METTL3 suppresses pancreatic ductal adenocarcinoma progression through activating endogenous dsRNA-induced anti-tumor immunity

Abstract

Purpose

Although immunotherapy improves clinical outcomes in several types of malignancies, as an immunologically ‘cold’ tumor, pancreatic ductal adenocarcinoma (PDAC) is arrantly resistant to immunotherapy. However, the role of N6-methyladenosine (m⁶A) modification in the immune microenvironment of PDAC is still poorly understood.

Methods

The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) datasets were used to identify differentially expressed m⁶A related enzymes. The biological role and mechanism of METTL3 in PDAC growth and metastasis were determined in vitro and in vivo. RNA-sequencing and bioinformatics analysis were used to identify signaling pathways involved in METTL3. Western blot, m⁶A dot blot assays, co-immunoprecipitation, immunofluorescence, and flow cytometry were used to explore the molecular mechanism.

Results

Here, we demonstrate that METTL3, the key regulator of m⁶A modification, is downregulated in PDAC, and negatively correlates with PDAC malignant features. Elevated METTL3 suppresses PDAC growth and overcomes resistance to immune checkpoint blockade. Mechanistically, METTL3 promotes the accumulation of endogenous double-stranded RNA (dsRNA) through protecting m⁶A-transcripts from further Adenosine-to-inosine (A-to-I) editing. The dsRNA stress activates RIG-I-like receptors (RLRs) to enhance anti-tumor immunity, finally suppressing PDAC progression.

Conclusion

Our findings indicate that tumor cell-intrinsic m⁶A modification participates in the regulation of tumor immune landscape. Adjusting the m⁶A level may be an effective strategy to overcome the resistance to immunotherapy and increase responsiveness to immunotherapy in PDAC.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13402-023-00829-2.

Introduction

Pancreatic ductal adenocarcinoma (PDAC), the major type of pancreatic cancer, is one of the most lethal malignancies [1]. The 5-years survival rate of PDAC is less than 8% due to high probabilities of recurrence and metastasis after surgical resection, chemotherapy resistance, and non-response to immunotherapy [1–3]. As an immunologically ‘cold’ tumor, PDAC is arrantly resistant to immunotherapy on account of CD8⁺ T cell deficiency and tremendous immunosuppression in the tumor microenvironment (TME) [4–6]. One of the core goals of immunotherapy is to supply enough active effector T cells. However, the onset and persistence of T cell responses depend on innate immune responses. Thus, a deep understanding of innate immunity would provide more clues for tumor immune evasion and new treatment strategies for PDAC.

Increasing studies show that pathways initially are revealed to participate in the detection of virus have also been found to drive the recognition of growing tumor cells and elicit an anti-tumor immune response [7, 8]. dsRNA derived from RNA virus is recognized by related sensors, leading to the initiation of antiviral defense to limit the spread of virus. Whereas, recent studies have shown that antiviral response triggered by endogenous dsRNA also plays an important role in inhibiting the progression of malignant tumors [9–11]. In the cytoplasm, RLRs are responsible for the sense of dsRNA. After being triggered by dsRNA, RLRs interact with mitochondrial antiviral signaling (MAVS) to induce signaling cascades that result in the activation of antiviral response programs, such as interferons (IFNs) and chemokines production[7, 12]. Strict regulation of endogenous dsRNA is important in shaping a balanced immune response. It was well reported that A-to-I editing disrupts the secondary structure of endogenous dsRNA and thus prevents dsRNA from being recognized by innate immune sensors [13–15]. Recent study showed that m⁶A modification suppressed A-to-I editing on the same transcript [16]. However, the physiological and pathological significance of this phenomenon in PDAC has not been elucidated.m⁶A modification, methylated at the N6 position of adenosine, is catalyzed by methyltransferase complex (aka ‘writers’), and reversibly removed by demethylases (aka ‘erasers’) [17]. The methyltransferase complex is comprised of a catalytic subunit methyltransferase-like 3 (METTL3) together with methyltransferase-like 14 (METTL14) and other proteins. The demethylases include alkB homolog 5 (ALKBH5) and fat mass and obesity-associated protein (FTO). The dynamic balance between deposition and removal of m⁶A modification of transcripts is critical for normal biological processes [17]. A mass of previous results have shown dysregulation of m6A modification resulted from the abnormal expression of ‘writers’ or ‘erasers’ could affect multiple tumor progression [18–20]. Besides, m⁶A modification has been involved in adaptive and innate immune cell–mediated immunity [21]. Nevertheless, whether and how tumor cell m⁶A modifications affect the immune landscape of PDAC are largely unknown.

In the present study, we identify the key enzyme of m⁶A, METTL3 is frequently downregulated in PDAC and positively associated with the prognosis of PDAC patients. Functionally, METTL3 suppresses PDAC progression through promoting the accumulation of endogenous dsRNA and activating anti-tumor immunity. Our data suggest that METTL3 may exert as a potential tumor suppressor to involve in PDAC progression.

Materials and methods

Cell culture

Human PDAC cell lines (AsPC-1, BxPC-3, Capan-1, CFPAC-1, PANC-1, PATU8988 and SW1990), human pancreatic ductal cell line HPDE6-C7, and murine PDAC cell line FC1199 were preserved in Shanghai Cancer Institute, Shanghai Jiao Tong University, and maintained at 37 °C in a 5% carbon dioxide incubator. AsPC-1, BxPC-3, and PATU8988 cells were cultured in RPMI-1640 medium (Gibco) with 10% fetal bovine serum (FBS, Gibco). Capan-1, CFPAC-1, PANC-1, HPDE6-C7, and FC1199 cells were cultured in Dulbecco's modified eagle's medium (DMEM, Gibco) with 10% FBS, SW1990 cells were cultured in Leibovitz's L-15 medium with 10% FBS.

In vitro cell behavior assays

For the cell proliferation assay, 1000 cells were plated and counted 5 days after seeding in complete culture media, Cell Counting Kit-8 assay (SB-CCK8S, Share-Bio, China) was performed to measure cell viability according to manufacturer’s instructions. For colony formation assay, 500 cells were seeded in 2 ml complete culture media in 6-well plates. After 12 days, colonies were stained using Crystal Violet and counted. For migration and invasion assays, transwell filter champers (Millipore, Danvers, MA) with or without Matrigel (BD Biosciences) were used according to the manufacturer’s instructions. Cells migrated through the membrane were fixed, stained and counted under light microscope. For OT-1/tumor cells co-culture assay, one day prior to OT-1 co-culture, a total of 0.5 × 10⁶ FC1199-OVA⁺ cells with or without METTL3 overexpressed were plated into 12-well plates. Activated OT-1 T cells were co-cultured with tumor cells at a ratio of 1:1 T cells: tumor cells for 24 h. Then the plate was washed with PBS and stained using Crystal Violet.

Isolation and activation of CD8⁺ OT-1 cells

CD8⁺ OT-1 T cells were isolated from spleen and lymph nodes from OT-1 mice. CD8⁺ cells were enriched using magnetic beads for MACS (130–104-075, Milteny Biotec). After CD8⁺ OT-1 T cells isolation, cells were activated for 24 h using 2 μg/ml of anti-Cd28 and anti-Cd3 antibody (102116 and 100340, BioLegend). CD8⁺ T cells were cultured in RPMI-1640 medium supplemented with 10% FBS, 1% penicillin/streptomycin (Gibco), 50 μM β-Mercaptoethanol (Gibco). Cells were kept at 37 °C in a 5% carbon dioxide incubator.

Construction of plasmids and cell transfection

For METTL3 overexpression version, METTL3 was amplified from cDNAs reverse-transcribed from 293 T cells, Mettl3 was amplified from cDNAs reverse-transcribed from FC1199 cells. PCR product was cloned into the EcoR I/BamH I digested pCDH vector using the Peasy-Uni Seamless Cloning and Assembly Kit (Transgen, Beijing, China). Fast Mutagenesis Kit V2 (Vazyme, Nanjing, China) was used to generate the catalytic mutant METTL3 (aa395-398, DPPW-APPA). Lentiviral particles were generated using a three plasmids system (pPACKH1-GAG, pPACKH1-REV, and pVSV-G). The transfection of plasmids and siRNAs was performed using the Lipofectamine 3000 kit (Invitrogen) or Lipofectamine® RNAiMAX reagent (ThermoFisher Scientific, #13778030) according to the manufacturer’s instructions. The siRNA sequences are listed in Supplementary Table 3.

Generation of artificial antigen Ova-expressing cell lines

Artificial tumor antigen OVA sequence was subcloned into a lenti-EF1-MCS-P2A-Blast vector via Gibson assembly to generate Ova-expressing vector (lenti-EF1-Ova-P2A-Blast). FC1199 murine PDAC cells were transduced with Ova-expressing lentivirus for 24 h. After blasticidin selection for 7 days, OVA expression level of transduced cancer cells was identified by immunoblotting.

RT-PCR and qRT-PCR

Total RNA extracted using Trizol reagent (Takara), and reverse-transcribed using PrimeScript RT-PCR kit (Vazyme, Nanjing, China). qRT-PCR analyses were performed with SYBR Premix Ex Taq (Vazyme, Nanjing, China) on a 7500 qRT-PCR system (Applied Biosystems) according to the manufacturer’s instructions. The mRNA levels were normalized by GAPDH. Primer sequences are listed in Supplementary Table 3.

RNA-seq and data analysis

Total RNA was isolated from METTL3-expressing (METTL3) and control PANC-1 cells (Vector) using Trizol (Takara). Poly(A) RNA from 1 μg total RNA was used to generate the cDNA library according to TruseqTM RNA Sample Prep Kit protocol, which was then sequenced using an Illumina NovaSeq 6000 platform. For RNA-seq analysis, paired-end clean reads were aligned to the human reference genome (GRCh38) with Hisat2, and the aligned reads were used to quantify mRNA expression by using featureCounts. DESeq2 was employed for data normalization and differential expression analysis of RNA-seq counts. The raw RNA-seq data had been deposited in NCBI SRA (https://www.ncbi.nlm.nih.gov/sra) under the accession number PRJNA946548.

Subcellular fraction and western blotting analysis

ProtLytic nuclear and cytoplasmic protein extraction kit (NCM Biotechnology, Jiangsu, China) was used to obtain nuclear and cytoplasmic fractions of cells. Cell lysates were collected from 10 cm plates using a total protein extraction buffer (Epizyme, Shanghai, China), the same amount of total protein lysates (20 μg) was separated by 10% SDS-PAGE and transferred onto a nitrocellulose membranes (Millipore, Danvers, MA). After blocking for nonspecific binding by 5% non-fat dried milk, the membranes were incubated with primary antibody overnight at 4 °C, followed by incubation with HRP-conjugated secondary antibody. Bands were detected by a Bio-rad ChemiDoc XRS system. Antibody information is listed in Supplementary Table 4.

Immunohistochemistry

Paraffin-embedded tumor tissue sections were deparaffinized and rehydrated with graded ethanol. After antigen retrieval, endogenous peroxidase removing, and nonspecific binding blocking, the slides were incubated using the primary antibodies at 4 °C overnight and treated with HRP secondary antibody at room temperature for 1 h. The positive staining was visualized with a liquid DAB substrate (DAB Substrate Kit, Cell Signaling), and the slides were counterstained with hematoxylin. All of the sections were observed and photographed using a microscope (Carl Zeiss, Germany). The tissue microarray containing 74 cases of PDAC tumorous tissues and corresponding nontumorous tissues purchased from Shanghai Outdo Biotech Inc (OD-CT-DgPan01-006). Scoring was conducted based on the percentage of positive-staining cells: 0–5% scored 0, 6–35% scored 1, 36–70% scored 2, and more than 70% scored 3; and staining intensity: no staining scored 0, weakly staining scored 1, moderately staining scored 2 and strongly staining scored 3. The final score was calculated using the percentage score × staining intensity score as follows: “-” for a score of 0–1, “ + ” for a score of 2–3, “ +  + ” for a score of 4–6 and “ +  +  + ” for a score of >6. These scores were determined independently by two senior pathologists in a blinded manner. Antibody information is listed in Supplementary Table 4.

Immunofluorescence

For cell immunofluorescence analysis, the cells were fixed with 4% paraformaldehyde and permeabilized with phosphate-buffered saline (PBS) containing 0.1% Triton X-100. After blocking for nonspecific binding by 10% BSA (Sangon, Shanghai, China), the cells were incubated with primary antibodies at 4 °C overnight and subsequently incubated with Alexa Fluor 594- or 488-conjugated secondary antibodies for 30 min at room temperature. Nuclei were visualized using 4ʹ,6-diamidino-2-phenylindole (DAPI) staining. Digital images were acquired with confocal microscopes equipped with a digital camera (Leica, Germany). Antibody information is listed in Supplementary Table 4.

Dot blot

Total RNA extracted using an RNeasy plus Mini Kit (QIAGEN, Hilden, Germany), following the manufacturer’s protocol. The RNA samples were loaded onto Amersham Hybond-N + membrane (GE Healthcare, Chicago, IL) and crosslinked to the membrane with UV radiation. Then the membrane was blocked with 5% nonfat dry milk (in 1 × PBST) for 1 h, incubated with the specific primary antibody overnight at 4℃ followed by HRP-conjugated secondary antibody for 1 h at room temperature, and then developed with Thermo ECL SuperSignal Western Blotting Detection Reagent (Thermo Fisher Scientific, Waltham, MA). Antibody information is listed in Supplementary Table 4.

A-to-I editing ratio analysis

Two reporter plasmids were individually transfected into PDAC cells with or without METTL3 overexpression. Next, total RNA extracted using Trizol reagent (Takara), and reverse-transcribed using PrimeScript RT-PCR kit (Takara). The cDNA was used as a template to amplify transcripts containing both the SON sequence and the IRAlus, and then cloned into T vector (Promega) for Sanger sequencing. Individual clones were sequenced using Applied Biosystems BigDye terminator mix version 3.1. The percentage of A-to-I editing ratio in each transcript was calculated from the edited allele burden (I/A × 100%).

Co-immunoprecipitation (Co-IP) Assay

To detect protein–protein interactions, cells were lysed in 500 μl co-IP buffer supplemented with a cocktail of proteinase inhibitors, phosphatase inhibitors. The protein G Dynabeads (Invitrogen, USA) were pre-cleaned and incubated with the corresponding antibodies or control IgG. After incubation at 4 °C overnight, beads were washed three times with co-IP buffer. SDS sample buffer was added to the beads and the immunoprecipitates were used for western blot analysis.

Animal model studies

Animal experiments were approved by Institutional Animal Care and Use Committee of East China Normal University (Shanghai, P.R. China). Male BALB/c nude mice and C57BL/6 mice (six-eight weeks old) were maintained under SPF conditions in a controlled environment of 20–22 °C, with a 12/12 h light/dark cycle, 50–70% humidity, and food and water provided ad libitum.

For xenograft experiments, six-eight weeks old male BALB/c nude mice were used. 2 × 10⁶ PANC-1 cells were subcutaneously injected into the right flank of each mouse. Tumor volumes were measured once a week and after five weeks the mice were humanely euthanized, the weight of tumors was measured.

For the pulmonary metastasis model, 1 × 10⁶ PANC-1 cells were injected into the tail vein of nude mice to produce lung metastatic lesions. After five weeks the mice were humanely euthanized, and their lung tissues were dissected and fixed with 4% phosphate-buffered neutral formalin. Lung tissues were analyzed by H&E staining.

For the hepatic metastasis model, 1 × 10⁶ PANC-1 cells suspended in 25 μl DMEM were injected into the spleen by surgery, and injected cells were transferred to the liver through the portal veins to generate liver metastatic lesion. After five weeks the mice were humanely euthanized, and their liver tissues were dissected and fixed with 4% phosphate-buffered neutral formalin. Liver tissues were analyzed by H&E staining.

For the orthotopic model of PDAC, six-eight weeks old male C57BL/6 mice were used. 1 × 10⁶ FC1199 cells suspended in 25 μl DMEM were injected into the pancreas of each mouse by surgery. After three weeks the mice were humanely euthanized. The tumors were dissected, and lymphocyte infiltration was detected by IHC.

For C57BL/6 mouse xenograft experiments, 1 × 10⁶ vector or METTL3 FC1199 cells were subcutaneously injected into the right flank of each C57BL/6 mouse, anti-PD-L1 or control immunoglobulin (Ctrl IgG) was intraperitoneally injected at day 9, 12, 15, 18, 21 and 24 after inoculation. Tumor volumes were measured every three days and when the tumor volume reached 1000 mm³ the mice were humanely euthanized.

Bioinformatics analysis

The gene expression data and clinical information for pancreatic ductal adenocarcinoma (PDAC) were downloaded from TCGA, which were processed by Broad Institute’s TCGA workgroup. And our reproduction abided by the rules of the TCGA request. Gene sets related to innate immunity in the gene set enrichment analysis (GSEA) MSigDB resource was used to identify the difference between the METTL3 group and vector group. GSEA was performed on the Broad Institute Platform and statistical significance (false discovery rate, FDR) was set at 0.25.

Statistical analysis

Results are presented as mean ± standard deviation of the mean. Statistical analyses were performed using Prism software (GraphPad Software 8), and unpaired, two-sided Student’s t-test was used to compare two groups unless indicated otherwise. Two-way ANOVA was used to compare multiple groups in the tumoir growth curves with two independent variables. A probability of 0.05 or less was considered statistically significant.

Results

METTL3 suppresses PDAC progression in vitro and in vivo

To examine the prognostic value of the four main m⁶A modification-related genes (METTL3, METTL14, FTO, and ALKBH5) in human PDAC, we analyzed the Cancer Genome Atlas (TCGA) datasets (https://tcga-data.nci.nih.gov/tcga) and found that only METTL3 was the most significant predictor for PDAC (Fig. 1A and Supplementary Fig. 1A). METTL3 was consistently downregulated in PDAC tissues compared to that in adjacent nontumorous tissues on analyses of GSE32676 and GSE62165 (Supplementary Fig. 1B). Consistent with these results, immunohistochemical (IHC) staining of tissue arrays containing 74 paired PDAC specimens and adjacent normal pancreas tissue showed that METTL3 expression levels were significantly downregulated in the PDAC tissues (Supplementary Fig. 1C). Moreover, METTL3 protein level was also downregulated in mouse PDAC tissues derived from Kras^G12D/+; Trp53^R172H/+; Pdx1-Cre (KPC) mouse model (Supplementary Fig. 1D). These data suggest METTL3 may prevent PDAC progression. To study the function of METTL3 in PDAC progression, we conducted METTL3 or catalytic mutant METTL3 (aa395-398, DPPW-APPA) [22, 23] overexpression cells (Supplementary Fig. 1E–F) and confirmed that METTL3 overexpression increased the m⁶A level in PDAC cells, instead of METTL3 mutant overexpression (Supplementary Fig. 1G). For METTL3 knockdown, two different short hairpin RNAs (shRNAs) were designed and successfully silenced METTL3 (Supplementary Fig. 2A). In vitro studies demonstrated that elevated METTL3 significantly suppressed the proliferation, migration, and invasion of PDAC cells, but METTL3 mutant had no obvious effects on these phenotypes (Fig. 1B–D). Conversely, loss of METTL3 remarkably promoted these phenotypes (Supplementary Figs. 2C–D). Moreover, in vivo experiments showed that the tumor growth of METTL3 overexpression group but not the mutant group was significantly reduced compared with that of the vector group as revealed by tumor volume and weight (Fig. 1E and Supplementary Fig. 2E). In the lung metastasis and liver metastasis models, we found that elevated METTL3 significantly suppressed PDAC metastasis, but METTL3 mutant did not affect the metastasis of PDAC (Fig. 1F–G and Supplementary Fig. 2F–H). These data indicate that METTL3 may act as a tumor suppressor in PDAC progression, and its methylation activity is required.

Fig. 1.

METTL3 inhibits proliferation and metastasis of PDAC cells in vitro and in vivo. (A) Kaplan–Meier analysis (log-rank test) of the overall survival in PDAC patients based on METTL3 expression. Data were derived from TCGA database. (B) CCK-8 assays were performed to detect short-term proliferation of PANC-1 and AsPC-1 cells with METTL3 or METTL3 mutant overexpression. (C) Colony formation assays were performed to detect long-term proliferation of PANC-1 and AsPC-1 cells with METTL3 or METTL3 mutant overexpression. Quantification of the colony formation assay results were shown in right panel. (D) Migration and invasion assays of PANC-1 and AsPC-1 cells with METTL3 overexpression or METTL3 mutant. Representative images (left panel) and quantification (right panel) of the cell migration and invasion assay results were shown. (E) Tumor volumes were monitored every week, and tumor growth curves were generated (left panel). Five weeks later, the tumors were extracted and weighed (right panel). (F) Quantification of lung metastatic nodes. (G) Quantification of the liver metastatic nodes. Error bars indicate s.d. ns means no significant, *P < 0.05, **P < 0.01, ***P < 0.001

METTL3 triggers anti-tumor immunity in PDAC cells

To decipher the molecular mechanism by which METTL3 suppressed PDAC progression, we performed comparative transcriptomic analysis (RNA-seq) of OV-METTL3 and OV-Vector PANC-1 cells. As a result, METTL3 overexpression affected the expression level of 1769 genes (Fold change > 2.0 or Fold change < -2.0 and P < 0.05 as significant change), of which 456 genes were downregulated and 1313 genes were upregulated (Fig. 2A and Supplementary Table 1). Enrichment analyses through gene ontology biological processes (GO_BP) and KEGG pathways about upregulated genes displayed that METTL3 triggered immune defense responses (Fig. 2B). Consistently, gene set enrichment analysis (GSEA) also revealed that METTL3 was implicated in innate immune response (Fig. 2C). To further elucidate the biological functions of METTL3 in PDAC, we compared the gene expression profiles between METTL3-low group and METTL3-high group by utilizing the RNA sequencing data in the TCGA cohort. GSEA results showed that the gene sets related to immune defense responses were enriched in samples with high METTL3 expression (Fig. 2D). These clinical data were consistent with the observation in PDAC cell lines, indicating that METTL3 may activate anti-tumor immunity in PDAC.

Fig. 2.

METTL3 elicits cellular immune responses. (A) A volcano plot showed the 1768 dysregulated genes that identified by RNA-seq analysis in PANC-1 cells with METTL3 overexpression (METTL3) versus control (Vector). (Fold change > 2.0 or Fold change < -2.0 and P < 0.05 as significant change). (B) Enrichment analyses of representative GO_BP and KEGG pathways for upregulated genes. (C) Gene set enrichment analysis (GSEA) plot for METTL3 overexpression group (METTL3) compared to control group (Vector) (NES, normalized enrichment score; FDR, false discovery rate). (D) GSEA plot of high METTL3 group compared to low METTL3 group based on the expression of METTL3 in TCGA database (NES, normalized enrichment score; FDR, false discovery rate). Error bars indicate s.d. **P < 0.01, ***P < 0.001

METTL3 activates RLRs-MAVS singling pathway in PDAC

Signaling pathways that participate in the detection of virus have also played an important role in the recognition and clearance of tumor cells [7]. Viral or host nucleic acid in the cytoplasm was detected by sensors and initiates RLRs-MAVS singling to promote the production of IFNs and a mass of cytokines [12, 24]. Therefore, we detected the protein level of several key genes involved in this pathway. The results displayed that RIG-I, MDA5, IRF7, IRF9, and STAT1/2 were upregulated by METTL3 overexpression, and the phosphorylation levels of STAT1/2 were also dramatically increased after METTL3 overexpression, instead of METTL3 mutant overexpression (Fig. 3A). Nuclear and cytoplasmic fractionation analysis presented that METTL3 promoted translocation of IRF3 and P65 into the nucleus (Fig. 3B). Immunofluorescence analysis also confirmed that METTL3 promoted phosphorylation and nuclear translocation of IRF3 (Fig. 3C). Signaling downstream of RLRs-MAVS depends on the essential adaptor protein MAVS [7, 12, 24], loss of MAVS abolished the upregulation of protein expression or nuclear translocation induced by METTL3 (Fig. 3D–E). Activation of RLRs-MAVS pathway culminates in the production of type I IFNs [7], and the resultant IFNβ could upregulate expression of a large number of interferon-stimulated genes (ISGs) [25, 26]. Consistent with previous studies, METTL3 overexpression remarkably increased IFNβ level (Supplementary Fig. 3A), and almost all the ISGs were found to be highly upregulated by METTL3 in PANC-1 cells (Supplementary Fig. 3B and Supplementary Table 2). Taken together, these findings suggest that METTL3 activates RLRs-MAVS singling pathway in PDAC.

Fig. 3.

METTL3 activates RLRs-MAVS pathway. (A) Western blot detected indicated protein levels involved in RLRs-MAVS pathway with or without METTL3 overexpression. (B) Protein levels of IRF3 and P65 in the nuclear and cytoplasmic fractions of PDAC cells. (C) Immunofluorescence detected the localization of p-IRF3 in PANC-1 cells. Scale bar = 10 μm. (D) Western blot detected indicated protein levels involved in RLRs-MAVS pathway with or without MAVS knockdown in PDAC cells. (E) Protein levels of IRF3 and P65 in the nuclear and cytoplasmic fractions of PDAC cells with or without MAVS knockdown

METTL3 promotes the accumulation of endogenous dsRNA through suppresses A-to-I editing

Pioneering studies have shown that RLRs could be activated by endogenous dsRNA even without infection [9–11]. We wonder if METTL3 activated the RLRs-MAVS pathway by increasing the endogenous dsRNA level in PDAC cells. We detected dsRNA in PANC-1 cells treated with either single-stranded RNA (ssRNA)-specific RNase A or dsRNA-specific RNase III [27], and found immunofluorescence signals and dot blot signals were sensitive to RNase III but not RNase A (Supplementary Fig. 4A–B). These results demonstrate that we indeed specifically label dsRNA. Then we observed dsRNA levels were markedly increased in PDAC cells with METTL3 overexpression, instead of METTL3 mutant (Fig. 4A–C and Supplementary Fig. 4C). Conversely, METTL3 knockdown remarkably reduced m⁶A and dsRNA level (Supplementary Fig. 4D). These results suggest METTL3 increases dsRNA level, and the m⁶A catalytic activity of METTL3 is indispensable. After being activated by RNA ligands, RIG-I and MDA5 bind with downstream adaptor MAVS for signaling cascade [7, 24]. We found interactions between both RIG-I and MDA5 with MAVS were enhanced by METTL3. Moreover, treatment with RNase III but not RNase A significantly impaired the associations (Supplementary Fig. 4E). These results further verify that the increased dsRNA level is responsible for the activation of RLRs-MAVS signaling cascade by METTL3.

Fig. 4.

METTL3 increases dsRNA level in PDAC cells. (A) Immunofluorescence detected dsRNA level in PANC-1 cells with METTL3 overexpression. Scale bar = 15 μm. (B) Flow cytometry detected dsRNA level in PANC-1 and AsPC-1 cells. Data are shown as mean ± SEM. n = 3 biologically independent samples. Significance was determined using unpaired two-sided t-test, ***P < 0.001. (C) Dot blot assays detected m⁶A and dsRNA level of PANC-1 and AsPC-1 cells using total RNA with or without METTL3 overexpression. MB, methylene blue staining was used as a loading control. (D) Immunofluorescence detected inosine level in PANC-1 cells with METTL3 overexpression. Scale bar = 15 μm. (E) Inosine dot blot assays using total RNA of PANC-1 and AsPC-1 cells with METTL3 or METTL3 mutant overexpression. (F) Inosine dot blot assays using total RNA of Capan-1 and CFPAC-1 cells with or without METTL3 knockdown. (G) Top: schematic drawings show the construction of chimeric reporter plasmids. The IRAlus element of Nicn-1 gene is shown as blue arrows with the indicated orientations. Bottom: A-to-I editing ratio (I/A × 100%) of IRAlus element in the chimeric reporter plasmids with or without METTL3 overexpression in PANC-1 cells. Error bars indicate s.d. ns means no significant, **P < 0.01, ***P < 0.001

It was well reported that A-to-I editing by ADAR1 could break the double chain structure of endogenous dsRNA and prevents it from being recognized by their sensors [28–31]. However, m⁶A modification suppressed A-to-I editing on the same transcript by suppressing ADAR1 binding to the m⁶A-modified RNA [16]. Based on these clues, we speculated that m⁶A may suppress A-to-I editing to enhance the dsRNA accumulation and then activate RLRs-MAVS signaling pathway. We detect inosine level and found METTL3 but not its mutant reduced inosine level in PANC-1 cells (Fig. 4D–E). Conversely, METTL3 knockdown remarkably increased inosine level (Fig. 4F). The m⁶A modification sites were highly enriched in the stop codon area [18], and the majority of the A-to-I editing occurred on IRAlus [32]. IRAlus comprise the most abundant dsRNA in human, which were present mostly in the 3’ untranslated regions (UTR) [32]. To further confirm the direct suppression of m⁶A on A-to-I editing, we constructed two chimeric reporter plasmids that contained both m⁶A and A-to-I regions. To construct the reporter plasmid, an 84 bp sequence of SON gene, which harbors three consensus m⁶A motifs [16, 33], and a 757 bp IRAlus enriched with A-to-I editing sites in the 3’ UTR of human Nicn1 gene were cloned sequentially downstream to the EGFP sequence [34]. This reporter plasmid produces a fused RNA containing both the SON sequence for m⁶A methylation and the IRAlus for A-to-I editing, together with egfp for EGFR as the transfection control. In addition, we constructed a mutant plasmid with replaced the ‘A’ to ‘T’ in the three consensus m⁶A motifs. As expected, A-to-I editing ratios of IRAlus in Reporter-SON-Nicn1 were significantly reduced after being transfected into PANC-1 cells with METTL3 overexpression but not METTL3 mutant overexpression. However, A-to-I editing ratios of IRAlus in Reporter-SON-MUT-Nicn1 were not affected by METTL3 (Fig. 4G). These data demonstrate that METTL3 suppresses A-to-I editing on the same transcripts in an m⁶A-dependent manner.

To further verify the contribution of RLRs on the inhibitory effects of METTL3 on PDAC progression, we performed rescue assays. We found MAVS knockdown remarkably restored proliferation, migration, and invasion that were inhibited by METTL3 overexpression (Supplementary Fig. 5A–E). Collectively, these results suggest METTL3 promotes the accumulation of endogenous dsRNA, and subsequently activates dsRNA-sensing proteins. RLRs are essential mediators for the inhibitory effects of METTL3 on PDAC progression.

METTL3 increases tumor inflammation

Type I IFNs play critical roles in dendritic cell (DC)-mediated cross-priming and activation of effector T cell response against tumor [35]. In order to explore whether METTL3 affects the TME, we conducted an orthotopic pancreatic cancer mice model using the murine PDAC cell line (FC1199) [36]. As a result, murine METTL3 overexpression significantly reduced the tumor weight (Fig. 5A). The proportions of TCRβ⁺ cells, CD4⁺ T cells and CD8⁺ T cells were also significantly increased in the METTL3-overexpressed tumors, but the proportions of macrophage and myeloid-derived suppressor cells (MDSCs) were markedly reduced in the METTL3-overexpressed tumors (Fig. 5B–C and Supplementary Fig. 6A–B). The IFNγ production of cytotoxic lymphocytes was significantly increased in the METTL3-overexpressed tumors (Fig. 5C). Moreover, intratumoral CD8⁺ T cells from the METTL3-overexpressed tumors displayed markedly increased expression of PD-1 and Tim-3 (Fig. 5C and Supplementary Fig. 6C), markers typically indicative of T-cell exhaustion [37]. To evaluate the correlation between METTL3 expression and CD8⁺ T cells infiltration in the PDAC patient tumorous tissues, we further analyzed the correlation between METTL3 and CD8A expression from TCGA database. We observed a significant positive correlation between METTL3 and CD8A expression (Fig. 5D), indicating that higher METTL3 level has more CD8⁺ T cells infiltration in the PDAC patient tumorous tissues. These results directly link our observations in mouse PDAC models to the data from PDAC patients.

Fig. 5.

METTL3 increases tumor inflammation and sensitizes PDAC to anti-PD-L1 treatment. (A) Representative images (left panel) and tumor weight (right panel) of the orthotopic xenograft tumors in C57BL/6 mice. n = 4 mice per group. Data are shown as mean ± SEM. Significance was determined using unpaired two-sided t-test. (B) H&E staining and IHC staining of CD8α in tissues derived from pancreas orthotopic inoculation mice model. (C) Flow cytometry of immune populations from orthotopic xenograft tumors with or without Mettl3 overexpression. n = 3 mice per group. Data are shown as mean ± SEM. Significance was determined using unpaired two-sided t-test. (D) Correlation analysis for METTL3 and CD8A in the TCGA-PDAC database. (E) Flow cytometry detected H-2 Kb and OVA peptide (SIINFEKL) levels in the FC1199 cellular surface. n = 4 biologically independent samples. Data are shown as mean ± SEM. Significance was determined using unpaired two-sided t-test. (F) Representative images of OT-1 T cells killing FC1199-OVA⁺ cells. (G) Tumor volume over time in C57BL/6 mice with indicated treatment. n = 7 mice per group. Data are shown as mean ± SEM. Significance was determined using a two-way ANOVA test. ns means no significant. Experimental schema depicting 1 × 10⁶ FC1199 cells with or without Mettl3 overexpression were inoculated subcutaneously into C57BL/6 mice, anti-PD-L1 or control immunoglobulin (Ctrl IgG) was intraperitoneal injected at day 9, 12, 15, 18, 21 and 24 after implantation. Mice were euthanatized when the tumor volume reached 1000 mm³ in the Vector group. Error bars indicate s.d. ns means no significant, *P < 0.05, **P < 0.01, ***P < 0.001

METTL3 sensitizes PDAC to anti-PD-L1 treatment

Activation of type I IFNs signaling cascade indicates that MHC-I, which belongs to interferon-stimulated gene, might be upregulated by METTL3. FACS results showed that METTL3 significantly increased the H-2 Kb protein level in the tumor cellular surface (Fig. 5E and Supplementary Fig. 6D). These data suggest that METTL3 may enhance tumor antigen presentation and sensitize PDAC to CD8⁺ T cells treatment. Therefore, FC1199 cells expressing the model antigen OVA (FC1199-OVA⁺) were generated. Results showed METTL3 increased the OVA peptide presentation on the cellular surface (Fig. 5E), and sensitized PDAC cells to the OT-1 T cells treatment (Fig. 5F). Increased CD8⁺ T cells abundance in the TME and PD-1 level of intratumoral CD8⁺ T cells imply that immune checkpoint blockade might further enhance the inhibition effect of METTL3. In the FC1199 subcutaneous tumor model, we noticed that METTL3 significantly inhibited tumor growth, but the anti-PD-L1 antibody alone failed to inhibit tumor growth. Excitingly, the combination of METTL3 overexpression and anti-PD-L1 antibody treatment inhibited tumor growth to the greatest degree (Fig. 5G). Collectively, these data suggest that METTL3 can potentiate the anti-tumor effects of immune checkpoint inhibitors.

Discussion

More than 100 types of chemical modification have been identified in eukaryotic RNAs, m⁶A has been identified as the most prevalent internal modification in mRNA and non-coding RNAs (ncRNAs) [18, 38]. As the reversible modification in RNA, m⁶A has been implicated in almost all vital bioprocesses, including tumor progression [18–20]. As a key subunit of methyltransferase complex, METTL3 participating in tumor progression has been extensively investigated in various cancers [18]. Previous study reported that morphology and proliferation were unaffected after METTL3-knockdown in pancreatic cancer cells [39], but another study showed that cell invasiveness was significantly inhibited by METTL3 knockdown [40]. Here, we found that METTL3 overexpression significantly suppressed the proliferation, migration, and invasion of PDAC cells (Fig. 1). The main reason for the controversial role of METTL3 in PDAC might because that m⁶A influences target transcripts fate decision through altering specific recognition by multiple m⁶A-binding proteins (as known as readers), and the differences in endogenous abundance of readers from different primary samples and cell lines may lead to these conflicting results. In this study, we identify METTL3 influences PDAC progression through increasing immunogenic dsRNA level instead of regulating the fate of a particular target mRNA. Future investigation is warranted to further identify the characteristics of these dsRNA in detail.

Innate immune system provides the first line of defense against pathogen infection, which also influences the pathway involved in cancer immunosurveillance [24]. Cytosolic pattern-recognition receptors detect pathogen-derived or endogenous dsRNA through molecular structural characteristics rather than specific sequences [32]. Endogenous dsRNA generate from both mitochondrial genome and repeat elements of the nuclear genome such as short-interspersed nuclear element (SINE), long interspersed nuclear element (LINE), and endogenous retroviral element (ERV) [32]. Recent study showed that intermolecular dsRNA from mitochondrial RNAs (mtRNA) could activate PKR and inhibit translation [41]. We found METTL3 did not affect the mtRNA level in PANC-1 cells (Supplementary Fig. 7A). These results are logical because METTL3 is spatially isolated from the mitochondrial genome. METTL3 is predominantly localized in the nucleus and m⁶A is deposited on transcripts co-transcriptionally in the nucleus [42]. dsRNA derived from transcriptional activation of ERV genes has been reported to activate antiviral immunity [9–11], here we found METTL3 did not affect ERV gene expression in PANC-1 cells (Supplementary Fig. 7B). Moreover, IRAlus and LINE-1 (L1) can also generate dsRNA [32]. Interestingly, the genome-wide analysis revealed that both IRAlus and L1 were the main substrates of ADAR1 [43, 44]. Latest study showed IRAlus are the major source of DNMTi-induced immunogenic dsRNA and stimulate immune responses [45]. These findings suggest that the increased dsRNA in METTL3 overexpressed PDAC cells may be mainly derived from IRAlus and L1. However, the exact source of dsRNA increased by METTL3 remains unknown and warrants further investigations.

Recent data suggest that m⁶A extensively participates in various aspects of immunity, including immune recognition, innate immunity activation, and anti-tumor immunotherapy [21, 46–48]. According to two additional studies, m⁶A enhances the antiviral effects of type I IFN response by promoting the translation of certain ISGs and IRF3 mRNAs [49, 50]. We here found that METTL3 promotes accumulation of endogenous dsRNA by protecting m⁶A-transcripts from further A-to-I editing, thereby initiating the type I IFN response (Figs. 2 and 3). These results highlight that METTL3 activates type I IFN response through regulating secondary structural characteristics of the transcripts with m⁶A rather than the expression of the transcripts with specific sequences. In fact, the correlation between m⁶A and endogenous dsRNA is still ill-defined. A recent study shows that m⁶A modification protects against endogenous dsRNA formation during mammalian hematopoietic development [51]. But another study found that m⁶A modification negatively regulates A-to-I RNA editing to promote endogenous dsRNA formations [16]. Our data strongly support that METTL3 promotes the accumulation of endogenous dsRNA in PDAC (Fig. 4). Crosstalk between dsRNA formation and m⁶A methylation might be more complicated, especially in different tissue and microenvironments. It was well reported that A-to-I editing of endogenous dsRNA limits the sensing by RLRs, the dsRNA stress activates RIG-I-like receptors to enhance anti-tumor immunity [30, 52]. Here we found that METTL3 promotes the formation of dsRNA and enhances tumor antigen presentation and sensitize PDAC to CD8⁺ T cells treatment (Fig. 5).

Conclusion

Overall, our findings reveal that METTL3 promotes the accumulation of dsRNA by protecting m⁶A-transcripts from further A-to-I editing. The resultant dsRNA leads to a ‘viral mimicry’ state that triggers anti-tumor immunity (Fig. 6). Our results suggest that elevated METTL3 in PDAC could turn ‘cold’ tumor ‘hot’ that can be reduce the resistance to immunotherapy.

Fig. 6.

Proposed working model for inhibitory effects of METTL3 on PDAC progression. METTL3 catalyzes m⁶A deposition on transcripts co-transcriptionally in the nucleus and protects m⁶A-transcripts from further A-to-I editing, leading to accumulation of dsRNA. The resultant dsRNA triggers anti-tumor immunity via eliciting RLR-MAVS signaling cascade, finally suppresses PDAC progression

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Lili Zhu, Botai Li and Rongkun Li designed and performed experiments, analyzed data and wrote the manuscript; Yanli Zhang and Lipeng Hu performed experiments and analyzed the data, Zhigang Zhang, Shuheng Jiang and Xueli Zhang edited the manuscript; Shuheng Jiang and Xueli Zhang supervised the study, obtained funding and provided critical review. All authors approved the final version of the manuscript.

Funding

The research was supported by grants from the National Postdoctoral Program for Innovative Talents, Initiative Postdocs Supporting Program (BX2021187), Shanghai Pstdoctoral Excellence Program (2021499), China Postdoctoral Science Foundation (2021M702161), the National Natural Science Foundation of China (81902370), and the Natural Science Foundation of Shanghai (22ZR1460000, 21ZR1461300).

Data availability

Full data will be available from the corresponding author upon reasonable request.

Declarations

Ethics approval

Animal experiments were performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals and relevant Chinese laws and regulations. The protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of Shanghai Jiao Tong University, the Animal Protocol number is A2020108. Ethical approval was obtained from the Research Ethics Committee of Ren Ji Hospital, School of Medicine, Shanghai Jiao Tong University.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors have no competing interests to declare that are relevant to the content of this article.