FTO-modulated m6A demethylation and upregulation of KIF11 mRNA promotes retinal microvascular dysfunction in diabetic retinopathy Authors: Chenyu Lin, Chenyang Cao, Shujing Cui, Jinyi Guo, Yu Xin, Jun Shao Published: 2026 ABSTRACT BackgroundThe N6-methyladenosine (m6A) demethylation of mRNAs is critical for the progression of diabetic retinopathy (DR). Fat mass and obesity-associated protein (FTO) has been reported to be overexpressed in DR and is considered to be an important epitranscriptomic regulator (m6A eraser) in retinal angiogenesis. The principle of Kinesin family member 11 (KIF11) gene in DR is unclear.MethodsMethylated RNA immunoprecipitation sequencing (MeRIP-Seq) was used to screen abnormal m6A modification in vitreous body samples of proliferative DR (PDR) patients. CCK-8 and EdU assays were employed to detect the proliferation and DNA synthesis in human retinal microvascular endothelial cells (hRECs), while Transwell, wound healing, and tube formation assays were used to evaluate migration and angiogenesis. In addition, the levels of RNA and m6A modified mRNA were tested using quantitative RT-PCR (qRT-PCR) and methylated RNA immunoprecipitation-qRT-PCR (MeRIP-qRT-PCR); the protein levels were tested via western blot. RNA binding protein immunoprecipitation-qRT-PCR (RIP-qRT-PCR) was used to test the interaction between FTO and m6A modified mRNA. A 16-week diabetic retinopathy (DR) rat model was induced by streptozotocin (STZ), and 5 µL of AAV9 virus (1 × 1012 vg/mL) was intravitreally injected every 8 weeks. The retinas were extracted and subjected to Evans blue leakage and retinal trypsin digestion assays, while the retinal paraffin sections were subjected to hematoxylin and eosin (H&E) staining, immunohistochemical or immunofluorescence assays.ResultsMeRIP-Seq and RIP-RT-qPCR indicated that KIF11 should be a downstream target of FTO. Regarding qRT-PCR and MeRIP-qRT-PCR, in vitreous body samples from patients with PDR and hRECs under hyperglycemic conditions, expressed enhanced levels of FTO and KIF11, whereas the m6A methylation of KIF11 was decreased. The overexpression of KIF11 enhanced the proliferation, DNA synthesis, migration, wound healing and tube formation of hRECs by affecting the downstream PI3K/AKT/mTOR and β-catenin/c-myc pathways, while the knockdown of KIF11 had the opposite effects. In STZ-induced DR rats, overexpression of FTO and KIF11 significantly promoted retinal leakage, acellular capillary formation, pericyte loss, fibrosis, and gliosis in DR progression, whereas the knockdown of FTO and KIF11 mitigated these phenomena.ConclusionOur findings demonstrated that the m6A demethylation of KIF11 mRNA, as modulated by FTO, provides a potential target for the clinical therapy of DR retinal microvascular dysfunction.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-026-08356-y. Introduction Over recent decades, worldwide trends for the prevalence of diabetes have dramatically increased [1–3]. As a severe microvascular complication of diabetes, diabetic retinopathy (DR) is a leading cause of vision loss and blindness, and its increasing prevalence is considered as a considerable global health challenge [4–6]. During the progression of DR in a sustained hyperglycemic and hypoxic microenvironment, it is possible to detect pathological features in the retina, including angiogenesis, vascular leakage, fibrosis, and gliosis [7–10]. Investigating the molecular mechanisms underlying the development of DR is highly beneficial for the clinical diagnosis and treatment of DR. M6A modification is one of the most typical post-transcriptional regulatory markers for different RNAs in eukaryotic cells, including mRNA, tRNA, rRNA, miRNA, circRNA and lncRNA [11, 12]. Recently, m6A modification has been reported to play vital roles in the splicing and translation of RNAs, and can exert influence on their stability, translocation, and high-level structure [13, 14]. RNA m6A is modified by writers (methyltransferases, e.g., METTL3/14/16, RBM15/15B, ZC3H3, VIRMA, CBLL1, WTAP, and KIAA1429), removed by erasers (demethylases, e.g., FTO and ALKBH5), and recognized by readers (e.g., YTHDF1/2/3, YTHDC1/2 IGF2BP1/2/3 and HNRNPA2B1) [15]. Recently, m6A modification has been reported to play essential roles in physiological and pathological progression in the hematopoietic, central nervous and reproductive systems [15]. In addition, studies have demonstrated that m6A modification can affect the progression of various cancers by targeting the mRNAs of oncogenes or tumor suppressor genes [16]. Neovascularization is essential for growth, invasion, and metastasis during tumor development, and m6A modification has been reported to regulate these progresses via important angiogenic factors (e.g., VEGF), and their downstream key signaling pathways [17]. Fat mass and obesity-associated protein (FTO) belongs to the Fe2+ and α-ketoglutarate-dependent oxygenase family, and is a m6A eraser that mediates the oxidative demethylation of RNAs [18]. FTO has been reported to regulate antitumor immunity [19], to promote osteoclastogenesis and alveolar bone resorption in apical periodontitis [20], and to promote hormone-secreting pituitary neuroendocrine tumors [21]. Recently, in the field of DR, FTO has been proven to fuel vascular endothelial dysfunction via the demethylation of TNIP1 [22] to aggravate microvascular anomalies by targeting CDK2 [23], to modulate macrophage homeostasis in diabetic vasculopathy [24], or to regulate the apoptosis of retinal pigment epithelial cells (RPEs) [25]. In addition, in our previous work, RNA sequencing (RNA-Seq, GEO, GSE117238) indicated that the expression of FTO was significantly promoted in hRECs cultivated under hyperglycemic conditions [26]. However, as a m6A eraser, the downstream targets and networks of FTO have yet to be systematically investigated. Kinesin family member 11 (KIF11) is considered as a biomarker of tumors, and high expression levels of this protein are usually associated with the progression of various cancers, including endometrial cancer [27], pancreatic cancer [28], and hepatocellular carcinoma [29]. In addition, KIF11 has been investigated as an antitumor target for meningioma [30]. Recently, KIF11 has also been reported as a potential molecular marker of proliferative diabetic retinopathy (PDR) [31]. However, the potential roles of KIF11 in the progression of DR have yet to be investigated. In the present study, vitreous bodies of DR and non-DR (rhegmatogenous retinal detachment, macular hole, and epimacular membrane) patients were subjected to MeRIP-Seq, and the expression levels of FTO and KIF11/m6A modified KIF11 were tested in human vitreous bodies and in hRECs. We found that FTO was overexpressed in DR vitreous bodies and in hRECs cultivated in high glucose media, followed by the demethylation and overexpression of KIF11. In vitro, high FTO or KIF11 levels enhanced proliferation, DNA synthesis, migration, wound healing, and tube formation in hRECs under hyperglycemic conditions via affecting the downstream PI3K/AKT/mTOR and β-catenin/c-myc pathways. In vivo, FTO or KIF11 significantly enhanced the development of DR in a streptozotocin (STZ)-induced rat model, whereas knockdown shRNAs led to the opposite effects. This research demonstrated the importance of FTO-modulated KIF11 m6A demethylation in the progression of DR. Our findings offer promising avenues for the development of clinical therapies for retinal microvascular dysfunction in DR. Methods Collection of vitreous body samples This research was designed and performed in accordance with the principles of the Declaration of Helsinki (2013 version) and the Association for Research in Vision and Ophthalmology (ARVO) statement for research involving human subjects. Our research was approved by the Ethics Committee of the Affiliated Wuxi People’s Hospital of Nanjing Medical University (KY24030, 2024/03/26) and written informed consent was obtained from all patients. Vitreous body samples were collected from patients with nondiabetic conditions (n = 90, including 26 rhegmatogenous retinal detachment (RRD), 30 macular hole (MH), and 34 epimacular membrane (ERM)) and patients with proliferative DR (PDR, n = 90) during pars plana vitrectomy. The main exclusion criteria were: ocular comorbidity (age-related macular degeneration, retinal vein occlusion, glaucoma with macular visual field defect, uveitis or history of uveitis), any eye injury or eye surgery before the study. Patients in the PDR group received a single intravitreal anti-VEGF injection (0.1 mL of 40 mg/mL EYLEA-aflibercept) within one week before pars plana vitrectomy surgery. The vitreous body samples were stored in liquid nitrogen for further use. To enhance signal abundance for MeRIP-Seq, vitreous body samples from each group were combined (30 samples were mixed randomly; final n = 3). Vitreous levels of inflammatory mediators, including IL-1β (ab100562), IL-4 (ab215089), IL-6 (ab178013), IL-8 (ab214030), and TNF-α (ab181421) were determined following the manufacturer’s instruction (Abcam, Cambridge, MA, USA). The concentration of serum creatinine (SCR), the central marker to assess kidney function, were measured using an automatic biochemical instrument analyzer (BS-480, Shenzhen Mindray Bio-Medical Electronics Co., Ltd, Shenzhen, China). The clinical characteristics of the patients are presented in Table S1. There were no significant age or sex differences between the control and PDR groups. Compared with the nondiabetic group (NC), the levels of inflammatory mediators were significantly promoted in PDR vitreous; however, there was no remarkable differences between RRD, MH and ERM in NC group. In addition, the concentration of SCR was significantly increased in PDR group, while there were no dramatical fluctuation in RRD, MH and ERM sub-groups. Cell culture, antibodies and reagents Short interfering RNAs (siRNAs) (Table S2) targeting human FTO and KIF11 and scramble RNAs were synthesized by GenePharma (Shanghai, China). Overexpression plasmids (in pAAV-CMV-MCS) and shRNAs (in pAAV-U6-shRNA-CMV-MCS plasmids, Table S2) for rat FTO (NM_001039713.1) and KIF11 (NM_001169112.1) genes were synthesized and packaged in AAV9 virus by OBIO Scientific Service (Shanghai, China). Lipofectamine 3000 (Life Technologies, Carlsbad, CA, USA) was used for transfection. hRECs were obtained from Cell Systems (Kirkland, WA, USA) and kept in liquid nitrogen for long-term storage; the cells were subsequently cultured and passaged using phenol red-free Dulbecco’s modified Eagle’s medium (DMEM; Life Technologies, Carlsbad, CA, USA). hRECs (passages 5–6) were plated in six-well plates (Corning, Acton, MA, USA) at a density of 3,000 cells/mL and then cultured in low-glucose (5.5 mM glucose; LG), simulated DR high-glucose (25 mM glucose; HG), or low-glucose + mannitol (20 mM; LM, osmotic control) media for 120 h. Cultured hRECs were further subjected to the following treatments: HG with KIF11 overexpression (HG + OE KIF11), HG with the NC for overexpression (HG + OE NC), HG with KIF11 siRNA (HG + siKIF11), HG with scramble (HG + scramble), HG with KIF11 overexpression and KIF11 siRNA (HG + OE KIF11 + siKIF11), HG with FTO overexpression and KIF11 siRNA (HG + OE FTO + siKIF11), HG with FTO overexpression (HG + OE FTO), HG with the negative control (NC) for overexpression (HG + OE NC), HG with FTO siRNA (HG + siFTO), HG with scramble (HG + scramble), HG with FTO overexpression and FTO siRNA (HG + OE FTO + siFTO), HG with KIF11 overexpression and FTO siRNA (HG + OE KIF11 + siFTO). In addition, the inhibitor for FTO, FB23 (60 nM) (HY-137187, MCE, NJ, USA) [32] together with Lipofectamine 3000 reagents, was used to prohibit its demethylation activity against m6A modification. The antibodies used in this study are listed in Table S3. Other chemicals were of analytical reagent grade and were purchased locally. MeRIP-Seq Total RNA was extracted from vitreous body samples acquired from patients with PDR and non-diabetic patients using TRIzol reagent (15596026, Invitrogen, CA, USA). 20 µg total RNA was fragmented using RNA Fragmentation Reagents (AM8740, Invitrogen, CA, USA) in Thermomixer at 70℃ for 15 min. And then, the fragmented RNA (~ 100 nt) was incubated for 4 h at 4 °C with Protein A/G Magnetic Beads (HY-K0202, MCE, NJ, USA) and anti-m6A polyclonal antibody (ABE572-I, Sigma-Aldrich, MO, USA), and then the captured RNA or input was used for library construction using an Ovation ® SoLo RNA-Seq System Core Kit (NuGEN, CA, USA). MeRIP-Seq was carried out by Novogene. Library preparations were sequenced on an Illumina NovaSeq 6000 platform with a paired-end read length of 150 bp. Sequencing was carried out with three independent replicates, m6A peak identification was carried out using the exome Peak R package (version 2.16.0), and m6A-enriched motifs in each group were identified by HOMER (version 4.9.1). Differential expression analysis was performed using the DESeq2 R package (version 1.20.0). For m6A peak identification: P value < 0.05 (PEAK_CUTOFF_PVALUE = 0.05, PEAK_CUTOFF_FDR = NA). For differential m6A peak analysis: FDR < 0.05 (PEAK_CUTOFF_PVALUE = NA, PEAK_CUTOFF_FDR = 0.05). Fragment length was set to 200 nt. GO enrichment analysis of differentially expressed genes was conducted using the clusterProfiler package in R. The sequencing data was deposited in NCBI GEO (GSE263660). QRT-PCR, MeRIP-qRT-PCR and RIP-qRT-PCR Total RNA was extracted from hRECs, human vitreous body samples or rat retina samples using TRIzol reagent (Invitrogen). The expression levels of FTO and KIF11 were then quantified according to previously reported protocols using β-Actin as a control [33, 34]. MeRIP was performed using an Imprint RNA Immunoprecipitation Kit (RIP-12RXN, Sigma-Aldrich). In brief, the corresponding cell or tissue extracts were incubated with beads coated with 5 µg of control IgG antibody or an anti-m6A antibody with rotation at 4 °C for 12 h. Subsequently, RNA was extracted for the detection of KIF11 expression by qRT-PCR using β-Actin as a control. The potential interaction between FTO and m6A modified KIF11 mRNA was investigated using a RIP-qPCR kit (B605109, Sangon). Following the handbook of the manufacturer, hRECs cells (~ 2 × 107) were washed using 2 mL PBS, and lysed in 400 µl Complete RIP Lysis Buffer. The supernatant was incubated at 4 °C with Protein A/G Magnetic Beads and anti-m6A or anti-FTO polyclonal antibodies overnight, and then the captured RNA or proteins were subjected to qRT-PCR or Western blot assays, respectively. The primers are listed in Table S4. To investigate the stability of KIF11 mRNA, hRECs were treated with 5 µM actinomycin D (ActD; SBR00013, Sigma-Aldrich, MO, USA) for 0, 4, and 8 h. And then, total RNA was then extracted subjected to qRT-PCR assay. Cell viability assays In accordance with previous protocols [33, 34], a Cell Counting Kit-8 (CCK-8) (Sigma-Aldrich, St. Louis, MO, USA) was used to monitor the proliferation of hRECs and an EdU labeling kit (RiboBio, Guangzhou, China) was used to test DNA synthesis in hRECs. Transwell assays (Corning Transwell chambers; Corning, NY, USA) and a wound healing assay kit (ab242285, Abcam, Cambridge, MA, USA) were employed to reveal the migratory properties of hRECs. The tube formation (angiogenic) potential of hRECs was tested on a basement membrane matrix (BD Biosciences, San Jose, CA, USA). Western blot assays For western blotting [33, 34], total protein was extracted from cells or retinas using RIPA lysis buffer supplemented with a protease inhibitor mixture and a phosphatase inhibitor, separated by 10% SDS-PAGE, and transferred to PVDF membranes. The membranes were then blocked with 5% skimmed milk, and subsequently incubated with primary and then secondary antibodies. After exposure to an enhanced chemiluminescent (ECL) reagent, the target proteins were visualized with an Odyssey Imaging System (LI-COR Biosciences, Lincoln, NE, USA). The WB assays contained three parallel tests. Co-immunoprecipitation (co-IP) assay As previously reported [34], hRECs was lysed in RIPA buffer, mixed with 4-µL volume of antibodies for FTO, KIF11, PI3K or β-catenin, and incubated at 4 °C for 12 h. Then, 40 µL of sepharose beads coupled with Protein A/G was added to the mixture for a further 4 h of incubation at 4 °C. The mixture was washed twice with PBS and three times with washing buffer (1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, 1 mM Na3VO4, 5 mM NaF, 10 mM sodium pyrophosphate, 10 mg/mL aprotinin, 5 mg/mL leupeptin, 1 mM PMSF), the sepharose beads were boiled with 50 µL of 1× SDS loading buffer for 10 min, and then subjected to western blot assay. Establishment of a STZ-induced DR rat model Animal experiments followed the Institutional Guidelines for the Humane Treatment of Animals, the Principles of Laboratory Animal Care (National Institutes of Health, Bethesda, MD, USA) (https://www.ncbi.nlm.nih.gov/books/NBK54050), and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research (https://www.arvo.org/About/policies/statementfor-the-use-of-animals-in-ophthalmic-and-visionresearch/). All experiments were approved by the Ethics Committee of The Affiliated Wuxi People’s Hospital of Nanjing Medical University (Reference: DL2024-012, 2024/06/19). Male Sprague–Dawley (SD) rats (Changzhou Cavens Laboratory Animals Co., Ltd, Changzhou, China) (aged 6 to 8 weeks) were used in this research. Five days after the intraperitoneal injection of STZ (60 mg/kg), blood glucose levels > 16.7 mmol/L indicated that the rat model of diabetes had been established. Following the intraperitoneal injection of ketamine (80 µg/g) and xylazine (4 µg/g), the animals were intravitreally injected with 5 µL of AAV9 virus (1 × 1012 vg/mL) for the first time; at the end of week eight, the animals were intravitreally injected with 5 µL of AAV9 virus (1 × 1012 vg/mL) for the second time. The following groups were set up: Normal, STZ, STZ + NC (AAV9 with empty vector), STZ + shFTO and STZ + OE FTO, ST + shKIF11 and STZ + OE KIF11 (n = 8 rats per group). In addition, in each group, there were two extra animals (n = 2), which were euthanized after 8 weeks; and the retinas were extracted for qRT-PCR, MeRIP-qRT-PCR and western blot assays. As per our previously published protocols [33, 34], after 16 weeks, retina samples were extracted and subjected to Evans blue leakage and retinal trypsin digestion assays. Retinal paraffin Sect. (5 μm thick) were prepared and stained with H&E or subjected to immunohistochemical or immunofluorescence assays. The investigators (the first and the third authors) were blinded during image quantification. Statistical analysis All data were analyzed using SPSS version 13.0 software (SPSS Inc., Chicago, IL, USA) and are presented as means ± standard deviation. Statistical significance was assessed using the Student’s t-test and one-way ANOVA followed by Turkey post hoc test (Turkey’s HSD). P-values < 0.05 were considered significant. Results The elevation of FTO-regulated m6A in DR both in vitro and in vivo Previously, our RNA-Seq assays (GEO, GSE117238) demonstrated that the expression of FTO was significantly enhanced hRECs in a simulated DR high glucose microenvironment (HG, high glucose group) [26]; these previous results concurred with the qRT-PCR assay results obtained in the current study (Fig. 1A). In addition, enhanced expression levels of FTO were previously detected in the retinal fibrovascular membranes of patients with DR [22]. In the present study, enhanced FTO levels were detected in vitreous body samples obtained from patients with PDR (Fig. 1B). As FTO is one of the typical erasers of m6A modification [15, 16], we investigated the potential role of FTO-regulated m6A fluctuation in DR. Total RNAs were extracted from vitreous body samples from patients with PDR and non-diabetic controls during pars plana vitrectomy. Then, we performed RNA immunoprecipitation sequencing (MeRIP-Seq) (GSE263660). The Principle Component analysis (PCA) and the Fragments Per Kilobase of transcript per Million mapped reads (FPKM) of the background RNA-Seq (Input) were revealed in Fig. S1A and S1B, respectively. The read depth and mapping rate of MeRIP-Seq were showed in Table S5 and S6, respectively. MeRIP-Seq revealed the distribution of m6A modification on mRNAs (Fig. 1C). Compared with non-diabetic retinopathy samples (the negative control (NC) group, including rhegmatogenous retinal detachment, macular hole, and epimacular membrane), 127 transcripts showed fewer enriched m6A modifications, and 19 transcripts showed increased m6A peak enrichments (Fig. 1D). Five transcripts showed increased expression levels together with enriched m6A modifications, 86 transcripts showed increased expression levels together with decreased m6A modifications, 14 transcripts showed decreased expression levels together with enriched m6A modifications, while 41 transcripts showed decreased expression level together with decreased m6A modifications (Fig. 1E). KIF11, one of the top 10 significantly fluctuated mRNAs (Table S7), which had been reported to be associated with the progression of tumors, was identified as a potential target of FTO-modulated m6A modifications in DR, and the candidate m6A-modified region of KIF11 gene was located at chr10: 92645375–92645555, on one consensus coding sequence (CCDS) but not 3’-UTR (Fig. S1C). Furthermore, Gene Ontology (GO) analysis was employed to reveal the pathways in which the genes experiencing differential m6A modification were mostly enriched (Fig. 1F). Regarding Fig. S2, in RIP-qRT-PCR assay, IP with anti-FTO or anti-m6A antibodies could both enrich the levels of FTO protein and KIF11 mRNA, which suggested that m6A modified KIF11 mRNA might be one of the downstream targets of FTO. Fig. 1FTO-modulated m6A modification in DR. (A) The expression levels of FTO in hRECs (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (B) The expression levels of FTO in vitreous body samples (1, vs. non-diabetes group; n = 90 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (C) Distribution of m6A sites plotted by mRNA transcripts. (D) Volcano plot showing the m6A enrichment of genes in vitreous body samples from patients with PDR. (E) A plot showing the m6A enrichment and mRNA expression of differentially expressed genes in vitreous body samples from patients with PDR. KIF11 was identified for its remarkable demethylation and increased level of mRNA. (F) Gene ontology (GO) analysis based on RNA-Seq for differentially expressed genes in vitreous bodies in DR FTO-modulated m6A modification in DR. (A) The expression levels of FTO in hRECs (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (B) The expression levels of FTO in vitreous body samples (1, vs. non-diabetes group; n = 90 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (C) Distribution of m6A sites plotted by mRNA transcripts. (D) Volcano plot showing the m6A enrichment of genes in vitreous body samples from patients with PDR. (E) A plot showing the m6A enrichment and mRNA expression of differentially expressed genes in vitreous body samples from patients with PDR. KIF11 was identified for its remarkable demethylation and increased level of mRNA. (F) Gene ontology (GO) analysis based on RNA-Seq for differentially expressed genes in vitreous bodies in DR Elevated levels of FTO in a high glucose environment promoted the viability of hRECs by affecting m6A modification and the expression of KIF11. Compared with hRECs cultivated in low glucose (the LG group) and low glucose with an osmotic control (the LM group) environments, we detected a significantly increased expression level of KIF11 (Fig. 2A) together with dramatically decreased m6A modification (Fig. 2B) in hRECs cultured in a high glucose environment (HG group). In addition, the high glucose environment also promoted the proliferation (Fig. 2C), DNA synthesis (Fig. 2D and E), wound healing (Fig. 2F and G), tube formation (Fig. 2H and I) and migration (Fig. 2J) of hRECs. Furthermore, western blotting was used to investigate potential molecular mechanisms. The high glucose environment elevated the protein levels of FTO (Fig. 2K and L) and KIF11 (Fig. 2K and M) along with significantly enhanced phosphorylation levels of β-catenin (Fig. 2K and N) and increased protein levels of c-Myc (Fig. 2K and O). The high glucose environment also dramatically promoted the phosphorylation levels of PI3K, Akt and mTOR (Fig. 2K and P-R). These results indicated that the elevation of FTO in a simulated DR high glucose environment reduced m6A modification and enhanced the expression of KIF11, while the overexpression of KIF11 protein could enhance the angiogenesis of hRECs via affecting the β-catenin/c-Myc and PI3K/Akt/mTOR axes, which have been reported to play vital roles in tumor angiogenesis [35]. Fig. 2m6A modification was associated with the viability of hRECs in a high glucose environment. (A, B) The total and m6A modified expression levels of KIF11 (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (C) CCK8 assays after 5 days of culture (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (D, E) EdU assays (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 100 μm. (F, G) Wound healing assays (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (H, I) Tube formation assays (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (J) Transwell assays, scale bar: 500 μm. (K) Western blot assays. (L) The relative protein levels of FTO. (M) The relative protein levels of KIF11. (N) The relative phosphorylation levels of β-catenin. (O) The relative protein levels of c-Myc. The relative phosphorylation levels of (P) PI3K, (Q) Akt, and (R) mTOR. 1, vs. LG; 2, vs. LM; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. LG: low glucose; LM: osmotic control (low glucose + mannitol) HG: high glucose m6A modification was associated with the viability of hRECs in a high glucose environment. (A, B) The total and m6A modified expression levels of KIF11 (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (C) CCK8 assays after 5 days of culture (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (D, E) EdU assays (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 100 μm. (F, G) Wound healing assays (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (H, I) Tube formation assays (1, vs. LG; 2, vs. LM; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (J) Transwell assays, scale bar: 500 μm. (K) Western blot assays. (L) The relative protein levels of FTO. (M) The relative protein levels of KIF11. (N) The relative phosphorylation levels of β-catenin. (O) The relative protein levels of c-Myc. The relative phosphorylation levels of (P) PI3K, (Q) Akt, and (R) mTOR. 1, vs. LG; 2, vs. LM; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. LG: low glucose; LM: osmotic control (low glucose + mannitol) HG: high glucose KIF11 promoted the viability of hRECs via affecting the β-catenin/c-Myc and PI3K/Akt/mTOR pathways In the high glucose environment, the overexpression of KIF11 significantly increased the mRNA levels and m6A modification of KIF11. In contrast, siRNA significantly reduced the expression levels of KIF11, which was rescued by the overexpression of KIF11 or FTO. The m6A modification of KIF11 was reduced by siRNA transfection, and was further decreased by co-transfection of an FTO overexpression plasmid (Fig. 3A and B). In addition, the overexpression of KIF11 promoted the proliferation (Fig. 3C), DNA synthesis (Fig. 3D and E), wound healing (Fig. 3F and G), tube formation (Fig. 3H and I) and migration (Fig. 3J) of hRECs in a simulated DR high glucose environment; siRNA for KIF11 showed the opposite effects, and could be rescued by the overexpression of KIF11 or FTO. Western blotting assays (Fig. 3K) revealed that the overexpression of KIF11 significantly promoted the protein levels of KIF11 (Fig. 3L) and c-Myc (Fig. 3N), along with the phosphorylation levels of β-catenin (Fig. 3M), PI3K (Fig. 3O), Akt (Fig. 3P) and mTOR (Fig. 3Q). In contrast, siRNA for KIF11 had opposite effects on these proteins; these effects could be reversed by the overexpression of KIF11 or FTO. Collectively, these results demonstrated that KIF11 could promote the angiogenic activation of hRECs via affecting the β-catenin/c-Myc and PI3K/Akt/mTOR pathways. Fig. 3KIF11 promoted the viability of hRECs in a simulated DR high glucose environment. (A, B) The total and m6A modified expression levels of KIF11 (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (C) CCK8 assays after 5 days of culture (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (D, E) EdU assay (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 100 μm. (F, G) Wound healing assays (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (H, I) Tube formation assays (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (J) Transwell assays, scale bar: 500 μm. (K) Western blot assays. (L) The relative protein levels of KIF11. (M) The relative phosphorylation levels of β-catenin. (N) The relative protein levels of c-Myc. The relative phosphorylation level sof (O) PI3K, (P) Akt, and (Q) mTOR. 1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. HG: high glucose; HG + OE KIF11: HG+KIF11 overexpression; HG + OE NC: HG+negative control; HG + si KIF11: HG+KIF11 siRNA; HG+Scramble: HG+Scramble RNA; HG + OE KIF11 + si KIF11: HG+ KIF11 overexpression+ KIF11 siRNA; HG + OE FTO + si KIF11: HG + FTO overexpression+ KIF11 siRNA KIF11 promoted the viability of hRECs in a simulated DR high glucose environment. (A, B) The total and m6A modified expression levels of KIF11 (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (C) CCK8 assays after 5 days of culture (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (D, E) EdU assay (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 100 μm. (F, G) Wound healing assays (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (H, I) Tube formation assays (1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (J) Transwell assays, scale bar: 500 μm. (K) Western blot assays. (L) The relative protein levels of KIF11. (M) The relative phosphorylation levels of β-catenin. (N) The relative protein levels of c-Myc. The relative phosphorylation level sof (O) PI3K, (P) Akt, and (Q) mTOR. 1, vs. HG + OE NC; 2, vs. HG + si KIF11; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. HG: high glucose; HG + OE KIF11: HG+KIF11 overexpression; HG + OE NC: HG+negative control; HG + si KIF11: HG+KIF11 siRNA; HG+Scramble: HG+Scramble RNA; HG + OE KIF11 + si KIF11: HG+ KIF11 overexpression+ KIF11 siRNA; HG + OE FTO + si KIF11: HG + FTO overexpression+ KIF11 siRNA FTO influenced the viability of hRECs via KIF11-affected β-catenin/c-Myc and PI3K/Akt/mTOR pathways In a high glucose environment, the overexpression of FTO significantly enhanced the mRNA levels of FTO and KIF11, while siRNA for FTO exerted suppressive effects (Fig. 4A-C). The overexpression of FTO rescued the effects of FTO siRNA on the expression of these two genes, while the overexpression of KIF11 could only enhance the level of KIF11. In addition, the m6A modification of KIF11 was dramatically decreased after FTO was overexpressed but was significantly promoted by siRNA for FTO; these effects were rescued by the expression of FTO but not KIF11. The proliferation (Fig. 4D), DNA synthesis (Fig. 4E and F), wound healing (Fig. 4G and H), tube formation (Fig. 4I and J) and migration (Fig. 4K) of hRECs were all significantly promoted following the overexpression of FTO in a simulated DR high glucose environment. In contrast, these characteristics of hRECs were prohibited by the application of siRNA for FTO; these effects were rescued by the overexpression of KIF11 or FTO. Western blotting (Fig. 4L) revealed that the overexpression of FTO significantly promoted the protein levels of FTO (Fig. 4M), KIF11 (Fig. 4N) and c-Myc (Fig. 4P), along with the phosphorylation levels of β-catenin (Fig. 4O), PI3K (Fig. 4Q), Akt (Fig. 4R) and mTOR (Fig. 4S). However, siRNA for FTO negatively regulated these proteins; these effects could be reversed by the overexpression of FTO. The overexpression of KIF11 rescued the levels of these proteins, except for FTO. Collectively, these results demonstrated that FTO could enhance the angiogenic activation of hRECs via KIF11-affected β-catenin/c-Myc and PI3K/Akt/mTOR pathways, and by reducing m6A modification and increasing the expression of KIF11. However, according to the co-IP assay, there should not be direct interaction between the proteins of FTO, KIF11, PI3K and β-catenin (Fig. S4). Fig. 4FTO promoted the viability of hRECs via KIF11 in a simulated DR high glucose environment. (A) The expression levels of FTO (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (B, C) The total and m6A modified expression levels of KIF11 (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (D) CCK8 assays after 5 days of culture (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (E, F) EdU assays (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 100 μm. (G, H) Wound healing assays (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (I, J) Tube formation assays (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (K) Transwell assays, scale bar: 500 μm. (L) Western blot assays. (M, N) The relative protein levels of FTO and KIF11. (O) The relative phosphorylation levels of β-catenin. (P) The relative protein levels of c-Myc. The relative phosphorylation levels of (Q) PI3K, (R) Akt, and (S) mTOR. 1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. HG: high glucose; HG + OE FTO: HG + FTO overexpression; HG + OE NC: HG+negative control; HG + si FTO: HG + FTO siRNA; HG+Scramble: HG+Scramble RNA; HG + OE FTO + si FTO: HG + FTO overexpression + FTO siRNA; HG + OE KIF11 + si FTO: HG+KIF11 overexpression + FTO siRNA FTO promoted the viability of hRECs via KIF11 in a simulated DR high glucose environment. (A) The expression levels of FTO (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (B, C) The total and m6A modified expression levels of KIF11 (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (D) CCK8 assays after 5 days of culture (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001). (E, F) EdU assays (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 100 μm. (G, H) Wound healing assays (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (I, J) Tube formation assays (1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 5 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001); scale bar: 200 μm. (K) Transwell assays, scale bar: 500 μm. (L) Western blot assays. (M, N) The relative protein levels of FTO and KIF11. (O) The relative phosphorylation levels of β-catenin. (P) The relative protein levels of c-Myc. The relative phosphorylation levels of (Q) PI3K, (R) Akt, and (S) mTOR. 1, vs. HG + OE NC; 2, vs. HG + si FTO; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. HG: high glucose; HG + OE FTO: HG + FTO overexpression; HG + OE NC: HG+negative control; HG + si FTO: HG + FTO siRNA; HG+Scramble: HG+Scramble RNA; HG + OE FTO + si FTO: HG + FTO overexpression + FTO siRNA; HG + OE KIF11 + si FTO: HG+KIF11 overexpression + FTO siRNA FB23, the inhibitor for the demethylation activity of FTO, did not affect the expression of FTO (Fig. S3A) but significantly reduced the total KIF11 level (Fig. S3B) and enhanced the m6A modification of KIF11 (Fig. S3C). Additionally, FB23 reversed the effects of overexpressed FTO on the proliferation (Fig. S3D), DNA synthesis (Fig. S3E and F), wound healing (Fig. S3G and H), tube formation (Fig. S3I and J) and migration (Fig. S3K) of hRECs in high glucose environment. Furthermore, after an 8 h ActD treatment, FB23 reduced the stability of total KIF11 mRNA, and it could reverse the promotion effects of overexpressed FTO (Fig. S3L). The results of western blot assay (Fig. S3M) were in good accordance with those of qRT-PCR analysis, FB23 could not reduce the protein level of FTO (Fig. S3N), but could significantly prohibit that of KIF11 (Fig. S3O). FTO and KIF11 promoted the progression of DR in a rat model STZ-induced DR rats were euthanized after 16 weeks. Evans blue assays (Fig. 5A) showed that when compared with the normal group (10.6 ± 1.1%), the leakage area was significantly increased in both the STZ (37.5 ± 5.5%) and STZ + NC (36.0 ± 3.3%) groups; this phenomenon was partially reversed after the inhibition of FTO (STZ + sh FTO, 12.6 ± 1.5%) or KIF11 (STZ + sh KIF11, 12.7 ± 2.1%). However, compared with the STZ + NC group, the retinal leakage area was further increased following the overexpression of FTO (STZ + OE FTO, 63.9 ± 4.3%) or KIF11 (STZ + OE KIF11, 64.9 ± 3.5%) (Fig. 5F). In addition, western blotting assays (Fig. 5G) showed that serum albumin leakage followed a similar trend. Fig. 5FTO mitigated the progression of DR via KIF11 in a rat model. (A) Evans blue (EB) staining of retinas, scale bar: 100 μm. (B) Following digestion with trypsin, the retinal vasculature was subjected to periodic acid-Schiff staining (scale bar: 25 μm; red arrows: acellular capillaries; green arrows: retained pericytes). The retinal Sect. (5 μm thick) were subjected to (C) hematoxylin and eosin (H&E) staining (scale bar: 200 μm. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium) or immunohistochemistry targeting (D) fibronectin and (E) PCNA (scale bar: 200 μm). (F) Retinal vascular leakage (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). (G) Retinal vascular leakage quantified by serum albumin western blotting (n = 3 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). (H) Acellular capillary number and (I) human retinal microvascular epithelial cell (hREC)/pericyte ratio (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). The relative positivity ratio for (J) fibronectin, and (K) PCNA (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). STZ: streptozotocin; NC: negative control; OE: overexpression; sh: shRNA FTO mitigated the progression of DR via KIF11 in a rat model. (A) Evans blue (EB) staining of retinas, scale bar: 100 μm. (B) Following digestion with trypsin, the retinal vasculature was subjected to periodic acid-Schiff staining (scale bar: 25 μm; red arrows: acellular capillaries; green arrows: retained pericytes). The retinal Sect. (5 μm thick) were subjected to (C) hematoxylin and eosin (H&E) staining (scale bar: 200 μm. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium) or immunohistochemistry targeting (D) fibronectin and (E) PCNA (scale bar: 200 μm). (F) Retinal vascular leakage (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). (G) Retinal vascular leakage quantified by serum albumin western blotting (n = 3 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). (H) Acellular capillary number and (I) human retinal microvascular epithelial cell (hREC)/pericyte ratio (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). The relative positivity ratio for (J) fibronectin, and (K) PCNA (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). STZ: streptozotocin; NC: negative control; OE: overexpression; sh: shRNA Next, we performed periodic acid-Schiff staining (Fig. 5B); compared with the normal group (2.4 ± 1.1), the number of acellular capillaries was significantly increased in both the STZ (11.8 ± 2.6) and STZ + NC (13.6 ± 2.5) groups; this trend remained following the inhibition of FTO (STZ + sh FTO, 3.4 ± 0.9) or KIF11 (STZ + sh KIF11, 3.2 ± 1.8). However, compared with the STZ + NC group, the number of acellular capillaries was further increased following the overexpression of FTO (STZ + OE FTO, 22.4 ± 3.7) or KIF11 (STZ + OE KIF11, 20.8 ± 1.8) (Fig. 5H). In addition, the hREC/pericyte ratio was dramatically increased in the STZ (11.3 ± 1.0) and STZ + NC (13.1 ± 0.9) groups when compared with that in the control group (6.0 ± 0.8); the inhibition of FTO (STZ + sh FTO, 6.0 ± 0.8) or KIF11 (STZ + sh KIF11, 7.2 ± 0.3) partially offset this increase, while the overexpression of FTO (STZ + sh FTO, 18.6 ± 1.0) or KIF11 (STZ + sh KIF11, 21.7 ± 2.6) further enhanced the hREC/pericyte ratio (Fig. 5I). Eyeballs from the rat model were paraffin-embedded and cut into 5-µm-thick sections before staining with hematoxylin and eosin (H&E) (Fig. 5C) or used for immunohistochemical staining assays. After scanning the sections and data analysis in ImageJ, we found that positivity for fibronectin (a biomarker of fibrosis) (Fig. 5D and J), PCNA (a biomarker of proliferation) (Fig. 5E and K) and inflammatory mediators (Fig. S6, IL-1β, IL-6 and TNF-α) in the STZ and STZ + NC groups was significantly increased when compared with the normal group; however, positivity for these proteins were lower in the STZ + sh FTO and STZ + sh KIF11 groups than in the STZ and STZ + NC groups. In addition, the overexpression of FTO or KIF11 significantly enhanced fibrosis, cell proliferation and the inflammatory responses in the retina. Furthermore, there were no significant differences between the STZ and STZ + NC groups, which suggested that the repeated intravitreal injection of AAV9 would not bring with additional inflammatory response in 16 weeks. Immunofluorescence (IF) assays (Fig. 6) showed that when compared with the normal group, the protein levels of FTO and KIF11 were both enhanced in the STZ and STZ + NC groups; this effect was partially reversed by the inhibition of FTO (STZ + sh FTO) or KIF11 (STZ + sh KIF11). The overexpression of FTO further enhanced the protein levels of FTO and KIF11; however, the overexpression of KIF11 could only enhance the levels of KIF11 (Fig. 6A, C and D). Fig. 6Immunofluorescence assays of the rat retina. Immunostaining for (A) FTO/KIF11 and (B) GFAP/GS on retinal paraffin sections (scale bar: 100 μm). GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. (C-F) The relative positivity ratio for (C) FTO, (D) KIF11, (E) GFAP, and (F) GS in the retina (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). STZ: streptozotocin; NC: negative control; OE: overexpression; sh: shRNA Immunofluorescence assays of the rat retina. Immunostaining for (A) FTO/KIF11 and (B) GFAP/GS on retinal paraffin sections (scale bar: 100 μm). GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. (C-F) The relative positivity ratio for (C) FTO, (D) KIF11, (E) GFAP, and (F) GS in the retina (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). STZ: streptozotocin; NC: negative control; OE: overexpression; sh: shRNA Compared with the normal group, the levels of glial fibrillary acidic protein (GFAP, a biomarker of retinal gliosis) were markedly increased (Fig. 6B and E) while those of glutamate synthase (GS) were decreased (Fig. 6B and F) in the STZ and STZ + NC groups, thus suggesting that retinal gliosis was significantly enhanced in the DR rat model; this trend was reversed by the inhibition of FTO or KIF11. In contrast, the overexpression of FTO or KIF11 enhanced retinal gliosis. FTO promoted the progression of DR via KIF11-affected β-catenin/c-Myc and PI3K/Akt/mTOR axes QRT-PCR analysis (Fig. 7A-C) demonstrated that the expression levels of FTO and total KIF11 were significantly increased in STZ-treated rats, while the expression levels of m6A modified KIF11 were decreased. Compared with the STZ + NC group, the inhibition of FTO (STZ + sh FTO) significantly reduced the expression of FTO and total KIF11, but enhanced the m6A modification of KIF11 mRNA. In contrast, the overexpression of FTO (STZ + OE FTO) significantly promoted the expression of FTO and total KIF11, but reduced the m6A modification of KIF11 mRNA. In addition, the inhibition of KIF11 reduced both the total and m6A modification levels of KIF11, while the overexpression of FTO increased these two levels; however, the regulating the levels of KIF11 did not affect the expression of FTO. Fig. 7FTO affected the β-catenin/c-Myc and PI3K/Akt/mTOR pathways via KIF11 in a rat model. Relative expression levels of (A) FTO, (B) total KIF11, and (C) m6A modified KIF11 (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). (D) Western blot assays of proteins extracted from retinal tissue. The relative protein levels of (E) FTO and (F) KIF11. (G) The relative phosphorylation levels of β-catenin. (H) The relative protein levels of c-Myc. The relative phosphorylation levels of (I) PI3K, (J) Akt, and (K) mTOR. 1, vs. Normal; 2, vs. STZ + NC; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. STZ: streptozotocin; NC: negative control; OE: overexpression; sh: shRNA FTO affected the β-catenin/c-Myc and PI3K/Akt/mTOR pathways via KIF11 in a rat model. Relative expression levels of (A) FTO, (B) total KIF11, and (C) m6A modified KIF11 (n = 5 biological replicates; 1, vs. normal; 2, vs. STZ + NC; * P < 0.05, ** P < 0.01, *** P < 0.001). (D) Western blot assays of proteins extracted from retinal tissue. The relative protein levels of (E) FTO and (F) KIF11. (G) The relative phosphorylation levels of β-catenin. (H) The relative protein levels of c-Myc. The relative phosphorylation levels of (I) PI3K, (J) Akt, and (K) mTOR. 1, vs. Normal; 2, vs. STZ + NC; n = 3 biological replicates, * P < 0.05, ** P < 0.01, *** P < 0.001. STZ: streptozotocin; NC: negative control; OE: overexpression; sh: shRNA Western blotting assays (Fig. 7D) demonstrated that following STZ treatment, the protein levels of FTO (Fig. 7E), KIF11 (Fig. 7F), p-β-catenin (Fig. 7G), c-Myc (Fig. 7H), p-PI3K (Fig. 7I), p-AKT (Fig. 7J), and p-mTOR (Fig. 7K) were increased in the retinal tissue of DR rats. Compared with the STZ + NC group, the inhibition of FTO (STZ + sh FTO) partially reversed these alterations, whereas the overexpression of FTO (STZ + OE FTO) further enhanced these trends. In addition, the inhibition of KIF11 (STZ + sh KIF11) yielded similar affects as the inhibition of FTO (STZ + sh FTO), while the overexpression of KIF11 (STZ + OE KIF11) had the same effects as the overexpression of FTO (STZ + OE FTO); however, regulating KIF11 did not affect the protein levels of FTO. Discussion The prevalence of diabetes is increasing worldwide [1–3]. As one of the most significant microvascular complications of diabetes, DR is being increasingly recognized as a critical global health challenge [4–6]. Several researchers have investigated the molecular mechanisms underlying the clinical pathologies responsible for the progression of DR, and various signaling pathways have been identified [7–10]. Currently, the intraocular injection of anti-VEGF antibodies is extensively used for the clinical therapy of DR, including pegaptanib, ranibizumab, and bevacizumab. However, owing to potential resistance and adverse effects on healthy vasculature, the long-term application of these antibodies is associated with many challenges [36–38]. Therefore, there is a critical need to identify other molecular mechanisms responsible for the development of DR in an attempt to identify novel targets for clinical therapy. As an important post-transcriptional regulatory event for various RNAs in eukaryotic cells [11, 12], m6A modification has been reported to affect the half-life, translocation, and structure of certain RNAs. Writers, erasers and readers have been reported to play vital roles in the m6A modifications of RNAs [13–15]. Recently, RNA m6A modification was identified as one of the most predominant forms of epigenetic regulation [39], and has been associated with various physiological and pathological events in cancer immunity [40], the development of cancers [41], Huntington’s disease [42], Parkinson’s Disease [43], acute myeloid leukemia [44], cardiac fibrosis [45], and diabetic microvascular complications [46]. As a typical m6A eraser [18], FTO has been reported to be associated with antitumor immunity [19]. Moreover, several FTO-mediated pathways have been investigated as potential regulators in the development of DR [22–25]. In our previous study, we detected significant enhancements in the expression of FTO in hyperglycemia-induced hRECs using RNA sequencing (RNA-Seq, GEO, GSE117238) [26]. The progression of DR is usually in a sustained hyperglycemic and hypoxic microenvironment in vivo [7–10]; in this work, hRECs were cultivated in a simulated DR high glucose environment, in order to minimize the confounding effects introduced by hypoxia; as the hypoxia-induced high expression of HIF 1-α has been considered to be associated a number of signal pathways, including Akt/mTOR, β-catenin/c-Myc, VEGF and so on [47–49]. In the present study, we systematically investigated the potential roles of FTO and its potential downstream targets or axes and collected samples of vitreous bodies from patients with PDR patients during pars plana vitrectomy. QRT-PCR indicated that FTO was overexpressed in vitreous body samples acquired from patients with PDR and that this trend concurred with findings in hyperglycemia-induced hRECs; these results were also consistent with previous reports involving both human or animal tissues [22]. MeRIP-Seq identified 127 transcripts exhibiting fewer enriched m6A modifications, and 19 transcripts exhibiting increased m6A peak enrichments (Fig. 1D); these findings may have been caused by the enhanced expression of the FTO eraser. As one of the top 10 differently m6A modified and expressed genes, KIF11 showed significantly reduced m6A peaks and significantly enhanced mRNA levels; consequently, we selected KIF11 as the downstream target of FTO. Regarding Fig. S1C, the candidate m6A-modified region was located at the CCDS of KIF11 gene (chr10: 92645375–92645555) in a 180 nt fragment. RIP-qRT-PCR (Fig. S2) indicated that FTO might be associated with m6A modified KIF11 mRNA; however, as the m6A modification of KIF11 has not been systematically investigated, to confirm the m6A-modified site requires multiple attempts. In our further work, using motif prediction tools, mutagenesis reporters will be designed on the 180 nt fragments. In MeRIP-Seq, non-diabetic vitreous samples (including RRD, MH and ERM) were used as negative control, the fasting blood glucose (FBG), inflammatory mediator and SCR levels in NC group were significantly lower than those in PDR group (Table S1). In each group, 30 vitreous body samples were randomly combined to enhance the signal abundance, although the individual differences in each group might affect the sequencing process, PCA assay (Figure S1A) and FPKM distribution of the background RNA-Seq (Figure S1B), gene tracks based on MeRIP-Seq of KIF11 (Figure S1C), read depth of MeRIP-Seq (Table S5), and mapping rate of MeRIP-Seq (Table S6) indicated that the sample mixing did not bring significant fluctuation in MeRIP-Seq. As previously reported, KIF11 represents a useful biomarker for tumors, and high expression levels of this protein are usually associated with the progression of various cancers [27–29] via promotion of tumor angiogenesis [35]. In addition, KIF11 has also been identified as a potential molecular marker for PDR [31] or familial exudative vitreoretinopathy [50]. In good accordance with the results of our MeRIP-Seq assays, the overexpression of KIF11 was detected in hyperglycemia-induced hRECs; this led to enhanced proliferation, DNA synthesis, wound healing, tube formation, and migration viabilities. In addition, the overexpression of KIF11 further enhanced these characteristics. In contrast, the administration of KIF11 siRNA led to directly opposing effects, which were partially rescued by the overexpression of KIF11 or FTO (Fig. 4). Furthermore, the overexpression of FTO significantly enhanced the levels of KIF11, further promoting proliferation, DNA synthesis, wound healing, tube formation, and migration viabilities, while the administration of FTO siRNA prohibited the expression of KIF11 and these key characteristics. Western blotting indicated that in a simulated DR high glucose environment, increased levels of FTO protein could promote the protein levels of KIF11 and the levels of proteins in the downstream PI3K/AKT/mTOR and β-catenin/c-myc axes, which have been investigated extensively in cancer or microvascular diseases [35, 51, 52]. However, in co-IP assay (Fig. S4), there were no direct interaction between KIF11, PI3K and β-catenin; how the expression of KIF11 could regulate these two pathways remained unclear. In the further work, inhibitors of KIF11 (Ispinesib), PI3K (LY294002), mTOR (Ramycin), or β-catenin (Nigericin or MSAB) would be employed together with bioinformatic tools, to investigate the unclear part in the integrated mechanisms. To prohibit the demethylation activity of FTO, the effects of its inhibitor FB23 were investigated in vitro. As reported, direct IC50 of FB23 for the demethylase activity of FTO was 60 nM, and extracellular IC50 for the proliferation of NB4 and MONOMAC6 were 44.8 µM (~ 20 nM, intracellular) and 23.6 µM (~ 15 nM, intracellular), respectively; for its derivative (FB23-2), the extracellular IC50 for the proliferation of NB4 and MONOMAC6 cells were 0.8–1.5 µM (50–200 nM intracellular); the limited inhibitory effect of these compounds on proliferation of these cells was likely due to the low cellular uptake [32]. In addition, FB23-2 (2 µM, extracellular) was reported to repress the FTO activity in human umbilical vein endothelial cells (HUVECs) and restrain the cell proliferation [53]. FB23 was insoluble in aqueous, and when the 10 mM stock solution was diluted to over 10 µM in serum-free DMEM medium, significant precipitation and suspended particles always appeared. In this work, as revealed in Fig. S3, hRECs in other groups were all transfected with Lipofectamine 3000, therefore, FB23 (60 nM) was added into the hRECs in high glucose medium together with Lipofectamine 3000 reagents, using the siRNA transfection protocol as reference. Regarding Fig. S3, FB23 could significantly reverse the effects of overexpressed FTO on the viability of hRECs via enhancing the m6A modification, reducing both of the expression level of KIF11 mRNA and protein level of KIF11. However, direct addition of FB23 (60 nM) did not work. It suggested that the Lipofectamine 3000 reagents might enhance the permeability of hRECs and help to promote the cellular uptake of FB23. The IC50 of FB23 for the proliferation of hRECs was not investigated in this work, because with the Lipofectamine 3000 reagents, higher concentration of FB23 could accelerate the precipitation. In the further work, the IC50 of FB23 (both extracellular and intracellular) for the proliferation of hRECs would be systematically investigated; the inactive FTO mutants would be brought into hRECs cells via CRISPR/Cas9 editing, and the overexpression of the native FTO would be employed in the rescue group. To investigate the specific roles of FTO and KIF11 in vivo, we established an STZ-induced rat model of DR and inserted these two genes into a pAAV-CMV-MCS plasmid. We also inserted shRNA sequences for FTO and KIF11 into a pAAV-U6-shRNA-CMV-MCS plasmid. These plasmids were packed into an AAV9 virus, and then administered to experimental animals by intravitreal injection every 8 weeks. After week 8, two extra animals in each group were euthanized, and the retina samples were subjected to qRT-PCR, MeRIP-qRT-PCR and western blot assays. The mRNA, m6A modification and protein levels of FTO and KIF11 at week 8 (Fig. S5) were in good accordance with the results at week 16 (Figs. 6 and 7); it indicated that the intravitreal injection of AAV9 virus successfully modulated the expression of FTO and KIF11 in the animal model of this work. In immunofluorescence assay (Fig. 6A and B), at week 16, although the FTO/KIF11 and GS/GFAP signals were all spread from GCL to RPE layers, in good accordance with the distribution of retinal vascular and Müller cells; these phenomena could not indicate that the AAV9 only transduced hRECs and Müller cells. In the further work, FTO and KIF11 would be colocalized with the specific bio-marker of retinal pigment epithelium (RPE) cells (CRALBP), Müller cells (CRALBP and SOX9) or other retinal cells. Compared with the normal group, the typical pathological features of DR, including retinal leakage, acellular capillary formation, pericyte loss, fibrosis, and gliosis [7–10], were detected in the STZ and STZ + NC group. The overexpression of FTO and KIF11 further promoted the progression of DR, whereas the knockdown of FTO and KIF11 partially mitigated the development of DR. In addition, qRT-PCR and western blotting assays demonstrated that in the rat model, FTO could promote the progression of DR via KIF11 and the downstream PI3K/AKT/mTOR and β-catenin/c-myc pathways; these findings concurred with our in vitro results. The fluctuation of FTO in vitro and in vivo was shown to influence the mRNA and protein levels of KIF11; however, there was no feedback effects from KIF11, thus indicating that FTO/KIF11 represents a unidirectional form of regulation. FTO is an established m6A eraser [18–21], and plays a role in various physiological or pathological progresses by removing the m6A modification of different mRNAs or non-coding RNAs [43–52, 54–57]. In the present research, we used vitreous body samples from patients with PDR and identified 127 transcripts exhibiting fewer enriched m6A modifications in MeRIP-Seq assays, thus indicating that KIF11 is not the only downstream target and that FTO might serve as a signal hub for the crosstalk between different pathways. Recently, FTO has been considered as a potential inhibitory target for the treatment of cancer, microvascular or neurological diseases [58–62]; however, FTO has also been reported to promote diabetic wound healing [63] and vascular remodeling [64] for clinical application. To inhibit the expression of FTO in eye might affect the m6A modification of a great number of mRNAs or non-coding RNAs, which would further interfere various pathways. Therefore, therapy strategies targeting FTO for DR should be considered and assessed very carefully. KIF11 has been previously reported as a biomarker for various tumors [27–30] but has not been systematically investigated in the context of DR progression. In this study, we performed MeRIP-Seq assays and selected KIF11 as one of the most crucial downstream targets of FTO. Our experiments proved that KIF11 promoted the development of DR in vitro and in vivo, thus indicating that KIF11 represents a potential target for the clinical therapy of DR. As previously reported, KIF11 plays vital roles in mitosis [65], whether the inhibition of KIF11 would brought side effects in eye is still unknown. In the further work, the intravitreal administration of KIF11 shRNA/AAV in animal model would be systematically studied, to investigate the potential toxicity and immunological responses; in addition, the commercial KIF11 inhibitor (Ispinesib) is also of great potential for the clinical therapy of DR, and different administration strategies, e.g., intravitreal microinjection and trans-corneal/blood eye barrier delivery would be tested. In summary, we found that elevated FTO-regulated m6A in DR in vitro and in vivo, FTO was associated with reduced m6A modification of KIF11, which could further affect the downstream PI3K/AKT/mTOR and β-catenin/c-myc axes. Both FTO and KIF11 should be of potential as candidate targets for the clinical treatment of DR retinal microvascular dysfunction. Conclusion We identified a unidirectional FTO/KIF11 regulatory mechanism in the progression of DR in vitro and in vivo. FTO might promote the development of DR by regulating KIF11 expression through m6A-dependent mechanisms, and by affecting the downstream PI3K/AKT/mTOR and β-catenin/c-myc signaling pathways (Fig. 8). These observations suggest that FTO/KIF11 axis represents a candidate molecular mechanism that requires further pharmacologic, safety, and translational validation for the clinical treatment of DR retinal microvascular dysfunction. Fig. 8The potential mechanism of FTO/KIF11 regulated pathways in DR The potential mechanism of FTO/KIF11 regulated pathways in DR Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1: The raw membranes of western blot and co-IP assays Supplementary Material 1: The raw membranes of western blot and co-IP assays Supplementary Material 2: Table S1-S7, and Figure S1-S6 Supplementary Material 2: Table S1-S7, and Figure S1-S6