METTL3 Overexpression Promotes ACC in Mice
Previous studies have indicated that METTL3 promotes cell proliferation and invasion of epithelial cancer cells in the pancreas22 and that METTL3 is required for maintaining β-cell function by regulating insulin secretion-related genes,23 which play a critical role in neonatal β-cell development and functional maturation.24 However, the relationship between METTL3-mediated m⁶A, exocrine pancreatic function, and ACC carcinogenesis and pathogenesis remains unclear.
In this study, we investigated the relationship between ACC and RNA methylation using a previously reported ACC mouse model.21 To overexpress Mettl3, we generated transgenic mice with the CAG promoter, which drives systemic expression via a cytomegalovirus enhancer fused to the chicken β-actin promoter and the poly(A) signal of the rabbit β-globin gene (M3 mice). These mice did not develop visible tumors within 22 weeks after birth. In contrast, mice with an elastase 1 promoter-driven luciferase-fused SV40 TAg in the pancreas (EL mice), a genetically engineered model of rapidly progressing ACC, developed apparent tumors by 22 weeks of age.21
SV40, a DNA virus discovered in 1959 in poliovirus vaccines cultured from monkey kidney cells,25 produces the TAg protein, which inhibits the tumor suppressors TP53 and RB, disrupting their tumor-suppressing pathways.26 Since EL mice are considered an ACC model,21 we sought to determine whether the RNA methylation process contributes to tumor formation in this context. To this end, we crossed M3 mice with EL mice, generating double-transgenic mice (WTg; M3/EL1) (Fig. 1a). Interestingly, these mice, which overexpress both Mettl3 and TAg, developed a highly aggressive pancreatic carcinoma phenotype and died within 25 weeks (Fig. 1b). Representative histological images are shown in Supplementary Fig. S1a. DAB-stained components were extracted by color deconvolution and binarized for quantitative comparison between WTg and EL mice (Supplementary Fig. S1b). Quantitative analysis revealed that METTL3 expression was significantly higher in the WTg, and Ki-67 levels also tended to be elevated in WTg mice. These findings indicate that tumor growth is markedly enhanced by METTL3.
Fig. 1
METTL3 promotes the development of pancreatic acinar cell carcinoma (ACC) in mice. a Schematic representation of the experimental mouse model. PEL1-TAg mice were crossed with PCAG-M3 mice to generate WTg mice. b Kaplan‒Meier survival curves comparing WTg and EL mice. The x-axis represents weeks after birth, and the y-axis indicates the overall survival rate. WTg mice exhibited a significantly worse prognosis than EL mice (log-rank test, p = 0.018). c Schematic representation of the Mettl3 deletion construct (Mettl3 del), which lacks the S-adenosyl-L-methionine (SAM)-binding domain, compared to the full-length Mettl3 mRNA and protein structure. d Bioluminescence imaging of tumors before and after conditional Mettl3 deletion. The left panels (blue outline) show mice with established tumors prior to treatment. The right panels (orange outline) display the same mice following tamoxifen administration (2 mg/30 g body weight for 5 days) to induce Cre-mediated Mettl3 deletion. e Experimental timeline for apoptosis detection. Following tamoxifen administration and conditional deletion of Mettl3, residual tumors were visualized using VivoGlo™ Luciferin. Forty-eight hours later, imaging with VivoGlo™ Caspase-3/7 Substrate revealed caspase-3/7-dependent apoptosis. f Experimental timeline for pharmacological inhibition. Mice were treated with the specific METTL3 inhibitor STM2457. Tumor burden was confirmed using VivoGlo™ Luciferin, followed 48 hours later by VivoGlo™ Caspase-3/7 substrate imaging to detect apoptosis. g Representative pathological specimens from the Kitasato University cohort of ACC patients. Each specimen shows ACC tissue and adjacent normal pancreatic tissue stained with hematoxylin and eosin (HE), anti-METTL3, anti-WTAP, and anti-Ki-67 antibodies. The lower right inset demonstrates the binarization process used for quantification. Scale bars represent 200 µm in ACC-1 and ACC-3, and 100 µm in ACC-2. h Quantification of positive staining for METTL3, WTAP, and Ki-67 in ACC versus normal pancreatic tissue. Positive pixel counts were compared as % area. Data are presented as the mean ± SD. Individual data points represent replicates (n = 6–9 per group). Statistical significance was determined by Student’s t-test (*p < 0.05, **p < 0.001)
SAM-binding METTL3 domain is indispensable for ACC formation
METTL3 participates in various biological processes, including mRNA translation, through interaction with the eukaryotic translation initiation factor (eIF) family.27 To investigate whether the SAM-binding domains of METTL3, which serve as methyl group donors, play a role in pancreatic tumor formation, we generated SAM-binding domain-deficient mutant mice (M3del) lacking amino acids 1129–1650, thereby deleting three SAM-binding domains (Fig. 1c). When crossed with EL mice, M3del/EL mice showed no visible tumor formation during the 22-week observation period, whereas EL mice displayed rapid and aggressive tumor development. These results demonstrate the critical role of the SAM-binding domains of METTL3 in pancreatic cancer development. Notably, METTL3 also interacts with eIF3h through its N-terminal region (amino acids 1–200),28 which remains intact in the M3del mutant. Thus, the observed phenotype primarily reflects loss of the catalytic activity of METTL3 rather than disruption of its protein‒protein interactions with the translation machinery. Collectively, these findings establish that WTg mice represent an ideal model for investigating the pathogenesis and potential therapeutic targets of RNA methylation in ACC.
Conditional deletion of transgenic METTL3 induces apoptosis in the ACC in mice
To determine whether METTL3 contributes to pancreatic cancer maintenance and assess its potential as a therapeutic target, we generated mice with tamoxifen-inducible Pdx1-Cre recombinase-driven deletion of transgenic Mettl3 (TAg/Mettl3) (Supplementary Fig. S1c). Before tamoxifen administration, bioluminescence imaging with luciferin revealed intra-abdominal pancreatic tumors that had partially invaded surrounding tissues and metastasized to the liver (Fig. 1d). In addition, liver metastasis was confirmed in mice prior to tamoxifen administration (Supplementary Fig. S1d). Conditional deletion of Mettl3 after tumor formation was achieved by administering tamoxifen (2 mg/30 g mouse for 5 days) to activate estrogen-responsive Cre recombinase in Pdx1-expressing pancreatic cells. Following tamoxifen administration, genotyping confirmed successful deletion of the tgM3 allele in vivo (Supplementary Fig. S1e). Tumor luminescence, as measured by VivoGlo™ Luciferin, significantly decreased or disappeared (Fig. 1d, Supplementary Fig. S1f). These results establish that conditional deletion of overexpressed transgenic Mettl3 disrupts the maintenance of ACC.
METTL3 inhibitor administration induces tumor apoptosis
A recent study reported that the potent METTL3 inhibitor STM2457 effectively treats hematopoietic malignancies.29 Having demonstrated that METTL3 represents a promising druggable target in ACC, we investigated whether STM2457 could induce apoptosis in a preclinical animal model with tamoxifen-inducible Pdx1-Cre recombinase-driven deletion of endogenous Mettl3 (TAg/endMettl3: PdxErt2/LSLendM3/EL1) (Supplementary Fig. S1g). For this purpose, we synthesized STM2457 according to a published protocol29 (Supplementary Fig. S1h).
After deleting endogenous METTL3 by administering tamoxifen, subsequent bioluminescence imaging using the caspase-3/7-activated prodrug luciferin demonstrated the induction of apoptosis specifically at tumor sites (Fig. 1e). As anticipated, STM2457 administration also resulted in rapid induction of caspase-3/7-dependent apoptosis in pancreatic tumors in vivo (Fig. 1f). We conducted imaging using the previously described protocol,30 first administering VivoGlo™ Luciferin to detect luminescence from SV40 tumor-incorporated luciferase, followed 48 hours later by VivoGlo™ Caspase-3/7 Substrate to detect apoptosis-related luminescence.
These findings provide strong support for targeting METTL3 as a potential therapeutic strategy for ACC in mice.
METTL3 expression and mutational landscape in human ACC
To determine whether METTL3 plays a central role in human ACC, we analyzed clinical specimens obtained through two university hospitals. We utilized samples from both institutions to ensure the robustness and reproducibility of our findings rather than to establish large-scale validation cohorts. Consistent with the rarity of ACC, only a small number of paraffin-embedded tumor blocks were available from over 10 years of records at each institution. These samples were stained for hematoxylin and eosin (HE), METTL3, WTAP, Ki-67, and METTL14 (Osaka University cohort only). In both cohorts, ACC tissues displayed higher positivity rates for METTL3, WTAP, and Ki-67 than normal pancreatic or duodenal tissue (Fig. 1g, h, Supplementary Fig. S2a, b). Since WTAP and METTL14 function cooperatively with METTL3 in the m⁶A methyltransferase complex, these findings suggest that m⁶A modification is important for tumorigenesis in ACC. Additionally, genomic analysis confirmed that KRAS hotspot mutations were absent in both human ACC clinical specimens and WTg mouse tumors (Supplementary Fig. S3a, b).
METTL3 regulates cell proliferation-related genes in ACC
To investigate the molecular mechanisms by which METTL3 influences ACC development and progression, we performed MeRIP-seq on tumor specimens obtained from each distinct mouse model system (Supplementary Fig. S4a; sample information is provided in Tables S1, S2). Our subsequent analyses of the resulting data showed that targeted Mettl3 depletion following tamoxifen administration resulted in attenuation of methylation peaks specifically within Mettl3-depleted mouse tumor samples. Depletion of Mettl3 was confirmed by the overall reduction in peak numbers (Supplementary Fig. S4b, c) and the loss of methylation at the 3’UTR of the Myc gene, a known methylation site (Supplementary Fig. S4d). Peak files underwent quality control (Supplementary Fig. S5a, b), and two replicates from each group were selected and integrated using IDR (irreproducible discovery rate) for subsequent analyses (Supplementary Fig. S5c). Motif analysis of peak files from TAg/Mettl3 mouse tumors identified DRACH as the enriched motif (p value = 1e-14; Supplementary Fig. S6a). Furthermore, comparison of peaks in mouse tumors before and after Mettl3 depletion revealed that the majority of peaks disappeared (Supplementary Fig. S6b).
Through systematic analysis of genes exhibiting methylation peaks across all experimental groups—including Kras/Mettl3, TAg/Mettl3-depleted, TAg/Mettl3, and TAg mice—we identified significant methylation enrichment in genes functionally associated with critical cellular pathways, particularly the insulin signaling pathway and pancreatic cancer-related molecular networks (Supplementary Fig. S7a, b). To evaluate the effect of Mettl3 overexpression on genes, detailed comparative analysis of methylation peak distributions between TAg/Mettl3 mice, which were Mettl3-overexpressing mice, and the other experimental groups (Kras/Mettl3, TAg/Mettl3-depleted, and TAg mice) revealed a striking pattern: genes critically involved in the cell cycle, nucleotide excision repair, and MAPK signaling pathway were preferentially methylated as a direct consequence of METTL3 overexpression (Fig. 2a, Supplementary Fig. S7b). This finding suggests a targeted regulatory mechanism whereby METTL3 selectively modifies specific gene sets essential for cellular proliferation.
Fig. 2
MeRIP-seq and RNA-seq analyses reveal the m⁶A landscape in pancreatic acinar cell carcinoma. a KEGG pathway enrichment analysis of genes associated with TAg/Mettl3-specific m⁶A peaks. Significant pathways were identified using a threshold of p < 0.01. b Differential gene expression analysis between TAg/Mettl3- and TAg/Mettl3-depleted mouse tumors. GO enrichment analysis was performed using the KEGG pathway and Gene Ontology Biological Process (GOBP) databases
Differential gene expression analysis was performed between mouse tumors in each group. We explicitly included the Kras/Mettl3 group in this analysis as a comparative reference to represent the canonical PDAC pathway. Although our data and public datasets confirm that KRAS mutations are rare in ACC, this comparison serves to contextualize the distinct molecular mechanisms of METTL3-driven ACC against the well-established KRAS-driven tumorigenesis in PDAC, rather than implying a driver role for KRAS in ACC. DNA replication and homologous recombination signaling pathways were activated in the Kras/Mettl3 group compared to the TAg group and similarly upregulated in the TAg/Mettl3 group relative to the Kras/Mettl3 group. Additionally, metabolic gene sets, including PPAR signaling, were enriched in the TAg/Mettl3 group compared to the TAg group (Supplementary Fig. S8–10). Notably, in the comparison between the TAg/Mettl3 and TAg/Mettl3-depleted groups, which directly reflects METTL3 function, gene sets involved in cell cycle regulation and the DNA damage response were activated in the TAg/Mettl3 group (Fig. 2b). This result is consistent with the MeRIP-seq peak analysis, which showed enrichment of genes related to cell proliferation upon Mettl3 overexpression.
Among the differentially expressed genes (DEGs) extracted from RNA-seq analysis, we focused our investigation on genes that are methylated by METTL3 that showed loss of methylation peaks in TAg/Mettl3-depleted conditions. Supplementary Fig. S11 shows the RNA-seq results as a volcano plot of the top 20 most significantly changed genes. METTL3-regulated genes are marked in colored bold text, with many upregulated genes showing METTL3-mediated regulation. Gene Ontology (GO) analysis of differentially expressed genes exhibiting peaks in the 3’ UTR, which are presumed to be regulated by METTL3, revealed consistent upregulation of cell cycle-related gene networks specifically in TAg/Mettl3 mice. Additionally, gene set enrichment analysis (GSEA) provided confirmation of significant enrichment of gene signatures associated with enhanced cell proliferation, DNA synthesis, and mitotic progression (Supplementary Fig. S12–18). These analytical approaches provide compelling evidence for a coordinated regulatory program.
Collectively, these experimental findings strongly suggest that METTL3 functions as a critical epigenetic regulator that actively promotes cell cycle progression and enhances cellular proliferation through targeted methylation of key cell proliferation-related genes in ACC, thereby contributing to tumor growth and progression through well-defined molecular mechanisms.
scRNA-seq analysis reveals enhanced malignancy in METTL3-overexpressing ACC
While MeRIP-seq enables transcriptome-wide identification of METTL3-dependent m⁶A modifications, it is inherently blind to cellular heterogeneity and cannot resolve how epitranscriptomic programs are distributed across tumor and stromal compartments. In contrast, scRNA-seq deconvolutes complex tissues into discrete cellular states, allowing regulatory programs to be mapped to specific tumor and tumor microenvironment (TME) populations—an essential approach for dissecting intercellular communication.4 To translate METTL3-driven RNA methylation landscapes into cell-type-resolved signaling networks and mechanistically dissect tumor microenvironment crosstalk, we conducted scRNA-seq analysis on tumor tissue samples obtained from both WTg (M3/EL1) and EL1 transgenic mice. Sample quality control was performed by filtering cells based on mitochondrial gene expression, RNA count, and gene features, followed by data integration using SCTransform normalization to reduce batch effects (Supplementary Fig. S19a). Subsequent cell clustering analysis was performed according to Seurat vignettes, and manual cell type annotation was conducted based on established marker gene expression patterns derived from previously published reports in the field31 (Supplementary Fig. S19b, S20a, b). We confirmed the identity of ACC cell clusters through examination of characteristic gene expression signatures that were highly consistent with previously established molecular profiles reported in the literature (Fig. 3a) (GSE48643).32
Fig. 3
Single-cell RNA sequencing analysis of pancreatic acinar cell carcinoma (ACC) in Mettl3-overexpressing mice. a Feature plots showing the expression of ACC identity markers (Cyp3a13, Muc1, and Apob), validated using the GSE48643 dataset, within the identified ACC clusters. Color intensity represents log-normalized expression levels. b Single-sample GSEA of ACC clusters. Specific enrichment scores for malignancy-related pathways, including cell division (MITOTIC_SPINDLE, E2F_TARGETS), epithelial-to-mesenchymal transition, and transforming growth factor-β signaling. Statistical significance was determined using the Wilcoxon rank-sum test (p < 0.01). c Circle plot illustrating putative ligand‒receptor interactions between cell types identified by CellChat analysis. Edge width is proportional to the communication strength between cell populations. d Chord diagram depicting specific ligand‒receptor interaction pairs and their relative contribution weights among ACC cells, endothelial cells, fibroblasts, and macrophages
Following the precise isolation and identification of ACC cell clusters from the complex cellular mixture, we performed comparative gene expression analysis between WTg and EL samples using single-sample gene set enrichment analysis (ssGSEA) to identify differentially regulated biological pathways (Supplementary Fig. S21). Our analytical results demonstrated significantly higher enrichment scores in WTg ACC clusters for gene sets specifically associated with various aspects of cancer malignancy and aggressive tumor behavior, including accelerated cell division processes, epithelial-to-mesenchymal transition (EMT) programs, and transforming growth factor-β (TGF-β) signaling cascades (Fig. 3b, Supplementary Fig. S20c). These findings corroborate the previously identified effect of Mettl3 overexpression on enhanced cell proliferation that we had initially discovered through our bulk RNA-seq analysis approach (Fig. 2), offering additional evidence that this proliferative phenotype is present specifically within the isolated ACC cell population rather than being a general tissue-level effect.
Furthermore, analysis of intercellular communication networks between ACC cells and other diverse cellular components within the tumor microenvironment revealed extensive and complex signaling interactions with multiple cell types, including fibroblasts, endothelial cells, and infiltrating macrophages (Fig. 3c, d, Supplementary Fig. S22–25). Collectively, these results suggest that Mettl3 overexpression promotes aggressive tumor progression through multiple complementary mechanisms, not only through direct intracellular effects on tumor cell behavior and gene expression but also by significantly enhancing paracrine and autocrine interactions with critical components of the surrounding tumor microenvironment, particularly fibroblasts and endothelial cells, which play crucial roles in supporting tumor growth, invasion, and metastasis.
Reclustering analysis of ACC and TME reveals the relationship between ACC and iCAFs via PRSS1 signaling
To investigate the complex interactions within the tumor microenvironment, we systematically extracted distinct cell populations comprising fibroblasts, endothelial cells, and macrophages from our dataset and subsequently performed reclustering analysis to achieve higher resolution cellular characterization (Supplementary Fig. S26a). Following cell annotation procedures that incorporated multiple marker genes and scoring algorithms (Supplementary Fig. S26b), we successfully identified and selected several critical tumor microenvironment constituent clusters. These specifically included tumor-associated macrophages (TAMs), inflammatory cancer-associated fibroblasts (iCAFs), myofibroblastic cancer-associated fibroblasts (myCAFs), and tumor endothelial cells (TECs). The identification and characterization of these cell types were accomplished using established scoring approaches that have been validated in previous tumor microenvironment studies (Supplementary Fig. S27, 28).
Subsequently, we conducted an evaluation of intercellular communication patterns and signaling networks between these tumor microenvironment constituent clusters and ACC to understand the complex crosstalk mechanisms that drive tumor progression (Supplementary Fig. S29, 30). Our analysis revealed significantly enhanced and preferential signaling from ACC cells to iCAFs in WTg mice compared to EL mice, indicating that genetic modifications in the WTg model specifically promote this particular communication pathway (Fig. 4a, Supplementary Fig. S31). This finding suggests that enhanced signaling represents a key mechanism through which ACC cells manipulate their surrounding microenvironment to create a more favorable niche for tumor growth and progression. Through ligand‒receptor analysis and pathway mapping, we identified that serine protease 1 (PRSS1) orthologs, specifically trypsin 10 (TRY10) and GM10334, serve as the primary mediators of signaling communication from ACC cells to iCAF populations via interaction with the coagulation factor II thrombin receptor (F2R) (Fig. 4b, Supplementary Fig. S34, 35). This discovery represents a novel mechanistic insight into how ACC cells communicate with their surrounding stromal components. The expression levels of these critical genes were markedly and consistently elevated in WTg mice compared to EL control mice, suggesting that their regulation occurs through RNA methylation mechanisms that are enhanced in the WTg model (Supplementary Figs. S34, 35).
Fig. 4
Reclustering analysis of pancreatic acinar cell carcinoma (ACC) and tumor microenvironment (TME) components. a Chord diagrams depicting ligand‒receptor signaling originating from ACC cells to other TME clusters, stratified by WTg (left) and EL (right) conditions. b Bubble plots showing significant ligand‒receptor interactions originating from the ACC. The analysis compares pathways upregulated in WTg (left) versus EL (right). Dot size represents statistical significance (p-value), and color intensity indicates the communication probability
To gain a more complete understanding of the bidirectional communication networks, we performed complementary signal analysis, examining the communication patterns from TME constituent cells back to ACC cells. This reciprocal analysis demonstrated that IGF1 and HAS1 signaling pathways originating from iCAFs were specifically and significantly activated in WTg mice compared to controls, indicating the establishment of a positive feedback loop between ACC cells and iCAFs (Supplementary Fig. S36). This bidirectional communication suggests the formation of a self-reinforcing signaling circuit that promotes sustained tumor growth. Further detailed investigation of the downstream molecular targets and biological consequences of the 10 most highly expressed ligands released from these TME components revealed important mechanistic insights. Specifically, we discovered that IGF1 signaling regulates a network of genes that are critically involved in cell cycle progression and proliferation, with particular emphasis on the regulation of PLK1, a key cell cycle kinase that controls mitotic progression and genomic stability (Supplementary Fig. S37). This finding provides a direct molecular link between TME-derived signals and the enhanced proliferative capacity of ACC cells.
Collectively, these findings suggest that METTL3 overexpression creates a complex signaling cascade wherein it enhances PRSS1 secretion from ACC cells, which subsequently acts on surrounding TME components, with particular specificity for iCAFs (Supplementary Figs. S38, 39). This interaction results in the activation of reciprocal tumor-promoting signals such as IGF1, ultimately leading to accelerated tumor growth through the establishment of a self-reinforcing communication network between cancer cells and their stromal microenvironment. This mechanism represents a novel therapeutic target for disrupting the supportive tumor microenvironment and potentially inhibiting tumor progression.
ACCs with high PRSS1 expression exhibit stemness characteristics
To examine the complex differentiation dynamics among ACC clusters and understand their developmental relationships, we performed RNA velocity analysis using scVelo methodology. This analysis systematically identified cluster 3 as the primary origin of differentiation trajectories within ACC clusters, suggesting that it represents a more primitive cellular state (Fig. 5a). The RNA velocity vectors consistently pointed from cluster 3 toward other clusters, indicating directional differentiation flow and establishing cluster 3 as the root of the developmental hierarchy. Previous research has established that a higher PRSS1/SPINK1 ratio correlates with a more undifferentiated cellular phenotype and enhanced stemness characteristics in acinar cells.33 Our quantitative analysis confirmed that cluster 3 exhibited the highest Prss1/Spink1 ratio among all identified clusters, with values significantly exceeding those observed in more differentiated clusters (Fig. 5b). This elevated ratio provides molecular evidence supporting the primitive nature of cluster 3 cells and their enhanced capacity for dedifferentiation.
Fig. 5
Pancreatic acinar cell cluster analysis reveals distinct molecular signatures and developmental trajectories. a UMAP feature plots displaying the gene expression levels of Prss1 and Spink1 and the Prss1/Spink1 ratio. The bottom-right panel visualizes RNA velocity streamlines, indicating the predicted developmental trajectory of acinar cells. b Violin plots displaying the distribution of the Prss1/Spink1 ratio across different acinar cell clusters. Points represent individual cells. c Heatmap illustrating the proportional changes in cell composition within pancreatic acinar cell carcinoma (ACC) clusters between WTg and EL conditions. Color intensity represents the relative abundance of each cluster. d Gene Ontology (GO) and pathway enrichment analysis of genes upregulated in Cluster 3 compared to other ACC clusters. The bar chart shows significant pathways related to pancreatic development, extracellular matrix organization, and metabolic processes. The x-axis represents the statistical significance of enrichment (-log10 p value; Fisher’s exact test)
Comparative analysis of cluster distribution between WTg and EL mice revealed distinct patterns of cellular composition across different experimental conditions. Notably, only cluster 3 showed a significantly higher proportion of cells in WTg mice compared to other clusters, with approximately 2-fold enrichment relative to control conditions (Fig. 5c). This selective enrichment suggests that Mettl3 overexpression specifically promotes the expansion or maintenance of cells with stemness characteristics. To further characterize the functional properties of cluster 3, we conducted GO analysis of upregulated genes in cluster 3 compared to other clusters. This analysis demonstrated significant enrichment of genes involved in acinar cell developmental processes, including pancreatic acinar cell differentiation, zymogen granule organization, and digestive enzyme biosynthesis pathways (Fig. 5d). The enriched developmental gene signatures provide additional molecular evidence supporting the primitive, stem-like nature of cluster 3 cells.
These results suggest that Mettl3 overexpression is mechanistically linked to elevated Prss1 expression through epigenetic regulation of developmental transcription programs. This molecular cascade contributes to cellular dedifferentiation and enhanced stemness characteristics by modulating key developmental pathways in acinar cells, ultimately promoting a more primitive cellular phenotype that may have implications for pancreatic regeneration and disease progression.
PRSS1 promotes proliferation and therapy resistance in ACC via fibroblast interactions
To investigate the functional consequences of PRSS1 expression in ACC, we established a stable PRSS1 knockdown cell line by overexpressing Luc2-GFP in the mouse ACC cell line 266-6 using a lentiviral vector system (Supplementary Fig. S40a–c). To substantiate the stemness characteristics suggested by our in silico analyses with functional evidence (Fig. 5), we performed a sphere formation assay using three-dimensional culture on low-attachment plates. Consistent with our RNA velocity and GO analyses, the results demonstrated that PRSS1 knockdown significantly reduced sphere-forming capacity, confirming the strong association between PRSS1 expression and stemness-related phenotypes (Supplementary Fig. S40d). In conventional two-dimensional monolayer cultures, PRSS1 knockdown resulted in significantly reduced cell proliferation rates compared to control cells (Fig. 6a, Supplementary Fig. S40e). To better recapitulate the tumor microenvironment, we employed three-dimensional coculture systems with mouse fibroblast 3T3 cells. Quantitative analysis of spheroid dimensions revealed that PRSS1 knockdown significantly reduced spheroid size, as measured by the largest cross-sectional area, reflecting total spheroid volume (Fig. 6b). These observations demonstrate that PRSS1 promotes cancer cell proliferation in both 2D and 3D culture conditions. We confirmed consistent Luc2-GFP expression across both culture systems (Supplementary Fig. S41a).
Fig. 6
PRSS1 knockdown attenuates pancreatic acinar cell carcinoma (ACC) progression and sensitizes cells to therapy. a Cell proliferation assay. 266-6 cells transduced with control (ctrl) or Prss1-targeting shRNA (shPrss1) vectors were seeded at the indicated densities. Absolute optical density (OD) values reflecting cell numbers are shown. Data are presented as the mean ± SD (n = 4). Statistical significance was determined by Student’s t-test (*p < 0.05, **p < 0.01). b Spheroid formation assay in 3D coculture (266-6 and 3T3 cells). Representative brightfield images (right) and quantification of spheroid surface area (left) are shown. Data are presented as box-and-whisker plots (±SD). *p < 0.01. Scale bars represent 100 µm. c Drug sensitivity assays assessed by bioluminescence. Cocultures were treated with cisplatin (CDDP), 5-fluorouracil (5-FU), STM2457, or 5-FU + STM2457 at concentrations corresponding to the effective dose (ED) of 10, 50, 75, and 90. For the combination treatment, STM2457 was maintained at a constant concentration of 10 μM. Data represent the mean ± SD (n = 6). *p < 0.05, **p < 0.01 (Student’s t test). d Synergistic effect analysis. Left panels: Dose‒response curves for 5-FU monotherapy and 5-FU + STM2457 combination in 266-6 ctrl cells. Right panel: Combination index (CI) values calculated using the Chou-Talalay method. CI < 1 indicates a synergistic effect. e, f In vivo therapeutic efficacy in a subcutaneous tumor model. 266-6 cells were inoculated into NOD-SCID mice, and treatment commenced on day 5 when tumors reached approximately 50 mm³. Mice were treated with vehicle, STM2457, FOLFIRINOX, or FOLFIRINOX + STM2457. Tumor growth curves represent the mean ± SD (n = 8 mice per group). Statistical significance was determined by one-way ANOVA followed by Bonferroni’s multiple comparisons test (*p < 0.05, **p < 0.01). g Clonogenic survival assay demonstrating radiosensitivity. Survival fractions of Ctrl and shPrss1 cells, with or without STM2457 (10 μM), were fitted to a linear-quadratic model. Data represent the mean ± SD (n = 3). Statistical significance was determined by ANOVA using the R package “CFAssay” (*p < 0.01, **p < 0.001)
When evaluating chemotherapeutic responses, we observed that PRSS1 knockdown had no significant effect on drug sensitivity in 266-6 cells cultured alone (2D), as evidenced by comparable dose‒response curves for cisplatin (CDDP), 5-fluorouracil (5-FU) and the selective METTL3 inhibitor STM2457 (Supplementary Fig. S41b). However, in 3D cocultures with 3T3 fibroblasts, PRSS1 knockdown significantly enhanced sensitivity to those chemotherapeutic agents (Fig. 6c). This observation suggests that PRSS1-mediated interactions between ACC cells and stromal fibroblasts play a critical role in modulating drug sensitivity and therapeutic resistance. Additionally, combination treatment with 5-FU, a standard therapeutic agent for ACC, and STM2457 demonstrated efficacy in both control and PRSS1 knockdown cells. Combination index (CI) analysis34 revealed synergistic effects (CI < 1) at low doses in shPrss1 cells and at therapeutic doses (>ED25) in control cells (Fig. 6d, Supplementary Fig. S42a). These findings suggest that adding STM2457 to 5-FU-containing regimens represents a promising therapeutic option for ACC.
To validate this combination strategy in vivo, we established a subcutaneous tumor model using 266-6 cells. When tumors reached approximately 50 mm³ on day 5 postinoculation, mice received FOLFIRINOX (day 5), STM2457 (days 5-7), or combination treatment, while control animals received vehicle (DMSO, PBS, and 5% dextrose). All treatment groups exhibited significantly reduced tumor volumes compared to controls, with the combination group showing the strongest tumor suppression effect, although differences between monotherapy and combination therapy did not reach statistical significance (Fig. 6e, f, Supplementary Fig. S42b). Importantly, no treatment group displayed significant body weight changes compared to controls, suggesting minimal off-target toxicity (Supplementary Fig. S42c). While the difference between the combination and monotherapies did not reach statistical significance, the observation that the combination group exhibited the strongest tumor suppression without increased toxicity provides in vivo proof-of-concept for our mechanistic model. These findings suggest that STM2457 may biologically enhance the response to FOLFIRINOX by disrupting TME-mediated resistance, pointing to a potential combinatorial benefit that warrants further investigation rather than definitive therapeutic superiority in this specific experimental setting.
To investigate the PRSS1-F2R-IGF1 axis, we first attempted PRSS1 overexpression but found that endogenous PRSS1 expression in 266-6 cells was already high, resulting in only minimal exogenous PRSS1 expression (approximately 6%), which precluded functional overexpression studies (Supplementary Fig. S42d). IGF1 measurements in culture supernatants from 266-6, 266-6 shPrss1, 3T3, and coculture conditions revealed that in 2D monocultures, 266-6 cells secreted higher levels of IGF1 than 3T3 cells. Interestingly, 266-6 shPrss1 cells exhibited elevated IGF1 secretion compared to wild-type 266-6 cells, consistent with reports that PRSS1 degrades IGF1.35 In 3D cocultures, 266-6 shPrss1 also showed higher IGF1 secretion, which was markedly suppressed by STM2457 treatment. Furthermore, F2R inhibitor (Vorapaxar) treatment significantly reduced IGF1 secretion from 3T3 cells, confirming that F2R activation mediates IGF1 production by fibroblasts, as demonstrated in Fig. 4 (Supplementary Fig. S43a). To determine the functional impact of IGF1 on cell proliferation within this coculture model, we neutralized secreted IGF1 using a specific antibody. This blockade significantly reduced the spheroid surface area (Supplementary Fig. S43b). Notably, treatment with STM2457 resulted in an even greater suppression of spheroid growth than IGF1 neutralization alone, suggesting that METTL3 regulates tumor progression through multifaceted pathways beyond the IGF1 axis. Given that PRSS1 is well recognized as a secreted protease,36 these results collectively indicate that METTL3-mediated stabilization of Prss1 mRNA leads to enhanced protein secretion. This, in turn, drives the intercellular signaling cascade that activates stromal F2R, demonstrating a direct functional link between intracellular RNA methylation and extracellular tumor-stromal communication.
To examine the role of METTL3 in regulating PRSS1, we performed mRNA stability assays. The half-life of IGF1 mRNA was reduced in 266-6 shMETTL3 cells compared to wild-type 266-6 cells, supporting our scRNA-seq data (Supplementary Fig. S36, 37), indicating that methylation of the Prss1 3’UTR stabilizes mRNA and enhances expression (Supplementary Fig. S43c). Notably, PRSS1 expression was paradoxically higher in 266-6 shMETTL3 cells than in wild-type cells, showing an inverse correlation (Supplementary Fig. S44a). Interestingly, analysis of the ACC cluster from scRNA-seq data also revealed a subpopulation of cells with an inverse correlation between PRSS1 and METTL3 expression (Supplementary Fig. S44b), highlighting the difficulty of recapitulating in vivo tumor heterogeneity in vitro. Additionally, it remains unclear whether 3T3 cells are fully induced into iCAFs in our 3D coculture system, underscoring the inherent limitations of in vitro models in fully replicating the complex tumor microenvironment.
Additionally, we assessed the impact of PRSS1 expression on radiosensitivity in 266-6 cells. Our results demonstrated that PRSS1 downregulation significantly increased radiosensitivity (Fig. 6g). The estimated α/β ratio derived from radiobiological modeling was higher in wild-type cells than in PRSS1 knockdown cells, consistent with the observed differences in radiosensitivity and proliferative capacity. Moreover, cotreatment with STM2457 produced synergistic effects, enhancing radiosensitivity in 266-6 cells regardless of PRSS1 expression status. This radiosensitization likely reflects STM2457-mediated disruption of cell cycle regulation and DNA repair mechanisms controlled by METTL3, as demonstrated in Figs. 2 and 3.
Collectively, these experimental findings establish that PRSS1 functions as a multifaceted promoter of malignant cellular behaviors, specifically promoting enhanced cell proliferation rates and conferring significant radioresistance in ACC cells. Furthermore, our data demonstrate that PRSS1 contributes substantially to chemoresistance mechanisms, particularly through specific molecular interactions with stromal fibroblasts in the tumor microenvironment (Fig. 7). These results provide important insights into the complex role of PRSS1 in therapeutic resistance and suggest that METTL3 is a potential target for improving treatment efficacy in ACC patients.
Fig. 7
Schematic representation of the proposed mechanism. METTL3-mediated m⁶A modification stabilizes Prss1 mRNA, promoting PRSS1 secretion. This activates cancer-associated fibroblasts (iCAFs) to secrete IGF1, establishing a tumor-stromal interaction loop that enhances ACC proliferation, stemness, and therapy resistance
