.