The m6A demethylase ALKBH5 promotes tumor progression by inhibiting RIG-I expression and interferon alpha production through the IKKε/TBK1/IRF3 pathway in head and neck squamous cell carcinoma

To evaluate the expression profile of m6A “writers”, “erasers”, and “readers” (WERs) in HNSCC, The Cancer Genome Atlas (TCGA) database was analyzed to show that half of the m6A WERs were dysregulated in HNSCC (Fig. 1a). Moreover, AKLBH5 was upregulated and had the highest abundance among the 14 WERs, making it an ideal research object in HNSCC (Fig. 1b). A dot blot assay using an m6A antibody was performed and revealed that the m6A level was lower in HNSCC tissues and cell lines (Fig. 1c, d). The percentage of m6A modification downregulated in HNSCC was compared with that in paired normal tissues using an m6A RNA Methylation Quantification Kit (Fig. 1e). The m6A modification level was lower in most HNSCC cell lines than in primary oral keratinocytes (Fig. 1f). We speculated that the downregulation of m6A in HNSCC may be attributed to demethylase overexpression. On the basis of the expression profile results, we then evaluated the expression of the demethylase ALKBH5 in samples from 138 HNSCC patients and 20 normal oral mucosal samples using immunohistochemistry (Fig. 1g). We found the upregulation of ALKBH5 in HNSCC tissues (Fig. 1h), and its higher expression correlated with advanced TNM stage and poor prognosis in HNSCC but not with other parameters (Fig. 1i, j and Supplementary Fig. S1a-d). ALKBH5 had a sensitivity of 65.7% for HNSCC diagnosis according to the receiver operating characteristic (ROC) curve (Supplementary Fig. S1e). In addition, ALKBH5 was highly expressed in the majority of human cancers, including HNSCC (Supplementary Fig. S1f).

To date, only two m6A demethylases, ALKBH5 and FTO, have been identified. Therefore, we also evaluated the expression of another m6A demethylase, FTO, in samples from 138 HNSCC patients and 20 normal oral mucosal samples using immunohistochemistry. Interestingly, FTO was also upregulated in HNSCC tissues, and its overexpression correlated with advanced TNM stage and poor prognosis in HNSCC (Supplementary Fig. S2a-i). FTO was highly expressed in half of human cancers, including HNSCC (Supplementary Fig. S2j). Consistent with the above results, ALKBH5 protein expression was higher in five tumor tissues than in the adjacent normal oral epithelium tissues (Fig. 1k). Similarly, HNSCC cell lines had higher protein and mRNA expression than normal oral keratinocytes (NOKs) (Fig. 1l, m). These results indicate that upregulated m6A demethylase promotes low m6A status and is correlated with poor prognosis in HNSCC.

To explore the role of ALKBH5 in tumor development, we knocked down ALKBH5 expression in HN4 and Cal27 cells using small interfering RNA (siRNA) (Fig. 2a and Supplementary Fig. S3). ALKBH5 inhibition significantly reduced proliferation (Fig. 2b) and colony formation (Fig. 2c) in cell lines. In addition, silencing ALKBH5 reduced DNA replication activity, as shown using an EdU assay (Fig. 2d and Supplementary Fig. S4). The proportions of G1 phase and apoptotic cells also notably increased after ALKBH5 knockdown (Fig. 2e and Supplementary Fig. S5. S6). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of genes with changes in expression after ALKBH5 silencing pointed to cell cycle and DNA replication processes (Fig. 2f). Furthermore, ALKBH5 inhibition suppressed migration and invasion (Supplementary Fig. S7). These results suggested that ALKBH5 acts as an oncogene driving HNSCC development.

To determine the function of ALKBH5 in tumor growth, knockdown and overexpression models were constructed in Cal27 cell lines using lentivirus (Fig. 2g). We showed that knockdown of ALKBH5 expression inhibited tumor growth and weight, while overexpression promoted tumor growth (Fig. 2h, i). Moreover, knockdown of ALKBH5 expression increased cell apoptosis, as revealed by TUNEL staining, while overexpression decreased apoptosis (Fig. 2j). The level of Ki-67, a proliferation marker, decreased after ALKBH5 knockdown and vice versa (Fig. 2k). The above results indicate that ALKBH5 plays a fundamental role in HNSCC development.

To understand the mechanism of ALKBH5 in gene expression, both transcriptome and epitranscriptome sequencing were conducted. ALKBH5 depletion increased the total m6A abundance in HNSCC cell lines (Fig. 3a, b). Methylated RNA immunoprecipitation (MeRIP) with an m6A-specific antibody followed by RNA sequencing (MeRIP-seq) and conventional RNA sequencing were combined to analyze the downstream targets of ALKBH5 in Cal27 cells. A total of 171 genes that were differentially expressed and differentially m6A-modified were screened (Fig. 3c and Supplementary Fig. S8). We mapped the m6A motif with independent biological replicates. The GGACU motif was identified to be highly enriched within the m6A site, consistent with the m6A consensus sequence RRACH (R = G or A; H = A, C or U) (Fig. 3d). In general, m6A-seq of control and ALKBH5-deficient cells identified 2080 and 1962 unique m6A peaks, respectively, and 1172 and 1071 unique m6A-modified genes, respectively (Fig. 3e, f). We further investigated the m6A distribution patterns within both the total and unique peaks. Similar patterns of total and common m6A distribution were observed in control and ALKBH5-deficient cells. Interestingly, a relative increase in m6A in coding sequences (CDSs) was observed in an ALKBH5-dependent manner (Fig. 3g).

To explore the biological processes downstream of ALKBH5 silencing, the differentially expressed genes (DEGs) were analyzed by functional enrichment analysis. The DEGs were analyzed and grouped using the cluster analysis method into apoptosis, migration, and immunity (Fig. 3h). Gene Ontology (GO) analysis showed that the DEGs were associated with tumor development processes, such as negative regulation of growth and cell cycle arrest, and innate immunity processes, such as cellular response to type I interferon and response to dsRNA (Fig. 3i).

A bubble chart that showed the enrichment of the biological process of interest using the DEGs obtained by sequencing was created (Fig. 4a). Then, the expression of the DEGs of interest in the bubble chart, including those involved in cellular response to type I interferon, were verified using qPCR after ALKBH5 silencing in two cell lines. The increase in DDX58 after ALKBH5 knockdown was the most significant and was relatively consistent between the cell lines (Fig. 4b, c). Volcano plots showed that DDX58 had relatively high mRNA expression and m6A methylation compared to the control (Supplementary Fig. S9). Other genes had similar patterns of expression/methylation as DDX58, as presented in the Supplementary Table. S2. Visualization analysis indicated notable m6A enrichment in the 3′-untranslated regions (UTRs) of DDX58 mRNA (Fig. 4d). MeRIP-PCR using an m6A-specific antibody confirmed the enrichment of m6A in the 3’UTR after ALKBH5 knockdown in HN4 and Cal27 cell lines (Fig. 4e). These results indicated that DDX58 was a target of m6A methylation mediated by ALKBH5 in HNSCC.

DDX58 encodes the protein RIG-I, which is a cytoplasmic receptor that recognizes viral RNA and plays a critical role in innate immunity and type I interferon production. We found that ALKBH5 knockdown significantly increased the protein and mRNA expression of RIG-I (Fig. 4f and Supplementary Fig. S10). Knockdown of RIG-I expression reversed this effect (Fig. 4g). Conversely, the overexpression of wild-type (WT) ALKBH5 but not mutant H204A decreased RIG-I expression (Fig. 4h), indicating that ALKBH5 affected RIG-I expression through its demethylation activity. Similarly, overexpression of RIG-I reversed its ALKBH5-mediated inhibition (Fig. 4i). Moreover, ALKBH5 overexpression inhibited IFNα secretion mediated by RIG-I and vice versa (Supplementary Fig. S11). To determine the effect of ALKBH5-dependent m6A regulation on RIG-I expression, we constructed a luciferase reporter gene with the 3’UTR from DDX58 mRNA, in which the two adenosine (A) bases within the m6A consensus sequences were mutated to cytosine (C) (Supplementary Materials). Luciferase assays showed that ALKBH5 depletion largely increased the activity of luciferase with WT DDX58 but not mutated DDX58 (Fig. 4j). To analyze the effect of the m6A level on DDX58 mRNA metabolism, RNA stability assays were conducted. The assays showed that forced knockdown of ALKBH5 enhanced the expression of DDX58 and prolonged the half-life of DDX58 mRNA transcripts in HNSCC cells (Fig. 4k, l). Thus, the ALKBH5-induced increase in DDX58 is at least in part due to the increased stability of DDX58 mRNA transcripts. These results suggest that ALKBH5-mediated m6A level upregulation of DDX58 mRNA suppresses RIG-I expression.

As RIG-I is a novel downstream target of ALKBH5, it was essential to explore the role of RIG-I in the protumorigenic effects of ALKBH5. Our previous study fully demonstrated that RIG-I in HNSCC acts as a tumor suppressor in vitro and in vivo [21]. As expected, the ectopic expression of RIG-I reversed the cell proliferation and increase in cell colony number mediated by ALKBH5 overexpression in HN4 and Cal27 cells (Fig. 5a, b). In addition, an EdU assay showed that RIG-I could decrease the DNA replication activity mediated by ALKBH5 (Fig. 5c). To further support the in vitro results, we subcutaneously injected Cal27 cells with or without ALKBH5 or with or without combined ALKBH5 and RIG-I expression into athymic nude mice. We found that overexpression of ALKBH5 increased the tumor size and weight, which were inhibited by RIG-I expression (Fig. 5d, e and Supplementary Fig. S12). Moreover, the increase in the Ki-67 index and the decrease in the TUNEL-positive percentage in the ALKBH5-expressing group were alleviated by combination with RIG-I expression (Fig. 5f, g, h). These results indicate that suppression of RIG-I expression is a key event in ALKBH5-mediated tumorigenesis in HNSCC.

It is widely known that the specific role of m6A “readers” in controlling the fate of methylated mRNAs is critical for m6A-dependent biological processes. To elucidate the specific m6A readers of DDX58 and determine the m6A-dependent mechanism of DDX58 regulation, ChIRP-MS was conducted using random DDX58 mRNA-specific biotin-labeled probes (Fig. 6a). Precipitated proteins were analyzed using a bubble chart, and spliceosome, ribosome, and RNA degradation were highlighted as three items involved in m6A-related mRNA metabolism (Fig. 6b and Supplementary Fig. S13). A total of 109 proteins were identified. The top 20 precipitated proteins are shown in Fig. 6c. We identified six unique peptides among seven heterogeneous nuclear ribonucleoprotein C (HNRNPC) peptides, which ranked as having the highest specificity (Fig. 6d). Furthermore, western blot analysis showed that the DDX58 probe bound notably to HNRNPC after ALKBH5 silencing and increased the m6A status (Fig. 6e). Conversely, an RNA binding protein immunoprecipitation (RIP) assay using two specific primers showed that the HNRNPC antibody bound to the 3’UTR of DDX58 mRNA after ALKBH5 inhibition (Fig. 6f). HNRNPC is associated with premRNAs in the nucleus and influences premRNA processing, transport, stability, and other aspects of mRNA metabolism. Therefore, primers designed with or without introns were used to detect premRNA and mature RNA to analyze the processing and maturation function of HNRNPC. We found that overexpression decreased the premRNA level and increased the level of mature DDX58 mRNA, while ALKBH5 silencing promoted the maturation of DDX58 mRNA (Fig. 6g, h).

To characterize the role of the binding of HNRNPC to DDX58 mRNA in RIG-I expression, we evaluated the ectopic expression of HNRNPC and found that it significantly increased the expression of RIG-I (Fig. 6i). RIG-I knockdown reversed this HNRNPC overexpression-mediated effect (Fig. 6j). In contrast, the efficient knockdown of HNRNPC expression using siRNA inhibited RIG-I expression (Fig. 6k and Supplementary Fig. S14). Similarly, RIG-I overexpression reversed the inhibition mediated by HNRNPC silencing (Fig. 6l). Moreover, we observed a positive correlation between HNRNPC and DDX58 or RIG-I in the GEPIA database and in our HNSCC TMA (Fig. 6m, n). In addition, luciferase assays showed that HNRPNC overexpression greatly increased the luciferase activity driven by the WT DDX58 3’UTR (containing m6A) but not the mutated DDX58, and vice versa (Fig. 6o, p). These results indicated that ALKBH5 silencing increased the m6A modification of DDX58 mRNA and the subsequent binding of HNRNPC, which promoted the maturation of DDX58 mRNA and enhanced RIG-I protein expression.

RIG-I recognizes viral RNA, recruits the mitochondrial antiviral signal protein (MAVS), and then activates the IKKε/TBK1/IRF3 pathway to induce type I interferon production. To explore whether ALKBH5 could also affect IFNα secretion through RIG-I regulation, the downstream signaling of RIG-I and IFNα production were detected. Interestingly, gene set enrichment analysis (GSEA) of our RNA sequencing data revealed significant enrichment of the RIG-I-like receptor pathway and interferon α/β signaling after ALKBH5 silencing (Fig. 7a). This enrichment suggests that ALKBH5 is involved in RIG-I signaling and interferon production. Increased phosphorylation of IKKε/TBK1/IRF3 after RIG-I overexpression in HNSCC cell lines was then observed (Fig. 7b). RIG-I silencing inhibited the activation of this pathway (Fig. 7c). Enzyme-linked immunosorbent assays showed increased secretion of IFNα in culture medium after RIG-I overexpression and vice versa (Supplementary Fig. S15). Moreover, we found that ectopic expression of ALKBH5 reduced IFNα secretion, while knockdown of ALKBH5 increased IFNα secretion (Fig. 7d, e). Similarly, silencing of HNRNPC expression also reduced IFNα secretion (Fig. 7f). HNRNPC overexpression reversed the inhibition of RIG-I expression and the subsequent IFNα secretion mediated by ALKBH5 overexpression (Fig. 7g). To explore the role of the IKKε/TBK1/IRF3 pathway, a specific IKKε/TBK1 inhibitor, bay-985, was used. This inhibitor reduced the upregulation of RIG-I expression and IFNα secretion after ALKBH5 knockdown (Fig. 7h).

As a well-known immunomodulatory cytokine, IFNα plays a critical role in the activation of immune cells, such as natural killer (NK) cells, T cells, and dendritic cells (DCs), and exerts an antitumor effect. First, we confirmed the antitumor effect of IFNα in HNSCC cell lines (Supplementary Fig. S16). Then, a xenograft model was established using the murine HNSCC cell line SCC7 in immunocompetent C3H/HeJ mice (Fig. 7i, Supplementary Fig. S17). IFNα reversed the tumor-promoting capacity mediated by ALKBH5 overexpression in vivo (Fig. 7j). The serum IFNα concentration was decreased after ALKBH5 overexpression (Fig. 7k). The tumor suppressive function of IFNα was demonstrated in our previous study [21, 24], and it could promote the infiltration of immune-killing cells in the tumor microenvironment (Supplementary Fig. S18). Moreover, the percentages of NK cells, T cells (CD8+ T and CD4+ T), DCs and M1 macrophages among tumor-infiltrating lymphocytes were significantly reduced after ALKBH5 silencing, and this pattern was reversed after murine IFNα injection (Fig. 7l-o and Supplementary Fig. S19). These results indicated that overexpression of ALKBH5 in HNSCC could suppress IFNα secretion through RIG-I regulation and then inhibit immune infiltration and promote tumor progression.

To investigate the clinical relevance of the ALKBH5/RIG-I/IFNα regulation axis, immunohistochemistry staining was performed in a tissue microarray with 138 HNSCC specimens (Fig. 8a). An inverse correlation between ALKBH5 expression and RIG-I and IFNα protein levels was observed (Fig. 8b, c). Moreover, there was a positive correlation between RIG-I expression and IFNα expression (Fig. 8d). These results verified the correlation in clinical samples. In summary, upregulated ALKBH5 plays an oncogenic role in HNSCC by inhibiting RIG-I-mediated IFNα secretion via the IKKε/TBK1/IRF3 pathway, which ultimately reduces immune-killing cell infiltration and promotes tumor progression (Fig. 8e).

A total of 138 pairs of tissue samples were obtained from patients with HNSCC who underwent radical resection at the Department of Oral Maxillofacial-Head and Neck Oncology, Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine. All tumor and adjacent normal tissues used in this study were collected with informed consent. This study was approved by the Ethics Committee of the Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine. The HNSCC cell lines HN4, HN6, and HN30 were provided by the University of Maryland. Cal27, SCC4, SCC25, and 293 T cells were purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). SCC7, a mouse-derived HNSCC cell line, was donated by Prof. Liu of Soochow University. Primary NOKs were primarily cultured.

Preheated RNA was spotted on a positively charged nylon membrane (RPN303B, Amersham, USA). The m6A status was probed with anti-m6A antibody (ABE572, Millipore, USA) and detected using ECL detection reagents (P36599↗, Millipore, MA, USA). The same membrane was also stained with methylene blue (319,112, Sigma, St. Louis, MO) as a control. The global m6A levels in mRNA were quantified with an EpiQuik m6A RNA Methylation Quantification Kit (Colorimetric) (Abcam) following the manufacturer’s protocol.

After deparaffinization, rehydration and antigen retrieval, slides were incubated with primary rabbit anti-human ALKBH5 (dilution 1:500; Abcam Antibody; ab195377), primary rabbit anti-human RIG-I (dilution 3.75 μg/ml; Abcam Antibody; ab238254) or rabbit anti-human IFNα (dilution 1:100; Thermo Fisher; PA5–115430) at 4 °C. We quantitatively scored the tissue slides according to the percentage of positive cells and staining intensity. The H-score (maximum score, 300) was assessed using the following formula: (3 × percentage of strongly staining) + (2 × percentage of moderately staining) + percentage of weakly staining.

Immunoblotting was performed as described in our previous study. The antibodies used in this study were as follows: ALKBH5, FTO, RIG-I and HNRNPC purchased from Abcam (Cambridge, MA, UK); MAVS, IKKε, p-IKKε, TBK1, p-TBK1, IRF3, and p-TBK1 purchased from Cell Signaling Technology (Danvers, MA, USA); GAPDH and α-tubulin purchased from Proteintech Inc. (Proteintech, Rocky Hill, NJ, USA). The TBK1/IRF3-specific inhibitor Bay-985 was purchased from Selleck (Houston, TX, USA).

Total RNA was isolated with TRIzol reagent (Invitrogen, USA), and qRT–PCR was conducted using a StepOnePlus Real-time PCR system (Applied Biosystems, Waltham, MA, USA) following the manufacturer’s instructions. The primer sequences are listed in Supplementary Table S. 1

Cells were transfected with siRNA using RNAiMAX and plasmids with Lipofectamine 3000 (Invitrogen, USA) according to the manufacturer’s instructions. Lentiviruses expressing ALKBH5 or RIG-I were constructed by Genomeditech Company (Shanghai, China).

Cell proliferation was detected using a Cell Counting Kit (CCK8; Dojindo, Kumamoto, Japan). Cells were seeded into 6-well plates at 500 cells/well for 10 ~ 14 days to analyze the clone-forming capacity. Transwell assays were performed using uncoated polycarbonate inserts (Millipore, Darmstadt, Germany) to test migration or BioCoat™ inserts (BD Biosciences, Franklin Lake, NJ, USA) to test invasion. EdU (RiboBio, Guangzhou, China) and TUNEL (Beyotime, Shanghai, China) staining were conducted according to the manufacturers’ protocols.

For apoptosis analysis, Annexin V-FITC/propidium iodide (PI) apoptosis kits (BD Biosciences) were used, and measurements were performed with a BD Beckman cytometer (BD Biosciences, Franklin Lakes, NJ, USA) and FlowJo software. For cell cycle analysis, cells were incubated with PI/RNase staining kit reagents (BD Biosciences, Franklin Lakes, NJ, USA). Anti-mouse CD3e APC and anti-mouse CD49b FITC antibodies were used for NK cell analysis. Anti-mouse CD11b FITC, anti-mouse CD11c PE-Cy7, and anti-mouse I-A/I-E APC antibodies were purchased for DC analysis. Anti-mouse CD8a PE, anti-mouse CD4 FITC, and CD3e APC antibodies were used for T cell analysis. Anti-mouse CD11b PE-Cy7, F4/80 PerCP-Cy5.5, CD80 AF647, and CD163-PE antibodies were used to analyze macrophages. All of the antibodies for immune cells were purchased from Peprotech (Rocky Hill, NJ, USA) and analyzed using a BD Fortessa cytometer (BD Biosciences, Franklin Lakes, NJ, USA).

The DNA fragments of the DDX58 3’UTR containing two m6A motifs, as well as the mutated motifs (m6A was replaced by C), were inserted into a PGL3-CMV-LUC-MCS plasmid vector (Genomeditech, Shanghai, China). The mutation sites are shown in the supplementary materials. Dual-luciferase reporter assays were performed using HEK 293 T cells (Beyotime, Shanghai, China).

For the subcutaneous implantation model, 1 × 10⁶ Cal27 cells stably transduced with lentivirus were injected into the left or right flanks of BALB/c nude mice (aged 4–6 weeks). Following stable transfection, 2 × 10⁵ SCC7 cells were subcutaneously inoculated into C3H mice (aged 6–8 weeks), which were purchased from the Shanghai Laboratory Animal Center (Shanghai, China). The width (W) and length (L) of the tumors were measured every week, and the volume was calculated using the formula V = (L × W²/2). All animal experiments were approved by the Animal Care and Use Committee of the Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine.

Established tumors were removed from mice, dissociated, and treated with 0.4 mg/mL collagenase IV, 0.4 mg/ml hyaluronidase, and 30 U/ml DNase I in RPMI 1640 medium containing 10% FBS for 1 h at 37 °C. All reagents were purchased from Sigma. Cells were washed twice, followed by filtration through a 70-μm strainer. The obtained single-cell suspensions were stained with anti-mouse CD45 PerCP-Cyanine 5.5 and Viability Dye 450 to distinguish immune cells and live/dead cells (Peprotech, Rocky Hill, NJ, USA).

MeRIP-Seq and RNA-Seq were performed by Cloudseq Biotech Inc. (Shanghai, China) according to standard procedures. Fragmented mRNA was incubated with an anti-m6A polyclonal antibody (Synaptic Systems, 202,003) in IPP buffer (10 mM Tris HCl, 150 mM NaCl, 0.1% NP40, pH 7.4), which was then immunoprecipitated by incubation with protein-A beads. The bound RNA was eluted from the beads with m6A (Berry & Associates, PR3732) and then extracted with TRIzol reagent (Thermo Fisher, Invitrogen, USA) following the manufacturer’s instructions. Purified RNA was used for RNA sequencing library generation with the NEBNext® Ultra™ RNA Library Prep kit (NEB). The input sample and the m6A IP samples were analyzed with 150 bp paired-end sequencing on an Illumina HiSeq 4000 sequencer.

First, adaptors and low-quality reads were filtered by cutadapt; then, the clean reads were aligned to the reference genome (UCSC HG19) by Hisat2 software (v2.0.4). The m6A peaks were analyzed by Model-based Analysis of ChIP-Seq (MACS) software. For the identification of consensus sequences, motifs were analyzed by HOMER. The metagenes of the m6A regions were generated with the R package MetaPlotR. Differentially methylated sites with a fold-change cutoff of ≥2 and a false discovery rate cutoff of ≤0.00001 were screened by diffReps61. The paired-end, adaptor-cleaned reads mapped to each gene were calculated by HTSeq software (v0.9.1) and normalized by edgeR software. GO and KEGG enrichment analyses were performed on differentially methylated and expressed genes.

To detect the m6A modification of target genes, the Magna MeRIP™ m6A Kit (17–10,499, Millipore, Billerica, MA) was used according to the manufacturer’s instructions. Briefly, 300 μg of total RNA was enriched with a monoclonal antibody against m6A.

The RNA of interest was immunoprecipitated with Protein A/G Magnetic Beads in 500 mL of 1x IP buffer supplemented with RNase inhibitors at 4 °C overnight. After immunoprecipitation, isolated RNA fragments were subjected to qRT–PCR. Thirty micrograms of the fragmented RNA sample was saved as the 10% input control and further analyzed.

Thirty-six biotin-labeled probes targeting DDX58 and a negative probe were directly synthesized by gene synthesis (RiboBio, Guangzhou, China). In brief, 3 × 10⁷ cells were fixed by adding 37% formaldehyde solution at a ratio of 1:36 at room temperature, followed by ultrasonic lysis. Five microliters of biotin-labeled probe was incubated in RNA protein hybridization buffer overnight at room temperature. Fifty microliters of streptavidin magnetic beads were added to immunoprecipitate the RNA-binding proteins. Eluted proteins were separated by SDS–PAGE for electrophoresis, and the bands of interest were excised for mass spectrometry analysis. The selected differentially expressed proteins were annotated with GO and KEGG analyses.

RIP analysis was performed using an RNA-binding protein immunoprecipitation kit according to the manufacturer’s instructions (Cat: 17–700, Millipore, Darmstadt, Germany). At 24 h after transfection, the cells were incubated with RIP buffer containing magnetic beads conjugated with IgG and HNRNPC antibodies overnight at 4 °C. The samples were then incubated with proteinase K to isolate immunoprecipitated RNA. The isolated RNA was analyzed by RT–PCR with primers targeting DDX58 truncations (Additional file: Table S1). 1

After transfection or treatment, cell supernatants were collected for detection with a human IFNα ELISA kit (Shycbio, Shanghai, China). Blood was collected from the hearts of mice, and the serum IFNα concentration was detected according to the manufacturer’s instructions (Shycbio, Shanghai, China).

We used unpaired Student’s t tests to compare means between groups and ANOVA for comparisons among more than three groups. All data are expressed as the mean ± SD. The survival analyses were performed using the Kaplan–Meier method to plot survival curves and the log-rank test to compare survival rates. The correlation between ALKBH5 and RIG-I and IFNα levels was analyzed by the Pearson correlation coefficient. p values <0.05 were considered to be significant. All statistical tests were two-sided.

This study was approved by the Ethics Committee of the Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine. The Laboratory Animal Care and Use Committees of the hospital approved all experimental procedures. Written informed consents were received from all patients.

Not applicable.

The authors declare that they have no competing interests.

All data supporting this study are available within this article and supplementary Information file. The MeRIP-seq and mRNA-seq data have been deposited into the Gene Expression Omnibus repository under accession number GSE185888↗. You may view the GSE185888↗ study at: https://www.ncbi↗. nlm.nih.gov/geo/query/acc.cgi?acc=GSE185888↗. The dataset used and/or analyzed during the current study are available from the corresponding author on reasonable request.