SRSF1 orchestrates splicing-directed dual-engine glucose metabolism driving chemoresistance in diabetes-associated pancreatic cancer

Background: Pancreatic ductal adenocarcinoma (PDAC) patients with concurrent type 2 diabetes mellitus (T2DM) exhibit significantly worse treatment responses and shorter survival rates. However, the molecular mechanisms underlying diabetes-driven chemoresistance and immune suppression remain poorly understood. This study aims to elucidate how the diabetic microenvironment promotes PDAC chemoresistance through the splicing-metabolism axis. Methods: We investigated SRSF1-mediated splicing regulation using clinical samples from 28 advanced PDAC patients receiving AG chemotherapy, T2DM-associated PDAC mouse models (KPC, KPSC), patient-derived organoids, and multiple in vivo models. Mechanistic studies included scRNA-seq, eCLIP-seq, ATAC-seq, RIP, ChIP-qPCR, and metabolic flux analysis. Results: This study revealed that pancreatic cancer with concurrent diabetes is prone to chemoresistance, accompanied by SRSF1-driven elevated back-splicing ratios and altered pyruvate metabolism. Mechanistic investigation found that lactate accumulation in the diabetic microenvironment induced K48 lactylation of SRSF1 protein, enhancing its stability and promoting back-splicing events. SRSF1 regulated the linear-to-back splicing conversion of PSMA3-AS1 to circ-PSMA3-AS1, ultimately promoting transcriptional upregulation of metabolic enzymes DLAT and LDHA. DLAT/LDHA provided metabolic resilience through enhanced mitochondrial fusion and glycolysis, enabling cells to activate both TCA cycle and glycolytic energy-producing systems to acquire chemoresistance. Additionally, this metabolic remodeling profoundly impacted the tumor immune microenvironment through upregulated PD-L1 and sustained lactate production, causing immune evasion, suppressing CD8+ T cell function, and promoting Treg expansion. This process constituted a self-reinforcing positive feedback loop, resulting in the intractability of diabetes-associated pancreatic cancer. Based on these mechanistic discoveries, we developed a triple combination therapy (Berberine + anti-PD1 + AG) targeting SRSF1, DLAT, and LDHA while reducing glucose and tumor lactate levels, significantly improving treatment efficacy in diabetic PDAC models. Conclusions: This study reveals the mechanistic framework underlying poor outcomes in diabetes-associated PDAC and provides a clinically translatable therapeutic strategy. The discovery of the splicing-metabolism-immunity axis opens new avenues for targeting metabolic vulnerabilities in cancer treatment, with particular significance for the growing population of diabetic cancer patients.