FOXM1 Participates in Scleral Remodeling in Myopia by Upregulating APOA1 Expression Through METTL3/YTHDF2

Abstract

Purpose: Apolipoprotein A1 (APOA1) is a potential crucial protein and treatment goal for pathological myopia in humans. This study set out to discover the function of APOA1 in scleral remodeling in myopia and its underlying mechanisms.

Methods: A myopic cell model was induced using hypoxia. Following loss- and gain-of function experiments, the expression of the myofibroblast transdifferentiation-related and collagen production-related factors Forkhead box M1 (FOXM1), APOA1, and methyltransferase-like 3 (METTL3) in the myopic cell model was examined by quantitative reverse transcription polymerase chain reaction (RT-qPCR) and western blotting. The proliferation and apoptosis were determined by Cell Counting Kit-8 assay and flow cytometry, respectively. Chromatin immunoprecipitation (ChIP) was employed to examine FOXM1 enrichment in the METTL3 promoter, methylated RNA immunoprecipitation (Me-RIP) to examine the N⁶-methyladenosine (m⁶A) modification level of APOA1, and photoactivatable ribonucleoside-enhanced crosslinking and immunoprecipitation (PAR-CLIP) to examine the binding between METTL3 and APOA1.

Results: Hypoxia-induced human scleral fibroblasts (HSFs) had high APOA1 and FOXM1 expression and low METTL3 expression. FOXM1 knockdown elevated METTL3 expression and downregulated APOA1 expression. FOXM1 was enriched in METTL3 promoter. APOA1 or FOXM1 knockdown or METTL3 overexpression reversed the hypoxia-induced elevation in vinculin, paxillin, and α-smooth muscle actin (α-SMA) levels and apoptosis and the reduction in collagen, type I, alpha 1 (COL1A1) level and cell proliferation in HSFs. METTL3 or YTH N⁶-methyladenosine RNA binding protein F2 (YTHDF2) knockdown or APOA1 overexpression reversed the impacts of FOXM1 knockdown on vinculin, paxillin, α-SMA, and COL1A1 expression and cell proliferation and apoptosis.

Conclusions: FOXM1 elevated the m⁶A methylation level of APOA1 by repressing METTL3 transcription and enhanced APOA1 mRNA stability and transcription by reducing the YTHDF2-recognized m⁶A methylated transcripts.

Currently, myopia is characterized as a refractive abnormality of the non-accommodated eye that has a spherical equivalence of –0.5 diopters or less.¹ This disease occurs as a result of genetic and environmental risk factors.² Myopia is distinguished by increased axial length of the eye and invokes diverse complications, such as scleral/choroidal thinning, peripheral deformation, glaucoma, cataract, macular hole, myopic foveoschisis, retinal detachment, myopic choroidal neovascularization, and dome-shaped macula.³ Considering the pathological complications and other related serious conditions of myopia, myopia is not only a negative factor for eye health, self-perception, and work/activity options but also a major contributor to blindness worldwide.⁴ Because scleral remodeling leads to excessive elongation of the eye, which is the basis of myopia,⁵ this study focuses on the molecular mechanisms related to scleral remodeling in myopia, hoping to propose novel insights for the management of myopia. 

Apolipoprotein A1 (APOA1) is the primary constituent of high-density lipoprotein and performs an essential function of reversing cholesterol transport.⁶ APOA1 exerts a role as a STOP signal through its ability to suppress excessive growth of the ocular axis.⁷ A previous study indicated that APOA1 is a potentially crucial protein and treatment target for pathological myopia in humans.⁸ Wen et al.⁹ indicated that the N⁶-methyladenosine (m⁶A) modifications have modulatory effects in the anterior capsule of the lens in patients with high myopia. SRAMP (sequence-based RNA adenosine methylation site predictor) database predictions indicate that APOA1 has multiple m⁶A modification sites, suggesting that APOA1 might undergo m⁶A modification in myopia. The m⁶A modifications are modified by the m⁶A methyltransferases (such as methyltransferase-like 3 [METTL3]), removed by the demethylases (such as fat mass and obesity-associated protein), and recognized by the m⁶A binding proteins (e.g., YTH domain family 1, 2, and 3 [YTHDF1, YTHDF2, YTHDF3]).¹⁰ Both METTL3 and YTHDF2 have been reported in ocular diseases. For example, METTL3 affects angiogenesis in a hypoxia-induced retinopathy model.¹¹ Histone lactylation triggers the occurrence of ocular melanoma through promotion of YTHDF2 expression.¹² Forkhead box M1 (FOXM1), a transcription factor of the Forkhead family, participates in normal cell proliferation.¹³ Moreover, a prior study observed FOXM1 expression in the entire ciliary marginal zone of the developing Xenopus retina,¹⁴ suggesting that FOXM1 might exert a critical influence on myopia. 

Taken together, we speculated that FOXM1 functions in myopia by modulating METTL3/YTHDF2 to affect APOA1 expression. Notably, Zhao et al.¹⁵ reported that visual signals associated with myopia led to decreased permeability of choroidal capillaries and decreased blood flow. Reduced choroidal capillary permeability and blood flow led to reduced levels of oxygen and nutrient supply to the sclera, resulting in scleral hypoxia and ultimately excessive elongation of the ocular axis, suggesting scleral hypoxia as a control target for myopia.¹⁶ Therefore, most of the cell models of myopia studied at present use hypoxia treatment of human scleral fibroblasts to simulate the myopia extracellular matrix (ECM) remodeling phenomenon.¹⁶ Based on the previous data, we constructed a myopic cell model using hypoxia and then ascertained the function played by FOXM1 in scleral remodeling of myopia through APOA1 by orchestrating METTL3/YTHDF2. 

Materials and Methods

Bioinformatics Analysis

The modification sites of APOA1 m⁶A were forecast using the SRAMP database (http://www.cuilab.cn/sramp). First, the GSE136701 dataset was downloaded through the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo). Next, the differentially expressed genes of three patients with high myopia and three control patients with hyperopia were analyzed using the limma package for R (R Foundation for Statistical Computing, Vienna, Austria), with |logFC| > 1 and P < 0.05 as screening criteria. The proteins interacted with were screened using the Starbase database (https://starbase.sysu.edu.cn/). The protein interaction was analyzed using the STRING database (https://cn.string-db.org/), and the results were plotted using CytoScape software. The datasets used or analyzed during the current study are available from the corresponding author on reasonable request. 

Cell Culture and Treatment

Human scleral fibroblasts (HSFs; Procell, Wuhan, China) were cultivated with Gibco Dulbecco's Modified Eagle's Medium (Thermo Fisher Scientific, Waltham, MA, USA) encompassing 1% penicillin and streptomycin and 10% fetal bovine serum. After vimentin and keratin identification, HSFs at the fourth passage were adopted for the experiments. Cells were exposed to 21% O₂ as the normoxia (NO) group and to 5% O₂ as the hypoxia (HO) group. The medium was replenished on the second day of passaging. The cells were placed in a conventional CO₂ incubator (Heal Force, Shanghai, China) for 2, 4, 7, and 10 hours to determine the optimal time of hypoxic action. The cells received lysis for follow-up experiments after reaching the corresponding predetermined time. 

Logarithmically growing HSFs underwent 0.25% trypsin digestion and dilution before being seeded into 96-well plates. When a 60% to 70% confluence was achieved, the cells were transfected or cotransfected with lentivirus (LV)-METTL3, LV-APOA1, LV-negative control (NC), small interfering RNA (si)-APOA1, si-FOXM1, si-YTHDF1, si-YTHDF2, and si-NC plasmids (75 nM; GenePharma, Shanghai, China) following the directions for the Invitrogen Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific). Subsequently, the transfected cells were subjected to the appropriate normoxic or hypoxic treatment for subsequent experimentations. The cells were classified as follows: NO (normoxia-treated HSFs), HO (hypoxia-treated HSFs), NO+NC (normoxia-treated HSFs transfected with si-NC), NO+si-APOA1 (normoxia-treated HSFs transfected with si-APOA1), NO+si-FOXM1 (normoxia-treated HSFs transfected with si-FOXM1), HO+NC (hypoxia-treated HSFs transfected with si-NC), HO+si-APOA1 (hypoxia-treated HSFs transfected with si-APOA1), HO+si-FOXM1 (hypoxia-treated HSFs transfected with si-FOXM1), NO+LV-NC (normoxia-treated HSFs transfected with LV-NC), NO+LV-METTL3 (normoxia-treated HSFs transfected with LV-METTL3), HO+LV-NC (hypoxia-treated HSFs transfected with LV-NC), HO+LV-METTL3 (hypoxia-treated HSFs transfected with LV-METTL3), HO+si-FOXM1+si-METTL3 (hypoxia-treated HSFs cotransfected with si-FOXM1 and si-METTL3), HO+si-FOXM1+LV-APOA1 (hypoxia-treated HSFs cotransfected with si-FOXM1 and LV-APOA1), or HO+si-FOXM1+si-YTHDF2 (hypoxia-treated HSFs cotransfected with si-FOXM1 and si-YTHDF2). The knockdown plasmid sequences are listed in Supplementary Table S1. 

Western Blotting

Total proteins were derived from the cells using the radioimmunoprecipitation assay lysis solution encompassing phenylmethylsulfonyl fluoride (PMSF). Following 30-minute incubation on ice, the proteins were centrifuged at 4°C and 8000g for 10 minutes to attain the supernatant. A bicinchoninic acid kit was applied for measuring the protein concentration. Then, the proteins (50 µg) were dissolved in 2× sodium dodecyl sulfate (SDS) loading buffer, boiled for 5 minutes at 100°C, and subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE). After the proteins were transferred to polyvinylidene fluoride membranes by the wet transfer method, 5% skimmed-milk powder was employed to block the membranes at room temperature for 1 hour. The membranes then underwent overnight incubation with diluted primary antibodies (Abcam, Cambridge, UK) against vinculin (ab219649, 1:1000), paxillin (ab32084, 1:1000), α-smooth muscle actin (α-SMA, ab7817, 1:1000), collagen, type I, alpha 1 (COL1A1, ab138492, 1:1000), YTHDF2 (ab220163, 1:1000), transforming growth factor-beta 1 (TGF-β1, ab215715, 1:1000), and alpha-tubulin (ab7291, 1:10,000) at 4°C. Subsequent to washing, the membranes received 2 hours of incubation with horseradish peroxidase–labeled secondary antibody immunoglobulin G (IgG, ab205718, 1:5000; Abcam) at room temperature and development with electrochemical luminescence. The membranes were scanned and analyzed on a gel imager, and the bands in the western blotting images were quantified in grayscale using Image J (National Institutes of Health, Bethesda, MD, USA) with alpha-tubulin as an internal reference. 

Immunofluorescence Staining

Cells underwent 30 minutes of 4% paraformaldehyde fixation, phosphate-buffered saline with Tween 20 (PBST) washing, and 15 minutes of 0.5% Triton X-100 permeabilization. The cells then underwent 1 hour of blocking with 5% bovine serum albumin followed by overnight incubation with primary antibodies (1:200, Abcam) against vimentin (ab92547) and keratin (ab185627) at 4°C. Following PBST washing, the cells were subjected to 1 hour of incubation at room temperature with fluorescent secondary antibody and nuclei staining with 4′,6-diamidino-2-phenylindole. Following a PBST washing, the slides were sealed with fluorescent mounting medium for photography under a fluorescent microscope. 

Quantitative Reverse Transcription Polymerase Chain Reaction

Total RNA was isolated from cells using TRIzol Reagent (Life Technologies, Carlsbad, CA, USA), with a NanoDrop 2000 ultra-micro spectrophotometer (Thermo Fisher Scientific) to determine RNA concentration and purity. The reverse transcription reaction was conducted in a PCR amplifier following the reverse transcription kit (11750150; Thermo Fisher Scientific) to synthesize cDNA templates. Specifically, 2 µg total RNA was added to a 0.5-mL microcentrifuge tube and supplemented with an appropriate amount of diethyl pyrocarbonate water to achieve a total volume of 11 µL. The tubes were filled with 10-µM Oligo(dT) Primer (12–18 µL) with gentle mixing and centrifugation. Following 10 minutes of heating at 70°C, the microcentrifuge tubes were immediately inserted into an ice bath for at least 1 minute followed by the addition of 10× PCR buffer, 25-mM MgCl, 10-mM dNTP mix, and 0.1-M dithiothreitol to form a 50-µL system, followed by gentle mixing, centrifugation, and 2 to 5 minutes of incubation at 42°C. The tubes were supplemented with 1 µL Invitrogen SuperScript II Reverse Transcriptase, incubated at 42°C for 50 minutes, and heated at 70°C for 15 minutes to terminate the reaction. The tubes were then inserted into ice, 1 µL RNase H was added, and the tubes were incubated for 20 minutes at 37°C to degrade the residual RNA. Real-time quantitative reverse transcription polymerase chain reaction (RT-qPCR) experiments were conducted following the instructions of the TransStart Green qPCR SuperMix kit (Transgen Biotech, Beijing, China) using a fluorescent qPCR analyzer (CFX Connect; Bio-Rad, Hercules, CA, USA), in which glyceraldehyde-3-phosphate dehydrogenase served as a normalizer for RNA expression. The reaction was pre-denatured for 10 minutes at 95°C and then denatured for 10 seconds at 95°C, annealed for 20 seconds at 60°C, and extended for 34 seconds at 72°C for 40 cycles. Data analysis was conducted using the 2^−ΔΔCt method.¹⁷ The primer sequences are listed in the Table. 

Table.

 

Primer Sequences

 

Measurement of RNA Lifetime

The RNA stability assay was performed as previously described.¹⁸ HSFs (1 × 10⁵ cells) with 50% confluence were seeded onto 10-cm plates, followed by 24-hour incubation in a conventional CO₂ incubator. After that, the cells on a 10-cm plate were reseeded onto three 6-cm plates. After 48 hours, the cells were collected by trypsin digestion, and actinomycin D was added to the cells at a concentration of 5 mg/mL at 6 hours, 3 hours, and 0 hour. Total RNA was purified using the RNeasy Kit with DNase-I digestion procedures attached to the column. RT-qPCR was conducted to examine the RNA quantity. 

Chromatin Immunoprecipitation

The EZ-ChIP kit (MilliporeSigma, Bedford, MA, USA) was utilized for chromatin immunoprecipitation (ChIP) assays. Cells were cultivated for 36 hours and fixed with formaldehyde, and the fixation was terminated using glycine. The cells were then scraped and centrifuged to obtain the cell precipitate. The precipitate was suspended with a cell lysis solution containing PMSF and centrifuged before the supernatant was discarded. After sonication in an ice-water bath to precipitate the DNA, 10% of the supernatant was taken as control and the remaining 90% of the lysate was incubated with anti-FOXM1 antibody (ab240753; Abcam) or normal rabbit IgG antibody (5946; Cell Signaling Technologies) coupled with magnetic beads and then centrifuged. DNA bound to FOXM1 protein was eluted with eluent and then purified using a DNA purification kit (Beyotime, Shanghai, China), followed by RT-qPCR. 

Cell Counting Kit-8 Assay

HSFs were seeded onto 96-well plates with 100 µL diluted cell suspension (1 × 10⁶ cells/mL) per well; there was a set of three replicates. Following 24 hours, 48 hours, or 72 hours of incubation in the incubator, each well of cells was incubated for 2 hours with 10 µL Cell Counting Kit-8 (CCK-8) reagent (Abcam) in the incubator. A microplate reader (BioTek Instruments, Winooski, VT, USA) was applied for measuring the absorbance value at 450 nm, and the average value of the three replicates was calculated with analysis of cell growth. 

Flow Cytometry

Following 48 hours of transfection, cells were digested with EDTA-free 0.25% trypsin (YB15050057; YBio, Shanghai, China), collected in flow tubes, and centrifuged, and the supernatants were removed. Following three washes with cold PBS, the cells were centrifuged, and the supernatants were removed. This assay was conducted following the specifications of the Annexin V-FITC Apoptosis Kit Plus (K201-100; BioVision, Milpitas, CA, USA). The cells were resuspended in the staining solution, shaken, mixed, incubated at room temperature for 15 minutes, and shaken and mixed with 1 mL HEPES buffer solution (PB180325; Procell). Apoptosis was evaluated by flow cytometry. 

Methylated RNA Immunoprecipitation

Total RNA was isolated from HSFs by the TRIzol method, and mRNA in the total RNA was isolated and purified using the PolyATtract mRNA Isolation System (A-Z5300; A&D Technology Corporation, Beijing, China). The immunoprecipitation (IP) buffer (2-mM EDTA; 140-mM NaCl; 20-mM Tris, pH 7.5; and 1% Nonidet P-40) was incubated for 1 hours with anti-m⁶A antibodies (1:500, ab151230; Abcam) or anti-IgG antibodies (ab109489, 1:100; Abcam) and protein A/G magnetic beads for binding. The isolated and purified mRNA and magnetic bead-antibody complexes were added to the IP buffer encompassing ribonuclease and protease inhibitors for overnight reaction at 4°C. The RNA was eluted with elution buffer and purified using phenol–chloroform extraction, and RT-qPCR was employed for APOA1 analysis.^19,20 The primer sequences are listed in the Table. 

Photoactivatable Ribonucleoside-Enhanced Crosslinking and Immunoprecipitation

HSFs underwent 14 hours of incubation with 200 mm 4-thiopyridine (Sigma-Aldrich, St. Louis, MO, USA) followed by cross-linking with 0.4 J/cm² at 365 nm. Following lysis, the cells were immunoprecipitated with METTL 3 antibodies (5 and 3 mg, respectively) at 4°C. The precipitated RNA was labeled with [γ-³²P]-adenosine triphosphate and visualized by radioautography. The photoactivatable ribonucleoside (PAR) fragments were digested by proteinase K to remove proteins, and RNA was isolated for qRT-PCR to examine APOA1 expression. 

Statistical Analysis

Prism 8.0 (GraphPad, Boston, MA, USA) was utilized for the statistical analysis, and all data are stated as mean ± standard deviation (SD). Unless otherwise specified, t-tests were performed to compare data between two groups, and one-way ANOVA tests were performed to compare data among multiple groups with Tukey's multiple comparison tests. P < 0.05 was considered statistically significant. 

Results

METTL3 Expression was Reduced in Hypoxia-Treated HSFs

Predictions of the m⁶A modification sites of APOA1 by the SRAMP database (http://www.cuilab.cn/sramp) indicated the presence of multiple m⁶A modification sites on APOA1 (Fig. 1A), suggesting that APOA1 might experience m⁶A modification in myopia. HSFs were treated with hypoxia to construct a myopic cell model. Immunofluorescence results indicated that the cultivated HSFs were positive for vimentin and negative for keratin (Fig. 1B), thus confirming that the cells were HSFs. After hypoxia treatment, myofibroblast transdifferentiation and collagen production in HSFs were examined using western blotting. The results indicated that protein expression of the focal adhesion proteins vinculin and paxillin and the myofibroblast marker α-SMA was considerably augmented, but COL1A1 protein expression was appreciably reduced in hypoxia-treated HSFs compared with the normoxia-treated cells (Fig. 1C), suggesting that the myopic cell model was successfully constructed using hypoxia. RT-qPCR results showed that METTL3 expression was downregulated in hypoxia-treated HSFs as compared to normoxia-treated HSFs (Fig. 1D). Methylated RNA immunoprecipitation (Me-RIP) assay of the m⁶A modification level of APOA1 in HSFs revealed a considerable reduction in the m⁶A modification level of APOA1 in hypoxia-treated HSFs compared with normoxia-treated HSFs (Fig. 1E). The results of photoactivatable ribonucleoside-enhanced crosslinking and immunoprecipitation (PAR-CLIP) assays illustrated that, compared to normoxia treatment, hypoxia treatment resulted in a substantial decrease in APOA1 expression pulled down by METTL3 antibodies (Fig. 1F). These results indicate that METTL3 mediated m⁶A modification of APOA1 in the hypoxia-induced myopic cell model. 

Figure 1.

 

Hypoxia treatment reduced METTL3 levels in HSFs. (A) SRAMP database prediction of m⁶A modification sites of APOA1. (B) Immunofluorescence evaluation of vimentin and keratin in HSFs. (C) Western blotting examination of myofibroblast transdifferentiation and collagen production in hypoxia-treated HSFs with one-way ANOVA to confirm P values and Tukey's multiple comparison tests. (D) RT-qPCR examination of mRNA expression of m⁶A-related factor in hypoxia- and normoxia-treated HSFs. (E) Me-RIP examination of APOA1 m⁶A modification levels in hypoxia- and normoxia-treated HSFs. (F) PAR-CLIP examination of the binding between METTL3 and APOA1 mRNA. Unless otherwise stated, the t-test was applied to confirm the P values; N = 3.

YTHDF2 Increases APOA1 Expression in Hypoxia-Treated HSFs

Next, we further examined whether APOA1 influences scleral remodeling in myopia through m⁶A methylation regulation. First, si-APOA1 was transfected into HSFs, and RT-qPCR displayed a dramatic decrease in APOA1 expression, suggesting that the system designed to interfere with APOA1 expression was successfully constructed (Fig. 2A). Western blotting demonstrated that si-APOA1 transfection markedly diminished vinculin, paxillin, and α-SMA protein levels and elevated COL1A1 protein levels in HSFs after hypoxia treatment (Fig. 2B). Moreover, LV-METTL3 transfection also reversed the hypoxia-induced elevation in vinculin, paxillin, and α-SMA protein levels and reduction in COL1A1 protein levels in HSFs (Fig. 2C). RT-qPCR and western blotting showed that LV-METTL3 transfection resulted in a pronounced reduction in APOA1 expression in HSFs after hypoxia treatment (Figs. 2D, 2E). 

Figure 2.

 

APOA1 expression was upregulated by YTHDF2 in hypoxia-treated HSFs. (A) RT-qPCR to examine APOA1 expression with t-tests to confirm P values. (B) Western blotting to evaluate myofibroblast transdifferentiation and collagen production in si-APOA1-transfected and hypoxia- or normoxia-treated HSFs. (C) Western blotting to evaluate myofibroblast transdifferentiation and collagen production in LV-METTL3-transfected and hypoxia- or normoxia-treated HSFs. (D, E) RT-qPCR (D) and western blotting (E) to detect APOA1 expression in LV-METTL3-transfected and hypoxia-treated HSFs. (F, G) RT-qPCR (F) and western blotting (G) to detect APOA1 expression in si-YTHDF1- or si-YTHDF2-transfected and hypoxia-treated HSFs. (H) PAR-CLIP assay to detect the interaction between YTHDF2 and APOA1. (I) detection of APOA1 mRNA degradation rate in si-YTHDF2-transfected and hypoxia-treated HSFs. Unless otherwise stated, one-way ANOVA was employed to confirm P values with Tukey's multiple comparison tests; N = 3.

To investigate whether m⁶A-methylated APOA1 is affected by YTHDF1 or YTHDF2, HSFs were transfected with si-YTHDF1 or si-YTHDF2. RT-qPCR and western blotting indicated that si-YTHDF1 transfection did not affect APOA1 expression, whereas si-YTHDF2 transfection apparently enhanced APOA1 mRNA and protein expression (Figs. 2F, 2G). The results of the PAR-CLIP experiment further confirmed the direct relationship between YTHDF2 and APOA1 (Fig. 2H). Moreover, si-YTHDF2 transfection lowered the degradation rate of APOA1 mRNA (Fig. 2I). These findings suggest that the reduction of m⁶A-methylated APOA1 transcripts recognized by YTHDF2 slowed the APOA1 degradation rate, thereby promoting APOA1 expression in hypoxia-treated HSFs. 

FOXM1 Indirectly Affects APOA1 Expression by Downregulating METTL3 Expression to Participate in Scleral Remodeling in Myopia

Differential gene expression microarrays in myopia were screened in the GEO database (https://www.ncbi.nlm.nih.gov/geo/), and the microarray GSE136701 was selected for differential gene analysis using the limma package in R. The proteins interacting with YTHDF2 were screened by the Starbase database (https://starbase.sysu.edu.cn/) and intersected with the differentially expressed genes in GSE136701, resulting in 19 differentially expressed genes that interact with YTHDF2 in myopia. These genes were analyzed for protein interactions using the STRING database (https://cn.string-db.org/) followed by the use of CytoScape software to analyze the degree of interaction, which showed that FOXM1 had the highest degree of interaction with METTL3 (Fig. 3A), indicating that FOXM1 might exert a crucial role in myopia through METTL3. GSE136701 microarray analysis indicated that FOXM1 expression was substantially higher in myopic patients compared to controls (Fig. 3B). RT-qPCR and western blotting revealed considerably higher FOXM1 expression in hypoxia-treated HSFs than in normoxia-treated HSFs (Figs. 3C, 3D). ChIP assays indicated that FOXM1 was notably enriched in the METTL3 promoter region in hypoxia-treated HSFs (Fig. 3E). The aforementioned results suggest that FOXM1 might act in the hypoxia-induced myopic cell model through METTL3. 

Figure 3.

 

FOXM1 participates in scleral remodeling in myopia by indirectly affecting APOA1 expression through inhibition of METTL3 expression. (A) Differentially expressed genes that interact with METTL3 in myopia were analyzed by the STRING database with CytoScape software to analyze the degree of interaction. (B) GSE136701 microarray to analyze FOXM1 expression in myopia patients. (C, D) RT-qPCR (C) and western blotting (D) to detect FOXM1 expression in hypoxia- or normoxia-treated HSFs. (E) ChIP assay to evaluate the enrichment level of FOXM1 in the METTL3 promoter region. (F, G) RT-qPCR (F) and western blotting (G) to detect FOXM1, METTL3, and APOA1 expression in si-FOXM1-transfected and hypoxia-treated HSFs. (H) Western blotting to assess myofibroblast transdifferentiation and collagen production in si-FOXM1-transfected and hypoxia-treated HSFs with one-way ANOVA to confirm P values and Tukey's multiple comparison tests. Unless otherwise stated, the t-test was applied to confirm the P values; N = 3.

HSFs were then transfected with si-FOXM1 and treated with hypoxia. RT-qPCR and western blotting results showed that si-FOXM1 transfection decreased FOXM1 and APOA1 expression and upregulated METTL3 expression (Figs. 3F, 3G). In addition, after si-FOXM1 transfection, there was no significant change in YTHDF2 expression (Supplementary Fig. S1), which suggested that APOA1 relied on the m⁶A modification of METTL3 and was recognized by YTHDF2, thus affecting the stability of APOA1. Furthermore, si-FOXM1 transfection reversed the hypoxia-induced elevation in vinculin, paxillin, and α-SMA protein levels and reduction in COL1A1 protein levels in HSFs (Fig. 3H). These results suggest that FOXM1 indirectly affects APOA1 expression by downregulating METTL3 expression to participate in scleral remodeling in myopia. 

Knockdown of FOXM1 Reverses the Inhibition of Proliferation and Promotion of Apoptosis in HSFs by Hypoxia Treatment Via the METTL3/APOA1 Axis

Cell proliferation and apoptosis were examined by CCK-8 assay and flow cytometry, which showed that hypoxia treatment inhibited proliferation but promoted apoptosis of HSFs, but these trends were abolished by transfection with si-FOXM1 (Figs. 4A–4C). In addition, regarding the above indicators, si-APOA1 or LV-METTL3 transfection exerted the same effects as si-FOXM1 transfection (Figs. 4D–4F). These findings indicate that knockdown of FOXM1 reverses the inhibition of proliferation and promotion of apoptosis in HSFs by hypoxia treatment via the METTL3/APOA1 axis. 

Figure 4.

 

Knockdown of FOXM1 orchestrated the METTL3/APOA1 axis to abrogate the influences of hypoxia treatment on proliferation and apoptosis of HSFs. (A) CCK-8 assay to determine the proliferation of HSFs treated with normoxia, hypoxia, or hypoxia + si-FOXM1, with two-way ANOVA to confirm P values. (B, C) Flow cytometry to determine the apoptosis of HSFs treated with normoxia, hypoxia, or hypoxia + si-FOXM1. (D) CCK-8 assay to determine the proliferation of HSFs treated with normoxia, hypoxia, or hypoxia + si-APOA1, or hypoxia + LV-METTL3, with two-way ANOVA to confirm P values. (E, F) Flow cytometry to determine the apoptosis of HSFs treated with normoxia, hypoxia, hypoxia + si-APOA1, or hypoxia + LV-METTL3. In C and F, one-way ANOVA was employed to confirm P values. The Tukey's test was used for post hoc multiple comparisons for all assays; N = 3.

FOXM1 Participates in Scleral Remodeling in Myopia by Regulating APOA1 Expression Through METTL3/YTHDF2

To further confirm whether FOXM1 affected APOA1 m⁶A methylation levels through METTL3 and regulated APOA1 expression by YTHDF2 specifically recognizing transcripts of APOA1 m⁶A methylation, thereby affecting scleral remodeling in myopia, HSFs were transfected or cotransfected with si-FOXM1, LV-APOA1, si-METTL3, and si-YTHDF2 and treated with hypoxia. Western blotting revealed that si-METTL3, si-YTHDF2, or LV-APOA1 transfection counteracted the si-FOXM1 transfection-induced decrease in vinculin, paxillin, and α-SMA protein expression and elevation in COL1A1 protein expression in hypoxia-treated HSFs (Fig. 5A). Moreover, si-METTL3, si-YTHDF2, or LV-APOA1 transfection abolished the reduction of apoptosis induced by si-FOXM1 transfection in hypoxia-treated HSFs (Fig. 5B). A prior study demonstrated that TGF-β1 was one of the cytokines most closely related to the formation of myopia.²¹ Therefore, we examined the expression of TGF-β1. Western blotting results revealed that, compared with the NO group, TGF-β1 expression was signally reduced in the HO group, whereas si-FOXM1 transfection elevated the expression of TGF-β1. Additionally, si-METTL3, si-YTHDF2, or LV-APOA1 transfection could reverse the increased effect of si-FOXM1 on TGF-β1 expression and inhibit the expression of TGF-β1 (Fig. 5C). These results indicated that FOXM1 elevates the m⁶A methylation level of APOA1 by repressing METTL3 transcription and enhances APOA1 mRNA stability and transcription by reducing the YTHDF2-recognized m⁶A methylated transcripts, thereby affecting scleral remodeling in myopia by downregulating TGF-β1 expression. 

Figure 5.

 

FOXM1 orchestrated APOA1 expression through METTL3/YTHDF2 to participate in scleral remodeling in myopia. (A) Western blotting to determine myofibroblast transdifferentiation and collagen production in HSFs. (B) Flow cytometry to determine the apoptosis of HSFs. (C) Western blotting to determine the protein expression of TGF-β1. One-way ANOVA was employed to confirm P values with Tukey's multiple comparison tests; N = 3.

Discussion

Myopia, the most prevalent ocular condition in the world, has become a worldwide public health problem with its increasing prevalence over the last decades.²² The sclera, a very elastic connective tissue with a complicated structure, accounts for approximately 85% of the outer eyeball tunic and serves crucial roles in vision.²³ The crucial constituents of the sclera are HSFs and the ECM, where the fibroblasts take care of the synthesis of scleral ECM constituents such as collagen, elastic fibers, and proteoglycans.^24,25 Myopia pathogenesis occurs with a transdifferentiation of the cellular phenotype of the sclera from fibroblasts to myofibroblasts and alterations in the ECM.¹⁶ The scleral remodeling performs an essential role in myopia onset and progression and is mainly dependent on variations in the scleral ECM composition, where the accumulation of scleral collagen decreases with myopia progression and the breakdown rises.²⁶ Hence, it is desirable to expand our understanding of scleral remodeling, which may be beneficial for the understanding of myopia development and myopia management. This study discovered that FOXM1 participates in scleral remodeling by upregulating APOA1 expression via METTL3/YTHDF2. 

Hypoxia in the sclera results in scleral ECM remodeling, which contributes to myopia occurrence.¹⁵ Accordingly, this study employed hypoxia-induced HSFs to construct a myopic cell model. A prior publication illustrated that hypoxic exposure to 5% oxygen facilitated myofibroblast transdifferentiation of HSFs and downregulated type I collagen expression.¹⁶ Fibroblast transdifferentiation generates contractile myofibroblasts that are recognized by particular biomarkers such as vimentin, paxillin, and α-SMA.^27,28 Consistently, the present study also discovered that hypoxia resulted in enhancement in myofibroblast transdifferentiation and diminishment in type I collagen expression as evidenced by upregulation of vinculin, paxillin, and α-SMA protein expression and downregulation of COL1A1 protein expression in HSFs. Moreover, hypoxia treatment inhibited proliferation but promoted apoptosis of HSFs. 

APOA1 has beneficial roles in heart, diabetes, atherosclerosis, thrombosis, neurological, and cancer diseases.²⁹ APOA1 is differentially expressed in the vitreous and is involved in ocular overgrowth.³⁰ During the restoration of induced myopia, APOA1 mRNA and protein expression was remarkably elevated in the choroid of chicks’ eyes.³¹ Xue et al.⁸ stated that APOA1 was upregulated in patients with pathological myopia and was a potential treatment target for human pathological myopia. In agreement with that study, the present study revealed that APOA1 was highly expressed in the hypoxia-induced myopic cell model. Moreover, silencing of APOA1 abrogated the influence of hypoxia on myofibroblast transdifferentiation and type I collagen expression. Database predictions identified multiple m⁶A modification sites on APOA1. Correspondingly, this study revealed that the m⁶A modification level of APOA1 was dramatically decreased in the hypoxia-induced myopia model. Of note, METTL3 and YTHDF2 were substantially diminished in patients with nuclear cataract and high myopia relative to patients with pure nuclear cataract.⁹ The present study observed low expression of METTL3 in the hypoxia-induced myopia model and inversely targeted APOA1. Overexpression of METTL3 also reversed the hypoxia-induced elevation in vinculin, paxillin, and α-SMA protein levels and reduction in COL1A1 protein level in HSFs. In addition, knockdown of YTHDF2 enhanced APOA1 mRNA and protein expression and slowed down the degradation of APOA1. Collectively, the hypoxic myopia model reduced METTL3 and YTHDF2, leading to elevated m⁶A methylation levels of APOA1, reduced recognized m⁶A methylated transcripts, and enhanced APOA1 mRNA stability and transcription. Database analysis identified that FOXM1 was differentially expressed in myopia and could interact with METTL3. FOXM1, a transcription factor of the conserved FOX family, is a polyfunctional oncoprotein and a potent biomarker for poor prognosis in numerous human neoplasms.³² A study uncovered that thioredoxin suppresses FOXM1 and paxillin expression to suppress the metastasis of nasopharyngeal carcinoma cells.³³ In renal interstitial fibrosis, downregulation of FOXM1 restrains the epithelial-to-mesenchymal transition, as evidenced by decreased α-SMA expression.³⁴ However, we have found no publications describing the influence of FOXM1 on ocular diseases. The present study disclosed that FOXM1 was highly expressed and bound to METTL3 in the hypoxia-induced myopia model. Silencing of FOXM1 upregulated METTL3 to diminish APOA1 expression. Silencing of APOA1 also counteracted the influence of hypoxia on myofibroblast transdifferentiation, type I collagen expression, proliferation, and apoptosis in HSFs, further reversed by silencing of METTL3 or YTHDF2 or overexpression of APOA1. 

In summary, this work is the first, to the best of our knowledge, to uncover the involvement of FOXM1 in scleral remodeling in myopia and to find that FOXM1 restrained myofibroblast transdifferentiation and enhanced type I collagen production in HSF cells by upregulating APOA1 expression via METTL3/YTHDF2. However, there are some limitations of our study. First, our study focused only on the mechanism between the FOXM1/METTL3/APOA1 axis and its effect on HSF biological phenotypes. It would be more valuable in the future to explore more potential interactions of the FOXM1/METTL3/APOA1 axis with other pathways or factors that may affect the remodeling process. Second, our experiments were conducted only at the cellular level. More data and animal studies are needed before the findings of this study can be translated into clinical applications. Overall, this study promises to expand our understanding of myopia and lead to the development of novel ideas for the management of myopia. 

Acknowledgments

This research was funded by the grants from Anhui Medical University Research Foundation (grant no. 2023xkj235); Scientific Research Foundation of Anhui Provincial Health Commission (grant no. AHWJ2023A20470) and Key Project of Natural Science in Colleges and Universities of Anhui Province (grant no. 2022AH052324). 

Disclosure: M. Xue, None; B. Li, None; Y. Lu, None; L. Zhang, None; B. Yang, None; L. Shi, None