ABSTRACT
Purpose: Chemotherapy is pivotal in the multimodal treatment of pancreatic ductal adenocarcinoma (PDAC). Technical advances unveiled a high degree of inter- and intratumoral heterogeneity. We hypothesized that intratumoral heterogeneity (ITH) impacts response to gemcitabine treatment and demands specific targeting of resistant subclones. Methods: Using single cell-derived cell lines (SCDCLs) from the classical cell line BxPC3 and the basal-like cell line Panc-1, we addressed the effect of ITH on response to gemcitabine treatment. Results: Individual SCDCLs of both parental tumor cell populations showed considerable heterogeneity in response to gemcitabine. Unsupervised PCA including the 1,000 most variably expressed genes showed a clustering of the SCDCLs according to their respective sensitivity to gemcitabine treatment for BxPC3, while this was less clear for Panc-1. In BxPC3 SCDCLs, enriched signaling pathways EMT, TNF signaling via NfKB, and IL2STAT5 signaling correlated with more resistant behavior to gemcitabine. In Panc-1 SCDCLs MYC targets V1 and V2 as well as E2F targets were associated with stronger resistance. We used recursive feature elimination for Feature Selection in order to compute sets of proteins that showed strong association with the response to gemcitabine. The optimal protein set calculated for Panc-1 comprised fewer proteins in comparison to the protein set determined for BxPC3. Based on molecular profiles, we could show that the gemcitabine-resistant SCDCLs of both BxPC3 and Panc-1 are more sensitive to the BET inhibitor JQ1 compared to the respective gemcitabine-sensitive SCDCLs. Conclusion: Our model system of SCDCLs identified gemcitabine-resistant subclones and provides evidence for the critical role of ITH for treatment response in PDAC. We exploited molecular differences as the basis for differential response and used these for more targeted therapy of resistant subclones.
ABSTRACT
The prognosis of pancreatic ductal adenocarcinoma (PDAC) is exceedingly poor. Although surgical resection is the only curative treatment option, multimodal treatment is of the utmost importance, as only about 20% of tumors are primarily resectable at the time of diagnosis. The choice of chemotherapeutic treatment regimens involving gemcitabine and FOLFIRINOX is currently solely based on the patient's performance status, but, ideally, it should be based on the tumors' individual biology. We established two novel patient-derived primary cell lines from surgical PDAC specimens. LuPanc-1 and LuPanc-2 were derived from a pT3, pN1, G2 and a pT3, pN2, G3 tumor, respectively, and the clinical follow-up was fully annotated. STR-genotyping revealed a unique profile for both cell lines. The population doubling time of LuPanc-2 was substantially longer than that of LuPanc-1 (84 vs. 44 h). Both cell lines exhibited a typical epithelial morphology and expressed moderate levels of CK7 and E-cadherin. LuPanc-1, but not LuPanc-2, co-expressed E-cadherin and vimentin at the single-cell level, suggesting a mixed epithelial-mesenchymal differentiation. LuPanc-1 had a missense mutation (p.R282W) and LuPanc-2 had a frameshift deletion (p.P89X) in TP53. BRCA2 was nonsense-mutated (p.Q780*) and CREBBP was missense-mutated (p.P279R) in LuPanc-1. CDKN2A was missense-mutated (p.H83Y) in LuPanc-2. Notably, only LuPanc-2 harbored a partial or complete deletion of DPC4. LuPanc-1 cells exhibited high basal and transforming growth factor (TGF)-ß1-induced migratory activity in real-time cell migration assays, while LuPanc-2 was refractory. Both LuPanc-1 and LuPanc-2 cells responded to treatment with TGF-ß1 with the activation of SMAD2; however, only LuPanc-1 cells were able to induce TGF-ß1 target genes, which is consistent with the absence of DPC4 in LuPanc-2 cells. Both cell lines were able to form spheres in a semi-solid medium and in cell viability assays, LuPanc-1 cells were more sensitive than LuPanc-2 cells to treatment with gemcitabine and FOLFIRINOX. In summary, both patient-derived cell lines show distinct molecular phenotypes reflecting their individual tumor biology, with a unique clinical annotation of the respective patients. These preclinical ex vivo models can be further explored for potential new treatment strategies and might help in developing personalized (targeted) therapy regimens.
