m6A modifications of circular RNAs in ischemia-induced retinal neovascularization

Ischemia-induced pathological neovascularization in the retina is a leading cause of blindness in various age groups. The purpose of the current study was to identify the involvement of circular RNAs (circRNAs) methylated by N6-methyladenosine (m6A), and predict their potential roles in oxygen-induced retinopathy (OIR) in mice. Methylation assessment via microarray analysis indicated that 88 circRNAs were differentially modified by m6A methylation, including 56 hyper-methylated circRNAs and 32 hypo-methylated circRNAs. Gene ontology enrichment analysis predicted that the enriched host genes of the hyper-methylated circRNAs were involved in cellular process, cellular anatomical entity, and protein binding. Host genes of the hypo-methylated circRNAs were enriched in the regulation of cellular biosynthetic process, the nucleus, and binding. According to the Kyoto Encyclopedia of Genes and Genomes analysis, those host genes were involved in the pathways of selenocompound metabolism, salivary secretion, and lysine degradation. MeRIP-qPCR verified significant alterations in m6A methylation levels of mmu_circRNA_33363, mmu_circRNA_002816, and mmu_circRNA_009692. In conclusion, the study revealed the m6A modification alterations in OIR retinas, and the findings above shed light on the potential roles of m6A methylation in circRNA regulatory functions in the pathogenesis of ischemia-induced pathological retinal neovascularization.


Introduction
As the shared pathogenesis of proliferative diabetic retinopathy (PDR), retinopathy of prematurity (ROP), and other retinal vasculopathies, ischemiainduced pathological retinal neovascularization is a common cause of irreversible blindness globally in various age groups [1][2][3]. Therefore, effective intervention has become crucial to reducing the severe vision loss caused by retinal neovascularization.
As the most prevalent mRNA modification in mammals [4], N6-methyladenosine (m 6 A) RNA methylation exerts potential epigenetic functions in various physiological and pathological processes [5][6][7]. We recently revealed the profiles of altered m 6 A epitranscriptomes of mRNAs and long non-coding RNAs in the retinas of oxygen-induced retinopathy (OIR) [8], a classical mouse model for retinal neovascular diseases [9]. Several studies suggested the regulatory roles of m 6 A modifications in retinal neovascular diseases. For example, it has been reported that YTHDF2 induces mRNA instability of ITGB1 in an m 6 A-dependent manner, and attenuated the development of diabetic retinopathy [10]. Yao et al. [11] indicated that the m 6 A writer enzyme METTL3 is a promising strategic target for the treatment of pathological neovascularization. Therefore, the potential role of m 6 A methylation in the pathogenesis of retinal neovascularization is worth exploring.
Circular RNA (circRNA) is a closed RNA that Ivyspring International Publisher lacks 5' and 3' ends [12]. Recent studies have revealed important physiological and pathological functions of circRNAs, and their potential as novel molecular biomarkers and therapeutic targets [13]. In the vitreous humour of PDR, a total of 131 circRNAs were dysregulated compared to the controls [14]. Besides, Li et al. [15] revealed altered exosomal circRNAs in the serum of PDR. In OIR mouse model, 539 circRNAs were significantly changed in the retinal neovascularization group, indicating the possible involvement of circRNA-associated ceRNA networks [16]. Significant alteration of circRNAs in peripheral blood mononuclear cells in premature infants suggests they may also be biomarkers and/or therapeutic targets for ROP [17]. Numerous studies have revealed profile alterations in m 6 A-modified circRNAs in many diseases, including cerebral infarction [18], severe acute pancreatitis [19], acute myeloid leukemia [20], and oral squamous cell carcinoma [21]. Interestingly, Huang et al. [22] recently demonstrated that circRNA circFAT1 enhanced autophagy and reduced pyroptosis in glucose-stressed retinal pigment epithelial cells, and reported the binding of circFAT1 and m 6 A reader YTHDF2.
To date, m 6 A modification of circRNAs in retinal neovascularization remains unknown, and requires further investigation. In the current study, the epitranscriptomic profile of m 6 A-modified circRNAs was established via microarray analysis in the OIR model in mice, followed by validation with MeRIP-qPCR. The potential functions of those circRNAs with altered m 6 A methylation levels were predicted by further bioinformatics analyses.

Materials & Methods
Establishment of OIR mouse model C57BL/6J mice obtained from the SJA Laboratory Animal Center (Changsha, Hunan, China) were used in the study. The OIR model was established as previously described [8,9]. Pups were exposed to hyperoxic conditions (75%) on postnatal day (P) 7, and moved back to room air at P12. Pups in the control group were exposed to room air continuously. At P17, the retinas of mice in both groups were harvested for use in subsequent experiments. Four retinas from two mice were mixed to constitute one sample, and three samples were assessed for microarray analysis of each group. The animal experiments were in accordance with the ARVO Statement and the procedures have been approved by the Institutional Animal Care and Use Committee of the Second Xiangya Hospital of Central South University.

RNA extraction and m 6 A immunoprecipitation
Total RNA was extracted from retinal tissues using TRIzol (Invitrogen, Carlsbad, CA, USA), the RNA purity and concentration were determined by a NanoDrop ND-1000 spectrophotometer. Either a Bioanalyzer 2100 or Mops electrophoresis was used to assess RNA integrity.
Total RNA and m 6 A spike-in control mixture was mixed with immunoprecipitation (IP) buffer, added the anti-m 6 A rabbit polyclonal antibody (Synaptic Systems, Goettingen, Germany), and then the preparation was incubated at 4°C for 2 h. For each sample, 20 μL Dynabeads M-280 sheep anti-rabbit IgG suspension (Invitrogen) was used, and the preparation was blocked with 0.5% BSA (4°C, 2 h). After washing, they were resuspended in the prepared mixture containing total RNA and antibody and incubated at 4°C for 2 h. After washing, the enriched RNA was eluted with elution buffer at 50°C for 1 h, and then acid phenol-chloroform and ethanol precipitation were used for RNA extraction.

Epitranscriptomic microarray analysis of circRNAs
IP and Sup RNAs were digested with RNase R (Epicentre, Inc.) to remove linear RNAs and enrich circRNAs. The Super RNA Labeling Kit (Arraystar, Rockville, MD, USA) was used to label Sup RNAs with Cy3, and IP RNAs with Cy5, and then preparations were purified via the RNeasy Mini Kit. The labeled cRNA mixture was fragmented, hybridized on an m 6 A-circRNA Epitranscriptomic Microarray slide (Arraystar) following the manufacturer's instructions, and scanned by a microarray scanner G2505C (Agilent Technologies, Santa Clara, CA, USA).
Agilent Feature Extraction software (V11.0.1.1) was used to analyze images of captured array. Average of the log2-scaled spike-in RNA intensities were used for normalization of the IP and Sup raw intensities. m 6 A quantities were calculated based on normalized IP amounts, and m 6 A methylation levels were calculated as the percentage of modification in the input according to normalized intensities. Raw data from the microarray analysis were deposited in the database of Gene Expression Omnibus (No. GSE213362). The associations between circRNA m 6 A methylation and expression levels were conducted by the intersection of the current m 6 A microarray data and the expression profile of circRNAs in OIR retinas in accordance with a previous study by our research group [16].

Methylated RNA immunoprecipitation-qPCR
Methylated RNA immunoprecipitation (MeRIP)-qPCR analysis was conducted to validate the microarray data. SuperScript TM III Reverse Transcriptase (Invitrogen) was used to synthesize the first-strand cDNA from IP RNA, and the system of QuantStudio TM 5 Real-Time PCR (Applied Biosystems, Foster City, CA, USA) with 2X PCR master mix (Arraystar) was used to conduct RT-qPCR. The IP fraction in the input was calculated as MERIP/input (%) as described in Xing et al. [23]. Data are presented as means ± SEM. Primer sequences are displayed in Table 1.

Bioinformatics analyses
Analyses of gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) were used to reveal the functional annotation and the pathways involved. The network of circRNA-miRNA-mRNA was constructed under the competing endogenous RNA (ceRNA) hypothesis.

Statistical analysis
The differences between the two groups were compared by Student's t-test. Significant differences in m 6 A-methylated circRNAs between the two groups in microarray analyses were determined by fold change ≥ 1.5 and p < 0.05. For MeRIP-qPCR data, p < 0.05 was deemed to indicate statistical significance.

OIR alters the profile of m 6 A methylation levels in circRNAs
A total of 88 circRNAs were differentially modified by m 6 A methylation in the retinas of the OIR group and the control group, including 56 hypermethylated circRNAs and 32 hypo-methylated circRNAs in the OIR group ( Figure 1A-1C). The top ten hyper-methylated and hypo-methylated circRNAs are listed in Table 2.

Associations between circRNA m 6 A methylation and expression levels
To investigate associations between circRNA expression and m 6 A methylation levels, circRNAs were intersected with altered m 6 A methylation and differentially expressed circRNAs by a previous study [16]. Using a threshold of ≥ 1.5-fold change, four interaction modes were identified (Figure 2), including 4 hyper-methylated and upregulated circRNAs, 20 hyper-methylated and downregulated circRNAs (6 circRNAs met the significance threshold of p < 0.05 in both methylation and expression), 2 hypo-methylated and upregulated circRNAs, and 24 hypo-methylated and downregulated circRNAs (10 circRNAs met the significance threshold of p < 0.05 in both methylation and expression).

Predictions of involved functions and pathways in host genes of the altered m 6 A-modified circRNAs
By GO enrichment analyses, host genes of the enriched hyper-methylated circRNAs were involved in cellular process, cellular anatomical entity, and protein binding ( Figure 3A). Those of the enriched hypo-methylated circRNAs were involved in the regulation of cellular biosynthetic process, the nucleus, and binding ( Figure 3B). KEGG analyses identified only three pathways, with host genes of hyper-methylated circRNAs involved in selenocompound metabolism and salivary secretion ( Figure  3C), and those of hypo-methylated circRNAs involved in lysine degradation ( Figure 3D).

Verification of altered methylation levels by
MeRIP-qPCR m 6 A methylation levels of four selected circRNAs assessed via MeRIP-qPCR are shown in Figure 4. The m 6 A methylation levels of mmu_circRNA_33363 and mmu_circRNA_002816 significantly decreased in OIR group, and m 6 A methylation level of mmu_circRNA_009692 was significantly increased (p < 0.05). The m 6 A methylation level of mmu_circRNA_27462 was slightly downregulated, but of no statistical significance (p > 0.05). The downward trend was consistent with the microarray analysis.

Discussion
M 6 A modification, as an essential epigenetic regulator, has been suggested to involve in vision function and pathogenesis of ophthalmic diseases [24]. Recently, with the wide application of high-throughput sequencing technologies, there is accumulating evidence that crosstalk between m 6 A modifications and circRNAs contributes to multifaceted physiological and pathological processes [25,26]. Therefore, a better understanding of m 6 A-modified circRNAs in the pathogenesis of retinal neovascularization is warranted.
The current study revealed the epitranscriptomic profile of m 6 A-modified circRNAs in OIR retinas. A total of 88 circRNAs with significantly altered m 6 A modifications were identified (Figure 1). The study also identified enriched biological functions via GO analysis, and the pathways involved via KEGG analysis, based on the host genes of hyper-methylated and hypo-methylated circRNAs ( Figure 3). These data and predictions shed light on the potential functions of m 6 A modifications in circRNA regulation of retinal neovascularization. For example, "lysine degrada- tion" was an enriched pathway of the host genes of hypo-methylated circRNAs. Dong et al. [27] demonstrated the presence and involvement of acrolein-lysine adduct in fibrovascular membrane in PDR patients. In our previous study, lysine was significantly upregulated in retinal tissues of mice with OIR [28]. Lysine degradation occurs mainly via saccharopine formation and the pipecolic acid pathway, and involves interaction with subcellular compartmentalization and enzyme deficiencies [29]. However, the specific roles of lysine degradation and the mechanisms involved require further investigation.
It has been suggested that understanding ceRNA crosstalk provides extraordinarily essential views of the regulatory roles of circRNAs [30]. A recent study indicated that m 6 A modification promotes the capability of binding of circALG1-an overexpressed circRNA in colorectal cancer-to miR-342-5p, enhancing ceRNA regulation, which may facilitate the development of a novel therapeutic method targeting colorectal cancer [31]. Li et al. [32] reported that m 6 A-dependent upregulation of circMETTL3 aggravated breast cancer progression by acting as a ceRNA via the circMETTL3/miR-31-5p/CDK1 axis. In the current study, MeRIP-qPCR results indicated significant alteration of m 6 A methylation levels of mmu_circRNA_33363, mmu_circRNA_002816, and mmu_circRNA_009692 ( Figure 4). Therefore, a circRNA-miRNA-mRNA network was constructed under hypothesis of ceRNA based on these three validated circRNAs ( Figure 5). GO analysis based on the 84 mRNAs involved showed that the majority of the enriched genes were involved in cellular process, cellular anatomical entity, and binding ( Figure 6A); indicating the potential importance of m 6 A-modified circRNAs in retinal neovascularization. Recent studies showed the involvement of RNA editing modifications in retinal and vascular pathogenesis [33,34], which indicated that the differential edited genes were enriched in pathways associated with angiogenesis, inflammation and apoptosis. It is also interesting to conduct the integrated analysis of multi-omics of the m 6 A methylation with other kinds of RNA modifications to further explore the cellular mechanisms in ischemia-induced retinopathy. Light blue nodes represent mRNAs, red nodes represent miRNAs, and brown nodes represent circRNAs with significantly altered m 6 A methylation levels. Edges with T-shaped arrows and those without arrows indicate directed and undirected relationships, respectively. In summary, the present study investigated the m 6 A modification alterations of circRNAs in the retinas of OIR. The possible relevant functions and pathways were also predicted by bioinformatics analyses. These provide novel insight into the involvement of m 6 A methylation in circRNA regulatory functions in OIR. Further investigations are needed to reveal the m 6 A epitranscriptomic profiles of circRNAs in clinical samples from retinal neovascular diseases, and to identify the specific roles of these m 6 A-modified circRNAs in the pathological process of ischemia-induced retinal neovascularization.