Figure S2. Cellular compositions of myeloid and T cell subclusters between hepatic metastases and healthy livers.

GSEA results for all ductal subclusters using genesets collected from published articles.

GO functional enrichment analysis of genes downregulated in immunotherapy treated mice.

Figure S4. Association between H2AFZ expression and genomic mutations in PDAC.

Cell markers for celltype annotation.

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignancy with a poor prognosis, in which the role of lipophagy, a selective autophagic process degrading lipid droplets (LDs), remains poorly characterized. This study investigated lipophagy and its key regulator, OSBPL10, in PDAC progression. Through immunofluorescence analysis of patient samples, transgenic mouse tissues, and cell lines, we find that lipophagy is elevated in PDAC and correlates with poor prognosis. Single-cell transcriptomic analysis identified OSBPL10 as a critical lipophagy regulator and an independent clinicopathological indicator. Functional assays, including orthotopic and subcutaneous xenografts, demonstrated that OSBPL10 promotes tumor growth. Mechanistically, OSBPL10 functionally cooperates with VAPA/VAPB to facilitate rapid lysosomal repair via ATG2A, thereby promoting lipophagy and lipid mobilization. Inhibition of lysosomal function abrogated the pro-lipophagic and pro-tumorigenic effects of OSBPL10. Collectively, our findings demonstrate that upregulated OSBPL10 drives PDAC progression by enhancing lipophagy through ATG2A-mediated rapid lysosomal repair, highlighting OSBPL10 as a potential therapeutic target in PDAC.

Figure S3. Analysis of bulk RNA-seq data on metastatic tumor tissues of anti-PD1 monotherapy, anti-LAG3 monotherapy and anti-PD1 plus anti-LAG3 combined therapy.

PURPOSE:Hepatic metastasis (HM) is the leading cause of death in pancreatic ductal adenocarcinoma (PDAC). However, the underlying cellular and molecular programs remain poorly understood, leading to limited therapeutics for this disease. EXPERIMENTAL DESIGN:In this study, we integrated single-cell RNA sequencing data from paired primary tumors and HMs, along with bulk RNA sequencing and IHC data from hundreds of patients to elucidate metastasis-associated programs. RESULTS:Our analysis identified a metastasis-prone malignant subpopulation, which is associated with a higher risk of HM and a transitional plastic state. This malignant subpopulation represents a poorly differentiated and highly proliferative phenotype, with H2AFZ potentially contributing to this phenomenon. Moreover, the presence of tumor cells in the liver was accompanied by an increased abundance of M2 macrophages, regulatory T cells, and exhausted T cells (Tex) in HMs compared with adjacent tissues, indicative of a shift toward an immunosuppressive environment. Notably, within the tumor environment of HMs, Tex exhibited elevated expression of PDCD1 and LAG3. The combined therapy targeting these two genes effectively inhibited tumor growth in mouse models of metastatic PDAC. CONCLUSIONS:In conclusion, we reveal a metastasis-associated malignant subpopulation and provide a promising therapeutic strategy for metastatic PDAC.

Signature genes of M1 and M2 macrophages derived from PMID: 33545035.

The complex interplay between nerves, immunity, and tumor progression remains poorly understood, particularly in the context of chemotherapy. Here, we investigated how neural remodeling influences tertiary lymphoid structures (TLSs) and clinical outcomes following neoadjuvant chemotherapy (NAT) in pancreatic ductal adenocarcinoma (PDAC). Using tissue samples from 86 treatment-naïve and 49 NAT-treated patients with PDAC, we demonstrated that chemotherapy significantly increases both nerve density (ND) and TLS abundance. Notably, nerve-proximal TLSs (N-TLSs) displayed more mature phenotypes and correlated positively with tumor regression. Spatial transcriptomics of nerve regions showed chemotherapy-induced transcriptional reprogramming of Schwann cells, marked by altered myelination programs and elevated pro-inflammatory signaling. The Schwann cell state shift coincides with TLS accumulation, maturation, and enhanced peri-neural immune infiltration. Collectively, our study indicates a spatially organized neuro-immune axis linking neural remodeling to TLS abundance and maturation after chemotherapy and nominates N-TLS abundance as a potential histological biomarker of treatment response in resected PDAC.

Signature genes of all ductal subclusters.

GO functional enrichment analysis of genes upregulated in immunotherapy treated mice.

Clinical information of patients with metastatic PDAC.

Figure S1. Functional interrogation of ductal subclusters and the prognostic value of PFN1.

Cellular adaptive volume regulation is essential for maintaining metabolic homeostasis and supporting survival, yet its role in desmoplastic pancreatic ductal adenocarcinoma (PDAC) remains incompletely understood. Through comprehensive bioinformatic and functional studies, we identified LRRC8A, the core subunit of volume-regulated anion channels (VRAC), as a central regulator linking volume homeostasis to PDAC progression. Beyond its established role in osmotic stress responses, genetic silencing or pharmacological inhibition of LRRC8A revealed its critical function in proliferation-associated volumetric expansion during S-phase. Functional validation through in vitro proliferation assays, in vivo xenograft models, and patient-derived pancreatic cancer organoids (PDO) demonstrated that LRRC8A critically drives PDAC progression. Mechanistically, LRRC8A coordinates plasma membrane dynamics, cortical cytoskeletal organization, membrane-delimited oncogenic signaling (KRAS/EGFR), and nucleolar ribosome biogenesis to support volumetric expansion during S phase. Co-immunoprecipitation coupled with mass spectrometry identified that LRRC8A forms complexes with Caveolin 1 (CAV1). Disruption of LRRC8A leads to decreased CAV1 protein levels, impaired activation of KRAS and EGFR oncogenic signaling, and suppressed ribosome biogenesis and global protein synthesis. Reciprocally, CAV1 knockdown or cholesterol depletion using lovastatin destabilized LRRC8A in plasma membrane, resulting in reduced cortical F-actin organization, oncogenic signaling and biosynthetic activity, indicating that LRRC8A and CAV1 are mutually stabilized and depend on cholesterol-rich membrane microdomains for proper integration and function. Furthermore, disruption of LRRC8A-CAV1 axis through LRRC8A inhibition or cholesterol depletion potently suppressed PDO growth in vitro. Collectively, our work establishes the LRRC8A-CAV1 complex as a key coordinator of biosynthetic expansion and a promising therapeutic target in pancreatic cancer.

GSEA results for all ductal subclusters using Gene Ontology (GO) terms.

Metabolic reprogramming is a hallmark of cancer. The“Warburg effect”, also known as aerobic glycolysis, is an essential part of metabolic reprogramming and a central contributor to cancer progression. Moreover, hypoxia is one of the significant features of pancreatic ductal adenocarcinoma (PDAC). Under hypoxic conditions, the “Warburg effect” occurs to meet the nutrient and energy demands of rapid genome replication, remodeling the tumor microenvironment (TME) and influencing tumor immunity. α-Enolase (ENO1) is a multifunctional protein, acting as a glycolytic enzyme that catalyzes the conversion of 2-phosphoglyceric acid to phosphoenolpyruvic acid. ENO1 was found to be overexpressed in multiple types of cancers. Here, we investigated the role of ENO1 in modulating the PDAC microenvironment.Using bioinformatic analyses, we demonstrated that ENO1 was highly expressed in PDAC patients, which was related to a poor prognosis. In vitro, Eno1 knockdown resulted in reduced PDAC cell proliferation and colony formation, along with enhanced apoptosis in PDAC cells. In vivo, tumorigenesis was suppressed in mouse PDAC models by Eno1 knockdown. Flow cytometry analysis revealed that high expression of Eno1 altered the tumor immune microenvironment (TIME), particularly the impaired tumor infiltration and function of CD8+ T cells. Mechanistic studies revealed that ENO1 upregulated PD-L1 to prevent CD8+ T cells infiltration through the hypoxia-inducible factor (HIF)-1α signaling pathway, leading to PDAC progression.In conclusion, our findings indicate that ENO1 might serve as a potential biomarker for PDAC and a novel onco-immunotherapeutic target via its role in altering the TIME.

The peripheral nervous system significantly determines the fate of solid tumors and their microenvironment. In neurotropic malignancies such as pancreatic and prostate cancer, denervation in animal models demonstrate significantly delays in tumor initiation and progression, underscoring the critical neural dependency of these cancers. While tumor innervation establishes a structural basis for the neuromodulatory effects, the degree of innervation exhibits marked heterogeneity across tumor types, and its regulatory mechanisms remain poorly characterized. In this study, we screened genes associated with innervation status in pancreatic cancer and identified the splicing factor SRSF12 as a critical gene related to tumor innervation. In clinical samples, SRSF12 was expressed at low levels in pancreatic cancer tissues, and its downregulation was linked to poor prognosis in patients. Then we crossed Kras mutation and Srsf12 knockout mice (KrasG12DSrsf12 fl/fl) together with Srsf12 fl/flPdx1cre mice and found that depletion of Srsf12 accelerated Kras-driven pancreatic tumorigenesis and enhanced tumor innervation. Furthermore, we demonstrated that SRSF12 inhibits neurite outgrowth primarily by generating a LAMA3 splice isoform that lacks the fourth and fifth LG (G45) domains. Mechanistically, G45 promotes tumor innervation by activating ITGB1 and FAK in neurons. Together, our findings delineate SRSF12 as a novel suppressor of tumor innervation and pancreatic tumorigenesis, while also identifying a tumor-specific target for SRSF12-deficient pancreatic cancer.

Rationale: Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy with a global prevalence and poor prognosis, largely due to immune escape mechanisms. However, the potential reasons for the decreased infiltration of cytotoxic T lymphocytes (CTLs) in PDAC remain inadequately understood. In this study, we aimed to elucidate the molecular mechanisms contributing to the low-CTLs infiltration in patients with PDAC. Methods: Bioinformatic analyses were used to identify key factors associated with low-CTLs infiltration in PDAC and the role of oligoadenylate synthetase-like (OASL) was mainly focused in our study. Immunohistochemistry (IHC) was used to assess the relationship between the expression of OASL and the prognosis of patients. Western blotting, Flow cytometry, Co-immunoprecipitation and Immunofluorescence were applied to elucidate the molecular mechanism by which OASL mediates immune escape in PDAC. The orthotopic PDAC models were constructed to evaluate the effects of OASL-knockdown on CD8+ T cells infiltration and tumor growth in vivo. Results: OASL was found to be significantly upregulated in PDAC and negatively correlated with the major histocompatibility complex class I (MHC-I) expression, which is associated with worse patient prognosis. Notably, OASL-knockdown leads to a significant increase in CD8+ T cell infiltration and slows tumor growth in vivo. Mechanistic studies revealed that OASL -knockdown restored the total and surface MHC-I level through impairing neighbor of BRCA1 gene 1 (NBR1)-mediated autophagy-lysosomal degradation of MHC-I. Conclusions: Targeting OASL enhances the immune response in PDAC, providing a novel therapeutic strategy to improve outcomes in PDAC patients.

Background: Pancreatic cancer (PCa) is one of the most malignant diseases in the world. Different from ferroptosis and apoptosis, disulfidptosis is a novel type of cell death. The role of disulfidptosis in PCa remains uncovered. Methods: Disulfidptosis-related lncRNAs were identified based on TCGA-PAAD database. The disulfidptosis-related predict signature was constructed and verified by bioinformatic analysis. TCGA and GTEx database and Renji tissue microarray (TMA) were applied to determine TMEM105 and its clinical significance. F-actin and PI staining were performed to detect disulfidptosis of PCa cells. The biological function of TMEM105 was investigated by loss-of-function and gain-of-function assays. RNA pull-down and LC-MS/MS analysis were employed to detect TMEM105 interacted proteins. The tissue samples from PCa patients with PET-CT information were utilized to validate the TMEM105-β-catenin-c-MYC-GLUT1 pathway in clinical settings. Results: A disulfidptosis-related predict signature, which was comprised of six lncRNAs, was constructed and validated by bioinformatic analysis. TMEM105 was identified as disulfidptosis-related lncRNA whose high expression predicted a poor prognosis in PCa. Functional studies revealed that TMEM105 promoted the growth and mitigated the disulfidptosis in PCa. Mechanically, TMEM105 upregulated the expression of β-catenin by maintaining the protein stability through the proteosome pathway. The forced expressed β-catenin increased the expression of glycolysis-related transcription factor c-MYC, thus induced the transcription activity of GLUT1. Conclusion: These results revealed the growth acceleration and the disulfidptosis mitigation function of TMEM105 in PCa. Targeting the TMEM105-β-catenin-c-MYC-GLUT1 pathway could be a potent therapy for PCa patients.