Molecular mechanisms underlying the development and progression of pancreatic neuroendocrine tumors (PanNETs) are still insufficiently understood. Efficacy of currently approved PanNET therapies is limited. While novel treatment options are being developed, patient stratification permitting more personalized treatment selection in PanNET is yet not feasible since no predictive markers are established. The lack of representative in vitro and in vivo models as well as the rarity and heterogeneity of PanNET are prevailing reasons for this. In this study, we describe an in vitro 3-dimensional (3-D) human primary PanNET culture system as a novel preclinical model for more personalized therapy selection. We present a screening platform allowing multicenter sample collection and drug screening in 3-D cultures of human primary PanNET cells. We demonstrate that primary cells isolated from PanNET patients and cultured in vitro form isletlike tumoroids. Islet-like tumoroids retain a neuroendocrine phenotype and are viable for at least 2 weeks in culture with a high success rate (86%). Viability can be monitored continuously allowing for a perwell normalization. In a proof-of-concept study, islet-like tumoroids were screened with three clinically approved therapies for PanNET: sunitinib, everolimus and temozolomide. Islet-like tumoroids display varying in vitro response profiles to distinct therapeutic regimes. Treatment response of islet-like tumoroids differs also between patient samples. We believe that the presented human PanNET screening platform is suitable for personalized drug testing in a larger patient cohort, and a broader application will help in identifying novel markers predicting treatment response and in refining PanNET therapy Molecular mechanisms underlying the development and progression of PanNET are still insufficiently understood. Efficacy of currently approved PanNET therapies is limited. While novel treatment options are being developed, patient stratification permitting more personalized treatment selection in PanNET is yet not feasible since no predictive markers are established. The lack of representative in vitro and in vivo models as well as the rarity and heterogeneity of PanNET are prevailing reasons for this.

Abdominal surgeries are lifesaving procedures but can be complicated by the formation of peritoneal adhesions, intra-abdominal scars that cause intestinal obstruction, pain, infertility, and significant health costs. Despite this burden, the mechanisms underlying adhesion formation remain unclear and no cure exists. Here, we show that contamination of gut microbes increases post-surgical adhesion formation. Using genetic lineage tracing we show that adhesion myofibroblasts arise from the mesothelium. This transformation is driven by epidermal growth factor receptor (EGFR) signaling. The EGFR ligands amphiregulin and heparin-binding epidermal growth factor, are sufficient to induce these changes. Correspondingly, EGFR inhibition leads to a significant reduction of adhesion formation in mice. Adhesions isolated from human patients are enriched in EGFR positive cells of mesothelial origin and human mesothelium shows an increase of mesothelial EGFR expression during bacterial peritonitis. In conclusion, bacterial contamination drives adhesion formation through mesothelial EGFR signaling. This mechanism may represent a therapeutic target for the prevention of adhesions after intra-abdominal surgery.

Thyrotropin receptor (TSHR), sodium iodide symporter (NIS), pendrin, and thyroid peroxidase (TPO) are essential for the uptake of iodine by follicular thyroid cells. The aim of this study was to establish immunohistochemistry (IHC) protocols for TSHR, NIS, pendrin, and TPO in canine tissues and characterize their expression in organoids derived from canine follicular cell thyroid carcinoma (FTC) and in the respective primary tumors. This constitutes a fundamental step to establish organoids as a model to study the uptake of iodine in canine FTC. Commercially available antibodies directed against human proteins were selected. Antibody specificity was confirmed by western blot using lysates of the HTori-3 human thyroid cell line and healthy canine thyroid gland. IHC was validated using HTori-3 cells and a set of canine normal tissues including healthy thyroid gland. The expression of TSHR, NIS, pendrin, and TPO was evaluated in 3 organoid lines derived from FTC and respective primary tumors. All 4 antibodies produced specific bands by western blot and cytoplasmic labeling in follicular cells by IHC in both human HTori-3 cells and canine thyroid gland. NIS also showed basolateral membrane immunolabeling in follicular cells. All 4 proteins were highly expressed in organoids derived from FTC. The expression was similar or higher compared to the primary tumors. The results of this study characterize organoids derived from canine FTC as a suitable in vitro model to investigate iodine uptake, opening new research possibilities in the field of canine thyroid cancer therapy.

Pancreatic neuroendocrine neoplasms are epigenetically driven tumors, but therapies against underlying epigenetic drivers are currently not available in the clinical practice. We aimed to investigate EZH2 (Enhancer of Zest homolog) expression in PanNEN and the impact of EZH2 inhibition in three different PanNEN preclinical models. EZH2 expression in PanNEN patient samples (n = 172) was assessed by immunohistochemistry and correlated with clinico-pathological data. Viability of PanNEN cell lines treated with EZH2 inhibitor (GSK126) was determined in vitro. Lentiviral transduction of shRNA targeting EZH2 was performed in QGP1 cells, and cell proliferation was measured. Rip1TAG2 mice underwent GSK126 treatment for three weeks starting from week 10 of age. Primary cells isolated from PanNEN patients (n = 6) were cultivated in 3D as islet-like tumoroids and monitored for 10 consecutive days upon GSK126 treatment. Viability was measured continuously for the whole duration of the treatment. We found that high EZH2 expression correlated with higher tumor grade (p < 0.001), presence of distant metastases (p < 0.001), and shorter disease-free survival (p < 0.001) in PanNEN patients. Inhibition of EZH2 in vitro in PanNEN cell lines and in patient-derived islet-like tumoroids reduced cell viability and impaired cell proliferation, while inhibition of EZH2 in vivo in Rip1TAG2 mice reduced tumor burden. Our results show that EZH2 is highly expressed in high-grade PanNENs, and during disease progression it may contribute to aberrations in the epigenetic cellular landscape. Targeting EZH2 may represent a valuable epigenetic treatment option for patients with PanNEN.

Pancreatic neuroendocrine neoplasms (PanNENs) are the second most common malignancy of the pancreas. Surgery remains the only curative treatment for localized disease. For patients with inoperable advanced or metastatic disease, few targeted therapies are available, but their efficacy is unpredictable and variable. Exploiting prior knowledge on pathogenetic processes involved in PanNEN tumorigenesis, we tested buparlisib (PI3K inhibitor) and ribociclib (CDK4/6 inhibitor), as single agents or in combination, in different preclinical models. First, we used cell lines representative of well-differentiated (INS-1E, NT-3) and poorly differentiated (BON-1) PanNENs. The combination of buparlisib with ribociclib reduced the proliferation of 2D and 3D spheroid cultures more potently than the individual drugs. Buparlisib, but not ribociclib, induced apoptosis. The anti-proliferative activity of the drugs correlated with downstream target inhibition at mRNA and protein levels. We then tested the drugs on primary islet microtissues from a genetic PanNET animal model (Men1-defective mice) and from wild-type mice: the drug combination was effective against the former without altering islet cell physiology. Finally, we treated PanNET patient-derived islet-like 3D tumoroids: the combination of buparlisib with ribociclib was effective in three out of four samples. Combined targeting of PI3K and CDK4/6 is a promising strategy for PanNENs spanning various molecular and histo-pathological features.

Cancer (CIBERONC) 3 Biobank and Biomodels Platform (PT20/0045), ISCIII research and development platforms in biomedicine and health

There are no therapeutic predictive biomarkers or representative preclinical models for high-grade gastroenteropancreatic neuroendocrine neoplasms (GEP-NEN), a highly aggressive, fatal, and heterogeneous epithelial malignancy. We established patient-derived (PD) tumoroids from biobanked tissue samples of advanced high-grade GEP-NEN patients and used this model for targeted ex vivo pharmacotyping, next-generation sequencing, and perturbational profiling. We used tissue-matched PD tumoroids to profile individual patients, compared ex vivo drug sensitivity to patients' clinical response to chemotherapy, and investigated treatment-induced adaptive stress responses. PD tumoroids recapitulated biological key features of high-grade GEP-NEN and mimicked clinical response to cisplatin and temozolomide ex vivo. When we investigated treatment-induced adaptive stress responses in PD tumoroids in silico, we discovered and functionally validated Lysine demethylase 5A and interferon-beta, which act synergistically in combination with cisplatin. Since rapid ex vivo drug screens in PD tumoroids matched clinical patient responses to standard-of-care chemotherapeutics for GEP-NEN, our functional precision oncology approach could expand personalized therapeutic options for patients with advanced high-grade GEP-NEN.