cGAMP suppresses FTO expression to promote m⁶A modification and potentiate antitumor immunity

Abstract

Cyclic GMP-AMP (cGAMP) serves as a pivotal second messenger in the cGAS‒STING innate immune signaling pathway and plays a critical role in antiviral and antitumor immunity; however, its regulatory mechanisms governing the epitranscriptome remain to be elucidated. In this study, m⁶A methylation sequencing revealed that cGAMP induces dynamic changes in m⁶A modification in L929 and B16F10 cells. Mechanistic investigations revealed that cGAMP treatment significantly downregulated the expression of the RNA demethylase FTO. Although FTO deficiency suppresses cGAMP-induced interferon-stimulated gene (ISG) expression and IFN-β secretion, FTO may act as a critical node bridging cGAMP signaling and the interferon response. Further analysis revealed that the dynamic m⁶A regulation of ISGs, such as DDX58, is closely associated with the prognosis of tumor patients and immune cell infiltration. In vivo experiments confirmed that the targeted inhibition of FTO in combination with cGAMP exerts a synergistic antitumor effect, significantly suppressing tumor growth and prolonging survival. Overall, this study reveals the core mechanism of the "cGAMP-FTO-m⁶A" axis in innate immunity and tumor progression. These findings provide a theoretical foundation for the development of novel immunotherapies targeting the epitranscriptome.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-026-06260-8.

Introduction

Solid malignancies remain a major threat to public health. In recent years, tumor immunotherapy has emerged as a promising therapeutic strategy, as evidenced by the increasing validation of immune checkpoint inhibitors (ICIs) in both preclinical and clinical studies, offering new perspectives on cancer treatment [1–3]. However, limited clinical response rates and the prevalence of drug resistance remain significant challenges [4–6]. These issues highlight the urgent need for a deeper understanding of immune interactions within the tumor microenvironment (TME). During the process of oncogenesis, tumor cells in the TME can self-regulate and establish a favorable niche for survival and expansion by suppressing local immune responses, thereby evading immune surveillance and eradication by host immune cells [7, 8]. Therefore, it is critical to obtain deeper insight into the mechanisms of tumor immune escape to identify novel therapeutic targets and improve the efficacy of cancer immunotherapy.

As a critical component of the innate immune system, the cGAS‒STING signaling pathway plays a pivotal role in antitumor immunity [9–12]. Upon sensing cytosolic tumor-derived DNA, the sensor cyclic GMP-AMP synthase (cGAS) synthesizes cyclic GMP-AMP (cGAMP) [13]. cGAMP subsequently binds to stimulator of interferon genes (STING) located on the endoplasmic reticulum, thereby triggering downstream signaling cascades [14]. This process induces the production of type I interferons (IFNs) and proinflammatory cytokines, thereby promoting dendritic cell (DC) activation and T cell–mediated tumor elimination [9, 15]. Given their potent capacity for immune activation, STING pathway agonists have emerged as promising therapeutic agents for cancer therapy [16–18]. With a deeper understanding of the cGAS‒STING pathway, accumulating evidence suggests that the biological effects of cGAMP extend beyond those of canonical signaling pathways. Notably, cGAMP may regulate tumor cell fate and modulate tumor–immune crosstalk through more complex epigenetic mechanisms [19–21]. Therefore, elucidating the mechanistic link between cGAMP signaling and RNA epitranscriptomic remodeling may provide new insights into antitumor immunity and facilitate the development of novel therapeutic strategies for cancer immunotherapy.

N⁶-methyladenosine (m⁶A) is the most prevalent internal RNA modification in eukaryotic cells and is involved in multiple aspects of RNA metabolism, including nuclear export, alternative splicing, translation, and degradation [22]. m⁶A modification is a dynamic and reversible process that is coregulated by methyltransferase complexes (writers), demethylases (erasers), and m⁶A-binding proteins (readers). Among the numerous m⁶A regulators, fat mass and obesity-associated protein (FTO), which was identified as the first mRNA demethylase, plays a critical role in tumorigenesis and tumor progression [23–25]. FTO functions as an oncogene by promoting tumor cell expansion, invasion, metastasis, and drug resistance. Mechanistically, FTO removes m⁶A modification from the transcripts of key oncogenes and immune-related genes, resulting in increased stability, altered splicing, nuclear export, and translational efficiency [26]. Furthermore, FTO has emerged as a promising therapeutic target in cancer because of its critical role in the tumor immune microenvironment. In addition, alterations in whole-genome transcriptomic profiles under the regulation of m⁶A modification can be observed when innate immunity is stimulated. However, whether cGAMP, the central second messenger of the cGAS–STING pathway, directly regulates FTO expression to reshape the m⁶A epitranscriptome and thereby modulate antitumor immunity remains largely unknown.

In this study, we systematically investigated the effect of cGAMP on m⁶A modification via m⁶A RNA methylation sequencing (MeRIP-seq). We found that cGAMP treatment led to the downregulation of FTO expression, thereby reshaping the m⁶A methylation landscape of a subset of transcripts, including DDX58. Notably, DDX58 expression is positively correlated with immune infiltration. In addition, we found that FTO deficiency significantly enhances the antitumor efficacy of cGAMP. In conclusion, our findings reveal a novel link between innate immune signaling and epitranscriptomic regulation. Moreover, this study not only provides new insights into tumor immune evasion but also highlights the therapeutic potential of targeting FTO to increase the efficacy of existing cancer immunotherapies.

Materials and methods

Cell culture

B16F10-luciferase (B16F10-luci), B16F10, and L929 cell lines were obtained from the American Type Culture Collection (ATCC, USA). B16F10-luci and B16F10 cells were cultured in RPMI-1640 medium (Invitrogen-Gibco), while L929 cells were maintained in DMEM (Gibco). All culture media were supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific) and 1% penicillin-streptomycin (Gibco). All the cells were incubated at 37 °C in a humidified atmosphere containing 5% CO₂.

Cell treatment and cell transfection

L929 cells were pretreated with DMSO, 10 µM FB23, and 5 µM FB23-2 for 48 h. Subsequently, 2 × 10⁵ cells were seeded per well in a 24-well plate for future experiments. HT-DNA and poly(I:C) transfection were conducted in cells stimulated with 2 µg/mL DNA using Lipofectamine 2000. Additionally, perfringolysin O (PFO) was mixed with 500 nM cGAMP (Biolog) for the cGAMP stimulation assay.

Quantitative real-time PCR (qRT-PCR)

Total RNA was extracted via TRIzol reagent (Takara, Japan) following the manufacturer’s protocol. cDNA synthesis was performed via the use of M-MLV reverse transcriptase (Invitrogen). qRT‒PCR was conducted via SYBR Green PCR Master Mix (Roche) on an Agilent Technologies AriaMx Real-Time PCR System. Gapdh was used as an internal control for normalization. Relative gene expression levels were calculated via the 2^−∆∆Ct method. The primer sequences are provided in Supplementary Table S1.

Western blotting analysis

The cells were lysed in RIPA buffer supplemented with 1% PMSF (Beyotime, China) at 4 °C for 15 min. The protein concentration was quantified via a BCA protein assay kit (Beyotime). Equal amounts of protein were separated by SDS-PAGE and transferred onto PVDF membranes. After being blocked with 5% nonfat milk for 2 h at room temperature, the membranes were incubated with primary antibodies overnight at 4 °C. Following three washes with TBST, the membranes were incubated with the appropriate secondary antibodies for 2 h at room temperature. The protein bands were detected via an Odyssey fluorescence scanner (LI-COR, Lincoln, NE, USA). The primary antibodies used are listed in Supplementary Table S2.

Enzyme-linked immunosorbent assay (ELISA)

The secretion of IFN-β in the cell culture supernatants was quantified using LumiKine™ Xpress mIFN-β 2.0 ELISA kit (InvivoGen), following the manufacturer’s instructions. Briefly, cell culture media were collected at the designated time points poststimulation and centrifuged at 1,000 × g for 5 min to remove cellular debris. The luciferase activity from cell culture medium was measured using a microplate reader (Bio-Tek Instruments), and the concentration of IFN-β was calculated on the basis of the standard curve.

MeRIP-seq data processing

First, the raw reads were processed with Trim Galore and FastQC to obtain clean and high-quality reads. The cleaned reads were then aligned to the mouse reference genome (mm10) with HISAT2, using the known splice sites and exons from the Ensembl gene annotation (GRCm38) for guidance. Duplicated reads were subsequently removed via the Sambamba markdup program.

m⁶A meRIP-qPCR

The m⁶A MeRIP qPCR service was provided by CloudSeq Inc (Shanghai, China). Total RNA was extracted using Trizol reagent, followed by immunoprecipitation with the GenSeq^® m⁶A MeRIP Kit (GenSeq, Inc., Shanghai, China) in accordance with the manufacturer’s instructions. Briefly, RNA from cells was fragmented, and Protein A/G beads were conjugated to the m⁶A antibody by rotating at room temperature for 1 hour. Subsequently, the fragmented RNA was incubated with the antibody-conjugated beads and rotated at 4°C for 1 hour to capture the m⁶A-modified RNA. The beads were then washed, and the bound RNA was eluted using 5’-monophosphate sodium salt (m⁶A). The eluted RNA was purified and subsequently subjected to qPCR (real-time PCR). Primers of Ifit1 and Ifit2 were shown below:

• Ifit1: F- ATTCCAGCCACACCCGACTA, R- CCACACTTGGTGCTTTGAGG.

• Ifit2: F- ACACAGACGTTTGCTAGTGGC, R-TCCCTGGACCCTCAGGACAAT.

Identification of m⁶A peaks and differentially methylated regions (DMRs)

MACS2 (version 2.2.7.1) was used to call m⁶A peaks from the IP sequencing data of each sample with the parameters "--nomodel --extsize 200 --SPMR -q 0.001". To assess replicate consistency, Pearson correlation coefficients of m⁶A intensities across all peaks were calculated between biological replicates, and only highly consistent replicates were retained for further analysis. High-confidence peaks with a fold enrichment ≥ 2 and q-value ≤ 0.01 were used for the detection of the differentially methylated regions. Finally, bedtools intersect -v was used to compare the high-confidence peak sets between the Mock and cGAMP groups: regions present only in the Mock group and showing no overlap with any cGAMP peaks were designated "lost" m⁶A sites, whereas regions present only in the cGAMP group and showing no overlap with any Mock peaks were designated “gained” m⁶A sites; both were collectively considered differentially methylated regions (DMRs). The R package ChIPseeker was employed to annotate these peak regions to genomic features (e.g., promoters, exons, introns) to identify m⁶A-modified genes. Furthermore, genes were classified into categories (non-m⁶A, with-m⁶A, and uniquely m⁶A-modified) on the basis of the m⁶A status of their CDS regions. For visualization purposes, the bedGraph files generated by MACS2 were converted to bigWig format using the bedGraphToBigWig tool.

Gene expression quantification and differential expression analysis

BAM files from input samples were processed using StringTie for transcript assembly and quantification, generating gene-level count matrices as well as FPKM and TPM values. Differential expression analysis was performed using the DESeq2 R package (v1.30.0). Raw read counts were used as inputs, and normalization and dispersion estimation were conducted according to the default DESeq2 pipeline. Differentially expressed genes (DEGs) were identified based on an adjusted p-value (Benjamini–Hochberg correction) < 0.05 and an absolute log2 fold change > 1.

Gene function analysis of DEGs and DMR target genes

Functional enrichment analyses were performed for both DEGs and genes associated with differentially methylated regions (DMRs). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were conducted using the clusterProfiler R package. Enriched terms with an adjusted p-value < 0.05 were considered statistically significant. In addition, HALLMARK pathway enrichment analysis was carried out using gene set enrichment analysis (GSEA) implemented in the fgsea package. To visualize m⁶A modification patterns for selected genes, genomic tracks were generated using the gviz R package.

Generation of CRISPR/Cas9 knockout cell lines

To generate FTO-knockout (KO) L929 and B16F10 cell lines, site-specific single-guide RNAs (sgRNAs) targeting Fto were designed via the CRISPR Direct Web server (https://crispr.dbcls.jp). Two specific sgRNAs and one nontargeting control gRNA were selected. The target sequences are detailed in Supplementary Table 1. Lentiviral particles were produced by cotransfecting HEK293T cells with the Lenti-Cas9-puro vector and packaging plasmids. To establish stable cell lines, L929 and B16F10 cells were infected with CRISPR-Cas9-sgRNA lentiviruses and subsequently selected with puromycin. Puromycin-resistant clones were isolated, and the knockout efficiency was verified via Western blotting.

Animal experiments

All animal experiments were approved by the Animal Ethics Committee of Fujian Normal University (Approval No. IACUC-20240049). A subcutaneous tumor model was established by injecting 1 × 10⁶ cells into the right flank of 5-week-old female C57BL/6 mice. The mice were housed in a climate-controlled environment with a 12 h light/dark cycle and had free access to standard chow and water.Treatment was initiated when the tumor volume reached 60–100 mm³. In this study, the treatment method employed for mouse tumor models involved intratumoral injection. The ADV shFTO was administered via multipoint injection at a concentration of 2 × 10¹¹ PFU. Additionally, 5 µg of cGAMP and 0.5 µg of diABZI were injected every two days. For the FTO inhibitor FB23-2, the dosage was administered at 2 mg/kg, and the drug was injected every two days. The tumors were measured with a Vernier caliper, and the tumor sizes were calculated using the following formula: 3.14×length×width×height/6.

Extraction and flow cytometry analysis of tumor infiltrating lymphocytes (TILs)

Appropriate amounts of mouse tumor tissue were removed and cut into pieces. These pieces were then placed in FACS buffer (PBS + 2% FBS) containing recombinant DNase I (MCE, China) and collagenase IV (2 mg/mL, MCE, China). The mixture was cultured at 37 °C with constant rotational oscillation for 60 min. Afterward, the cells in the mixture were filtered through a 70-micron filter and treated with red blood cell lysate. These single-cell suspensions were subsequently cultured with the corresponding surface antibodies at 37 °C for 30 min. The stained cells were then analyzed using a FACSymphonTM A5 flow cytometer (BD Biosciences), and data processing was conducted using FlowJo software (BD Biosciences). The antibodies used for FACS are listed in Supplementary Table S3.

Statistical analysis

All the statistical analyses were performed using R software (version 4.1.2). Bioinformatics analyses and data visualization were conducted in R. For comparisons between two groups, a two-tailed Student’s t-test was applied, while comparisons among more than two groups were performed using one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test. Overall survival (OS) was analyzed using the Kaplan–Meier method, and differences between groups were assessed using the log-rank test. Correlations between gene expression levels were evaluated using Spearman’s rank correlation coefficient. All the data are presented as the mean ± standard error of the mean (SEM). A p-value < 0.05 was considered to indicate statistical significance, whereas p ≥ 0.05 was considered not significant (ns). Statistical significance is indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Results

cGAMP treatment induces dynamic changes in m⁶A modification

To investigate whether cGAMP regulates cellular m⁶A modification, we performed m⁶A sequencing (m⁶A-seq) analysis in L929 and B16F10 cells treated with either cGAMP or mock control. Consistent with the known distribution pattern of m⁶A modification [27–29], non-m⁶A regions were markedly more abundant than m⁶A-modified regions across all the samples. Our results revealed that cGAMP treatment induced widespread dynamic changes in m⁶A modification, with approximately 20%-30% of m⁶A peaks exhibiting altered methylation status, including both gain or lost peaks (Fig. 1A–C). These dynamic m⁶A peaks displayed canonical features of m⁶A modification, with enrichment in coding sequences (CDSs) and 3’ untranslated regions (3’ UTRs) (Fig. 1D, E). Comparative analysis further revealed distinct patterns of gain and lost m⁶A peaks between B16F10 and L929 cells (Fig. 1F), indicating cell type-specific differences in m⁶A dynamics upon cGAMP treatment. Moreover, an integrated analysis of differentially expressed genes (DEGs) and transcripts associated with differentially methylated regions (DMRs) indicated a potential link between m⁶A modification changes and gene expression regulation (Fig. 1G, H).

Fig. 1.

cGAMP treatment induces dynamic changes in m⁶A modification. (A) Number and proportion of m⁶A peaks identified in each sample. m⁶A peaks were classified as "common" (shared between conditions) or "unique" (specifically detected in either Mock or cGAMP-treated samples) based on peak overlap analysis. (B) Number and proportion of transcripts categorized according to m⁶A modification status within coding sequence (CDS) regions. Transcripts were classified as "non-m⁶A" (no m⁶A peaks in either condition), "common m⁶A" (harboring shared m⁶A peaks in both conditions), or "unique m⁶A" (harboring condition-specific m⁶A peaks), based on peak annotation using ChIPseeker. (C) Classification of transcripts based on dynamic m⁶A changes between conditions. Transcripts were grouped into "maintained", "gain", or "lost" categories according to the presence or absence of m⁶A peaks under different conditions. (D) Genomic distribution of m⁶A peaks across gene features, including 5′UTRs, coding sequences (CDS), 3′UTRs, introns, intergenic regions, and promoter–TSS regions. (E) Metagene profiles showing the distribution of m⁶A peak density along transcripts (5′UTR, CDS, and 3′UTR). (F) Venn diagrams showing the overlap of differentially methylated regions (DMRs), defined as peaks with significant changes in m⁶A enrichment (adjusted p-value < 0.05 and |log2 fold change| > 1). (G, H) Overlap analysis between genes associated with DMRs and differentially expressed genes (DEGs) in L929 (G) and B16F10 (H) cells

To validate the activation of the cGAS–STING pathway, transcriptome analysis was performed. cGAMP treatment significantly upregulated the expression of interferon-stimulated genes (ISGs) (Supplementary Fig. 1A), activated pathways related to cytosolic DNA sensing (Supplementary Fig. 1B), and increased the secretion of interferon-α and interferon-γ (Supplementary Fig. 1C, D). Interestingly, although cGAMP induced a robust innate immune response in both normal (L929) and tumor (B16F10) cells, B16F10 cells exhibited stronger induction of immune-related genes (Supplementary Fig. 1A, E, F). Collectively, these results demonstrate that cGAMP induces widespread and dynamic changes in m⁶A modification, which may contribute to the regulation of gene expression in the context of innate immune activation.

cGAMP treatment inhibits FTO expression

Dynamic m⁶A modification is orchestrated by methyltransferase complexes (writers), demethylases (erasers), and m⁶A-binding proteins (readers) [30, 31]. To investigate whether cGAMP regulates key components of m⁶A modification, we first analyzed mock- or cGAMP-treated L929 and B16F10 cells via Western blotting. The results indicated that cGAMP treatment significantly inhibited FTO expression in L929 and B16F10 cells but had no significant effects on other m⁶A-modifying enzymes (Fig. 2A, B). Notably, in B16F10 cells, cGAMP treatment also demonstrated a certain inhibitory effect on ALKBH5 (Fig. 2B). Treatment with the core messenger of the cGAS‒STING innate immune signaling pathway, cGAMP, significantly inhibited tumor growth (Fig. 2C). The Western blot results indicated that cGAMP treatment markedly reduced the expression of FTO (Fig. 2D, E) but did not affect other m⁶A-modifying enzymes (Fig. 2F-K), which aligns with the cellular results. Furthermore, qRT‒PCR analyses of tumor samples revealed that cGAMP stimulation induced the expression of interferon-stimulated genes, including Cxcl10 (Fig. 2L), Isg15 (Fig. 2M), and Ifit3 (Fig. 2N), thereby promoting antiviral and antitumor immune functions [32, 33]. Notably, cGAMP stimulation also significantly suppressed the mRNA expression of Fto (Fig. 2O) and slightly inhibited Alkbh5 (Fig. 2P) and Mettl3/14 (Fig. 2Q, R). Overall, cGAMP treatment increases global m⁶A levels, which is mediated by the downregulation of FTO expression.

Fig. 2.

cGAMP induction suppresses FTO expression. (A, B) Western blot analysis of m⁶A writers, erasers, and readers in L929 (A) and B16F10 (B) cells treated with cGAMP for the indicated times (0, 6, and 12 h). GAPDH was used as a loading control. (C) Growth curves of B16F10 melanoma tumors in mice treated with vehicle (Mock) or cGAMP (n = 5 per group). ****P < 0.0001. (D) Western blot analysis of m⁶A-related proteins in tumor tissues from the Mock and cGAMP treatment groups. (E-K) Quantification of the protein levels shown in (D). (L-R) qRT-PCR analysis of ISGs (L-N) and m⁶A regulators (O-R) in tumor tissues. Data are presented as the mean ± SEM. Statistical significance was determined using an unpaired Student’s t-test for comparisons between two groups. *P < 0.05, **P < 0.01; ns, not significant

FTO deficiency impairs the cGAS‒STING interferon response

To determine whether FTO is involved in regulating the interferon response, we generated FTO-knockout (KO) L929 and B16F10 cell lines. Upon cGAMP stimulation, the mRNA expression of Cxcl10, Isg56, Isg15, and Ifit3 in the control group significantly increased over time. However, this upregulation was significantly suppressed in L929-FTO^−/− cells (Fig. 3A–D). A similar trend was observed in B16F10-FTO^−/− cells, although the degree of suppression was lower than that in L929 cells (Fig. 3E–H). These results suggest that FTO is a critical molecule in cGAMP-mediated activation of ISG expression. Given that the transcriptional activation of ISGs typically depends on IFN-β secretion and signal transduction [34], we further investigated whether FTO participates in the regulation of upstream IFN-β secretion. Protein analysis indicated that FTO deficiency in B16F10 cells did not significantly affect the expression levels of STING, TBK1, or IRF3 (Supplementary Fig. 2A). Similarly, the bioinformatics analysis produced comparable findings (Supplementary Fig. 2B). In L929 cells, treatment with the FTO inhibitors FB23 or FB23-2 significantly reduced IFN-β secretion under stimulation with HT-DNA or cGAMP (Fig. 3J–K). In B16F10 cells, following treatment with FB23-2, there was no significant trend toward downregulation of IFN-β induced by HT-DNA and cGAMP (Supplementary Fig. 2C). This observation may explain why the downregulation of ISGs induced by cGAMP according to the qRT‒PCR results is more pronounced in L929-FTO^−/− cells. Interestingly, the regulatory pattern of FTO differed under poly(I:C) stimulation (Fig. 3L), potentially reflecting the distinct signaling pathways activated by different stimuli. Collectively, these findings reveal that FTO serves as a key regulatory node in cGAMP-induced innate immune responses, modulating the cellular immune state in a cell type-specific manner.

Fig. 3.

FTO deficiency inhibits the cGAS‒STING pathway. (A-H) qRT‒PCR analysis of Cxcl10, Isg56, Isg15, and Ifit3 mRNA levels in mock-treated vs. FTO-knockout (FTO^−/−) L929 (A-D) and B16F10 (E-H) cells stimulated with cGAMP for the indicated times. (I) qRT‒PCR detection of Ddx58 expression in B16F10 and B16F10-FTO^−/− stimulated with cGAMP for the indicated times. (J-L) ELISA analysis of IFN-β secretion in L929 cells treated with vehicle (Mock) or FTO inhibitors (FB23, FB23-2) followed by stimulation with HT-DNA (J), cGAMP (K), or poly(I:C) (L). The data are presented as the mean ± SEM. Statistical significance was determined using an unpaired Student’s t-test (for comparisons between two groups) or one-way ANOVA followed by Tukey’s post hoc test (for comparisons among more than two groups). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant

cGAMP participates in the transcriptional regulation of ISGs accompanied by dynamic m⁶A changes

Interferon-stimulated genes (ISGs) serve as core effector molecules in innate and antitumor immunity [3, 35]. On the basis of the enrichment analysis of DEGs and DMR modified transcripts, we hypothesized that cGAMP-induced immune activation may be associated with dynamic changes in m⁶A modification. Our results showed that cGAMP treatment broadly induced the expression of ISGs (Fig. 4A). Notably, a subset of ISGs, including Ifit2, Ifit1, and Ddx58, exhibited concomitant increases in m⁶A modification levels, a pattern consistently observed in both the L929 and B16F10 cell lines (Fig. 4B). Methylation peak analysis indicated that cGAMP stimulated B16F10 cells, resulting in a significant increase in the m⁶A methylation peak of Ddx58 (Fig. 4C), Ifit1 (Supplementary Fig. 3A) and Ifit2 (Supplementary Fig. 3B). m⁶A meRIP‒qPCR revealed that, as indicated by m⁶A sequencing, cGAMP stimulation significantly increased the m⁶A levels of Ifit1 (Supplementary Fig. 3C) and Ifit2 (Supplementary Fig. 3D). However, owing to the high degree of sequence duplication in the m⁶A modification site region of Ddx58, m⁶A meRIP‒qPCR verification is not feasible. Furthermore, the results of qRT‒PCR demonstrated that FTO deficiency significantly increased the expression of Ddx58 in response to cGAMP stimulation (Fig. 3I). Western blot analysis also revealed that FTO deficiency led to a notable increase in DDX58 expression (Supplementary Fig. 2D, E).

Fig. 4.

m⁶A-mediated regulation of ISGs is correlated with prognosis and immune infiltration in melanoma. (A) Heatmap showing the mRNA expression levels of ISGs in mock- and cGAMP-treated cells. The color scale represents the log2(TPM + 1). (B) Heatmap showing the dynamic changes in m⁶A modification (gain, loss or maintenance) of ISGs following cGAMP treatment. (C) Genomic tracks showing that the Ddx58 m⁶A modification level significantly increased after cGAMP treatment. The yellow box indicates the newly generated m⁶A modification. (D) Box plot showing the differential expression of DDX58 between SKCM tissues and healthy tissues. (E) Kaplan‒Meier curves indicating that high DDX58 expression is associated with better overall survival in SKCM patients. (F) Forest plot of Cox regression results displaying hazard ratios of various ISGs; HRs < 1 indicate protective factors. (G-K) Correlation analysis showing that high DDX58 expression is associated with increased immune infiltration levels (e.g., CD4⁺ T cells, NK cells) in the tumor microenvironment. Spearman correlation coefficients and two‑sided P‑values are labeled

Afterward, we focused on the clinical relevance of these findings. Analysis of public clinical datasets revealed that DDX58 expression was significantly lower in skin cutaneous melanoma (SKCM) tissues than in normal tissues (Fig. 4D). Moreover, higher DDX58 expression was significantly associated with better overall survival (OS) (Fig. 4E, F). Further immune infiltration analysis suggested that DDX58 expression was positively correlated with the infiltration of immune cells, particularly activated natural killer (NK) cells, suggesting a potential role for DDX58 in shaping the tumor immune microenvironment (Fig. 4G–K). In summary, these results suggest that cGAMP-induced transcriptional activation of ISGs is accompanied by dynamic changes in m⁶A modification, and highlight DDX58 as a potential mediator of antitumor immune responses.

FTO drives tumor growth

Previous studies have suggested that FTO participates in antitumor immune surveillance via the IFN-β-ISG axis. However, in tumor cells, FTO may also directly promote tumor progression by regulating the m⁶A modification of proliferation-related genes [20, 21, 36]. To clarify the role of FTO in tumorigenesis, we first evaluated the effect of the FTO inhibitor FB23-2 on B16F10 cell proliferation in vitro. Colony formation assays revealed that FB23-2 inhibited the colony-forming ability of B16F10 cells in a concentration-dependent manner, whereas the METTL3 inhibitor STM-2457 had no significant inhibitory effect (Fig. 5A, B). Moreover, B16F10 cell viability was significantly reduced when the FB23-2 concentration was ≥ 5 µM, further confirming its inhibitory effect on tumor cell proliferation (Fig. 5C). Consistent with these findings, genetic knockout of FTO (FTO^−/−) also significantly suppressed the colony formation ability of B16F10 cells (Fig. 5D, E).

Fig. 5.

FTO promotes melanoma tumor growth in vitro and in vivo. (A) Representative images of colony formation assays in B16F10 cells treated with the indicated concentrations of FB23-2 (FTO inhibitor) or STM-2457 (METTL3 inhibitor). (B) Quantification of colony numbers from (A). *P < 0.05, **P < 0.01; ns, not significant. (C) Viability of B16F10 cells treated with increasing concentrations of FB23-2. (D) Representative images of colony formation assays in WT and FTO-knockout (FTO^−/−) B16F10 cells. (E) Quantification of colony numbers from (D). **P < 0.01. (F) Tumor growth curves of mice inoculated with B16F10 cells and treated with Mock (vehicle) or FB23-2. ****P < 0.0001. (G) Kaplan‒Meier survival curves of tumor-bearing mice treated with Mock or FB23-2. **P < 0.01. (H) Tumor growth curves of mice inoculated with mock (control), FTO-knockout (FTO^−/−), or ADV-shFTO (FTO knockdown) B16F10 cells. Notably, both genetic knockout and adenoviral interference with FTO expression suppressed tumor growth. **P < 0.01, ***P < 0.001. (I) Kaplan‒Meier survival curves of the mice in the indicated groups. *P < 0.05, **P < 0.01.The data are presented as the mean ± SEM. Statistical significance was determined using an unpaired Student’s t-test (for two groups), one-way ANOVA followed by Tukey’s post hoc test (for more than two groups), or the log-rank test (for Kaplan‒Meier survival analysis). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant

To verify the protumorigenic function of FTO in vivo, we established a mouse tumor model and treated the mice with FB23-2. The results demonstrated that FB23-2 effectively delayed tumor growth (Fig. 5F) and prolonged survival (Fig. 5G) by targeting FTO activity. Furthermore, we investigated the effects of genetic FTO depletion in FTO-knockout (KO) cells and adenovirus-mediated FTO knockdown (ADV-shFTO) cells in a subcutaneous tumor model. The results revealed that tumor volume was significantly lower in both the FTO-KO group and the ADV-shFTO group than in the mock control group (Fig. 5H). Consistently, both Fto deficiency and interference significantly prolonged the survival of tumor-bearing mice (Fig. 5I). Collectively, these in vitro and in vivo data support the conclusion that FTO plays a critical role in promoting tumor progression.

FTO deficiency promotes the antitumor effect of cGAMP

Previous studies have indicated that FTO plays a protumorigenic role in tumor progression. To evaluate the antitumor efficacy of combining targeted FTO inhibition with cGAMP in vivo, we established a B16F10 tumor model. The results revealed that the tumor volumes in both the ADV-shFTO group and the ADV-shFTO + cGAMP combination group were significantly smaller than those in the mock group (Fig. 6A). Notably, the combination treatment significantly prolonged survival, demonstrating a synergistic effect between FTO knockdown and cGAMP in inhibiting tumor growth (Fig. 6B). Mechanistic studies revealed that FTO modulation activated the cGAS‒STING pathway and triggered the IFN-β-ISG signaling axis, thereby enhancing the immunogenicity of tumor cells and promoting their recognition and killing by immune cells. Notably, in B16F10 cells, although FTO inhibition only slightly affected the activation intensity of the IFN-β pathway, the cGAS‒STING signal was still effectively activated. These findings suggest that interventions targeting FTO or cGAMP alone are insufficient to fully unleash the antitumor effect, whereas their combination can simultaneously act on complementary regulatory levels, thereby inhibiting tumor progression more effectively.

Fig. 6.

FTO plays dual roles in both tumor cells and CD8⁺ T cells to regulate melanoma progression and the immune response. (A) Tumor growth curves of mice inoculated with B16F10 melanoma cells expressing mock or ADV-shFTO and treated with or without cGAMP. Notably, the combination of FTO knockdown and cGAMP treatment resulted in the most potent suppression of tumor growth. ***P < 0.001, ****P < 0.0001. (B) Kaplan‒Meier survival curves of tumor-bearing mice from the indicated groups in (A). **P < 0.01. (C) Growth curves of control (FTO^fl/fl) and CD8-specific FTO knockout (FTO^fl/flCD8^cre) mice inoculated with B16F10-Luci cells and treated with NaCl or cGAMP. Notably, FTO deficiency in CD8⁺ T cells significantly enhances the antitumor effect of cGAMP. *P < 0.05, ****P < 0.0001. (D) Kaplan‒Meier survival curves of tumor-bearing mice from the indicated groups in (C). *P < 0.05. (E) Growth curves of FTO^fl/fl and FTO^fl/flCD8^cre mice inoculated with B16F10-Luci cells and treated with NaCl or the STING agonist diABZI. Notably, the loss of Fto in CD8⁺ T cells further potentiated the therapeutic response to diABZI. ***P < 0.001, ****P < 0.0001. (F) Kaplan‒Meier survival curves of the mice in the indicated groups in (E). *P < 0.05, **P < 0.01. The data are presented as the means ± SEM. Statistical significance for tumor growth curves was determined using one-way ANOVA followed by Tukey’s post hoc test. Overall survival was analyzed using the Kaplan‒Meier method and the log-rank test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

CD8⁺ T cells are the primary effector cells involved in antitumor immunity. Treatment with cGAMP significantly increased the proportion of activated CD4⁺ and CD8⁺ T cells among tumor-infiltrating lymphocytes (Supplementary Fig. 4A–E), with a notably greater proportion of activated CD8⁺ T cells than activated CD4⁺ T cells. To investigate whether FTO is involved in regulating their function, we established tumor models in mice with a conditional knockout of FTO in CD8⁺ T cells. The results showed that FTO deficiency in CD8⁺ T cells further enhanced the antitumor effects of cGAMP (Fig. 6C, D) and the STING agonist diABZI (Fig. 6E, F). These findings suggest that FTO in CD8⁺ T cells functions as an "inhibitory" molecule of the immune agonist response, and its deficiency helps release the immunosuppressive state, thereby potentiating the antitumor immune effects induced by agonists.

Discussion

Although the induction of cGAS–STING by cGAMP and its role in antitumor therapy have been extensively studied, the understanding of whether cGAMP can inhibit tumors through alternative mechanisms is limited. In this study, we demonstrated that cGAMP regulates m⁶A modification by downregulating the expression of FTO, which in turn inhibits the production of IFN-β induced by either HT-DNA or cGAMP. Notably, FTO deficiency suppresses tumor growth and further potentiates the antitumor efficacy of cGAMP and diABZI. Our findings position FTO as a pivotal molecular node connecting cGAS–STING signaling with gene expression regulation, providing a new framework for understanding epigenetic regulation within innate immunity.

First, we demonstrated that cGAMP globally regulates m⁶A modification in a cell type-specific manner. This specificity manifests not only in the significant differences in m⁶A maintenance, loss, and gain patterns between cell lines but also in the differential remodeling of the m⁶A peak distribution across mRNA transcripts. These findings suggest that cGAMP-mediated regulation of the epitranscriptome is not uniform but rather highly context dependent, and relies on the intrinsic epigenetic state and signaling network of the cell. Mechanistically, we found that cGAMP treatment significantly downregulated the expression of the m⁶A demethylase FTO. This downregulation occurs concomitantly with the robust upregulation of interferon-stimulated genes (ISGs) [36]. L929 cells exhibited a significant inhibition of IFN-β induced by HT-DNA and cGAMP following treatment with FB23-2, whereas no notable changes were observed in B16F10 cells. These findings suggest that the regulation of the cGAS‒STING pathway by FTO may exhibit distinct patterns across different cell types. However, FTO deficiency typically inhibits ISGs, indicating that FTO functions upstream of IFN-β secretion to amplify innate immune signals via the IFN-β-ISG axis.

In the context of tumor biology, our study reveals a complex and paradoxical dual role for FTO. On the one hand, within tumor cells, FTO has distinct oncogenic functions [37–39]. We demonstrated that inhibiting or knocking out FTO effectively suppressed colony formation and in vivo tumorigenicity, whereas FTO expression promoted tumor growth. These findings align with those of previous studies suggesting that FTO regulates proliferation-related genes via demethylation. On the other hand, within CD8⁺ T cells in the tumor microenvironment, FTO acts as an "immune brake" [40]. Conditional knockout of FTO in CD8⁺ T cells significantly increased the antitumor efficacy of STING agonists, indicating that FTO dampens the intensity of antitumor immune responses in effector cells. This cell type-dependent function elegantly explains why the combination of FTO targeting and cGAMP yields synergistic antitumor effects: This strategy achieves a "two-pronged attack" by directly suppressing autonomous tumor growth while indirectly relieving the suppression of CD8⁺ T-cell function, thereby remodeling the immune microenvironment. Importantly, this cell type-dependent dual role elegantly reconciles an apparent paradox observed in our study: while FTO deficiency impaired cGAMP-induced IFN-β production in vitro, the combination of FTO inhibition and cGAMP robustly enhanced antitumor efficacy in vivo. We postulate that in the complex context of the TME, the systemic benefits of FTO inhibition—namely, directly halting tumor cell proliferation and alleviating CD8⁺ T cell suppression—far outweigh the partial reduction in initial IFN-β secretion. Furthermore, cGAMP administration ensures that cGAS‒STING signaling remains effectively activated even when the FTO-dependent amplification loop is inhibited. Therefore, the synergistic efficacy of this combination does not rely on maximizing peak IFN-β levels, but rather on reshaping a microenvironment where intrinsic tumor growth is crippled and immune effector cells are fully sensitized to STING-mediated activation.

Despite the promising synergistic effects observed in our preclinical models, translating the combination of FTO inhibition and STING agonists into the clinic necessitates careful consideration of several translational challenges. Given that FTO is ubiquitously expressed and regulates fundamental physiological processes, systemic FTO inhibition poses a risk of on-target/off-tumor toxicity. Furthermore, nonselective dual therapy could inadvertently impair the function of other essential immune cell populations within the complex tumor microenvironment or cause unintended collateral damage to normal healthy tissues, thereby raising valid concerns regarding long-term safety. Therefore, the successful clinical application of this combinatorial strategy will likely depend on the development of targeted codelivery systems—such as advanced tumor- or immune-cell-directed nanocarriers—to precisely localize therapeutics, thereby maximizing local efficacy while minimizing systemic adverse effects [41, 42]. Additionally, establishing robust predictive biomarkers will be crucial for identifying patients most likely to benefit from this combination therapy.

In conclusion, this study systematically elucidates the central role of the cGAMP-FTO-m⁶A regulatory axis in innate immune responses and tumor progression. We discovered that cGAMP remodels the m⁶A methylome in a cell type-specific manner and activates the IFN-β-ISG signaling pathway by downregulating the demethylase FTO, thereby augmenting the antitumor immune response. Crucially, this study reveals the dual function of FTO in the tumor microenvironment: it promotes proliferation in tumor cells while suppressing immune activity in CD8⁺ T cells. Notably, in the B16F10 model, although FTO inhibition only marginally affected IFN-β signaling, cGAS‒STING signaling remained effectively activated; consequently, the combined application of both agents exerted a synergistic antitumor effect by simultaneously targeting tumor cell-autonomous proliferation and regulating the immune microenvironment. These findings not only provide a novel mechanism for understanding the crosstalk between innate immunity and epigenetic regulation but also lay a solid theoretical foundation for developing combinatorial immunotherapy strategies based on FTO inhibitors.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: [Junzhong Lai], [Yajuan Fu], [Xiaoyu Yang]; experiments performed and data analysis: [Feng Pan], [Guo Dong], [Zhihua Feng]; bioinformatics analysis: [Xiaolan Chen]; writing of the manuscript: [Feng Pan], [Xiaolan Chen], [Junzhong Lai]; edited and revised manuscript: [Junzhong Lai ], [Xiaoyu Yang]; funding acquisition: [Xiaoyu Yang]; supervision: [Yajuan Fu].

Funding

The present study were supported by the Joint Funds for the Innovation of Science and Technology, Fujian Province (Grant number 2025Y9597) and Special Subsidy Fund for the Traditional Chinese Medicine of Fuzhou Health Commission, Fujian, China (Grant No. 2025-ZYY-W001).

Data availability

Data and materials are available upon reasonable request if applicable.

Declarations

Ethics approval

All animal experiments were approved by the Animal Ethics Committee of Fujian Normal University (Approval No. IACUC-20240049).

Consent to participate

Informed consent was obtained from all the individual participants included in the study.

Competing interests

The authors have no relevant financial or nonfinancial interests to disclose.