Exosomal FOXL1 from bone marrow mesenchymal stem cells activates the METTL3/ATXN2L pathway to ameliorate high glucose-induced human retinal microvascular endothelial cell injury

Abstract

Background

Bone marrow mesenchymal stem cells (BMSCs) and their secreted exosomes have been shown to possess therapeutic potential in various diseases, including diabetic retinopathy (DR). Retinal microvascular endothelial cell (RMEC) injury is a key factor in DR, and understanding the underlying molecular mechanisms is crucial for the treatment of DR. The study investigated the role of MSC-derived exosomes in RMEC injury and the underlying mechanism.

Methods

Human retinal microvascular endothelial cells (HRMECs) were exposed to high glucose (HG) to establish an in vitro DR model. Exosomes were isolated from BMSCs using differential centrifugation and co-incubated with HRMECs for functional studies. mRNA expression of ataxin 2 like (ATXN2L), methyltransferase-like 3 (METTL3), and forkhead box L1 (FOXL1) was assessed by quantitative real-time polymerase chain reaction. Protein expression was evaluated by western blotting. Cell viability was measured with a cell counting kit-8 assay, and pro-inflammatory cytokines (TNF-α, IL-1β, and IL-6) were analyzed by enzyme-linked immunosorbent assays. Apoptosis was analyzed through flow cytometry. MDA levels, GSH-Px activity, and ROS levels were determined by colorimetric methods and fluorescence microscopy, respectively. The association of METTL3 with ATXN2L and FOXL1 was investigated using a dual-luciferase reporter assay and RNA immunoprecipitation assay.

Results

HG treatment increased the secretion of pro-inflammatory factors, apoptosis rate, and oxidative stress in HRMECs. BMSC-derived exosomes inhibited inflammation, apoptosis and oxidative stress in HRMECs by transferring FOXL1 into HRMECs. FOXL1 functioned as an RNA-binding protein of METTL3, which stabilized ATXN2L mRNA expression through m6A methylation in HRMECs. ATXN2L expression was reduced in DR patients’ serum and HG-treated HRMECs. Overexpression of ATXN2L mitigated the high glucose-induced inflammation, apoptosis, and oxidative stress in HRMECs.

Conclusion

Exosomal FOXL1 from BMSCs stabilized METTL3 to increase ATXN2L expression, thus offering a protective effect against high glucose-induced injury in HRMECs. This finding holds clinical significance for the development of targeted therapies for DR.

Introduction

Diabetic retinopathy (DR) is recognized as a serious and widespread complication of diabetes, emerging as a predominant cause of vision loss and blindness among adults between the ages of 25 and 65 [1]. Studies indicate that prolonged high blood sugar levels, hypertension, and dyslipidemia are risk factors that accelerate the progression of DR [2]. According to statistics from 2020, more than 103 million diabetic patients worldwide are afflicted with diabetic retinopathy, and this number is expected to rise to 160 million by 2045 [3]. Anti-vascular endothelial growth factor therapy is currently the mainstay treatment [4]. Although this method has become the standard of care, its treatment process may impose significant economic and psychological burdens on patients. Consequently, clinicians are faced with the important task of finding and utilizing new drugs to more effectively control and treat diabetic retinopathy. Against this backdrop, the development of safer and more effective treatments is of great significance for improving the quality of life for diabetic patients.

Bone marrow mesenchymal stem cells (BMSCs) possess remarkable stem cell properties, including the ability to self-replicate, express specific surface markers of bone marrow mesenchymal stem cells, and multilineage differentiation potential [5]. These characteristics make BMSCs a highly promising therapeutic resource. Exosomes are membrane-bound extracellular vesicles with a diameter ranging from 30 to 200 nanometers. They are capable of carrying critical information and macromolecules from their parent cells, thus playing a key role in intercellular communication [6]. Studies have shown that exosomes are secreted by almost all types of cells, including BMSCs [7]. It has been shown that exosomes derived from BMSCs contribute to the maintenance of blood glucose homeostasis and the slowing of the progression of diabetes and its complications [8]. Delving into the specific mechanisms of action of BMSC-derived exosomes in DR is of vital importance for the development of innovative treatment methods for DR.

The Forkhead box (FOX) protein family is a large superfamily of transcription factors and serves important parts in regulating cellular metabolism [9]. Many members of the FOX protein family act as intersections of various cellular signaling pathways, thus playing key roles at multiple stages of embryonic development [10]. Previous evidence has shown their regulation in diabetes mellitus [11]. FOXL1, as a conserved member of the FOX gene family, is a transcription factor that contains a typical winged-helix DNA-binding domain [10]. Studies have shown that FOXL1 is lowly expressed in gastric cancer and its overexpression promotes cancer cell proliferation and migration [12]. Additionally, FOXL1 is involved in the regulation of central nervous system development, where it can influence the proliferation and differentiation of neural cells by inhibiting the expression of the key protein Sonic Hedgehog in the Sonic Hedgehog signaling pathway [13]. Although the role of FOXL1 in multiple biological processes has been widely studied, there have been no reports to date on its specific role in the development of DR.

The n6-methyladenosine (m6A) modification is one of the most common chemical modifications on RNA and plays a significant role in metabolic diseases such as diabetes [14]. The regulatory mechanism of m6A modification involves three steps: “writing”, “reading”, and “erasing”, which are mediated by methyltransferases, reader proteins, and demethylases, respectively. Among them, methyltransferase-like protein 3 (METTL3) may play a key role in the progression and pathogenesis of diseases like diabetes and is considered a potential therapeutic target [15]. Ataxin 2 like (ATXN2L) has recently been found to be involved in the regulation of stress granule formation [16]. A previous study has revealed that ATXN2L may contribute to the onset of type 2 diabetes in young individuals [17]. Furthermore, reports suggest that ATXN2L is implicated in the advancement of diabetic complications, such as diabetic peripheral neuropathy [18]. However, there is little data regarding its role in DR.

Drawing on the aforementioned evidence, this study delved into the function of BMSC-derived exosomes in DR and the molecular mechanisms at play. The research revealed that exosomes derived from BMSCs were capable of delivering the FOXL1 protein to human retinal microvascular endothelial cells (HRMECs), thereby modulating the expression of METTL3 and ATXN2L. This process ultimately contributed to the mitigation of HRMEC damage. The findings underscore the therapeutic potential of BMSC-derived exosomes in the treatment of DR.

Materials and methods

Clinical samples

The blood samples were collected from 17 DR patients and 17 healthy individuals undergoing routine physical examinations at Henan Provincial People’s Hospital. The patients were diagnosed with DR by professional physicians. The collected blood samples were stored in a 4-degree refrigerator. The collection of blood samples was consented to by all participants, and the use of the blood samples for this study was agreed upon by the participants. The Ethics Committee of Henan Provincial People’s Hospital approved the study.

Cell culture and treatment

Human retinal microvascular endothelial cells (HRMECs) and BMSCs were provided by EK-Bioscience (Shanghai, China). These cells were cultured in DMEM (GENIA Bioscience, Beijing, China) added with 10% fetal bovine serum (Yuanye Bio-Technology, Shanghai, China) and 1% penicillin/streptomycin (Solarbio, Beijing, China) at 37˚C with 5% CO₂. HRMECs were seeded into cell culture dishes. After attachment, the cells were starved for 24 h in serum-free medium, followed by culturing for 24 h in medium containing 30 mM glucose (ybiotech, Shanghai, China) to establish a DR cell model. The control group cells were cultured with 5.5 mM glucose. Additionally, 30 mmol/L mannitol (Yuanye Bio-Technology) was added as an osmotic control group.

Cell transfection

ATXN2L overexpression plasmid (ATXN2L), METTL3 overexpression plasmid (METTL3), FOXL1 overexpression plasmid (FOXL1), the small hairpin RNAs against ATXN2L (sh-ATXN2L), METTL3 (sh-METTL3), and FOXL1 (sh-FOXL1), and controls (pcDNA and sh-NC) were provided by GenePharma (Shanghai, China). HRMECs were digested using trypsin (Solarbio) and counted before seeding them into 12-well plates. When the cells were observed under the microscope to be in good condition and the cell density reached approximately 60%, the transfection experiment could be initiated. The shRNAs or plasmids were added to Opti-MEM medium (Solarbio), gently mixed by pipetting, and then incubated for 5 min. Concurrently, Lipofectamine 2000 (Thermo Fisher, Waltham, MA, USA) was also added to Opti-MEM medium, gently mixed by pipetting, and then incubated for 5 min. The mixtures were combined, gently mixed by pipetting, and then incubated for 20 min. Subsequently, the mixtures were added to the corresponding wells, and the plates were gently shaken to ensure even distribution. The plates were then placed back in the incubator for 4–6 h. The cells were continued to culture in fresh complete medium containing 10% fetal bovine serum (Yuanye Bio-Technology) for 24–48 h.

Flow cytometry

An appropriate amount of trypsin (Solarbio) was added to BMSCs for digestion to obtain cell pellets, which were then resuspended in PBS. After cell counting, the cell concentration was adjusted, and the cells were incubated with the primary antibodies against CD45 (E-AB-F1137D, Elabscience, Wuhan, China), CD73 (E-AB-F1242D, Elabscience) and CD105 (E-AB-F1310D, Elabscience) at 4 °C for 20 min. The samples were subsequently washed with a flow cytometry staining buffer, followed by centrifugation at 300 g for 5 min to discard the supernatant. The cells were resuspended in flow cytometry staining buffer and analyzed using an AttuneTM NxT flow cytometer (Thermo Fisher).

HRMECs were collected by centrifugation at 4 °C at 300 g and washed twice with pre-cooled PBS. The cells were resuspended in Annexin-V Binding Buffer (Solarbio), and Annexin V-FITC (Solarbio) and propidium iodide (Solarbio) were added. After gentle mixing, the cells were incubated in the dark for 10 min. Upon completion of the incubation, Annexin-V Binding Buffer was added, and the mixture was gently mixed before being placed on ice. The samples were then analyzed using an AttuneTM NxT flow cytometer (Thermo Fisher).

Isolation of exosomes from BMSCs

When BMSCs reached approximately 60% confluence, the culture medium was replaced with serum-free DMEM (GENIA Bioscience). The cells were cultured for an additional 48 h, and the supernatant was collected. The cell supernatant was centrifuged at 1500 g for 20 min to remove dead cells. Further centrifugation at 10,000 g for 40 min was performed to remove cell debris and organelles. The supernatant was then transferred to ultracentrifuge tubes and centrifuged at 100,000 g for 1.5 h, after which the supernatant was discarded. The extracellular vesicle pellets were washed with sterile PBS, resuspended by pipetting, and centrifuged again at 100,000 g for 1.5 h. The supernatant was discarded, and the extracellular vesicle pellets were dissolved in an appropriate amount of sterile PBS. The solution was then sterilized through a 0.22 μm sterile filter and stored at -80 °C.

Transmission electron microscope (TEM) analysis of EVs

Twenty microliters of the freshly extracted sample were pipetted onto the surface of the test strip, and a sample-loaded copper grid was inverted onto the extracellular vesicles and incubated for 5 min. After removing the excess liquid, the copper grid was inverted onto a drop of 3% phosphotungstic acid solution (ybiotech) for an additional 5 min of staining. The copper grid was then placed under an incandescent lamp to dry, followed by observation and photography under a Tecnai T20 TEM (FEI, Hillsboro, Oregon, USA).

Nanoparticle tracking analysis (NTA) of EVs

The extracellular vesicles were diluted with PBS in appropriate proportions to achieve the optimal detection concentration (1.0 × 10⁸-2.5 × 10⁹ particles/mL). The diluted extracellular vesicles were injected into the sample chamber using a disposable syringe to avoid bubble formation. Five recordings of 30 s each were captured, and the data from at least 5000 individual particle tracks were analyzed every sample. The instrument performed particle counting and size analysis based on the Brownian motion of the extracellular vesicles. NTA 3.2 software (NanoSight, Malvern, UK) was used for data analysis.

Western blotting assay

RIPA buffer (Beyotime, Shanghai, China) was added to the exosomes, blood samples, and cell samples, and the samples were vortexed to ensure complete lysis. Tissue samples were subjected to ultrasonic disruption. After centrifugation, the supernatant was transferred to new EP tubes, and protein loading buffer (Beyotime) was added before boiling in water for 10 min. The gel was placed into the electrophoresis tank. Thirty micrograms of protein per well were loaded, with the volume calculated and added according to the determined protein concentration for each sample. Electrophoresis was run at 60 V for 40 min to allow the proteins to pass through the stacking gel. The voltage was then adjusted to 90 V, and electrophoresis was continued for 2 h. The current was set to 350 mA, and wet transfer was performed for 120 min. The primary antibody against ATXN2L (ab184834, 1:1000, Reactivity: Human, Source: Rabbit, Abcam), METTL3 (ab195352, 1:1000, Reactivity: Human, Source: Rabbit, Abcam), FOXL1 (2299-MSM2-P0, 1:100, Reactivity: Human, Source: Mouse, Thermo Fisher), TSG101 (MA1-23296, 1:2000, Reactivity: Human, Source: Mouse, Thermo Fisher), CD63 (PA5-92370, 1:200, Reactivity: Human, Source: Rabbit, Thermo Fisher), CD81 (MA5-32333, 1:2000, Reactivity: Human, Source: Rabbit, Thermo Fisher), or β-actin (MA1-140, 1:10000, Reactivity: Human, Source: Rabbit, Thermo Fisher) was applied to PVDF membranes (Millipore, Bradford, MA, USA) and incubated overnight in a 4 °C refrigerator. The secondary antibodies, Goat anti-Mouse IgG (31430, 1:8000, Thermo Fisher), and Goat anti-Rabbit IgG (31460, 1:20000, Thermo Fisher) were then applied to the PVDF membranes and incubated at 37 °C for 1 h. The developing solution (Beyotime) was added dropwise to the PVDF membranes, and the target protein expression was detected by exposing the bands. Antibodies were diluted using Tris-Buffered Saline with Tween-20 (Bei Jing Think-Far Technology Co.,Ltd, Beijing, China).

Quantitative real-time polymerase chain reaction (qRT-PCR)

Trizol reagent (Khayal Bio-Technology, Wuhan, China) was added to the blood or cell samples, and the mixtures were vortexed to ensure complete lysis. Subsequently, RNA was extracted using the RNAsimple kit (Tiangen, Beijing, China). After determining the RNA concentration, cDNA synthesis was performed according to the instructions provided with the FastKing RT Kit (Tiangen). The detection system was prepared with primers (shown in Table 1) and SuperReal PreMix Color (Tiangen), and PCR amplification was conducted on an IQ5 thermocycler (Bio-Rad, Hercules, CA, USA). Statistical analysis was performed based on the amplification curves and the number of cycles required for detection.

Table 1.

Primer sequences used in qRT-PCR

Name	Primers for qRT-PCR (5’-3’)
METTL3	Forward	ATCCCCAAGGCTTCAACCAG
	Reverse	GCGAGTGCCAGGAGATAGTC
ATXN2L	Forward	GATGACTGGGAGGACTGCG
	Reverse	CTAGTCCCTGCCCTAGGTGT
FOXL1	Forward	TTCAACGCTTCCCTGATGCT
	Reverse	GAACCGTGCCATTGTTTGCT
β-actin	Forward	CTTCGCGGGCGACGAT
	Reverse	CCACATAGGAATCCTTCTGACC

Cell counting kit-8 (CCK-8) assay

HRMECs were seeded into 12-well plates and subjected to various treatments before being digested with trypsin and counted. The treated cells were then seeded into 96-well plates. After 48 h, the medium was discarded, and fresh medium containing CCK-8 (Beyotime) was added to each well. The plates were then placed in a cell culture incubator for 3 h of light-free incubation. The absorbance of each well was measured at 450 nm using an RNE90002 microplate reader (REAGEN, Shenzhen, China).

Enzyme-linked immunosorbent assay (ELISA)

The commercial ELISA kits including Human TNF-α ELISA Kit (PT518, Beyotime), Human IL-1β ELISA Kit (PI305, Beyotime), and Human IL-6 ELISA Kit (PI325, Beyotime) were used for analysis. Test samples (cell supernatant collected from each group) and various working solutions (Beyotime) were prepared for the assays. The standard samples and test samples were added to 96-well plates, which were pre-coated, and incubated in a 37 °C incubator for 90 min. After discarding the liquid in the plates, the wash buffer was added. After washing, a biotinylated antibody working solution against TNF-α, IL-1β, or IL-6 (Beyotime) was added to each well and incubated for 60 min. Following another round of washing, the enzyme conjugate working solution (Beyotime) was added. The chromogenic agent (Beyotime) was added to each well. The plates were sealed with adhesive tape and incubated in a 37 °C incubator for 15 min. Finally, the stopping solution was added, and the OD values were measured using an RNE90002 microplate reader (REAGEN).

Oxidative stress analysis

The study detected malondialdehyde (MDA) levels, glutathione peroxidase (GSH-PX) activity, and reactive oxygen species (ROS) levels to determine oxidative stress in HRMECs using MDA Content Assay Kit (BC0020, Solarbio), GSH-PX detection reagent (QYS-23071, Qiyi Biotech, Shanghai, China), and Cellular ROS Assay kit (ab113851, Abcam), respectively. MDA and GSH-PX levels were analyzed using a colorimetric method, whereas ROS levels were assessed by an IX71 fluorescent microscope (Olympus, Tokyo, Japan).

Dual-luciferase reporter assay

HRMECs were digested with trypsin (Solarbio) and counted before being seeded into 24-well plates. Wild-type (WT) reporter plasmids (ATXN2L-Luc-WT, METTL3 WT, GenePharma) and mutant (MUT) reporter plasmids (ATXN2L-Luc-MUT1, ATXN2L-Luc-MUT2, ATXN2L-Luc-MUT3, METTL3 MUT, GenePharma) were transfected into the cells with METTL3 overexpression plasmids, FOXL1 overexpression plasmids, or pcDNA 3.1 vector. After 48 h, Passive Luciferase lysis buffer (Solarbio) was added, and the mixtures were centrifuged for 15 min. The supernatant was collected and used to analyze the luciferase activity values with a GloMax Explorer multifunction detector (YPH biotech, Beijing, China).

m6A methylated RNA Immunoprecipitation (MeRIP) assay

This experiment was conducted to detect the methylation level of ATXN2L in HRMECs transfected with METTL3 overexpression plasmid or pcDNA 3.1 vector. RNA extraction was performed as mentioned above, followed by RNA fragmentation. PGM beads (CLOUNSEQ, Shanghai, China) were gently resuspended using a pipette and washed with 1×IP buffer (CLOUNSEQ). The centrifuge tubes were placed on a magnetic stand until the solution was clear, and then the supernatant was discarded and incubated with the antibody against m6A (CLOUNSEQ) or IgG (CLOUNSEQ). The RNA mixtures were added to the prepared beads and incubated for 1 h at 4 °C with a rotating mixer. After washing the beads, the supernatant was collected, and the RNA was purified, followed by qRT-PCR analysis of ATXN2L mRNA levels.

RNA Immunoprecipitation (RIP) assay

HRMECs were lysed and collected using a complete RIP lysis buffer (Millipore). Magnetic beads (Millipore) were resuspended in RIP lysis buffer and washed on a magnetic stand. The supernatant was discarded, and either IgG (Abcam) or FOXL1 antibody (Thermo Fisher) was added to the beads, followed by incubation with rotation for 30 min. The supernatant was then discarded on the magnetic stand, and RIP wash buffer was added again to resuspend the bead-antibody mixture and placed on ice. After incubating the above bead-antibody complexes with the lysates, RNA was purified, and qRT-PCR was used to detect the level of METTL3 mRNA.

mRNA stabilization analysis

HRMECs were seeded into 6-well plates for 24 h. Actinomycin D (5 µg/mL, Abcam) was added to the wells and treated for 2, 4, or 6 h, respectively. The cells were then collected for RNA purification. ATXN2L mRNA levels were detected by qRT-PCR, and the stability of ATXN2L mRNA expression was measured by analyzing ATXN2L mRNA expression.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 8.0 software. The data were analyzed using Student’s t-test to assess differences between the two groups. One-way analysis of variance (ANOVA) and Tukey’s multiple comparison test were employed to evaluate differences among more than two groups. Prior to conducting these tests, the assumptions of normality and homogeneity of variance were assessed. Normality was evaluated using the Shapiro-Wilk test, and homogeneity of variance was assessed using Levene’s test. P < 0.05 was considered statistically significant.

Results

ATXN2L expression was downregulated in the serum of DR patients and HG-induced HRMECs

The study analyzed the expression of ATXN2L in DR patients through the GEO database (Accession, GSE251731). The results showed that its expression was downregulated in DR patients (Fig. 1A and B). The result was also confirmed by the qRT-PCR analysis of serum samples from DR patients and healthy volunteers. As shown in Fig. 1C, its mRNA expression was significantly decreased in DR patients when compared with the healthy volunteers. Moreover, its expression at mRNA and protein levels was also downregulated in HG-induced HRMECs (Fig. 1D and E). These data demonstrate that ATXN2L expression is downregulated in DR patients and high glucose-induced HRMECs.

Fig. 1.

ATXN2L expression was downregulated in the serum of DR patients and HG-induced HRMECs. (A and B) The expression of ATXN2L in DR patients was analyzed through the GSE251731 database. (C) ATXN2L mRNA expression was detected by qRT-PCR in the serum samples from healthy volunteers (N = 17) and DR patients (N = 17). (D and E) The mRNA and protein expression of ATXN2L were detected by qRT-PCR and western blotting assays, respectively, in HRMECs treated with NG, mannitol or HG (N = 3). *P < 0.05

ATXN2L overexpression ameliorated high glucose-induced promoting effects on inflammation, apoptosis and oxidative stress in HRMECs

The study then analyzed the effects of ATXN2L overexpression on HG-induced inflammation, apoptosis and oxidative stress in HRMECs. As shown in Fig. 2A, HG treatment inhibited ATXN2L protein expression, whereas the effect was relieved after ATXN2L overexpression. Subsequently, HG treatment inhibited cell viability and promoted the production of TNF-α, IL-1β, and IL-6, whereas these effects were attenuated after transfection with ATXN2L overexpression plasmid (Fig. 2B-E). Moreover, HG-induced cells displayed increased cell apoptotic rate, MDA levels and ROS levels and weakened GSH-Px activity, however, these effects were relieved after ATXN2L overexpression (Fig. 2F-I). These data demonstrate that the upregulation in ATXN2L expression could weaken HG-induced inflammation, apoptosis and oxidative stress in HRMECs.

Fig. 2.

ATXN2L overexpression ameliorated high glucose-induced promoting effects on inflammation, apoptosis and oxidative stress in HRMECs. HRMECs were divided into four groups, including negative control (normal glucose, NG) group, HG group, HG + pcDNA group, and HG + ATXN2L group. (A) ATXN2L protein expression was analyzed by western blotting assay (N = 3). (B) Cell viability was detected by CCK-8 assay (N = 3). (C-E) The levels of TNF-α, IL-1β, and IL-6 were analyzed by ELISAs (N = 3). (F) Cell apoptosis was analyzed by flow cytometry (N = 3). (G and H) MDA levels and GSH-Px activity were analyzed through the colorimetric method (N = 3). (I) ROS levels were detected by fluorescence microscopy (N = 3). *P < 0.05

METTL3 stabilized ATXN2L mRNA expression through m6A methylation modification

Subsequent investigation showed that there were three methylation modification sites in ATXN2L mRNA (Fig. 3A). These methylation modification sites and mutant sites are shown in Fig. 3B. The study constructed luciferase reporter plasmids based the predicted modification sites (site 1, site 2, and site 3) as well as mutant sites to identify the association of ATXN2L with a methylase METTL3, named ATXN2L-Luc-WT, ATXN2L-Luc-MUT1, ATXN2L-Luc-MUT2, and ATXN2L-Luc-MUT3. As shown in Fig. 3C, METTL3 overexpression significantly inhibited the luciferase activities of ATXN2L-Luc-WT, ATXN2L-Luc-MUT1, and ATXN2L-Luc-MUT3 in HRMECs, but it did not affect the luciferase activity of ATXN2L-Luc-MUT2, indicating that METTL3 regulated ATXN2L expression by methylated the site 2. Moreover, the results showed that the affinity of the m6A antibody to ATXN2L was enhanced after METTL3 overexpression (Fig. 3D). The data also indicated that METTL3 overexpression lengthened the transcript half-life of ATXN2L in HRMECs (Fig. 3E). As shown in Fig. 3F and G, METTL3 mRNA expression was downregulated in the serum of DR patients and HG-induced HRMECs. Comparatively, its protein expression was downregulated in HG-induced HRMECs when compared with the control groups (Fig. 3H). Further, the data showed that METTL3 overexpression upregulated ATXN2L expression at mRNA and protein levels, whereas these effects were relieved after transfection with shRNA of ATXN2L (Fig. 3I and J). These data suggest that METTL3 upregulates ATXN2L mRNA expression through m6A methylation modification.

Fig. 3.

METTL3 stabilized ATXN2L mRNA expression through m6A methylation modification. (A) The Cui Lab database was used to predict the methylation modification sites in ATXN2L mRNA. (B) The schematic diagram showed the m6A methylation sites of ATXN2L and the mutated sites of ATXN2L. (C and D) Dual-luciferase reporter assay and RIP assay were performed to identify the association of METTL3 and ATXN2L in HRMECs (N = 3). (E) The effect of METTL3 overexpression on the ATXN2L mRNA stabilization was analyzed by the actinomycin D assay (N = 3). (F) METTL3 mRNA expression was detected by qRT-PCR in the serum of DR patients (N = 17) and healthy volunteers (N = 17). (G) METTL3 mRNA expression was detected by qRT-PCR in normal glucose-treated HRMECs, mannitol-treated HRMECs, and high glucose-induced HRMECs (N = 3). (H) METTL3 protein expression was detected by western blotting assay in normal glucose-treated HRMECs, mannitol-treated HRMECs, and high glucose-induced HRMECs (N = 3). (I and J) HRMECs were divided into the pcDNA group, METTL3 group, METTL3 + sh-NC group, and METTL3-ATXN2L group, and the mRNA and protein expression of ATXN2L were detected by qRT-PCR and western blotting assay in HRMECs (N = 3). *P < 0.05

ATXN2L knockdown attenuated METTL3 overexpression-induced effects in HRMECs

The study then analyzed the association of METTL3 and ATXN2L in regulating inflammation, apoptosis and oxidative stress in HRMECs. The results showed that METTL3 overexpression promoted cell viability and inhibited the production of TNF-α, IL-1β, and IL-6 in HRMECs, whereas these effects were counteracted when ATXN2L knockdown (Fig. 4A-D). In addition, METTL3 overexpression led to decreased apoptotic rates, MDA levels and ROS levels and increased GSH-Px activity, but ATXN2L silencing restored these effects (Fig. 4E-H). These data demonstrate that METTL3 inhibits inflammation, apoptosis and oxidative stress by regulating ATXN2L in HRMECs.

Fig. 4.

ATXN2L knockdown attenuated METTL3 overexpression-induced effects in HRMECs. HRMECs were divided into the pcDNA group, METTL3 group, METTL3 + sh-NC group, and METTL3-ATXN2L group. (A) Cell viability was detected by CCK-8 assay (N = 3). (B-D) The levels of TNF-α, IL-1β, and IL-6 were analyzed by ELISAs (N = 3). (E) Cell apoptosis was analyzed by flow cytometry (N = 3). (F and G) MDA levels and GSH-Px activity were analyzed through the colorimetric method (N = 3). (H) ROS levels were detected by fluorescence microscopy (N = 3). *P < 0.05

METTL3 knockdown attenuated FOXL1 overexpression-induced effects in HRMECs

The study predicted RNA-binding proteins that could bind to the promoter region of METTL3 through the JASPAR database. Based on the scoring, the predicted RNA-binding proteins were ranked, and the top 10 were selected as candidates. Subsequently, by examining which of these RNA-binding proteins have not been reported to be related to METTL3, and have not been studied in DR disease nor predicted in databases, the study identified METTL3 as the sole candidate. The binding sites of FOXL1 for the promoter region of METTL3 are shown in Fig. 5A. Subsequently, the result showed that the FOXL1 antibody could significantly enrich METTL3 in HRMECs (Fig. 5B). Moreover, FOXL1 overexpression significantly increased the luciferase activity of wild-type reporter plasmid (METTL3 WT), but it did not affect the luciferase activity of mutant reporter plasmid (METTL3 MUT), as shown in Fig. 5C. Through the prediction of the GEO database (Accession, GSE251731), we discovered a low expression of FOXL1 in DR patients (Fig. 5D). Comparatively, FOXL1 mRNA expression was downregulated in the serum from DR patients when compared with those from healthy volunteers (Fig. 5E). The results also showed a low expression of FOXL1 at mRNA and protein levels in HG-induced HRMECs (Fig. 5F and G). As shown in Fig. 5H and I, FOXL1 overexpression upregulated the protein expression of METTL3 and ATXN2L in HRMECs, but these effects were relieved after METTL3 knockdown. The study showed that FOXL1 overexpression promoted cell viability and inhibited the production of TNF-α, IL-1β, and IL-6 in HRMECs, however, METTL3 depletion relieved these effects (Fig. 5J-M). Further data revealed that FOXL1 overexpression decreased apoptotic rate as well as MDA and ROS levels and promoted GSH-Px activity, whereas these effects were restored after METTL3 silencing (Fig. 5N-Q). Thus, FOXL1 inhibited inflammation, apoptosis and oxidative stress by regulating METTL3.

Fig. 5.

METTL3 knockdown attenuated FOXL1 overexpression-induced effects in HRMECs. (A) The schematic illustration showed the binding sites of FOXL1 for METTL3. (B and C) The RIP and dual-luciferase reporter assays were performed to identify the association of METTL3 and FOXL1 in HRMECs (N = 3). (D) The GEO database was used to predict the expression of FOXL1 in DR patients (N = 4) and healthy volunteers (N = 4). (E) FOXL1 mRNA expression was analyzed by qRT-PCR in the serum of DR patients (N = 17) and healthy volunteers (N = 17). (F and G) The mRNA and protein expression of FOXL1 were detected by qRT-PCR and western blotting assays in normal glucose-treated HRMECs, mannitol-treated HRMECs, and high glucose-induced HRMECs (N = 3). (H-Q) HRMECs were divided into four groups, including the pcDNA group, the FOXL1 group, the FOXL1 + sh-NC group, and FOXL1 + sh-METTL3 group. (H and I) METTL3 and ATXN2L protein expression was detected by western blotting assay (N = 3). (J) Cell viability was detected by CCK-8 assay (N = 3). (K-M) The levels of TNF-α, IL-1β, and IL-6 were analyzed by ELISAs (N = 3). (N) Cell apoptosis was analyzed by flow cytometry (N = 3). (O and P) MDA levels and GSH-Px activity were analyzed through the colorimetric method (N = 3). (Q) ROS levels were detected by fluorescence microscopy (N = 3). *P < 0.05

BMSC-derived exosomes might ameliorate the HG-induced cell injury by transferring FOXL1 into HRMECs

MSCs are multipotent cells that have been proposed for several disorders [19]. The study isolated exosomes from BMSCs and analyzed their effects on HG-induced cell injury. The morphological characteristics of these BMSCs are shown in Fig. 6A. We discovered that hematopoietic marker CD45 was negative in these BMSCs, but stem cell markers including CD73 and CD105 were positive in these BMSCs (Fig. 6B). The TEM analysis showed that the vesicles isolated from BMSCs had similar morphology with exosomes (Fig. 6C). Moreover, exosome markers including TSG101, CD63 and CD81 were mainly expressed in the isolated vesicles (Fig. 6D). Moreover, the size of these isolated vesicles mainly ranged around 130 nm (Fig. 6E). After co-cultured HRMECs with these exosomes labeled with 2 µM PKH67 for 24 h, we observed that green fluorescent mainly distributed in the cytoplasm of HRMECs (Fig. 6F), indicating that BMSC-derived exosomes could be transferred into HRMECs. The study also isolated exosomes from FOXL1-deficient MSCs and determined their effects on HG-induced inflammation, apoptosis and oxidative stress. In HG-induced HRMECs, BMSC-derived exosomes increased the protein expression of FOXL1, METTL3, and ATXN2L, promoted cell viability and inhibited the production of TNF-α, IL-1β, and IL-6, however, these effects achieved by FOXL1-deficient exosomes were less (Fig. 7A-E). Moreover, BMSC-derived exosomes inhibited cell apoptosis, MDA levels and ROS levels and promoted GSH-Px activity, but FOXL1-deficient exosomes displayed fewer effects on cell apoptosis, MDA levels, ROS levels and GSH-Px activity (Fig. 7F-I). Thus, BMSC-derived exosomes might ameliorate the HG-induced cell injury by transferring FOXL1 into HRMECs.

Fig. 6.

BMSC-derived exosomes could be transferred into HRMECs. (A) Analysis of morphological characteristics of BMSCs using a microscope. (B) Flow cytometry was used to quantify the expression of CD45, CD73, and CD105 (N = 3). (C-E) Identification of isolated exosomes through TEM, western blotting assay, and NTA (N = 3). (F) HRMECs were co-cultured with exosomes labeled with PKH67, and the location of exosomes was observed under a microscope (N = 3). *P < 0.05

Fig. 7.

BMSC-derived exosomes might ameliorate the HG-induced cell injury by transferring FOXL1 into HRMECs. HG-induced HRMECs were incubated with exosomes isolated from BMSCs transfected with sh-NC or sh-FOXL1, thus dividing into Exo + sh-NC group and Exo + sh-FOXL1 group. The cells that were not treated and those incubated with exosomes isolated from BMSCs that had not been transfected served as control groups, named the Control group and the Exo group, respectively. (A) The protein expression of FOXL1, METTL3, and ATXN2L was detected by western blotting assay (N = 3). (B) Cell viability was detected by CCK-8 assay (N = 3). (C-E) The levels of TNF-α, IL-1β, and IL-6 were analyzed by ELISAs (N = 3). (F) Cell apoptosis was analyzed by flow cytometry (N = 3). (G and H) MDA levels and GSH-Px activity were analyzed through the colorimetric method (N = 3). (I) ROS levels were detected by fluorescence microscopy (N = 3). *P < 0.05

Discussion

DR is a major cause of blindness and visual impairment, particularly posing a significant hazard to the working-age adult population [20]. Currently, anti-VEGF therapy has become the standard treatment for diabetic retinopathy. However, this method may impose considerable economic and psychological burdens on patients, and it requires long-term, repeated treatments [4]. Therefore, delving into the pathogenesis of diabetic retinopathy is of great importance for the development of new treatment strategies. Against this backdrop, the role of exosomes in diseases such as diabetic retinopathy has garnered widespread attention [21, 22]. This study aimed to analyze the role and potential mechanisms of exosomes derived from BMSCs in regulating the progression of DR in vitro. The result showed that BMSC-derived exosomal FOXL1 activated the METTL3/ATXN2L pathway to protect against high glucose-induced HRMEC injury.

FOXL1 is a transcriptional regulator that has been confirmed to play a role in disease progression. For example, Nakada et al.. reported that FOXL1 deficiency in function resulted in eye development [13]. FOXL1 aggravated the progression of lung cancer by the miR-1471/FOXL1 axis [23]. The current work was the first one to report its role in DR progression. The result showed that FOXL1 expression was downregulated in the serum of DR patients and high glucose-induced HRMECs. Our study also demonstrated that BMSC-derived exosomal FOXL1 ameliorated HG-induced HRMEC inflammation, apoptosis, and oxidative stress, highlighting a novel mechanism in the treatment of diabetic retinopathy. Unlike previous studies that have focused on the role of specific microRNAs, such as microRNA-486-3p [24] and miR-483-5p [25]in mediating the protective effects of BMSC-derived exosomes through pathways like the TLR4/NF-κB axis, our research identifies FOXL1 as a critical protein component of these exosomes. Additionally, while another study has shown that MSC-derived exosomes can alleviate senescence of retinal pigment epithelial cells through the activation of the PI3K/AKT pathway [26]our findings provide new insights into the molecular cargo of BMSC-derived exosomes, specifically implicating FOXL1 in the protection against HRMEC damage. This discovery expands the understanding of the diverse mechanisms by which BMSC-derived exosomes exert their therapeutic effects for diabetic retinopathy.

Previous research has demonstrated the involvement of METTL3 in various aspects of DR pathogenesis, such as inhibiting endothelial-mesenchymal transition via the lncRNA SNHG7/KHSRP/MKL1 axis [27]aggravating diabetes-induced pericyte dysfunction through the YTHDF2-PKC-η/FAT4/PDGFRA pathway [28]and ameliorating HG-triggered apoptosis and oxidative stress by upregulating PSAT1 [29]. Our findings uniquely positioned FOXL1 as an upstream regulator of METTL3 in HRMECs. METTL3 expression was downregulated in the serum of DR patients and HG-induced HRMECs. Importantly, our result showed that FOXL1 inhibited cell apoptosis, inflammation and oxidative stress by regulating METTL3 expression in HRMECs. This highlights a previously unexplored regulatory mechanism, wherein FOXL1 acts through METTL3 to exert its protective effects, thereby offering a new perspective on the molecular pathways involved in the pathophysiology of DR.

Further data revealed that METTL3 stabilized ATXN2L mRNA through the m6A methylation modification in HRMECs. ATXN2L, when upregulated by epidermal growth factor, promoted gastric cancer cell invasiveness and oxaliplatin resistance [30]. This finding highlights ATXN2L’s role in cancer progression and its potential as a therapeutic target for gastric cancer. ATXN2L’s ability to be recruited by HDAC3 was inhibited by jatrorrhizine, which ameliorated Schwann cell myelination in diabetic peripheral neuropathy mice [18]. This study underscores ATXN2L’s involvement in neuronal processes and its potential role in treating diabetic complications. However, there is no data regarding its effect on DR. Our data showed that ATXN2L was lowly expressed in DR patients and high glucose-induced HRMECs. Our study extended the understanding of ATXN2L’s role beyond cancer and neuronal processes to include its protective functions in diabetic retinopathy. Specifically, we demonstrated that ATXN2L overexpression inhibited cell apoptosis, inflammation, and oxidative stress by regulating METTL3 in HRMECs. Moreover, FOXL1 overexpression upregulated ATXN2L expression by modulating METTL3, and BMSC-derived exosomal FOXL1 regulates high glucose-induced injury of HRMECs through the METTL3/ATXN2L axis.

However, the study used an in vitro model of HRMECs exposed to high glucose to simulate DR. While this model is useful for understanding cellular mechanisms, it may not fully recapitulate the complex in vivo environment of the retina in diabetic patients. While the study suggests that FOXL1 stabilizes ATXN2L mRNA through m6A methylation, the exact mechanism by which ATXN2L then mitigates inflammation, apoptosis, and oxidative stress is not fully elucidated. Additional research is needed to understand the downstream targets and pathways involved.

Taken together, the findings suggest that BMSC-derived exosomes could be a novel therapeutic strategy for diabetic retinopathy by specifically targeting the FOXL1/METTL3/ATXN2L pathway. This could lead to the development of new treatments that reduce the burden of this complication in diabetic patients.

Author contributions

Chao Niu designed and performed the research; Daoquan Dong, Longjiang Cui, Yingli Dong, Wei Wang analyzed the data; Chao Niu wrote the manuscript. All authors read and approved the final manuscript.

Funding

None.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Written informed consents were obtained from all participants and this study was permitted by the Ethics Committee of Henan Provincial People’s Hospital.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.