All the frozen clinical samples (normal from the patients with hepatic hemangioma, HCC tumor and paired tumor adjacent tissues from the patients with hepatitis B virus infection) in this study were randomly collected from Eastern Hepatobiliary Surgery Hospital, Naval Medical University (Shanghai, China). Written informed consents were obtained from all patients amenable to the policies of the committee, and the study was approved by the Ethics Committee of Naval Medical University.

Wild-type C57BL/6 male mice were purchased from JiHui experimental animal Co. (Shanghai, China). For murine NAFLD model, the 5-week-old C57BL/6 male mice were assigned randomly to five groups (n = 5): control diet group (Control), high‐fat diet group (HFD, Dyets, 60% kcal from fat diet), HFD + 0.1% rhein diet group (HFD + Rhein, Dalian Meilun, China), HFD + AAV-NC group, and HFD + AAV-oeFTO group. The 5-week-old male mice were fed with these diets for a total of 14 weeks. For the murine chronic liver injury model, the 2-week-old male mice received intraperitoneal injection of 2 mg/kg N-nitrosodiethylamine (DEN) (CAS: 55-18-5, RHAWN). When the mice with DEN administration were 4 weeks old, mice received intraperitoneal injections of 20% CCl₄ in olive oil at 5 ml/kg twice a week for 14 weeks and were fed with a control diet. For studies of FTO overexpression, each C57BL/6 male mouse was intravenously injected with a single dose of 1 × 10¹¹ vector genomes of adeno-associated virus (AAV) serotype 8 carrying either FTO (AAV-oeFTO: pAAV-CMV-Fto-3FLAG) or control gene (AAV-NC: pAAV-CMV-MCS-3FLAG) via the portal vein. For better transfection, six weeks later after the injection, mice were started to use for further experiments. AAV-NC and AAV-oeFTO were constructed and purchased from OBiO Technology (Shanghai, China). All mice were housed under a 12 light/12dark cycle at a temperature of 21 ± 2 °C and with a humidity of 60 ± 5% at the animal facility of Naval Medical University. All experiments using mice were carried out following animal protocols approved by the Ethics Committee of Naval Medical University.

HCC-derived Huh-7 cell line was obtained from the Chinese Academy of Sciences Cell Bank (Shanghai, China) and cultured in Dulbecco’s modified Eagle’s medium (HyClone) containing 10% fetal bovine serum (Gibco, USA) and 1% penicillin/streptomycin at 37°C, in an atmosphere of 5% CO₂.

The lentiviral knockdown vector pLKD-CMV-G&PR-U6-shFTO (5′-GGAGCTCCATAAAGAGGTT-3′) and the lentiviral overexpression vector pLenti-CMV-FTO-3flag-PGK-EGFP-2A-Puro were constructed by OBiO Technology (Shanghai, China). Recombinant lentiviruses containing full-length FTO (oeFTO), FTO control (oeNC), shRNA targeting FTO (shFTO), and control shRNA (shNC) were used to transfect cells following the protocol provided by OBiO Technology (Shanghai, China). 2 mg/mL puromycin was used to select for stable cells.

The wide type target site sequence and the 3′-UTR mutated sequence of the IL-17RA were synthesized and the plasmid overexpressing IL-17RA was constructed and provided by Genomeditech Co., LTD (Shanghai, China). IL-17RA was cloned into vector PGMLV-6751 (BamHI, XhoI) (forward primer1, 5’-GCGAATTCGAAGTATACCTCGAGGCCACCATGGGGGCCGCAC-3’; reverse primer1, 5’-CCAGGCCCCGGAATTGGTTCTGGAGTGTCTGGCATTTCTG-3’; forward primer2, 5’-AATTCCGGGGCCTGGAAGTGAAAAATACAGTGATGAC-3’; reverse primer2, 5’-CGTCATGGTCTTTGTAGTCGGATCCTGCACTGGGCCCCTCTGACTC-3’). The wild type plasmid overexpressing IL-17RA (PGMLV-CMV-H_IL17RA-3×Flag-EF1-ZsGreen1-T2A-Puro) was named oe_IL-17RA_WT, and site-directed mutants plasmid (PGMLV-CMV-H_IL17RA (c.A2319C)-3×Flag-EF1-ZsGreen1-T2A-Puro) overexpressing IL-17RA was named oe_IL-17RA_MUT. (The cloned sequences are listed in Supplementary Tables S1, 2 ). The empty plasmids were transfected as negative controls. For cell transfection, Huh7 cells were cultured in 6-well plates to reach a confluence of 70 – 90% and then transfected with IL-17RA overexpression plasmids (WT and MUT) and NC plasmids using Lipofectamine 3000 reagent (Invitrogen) according to the manufacturer’s instructions. After 48 hours of transfection, transfection efficiency was detected by quantitative real-time polymerase chain reaction (qRT-PCR) and Western blot.

MeRIP-seq was performed as described previously by Shanghai Jiayin Biotechnology Ltd (Shanghai, China) (38). In brief, total RNA was isolated and fragmented into ~100-nucleotide-long fragments. Approximately 5% of fragmented RNA as input RNA, other RNA was analyzed by immunoprecipitation using affinity-purified anti-m6A polyclonal antibodies (ABE572, Millipore, Germany). Sequencing was carried out using an Illumina NovaSeq 6000 platform. The method of analysis is based on the previously described m6A-seq protocol.

The RIP experiment was conducted with the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) following the manufacturer’s instructions. The m6A antibody (synaptic systems) and anti-rabbit IgG (Millipore, USA) were used for the RIP experiment. Total RNA was performed as input control. The interested RNAs were detected by qPCR. Relative enrichment was normalized to the input: %Input =1/10 × 2^{Ct [IP] – Ct [input]}.

Total protein was extracted from the frozen clinical tissues, murine model tissues or cells with radioimmunoprecipitation assay (RIPA) Lysis Buffer and PMSF (Beyotime, Shanghai, China). The protein was electrophoresed on 7.5%-12% SDS–polyacrylamide gels according to the different proteins molecular weight and transferred to nitrocellulose membranes (Bio-Rad, Hercules, USA). The membranes were incubated with the anti-IL-17RA receptor antibody, anti-FTO antibody, anti-ALKBH5 antibody (Abcam, Cambridge, USA), anti-METTL3 antibody (Zen bio, China), anti-METTL14 antibody (Sigma-Aldrich, Germany), β-Actin antibody (Proteintech, USA) overnight at 4°C. And then the membranes were incubated with secondary antibody IRdye 680RD goat (polyclonal) anti-mouse IgG (H+L) and IRdye 800CW goat (polyclonal) anti-rabbit IgG (H+L) (Licor Biosciences, Nebraska, USA) for 2 hours at room temperature. The membranes were detected using an Odyssey infrared scanner (Licor Biosciences, Nebraska, USA). The antibodies are listed in Table S3 .

Total RNA from the frozen clinical tissues, murine models, or cells were extracted with Trizol reagent (Takara, Dalian, China). For reverse transcriptions, cDNA was generated using a Reverse Transcription Kit (Takara, Dalian, China). Quantitative real-time PCR was performed using SYBR^® Green (Takara, Dalian, China) in a StepOne™ Real-Time PCR System (Applied Biosystems). β-actin was employed as an endogenous control. The relative mRNA levels were calculated using the comparative Ct method. The PCR primers are listed in Table S4 .

Liver tissues from murine models were fixed with 4% paraformaldehyde (Servicebio, China) for more than 24 hours, then embedded in paraffin and sectioned at 4-5 μm slices. For H&E staining, hematoxylin and eosin were used in the micro-section to analyze the histopathological damage in the liver from all experiment groups. For Sirius red staining, the slices were stained with picrosirius red solution for 8-10 min and then dehydrated and mounted in xylene. For Oil red O staining, lipid droplet accumulation was stained using 8-10 μm of frozen liver slices following a standard protocol. All the images were reviewed under the microscope (Eclipse Ci-L, Nikon, Japan) at 20× magnification and analyzed under Image-Pro Plus 6.0 (Media Cybemetics, USA) by Servicebio company (Wuhan, China). The NAFLD activity score (NAS) was calculated from the sum of the individual scores for steatosis, inflammation and ballooning. The NAFLD activity score is listed in Table S5 .

The obtained mice blood was centrifuged at 3000 × g for 10 min, and the supernatant was collected. The serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), triglyceride (TG), and the tissues levels of TG and hydroxyproline (HYP) were measured by using ALT assay kit (C009-2), AST assay kit (C010-2), TG assay kit (A110-1), and HYP assay kit (A030-2) following Nanjing Jiancheng Institute of Biotechnology protocols (Jiangsu, China).

Mouse serum was prepared by centrifugation at 1000 × g for 10 min at 4°C and stored in aliquots at −80°C until analysis. Cytokine concentration was measured in all of the mouse serum samples using a Mouse High Sensitivity T Cell Magnetic Bead Panel (MHSTCMAG-70K-08, EMD Millipore, Billerica, MA, USA) on the Luminex 100/200 multiplex immunoassay system according to the manufacturer’s instructions. The data were analyzed using Belysa 1.0 software (luminex 200).

All experiments were repeated at least three times and expressed as the means ± SEM. SPSS (version 17.0) software was used to analyze the data in this study. For comparisons involving three groups, data were analyzed using a one-way analysis of variance (ANOVA) followed by the least significant difference (LSD) test. For comparisons involving only two groups, data were analyzed with Student’s t test (two-tailed). R software (version 4.0.3) including ggplot2 and pheatmap packages was used for plot drawing. The following data significance levels were used for comparisons between independent groups: ns, not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

To investigate the dynamic changes and potential mechanisms of m6A modification during the progression of HCC, we performed MeRIP-seq using normal, HCC tumor and paired tumor adjacent tissues. The m6A consensus modified motif GGAC was identified in all samples obtained based on E-value ( Figure 1A ). Then, we evaluated dimensionality reduction on the sequencing data by principal component analysis (PCA), and results showed that three kinds of samples could be classified accurately ( Figure 1B ). Heatmap of m6A methylation levels also showed significant differences among the three kinds of samples ( Figure 1C ). To further explore the function of differentially m6A modified genes in the progression of HCC, KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis was performed. The top three pathways including herpes simplex virus 1 infection, NF-kappa B signaling pathway and cytokine-cytokine receptor interaction were significantly enriched ( Figure 1D ). Considering the important role of cytokines during the process of HCC (39), we took a closer look at the genes of m6A modification related to the cytokine-cytokine receptor interaction by heatmap and found that these genes were significantly decreased in tumor adjacent and HCC tumor tissues ( Figure 1E ). Compared with normal liver tissues, there were no significant difference between tumor adjacent tissues and HCC tumor tissues, especially IL-17RA ( Figure 1E ). Mapping of m6A methylation also revealed that m6A mRNA methylation level of IL-17RA exhibited a significant decrease in the tumor adjacent tissues, compared with normal tissues ( Figure 1F ). Importantly, the upregulation of IL-17RA in the tumor adjacent tissues was confirmed at protein level determined by Western blot ( Figures 1G, H ). Together, these results suggested that m6A mRNA methylation might be responsible for the abnormal expression of cytokine receptor IL-17RA in human livers, contributing to liver inflammation and fibrosis.

The above MeRIP-seq data suggested that m6A mRNA methylation of IL-17RA was significantly decreased in the tumor adjacent tissues. In order to unravel the driving forces behind the markedly changed m6A patterns in the tumor adjacent tissues, we analyzed the expression of proteins related to coordinate m6A modifications by Western blot, including m6A methylation writers (METTL3, METT14) and erasers (FTO, ALKBH5) ( Figure 2A ). Compared with normal tissues, the upregulation of FTO in the tumor adjacent tissues was confirmed at protein level ( Figure 2B ), while the expression of ALKBH5 showed no significant difference between normal and tumor adjacent tissues ( Figure 2C ). Besides, the expression of METTL14 and METTL3 at protein level were significantly upregulated in the tumor adjacent tissues ( Figures 2D, E ). Therefore, these results showed that FTO is a main reason for the decrease of m6A mRNA methylation of IL-17RA.

To further investigate the relationship between FTO and IL-17RA, we utilized Huh7 cells to construct cell lines stably underexpressing or overexpressing FTO using lentiviral vectors and the results showed the decreased and markedly increased FTO expression at transcript and protein levels, respectively ( Figures 3A, B, E, G ). Then, the mRNA level of IL-17RA was examined, which indicated no significant change in shFTO or oeFTO cell lines ( Figures 3C, D ). However, the protein level of IL-17RA extremely decreased in shFTO cell line and elevated in oeFTO cell line, implying the existence of post-transcriptional modification of IL-17RA ( Figures 3F, H ). Therefore, the wild type plasmid overexpressing IL-17RA (oe_IL-17RA_WT) and the site-directed mutants (in the site of m6A modification of IL-17RA) plasmid overexpressing IL-17RA (oe_IL-17RA_MUT) were constructed and transinfected into Huh7 cells, the result of which showed no obvious change at transcript level between cells with oe_IL-17RA_WT and oe_IL-17RA_MUT ( Figure 3I ). On the contrary, cells with oe_IL-17RA_MUT exhibited higher protein expression of IL-17RA compared with cells of oe_IL-17RA_WT ( Figure 3J ). Meanwhile, we found relative m6A enrichment of IL-17RA mRNA reduced in Huh7 cells after FTO overexpression by m6A-RIP-qPCR ( Figure 3K ). Taken together, the results above demonstrated the protein of IL-17RA was regulated by FTO via m6A modification in vitro.

In order to further verify that the modulator function of FTO on IL-17RA in vivo, murine NAFLD model was established as performed previously ( Figure 4A ). Compared with HFD group, NAS score of H&E staining in HFD + Rhein [inhibitor of FTO (40)] group showed apparently lower, manifesting in reduced inflammation infiltration and steatosis ( Figures 4B, C ). Meanwhile, positive staining area of Oil red O used for estimation of liver fat accumulation in mice group with HFD treated was at almost twice the group with HFD + Rhein ( Figure 4D ). Indeed, treatment with Rhein not only reduced TG level in serum and liver tissues, but the level of ALT and AST level as well ( Figures 4E, F, H, I ). Of note, Rhein feeding unexpectedly reduced the body weight and liver weight of the mice, which exhibited no significant change compared with the control group, suggesting the probably functional role of Rhein in the people of HFD ( Figures 4G, J ). Contrarily, mice administrated with AAV-oeFTO via intravenous injection performed higher total NAS score, positive staining area of Oil red O, ALT and AST level compared with HFD + AAV-NC group ( Figures 4C, D, H, I ). In addition, mice overexpressing FTO showed no significant change in TG, body weight and liver weight in comparison to HFD + AAV-NC group ( Figures 4E, F, G, J ). Consequently, we confirmed the functional role of FTO in liver damage in murine NAFLD model.

Besides the markers related to liver damage, we also examined certain inflammatory factors including IL-17A, IL-1α and IL-1β in murine NAFLD model and results showed treatment with Rhein obviously suppressed the increase of IL-17A, IL-1α and IL-1β due to the high-fat diet in murine NAFLD model ( Figures 5A–C ). Reversely, mice overexpressing FTO using AAV-oeFTO vector exhibited higher levels of proinflammatory factors ( Figures 5A–C ). Additionally, the relationship between FTO and IL-17RA was determined as well in vivo. Consistent with the results in vitro, FTO protein level was suppressed in HFD + Rhein treatment group and in the meantime, the expression of IL-17RA also decreased ( Figures 5D, E ). On the contrary, overexpression of FTO contributed to the increase of IL-17RA ( Figures 5F, G ). Then, m6A-RIP-qPCR was performed to quantify the m6A methylation modification of IL-17RA and we found that the reduction of FTO by Rhein led to the higher relative m6A enrichment of IL-17RA mRNA and vice versa ( Figures 5H, I ). Thus, we concluded that FTO facilitates the expression of IL-17RA via suppression of m6A methylation modification to accelerate the liver inflammation in murine NAFLD model.

Similar results showing FTO contributes to liver damage were obtained using another model of DEN + CCl₄-induced chronic liver injury in murine ( Figure 6A ). Inflammation score of H&E staining in the group exposure to DEN + CCl₄ increased significantly compared with the control group and addition of AAV containing full-length FTO further aggravated the condition ( Figures 6B, C ). Besides, positive staining area of Sirius red used for estimation of liver fibrosis in mice group with DEN + CCl₄ + AAV-oeFTO treated was bigger than the counterpart ( Figure 6D ). In fact, overexpression of FTO by AAV also promoted the level of hydroxyproline, TG, ALT and AST in liver tissues or serum ( Figures 6E–G, I, J ), but had no effect on the body and liver weight of mice ( Figures 6H, K ). In line with the murine NAFLD model, mice in DEN + CCl₄ + AAV-oeFTO group displayed higher IL-17A, IL-1α and IL-1β level in serum ( Figures 7A–C ). Similarly, we found a positive correlation relation between the expression of FTO and IL-17RA ( Figures 7D, E ). In addition, the result of m6A-RIP-qPCR showed a significant reduction of m6A methylation modification of IL-17RA in DEN + CCl₄ + AAV-oeFTO group, suggesting that FTO could force the translation of IL-17RA via downregulation of methylation level of IL-17RA mRNA ( Figure 7F ).