FTO alleviates Aβ1-40 induced retinal pigment epithelium degeneration via PKA/CREB signaling pathway

Amyloid-β (Aβ), a component of age-related macular degeneration (AMD) hallmark drusen, induces retinal pigment epithelium (RPE) cell degeneration and promotes the progress of AMD. Evidence shows that epigenetics mechanism is involved in the regulation of AMD. In this study, we aimed to investigate the roles of N6-methyladenosine (m6A) and its demethylase the fat mass and obesity-associated gene (FTO) in Aβ-mediated degeneration. The molecular characteristics and morphology of FTO were examined by quantitative Real-Time PCR (qRT-PCR), Western blot and immunouorescence. Inhibition of FTO was conducted to analyze its function on cell survival. Ocular Coherence Tomography and Fundus Photography was performed to evaluate the fundus of animal models. m6A-mRNA Epi-transcriptomic microarray, bioinformatics analysis and experimental verication were used to explore the potential molecular mechanism and signaling pathway. found and that form acid Mechanistically, we identied demethylated decreased


Abstract Background
Amyloid-β (Aβ), a component of age-related macular degeneration (AMD) hallmark drusen, induces retinal pigment epithelium (RPE) cell degeneration and promotes the progress of AMD. Evidence shows that epigenetics mechanism is involved in the regulation of AMD. In this study, we aimed to investigate the roles of N6-methyladenosine (m6A) and its demethylase the fat mass and obesity-associated gene (FTO) in Aβ-mediated degeneration.

Methods
The molecular characteristics and morphology of FTO were examined by quantitative Real-Time PCR (qRT-PCR), Western blot and immuno uorescence. Inhibition of FTO was conducted to analyze its function on cell survival. Ocular Coherence Tomography and Fundus Photography was performed to evaluate the fundus of animal models. m6A-mRNA Epi-transcriptomic microarray, bioinformatics analysis and experimental veri cation were used to explore the potential molecular mechanism and signaling pathway.

Results
We found overexpression of FTO in Aβ model and demonstrated that inhibition of FTO by the sodium form of Meclofenamic acid (MA1) aggravated RPE impairment. Mechanistically, we identi ed protein kinase A (PKA) as FTO's mediating target and found that FTO epigenetically demethylated PKA mRNA and decreased PKA expression, leading to suppressed PKA/ cyclin AMP-responsive element binding (CREB) signaling pathway. Moreover, inhibition of FTO promoted PKA/CREB signaling pathway inducing greater RPE degeneration and death.

Conclusions
These data demonstrated the functional signi cance of FTO in Aβ-induced RPE degeneration and the regulatory mechanism of PKA/CREB signaling pathway, implying FTO as a potentially therapeutic target for AMD.

Background
Age-related macular degeneration (AMD) is the leading cause of vision loss among people over 60. It is estimated that more than 196 million cases of AMD will occur in 2020 [1]. With the advent of antivascular endothelial growth factor (anti-VEGF), the clinical management of this disease have been revolutionized. However, despite advances in the understanding and treatment of AMD in the past decade, disease prevalence increases with the ageing human population [2]. A necessity is risen to clarify the pathological mechanism and nd out targets for early intervention.
Genetically con rmed that monozygotic twins can present initially with similar early macular pathology, but in the late stages of the disease, they can display different phenotypes [3]. Epigenetic mechanisms may explain some of the phenotypic variability in AMD. Epigenetics regulation includes DNA methylation, histone modi cation, chromatin remodeling and the regulation of non-coding RNAs [4]. Human AMD samples displayed signi cant inter-individual variation in RPE transcript expression [5], supporting that heterogenic stress responses underlie AMD pathological process. C57BL/6 mice intravitreally injected with Amyloid-β 1−40 (Aβ 1−40 ) showed degenerative changes and apoptosis within the retina, and microRNAs (mi-RNAs) expression pro ling proved differential expression of miRNAs between the Aβ model and the control group [6]. However, these ndings cannot fully explain the data obtained from other researches. Since translational repression and mRNA degradation are the outcomes of base pairing between miRNA and mRNA, mRNA modi cation may also have an impact on AMD pathogenesis [7]. Besides, miRNA regulation is species speci c, but mRNA modi cation is highly conserved in humans and mice, such as N6-methyl adenosine (m6A) RNA methylation. In this way, the mRNA modi cation mechanisms obtained from mice can be more easily translated into human diseases. Hence, examining post-transcriptional mRNA regulation can better understand epigenetic mechanisms under retinal degeneration and provide a more comprehensive picture of the biology of AMD progression.
N6-methyl adenosine (m6A) RNA methylation is the most abundant internal modi cations in RNA, regulating mRNA stability, protein expression, decay and several other cellular processes [8]. Emergent evidence has shown that the m6A modi cation plays an important role in the occurrence and development of human diseases. m6A levels in mRNA are found to be dynamic, varying during development, aging and in response to cellular stresses [9][10], indicating the role of m6A in age-related degeneration. Fat volume and obesity related (FTO) protein was the rst m6A demethylase [11]. It is generally expressed, especially in the brain. Some of FTO variants have impact on neural development that can cause severe growth retardation and diseases like Alzheimer's disease (AD). APOE allelic variant is a signi cant risk factor for AD [12]. Prospective study showed that FTO interacting with APOE, contributes to AD progression [13]. Interestingly, APOE is also a risk factors of AMD and is related to elevated cholesterol levels and Aβ metabolism [5]. Besides, retina shares the same precursor of the central nervous system (CNS) with brain [14]. With commonalities between AMD and AD of humans and APOE4 transgenic animals [15][16], and the fact that advanced AMD showed an increased risk for developing AD [17], we speculate overlapping molecular mechanism of AMD and AD concerning about APOE4 and Aβ as well as FTO, despite these being two completely distinct diseases.
AMD initially affects the retinal pigment epithelium (RPE), which is a mediator of retinal homeostasis [18]. Impaired RPE causes the formation of extracellular deposits collectively called drusen, which in turn induces RPE degeneration [19]. Increasing evidences support that Aβ, a main component of drusen, have played a role in AMD pathogenesis [20]. Evidence shows that Aβ induced cell degeneration via protein kinase A (PKA)/ cyclin AMP-responsive element binding (CREB) signaling pathway [21]. The phosphorylation of transcription factor CREB was downregulated in AD models APP/PS1 mouse, indicating Aβ's impact on PKA/CREB signaling pathway [22][23]. However, the activity of PKA in degenerated RPE cells changes under different conditions. While some reported that RPE cells from APP/PS1 mouse showed RPE degeneration and higher enzymatic activation of PKA [24], others suggested that RPE cells from Klotho-knockout mice showed RPE degeneration and lower PKA activation, which downregulated PKA/CREB signaling pathway and led to decreased phagocytosis of RPE, compared with normal human primary RPE cells [25]. This differential activity of PKA in degenerated RPE cells under Aβ treatment may concern about post-transcriptional regulation. Interestingly, FTO was found to alleviate the dephosphorylation of CREB and promote PKA/CREB signaling pathway and BDNF expression [26]. Collectively, above ndings point towards a possible association of FTO with RPE degeneration, via Aβ mediated PKA/CREB signaling pathway. Therefore, we identi ed the role of FTO in Aβ 1−40 induced RPE degeneration and investigated the underlying molecular mechanism by which FTO dependent m6A modi cation affects RPE damage and survival. We rst presented evidence that FTO played a protective role in Aβ 1−40 induced RPE degeneration at the cell and even tissue level. Then mechanistically, FTO was found to selectively demethylate PKA, regulating its mRNA and protein expression, and suppressing PKA/CREB signaling pathway, which partly rescued RPE from degradation. Our results revealed new functions of the m6A mechanism mediated by FTO in RPE damage and survival. We also proposed that targeting retinal epitranscriptome through FTO could be an effective treatment strategy for AMD.

Methods
Amyloid oligomerization. Lyophilized Aβ 1−40 (A1075, Sigma) was dissolved in deionized distilled water at a concentration of 1.386 mM and diluted by PBS at a concentration of 350 µM according to the instructions. Amyloid-β were incubated at 37 °C for 4 days and stored at 4 ℃.
Primary mouse RPE cell isolation and culture. The eyeballs were obtained from 3-week-old C57BL/6 mice.
To isolate primary RPE cells, connective tissues of eyeballs were cut off and RPE-choroid complex were separated from entire eyeballs. RPE cells were then gently swapped and transferred in Dulbecco's modi ed Eagle's medium /Ham's F-12 medium (Gibco) with 25% fetal bovine serum (Thermo Fisher Scienti c) and 1% P/S antibiotics (Beyotime). The cells were cultured at 37 °C in 5% CO2 cell incubator.
Animal model and treatment. To establish Aβ 1−40 mediated AMD mice, mice were anesthetized with 10% chloral hydrate, and its pupils were dilated by tropicamide (Santen Oy). A glass micropipette was used to mice. Mice with endophthalmitis will be excluded from the analysis.
Quantitative Real-Time PCR (qRT-PCR). Total RNAs of RPE-choroid tissues or RPE cells were extracted using RNAsimple Total RNA Kit (TIANGEN Biotech) and quanti ed. Complementary DNAs were synthesized from total RNAs using PrimeScript RT Master Mix (Takara Bio Inc.). Using TAKARA kit to convert RNA to cDNA (37 ℃ 15 minutes, 85 ℃ 5 seconds, 4 ℃ in nite time).
Quantitative RT-PCR system were performed by a SYBR green-based method and run by a 40-cycle ampli cation program (95 °C for 5 s, 60 °C for 30 s and 72 °C for 42 s) in real-time PCR detection system (Eppendorf). Relative enrichment was calculated using the 2 − ΔΔCt method, and internal controls (GAPDH) were used for normalization. All the primers sequences used in the experiments were listed in Table 1[see Additional le 1].
Western blot analysis. RPE-choroid tissues or RPE cells were lysed in RIPA buffer (Beyotime) containing the complete cocktail of protease inhibitors (Beyotime) and phosphorylase inhibitors (Beyotime). Protein concentrations were determined by BCA protein assay kit (Beyotime). Proteins were separated by 10 or 12% SDS-PAGE and transferred to polyvinylidene di uoride membranes. The membranes were blocked with blocking buffer (Tris-buffered saline Tween-20 (TBST), containing 5% nonfat dry milk) for 1 h at room temperature and blotted with related antibodies at 4 °C overnight. Secondary antibodies were incubated for 1 h at room temperature. The membranes were washed 3 times by TBST between each procedure. The membranes with target protein were exposed in visualizer (GE Healthcare). Antibodies were purchased from the following: anti-FTO (1:1000, Abcam), anti-PKA (1:2500, Abcam), anti-p-CREB Immunostaining. To obtain eyeball tissue slides, the mice were sacri ced after anesthesia and xed by PFA for 30 minutes. After removing its connective tissues, the eyeballs were extracted, xed by PFA for 2 hours, and dehydrated by 3% sucrose aqueous solution overnight. Dehydrated eyeballs were embedded by OCT, frozen at -80 °C, sliced for observation preparation.
Slides were observed using a uorescence microscope (Olympus) and a Leica TCS SP8 confocal laser scanning microscope (Leica TCS NT).
Cell proliferation CCK8. RPE cells were seeded at 1000 cells per well in 96-well plates with medium. Cell viability was assayed using Cell Counting Kit-8 (Beyotime) at the concentration in 0, 10, 50, 100, 500, 1000 µM/ml, according to operation manual and detected at 450 nm by Absorbance detector. The results of cell viability were expressed as a ratio (%) of treated group divided by untreated group.
Ocular Coherence tomography and Fundus Photography. Each mouse was anesthetized with pupils dilated and placed on a clean plate of OCT/FP detecting equipment (Cold Spring Harbor). The OCT and FP imaging was acquired through the pupil using a scan probe. Pictures were captured and saved in jpg format.
m6A-mRNA Epi-transcriptomic microarray. To perform m6A-mRNA&lncRNA Epitranscriptomic microarray, total RNAs from RPE-choroid complex of Aβ mice were extracted according to the protocols. Microarray experiments were performed by KangChen Bio-tech, Shanghai, China. methylated m6A-RNA binding protein immunoprecipitation. RPE-choroid complex tissues were xed using 1% methanal for 10 minutes and neutralized by Glycine Solution for 10 minutes at room temperature. After washed twice by PBS, tissues were lysed and incubated with m6A antibody for immunoprecipitation at 4 °C overnight. Enrichment of m6A containing mRNAs were pulled down by agar beads and washed by PBS twice. After collected, Total RNAs were extracted from beads using RNeasy Mini Kit (QIAGEN) and were converted to cDNA for further qRT-PCR with the primers listed in Above. Enrichment of m6A containing mRNA was analyzed by qRT-PCR with the primer's sequences as follows: PKA, forward 5′-ATGGTTATGGAATACGTCCCTGG-3′, reverse 5′-AATTAAGAGGTTTTCCGGCTTGA-3′; Statistics Analysis. Each experiment was repeated at least thrice. All the data were analyzed using Graphpad prism 8.0. Multiple t-test were used to compare the statistical difference between related groups. P-values < 0.05 was considered as statistical signi cance.

FTO was increased in Aβ induced RPE degeneration
To identify whether Aβ 1-40 treatment alters the expression of m6A regulators in RPE cells, we measured the levels of several major RNAs associated with m6A methylation (METTL3 and METTL14), demethylation (FTO and ALKBH5) and recognition (YTHDF1, YTHDF2 and YTHDF3). The primary mouse RPE cells were cultured with 2.5μM Aβ 1-40 for 24 hours. The result of quantitative real-time PCR (qRT-PCR) shown that FTO was up-regulated (2.06±1.38-fold) in Aβ group, compared with phosphate-buffered saline (PBS) group (Figure 1a). We further established the intravitreally injected Aβ 1-40 mice and detected the mRNA level of m6A modi cation RNAs in the RPE-choroid complex at day 4 ( Figure 1b). qRT-PCR data revealed that expression of FTO was increased (1.53±0.28-fold) at RNA levels as compared to its respective controls. To investigate whether Aβ 1-40 affects FTO protein expression in vivo, FTO expression in RPE-choroid complex was assessed at different time points exposed to Aβ (Figure 1c). In line with initial observation, notable increases of FTO were observed at day 2, day 4, and day 6, with the expression peaked at day 4. Therefore, we chose day 4 for subsequent studies. These ndings suggest that Aβ  can induce FTO transcription and expression in RPE cells.
To assess whether the Aβ 1-40 deposition induces RPE cell impairment, we used fundus photography (FP) to exam the morphology of fundus in Aβ mice at day 4 ( Figure 1d). In contrast to vehicle-injected eyes, the fundus of Aβ mice exhibited areas of depigmentation indicating RPE atrophy and underlying photoreceptor degeneration. This nding suggests that Aβ 1-40 deposition can spread through the retina and induce RPE degeneration. Next, to investigate the potential role of FTO in Aβ 1-40 -mediated RPE degeneration model, we analyzed immuno uorescence staining of Aβ-mice retinal sections and found that FTO showed strong uorescence signal in the nucleus and perinucleus of RPE cells, whereas slight nuclear signaling was observed in the PBS group (Figure 2a). For further con rmation, primary mouse RPE cells exposed to Aβ 1-40 for 24 h were analyzed with immuno uorescence staining. FTO uorescence signal was increased in the Aβ 1-40 group, and mainly concentrated in the nucleus and perinucleus area, To clarify the relationship between RPE degeneration and FTO expression, FP and ocular coherence tomography (OCT) were used to assess the fundus and retina of IV-PBS mice, IV-Aβ mice, IV-Aβ/MA1 mice and IV-MA1 mice (Figure 3e). While the fundus of IV-MA1 mice and IV-Aβ mice showed slight and moderate depigmentation comparing to IV-PBS mice, augmented depigmentation was observed in the fundus of IV-Aβ/MA1 mice, indicating that Aβ treatment induced RPE degeneration was aggravated by FTO inhibition. OCT identi ed no signi cant changes between IV-PBS mice and IV-MA1 mice. However, IV-Aβ/MA1 signi cantly aggravated retinal structure disorder, compared to IV-Aβ, con rming that the inhibition of FTO expression exacerbated RPE degeneration in Aβ-induced damaging model. These results suggest that FTO may play a protective role in IV-Aβ mice by rescuing RPE degeneration.
m6A-mRNA Epi-transcriptomic microarray identi ed PKA as a downstream target of FTO To investigate the molecular mechanism of FTO and identify its downstream targets in Aβ induced RPE degeneration, we performed m6A-mRNA Epi-transcriptomic Microarray to detect m6A levels of RNA extracted from RPE-choroid complex in Aβ mice and PBS mice. The volcano plot displayed that, m6A levels of 571 genes were differentially methylated among 37739 transcripts detected in the microarray (Figure 4a). 78.3% mRNAs (447 transcripts) were demethylated (green point) and 21.7% mRNA (124 transcripts) were methylated (red point). Since the increase of other m6A writers and readers could not fully explain the decrease in m6A, we suggest that FTO overexpression may result in the m6A demethylation in Aβ 1-40 induced RPE degeneration.
Gene Ontology (GO) analysis report revealed that Aβ 1-40 induced differential modi ed m6A RNAs mainly enriched in biological regulation of biological process, binding of molecular function, cell and cell part of cellular component (Figure 4b). Among these, we selected protein kinase A (PKA), a phosphorylase in cAMP/PKA/ cAMP-response element-binding protein (CREB) signaling pathway as a candidate target of FTO-mediated m6A modi cation for further investigation. Microarray identi ed that m6A level of PKA was downregulated in Aβ mice, and pathway analysis report suggested that cAMP/PKA/CREB signaling pathway was involved in regulation of above biological process (Figure 4c). PKA was activated by cAMP level and could phosphorylate CREB, a nuclear transcription factor, to its active state, promoting gene transcription and translation, such as brain derived neurotrophic factor (BDNF) (Figure 4d). To verify m6A level of PKA mRNA in Aβ mice, we performed m6A-RNA binding protein immunoprecipitation (MeRIP). qPCR analyses of MeRIP enriched RNA from Aβ mice con rmed that PKA transcripts were hypomethylated in Aβ mice (Figure 4e), suggesting that FTO overexpression may result in PKA mRNA hypomethylation. Taken together, these ndings hinted that PKA was a downstream target of FTO.

FTO suppressed PKA transcription and promoted translocation
To investigate the effect of FTO on PKA expression, we measured FTO and PKA protein level in Aβ mice and IV-Aβ/MA1 mice. While the expression of FTO was signi cantly increased in Aβ mice, PKA was decreased in comparison to the controls (Figure 5a). When the expression of FTO was inhibited by MA1 in IV-Aβ/MA1 mice, protein level of PKA was reversed, comparing to IV-Aβ mice (Figure 5b). Later, we con rmed the regulatory effect of FTO on PKA expression in vitro (Figure 5c). PKA was decreased in RPE cells of Aβ group comparing to PBS group. Meanwhile, the expression level of PKA was higher in Aβ/MA1 group compared to Aβ group. These results show that PKA was signi cantly suppressed by FTO in Aβinduced RPE degeneration model and this suppression could be reversed by MA1 treatment. Next, we used confocal microscopic analysis to detect the immuno uorescence of FTO and PKA expression. FTO and PKA showed signi cant overlaid uorescence signal around the nuclei of RPE cells in Aβ mice, whereas limited FTO and PKA signaling was observed in the PBS group (Figure 5d). Strong immuno uorescence of PKA was observed in the nuclear of primary RPE cells exposed to Aβ for 24h ( Figure 5e), indicating that FTO regulated not only PKA's expression, but also PKA's location. Taken together, these results indicate that PKA expression and translocation can be triggered by FTO overexpression.
Aβ induced FTO expression regulated PKA/CREB signaling pathway To investigate whether FTO mediated PKA/CREB signaling pathway, we detected the mRNA level of FTO and factors of PKA/CREB signaling pathway in Aβ mice (Figure 6a). qRT-PCR showed that FTO expression was increased at mRNA level in Aβ mice, but PKA showed slightly decrease without statistical signi cance. Downstream targets CREB and BDNF were signi cantly downregulated. We further evaluated the protein level of FTO and PKA/CREB signaling pathway by Western blot in Aβ mice ( Figure  6b). Consistently, a signi cant increase of FTO was observed, while others were decreased, suggesting that FTO induced PKA down-regulation would result in the suppression of PKA/CREB signaling pathway.
To con rm the regulatory role of FTO in PKA/CREB signaling pathway, we established FTO-inhibition model in vivo, and detected the mRNA level of FTO and factors of PKA/CREB signaling pathway ( Figure  6c). qPCR data showed that inhibiting FTO reduced the mRNA of FTO and PKA, but promoted mRNA expression of downstream CREB and BDNF, comparing to IV-Aβ mice. We further validated the regulatory effect of FTO through Western blot (Figure 6d). Compared to IV-Aβ mice, suppression of FTO signi cantly increased the protein levels of PKA, p-CREB and the precursor of BDNF (pro-BDNF), demonstrating that FTO inhibition signi cantly abolished the suppression of PKA/CREB signaling pathway.
We further validate the effect of FTO on PKA/CREB signaling pathway in vitro. qRT-PCR results of RPE cells from Aβ group showed that FTO was increased at mRNA level as well as PKA and CREB but BDNF showed slightly decrease without statistical signi cance, comparing to PBS group (Figure 6e). qRT-PCR results of RPE cells from Aβ/MA1 group showed signi cantly reduction in mRNA levels of FTO, PKA and BDNF, but an increase in CREB, comparing to Aβ group (Figure 6f). Next, we performed Western blot and found that downregulation of FTO caused upregulation of PKA, p-CREB and pro-BDNF (Figure 6g). Taken together, these results indicated that the FTO dependent protective effects on RPE survival and repair was mediated via PKA/CREB signaling pathway.

Discussion
In this study, we investigated FTO-dependent m6A modi cation underlying Aβ induced RPE degeneration. Major m6A modifying enzymes include methyltransferase, demethylase, and methyl recognition protein.
While methyltransferase and demethylase dynamically regulate m6A levels, m6A readers recognize m6A sites and facilitate subsequent process [28]. Exploring expression levels of these enzymes will help us understand the overall m6A levels and infer potential pathways in the pathological process. While we selected FTO as our research subject in Aβ induced damaging model, the mRNAs level of YTHDF1 was increased in vitro and in vivo (Figure 1a and b). As 'm6A reader', YTHDF1 recognize speci c m6A mRNA sequence and promote mRNA translation [29]. In addition, current study discovered the stimulusdependent characteristics of YTHDF1 [30]. YTHDF1-dependent translation was increased in neurons stimulated by potassium chloride depolarization but showed no alteration before the stimulation. However, this characteristic was restricted in certain cells and YTHDF1's translation was consistently activated in other cells like cancer cells. This nding implied that Aβ may trigger the stimulus-dependent characteristics of YTHDF1, leading to its changeable mRNA level. In our study, we found that over expression of FTO was related to RPE's protection, and suppression of FTO aggravated RPE's degeneration, the above ndings show no signi cant connection with YTHDF1 in the regulation of Aβ induced RPE injury.
MA selectively inhibits FTO by competitively binding with m6A enriched mRNAs. As the sodium form of MA, MA1 still exhibits similarly inhibitory activity on FTO with MA [25]. Despite of its inhibitory function, we observed slight damage in the retina of MA1 mice (Figure 3e). Although earlier studies have reported that high doses of MA may associate with retinal toxicity by inhibiting gap junction conductance of RPE cells [31-32-33], others showed that MA can reduce the inhibition of RPE phagocytosis in acute acrolein and acrolein/hydrogen peroxide combined toxicity models [34], indicating that moderate doses of MA were feasible in the retina. Therefore, we chose 100μM MA1 by CCK8 cell viability test to reduce the side effect of MA1 on RPE cells (Figure 3a and b). Previous study showed that MA could effectively inhibited FTO's function and expression, in uencing its downstream pathways [35]. In our study, we used MA1 to inhibit FTO demethylation in Aβ induced damaging model. We found that FTO inhibited PKA/CREB signaling pathway and downstream target pro-BDNF, and this effect was reversed due to the treatment of MA1. m6A modi cations have multiple effects on mRNA, including mRNA stability, localization, degradation, or translation, which depend on the modi ed sites of mRNA [36]. From our m6A mRNA-epitranscriptomic microarray analysis, we identi ed PKA as a candidate of FTO regulation. Our study found that PKA methylation levels were downregulated due to an increase in FTO (Figure 4e). In addition, as PKA gene level slightly decreased without statistic difference, the protein level of PKA was signi cantly reduced in Aβ mice (Figure 6a and b), indicating that decreased m6A level of PKA may not directly related to its stability, but to its translation process. Later we found that inhibition of FTO reduced PKA gene level, however, increased PKA protein level in IV-Aβ/MA1 mice (Figure 6c and d), suggesting that decreased FTO may promote the translation of PKA mRNA, which is consistent with previous review. We further con rmed in vitro that decreased m6A level of PKA disturbed translation process in RPE cells (Figure 6e to g), but we were unable to clarify the speci c mechanism of m6A modi cation on PKA mRNA, because the m6A-mRNA epitranscriptomic microarray could only recognizes the entire sequence enriched with m6A, such as PKA transcript. Further investigation will be needed to study precise m6A sites on PKA mRNA in order to better understand the mechanism of methylation modi cation.
As a functional target of PKA/CREB signaling pathway, mature BDNF(m-BDNF) is derived from pro-BDNF after proteolytic cleavage in the extracellular space [37]. The formation rate of m-BDNF from pro-BDNF was mediated by the expression of pro-BDNF and tPA-and p11-mediated cleavage process [38]. While previous research mainly focused on the neuroprotective factor m-BDNF [39], the pro-apoptosis characteristic of pro-BDNF in AMD are yet fully illuminated. Pro-BDNF binds to p75 neurotrophin receptor (p75NTR) and coreceptor sortilin with high a nity, thereby triggering apoptosis [40]. pro-BDNF dominated over m-BDNF in OXYS mice retina during the development of AMD-like retinopathy, indicating participation of pro-BDNF and the imbalance of neurotrophic factors in the pathogenesis of AMD [37]. The expression of m-BDNF did not change during aging, meanwhile the level of pro-BDNF was signi cantly increased in the brain of rat [41][42]. We held the opinion that the transformation from pro-BDNF to m-BDNF was impaired because of the dysfunction of cleavage enzymes in the elderly, resulting in injury aggravating, but we haven't con rmed that yet. Since apoptosis is the major pathologic process for RPE in AMD [43], pro-BDNF play an important role in the progress of RPE degeneration. Besides, pro-BDNF increased Aβ level in neuron in vitro, and accelerated Aβ deposition in the brain [44], suggesting that pro-BDNF may play a crucial negative role in the pathogenesis and progression of AD and ageing mice [45], which may also happen in AMD. In our study, we identi ed pro-BDNF decrease in Aβ induced RPE degeneration. Since downstream target pro-BDNF may trigger apoptosis [46], down-regulated PKA/CREB signaling pathway was in favor of RPE protection. Owing to above results, we suspected that FTO mediated pro-BDNF decrease was bene cial for RPE's survival.

Conclusions
In summary, we provided evidence that FTO was upregulated in Aβ induced AMD models. High expression of FTO was closely related to protection of RPE by downregulating m6A level of PKA mRNA and PKA expression. Inhibition of FTO aggravated RPE degeneration by activating PKA/CREB signaling pathway. Taken together, our ndings suggested that FTO and PKA/CREB signaling pathway may serve as a protective target for AMD.

Additional Files
Additional le 1. Quantitative RT-PCR primers.xls Additional le 2. Figure 7 Working schematic.pdf Aβ induced upregulation of FTO epigenetically demethylated PKA and disturbed its translation, leading to suppressed PKA/CREB signaling pathway and pro-BDNF. Inhibition of FTO reversed the suppression of PKA/CREB signaling pathway and promoted pro-BDNF expression, which aggravated RPE degeneration.

Declarations
Ethics approval and consent to participate

Consent for publication
Not applicable

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare no con icts of interest. Y.H., J.C., J.S., Y.W., P.H., J.F., T.L. and X.S. contributed to the analysis and interpretation of the data. Y.H. and J.C. wrote the paper. J.C., J.S., T.L. and X.S. contributed to the manuscript revision. All authors read the manuscript and approved the nal version. The fundus of Aβ mice exhibited areas of depigmentation (arrows) compared with that in PBS mice. Histograms represent the mean and S.E.M. N= 5, *P ≤ 0.05 **P ≤ 0.01, ***P ≤ 0.001 via Student's t-tests.     Scale bar: 10μm. (e) Immunoreactivity for PKA (red) and RPE65 (green)was detected in the RPE cells (arrows) in the primary mouse RPE cells exposed to Aβ1-40 at 24 h. Scale bar: 25μm.