Immune marker staining, collagen features, and αSMA-associated stromal analyses in KC/AKC and KPC/AKPC tumors.

Clinical characteristics and mutation profiles of the patient-derived organoids used in this study.

Analyses of TGFβ ligand/receptor expression, reporter activity, inferred signaling interactions, and PSC differentiation in ATM-deficient tumor models.

PSC-induced chemotherapy resistance and in vitro and ex vivo assays assessing apoptosis, invasion, wound closure, and drug response after perturbation of tumor–CAF signaling.

Fibrosis and CAF marker analyses, MRI imaging, survival, proliferation, apoptosis, and flow cytometry from orthotopic treatment experiments.

Abstract Pancreatic ductal adenocarcinoma (PDAC) is among the most aggressive and metastatic malignancies worldwide. Migrating cancer stem cells (miCSCs), marked by CD133⁺CXCR4⁺ expression is a key driver of PDAC progression, which currently lack effective therapeutic targets. Activated pancreatic stellate cells (PSCs) within the tumor microenvironment secrete CXCL12, the ligand for CXCR4, thereby promoting stemness, epithelial-to-mesenchymal transition (EMT), and chemoresistance in miCSCs. Despite advances in understanding PDAC biology, clinically effective strategies that target CXCR4⁺ CSC populations remain limited. In order to investigate the molecular mechanisms sustaining miCSCs, we performed protein–protein interaction network analysis, which identified the transcription factor BMI1 as a key downstream effector of the CXCL12/CXCR4 axis. Functional studies using shRNA-mediated knockdown of CXCR4 and BMI1 were conducted to assess their roles in miCSC migration, EMT, and self-renewal. We further evaluated the therapeutic potential of the endogenous CXCR4 antagonist EPI-X4 and its optimized derivative JM#21 in PDAC cell lines. We addressed the peptide stability by encapsulating JM#21 into mesoporous silica nanoparticles (MSNs) designed for improved half-life and sustained release under physiological conditions. BMI1 was confirmed as a critical mediator of CXCL12/CXCR4-driven stemness and EMT. Knockdown of CXCR4 or BMI1 significantly impaired miCSC maintenance and migration towards CXCL12. Both EPI-X4 and JM#21 potently inhibited CXCL12-mediated signaling, reduced EMT and stemness markers, and suppressed miCSC migratory potential. JM#21 displayed superior efficacy and re-sensitized previously resistant PDAC cell lines to gemcitabine and paclitaxel. Functional assays demonstrated that nanoparticle-loaded JM#21 more effectively suppressed EMT markers and self-renewal than the free peptide, highlighting the advantage of nanoparticle delivery in therapeutic applications. Given their biocompatibility and modularity, silica nanoparticles offer a promising platform for stabilizing peptide drugs. Our findings reveal that tumor–stroma crosstalk via the CXCL12/CXCR4/BMI1 axis plays a central role in sustaining miCSC-driven metastasis and therapy resistance in PDAC. Targeting this signaling pathway with nanoparticle-stabilized JM#21 represents a novel and clinically promising therapeutic strategy to disrupt PDAC progression and improve the efficacy of existing combination treatments.

ROS measurements, cytoskeletal marker expression, migration assays, and PSC differentiation analyses following genetic and pharmacologic perturbation of ROS and cytoskeletal pathways in ATM-deficient tumor models.

Number and proportion of cell types identified by single-nucleus multiomic analysis.

Single-cell analyses of BMP4 and CAF lineage signatures, CAF differentiation assays in PSC- (with BMP pathway perturbation) and CAF-based models, and subcutaneous assay with MRTX1133 and galunisertib.

Defining the basal-like A-specific enhancer landscape in PDAC

Defining the specificity and P63-dependence of B-STEP

Supplementary Table S3 contains list of genes associated with Gö4Pdx4 super enhancers overlapping TP63.

Supplementary Table S4 contains L3.6pl TP63-dependent enhancer regions.

Verification of P63-dependence and B-STEP enhancer-gene regulations.

Supplementary Table S5 contains list of genes associated with L3.6pl TP63 dependent enhancers.

The tumor microenvironment (TME) actively contributes to pancreatic ductal adenocarcinoma (PDAC) pathogenesis through dynamic bidirectional tumor–stroma interactions. Here, we demonstrate that homologous recombination-defective (HRD) tumor epithelium reprograms the TME in a genotype-specific manner to enhance cancer aggressiveness. Using genetically engineered mouse models, pancreatic stellate cell (PSC) and cancer-associated fibroblast (CAF) co-culture systems, single-nucleus multiomics, and human PDAC models, we show that tumoral loss of ATM serine/threonine kinase drives CAFs toward αSMA+ myofibroblastic differentiation, independently of P53 status. These myCAFs, in turn, promote cancer aggressiveness and chemoresistance. Mechanistically, ATM deficiency increases reactive oxygen species and contractility signaling, enhancing TGF-β1 secretion. Pharmacological TGF-β inhibition reverses myCAF differentiation, sensitizes tumors to chemotherapy, and impairs tumor progression in both murine and human ATM-null models. Our findings reveal that ATM-deficient tumors shape a cancer-promoting niche via TGF-β signaling and identify dual targeting of intrinsic and extrinsic vulnerabilities as a promising precision oncology strategy. SIGNIFICANCE HRD pancreatic cancers reprogram the tumor microenvironment in a genotype-specific manner through TGF-β-driven myCAF-enrichment. Targeting this stromal axis alongside platinum-based chemotherapy improves therapeutic efficacy in ATM-deficient models. These findings highlight the need to integrate epithelial genotype and stromal context for truly personalized treatment strategies in PDAC. ### Competing Interest Statement T. Seufferlein reports grants and personal fees from Celgene and Sanofi, personal fees from Amgen, AstraZeneca, Bayer, the Falk Foundation, Lilly, Merck-Serono, Merck, Pierre Fabre, Roche, Servier, and Shire, and grants from Boehringer Ingelheim outside the submitted work. A. Stenzinger reports grants and personal fees from Bayer and BMS, personal fees from AGCT, Astra Zeneca, Eli Lilly, Illumina, Janssen, MSD, Novartis, Pfizer, Roche, Seattle Genetics, Takeda, and Thermo Fisher, and grants from Chugai and Incyte outside the submitted work. L. Perkhofer reports nonfinancial support from Ipsen, personal fees from AstraZeneca and Servier outside the submitted work. A. Kleger reports personal fees from the Falk foundation outside the submitted work. No disclosures were reported by the other authors. German Cancer Aid, https://ror.org/01wxdd722, 70114761, 70115292 Deutsche Forschungsgemeinschaft, https://ror.org/018mejw64, KL 2544/6-1, PE 3337/1-1, GRK 2254/1 University Hospital Ulm, https://ror.org/05emabm63, L.SBN.0193

The file contains additional supplementary methods

Supplementary Table S2 contains GöPdx4 super enhancers regions overlapping TP63

Supplementary Table S6 contains list of qPCR primers used in the study.

Abstract Background Pancreatic ductal adenocarcinoma (PDAC) requires innovative therapeutic strategies to counteract its progression and metastatic potential. Since the majority of patients are diagnosed with advanced metastatic disease, treatment strategies targeting not only the primary tumor but also metastatic lesions are needed. Tumor-Associated Macrophages (TAMs) have emerged as central players, significantly influencing PDAC progression and metastasis. Our objective was to validate an innovative therapeutic strategy involving the reprogramming of TAMs using lipid nanosystems to prevent the formation of a pro-metastatic microenvironment in the liver. Results In vitro results demonstrate that M2-polarized macrophages lose their M2-phenotype following treatment with lipid nanoemulsions composed of vitamin E and sphingomyelin (VitE:SM), transitioning to an M0/M1 state. Specifically, VitE:SM nanoemulsion treatment decreased the expression of macrophage M2 markers such as Arg1 and Egr2, while M1 markers such as Cd86, Il-1b and Il-12b increased. Additionally, the TGF-βR1 inhibitor Galunisertib (LY2157299) was loaded into VitE:SM nanoemulsions and delivered to C57BL/6 mice orthotopically injected with KPC PDAC tumor cells. Treated mice showed diminished primary tumor growth and reduced TAM infiltration in the liver. Moreover, we observed a decrease in liver metastasis with the nanoemulsion treatment in an intrasplenic model of PDAC liver metastasis. Finally, we validated the translatability of our VitE:SM nanosystem therapy in a human cell-based 3D co-culture model in vivo, underscoring the pivotal role of macrophages in the nanosystem’s therapeutic effect in the context of human PDAC metastasis. Conclusions The demonstrated effectiveness and safety of our nanosystem therapy highlights a promising therapeutic approach for PDAC, showcasing its potential in reprogramming TAMs and mitigating the occurrence of liver metastasis. Graphical abstract