How Multilayered RNA Splicing Networks Drive Pancreatic Cancer Plasticity and Resistance

Pancreatic ductal adenocarcinoma ranks among the most lethal human malignancies, presenting an intractable challenge in oncology where five-year survival rates remain between five and ten percent across unselected cohorts. Because the disease typically progresses without noticeable early symptoms and lacks reliable non-invasive biomarkers for early detection, more than eighty percent of patients present with locally advanced, unresectable, or metastatic tumors, dropping five-year survival below two percent. Even for the small subset of fifteen to twenty percent of patients eligible for primary surgical resection, post-operative disease recurrence reaches approximately eighty percent, driven by a lethal combination of intrinsic chemoresistance, early systemic dissemination, and phenotypic plasticity. Genomically, tumor initiation follows a stereotypical sequence originating within pancreatic intraepithelial neoplasia, spearheaded by near-universal activating KRAS mutations occurring in over ninety percent of cases, followed by subsequent inactivation of key tumor suppressor genes including TP53, CDKN2A, and SMAD4. Although targeted therapies against driver mutations have historically faced failure due to biochemical constraints and compensatory feedback loops, the therapeutic landscape saw a milestone in August 2026 when the FDA approved the multi-selective RAS(ON) inhibitor daraxonrasib for metastatic disease following systemic therapy. Nevertheless, because durable benefit remains restricted to patient subsets and adaptive resistance inevitably emerges, attention has shifted toward the dynamic post-transcriptional mechanisms that dictate how oncogenic programs are executed at the RNA level.

Alternative splicing has emerged as a central engine of transcriptomic plasticity in pancreatic ductal adenocarcinoma, dynamically expanding proteomic diversity from single genes to facilitate tumor adaptation under intense microenvironmental stress. Splicing dysregulation directly supports hallmark oncogenic programs including sustained proliferation, metabolic reprogramming, epithelial-to-mesenchymal transition, immune evasion, and drug resistance. Crucially, whereas hematological cancers frequently acquire recurrent heterozygous hotspot mutations within core spliceosomal genes such as SF3B1, SRSF2, and U2AF1 to alter branch point and splice-site recognition, solid tumors like pancreatic adenocarcinoma rarely harbor recurrent spliceosomal mutations. Instead, pancreatic cancer cells disrupt splicing fidelity through quantitative alterations in the expression, activity, and stoichiometric balance of non-mutated trans-acting splicing regulators. This stoichiometric reprogramming begins exceptionally early during neoplastic transformation, producing a widespread molecular field effect detectable in histologically normal pancreatic tissues adjacent to tumors prior to visible morphological transformation or desmoplastic stroma formation. Specifically, early downregulations and altered ratios of essential serine/arginine-rich factors, including SRSF1, SRSF2, SRSF5, and SRSF6, impair the recognition of weak exonic splicing signals, thereby shifting homeostatic control toward repressive RNA-binding proteins that preferentially enforce oncogenic isoform expression.

To systematically evaluate how RNA processing intersects with pancreatic tumor biology, investigators have established a three-tiered conceptual framework that organizes regulatory mechanisms by their functional level and empirical support. Level one mechanisms directly govern splice-site selection on pre-mRNA transcripts, encompassing the local abundance of spliceosomal components, RNA-binding protein networks, and RNA-level structural modifications or adenosine-to-inosine editing events within pre-mRNA substrates. Level two mechanisms act indirectly by controlling the transcription, nuclear localization, or post-translational modification of splicing factors themselves, which is the predominant mode through which oncogenic transcription factors, upstream signaling cascades, and global chromatin states operate. Level three comprises autonomous post-transcriptional processes that intersect with, but operate independently from, alternative splice-site selection, such as adenosine deaminase acting on RNA-mediated editing of repetitive elements to modulate innate immune sensing, non-coding RNA stability networks, and competing endogenous RNA circuits. Classifying regulatory events across these functional tiers is essential because direct splice-site modulation requires demonstrating acute shifts in isoform output upon direct perturbation in relevant cellular contexts, whereas indirect regulatory cascades require tracing upstream oncogenic inputs to specific splicing factor alterations before claiming direct splicing governance.

Among direct level one mechanisms with robust evidence in pancreatic models, RNA-binding protein imbalances drive essential phenotypic transitions, including metabolic adaptation to the dense, nutrient-deprived, and hypoxic microenvironment. In pancreatic tumors, overexpression of the repressive ribonucleoprotein PTBP1 promotes intronic binding flanking exon nine of the pyruvate kinase pre-mRNA, repressing its inclusion and favoring mutual exclusion in favor of exon ten. This event yields the PKM2 isoform rather than PKM1; because PKM2 predominantly forms a low-activity dimer, glycolytic intermediates are diverted toward the pentose phosphate and macromolecular biosynthetic pathways to maintain redox homeostasis and cellular survival. Similarly, the level one splicing regulator RBM10 acts as a critical tumor suppressor whose loss, focal deletion, or somatic mutation correlates with advanced clinical stage, perineural invasion, and aggressive disease. RBM10 directly determines splice-site choice on transcripts such as NUMB, governing Notch signaling inhibition, and enforces the exclusion of exons seven and eight in hTERT to suppress telomerase activity. Beyond the major spliceosome, selective minor spliceosome dependency represents another direct mechanism in cancer, where processing bottlenecks in rare U12-type introns within essential signaling genes like MAPK and RAF may influence cell proliferation in pancreatic malignancy.

At the intersection of transcriptional networks and post-transcriptional processing, level two indirect mechanisms demonstrate how classical genetic drivers govern the splicing apparatus, best exemplified by the cooperation between mutant p53 and oncogenic KRAS. In pancreatic ductal adenocarcinoma models, gain-of-function mutant p53 proteins do not bind pre-mRNA transcripts directly, but instead transcriptionally upregulate heterogeneous nuclear ribonucleoprotein K. Elevated levels of this multifunctional regulator lead to binding at cytosine-rich regulatory regions across the pre-mRNAs of multiple GTPase-activating proteins, inducing aberrant exon inclusion and intron retention that generate truncated, dysfunctional GAP isoforms. Because normal GTPase-activating proteins serve as essential negative regulators by accelerating the hydrolysis of KRAS-bound GTP to GDP, their functional impairment locks KRAS in a constitutively active state, sustaining downstream RAF-MEK-ERK and PI3K-Akt signaling. This feed-forward circuitry is strictly restricted by tumor genotype, requiring concurrent TP53 and KRAS mutations. Disrupting this axis by blocking cytosine-rich exon inclusion or pharmacologically modulating the spliceosome with small molecules such as H3B-8800 selectively halts tumor growth in mutant preclinical models, revealing a genotype-dependent synthetic vulnerability generated by aberrant splicing control.

Co-transcriptional mechanisms further integrate chromatin states and transcription elongation kinetics directly into splice-site decisions, functioning primarily as level one modulators of the nascent pre-mRNA substrate alongside level two inputs that alter splicing factor expression. Because spliceosome assembly occurs cotranscriptionally while RNA polymerase II is actively engaged, splicing outcomes are governed by recruitment mechanisms mediated by the polymerase carboxyl-terminal domain and by kinetic coupling models where transcription elongation speed dictates the temporal window for splice-site recognition. In pancreatic cancer, microenvironmental cues, extensive desmoplasia, and epithelial-to-mesenchymal transition reshape chromatin landscapes to dictate isoform switching at loci including CD44, FGFR2, and TMC7. During mesenchymal transition, transcription factors such as Snail and ZEB1 recruit methyltransferases G9a and SUV39H1 to deposit repressive H3K9me2 and H3K9me3 marks across variable CD44 exons. Concurrently, loss of epithelial splicing regulators ESRP1 and ESRP2 allows the mesenchymal factor hnRNPM to coordinate a switch from variable CD44v isoforms to the standard CD44s isoform, generating invasive, gemcitabine-resistant cancer stem cell populations mediated by GABRP signaling. Additionally, localized tumor hypoxia stabilizes hypoxia-inducible factor-1alpha to remodel chromatin and elongate through TMC7 exon seventeen in cooperation with the MBNL1-018 splicing variant, whereas histone methyltransferase NSD2 deposits H3K36me2 marks to orchestrate mesenchymal splicing across transcripts including FLNA, LMBRD1, and COL6A3.

Epitranscriptomic modifications and competing non-coding RNA networks add further regulatory complexity to the post-transcriptional landscape of pancreatic cancer cells. Deposition of internal N6-methyladenosine marks by the METTL3-METTL14-WTAP methyltransferase complex controls alternative splicing through RNA secondary structure remodeling that exposes protein-binding motifs, or via direct recruitment of nuclear readers like YTHDC1 to dictate exon inclusion. In pancreatic cancer, the kinase CLK1 phosphorylates SRSF5 to alter METTL14 alternative splicing, amplifying m6A deposition, while METTL3 overexpression stabilizes the spliceosomal helicase DDX23 to promote gemcitabine resistance. Adenosine-to-inosine editing catalyzed by the ADAR family represents another epitranscriptomic layer capable of creating or abolishing splice sites. In pancreatic adenocarcinoma, however, the predominant documented role of ADAR1-p150 operates at level three, where it deaminates double-stranded inverted Alu repeats within intronic and untranslated regions to prevent innate immune sensing by MDA5 and RIG-I receptors, thereby attenuating interferon signaling and reinforcing an immunologically cold microenvironment. Intersecting with these pathways are circular RNAs and long non-coding RNAs that function as competing endogenous RNA sponges, such as circARFGEF2 which sequesters miR-1205 to activate oncogenic JAK2-STAT3 signaling, and circMYO1C whose m6A modification stabilizes PD-L1 expression to drive immune evasion.

These intricate regulatory dependencies have spurred intense interest in splicing-directed therapeutic strategies, spanning core spliceosome modulators, splicing kinase inhibitors, and transcript-specific splice-switching antisense oligonucleotides. Pharmacological small molecules targeting the SF3b subunit of the U2 small nuclear ribonucleoprotein, such as pladienolide derivatives, E7107, and orally bioavailable H3B-8800, competitively disrupt branch point recognition to induce intron retention in essential survival genes, triggering selective apoptosis in cancer cells operating near their splicing capacity. Splice-switching antisense oligonucleotides screened in patient-derived organoid platforms demonstrate the feasibility of redirecting transcript-level splice choices to restore normal protein isoforms. Furthermore, splice variant protein products present cell-surface therapeutic vulnerabilities, exemplified by claudin-18.2, a stomach-specific isoform generated through alternative first exon usage that is aberrantly retained in approximately thirty percent of resected pancreatic tumors and targeted by monoclonal antibodies like zolbetuximab, antibody-drug conjugates, and CAR-T cell therapies. Pharmacological spliceosome disruption also enhances tumor immunogenicity by generating aberrant intron-retaining peptides presented on major histocompatibility complex class I molecules, providing a compelling mechanistic rationale for combining splicing modulators with PD-1 or PD-L1 immune checkpoint inhibition.

Despite encouraging preclinical findings, translating RNA-targeted therapies into clinical practice faces formidable obstacles that explain why no splicing-directed therapeutic has yet demonstrated confirmed clinical efficacy in pancreatic cancer. Historical clinical trials in other oncology indications illustrate narrow therapeutic windows, evidenced by the termination of E7107 due to dose-limiting ocular and gastrointestinal toxicities and the lack of objective clinical responses with H3B-8800 in myeloid malignancies. In pancreatic ductal adenocarcinoma, clinical translation is further hindered by the dense fibrotic stroma that severely impedes drug perfusion, the rapid emergence of compensatory splicing adaptation under drug pressure, and low baseline major histocompatibility complex class I expression that dampens neoantigen-driven immune responses. Overcoming these barriers will require transitioning from traditional gene-level bulk sequencing to single-cell and long-read spatial transcriptomic profiling capable of resolving cell-type-specific isoform switches within distinct microenvironmental niches. Ultimately, successfully navigating these challenges will hinge on integrating multi-omic profiling, organoid validation platforms, and robust isoform biomarkers to stratify patients who can derive tangible clinical benefit from splicing-directed precision medicine.

Subject of Research: Multilayered regulation of alternative RNA splicing in pancreatic ductal adenocarcinoma

Article Title: Alternative splicing as a multilayered regulatory hub in pancreatic ductal adenocarcinoma

Article References: Jiménez-Lozano, S., Hernández-Munain, C., & Suñé, C. (2026). Alternative splicing as a multilayered regulatory hub in pancreatic ductal adenocarcinoma. Frontiers in Molecular Medicine, 6, Article 2009133. https://doi.org/10.3389/fmmed.2026.2009133

Image Credits: AI Generated

DOI: 10.3389/fmmed.2026.2009133

Keywords: pancreatic ductal adenocarcinoma, alternative splicing, epitranscriptomics, tumor plasticity, RNA-binding proteins, epigenetic regulation, spliceosome, targeted cancer therapy, Alternative, splicing, multilayered, regulatory