AZGP1 Attenuates Subretinal Fibrosis and Inhibits Epithelial-Mesenchymal Transition by Blocking the PI3K/AKT Signaling Pathway

Abstract

Purpose

Subretinal fibrosis (SRF) represents a significant contributor to irreversible vision loss in patients with neovascular age-related macular degeneration (nAMD). This study aimed to elucidate the underlying mechanism of SRF and identify potential therapeutic targets.

Methods

The SRF model was established using a two-stage laser-induced protocol in C57BL/6J mice. RNA-seq analysis was conducted to identify differentially expressed genes (DEGs) at 10 days and 30 days post-second laser. Quantitative RT-PCR was used to validate the expression levels of selected DEGs including zinc-alpha-2-glycoprotein 1 (AZGP1). Recombinant AZGP1 (rAZGP1) was intravitreally administrated to investigate its effects on SRF. The ARPE-19 cells were used to demonstrate the role of AZGP1 in modulating epithelial-mesenchymal transition (EMT).

Results

RNA-seq of RPE/choroid complex identified a total of 66 DEGs between samples collected at 10 days and 30 days post-second laser compared with controls (log2(fold change) ≥ 1, false discovery rate [FDR] < 0.05), with Azgp1 being one of the most significant downregulated genes. Intravitreal injection of rAZGP1 markedly reduced collagen I and CD31 positive areas in RPE/choroid flat-mounts. Co-localization of AZGP1 and RPE65 was observed in patients with nAMD (GSE135922) and SRF mouse models. Treatment with rAZGP1 resulted in significantly lower expressions of collagen I, α-SMA, and fibronectin in ARPE-19 cells after TGFβ1 induction. Both knockdown and overexpression studies demonstrated that AZGP1 regulated the PI3K/AKT signaling pathway within ARPE-19 cells.

Conclusions

The abnormal expression pattern of AZGP1 is critical for the development of SRF. Exogenous supplementation with AZGP1 may represent a promising strategy for ameliorating SRF by inhibiting EMT within RPE through the PI3K/AKT pathway.

Neovascular age-related macular degeneration (nAMD), characterized by choroidal neovascularization (CNV), constitutes 10% to 20% of all cases of AMD,¹ but is responsible for approximately 90% of vision loss among patients with AMD.² Anti-VEGF therapy is the consensus in CNV treatment.³ Whereas most instances of leakage can be effectively managed following intravitreal injection of anti-VEGF agents, a significant number of patients develop subretinal fibrosis (SRF) within 2 years,⁴ which remains the primary contributor to poor prognosis and irreversible vision loss in individuals with nAMD.

The conversion of neovascularization to fibrovascular membranes has been observed in macular fibrosis lesions from patients with AMD.⁵ Key components of these fibrovascular membranes include myofibroblasts, retinal pigment epithelial (RPE) cells, macrophages, vascular endothelial cells, fibroblasts, and accumulated extracellular matrix (ECM) components such as fibronectin and collagen.⁶ Given that myofibroblasts are responsible for synthesizing the majority of the ECM, strategies aimed in inhibiting the transition of other cell types into myofibroblasts are considered effective to limit SRF.⁷ However, the cellular origins of myofibroblasts in the subretinal region remain poorly understood. Studies have demonstrated that RPE cells can undergo epithelial-mesenchymal transition (EMT) after injury, with elevated expression levels of mesenchymal markers like fibronectin and α-SMA, indicating its transformation into myofibroblasts.^8,9 Literatures show that several signaling pathways are involved in EMT of RPE cells, including the TGF-β/Smad pathway, PI3K/AKT pathway, and Wnt/β-catenin pathway.^10–12 Notably, whereas studies have shown prospects of EMT inhibitors in SRF treatment,^13,14 further investigation is warranted to elucidate the intrinsic cause underlying RPE disturbance during SRF.

Zinc-alpha-2-glycoprotein 1 (AZGP1) is a secreted protein predominantly synthesized by epithelial cells and adipocytes.¹⁵ The physiological roles of AZGP1 are closely associated with lipid metabolism.¹⁶ However, research has also demonstrated its antifibrotic potential in modulating EMT.^17,18 Deficient hepatic expression of AZGP1 was found in patients with hepatocellular carcinoma, which triggered EMT induced by TGFβ1-ERK2 signaling and promoted invasion.¹⁹ Furthermore, exogenous supplementation of recombinant AZGP1 (rAZGP1) protein significantly alleviated inflammation and fibrosis across cardiac, hepatic, and renal tissues.^19–21 In light of these findings, we hypothesize that AZGP1 may exert antifibrotic effects on SRF secondary to nAMD.

In this study, we demonstrated the downregulation of AZGP1 in a mouse model of SRF through RNA sequence analysis, and observed the co-localization of AZGP1 with RPE cells. We also highlighted that supplementation of AZGP1 to inhibit PI3K/AKT-mediated EMT may be an effective therapeutic strategy for SRF management.

Materials and Methods

Mice

All mice used in this study were male C57BL/6J mice aged 6 to 8 weeks, purchased from LingChang (Shanghai, China). The mice were housed under specific pathogen-free conditions on a 12-hour light and 12- hour dark cycle, with water and food (grain-based chow) that were provided at all times. All animal experiments were conducted according to the protocols approved by the Committee of Animal Care of Soochow university (Approval number: 202407A0083) and adhered to the Association for Research in Vision and Ophthalmology's Statement for Use of Animals.

Subretinal Fibrosis Induction

Mouse model of SRF was established following a two-stage laser-induced protocol.²² Briefly, the operated eye was dilated with 1% tropicamide (Santen Pharmaceutical Co., Ltd). The mice were anaesthetized intraperitoneally with 1.8% avertin (0.15 mL per 10 g body weight). Then, laser burns (250 mv power, 0.1 second duration, and a spot size of 50 µm) were applied to each eye in the fundus, approximately 2 discs in diameter away from the optic disc utilizing a VITRA photocoagulator (Quantel Medical, France). For RPE/choroid flat mounts, four laser burns were inducted, whereas for RNA and protein extraction, eight laser burns were carried out. The second laser burn was conducted to each first lesion burn 7 days later. Lesions with hemorrhages or fusion were excluded from analysis. Specifically, lesions with the maximum of bleeding diameter bigger than lesion diameter (LD) were excluded and the whole eyes were excluded with bleeding diameter bigger than two LDs according to the recommendation given by Gong et al.²³

RNA-Seq Transcriptome

Mice were euthanatized 10 days and 30 days following the second laser and RPE/choroid complex was separated from the eyeball. Total RNA was extracted from RPE/choroid complex using RNeasy Kits (Qiagen, Germany) according to the manufacturer's protocol. The concentration and quality of RNA were assessed using the Standard Sensitivity RNA Analysis Kit (Agilent, Santa Clara, CA, USA). Single-stranded cDNA libraries from four subretinal fibrosis mice and three healthy control mice were sequenced using the DNBSEQ platform (BGI, Shenzhen, China). Reads containing adapter and N or low-quality reads were filtered by using the SOAPnuke software (version 1.5.2, China). Reference genome alignment was conducted by the HISAT2 software (version 2.0.4). Differentially expressed genes (DEGs) with the criteria of false discovery rate (FDR) < 0.05 (the P value adjusted by the Benjamini–Hochberg method) and log2(fold change) ≥1 were determined by DESeq2.

RNA Extraction, Reverse Transcription, and Quantitative RT-PCR

Total RNA from RPE/choroid complex was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. Total RNA from ARPE-19 cells was extracted by the RNA extraction kit (TaKaRa, Tokyo, Japan). Reverse transcription was conducted by the PrimeScript RT reagent Kit (Takara, Kyoto, Japan). For quantification of gene expression, CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) with TB Green Premix Ex Taq (TaKaRa, Tokyo, Japan) was used to perform quantitative RT-PCR. The two-step protocol was as follows: 95°C for 30 seconds followed by 40 cycles of 95°C for 5 seconds, and 60°C for 30 seconds. Relative expression of mRNA was calculated using the 2 − Δ(ΔCT) comparative method. All primers used in this study were sourced from Tsingke (Beijing, China) and are detailed in Table 1.

Table 1.

Sequence of Primers Used in the Experiments

Primer Name	Species	Forward Sequence	Reverse Sequence
Azgp1	Mouse	AGCAAAGGTTTTCCGAGGTTT	GAGACCCTGTAGTGTCCTTGTAA
Crlf1	Mouse	CTCCCTGCAAGCTACCTGC	AGGGTGGAGGTGTTAAGGAGG
Lcn2	Mouse	TGGCCCTGAGTGTCATGTG	CTCTTGTAGCTCATAGATGGTGC
Lox	Mouse	TCTTCTGCTGCGTGACAACC	GAGAAACCAGCTTGGAACCAG
Thbs1	Mouse	GGGGAGATAACGGTGTGTTTG	CGGGGATCAGGTTGGCATT
Npy	Mouse	ATGCTAGGTAACAAGCGAATGG	TGTCGCAGAGCGGAGTAGTAT
Edn2	Mouse	CACCTGCGTTTTCGTCGATG	CCAGTGTCTTCGATGGCAGAA
Myb	Mouse	AGACCCCGACACAGCATCTA	CAGCAGCCCATCGTAGTCAT
Il1rn	Mouse	GCTCATTGCTGGGTACTTACAA	CCAGACTTGGCACAAGACAGG
Olfm4	Mouse	CAGCCACTTTCCAATTTCACTG	GCTGGACATACTCCTTCACCTTA
Tmem125	Mouse	GCCGTGCAGGATATGAACTG	GCCAAGGTAATGCCCACAGA
Tnc	Mouse	ACGGCTACCACAGAAGCTG	ATGGCTGTTGTTGCTATGGCA
Top2a	Mouse	CAACTGGAACATATACTGCTCCG	GGGTCCCTTTGTTTGTTATCAGC
Aadac	Mouse	TACCGCTTCCAGATGCTATTGA	ACTGATTCCCAAAAGTTCACCAA
Azgp1	Human	AACCAAGATGGTCGTTACTCTCT	CCTGCTTCCAATCCTCCATTC
Col1a1	Human	GTGCGATGACGTGATCTGTGA	CGGTGGTTTCTTGGTCGGT
Fn1	Human	CGGTGGCTGTCAGTCAAAG	AAACCTCGGCTTCCTCCATAA
Hhatl	Mouse	CCTTCCGGGAGTCTGTGAGA	CATCGTGCAGAGTTTGGCA
Hlx	Mouse	GCAGCAATCACCAACGCAG	GGGTCAAATTCCGCAGACAAA
Htr3a	Mouse	CCTGGCTAACTACAAGAAGGGG	TGCAGAAACTCATCAGTCCAGTA
Gm4736	Mouse	GATGATGAAGATGGCAGTGAG	ATGGTGGTTTCCTGAGTGAG
β-actin	Mouse	CACTGTCGAGTCGCGTCC	CGCAGCGATATCGTCATCCA
GAPDH	Human	CGCTGAGTACGTCGTGGAGTC	GCTGATGATCTTGAGGCTGTTGTC

Protein Extraction and Western Blotting

RPE/choroid complex and ARPE-19 cells were lysed with RIPA buffer (Beyotime, Shanghai, China) added with proteolytic protease and phosphatase inhibitor cocktail (NCM Biotech, Suzhou, China). Following protein quantification, samples were separated by 10% SDS-page gels and transferred to the PVDF membrane. Primary antibodies used to incubate overnight at 4°C included: AZGP1 (1:1000; anti-mouse; Santa Cruz Biotechnology, Santa Cruz, CA), AZGP1 (1:1000; anti-human; Proteintech, Chicago, IL, USA), GAPDH (1:50000; Proteintech, Chicago, IL, USA), Collagen I alpha 1 (1:1000; Zenbio, Chengdu, China), α-SMA (1:500; Santa Cruz Biotechnology, Santa Cruz, CA), AKT1 (1:1000, Abclonal, Wuhan, China), phosphor-AKT1 (1:1000, Abclonal, Wuhan, China), PI3-kinase p85α (1:500; Santa Cruz Biotechnology, Santa Cruz, CA), and phosphor-PI3-kinase (1:1000, Abmart, Shanghai, China). Blots were quantified using Image J software. Band signals were normalized to GAPDH. Uncropped gels with molecular weight markers are provided in Supplementary Materials.

Cryosection and Immunofluorescence Staining

Localization of AZGP1 in cross-sections of mouse eyes was determined by immunofluorescence staining. Briefly, mouse eyeballs were immediately fixed with FAS eyeball fixative (Servicebio, Wuhan, China) after collection for 24 hours. After dehydration, the eyeballs were embedded with optimal cutting temperature (OCT) and sectioned by 10 µm thickness for cryosections. For immunofluorescence staining, the cryosections were fixed with methanol for 30 minutes and heated 30 minutes for antigen retrieval. The cryosections were blocked with 3% BSA for 30 minutes at room temperature and then incubated with primary mouse antibody to AZGP1 (1:100; Santa Cruz Biotechnology, Santa Cruz, CA) and rabbit antibody to RPE65 (1:200; Proteintech, Chicago, IL, USA) overnight at 4°C, followed by Alexa Fluor 488 anti-mouse secondary antibody (1:500; Beyotime, Shanghai, China) and Alexa Fluor 594 anti-rabbit secondary antibody (1:500; Abcam, CA, USA). Images were captured using the ECLIPSE Ni-U fluorescence microscope (Nikon, Japan).

Intravitreal Injection of Mouse Recombinant AZGP1 Protein

Mouse recombinant AZGP1 (rAZGP1; purity ≥90%) was obtained from MedChemExpress (MCE; Monmouth Junction, NJ, USA). Denatured rAZGP1 was prepared by heating the protein solution at 99°C for 60 minutes.²⁴ One microliter (1 µL) of rAZGP1 (1 µg), denatured rAZGP1 or relevant dilution buffer (vehicle) was administered via intravitreal injection one day following the second laser according to previous studies.²⁵ Eyes exhibiting significant hemorrhage post-injection were excluded from the analysis.

SRF Lesion Labeling in Choroid/RPE Flat-Mount

The mice with four laser burns were euthanized for choroid/RPE flat-mounts. Briefly, the anterior segment was meticulously excised along the corneoscleral limbus and then the retina was peeled away. The choroid/RPE complex was cut into flat-mount. For lesion labeling, choroid/RPE flat-mount was permeabilized and blocked with 5% BSA and 0.3% Triton X-100 for 2 hours at room temperature. Then, the primary rabbit antibody to Collagen I alpha 1 (1:200; Zenbio, Chengdu, China) and rat antibody to CD31 (1:50; BD Pharmingen, Franklin Lakes, NJ, USA) were incubated overnight at 4°C, followed by incubation with Alexa Fluor 594 anti-rabbit secondary antibody (1:500; Abcam, Fremont, CA, USA) and Alexa Fluor 488 anti-rat secondary antibody (1:1000; Invitrogen, Carlsbad, CA, USA). Fluorescence images were taken by the ECLIPSE Ni-U fluorescence microscope (Nikon, Japan). The area labeled with Collagen I and CD31 was highlighted to be the fluorescent regions by adjusting the threshold in Image J software (version 1.54g; Java 1.8.0). The fluorescent areas with the same adjustment of all photographs were then measured and analyzed.

Cell Culture and Treatments

ARPE-19 cell line was purchased from Vigen Biotechnology Co., Ltd (Jiangsu, China). The cells were cultured in DMEM/F12 medium (Gibco, Grand Island, NY, USA) with 10% fetal bovine serum (FBS) and 50 U/mL penicillin and streptomycin under standard conditions. Primary mouse RPE were isolated from the eyes of mice aged 18 to 21 days and cultured according to the previous study.²⁶ To induce EMT, the cells were treated with 10 ng/mL TGFβ1 (MCE, Monmouth Junction, NJ, USA), alone or in combination with human rAZGP1 protein (1 µg/mL; MCE, Monmouth Junction, NJ, USA). After 48 hours, the cells were harvested for subsequent experiments. ARPE-19 cells used were within 10 passages, and primary mouse RPE used were within 2 passages.

Immunocytochemistry

Cells were washed with PBS and then fixed with 4% PFA for 15 minutes, permeabilized and blocked with 5% BSA (Solarbio Science & Technology Co., Ltd, Beijing, China) and 0.3% Triton X-100 in PBS for 30 minutes at room temperature. The cells were then incubated with primary mouse antibody to α-SMA (1:100; Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight at 4°C, followed by Alexa Fluor 488 anti-mouse secondary antibody (1:500; Beyotime, Shanghai, China). Slides were mounted with DAPI Fluoromount-G (SouthernBiotech, Birmingham, AL, USA).

Small Interfering RNA Transfection in the ARPE-19 Cell Line

The small interfering RNAs (siRNAs) targeting human Azgp1 and negative control were synthesized by Yixuesheng Biotechnology (Shanghai, China). The siRNA was transfected with Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) at a cell confluence of 30%. The efficiency of transfection was detected via FAM-labeled siRNA using fluorescence microscope after 6 hours. The interfering efficacy was demonstrated by expression level of mRNA and protein after 48 hours. TGFβ1 (10 ng/mL; MCE, Monmouth Junction, NJ, USA) with or without the AKT inhibitor Capivasertib (1 µM; MCE, Monmouth Junction, NJ, USA) was added for an additional 48 hours for subsequent experiments. The sequence of siRNAs is detailed in Table 2.

Table 2.

Sequence of Human siRNAs Used in the Experiments

Name	Sense	Antisense
si-Azgp1	GUGAGAUCGAGAAUAACAGAATT	UUCUGUUAUUCUCGAUCUCACTT
Negative control	UUCUCCGAACGAGUCACGUTT	ACGUGACUCGUUCGGAGAATT

Establishment of AZGP1 Overexpressed ARPE-19 Cell Line

Lentivirus vectors (LVs) that expressed GFP-tagged Azgp1 (LV-Azgp1) and GFP-tagged empty vector (LV-empty) were synthesized and purchased from ANGEL Biotech Co., Ltd (Suzhou, China). ARPE-19 cells were transduced with LV-Azgp1 or LV-empty followed by puromycin dihydrochloride (10 µg/mL) to establish stable transfected cell line. Cells were added with TGFβ1 (10 ng/mL; MCE, Monmouth Junction, NJ, USA) for the following experiments.

Analysis of Single Cell RNA Sequence Data

Raw single cell RNA sequence count data was obtained from GSE135922 and the cells involved have already been clustered. Seurat objects were created for each sample, including both peripheral regions and macular regions (2 control human RPE/choroid complexes, 1 nAMD human RPE/choroid complex, 3 control human CD31 enriched RPE/choroid complexes, and 1 nAMD human CD31 enriched RPE/choroid complex). Seurat objects were merged, normalized, and clustered based on the raw data information. Cluster “RPE” and “RPE/Rod/Melanocyte” were selected to analyze the co-expression of AZGP1 and RPE65 using the FeaturePlot command through t-SNE.

Statistical Analysis

GraphPad Prism version 9.5.1 was used for the statistical analysis. Data were presented as mean ± SD and P < 0.05 was considered statistically significant. For statistical analysis, the 2-sided t-test was used for individual groups, whereas the 1-way ANOVA followed by a Turkey's test was used for multiple groups. Bioinformatic analysis including volcano plot, heatmap and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of DEGs were performed using the Dr. Tom System (https://biosys.bgi.com).

Results

Abnormal AZGP1 Expression in the RPE/Choroid Complex of Mice Following a Two-Stage Laser Induction

To investigate the potential molecular alterations during SRF, we first assessed the differential gene signature of the RPE/choroid complex from mice at 10 days (SRF10D) and 30 days (SRF30D) post-second laser induction by RNA-seq transcriptome analysis. In comparison to control mice that underwent no laser treatment, there were 353 DEGs in SRF10D and 214 DEGs in SRF30D (Figs. 1A–C). Furthermore, there were 66 common DEGs shared between SRF10D and SRF30D compared with the controls, comprising 51 upregulated genes, 5 downregulated genes in both time points, and 10 genes downregulated in SRF10D, but upregulated in SRF30D (see Figs. 1A–C, Supplementary Table S1), which indicated persistent changes of these genes throughout the duration of SRF. Notably, Azgp1 was consistently downregulated in both SRF10D and SRF30D samples (see Figs. 1B–D). Previous studies have showed that AZGP1 possesses inhibitory effects on vascular fibrosis and inflammation.^21,27 Consequently, we proceeded to validate the expression levels of Azgp1 along with 17 other selected DEGs using RT-qPCR (Figs. 1E–G). The majority of the DEGs were confirmed at least at one time point, the remaining one, GM4736, showed no trend in RT-qPCR (see Figs. 1E–G).

Figure 1.

RNA-seq transcriptome of RPE/choroid revealed different gene signature between control and subretinal fibrosis mice. RPE/choroid was collected at 10 days (SRF10D), 20 days (SRF20D), and 30 days (SRF30D) post-second laser induction. (A) Venn diagram showing 66 common differentially expressed genes (DEGs) in SRF10D and SRF30D compared with the healthy controls. Log2(fold change) ≥ 1, FDR < 0.05. Each sample contains 2 RPE/choroid complex, 3 samples in control group, 2 samples in SRF10D and SRF30D group. (B, C) Volcano plot showing DEGs in SRF10D (B) and SRF30D (C) compared with the healthy controls. (D) Heatmap showing the selected 18 DEGs among SRF10D, SRF30D, and the healthy control group. (E–G) RT-qPCR verification of selected notable DEGs in control and SRF mice. Mean ± SD, N = 3 to 4 per group, all data are presented as mean ± SD, *P  <  0.05, **P  <  0.01, ***P  <  0.001, the P value was obtained by 1-way ANOVA followed by multiple comparisons.

To further elucidate the expression of AZGP1 at the protein level during SRF, we collected RPE/choroid complexes from mice at 3 days (SRF3D), 7 days (SRF7D), 14 days (SRF14D), 21 days (SRF21D), and 28 days (SRF28D) post-second laser induction for Western blotting analysis. Notably, there were no differences in AZGP1 expression among control mice aged 8 to 11 weeks (Figs. 2A, 2B), and, therefore, week-matched controls were not utilized in the study. Additionally, no clear trend of AZGP1 was observed at SRF3D and SRF7D, followed by a marked decrease at SRF14D, SRF21D, and SRF28D (see Figs. 2A, 2C). Immunofluorescence staining of cross-sections of mouse eyes corroborated these results (Fig. 2D). These findings suggest that AZGP1 is abnormally expressed in the RPE/choroid complex of mice following two-stage laser induction, with levels appearing to decline in the late stages of SRF.

Figure 2.

AZGP1 was abnormally expressed in RPE/choroid from SRF mice compared with the healthy controls. Eyes were collected 3 days (SRF3D), 7 days (SRF7D), 14 days (SRF14D), 21 days (SRF21D), and 28 days (SRF28D) after the 2-stage laser induction. (A–C) Western blotting of the RPE/choroid complex showing discrepancies in AZGP1 expression after SRF induction (N = 3 to 4). (D) Immunofluorescence staining showed expression of AZGP1 (green; white arrow) and α-SMA (red) in cross-sections of SRF and the control mouse eyes. Cell nuclei were counterstained with DAPI (blue; N = 3). GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. Scale bar = 50 µm. All data are presented as mean ± SD, *P  <  0.05, **P  <  0.01, ***P  <  0.001, the P value was obtained by 1-way ANOVA followed by multiple comparisons for multiple groups and two-sided t-test between individual groups.

Supplementation of rAZGP1 Alleviated Subretinal Fibrosis in the SRF Mouse Model

Given that AZGP1 expression in the RPE/choroid complex was downregulated 14, 21, and 28 days following the second laser induction, we next investigated whether exogenous supplementation of AZGP1 would exert antifibrotic effects on SRF. As illustrated in Figure 3A, mice were subjected to two-stage laser-induction, followed by intravitreal injection of rAZGP1 (1 µg per eye) 1 day after the second laser treatment (see Fig. 3A). Administration of rAZGP1 did not induce retinal toxicity (Supplementary Figs. S1A–S1E), and the AZGP1 levels in the RPE/choroid complex were significantly elevated following intravitreal injection of rAZGP1 (Supplementary Figs. S1F, S1G). Expectedly, our findings demonstrated that rAZGP1 injection significantly diminished Collagen I and CD31 positive areas in choroid/RPE flat mounts at 14 days post-second laser compared with vehicle and denatured rAZGP1 injection (Figs. 3B–D). Because α-SMA served as a specific marker of myofibroblasts,²⁸ we also assessed the expression levels of α-SMA and Collagen I via Western blotting. In alignment with immunofluorescence results, significant decrease of α-SMA and Collagen I was found in mice that received rAZGP1 injection (Figs. 3E–G, Supplementary Fig. S1H). These data indicate that AZGP1 may alleviate laser-induced subretinal fibrosis and vascularization in mice.

Figure 3.

AZGP1 alleviated subretinal fibrosis in SRF mouse model. (A) Experimental protocols. Mice were subjected to the two-stage laser induction. The rAZGP1 (1 µg in 1 µL dilution buffer per eye), denatured rAZGP1 (de rAZGP1) or dilution buffer (vehicle) was intravitreally injected 1 day post-second laser. At day 14, choroid/RPE flat-mounts were made for further immunofluorescence staining, and RPE/choroid complex were collected for further Western blotting. (B–D) Representative immunostaining images and quantification for collagen I (red) and CD31 (green) in choroid/RPE flat-mounts of mice with and without intravitreal injection (IVI) of rAZGP1 after SRF induction. Cell nuclei were counterstained with DAPI (blue). Scale bar = 100 µm, N = 21 to 27 lesions from 8 eyes per group. (E–G) Western blotting of the RPE/choroid complex and quantification showing significant decrease of collagen I and α-SMA after IVI of rAZGP1 (N = 3). All data are presented as mean ± SD. *P  <  0.05, **P  <  0.01, ***P  <  0.001, the P value was obtained by 1-way ANOVA followed by multiple comparisons.

AZGP1 Was Co-Localized With RPE Cells of Human and Mouse Eye Samples

Studies have reported that the detachment and dissociation of RPE cells occur at the early stage of CNV, which play critical roles in initiating EMT process regulated by TGFβ.²⁹ Moreover, RPE cells are considered as the main sources of TGFβ2 in the outer retina.³⁰ Upon stimulation by TGFβ signaling, RPE cells can undergo transformation into myofibroblasts during SRF,⁹ indicating the important role of RPE cells throughout the progress of nAMD. Therefore, we subsequently investigated the expression of AZGP1 in RPE cells and whether AZGP1 could modulate the role of RPE cells involved in SRF.

Upon analyzing the online single-cell RNA-Seq datasets (GSE135922) of samples from patients with AMD and healthy individuals,³¹ we found that AZGP1 is mainly expressed in the “RPE” cluster and the “RPE/Rod/Melanocyte” cluster (Fig. 4A). Co-localization of AZGP1 with the RPE-specific marker RPE65 was also observed in both healthy individuals and those with macular degeneration (Fig. 4B). We further examined the cellular localization of AZGP1 in cross-sections of mouse eyes 7 days post-second laser treatment. Notably, there was co-localization of AZGP1 and RPE65 adjacent to the laser-induced SRF lesion, where the arrangement of RPE cells appeared relatively normal (see Fig. 4B). Conversely, diminished levels of AZGP1 were found within the SRF lesion characterized by ruptured and dissociated RPE cells (see Fig. 4B), which were in agreement with our expected expression pattern of AZGP1. Additionally, we identified co-localization of α-SMA and RPE65 within the mouse SRF lesion (Fig. 4C, Supplementary Fig. S2). Taken together, these findings suggest that AZGP1 is co-localized with RPE cells in human and mouse eye samples. Further investigation of its role in EMT progress and transformation into myofibroblasts among RPE cells is warranted.

Figure 4.

AZGP1 was co-localized with RPE65 both in human and mouse eye samples. (A) The expression of Azgp1 was analyzed in several clusters associated with fibrosis of human RPE/choroid complex from GSE135922. (B) Co-localization of AZGP1 (red dots) and RPE65 (green dots) in RPE and choroidal cell populations from patients with neovascular AMD and controls (single-cell RNA-seq data from NCBI GEO database, ID: GSE135922). CTRL, normal macula, N = 5; nAMD, neovascular age-related macular degeneration, N = 2. (C) Immunofluorescence staining showed co-localization (white arrow) of AZGP1 (green) and RPE65 (red) in cross-sections of SRF mouse eyes (7 days after the second laser induction). (D) Immunofluorescence staining showed co-localization (white arrow) of α-SMA (green) and RPE65 (red) in SRF mouse eyes (7 days after the second laser induction). Cell nuclei were counterstained with DAPI (blue; N = 3). Scale bar = 50 µm.

AZGP1 Limited TGFβ1-Induced EMT and Inhibited PI3K/AKT Signaling Pathway in ARPE-19 Cells

To further elucidate the role of AZGP1 in modulating EMT, we used TGFβ1 to induce an EMT phenotype in ARPE-19 cells. Notably, the expression level of AZGP1 significantly decreased following 48 hours of TGFβ1 stimulation (Figs. 5A–C). As shown in Supplementary Figures S1F to S1G, administration of rAZGP1 increased the total level of AZGP1 detected by Western blotting in RPE/choroid complexes, the protein level of AZGP1 also showed no reduction, whereas co-treatment with rAZGP1 compared to TGFβ1 stimulation alone in ARPE-19 cells (see Figs. 5B, 5C).

Figure 5.

AZGP1 inhibited TGFβ1 induced EMT in RPE cells. Cells were treated with 10 ng/mL TGFβ1 for 48 hours to induce EMT with or without 1 µg/mL rAZGP1. (A) The mRNA level of AZGP1 was detected by RT-qPCR in the healthy controls, the TGFβ1, and the TGFβ1 + rAZGP1 groups in ARPE-19 cells (N = 3). (B, C) Western blotting showed protein level of AZGP1 in the healthy controls, the TGFβ1, and the TGFβ1 + rAZGP1 groups in ARPE-19 cells (N = 3 to 4). (D–G) Immunofluorescence staining showed reduced expression of α-SMA (green) with rAZGP1 treatment in ARPE-19 cells (D, E) and primary mouse RPE cells (F, G). Scale bar = 50 µm, N = 3. (H, I) mRNA expression of Col1a1 (H) and Fn1 (I) were detected by RT-qPCR in ARPE-19 from the healthy control, the TGFβ1, and the TGFβ1 + rAZGP1 groups (N = 3). (J–M) Western blotting showed the protein expression of Fibronectin1, collagen I, and α-SMA in ARPE-19 from the healthy control, the TGFβ1, and the TGFβ1 + rAZGP1 groups (N = 4). Control, control group; TGFβ1, TGFβ1 induction; TGFβ1 + rAZGP1, TGFβ1 induction with rAZGP1 treatment. All data are presented as mean ± SD, *P  <  0.05, **P  <  0.01, ***P  <  0.001, the P value was obtained by 1-way ANOVA followed by multiple comparisons.

Immunofluorescence staining demonstrated a marked increase in α-SMA positive ARPE-19 cells after 48 hours of TGFβ1 treatment, whereas co-treatment with rAZGP1 significantly reduced α-SMA expression (Figs. 5D, 5E). Furthermore, rAZGP1 partly restored the myofibroblast-like morphology characterized by spindle-shape cells induced by TGFβ1 (see Fig. 5D). These results were further verified using primary mouse RPE (Supplementary Fig. S3, Figs. 5F, 5G). To further investigate molecular alterations associated with EMT in ARPE-19 cells, RT-qPCR and Western blotting were conducted. As illustrated in Figures 5H to 5M, markers associated with EMT such as Collagen I, Fibronectin1, and α-SMA exhibited significant upregulation in the TGFβ1 treatment group and were largely reversed by rAZGP1 administration (see Figs. 5H–M). These findings suggest that AZGP1 possesses the capacity to inhibit TGFβ1-induced EMT in ARPE-19 cells.

PI3K/AKT has been recognized as a principal regulator of fibrosis.^32,33 A previous study reported that AZGP1 inhibited cell migration through regulating the Akt pathway.³⁴ Furthermore, the KEGG pathway analysis of the 66 common DEGs mentioned above revealed that the PI3K/AKT signaling pathway ranked among the top 20 significant pathways in SRF10D and SRF30D compared with the control group (Fig. 6A), indicating its potential involvement in SRF. Consequently, we explored whether the inhibitory effect of AZGP1 on EMT was mediated through regulation of the PI3K/AKT pathway in vitro. As anticipated, treatment with rAZGP1 significantly decreased AKT phosphorylation in ARPE-19 cells induced by TGFβ1 (Figs. 6B, 6C). To further validate the effect of AZGP1 on the PI3K/AKT pathway, siRNA was transfected to knockdown AZGP1 expression in ARPE-19 cells. The efficacy of knockdown was assessed via RT-qPCR and Western blotting (Supplementary Figs. S4A–D), with sequence #1 at a concentration of 50 nM selected for subsequent experiments. As demonstrated in Figures 6D and 6F, knockdown of AZGP1 markedly aggravated TGFβ1-induced phosphorylation of AKT, which was mitigated upon administration of Capivasertib (an AKT inhibitor; Figs. 6D, 6F). Notably, knockdown of AZGP1 significantly increased the phosphorylation of the p55 subunit (Tyr199) of PI3K, but not the p85 subunit (Tyr467) of PI3K (see Figs. 6E, 6G). Additionally, lentivirus expressing Azgp1 was used to upregulate AZGP1 levels in ARPE-19 cells (Supplementary Figs. S4E–H). As shown in Figures 6H and 6I, overexpression of AZGP1 significantly diminished AKT, PI3K (p85), and PI3K (p55) phosphorylation levels, and decreased Collagen I expression induced by TGFβ1. Collectively, these findings suggest that AZGP1 inhibits the PI3K/AKT signaling pathway to exert anti-EMT effects within ARPE-19 cells.

Figure 6.

AZGP1 inhibited the PI3K-AKT pathway in ARPE-19 cells. (A) KEGG pathway analysis showing the top 20 significant pathways of 66 common DEGs in SRF10D and SRF30D compared with the healthy controls. (B, C) Western blotting showed the protein expression of phospho-AKT (Ser473) and AKT in ARPE-19 from the healthy control, the TGFβ1, and the TGFβ1 + rAZGP1 groups (N = 4). Control, control group; TGFβ1, TGFβ1 (10 ng/mL) induction for 48 hours; TGFβ1 + rAZGP1, TGFβ1 (10 ng/mL) induction with rAZGP1 (1 µg/mL) treatment for 48 hours. (D–G) ARPE-19 cells were transfected with negative control siRNA (NC) and Azgp1 targeted siRNA. TGFβ1 (10 ng/ml) and Capivasertib (1 µM) were added after RNA interference for additional 48 hours. Western blotting showed the phosphorylation of AKT (D, F) and PI3K (E, G) in ARPE-19 cells (N = 3). (H, I) Western blotting showed the expression level of Collagen I, phosphor-AKT, AKT, phosphor-PI3K, and PI3K in ARPE-19 cells transduced with LV-Azgp1 and LV-empty after treated with 10 ng/mL TGFβ1 for 48 hours (N = 3). The result of p-AKT/AKT was based on the band of p-AKT and AKT (incubated after stripping), whereas the result of p-PI3K/PI3K was based on the band of p-PI3K and PI3K normalized to GAPDH, respectively. All data are presented as mean ± SD, *P  <  0.05, **P  <  0.01, ***P  <  0.001, the P value was obtained by 1-way ANOVA followed by multiple comparisons for multiple groups and two-sided t-test between individual groups.

Discussion

The nAMD is characterized by the abnormal neovascularization in the choroid, often accompanied by exudation, leading to visual impairment. As the prevalence of AMD rises, it imposes an increasing social and economic burden.³⁵ Progressive lesion fibrosis affects up to 32% of patients with nAMD even after sustained treatment with anti-VEGF agents, such as ranibizumab, bevacizumab, and aflibercept for 1 year.^4,36 In addition, the incidence of SRF reaches 36%, 56%, and 62.7% at 2 years, 5 years, and 10 years, respectively.^4,37 These reports indicate that the incidence of SRF is highest in the first year and early prevention and treatment are critical. However, effective strategies for managing SRF remain elusive; therefore, further exploration into the mechanisms underlying SRF development is essential to identify potential therapeutic targets. In the current study, we showed abnormal expression of AZGP1 in a mouse model of SRF induced by two-stage laser-induction. Additionally, we illustrated the protective role of AZGP1 in alleviating SRF through intravitreal injection of rAZGP1 in mice. Finally, we investigated the anti-EMT effects of AZGP1 on RPE cells via inhibition of the PI3K/AKT pathway.

As an adipokine, AZGP1 has been shown to regulate lipid metabolism.³⁸ Recent studies have indicated that AZGP1 also modulates immune responses.^39,40 The level of AZGP1 was found to be negatively correlated with the severity of atopic dermatitis.⁴⁰ AZGP1 conferred its immunoregulatory effects by enhancing Foxp3 expression while reducing IL-4, IL-17, and IFN-γ levels.⁴⁰ Literatures have also demonstrated the antifibrotic and anti-EMT role of AZGP1 in organs such as the kidneys, lungs, and heart,^21,41 suggesting the potential of AZGP1 as a novel target for fibrosis treatment. For instance, Lenvatinib is a candidate agonist of AZGP1 that inhibits EMT by blocking the TGF-β1/Smad3 signaling pathway in an AZGP1-dependent manner.¹⁸ Of note, previous studies have confirmed that the level of AZGP1 in RPE/choroid decreased in aged mice as compared with younger counterparts,⁴² indicating the potential role of AZGP1 in AMD and SRF, which is an age-related disease. In our study, we detected a significant decrease of AZGP1 level within the RPE/choroid complex from mice that 10 days and 30 days post-second laser induction by RNA-seq transcriptome analysis (see Fig. 1), which was further verified by Western blotting and immunofluorescence staining at 14 days, 21 days, and 28 days post-second laser induction (see Fig. 2). The selected time points lacked precise synchronization with the RNA sequencing, which is a limitation that warrants attention and improvement in future studies. Interestingly, the expression of AZGP1 in choroid/RPE complex does not appear to significantly decrease in the early stage following the second laser exposure (see Fig. 2). This variation observed on day 3 and day 7 may be attributed to feedback regulation. Besides, the expression of AZGP1 is regulated by a multitude of factors, including metabolic processes, nutrition status, and various signaling pathways.^16,43 Despite the relative high level of AZGP1 initially observed, its functional efficacy may be compromised early on. Consequently, we investigated the impact of supplemental rAZGP1 administration on alleviating SRF. The strategy involving intravitreal injection of rAZGP1 effectively reduced subretinal lesion areas labeled by collagen I and CD31, and decreased expressions of collagen I and α-SMA (see Fig. 3). These results highlight AZGP1 as a potential target for SRF treatment and paved the way for further mechanistic investigations.

RPE cells are critical for maintaining retinal homeostasis, with functions such as phagocytosis of photoreceptor outer segments, secretion and transportation of important trophic factors, and regulation of immune response.⁴⁴ Under injured conditions, RPE cells undergo EMT induced by cytokines such as TGFβ and followed by the transition into α-SMA positive myofibroblasts, which has been considered as a primary source of myofibroblasts in SRF. The data presented in Figures 1 to 3 indicated that AZGP1 exhibits antifibrotic effects in the SRF model, but its specific cellular localization remained unknown. In the current study, we showed co-localization of AZGP1 and RPE65 in both human and murine models (see Fig. 4). Notably, only AZGP1-positive cells were observed in Figure 4A, which may correspond to rod cells and melanocytes within the “RPE/Rod/Melanocyte” cluster. Given that the roles of rod cells and melanocytes in subretinal fibrosis have not been extensively reported, RPE cells are considered to be the key cell type associated with AZGP1 in SRF. However, due to the limited number of human samples available for analysis, it remains challenging to compare the expression of AZGP1 in RPE cells between health individuals and patients with AMD. Future studies necessitate an increased sample size to clarify the temporal and spatial expression patterns of AZGP1 throughout the progression of AMD. Additionally, we provided insights into the impact of rAZGP1 administration on mitigating TGFβ1-induced EMT process in ARPE-19 cells (see Fig. 5) while also regulating the PI3K/AKT signaling pathway (see Fig. 6). Of note, given that AZGP1 is an adipokine and aberrant lipid metabolism has been recognized as a critical factor involved in AMD,^45,46 future investigations of the association of disturbed AZGP1 expression and lipid metabolism in SRF are also warranted.

There are several limitations in this study. First, only male mice were used in all experiments. Previous studies have reported significant differences between male and female mice in laser-induced CNV models, particularly regarding cytokine levels and lesion sizes.⁴⁷ Consequently, the influence of sex differences merits careful consideration in future research. Second, it is hard to isolate the antifibrotic effects of AZGP1 on SRF, as the lesions are partly vascularized. The two-stage laser method for SRF modeling may reflect the clinical characteristics of patients with nAMD more accurately and result in larger fibrotic lesions compared with the single-stage laser method.⁷ However, CNV can still be detected upon to 40 days post-laser coagulation.^22,48 Moreover, a recent study has shown that loss of AZGP1 also promoted angiogenesis in prostate cancer.⁴⁹ Consequently, alternative models of SRF exhibiting reduced CNV are necessary to further elucidate the antifibrotic role of AZGP1. Third, a more comprehensive approach is required regarding both the timing and methodology for exogenous administration of rAZGP1 other than intravitreal injection in mice. Recently reported novel peptides capable of penetrating cells have been shown to effectively deliver recombinant proteins into retinal cells in vivo.⁵⁰ Additionally, gene therapy via subretinal delivery of viral vectors has emerged as a promising strategy for treating ocular diseases in recent years.^51–54 Fourth, other cells that are involved in SRF, such as endothelial cells, pericytes, macrophages, and fibroblasts, are not discussed in this study, of which AZGP1 may directly or indirectly confer similar effects. Further studies are warranted to explore the role of AZGP1 with these cell types.

In summary, this study provided evidence supporting AZGP1 as a mediator of SRF and revealed that the level of AZGP1 in the RPE/choroid was significantly decreased in SRF mouse model. Our results also indicated that exogenous supplementation of AZGP1 protein to refine the intrinsic antifibrotic effects of RPE cells could be a novel strategy to limit SRF.